Democracy and Trade Policy in Developing Countries 9780226358956

Citation preview

Democracy and Trade Policy in Developing Countries

Chicago Series on International and Domestic Institutions

edited by william g. howell and jon pevehouse Other Books in the Series the american warfare state: the domestic politics of military spending  by Rebecca U. Thorpe (2014) the wartime president: executive influence and the nationalizing politics of threat  by William G. Howell (2013) the judicial power of the purse: how courts fund national defense in times of crisis  by Nancy C. Staudt (2011) securing approval: domestic politics and multilateral authorization for war  by Terrence L. Chapman (2011) after the rubicon: congress, presidents and the politics of waging war  by Douglas L. Kriner (2010)

Democracy and Trade Policy in Developing Countries

Bumba Mukherjee

The University of Chicago Press Chicago and London

Bumba Mukherjee is associate professor of  political science at Pennsylvania State University and a visiting fellow at the University of  Notre Dame’s Kellogg Institute for International Studies. He is coauthor of two books: Democracy, Electoral Systems, and Judicial Empowerment in Developing  Countries and The Politics of  Corruption in Authoritarian Regimes. The University of Chicago Press, Chicago 60637 The University of Chicago Press, Ltd., London

© 2016 by The University of Chicago All rights reserved. Published 2016. Printed in the United States of America 22 21 20 19 18 17 16   1 2 3 4 5 ISBN-­13: 978-­0-­226-­35878-­9 (cloth) ISBN-­13: 978-­0-­226-­35881-­9 (paper) ISBN-­13: 978-­0-­226-­35895-­6 (e-­book) DOI: 10.7208/chicago/9780226358956.001.0001 Library of Congress Cataloging-in-Publication Data Names: Mukherjee, Bumba, author. Title: Democracy and trade policy in developing countries / Bumba Mukherjee. Other titles: Chicago series on international and domestic institutions. Description: Chicago ; London : The University of Chicago Press, 2016. | Series: Chicago series on international and domestic institutions | Includes bibliographical references and index. Identifiers: LCCN 2015038334 | ISBN 9780226358789 (cloth : alk. paper) | ISBN 9780226358819 (pbk. : alk. paper) | ISBN 9780226358956 (e-book) Subjects: LCSH: Protectionism—Political aspects—Developing countries. | Protectionism—Political aspects—Developing countries—Mathematical models. | Democratization—Economic aspects—Developing countries. | Developing countries—Commercial policy. Classification: LCC HF2580.9 .M85 2016 | DDC 382/.73091724—dc23 LC record available at http:// lccn.loc.gov/2015038334 ♾ This paper meets the requirements of ANSI/NISO Z39.48-­1992 (Permanence of Paper).

C ont e nts

Acknowledgments / vii One

/ Democracy and Trade Policy in Developing Countries / 1

T wo

Three

/ Trade Protection and Electoral Malpractice in New Democratic Regimes / 29

/ Trade Protection and Electoral Fraud in New Democracies: The Empirical Evidence / 59 Four

Five

Six

Seven

/ Political Particularism and Trade Policy in Developing Democracies / 107

/ Empirical Tests for Political Particularism, Trade Protection, and Contributions / 129 / Democracy, Political Particularism, and Trade Liberalization in Brazil / 155

/ Trade Politics and Contributions in India and South Africa / 187 Ei g h t

/ Conclusion / 225

Appendix: Mathematical Proofs / 241 Notes / 257 References / 279 Index / 313

Ac k no w l e d g m e nts

Many people have helped me in the course of writing this book. First and foremost, I would like to express my gratitude to the three editors who gently but persistently guided my manuscript to fruition as a book: John Tryneski, Jon Pevehouse, and David Pervin. Their immense patience, unstinting support, encouragement, and advice helped me to both improve and complete this book. I am fortunate to have found a home for this man­ uscript with the University of Chicago Press and would particularly like to thank Rodney Powell, Jillian Tsui, Jenni Fry, and Shenyun Wu for their help with preparing the manuscript. I would like to thank Sergio Bejar and Minhyung Joo for outstanding research assistance and am grateful to Ben­ jamin Bagozzi and Raymond Hicks for their help on the project. I would also like to express my deep gratitude to the anonymous reviewers for both useful critiques and valuable insights that played an important role in the development of this book. Their guidance, feedback, and (in many ways) enthusiastic encouragement made this project feasible and provided the necessary motivation for completing the book. This book has benefitted considerably from their feedback. I take full responsibility for all errors and omissions in this book. This book was also inspired by colleagues and the manuscript improved substantially because of their active involvement in its development. While one lacks the space to list all the individuals whose advice was extremely useful, I would especially like to thank the following, whose feedback on related projects in conferences helped me to develop the ideas presented in this book: Helen Milner, Daniel Kono, Eric Reinhardt, Douglas Nelson, Jeffrey Frieden, David Leblang, Quan Li, and Dale Smith. Their contributions and feedback were invaluable. They pointed out what was valuable and not, shared new ideas, and offered feedback on related papers and projects, which

viii / Acknowledgments

gave me the courage to make major changes to the content, organization, and tone of the book. Some of the insights presented in this project are also partially drawn from earlier articles I published in the Annual Review of Political Science (with Helen Milner) and the  Journal of Politics (with Quan Li and Dale Smith). This book has also greatly benefited from my work on other projects in International Economics and International Political Economy that were by colleagues at other institutions, including David Singer, Alexandra Guisinger, William Bernhard, and Mark Hallerberg. In my daily work on this book, I have been blessed with an extremely supportive family and a great group of colleagues at Penn State University. I am most grateful to my wife, Vineeta Yadav, for her unwavering support and am immensely thankful for company from my (now) 4-year-old son, Reyhan. At some point in the future, Reyhan will probably understand why I had to spend long hours working on this book rather than playing with him during the weekends. My extended family including my sister and in-laws also supported my work on this book in numerous ways. Finally, I dedicate this book to my parents, Usha and Bishwanath Mukherjee, who consistently made many sacrifices throughout their lives to ensure that I received the best possible education.

One

Democracy and Trade Policy in Developing Countries

Lech Wałęsa made a remarkable speech in Poland’s newly democratized par­ liament (the Sejm) a few months after he won the first presidential election in December 1990. He noted in the speech that the “birth of democracy” in Poland would “pave the way for economic freedom” as well as significant re­ forms in trade and fiscal policies.1 But the speech was not merely about how democratization promotes economic reforms. In addition, he emphasized that Poland’s entry into the global economy would “strengthen democratic norms” and “discourage electoral malpractice” in the country.2 Wałęsa’s per­ spective was not new. During the Paris Peace Conference of 1919, President Woodrow Wilson often championed the idea that international commerce and democracy are closely intertwined.3 And in more recent years, President Bill Clinton explicitly stated in Between Hope and History, “just as democracy helps make the world safe for commerce, commerce helps make the world safe for democracy. It’s a two-­way street” (1996: 36). Thus there is little doubt that political leaders often proclaim that democracy promotes economic (e.g., trade) globalization and that such globalization helps consolidate de­ mocracy by dissuading newly democratic leaders from rigging posttransition elections. Yet the idea that the transition to democracy facilitates trade reforms and that trade openness facilitates the consolidation of democratic practices such as elections resonates more deeply among scholars who study trade policy in developing countries. Students of international political economy (IPE) have in fact consistently predicted that democratization explains the “rush to free trade” across the developing world from the mid-­1980s on­ ward (Guisinger 2001; Frye and Mansfield 2003, 2004; Milner and Kubota 2005; Eichengreen and Leblang 2008; Tavares 2008; Milner and Mukherjee 2009). This claim is simple but powerful. After all, there has been a dramatic

2 / Chapter One

Trade Flows (% of World GDP)

50

40

30

20

10

0

Developing Countries Developed Countries

1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

Figure 1.1  Trade flows in developed and developing countries

growth in the annual volume of trade flows across developing countries in the last two decades (see figure 1.1) because of sweeping trade reforms in developing economies. And the “rush to free trade” that engendered this surge in trade flows has occurred concurrently or immediately after the “third wave of democratization” (Huntington 1991). Some pundits have fur­­ ther noted that growing trade openness in the developing world tends to credibly constrain incumbents in newly democratized regimes “from turn­ ing the clock back on democracy” via electoral fraud.4 Thus numerous scholars and pundits subscribe to the “optimistic” view of a positive link between democracy and trade openness—­a view that is also held by political leaders. Such optimism is, however, in stark contrast to another widely held but “pessimistic” perspective about the prospects of democracy and economic reform. This pessimistic viewpoint paints a more cynical picture of democracy and economic (including trade) reforms as be­ 1 ing inherently incompatible (Przeworski 1991; Nelson 1993; Armijo et al. 1994; Haggard and Kaufman 1995). It also argues that greater exposure to economic globalization will jeopardize nascent democratic institutions in new democracies, thus creating more opportunity for electoral malpractices. As famously stated by Przeworski in Democracy and the Market, “[Market] reforms . . . are socially costly and politically risky . . . they hurt large so­ cial groups and evoke opposition from important political forces. And if that happens, democracy may be undermined or reforms abandoned, or both” (1991: 136). The fact that scholars hold conflicting views about the association between democratization and trade reforms is not surprising,

Democracy and Trade Policy in Developing Countries   /  3

as disagreements are part and parcel of academia. But the sharp divergence between the optimistic and the pessimistic perspectives delineated above leads to the following conundrum: Which view is correct? The answer to this question is not straightforward. Historical evidence from developing countries as diverse as Argentina, Brazil, Ghana, and the Philippines reveals that the transition to democ­ racy indeed led to the “vigorous pursuit of market-­based economic poli­ cies” (Dominguez 1998: 70), including trade liberalization, in these states. Furthermore, in contrast to the bleak prediction put forth by the pessimists, economic reforms economic reforms discouraged political elites in the aforementioned new democracies from manipulating elections in order to preserve democracy.5 However, the positive association between democra­­ tization and trade that occurred in Argentina, Brazil, Ghana, and the Phi­­l­ ippines is neither common nor ubiquitous. For instance, democratization did not lead to trade reforms in Bolivia, Haiti, Nepal, Nigeria, Peru, Pakistan, and Turkey. The blatant rigging of election results also occurred in some of these states.6 The variation suggested in the preceding examples is hardly unique. A careful examination of the relevant data shows that the relationship among democratization, trade policies, and electoral fraud in new democracies is more nuanced than suggested in the “optimistic” and “pessimistic” interpre­ tations of democracy and trade reforms. To see this in detail, consider fig­­ ure 1.2. This figure plots the annual average of the import duty coverage ratio (a common measure of tariff barriers) prior to and during the first five years after a democratic transition across a comprehensive list of fifty-­six developing countries that have experienced a transition to democracy in the previous three decades.7 These fifty-­six countries and the year in which they made a transition to democracy (as identified in the Polity V and the Przeworski et al. [2000] data sets) are listed in table 1.1. Figure 1.2 shows that although import duties decreased in the first five posttransition years for 58% of the fifty-­six “new democracies,” they in­ creased or retained the status quo in the remaining 42% of the developing countries that also successfully made a transition to democracy. I find a sim­ ilar pattern for nontariff barriers (NTBs) as well (see figure 1.3), since NTBs decreased during the five posttransition years for some new democratic regimes but not others. The data from the fifty-­six new democratic states reveal another intriguing phenomenon (which, as I show later, is related to trade politics): namely, incumbents across a certain share of these new de­ mocracies (almost 30%) have engaged in electoral fraud by rigging the first postdemocratic transition election, while governments in the remaining

4 / Chapter One Table 1.1.  Democratic transitions in the developing world Country

Transition year

Country

Transition year

Albania Argentina Bangladesh Benin Bolivia Brazil Bulgaria Cape Verde Central African Republic Chile Comoros Croatia Czech Republic Dominican Republic Ecuador El Salvador Estonia Ethiopia Ghana Guatemala Guyana Haiti Honduras Hungary Indonesia Korea, Republic of Latvia Lesotho

1992 1983 1991 1991 1982 1985 1990 1991 1993 1990 1990 2000 1993 1978 1979 1994 1991 1995 1996 1996 1992 1994 1982 1990 1999 1988 1991 1993

Lithuania Madagascar Malawi Mali Mexico Mongolia Mozambique Nepal Nicaragua Nigeria Pakistan Panama Peru Philippines Poland Slovak Republic Slovenia South Africa Thailand Uruguay Paraguay Russia Senegal Suriname Tanzania Turkey Ukraine Zambia

1991 1993 1994 1992 1997 1992 1994 1991 1990 1999 1988 1994 1980 1987 1990 1993 1992 1994 1992 1985 1993 1993 2000 1991 1995 1983 1991 1991

Note: The sources used to identify the year of democratic transition in these countries are listed in this chapter and in chapter 4.

new democracies have respected the electoral results from the first post­ transition election. Thus the aforementioned examples, figures 1.2 and 1.3, and the data neither fully confirm the optimistic perspective nor fully confirm the pessi­ mistic viewpoint about democratic transitions, trade reforms, and electoral fraud in developing states. Since scholars hold sharply divergent views about democratic transitions and economic (including trade) reforms—­that is, they believe that democratization has a positive or negative effect on trade liberalization—­it is perhaps not surprising that relatively few studies have focused on variation in trade barriers across newly democratized states. The political consequences of this variation have also been largely overlooked in the literature on democracy and trade. This omission is unfortunate, as

Democracy and Trade Policy in Developing Countries   /  5 35

Decline (57%) Increase (43%)

30

Democratic Transition at t

Import duties %

25 20 15 10

Pre-transition years

Initial post-democratic transition years

5 0 t-6

t-4

t-2

t

t+1 Time

t+2

t+3

t+4

t+5

t+6

t+5

t+6

Figure 1.2 New democratic regimes and import duties

35

Decline (52%) Increase (48%)

30

Democratic Transition at t

Core NTBs %

25 20 15 10 5

Years prior to Democratic transition

Initial post-democratic transition years

0 t-6

t-4

t-2

t

t+1 Time

t+2

t+3

t+4

Figure 1.3 New democratic regimes and nontariff  barriers 1

explaining why trade barriers vary across newly democratized states and an­­ alyzing the political impact of this variation could substantially enrich our knowledge about the political economy of trade policies in the develop­ ing world. Thus the first research objective of this book is to address the following questions that emerge from the examples and figures discussed above: Why do trade barriers decrease in the immediate years following a democratic transition in some newly democratic nations across the devel­ oping world but not others? What conditions influence incumbents in new democratic regimes in the developing world to adopt trade reforms? When

1

6 / Chapter One

are incumbents in new democracies more likely to respect the results from the first postdemocratic transition election rather than engage in electoral fraud? I develop a systematic theoretical framework in the following chapters that provides answers to these questions. These answers are also rigorously evaluated in the book. Before I summarize the theoretical framework and the empirical research design, it is important to note here that a full un­ derstanding of the link between democracy and trade policy in developing countries cannot be achieved by merely analyzing trade politics during the immediate postdemocratic transition period in “new” democracies. One rea­­ son for this is that newly democratized states do not remain nascent democ­ racies forever. Instead, as indicated in the first column of table 1.2, over 70% of these new democracies have evolved into full-­fledged democracies or, in other words, into “newly consolidated” democracies.8 Apart from these newly consolidated democracies, we also know for a fact that many established democracies exist in the developing world, including Costa Rica and India. These established democracies are listed under the sec­­ ond column heading of table 1.2. Since many of these newly consolidated and established democracies actively participate in the global trading system,9 it is natural to ask what the observable pattern of trade barriers has been across these democracies from around 1978 to 2008. Let us briefly examine the map in figure 1.4 to answer this question. This figure classifies the average import duty coverage ratio (in percentage terms) into four categories, ranging from low to high protection, for all develop­ ing states that are observed as democracies (as classified by Przeworski et al. [2000]) in the 1972 to 2008 time period.10 The map shows that average im­­ port duties, a key measure of trade protection, varies significantly across newly consolidated and established developing country democracies. I also find that the variation illustrated in the map holds for NTBs for this set of democracies. What accounts for variation in trade restrictions across “newly consolidated” and established developing country democracies? When do governments in newly consolidated and established developing democra­ cies reduce trade barriers? Scholars of international trade examine trade policy outcomes by often employing either “demand-­side” accounts of trade, which explore the trade policy preferences and strategic behavior of societal groups,11 or “supplyside” theories,12 which analyze how domestic political institutions affect trade barriers. A handful of studies in IPE also combine these two approaches to explain the choice of trade barriers.13 There is little ambiguity in the fact that existing research provides deep insights with respect to explaining the po­ litical economy of trade protection. Yet extant studies are typically designed

Table 1.2.  List of developing democracies Newly consolidated democracies

Established democracies

Country

Years

Country

Years

Albania Armenia Bangladesh Benin Botswana Bulgaria Central African Republic Chile Comoros

Argentina Bahamas Barbados Belize Bolivia Brazil Colombia Costa Rica Cyprus

1983–­2008 1972–­2008 1972–­2008 1972–­2008 1982–­2008 1985–­2008 1972–­2008 1972–­2008 1983–­2008

Congo Croatia Czech Republic Estonia Fiji Georgia Ghana Guinea Bissau Guyana Haiti Hungary Indonesia

1992–­2008 1990–­2008 1991–­2008 1991–­2008 2003–­2008 1990–­2008 1993–­2008 1990–­2008 1990–­1994, 2003–­2008 1992–­1996 1990, 1999–­2008 1990–­2008 1990–­2008 2003–­2008 2003–­2008 1996–­2008 2000–­2008 1992–­2008 1994–­2008 1990–­2008 1999–­2008

Dominica Dominican Republic Ecuador El Salvador Grenada Honduras India Jamaica Korea, Republic of Malta Mauritius Pakistan

Kenya Latvia Lebanon Lesotho Lithuania Madagascar Malawi Mali

2002–­2008 1990–­2008 2003–­2008 2003–­2008 1990–­2008 1993–­2008 1994–­2008 1992–­2008

Panama Papua New Guinea Peru Sri Lanka Suriname Thailand Trinidad and Tobago Turkey

Mexico Moldova Mongolia Namibia Nepal

2000–­2008 1990, 1996–­2008 1992–­2008 1990–­2008 1991–­2001, 2005–­2008 1990–­2008 1993–­2008 1999–­2008 2003–­2008 1990–­2008 1990–­2008 2000–­2008 1990–­2008 1990–­2008 1994–­2008 1991–­2008 1991–­2008

Uruguay Vanuatu Venezuela

1982–­2008 1978–­2008 1979–­2008 1984–­2008 1972–­2008 1982–­2008 1972–­2008 1972–­2008 1988–­2008 1972–­2008 1972–­2008 1972–­1976, 1988–­1998 1989–­2008 1972–­2008 1980–­2008 1972–­2008 1972–­2008 1975, 1983–­2008 1972–­2008 1973–­1979, 1983–­2008 1972, 1985–­2008 1972–­2008 1972–­2001

Nicaragua Niger Nigeria Paraguay Poland Romania Senegal Slovak Republic Slovenia South Africa Ukraine Zambia

8 / Chapter One

3-11%

18-25%

11-18 %

> 25 %

Non-democracies & advanced OECD countries

Figure 1.4  Import duty coverage ratio in developing democracies

toward accounting for cross-­sectional rather than temporal variation in trade policies in developed or developing countries. For instance, demand-­side theories invest substantial effort in explaining how cross-­sectional varia­ tion in factor endowments leads to distinct trade policies across countries.14 Numerous supply-­side studies explore how different domestic institutions engender variation in trade protection across countries. While useful, exist­ 1 ing approaches are not fully equipped to truly examine the temporal dy­ namics of trade politics in the developing world. Hence I develop two main theoretical stories to address this book’s two main research objectives and overcome the limitations mentioned in the preceding paragraph. The first theoretical story focuses on trade politics in new democracies and the impact that it has on the prospects of postelec­ tion fraud in these states. The second theoretical story develops proposi­ tions to explain variation in trade barriers across newly consolidated and established developing country democracies. I avoid choosing between the “demand-­side” approach that examines the trade policy preferences of soci­ etal groups and “supply-­side” theories that analyze the impact of domestic institutions on trade protection. Rather, each of the two theoretical stories

Democracy and Trade Policy in Developing Countries   /  9

in this manuscript is developed from a game-­theoretic model that reconciles these two competing approaches to answer the book’s research puzzles. The first theoretical story is drawn from a game-­theoretic model of trade politics in new democratic regimes that addresses the set of questions men­ tioned earlier about variation in trade policies across newly democratized states. This model explores how strategic interaction among political par­ ties and two key productive groups in society—­labor and capital (industry owners)—­affects trade policy in the institutional context of a new demo­ cratic regime. The model examines how political dynamics in the imme­ diate postdemocratic transition years (the “supply side”) interact with the factor mobility of  labor and capital—­specifically, the degree of interindustry occupational mobility of labor and asset specificity of industries (the “de­ mand side”)—­to explain the level of trade protection across new democra­ cies in the developing world. A key contribution of the model is that it develops a unified theoretical framework that explores how electoral competition in new democracies and the leverage of interest groups that provide financial contributions affect trade policies in developing countries. By doing so, the theory provides a first step toward explaining the puzzle of variance in trade barriers across new democracies in the developing world. It also helps develop a general theoretical framework to explain trade politics in new democratic regimes. Furthermore, this theoretical story explains how (i) politics in new demo­ cratic regimes affects the amount of campaign contributions provided by industries that favor trade protection and (ii) trade reforms in new democra­ cies influence the likelihood of electoral fraud. The second theoretical story is derived from an additional game-­theoretic model that accounts for variation in trade barriers across newly consoli­ dated and established developing country democracies. This model is mo­ tivated by a vast literature positing that political particularism has a critical influence on fiscal and foreign economic policies in developing country democracies.15 As suggested in this literature, “particularism” captures the extent to which political candidates (including party leaders) will seek to cultivate their personal vote during and after elections (Carey and Shugart 1995; Hix 2004; Hicken 2006). The term particularism is theoretically con­ ceptualized as a continuum that ranges from a low to a high level (Wallack et al. 2003; Hix 2004; Johnson and Wallack 2005). Democracies that exhibit high levels of particularism are classified as particularistic (i.e., candidate-­ centered) systems. In particularistic democracies, political candidates have incentives to cultivate their personal vote and compete with candidates from

10 / Chapter One

their own party and other opposition parties (Carey and Shugart 1995; Mainwaring and Shugart 1997; Hix 2004). On the other side of the con­ tinuum, democracies that exhibit low levels of particularism are labeled as party-­centered systems. In contrast to particularistic democracies, party leaders and members do not cultivate their personal vote in party-­centered systems (Hix 2004; Hicken 2006). Instead, electoral politics in these systems strictly occur along the lines of interparty competition. Building on this conceptualization of particularism, the second model generates some theoretical arguments that focus on the link between politi­ cal (i.e., electoral) particularism and trade politics. Yet this model departs from existing accounts of particularism and trade politics in that it analyzes how particularism influences the responsiveness of politicians to the de­ mands of labor and capital when the trade policy preferences of these two groups is determined by their relative factor mobility. The model in this case is largely similar to the previously described model with one exception—­it explicitly formalizes the extent to which politicians have incentives to cultivate a personal vote. It then examines how such in­ centives to cultivate a personal vote determine the responsiveness of politi­ cians to the demands of labor and capital. The model does so by examining how party leaders in developing democracies that exhibit high levels of par­ ticularism design trade policies when the trade policy preference of work­ ers is given by the level of their interindustry occupational mobility, and the trade policy preference of industry owners (capital) is determined by the level of asset specificity of the industries that they own. The details of the second theoretical story that emerges from this model are summarized be­ low. It is worthwhile to note here, however, that this latter theoretical story allows one to identify the precise conditions that explain when and how particularism affects trade barriers and financial contributions by capital in democratic developing countries. The testable hypotheses and the main cau­­sal claims derived from each of the two game-­theoretic models are empiri­­ cally evaluated via a multimethodological approach. This approach includes statistical tests employing both large-­n pooled data and within-­country time-­series data for trade barriers. It also includes the historical analysis and detailed comparison of three case studies. Finally, statistical tests are also conducted on firm-­level survey-­response data sets. The rest of this chapter is divided into five sections. The next section briefly explains why studying the politics of trade protection in develop­ ing countries is important. “The Politics of  Trade in Current Research” sec­ tion briefly reviews existing theoretical explanations for trade policy that are usually classified into “demand-­side” and “supply-­side” explanations. The

Democracy and Trade Policy in Developing Countries   /  11

section that summarizes the book’s theory and hypotheses describes the theoretical arguments that both indicate when newly democratic nations reduce trade barriers and address the link between electoral particularism and trade policies in developing democracies. The empirical research design section discusses the multimethodological approach employed in this book to empirically explore the main theoretical claims. The chapter ends with an outline of this book.

Why Analyzing Trade Politics in Developing Countries Is Important Exploring the link between democracy and trade politics in developing states has substantive implications that speak directly to the relevant academic literature. It also has critical ramifications that go beyond the academic lit­ erature. From an academic perspective, first note that the two predominant schools of thought in the study of trade olitics—­demand-­and supply­side theories—­largely focus on examining trade policy preferences and out­­ comes in either advanced industrial democracies or a “mixed” sample of de­ v­­eloped and developing countries (e.g., Busch and Reinhardt 1999, 2000; Scheve and Slaughter 2001; McGillivray 2004; Hankla 2005; Mayda and Rodrik 2005). A small number of studies attempt to explain why trade barri­ ers vary across democracies in the developing world. These studies are, how­ ever, restricted to democracies in Latin America or eastern Europe.16 Thus students of IPE have not to date systematically examined variation in trade protection in a global sample of developing states and democracies across the developing world. Unlike these previous studies, this manuscript com­ bines the two main schools of thought in innovative ways to develop a full-­ fledged theory of the political economy of trade policy that applies to a global sample of developing countries. Integrating demand-­and supply-­ side approaches to construct the theory has two main advantages. The first advantage is that it helps one to account for not just cross-­ sectional but also—­as we will see later—­temporal (within-­country) vari­ ation in trade policies. This is crucial, as existing theories of trade politics offer valuable insights to explain cross-­sectional patterns of  trade protection but tend to underestimate temporal changes in trade restrictions. Explaining changes in trade policy over time (as done here) will arguably increase our ability to identify the precise conditions under which certain democratic institutions affect trade barriers. This will lead to less ambiguous and thus more accurate predictions about democracy and trade. The second ad­ vantage of integrating the two competing approaches is that it facilitates

12 / Chapter One

development of a theory that departs from existing polarized debates. In­­ stead of choosing between these two opposing views about democracy and trade, this book provides a coherent theoretical explanation of when, how, and what type of democratic institutions influence trade policy outcomes in developing countries. This permits scholars to comprehensively understand the intricacies and nuances of the link between democracy and trade policy in developing states. That said, the analysis presented in this book is more than just an academic exercise. Rather, two key reasons relevant to the real world of trade policy making and international economic relations justify an in-­depth assessment of the political economy of trade protection in the developing world. First, recall that figure 1.1 shows that the annual volume of trade flows in develop­ ing countries as a proportion of total world output has grown dramatically in the last two decades. Such increasing exposure to global trade flows that re­ sults from trade liberalization is likely to benefit some citizens in developing states, including those in developing democracies. But it will also have ad­ verse distributional consequences. Indeed, greater exposure to foreign com­ petition from increased trade openness tends to be costly for citizens who lack the skills to cope with globalization. Such “costs” can endanger fragile democratic institutions in developing states. The theoretical and empirical results in this book, however, provide some insights on how democratic gov­­ ernments can adjust to the costs of globalization and when democracy is likely to survive in the developing world. This helps us understand how frag­ ile democracies can evolve into more consolidated democracies in the con­ text of economic globalization. Second, the emerging role of prominent developing countries in inter­ national trade has made it more difficult for the developed “north” and the developing “south” to cooperate on multilateral trade issues at the World Trade Organization (W TO). For instance, it was widely believed that the three most prominent developing countries (which includes two democracies)—­ Brazil, China, and India—­were responsible for the breakdown of the Doha round of multilateral trade negotiations in 2001 (Buiter 2003; Bergsten 2008). Similarly, in the Cancun round of multilateral trade negotiations, developing nations such as Indonesia and the Philippines demanded “spe­ cial safeguards” to protect their economies. This is pointed out by Narlikar and Tussie (2004: 951), who state that “at . . . the conference, the coalition came to comprise over 20 members . . . The groups, particularly under the leadership of Indonesia and Philippines, proposed that developing coun­ tries be allowed to self-­designate certain strategic products that would not be subject to tariff reductions or new commitments.”

Democracy and Trade Policy in Developing Countries   /  13

The demand put forth by Indonesia and the Philippines has escalated into a trade conflict and ensured that the Cancun round ended with lit­ tle progress or agreement on trade issues (Buiter 2003; Narlikar and Tussie 2004). The failure of the Doha round of negotiations and subsequent Can­ cun meeting raises these policy-­relevant questions: Which domestic condi­ tions induce developing countries to adopt a more rigid bargaining position when negotiating a reduction in trade barriers with developed nations at the W TO? What prevents developed and developing nations from designing an optimal multilateral trade agreement that can benefit all parties? It is beyond the scope of this book to provide detailed answers to these questions. However, as discussed in the concluding chapter, the book’s over­ all analysis helps us explain when and why developing states are likely to stall multilateral trade negotiations. It also identifies the domestic political conditions in developing democracies that make it harder for these coun­ tries to strike a mutually beneficial trade agreement with other nations. As such, the analysis presented here may serve as a useful guide to policy mak­ ers who are interested in increasing the likelihood of cooperation in the issue-­area of multilateral trade negotiations.

The Politics of  Trade in Current Research The literature on the cross-­national determinants of  trade policy constitutes a key area of research within IPE. This literature can be broadly divided into two schools of thought: “demand-­side” and “supply-­side” explanations for trade policy. Demand-­side theories of trade policy can be further divided into three main categories. The first category focuses on societal demands for trade protection that result from economic downturns, unemployment rates, and exchange rate changes (Nowzad 1978; Magee 1980; Hughes and Waelbroeck 1981; Bergsten and Cline 1983; Bergsten and Williamson 1983; Shapiro and Page 1994). The second category concentrates on linking cross­sectoral variations in trade protection to particular characteristics of indus­ tries, such as sectoral size, import penetration, export dependence, and geo­ graphic concentration of industries (Magee 1980; Anderson and Baldwin 1987; Trefler 1993; Lee and Swagel 1997; Busch and Reinhardt 1999, 2000). More recent models on the effects of industrial characteristics on trade poli­ cies study how lobbying by industries exposed to import competition affect trade protection (Grossman and Helpman 1994; Baldwin and Magee 2000). The third category of demand-­side theories primarily employs either the Heckscher-­Ohlin (hereafter H-­O) model or the Ricardo-­Viner model to ex­ plain the trade policy preference and strategic behavior of different societal

14 / Chapter One

groups in the issue-­area of trade politics (e.g., Rogowski 1989; Scheve and Slaughter 2001; Hiscox 2002; Baker 2005; Mayda and Rodrik 2005; Mans­ field and Mutz 2009). According to some of these studies, the trade policy preferences of individuals, including voters, are determined by their degree of occupational specificity, education, income, their beliefs about whether or not trade openness is beneficial for them, and their sociotropic perceptions of trade’s influence on the national economy. Other prominent demand-­ side studies in the third category instead focus on understanding how  factoral cleavages over trade policy, most notably between capital and labor, affect trade politics (e.g., Rogowski 1989) or how factor mobility shapes political cleavages associated with trade politics (Hiscox 2001, 2002). Existing studies of the empirical validity of demand-­side theories typi­ cally employ cross-­national surveys. These studies find that the trade policy preferences of citizens are influenced by their factor endowments, human capital, or sociotropic concerns (see, for example, Scheve and Slaughter 2001; Baker 2005; Mayda and Rodrik 2005; Hainmueller and Hiscox 2006; Mansfield and Mutz 2009).17 In contrast, a handful of studies empirically evaluate the effect of demand-­side variables on trade policy outcomes in a sample of just developing countries. The few studies that test the impact of demand-­side variables (e.g., factor endowments) on trade barriers in devel­ oping states typically use a cross-­sectional sample of developing countries (Dutt and Mitra 2002; Dhingra 2009). The fact that these studies employ just cross-­sectional samples limits the empirical generalizability of the results that they report. Additionally, concerns that demand-­side theories place too little emphasis on state institutions has encouraged scholars to pay atten­ tion more recently to supply-­side variables in trade policy decisions. Supply-­side explanations primarily consider the relationship between domestic political institutions and trade protection. Some of these supply-­ side studies explore the effect of international agreements or institutions such as the General Agreement on Tariffs and Trade (GATT) and WTO on international trade flows or the settlement of interstate trade disputes (Busch and Reinhardt 2006; Tomz et al. 2007; Davis and Bermeo 2009). The more prominent supply-­side explanations of  trade policy, however, examine the relationship between democracy and trade protection or trade flows. Re­­ search in this issue-­area indicates that democracies are more likely to have stable trade flows (McGillivray and Smith 2004) and lower levels of trade protection (Frye and Mansfield 2004; Milner and Kubota 2005; Eichengreen and Leblang 2008). Yet not all researchers agree with the claim that democ­ racy helps reduce trade barriers. For example, Kono (2006) finds that de­ mocracies have a negative effect on tariffs but a positive impact on N TBs.

Democracy and Trade Policy in Developing Countries   /  15

Other scholars suggest that democratic governments are relatively more pro­ tectionist since they are more susceptible to demands for protection from interest groups (Garrett 2000; Yu 2006). As stated by Garrett (2000: 973), “On the one hand, democracy makes leaders more accountable to their citi­ zens, promoting trade liberalization to the extent that this is good for soci­ ety as a whole. On the other hand, democracy also empowers distributional coalitions with intense interests, making higher levels of protectionism more likely.” The possibility that democracy may not always promote free trade (as suggested by Garrett 2000) has motivated scholars to increasingly focus on variation in trade barriers across democracies. This has led them to analyze how different democratic political institutions account for variation in trade barriers across democracies. These institutions include electoral systems, the number of veto players, the degree of electoral particularism across democ­ racies, political partisanship, and the number of effective parties in democ­ racies (see, for example, Nielsen 2003; Hankla 2006; Henisz and Mansfield 2006; Ehrlich 2007; Kono 2009). There is little doubt that demand-­side and supply-­side theories of trade politics have helped researchers deeply understand the political economy of trade policy. But there still exists considerable debate about the empirical va­ lidity of  these theories for particularly developing countries, which are the fo­ cus of  this book. Studies that use a time-­series-­cross-­sectional (hereafter  TSCS) sample of developing countries find, for instance, that neither industry-­based characteristics (such as import dependence and intraindustry trade) nor mac­ roeconomic indicators (such as exchange rate changes or factor endowments) directly influence trade barriers in developing states (Bandyopadhyay and Roy 2007; Mukherjee et al. 2009; Milner and Mukher­jee 2010). Supply-­side explanations of trade policy also find mixed statistical sup­ port in samples that only include developing countries. For example, in a sample of just eighteen developing countries from Latin America, Nielson (2003) finds that particularism and the share of legislative seats held by left-­leaning parties statistically affects trade restrictions in these nations. But other supply-­side factors such as dummy for presidential systems and aver­ age district magnitude are statistically insignificant in Nielson’s (2003) em­ pirical models. In a larger TSCS sample of over ninety developing countries, researchers find that the independent effect of  electoral systems, partisanship, district magnitude, and the number of parties on tariffs is insignificant and not robust (Guisinger 2001; Milner and Mukherjee 2010). The mixed support for the independent effect of demand-­side and supply-­side variables on trade restrictions in developing states should not

16 / Chapter One

be taken to mean that these variables cannot account for trade policy in the developing world. Instead, as acknowledged by various scholars in IPE,18 one needs to reconcile demand-­and supply-­side explanations in a unified theoretical framework to account for the political economy of trade policy. Doing so may enhance the explanatory power of these approaches when ex­­ plaining heterogeneity in trade protection across developing countries. In particular, as described below, I draw on the relative analytic strengths of both camps in an attempt to further bridge the divide that separates them as well answer the book’s central research questions. I do so because of two main reasons. First, if scholars focus on just the impact of certain domestic political institutions on trade policy, then they may find it difficult to account for the observed variation in trade barriers across developing countries and even across developing democracies. For instance, I mentioned earlier that research on the link between democracy and trade policy often suggests that the emergence of democracy reduces trade barriers in developing states (Frye and Mansfield 2004; Milner and Kubota 2005). While logically con­ sistent, this research cannot fully explain why we observe in some (but not all) new democracies an increase in the level of tariffs and N T Bs during the immediate posttransition years. As shown in the following chapters, how­ ever, one can provide a more compelling answer to this specific question by integrating demand-­and supply-­side perspectives on the political economy of trade protection. Second, since 2000, supply-­side studies of international trade that ana­ lyze the impact of specific democratic political institutions on trade barriers in the developing world have focused on a small set of developing country democracies in specific regions such as Latin America or eastern Europe (e.g., Frye and Mansfield 2004; Brooks and Kurtz 2007). Demand-­side explana­ tions for trade policy preferences in developing states also tend to focus on one or a few countries in Asia or Latin America (Baker 2005; Hicks et al. 2014). Thus existing research on the relationship among trade policy pref­ erences, democratic institutions, and trade policy outcomes across a global sample of developing states is still in its infancy and has not been system­ atically developed by scholars. A key goal of this book is therefore to ana­ lyze how the trade policy preferences of key societal actors (the “demand side”) combines with specific democratic political institutions (the “sup­ ply side”) to influence trade politics in a much larger set of developing democracies that spans across all regions of the globe. The game-­theoretic models presented in this book are thus designed to integrate demand-­side claims with supply-­side approaches in order to provide both detailed causal

Democracy and Trade Policy in Developing Countries   /  17

mechanisms and generalizable predictions about trade politics in different institutional contexts across the developing world. But this book is not just about combining two main competing theo­ retical perspectives to address a series of substantively important questions about trade politics in developing countries; the theory developed here also provides clear microfoundations about how the relative factor mobility of labor affects the strategic behavior of capital (industry owners) and the cam­­ paign contributions that they provide to political parties in order to influence trade policy. The theory also highlights in sufficient detail how specific de­­m­ ocratic institutions affect the strategic calculations of capital and their opti­ mal choice of campaign contributions, given the relative factor mobility of labor. Hence the theoretical analysis about the role of “capital” in develop­ ing economies presented in the following chapters departs from endoge­ nous models of trade policy that tend to downplay or (sometimes) ignore how specific democratic institutions and the factor mobility of  labor affects the strategies of industry owners in the issue-­area of trade policy.19

The Theory and the Hypotheses Trade Politics in Newly Democratized Regimes This book constructs a game-­theoretic model of trade politics in new demo­ cratic regimes to explain when incumbents in these developing economies will adopt trade reforms. The model analyzes how strategic interaction dur­ ing elections in the immediate postdemocratic transition period among the ruling party, the opposition party, and two key societal actors—­labor and capital (industry owners)—­affects the following outcomes: tariffs, N T Bs, as well as the financial payments (campaign contributions and illegal dona­ tions) provided by the owners.20 The model builds on yet substantially ex­­ pands two approaches that economists rely on to study the political economy of trade policy. The first approach utilizes endogenous models of trade pol­ icy. The second approach utilizes Meyer’s (1989) well-­known median voter model of trade politics. These two approaches provide the game-­theoretic foundations in my model that permit me to explore how political parties rationally design trade policies in response to the (i) optimal trade policy response of the median voter (a worker)21 and (ii) campaign contributions offered by the owners of asset-­specific industries. The trade policy preference of labor—­which includes the median voter— ­is given by the extent to which they are occupationally mobile across indus­ tries in the economy. I assume and formalize from extant research in the

18 / Chapter One

model that workers who are more occupationally mobile tend to favor trade reforms, as they benefit from trade openness.22 The opposite is also true: less occupationally mobile workers tend to oppose trade reforms because they are adversely affected by trade openness. This assumption is drawn from recent studies that suggest the interindustry occupational mobility of work­ ers critically influences their attitudes toward trade openness in developing countries.23 The rationale for adopting this assumption is discussed in more detail in chapter 2. With respect to capital (the industry owners), I largely focus on the trade policy preference and strategic behavior of industry own­ ers of asset-­specific industries that often operate in the import-­competing sector in developing economies. The rationale for doing so is twofold. First, industry owners in developing states often own and manage import-­competing industries that are highly asset specific (Krueger 1993; Bhagwati and Sreenivasan 2002; Harrison and Rodriguez-­Clare 2009; Cadot et al. 2011). Second, asset-­specific industries in the import-­competing sector tend to be concentrated in developing countries (Rodrik 1988; Lawrence 2000; Harrison and Rodriguez-­Clare 2009; Cadot et al. 2011). Owners of these industries in the import-­competing sector are therefore well organized compared to the more fragmented (and disorganized) export-­competing in­­ dustry owners (Rodrik 1988; Zahariadis 2001; Bhagwati and Sreenivasan 2002).24 Hence lobbying for trade protection in developing countries (which includes developing democracies) is typically carried out by industry owners of asset-­specific industries. It is therefore empirically plausible to focus on the preference and behavior of “capital” from asset-­specific industries in the model. In addition to the basic theoretical framework described above, I explic­ itly formalize in the model the following three key political features of new democratic regimes in developing countries: the expansion of the selector­ ate, the introduction of electoral competition after the transition to democ­ racy, and the possibility that political elites in new democracies may attempt to rig the election results ex ante if they believe the ex post electoral outcome may threaten their survival in office (Mansfield and Snyder 1995a, 2005; Schedler 2002; Bueno De Mesquita et al. 2003; Kalandadze 2009). How do the political parties rationally respond to the trade policy preferences of work­ ers (labor) and the industry owners (capital) as well as the monetary contri­ butions provided by the owners during elections given these three political features? To start, the model suggests that the expansion of the selectorate and the introduction of elections in a new democracy unleash intense political com­ petition among the political parties. Intense political contestation among

Democracy and Trade Policy in Developing Countries   /  19

the parties in a new democracy generates uncertainty within the ruling and opposition parties about their prospects for winning office. Such uncertainty induces the opposition and the ruling parties to situate the equilibrium level of tariffs and N T Bs at the median voter’s optimal trade policy prefer­ ence to maximize their likelihood of winning the election. Because the par­ ties situate the tariffs and N T Bs in equilibrium at the median voter’s optimal preference, the equilibrium trade policy is effectively biased toward the pref­ erence of labor, since the median voter is a worker in the labor force of the developing economy. This in turn has two consequences in the model. The first consequence is that it ensures that the ruling party’s decision to decrease or increase trade barriers depends on not only the level of the interindustry occupational mobility of labor—­the electoral majority—­but also the median voter (a worker) in the polity. If the workers’ and hence the median voter’s level of occupational mobility is sufficiently high, then they support trade reforms. This is because occupationally mobile workers in de­ veloping countries adjust to and benefit from trade liberalization.25 Parties in the new democratic regime, including the ruling party, rationally respond to such support for trade reforms by decreasing tariffs and N T Bs in equilib­ rium. Hence the first testable hypothesis that emerges from the model is that governments in new democracies in the developing world will reduce tariffs and N T Bs during the initial posttransition years when the level of interin­ dustry labor mobility is sufficiently high. The second consequence of the parties’ prolabor trade policy bias in a new democratic regime is that the equilibrium level of tariffs and NTBs diverges away from the optimal trade policy preference the owners of asset-­specific in­ dustries (labeled as “capital” in the model) that favor trade protection. One reason for this is because the owners cannot credibly threaten to depose the incumbent in a new democracy. The politically empowered workers in a new democracy can in contrast credibly endanger the ruling party’s political sur­ vival, as they constitute the electoral majority. This induces the incumbent to rationally shift the equilibrium level of trade barriers away from the trade policy preference of asset-­specific industries and—­as emphasized earlier—­ toward the preference of labor. Shifting the equilibrium trade policy away from the preference of the owners of asset-­specific industries also implies that the ruling and the op­ position party cannot credibly commit themselves ex ante to pursue the trade policy interests of these owners. As a result, these owners are more likely to believe that the monetary payments (contributions and illegal donations) that they offer will not influence the political parties to design and imple­ ment trade policies that suit their (the owners’) interests. This incentivizes

20 / Chapter One

the owners’ to reduce their payments to the parties. Hence the second hy­ pothesis posits that the legal campaign contributions provided by industry owners to influence trade policy in new democracies will decline when la­ bor mobility is sufficiently high. Adopting trade reforms also has implications for the likelihood of elec­ toral fraud in newly democratized regimes. In this regard, recall that I for­ malize the possibility that the ruling party may rig the election results ex ante if it anticipates losing the election. The model shows that liberalization of trade policies by incumbents in new democratic regimes will generate “electoral dividends” for these leaders and increase their probability of  legit­ imately winning the election. Incumbents that adopt trade reforms in new democracies will therefore anticipate less political threats to their survival in office. This substantially dampens their incentives to rig the election. Thus the third hypothesis predicts that the adoption of trade liberalization in new democracies reduces the likelihood of electoral fraud. Particularism, Factor Mobility, and Trade Policy This book constructs an additional game-­theoretic model to explore when incumbents in newly consolidated and established democracies across the developing world reduce trade barriers. The model investigates the relation­ ships among political particularism, trade protection, and campaign contri­ butions in newly consolidated and established developing democracies.26 It does so by studying how particularism affects strategic interaction during elections between ruling and opposition party leaders and the labor and cap­ ital societal groups. The basic structure of this model is similar to the pre­­ vious model on trade politics in new democracies—­that is, the trade policy preference of labor (including the median voter) is given by the extent of the workers’ interindustry occupational mobility, while the policy preference of capital is determined by the level of asset specificity of their respective in­­ dustries. Labor exercises its leverage over trade policy via voting, while the owners of  industries who prefer more trade protection lobby party leaders to obtain higher trade barriers. Unlike the previous model on trade politics in new democracies, this latter game-­theoretic model formally incorporates the following two key features of political particularism in developing democracies: the extent to which politicians cultivate their personal vote and extract rent for per­ sonal benefits (for these features, see for example, Carey and Shugart 1995, 1998; Wallack et al. 2003).27 The model provides three main results. First, it

Democracy and Trade Policy in Developing Countries   /  21

shows that in highly particularistic (i.e., candidate-­centered) democracies, leaders from the ruling and opposition parties cannot rely on their rank-­ and-­file party members to mobilize electoral support for the trade policy position that they adopt owing to personal vote cultivation. Hence the lead­ ers “hedge” their electoral strategy in the specific context of trade policy by shifting the equilibrium level of tariffs and N T Bs to the median voter’s optimal trade policy preference. This leads to a prolabor bias in trade policy, as the median voter is a worker. Shifting the equilibrium trade policy to the median voter’s optimal choice maximizes the party leaders’ likelihood of winning the election even though they cannot rely on rank-­and-­file party members to garner political support. This also leads to a situation where the ruling party leaders’ decision to decrease or increase trade barriers is critically determined by the level of the median voter’s (who is drawn from the labor force) interindustry occupa­ tional mobility. If labor’s occupational mobility is sufficiently high, then they will be more likely to support trade liberalization (as explained ear­ lier). Party leaders in a particularistic democracy will strategically respond to such support for trade liberalization by decreasing tariffs and NTBs in equilibrium. Hence the main hypothesis that emerges from the story in this case is that incumbents in developing democracies that exhibit a high level of particularism will reduce tariffs and N T Bs when the degree of labor mo­ bility is sufficiently high. Second, given that party leaders in highly particularistic developing de­ mocracies design trade policies that appeal to labor, they effectively shift the equilibrium level of trade protection away from the trade policy prefer­ ence of industry owners (i.e., capital). The owners of asset-­specific industries in these democracies will observe the prolabor bias in trade policy. They will therefore believe that the financial contributions that they provide to politicians will not politically induce the latter to adopt trade policies that suit their (the owners’) interests. This will encourage them to reduce the contributions that they offer to the party leaders. This leads to the follow­ ing hypothesis: industry owners in particularistic democracies will reduce the contributions offered to political candidates when the degree of labor mobility is sufficiently high. The model’s third theoretical result focuses on trade protection and financial payments offered by industry owners in developing democracies that exhibit low levels of particularism (i.e., party-­centered systems). This set of results first reveals that party leaders in party-­centered systems have incentives to obtain substantial financial resources, as these resources are

22 / Chapter One

invested to “build” their party. This influences the party leaders to solicit financial contributions from capital owners, especially the owners of asset-­ specific industries who provide contributions in exchange for more trade protection. Soliciting contributions thus makes party leaders more receptive to the protectionist demands of the owners of these industries. This drives the owners of asset-­specific industries to offer substantial contributions to politicians in party-­centered developing democracies.

Empirical Research Design I adopt a multimethodological approach to test the hypotheses and the causal claims presented in the book. This approach includes tests conducted on large-­n TSCS data sets, survey-­response data sets, comparative case-­study research, and statistical evaluation of within-­country data on trade barri­ ers and campaign contributions. I briefly discuss this multimethodological strategy for the empirical tests below. The results from the model on democratization and trade reforms as well as the model on electoral particularism and trade policy generate a series of testable hypotheses, summarized earlier. Recall that the main prediction from the first set of the testable hypotheses is that the interactive effect of new democratic regimes and the interindustry occupational mobility of la­ bor on tariffs and N T Bs will be negative in developing economies. These hy­ potheses also posit that the adoption of trade reforms in new democracies curtails the likelihood of electoral fraud. I test these claims in a TSCS data set of 129 developing countries that are observed between 1972 and 2008. This large-­n statistical analysis allows one to test the empirical validity and generalizability of the theoretical predictions about trade barriers in new democracies. Finally, the first set of hypotheses also predict that the amount of le­ gal campaign contributions provided by industry owners to influence trade policy in new democracies declines when labor mobility is sufficiently high. The causal claim underlying this prediction is that industry owners in new democracies will believe that the contributions that they provide to parties will not influence the parties’ trade policy decisions when labor mobility is high. This will induce them to reduce the contributions that they provide to the parties. Testing the prediction posited above is impossible since cross-­ sectional or pooled data on the amount of campaign contributions given to political parties or candidates by industrialists in developing countries is not available. I thus focus on directly evaluating the aforementioned cau­sal claim by using firm-­level survey responses gathered from the World Economic

Democracy and Trade Policy in Developing Countries   /  23

Forum’s Executive Opinion Survey (EOS) database from eighty-­five develop­ ing countries.28 The main prediction in the second set of  testable hypotheses is that higher levels of political particularism reduces trade barriers (tariffs and N T Bs) in “newly consolidated” and established developing democracies when labor mobility is sufficiently high. I statistically test this claim in a comprehensive TSCS data set of democracies in the developing world. The second set of hypotheses also predicts that contributions provided by industry owners to politicians in particularistic democracies declines when labor mobility is reasonably high. As noted earlier, it is impossible to test this prediction owing to the paucity of data on the amount of campaign contributions in developing countries. I therefore use firm-­level survey responses gathered from the World Economic Forum’s EOS database for fifty-­six developing de­ mocracies to directly assess the key causal claim that leads to the prediction stated above. This causal claim posits that industry owners in particularistic democracies reduce the contributions that they offer to politicians, as they believe that their contributions will not influence the trade policy decisions of party leaders when labor mobility is high. The book’s large-­n statistical analysis is complemented by a detailed “small-­n” case-­study analysis. A key advantage of the case-­study analysis is that it helps one to directly analyze and reconstruct the process of political decision making with respect to trade policies implemented by governments in the developing world. This allows researchers to ascertain causal mecha­ nisms more directly than can be done via large-­n statistical analysis (Ragin 1987; King et al. 1994; Collier et al. 2002). A contextual comparison of in­ dividual cases also ensures that the generalizations made from the large-­n analysis are not too sweeping. I thus compare and contrast the following three cases to empirically assess the hypotheses from the formal models and the underlying causal mechanisms that generate these hypotheses: Brazil, India, and South Africa. These three country cases are important because they are a part of the BRICS (Brazil, Russia, India, China, and South Africa) group of countries and are therefore also key players in the broader group of emerging markets in the developing world. The selection of these three cases has both a theo­ retical and a methodological rationale. The total share of trade flows (as a proportion of world output) from these three countries has also grown by a substantially large 17% since the mid-­1990s. Further, these three countries are also critical players in the G-­20 group of nations. In this group, their co­ operation on a multilateral basis for trade and financial policy issues is vital. In addition to these broader reasons, the choice of  these three countries for

24 / Chapter One

the case-­study analysis is based on sound theoretical and methodological reasons. To start with, Brazil, India, and South Africa provide ample variation on the key dependent variables of interest—­tariffs and N T Bs—­and by exten­ sion variation in the degree of “legal” campaign contributions provided by protectionist industry owners to political parties or candidates across these three cases. In fact, I show in the case-­study chapters that the level of trade protection in India was and is relatively higher than in Brazil and South Africa.29 Moreover, Brazil has implemented more comprehensive trade re­ forms compared to India and South Africa.30 Second, there exists variation in the independent variables of interest for these cases. Brazil and South Africa, for instance, transitioned to full-­fledged democracies in the previous two decades,31 while India has been an established democracy since 1947. Brazil is moreover a particularistic (i.e., candidate-­centered) system,32 while India and South Africa are both party-­centered systems.33 Notwithstanding these differences, the three cases have much in common. This allowsone to control for other competing explanations. For instance, the sizes of Brazil’s and India’s economies are largely similar in nominal terms,34 while Brazil and South Africa are both middle-­income countries.35 Successive governments in each of these three countries also followed import­substitution-­industrialization (ISI) policies for at least three decades after World War II and then adopted trade reforms in recent years (Panagariya 2008; Cardoso 2009). The similarities between the three cases and variation in the independent and dependent variables of interest across these three cases provides one with an opportunity to assess the causal mechanisms un­­ derlying the predictions from the formal models. I also use some within-­ country data from each of the three countries mentioned above to test the causal claims presented in this book. Specifically, for each of these three cases, I gathered tariff data at the industry-­year level. I employ this industry-­year tariff data to closely assess my predictions about the effect of interindustry occupational mobility of labor and the protectionist interests of industries on tariffs separately in (i) Brazil (observed recently as a new democracy and then a candidate-­centered sys­ tem), (ii) India (an established democracy and a party-­centered system), and (iii) South Africa (also recently observed as a new democracy and then a party-­centered system). Second, I gathered data on campaign contribu­ tions from Brazil and survey data on perceptions of contributions given to political parties in India and South Africa. These within-­country data are employed to assess the causal claims that I propose about political institu­ tions and campaign contributions in developing states.

Democracy and Trade Policy in Developing Countries   /  25

Book Outline This book is organized as follows. Chapter 2 presents and discusses the re­ sults from a formal model in which I explore how politics in newly democ­ ratized regimes affects trade protection in developing countries. This model examines how strategic interaction among political parties, labor, and capi­ tal in a new democracy affects trade barriers and campaign contributions. The strategic interaction formalized in the model occurs in a context where the relative factor mobility of labor and capital influence their trade policy preferences. The model then explores the effect of trade reforms on the like­ lihood of electoral fraud in new democracies. The model’s main results gen­ erate testable hypotheses that are listed in the chapter. Chapter 3 presents the statistical results from testing the predictions de­ rived from the formal model on trade protection and campaign contribu­ tions in new democracies. This chapter first discusses the large-­n sample employed for the statistical tests, the operationalization of the dependent variables for trade protection (tariffs and N T Bs), and the independent vari­ ables and control variables. Considerable space is devoted in this chapter to explain the procedure employed to identify and code the set of new de­­m­ ocratic regimes in the pooled data set of developing countries. I also explain the steps taken to operationalize the other key independent variable of interest—­that is, the interindustry occupational mobility of labor. This is followed by a presentation of the statistical estimates that evaluate the re­ lationship among new democracies, labor mobility, and trade protection. Chapter 3 also presents results from a survey-­response data set from the World Economic Forum’s EOS database (2004b). This survey-­response da­ tabase is employed to evaluate my theoretical claims about the extent to which legal campaign contributions (offered by industry owners) influence economic (including trade) policies in new democracies as perceived by the industry owners. This chapter ends with a discussion of the results obtained from data that tests the effect of trade reforms in new democracies on the likelihood of electoral fraud. Chapter 4 begins with the premise that the level of electoral particular­ ism is a critical factor that influences the design of trade policies in “newly consolidated” and established developing country democracies. Based on this premise, I construct a formal model in chapter 4 that examines how strategic interaction among party leaders from the ruling and opposition parties and the two groups in society, labor and capital, affects trade barri­ ers. The model particularly explores how personal vote cultivation by politi­ cians in particularistic (i.e., candidate-­centered) democracies interacts with

26 / Chapter One

interindustry labor mobility to influence tariffs and N T Bs. I also study the effects of party-­centered systems (i.e., democracies characterized by low lev­ els of particularism) on tariffs and NTBs. Finally, the model in this chapter also provides theoretical predictions about how the level of particularism influences financial contributions provided by owners to influence trade policy, given the relative factor mobility of labor. Chapter 5 begins with a brief description of the pooled data set of devel­ oping used for testing the theoretical claims about political particularism and trade protection in developing democracies. I describe the operationalization of the key independent variables of interest: the level of political particular­ ism in developing democracies and the interindustry occupational mobility of labor. This chapter then presents the statistical results obtained from test­ ing the predictions about particularism, labor mobility, and trade protec­ tion in developing democracies. I also use the World Economic Forum’s EOS database to assess claims about the extent to which campaign contributions offered by industry owners influence economic (including trade) policies in particularistic democracies as perceived by these owners. Chapters 6 and 7 supplement the large-­n tests in the preceding chapters with in-­depth case studies of the politics of trade protection, trade reforms, campaign contributions, and “illegal” donations. More specifically, in chap­ ter 6, I focus on the politics of trade liberalization and the phenomenon of financial payments (legal and illegal) provided by industry owners in Brazil. Given that Brazil is observed as a new democracy from 1986 to 1991 and as a particularistic (i.e., candidate-­centered) system thereafter, the Brazilian case allows one to evaluate the causal claims about how politics in new de­ mocracies and particularistic systems combine with the degree of  factor mo­ bility of labor to influence trade barriers. I also use some data on campaign contributions from Brazil to assess the theoretical claims about contribu­ tions by industry owners in new democracies and particularistic systems. I examine the cases of India and South Africa in chapter 7. These two cases provide an interesting contrast to Brazil, as these countries are both de­ veloping democracies with party-­centered systems. I thus use these two cases to assess my causal arguments in chapter 3 that explore how trade politics in party-­centered systems influence trade politics and campaign contributions in developing democracies. Moreover, since South Africa experienced a tran­ sition to democracy in 1994, I analyze the South African case to qualitatively test my hypothesis about how politics in a new democracy influences trade barriers and contributions provided by import-­competing industry owners. Chapter 8 is the concluding chapter. This chapter reviews the theory and the findings presented in the book. It discusses the theoretical and empirical

Democracy and Trade Policy in Developing Countries   /  27

contributions that this study provides not just for the IPE literature but also, more generally, for researchers who study the effect of domestic political institutions on economic outcomes. I also describe some prescriptive steps that policy makers may adopt to prevent a backlash against economic glob­ alization in their respective countries. The concluding chapter outlines the implications that this study may have for the prospects of multilateral trade cooperation between developed and developing countries. The book ends with an appendix that contains the mathematical proofs for the model on democratization and trade liberalization and the model on electoral par­ ticularism and trade policy.

Two

Trade Protection and Electoral Malpractice in New Democratic Regimes

In this chapter, I develop a game-­theoretic model of trade politics and trade policy in new democratic regimes in the developing world. The model explores how competition among political parties during the initial postdemocratic transition years affects trade protection (tariffs and nontariff barriers [NTBs]) as well as campaign contributions that are offered to influence trade policy in developing states. It specifically analyzes the decision by the parties during the immediate posttransition period to adopt trade reforms in an economy where the trade policy preferences of the two productive groups in society, labor (i.e., workers) and capital (i.e., owners of industries),1 are determined by their degree of factor mobility.2 As described below, the trade policy preference of workers is influenced by the extent of their occupational mobility between industries and the preference of capital is affected by the degree of asset specificity of the industries they own. Building on this, the model analyzes how the optimal choice of trade barriers by the parties—­when responding to labor’s and capital’s respective preferences—­is influenced by electoral competition in a new democracy and two additional political features of new democratic regimes that are described below. The results also provide predictions about how trade reforms in new democracies affect the probability of electoral fraud. The rest of this chapter is organized as follows. The first section presents the model in detail and is divided into three parts: The first part provides a descriptive background of the players and the economy, as well as the structure of political competition (in a new democracy) in the model. The second part technically presents the following components of the model: the players’ utility functions, the key political features of a new democracy (including the introduction of elections), the sequence of moves, and the solution concept. The third part discusses the model’s equilibrium and

30 / Chapter Two

comparative static results that generate the causal claims and testable hypotheses. Beyond this, the second section extends the model to explore how trade politics in a new democracy in the developing world affect NTBs. This extension of the model also provides some testable hypotheses. The conclusion summarizes the model’s main theoretical predictions and briefly discusses the substantive theoretical implications of the model’s results.

A Model of the Political Economy of  Trade in New Democracies To begin with, the model explores how strategic interaction in a developing economy among capital, labor, the opposition political party, and the ruling party affects the level of tariffs and NTBs as well as campaign contributions in a new democracy. The term labor refers to workers in the economy, while “capital” denotes the owners of industries who own substantial amounts of capital and can use their financial leverage to influence trade policy. With respect to societal actors, I focus on labor and capital in the model because of two reasons. First, incorporating workers is standard practice in existing theoretical models of trade policy in developed and developing democracies (e.g., Mayer 1984; Grossman and Helpman 1994, 2002; Goldberg and Maggie 1999; Dutt and Mitra 2002; Milner and Kubota 2005). Current research on trade politics in developing democracies in fact suggests that workers constitute the electoral majority in developing democracies and that their preference matters for the design of trade policy in these states (Dutt and Mitra 2002; Baker 2003, 2005; Milner and Kubota 2005). It is thus empirically realistic to focus on the preference and political behavior of labor in the theoretical analysis. Second, in addition to workers, research on trade politics in developing states reveals that owners of especially asset-­specific industries in these countries often coalesce into interest groups to influence trade policies by offering contributions to parties (Payne 1992; Eaton 2002; Weyland 2002). For instance, the owners of several industries in Brazil that exhibit a high level of asset specificity have coalesced into the formal organization called the Federacao das Industrias do Estado de São Paulo (FIESP). These owners routinely use FIESP as a platform for providing campaign contributions and lobbying for trade policy that is compatible with their interests (Hudson 1997; Weyland 2002; Ferreira and Facchini 2005). Research on trade politics in other developing democracies such as the Philippines, Indonesia (which made a transition to democracy in 1998), and South Africa (which experienced a transition to democracy in 1995) also indicates that factor-­specific

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  31

industry owners attempt to influence trade policy in these states by providing contributions to politicians (Bond 2000; Eaton 2002; Athukorala 2005). These examples motivate one to incorporate the preference and strategic behavior of capital in the analysis. A central objective of the model is to explore how strategic interaction among the actors identified above and political parties in a new democracy affects the prospects for trade liberalization and campaign contributions during the immediate posttransition period. Strategic interactions between the societal actors and the parties occur in a small, open economy of a newly democratic developing country that produces import and export goods. There are two factors of  production in the economy (labor and capital), and these two factors are employed to produce the imported good. Further, the trade policy preferences of  labor and capital are well defined in the model. What are these preferences? There is an ongoing debate among scholars of trade politics about the determinants of trade policy preferences of individuals in developing countries.3 Some studies of trade politics in developing countries subscribe to the assumption in the Heckscher-­Ohlin model and the Stolper-­Samuelson theorem that the factors of production are perfectly mobile among industries (Dutt and Mitra 2002; Milner and Kubota 2005). For instance, using the logic of the Stolper-­Samuelson theorem, Milner and Kubota (2005: 116) suggest that “liberalizing trade policy in LDCs [least developed countries] results in a gain in income for, and a reduction in the prices of imported goods bought by, those well endowed with the relatively abundant factor—­that is, labor—­in these economies.”4 Workers in developing countries therefore tend to support trade reforms according to this perspective. The alternative perspective suggests that the Ricardo-­Viner approach that assumes complete factor specificity is the most appropriate way to model the income effects of trade and the determinants of trade policy attitudes of individuals in developed and developing nations (Magee 1980; Grossman and Helpman 1994; Rodrik 1995). A more recent line of research moves away from the assumption of perfect factor mobility and complete factor specificity. This research conceptualizes the degree of factor mobility of labor and capital as a continuous variable (Alt et al. 1996; Hiscox 2001a, 2002; Wacziarg and Wallack 2004; Mukherjee et al. 2009).5 The main theoretical rationale underlying this conceptualization is the idea that the degree of mobility (or specificity) of assets to particular types of economic activities is a crucial determinant of the income effects of trade and thus the trade policy preferences of individuals (Hiscox 2001a, 2002; Alt et al. 1996, 1999; Wacziarg and Wallack

32 / Chapter Two

2004; Rickard 2005; Hainmueller and Hiscox 2008). As emphasized by Hainmueller and Hiscox, Depending upon whether assets are assumed to be highly specific to different industries or highly mobile between them, general equilibrium models generate very different predictions about the distributional implications of tariffs, or indeed any policy that alters relative commodity prices. Asset specificity is thus rightly regarded as a critical parameter in the political-­economy models of a wide range of trade . . . and industrial policies. The policy preferences of different individuals in an economy, and whether and how they attempt to influence government decision making, will depend critically upon their ability to shift assets between alternative types of economic activities. (2008: 1)

These models build on this recent strand of research summarized above by focusing on the impact of  factor mobility on the trade policy attitudes of  the societal actors, workers, and capital owners in the theoretical story. In this regard, two assumptions are adopted in this model. The first assumption is that the trade policy preference of the capital owners is determined by the degree of asset specificity of the industries they own. More specifically, the owners of asset-­specific industries in the theoretical model favor higher trade barriers. They thus lobby the parties for more trade protection and offer “legal” campaign contributions to the parties to induce politicians to implement trade policies that are compatible with their (the owners’) interests.6 It is not difficult to discern why the capital owners of asset-­specific industries favor higher trade restrictions. These owners face greater adjustment costs and declining income when the economy is exposed to more import competition (Alt et al. 1996; Hathaway 1998; Hiscox 2001a). The increased import competition that results from trade liberalization also endangers their economic survival in the market. Additionally, these capital owners earn larger quasi rents from protection of their industries from import competition.7 Therefore the capital owners have incentives to lobby for protection by offering legal contributions to the parties in the newly democratized state.8 Furthermore, given that close financial ties often exist among industry owners (i.e., big business elites) and political parties as well as leaders in new democratic regimes—­which leads to corruption—­the owners in the model also offer “illegal” political donations to the parties.9 These illegal donations are offered by the industry owners more generally to obtain a variety of favors, including government and procurement contracts, the ability to buy state-­owned assets at below-­ market prices, and the ability to influence policies (e.g., trade policy) to suit

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  33

their interests. These donations provide the owners with access to policy makers and exert pressure on the parties, especially the ruling party. The second assumption is that the trade policy preference of the workers in the newly democratized state is determined by their factor mobility. Following recent decision-­theoretic models of trade protection with “search frictions” that study how the factor mobility of workers determines their trade policy preference, it is formally assumed in the model that workers with higher levels of interindustry occupational mobility benefit materially from trade reforms (e.g., Moscarini 2001; Costinot 2009). One reason for this is because occupationally mobile workers face low net adjustment costs of shifting from one industry to another industry (Moscarini 2001; Matschke 2006; Costinot 2009). Occupationally mobile workers can shift more easily from their industry if it is threatened by trade liberalization and transfer to other industries that are not hurt by import competition (Hamermesh and Pfann 1996; Matschke 2006). As suggested by Costinot (2009: 130), when “workers are more mobile,” they “lose less when switching sectors. This implies better outside options once unemployed, which reduces their incentives to be protectionist.” Further, occupationally mobile workers can also relocate more easily to “rising” industries that both gain from trade openness and offer higher wages to workers when trade policies are liberalized (Prichett 2006; Harrison 2007; Costinot 2009).10 Mobile workers therefore earn higher wages when trade policies are liberalized, and this drives them to support trade reforms (Prichett 2006; Costinot 2009). In contrast, workers that are occupationally specific lose from trade reforms, as more import competition may adversely affect the industries in which they are employed and consequently hurt their employment prospects (Harrison and Revenga 1995, 1998; Davidson and Matusz 2000; Harrison 2007; Costinot 2009). Their wages also decline when trade barriers are reduced. This is emphasized by Pavcnik et al. (2003: 5) in their research on labor markets and trade policy in developing countries: “If workers cannot switch industry employment easily, . . . workers in industries with larger tariff reductions are expected to observe a decline in their wages relative to workers with the same observable characteristics in industries with smaller tariff declines.” Thus, unlike mobile workers, occupationally specific workers are more likely to resist trade reforms (Harrison and Hanson 1999; Pavcnik et al. 2003; Costinot 2009). As described more formally in the next section, the political parties in the new democracy optimally choose the level of trade protection, given

34 / Chapter Two

the trade policy preferences of labor and capital (described above), which is common knowledge. They do so in the context of three main political features of new democratic regimes identified in the literature on democratic transitions.11 These three features include the shift from autocratic rule to a new democracy, the introduction of elections,12 and the possibility that the ruling party in the new democratic state may rig the election, particularly if it anticipates losing the election ex ante (Schedler 2002; Levitsky and Way 2006).13 Furthermore, it is assumed in the model that workers constitute a significant demographic and thus electoral majority. This is plausible particularly in the context of developing states, as these countries are either low-­or middle-­income nations where workers constitute the demographic majority by far (Bates and Krueger 1993; Milner and Kubota 2005). Because workers constitute the vast majority of the electorate in the newly democratized regime, it follows that the “median voter” is a worker as well.14 As we will see below, exploring the manner in which the parties respond to the preferences of labor and capital under the political conditions delineated above helps one discern when parties in new democracies can optimally choose to reduce tariffs and NTBs and how industry owners in turn could respond to the parties’ proposed trade policies. The Formal Model: The Economy and the Players The presentation of the model begins with a brief formal description of the economy. I then list the players, describe their utility functions, and discuss formally the nature of political competition in the model. Finally, I present the results from the model. I analyze a small, open economy of a developing country producing two goods with two factors of production—­that is, labor and capital. The domestic price of the imported good (in terms of the exported good) is p p* t J, where p* is the fixed world price while t J is the specific tariff level of the imported good that is proposed by the two parties [ J ∈ {A , B}]. Let y(p) denote the supply of the import-­competing good, with y( p) > 0. Imports are then m(p ) = d(p ) L − y(p), where d(p) denotes the demand of the imported good. Tariff revenue in the model is defined as T = t Jm(p),15 which by assumption is redistributed lump-­sum to citizens in the economy. Finally, m′(t J ) < 0, which means that imports fall as tariffs (i.e., price of the importable) increase; this is a standard assumption in political economy models of trade. There are two political parties [ J ∈ {A , B}] in the model: the ruling party/ government (labeled as A) and the opposition party (denoted as B). The set

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  35

of citizens in the economy is defined as i ∈{1,.... N} . Furthermore, following Goldberg and Maggie (1999: 1138), I assume that in the set {1, ...N}, there exists a fraction of people (denoted as C) who own asset-­specific industries16 and therefore have incentives to lobby for protection.17 The citizens that do not own asset-­specific industries are workers (denoted as l) who constitute a significant demographic and thus electoral majority, which holds true for developing states. After the transition to democracy—­in other words, in a new democracy—­the two parties propose their respective tariff policy, t J [ J ∈{A, B}], in the context of electoral competition. That is, the two parties, A and B, propose policies tA and tB, respectively, to the voters. Therefore, more formally, the electoral issue in my model of trade policy belongs to the interval Γ ⊂ ℜ; this implies that for any given pair of policies (tA , t B ) proposed by the parties, it follows that (tA,tB)∈ Γ × Γ. To examine how electoral competition affects trade policy in the immediate postdemocratic transition period, one needs to first define the utility functions and strategies of the players and then define the structure of political competition in the model. Utility Functions, Political Competition, and Sequence of Moves The utility function of the workers (labor) in the model is drawn largely from Feenstra (2004). More formally, each l owns a unit of labor Ll = 1, and all workers have the same optimal consumption d( p), with d ′( p ) < 0. Their remaining income is spent on the numéraire good, I l − pd(p ). Thus, following Feenstra (2004),18 individual utility of workers’ is given by (2.1)

u l (p,I l) = I l − pd (p) + U [d( p)]

where I l denotes their income. The income of each l (i.e., I l ) who owns a unit of labor is determined by his or her wage rate (w), the net material benefit (v) realized from his or her level of human capital (H l )—­where H l influences his or her degree of interindustry occupational mobility,19 demobile noted as θ l (defined below)—­and his or her share of tariff revenue (T/L). Therefore, individual income is mobile I l = 1 (wL + θ lmobile vH’ L + T ) = 1 (wL + θ l vH + T ) L L where H is the aggregate human capital level in the economy and where θ lmobile = H l / ( H / L)—­ that is, where each citizens’ ability to be occupationally mobile across industries is affected by his or her level of human capital relative to the (H/L) ratio in the economy.20 Without loss of gen­ mobile erality, let θ lmobile∈ [0,1]. Substituting θ l in (2.2) and simplifying leads to

(2.2)

36 / Chapter Two

I l = w + vH l + (T / L). Note that total gross domestic product (GDP) in the economy is y0 ( p ) + py( p) = w L + vH, where y0 ( p ) represents the output of the numeraire good.21 Income in (2.2) can thus be rewritten as 1 mobile l (2.3) I = [(θ l − 1) vH + y 0 ( p) + py ( p) + T ] L I assume in (2.2) and (2.3) that individuals with higher levels of human capital are more occupationally mobile and that more occupationally mobile individuals (which results from higher levels of human capital) obtain greater benefits from shifting jobs. These intuitive assumptions are drawn from existing studies suggesting that unlike workers with low levels of occupational mobility, workers with higher levels of human capital and thus greater mobility realize higher net benefits from shifting jobs, since their adjustment costs of shifting jobs are low (Matschke 2004; Cameron et al. 2007). Moreover, as discussed earlier, existing studies predict that individuals characterized by higher (lower) levels of interindustry occupational mobility tend to favor (oppose) trade reform and support (resist) reduction of trade barriers because their net gain from more trade openness is likely to be positive (negative).22 This claim—­which is discussed later—­is formally captured in the model.23 The trade policy preference of workers, which is determined by their level of interindustry occupational mobility, is common knowledge to the players. I mentioned earlier that citizens who do not own asset-­specific industries (i.e., the workers) in the polity demographically outnumber the owners of asset-­specific industries in developing countries. Hence, in the polity of the new democratic regime, the workers constitute a significant electoral majority in the population of voters. The population of voters is distributed according to the distribution function F : ℜ → [0,1], with an associated density function f : ℜ → ℜ + where f  is a continuous and single-­peaked density function that takes positive values on Γ and is 0 everywhere else. Additionally, I define the set of voters who have single-­peaked preferences—­these preferences are determined by their level of factor mobility—­with respect to tariff policy and who are distributed according to the distribution function F as γ (tA , t B ).24 From γ (tA , tB) and the fact that the trade policy preference of voters is common knowledge (since the degree of factor mobility of labor and capital is common knowledge), I assume that given the two proposals (tA , tB), party J ∈{A , B} knows ex ante that they can (based on their respective tariff policy proposal) obtain the votes of those citizens with ideal policy in the following interval γ J (tA , tB) = [ tJa (tA ,t B), t Jb( tA , t B)] . I emphasized earlier that workers constitute a significant demographic

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  37

and electoral majority in developing countries. Observe that the demographic majority of workers implies that in the population of voters (distributed according to F ), the median voter (denoted as mv) is also a worker whose mv mv (i) degree of labor mobility is labeled as θ , (ii) income level is I , and (ii) utility is from (2.1), defined as u mv (p, I mv ) = ul ( p, I l ). In the model, each voter’s, including the median voter’s, strategy is to choose to either reelect or vote the incumbent out of office based on his or her evaluation of the effect that the incumbent’s proposed tariff policy has on his or her utility. Thus each voter, including the median voter, chooses to reappoint the government if his or her realized utility function is maximized from the tariff policy that is proposed by the government during elections. The voters’, including the median voters’, voting behaviors in turn affect the government’s probability of reelection (which is formally defined below) after it sets tariff policy tA in response to t B. Here, we turn to explore the owners’ utility function. I construct the utility function of the owners of asset-­specific industries from existing models of lobbying and trade policy (e.g., Grossman and Helpman 1994; Goldberg and Maggie 1999). The rationale for focusing on the trade policy preferences and strategic behavior of the owners of asset-­specific industries was described in chapter 1. To start, the utility function of the owners consists of the net difference between the material benefits they obtain from trade policy and the monetary payments offered to the parties to influence trade policy. The material benefit that the owners obtain from trade policy is given additively by three factors (see Goldberg and Maggie 1999): the returns on capital invested in producing the imported good, which is affected by the tariff policy and thus its domestic price ([k(t J )]); their surplus from producing the imported good (s(t J) = m(t J ) + d(t J)); and the tariff revenue generated from tariffs placed on imports tJ m(t J ), which is redistributed to them from the government. Put together, then, the benefit that the owners obtain from trade policy is defined as n(t J ) = k(t J ) + (c / C ) [s(t J ) + t J m(t J )], where C denotes each owner while C (defined earlier) indicates the share of the owners of asset-­specific industries in the economy. In the model, the owners seek trade protection and thus attempt to influence tariff policy by lobbying and providing legal campaign contributions to both parties. The owners also (as mentioned earlier) provide “illegal” donations to the parties. Yet offering contributions and donations is costly, which implies that the owners will provide monetary payments (both contributions and donations) to influence tariff policy only when

38 / Chapter Two

they have strong incentives to do so. Therefore, in the model, the owners’ incentives to provide contributions and donations are determined by the degree of asset specificity of their respective industries. As suggested in extant studies, the higher (lower) the level of asset specificity of the industry run by the owners, the greater (lesser) their incentives to not only offer legal contributions to influence tariff policy25 but also provide “donations” to exert influence or pressure on policy makers. I formalize the lobbying dynamics described above as follows. To begin with, the degree of asset specificity of each industry run by the owners is given by the parameter φ. The continuous parameter φ varies from low to high asset specificity. Further, φ ∈ ℜ + and φ ≥ 1, and moreover, φ is bounded from above.26 Given φ, the total monetary payments—­that is, legal campaign contributions and illegal donations—­that the owners offer to each party is 1 1 defined as r (t J , φ , dJ ) = φt J + φ dJ. In the aforementioned expression, φ tJ is 2 2 allocated by the owners for legal contributions that are given to specifically influence tariff policy, and dJ is the illegal donations that they provide more generally to influence policy makers.27 r (t J, φ , dJ) strictly increases in φ. Hence the higher (lower) the level of asset specificity of their industry, the greater (lesser) the amount of monetary payments that they offer. Given n(tJ ) and r (tJ ,φ , d J ), the net utility function of the owners is c

(2.4)

u (t J , φ ) = n(tJ ) − r (t J ,φ , dJ) 1  = k(t J ) + (c / C )[s(tJ ) + t J m(t J )− φ  t 2J + d J  2 

I build upon not only existing political economy models of trade policy but also the rich literature on economic dynamics (e.g., corruption) in new democracies in the developing world29 to define the objective function of the government and the opposition party. To begin with, the government (A) is a rent seeker that is interested in extracting rents from tariff policy and the monetary payments that it obtains from the owners of asset-­specific industries. But it is also interested in obtaining political support from citizens in the polity. I formalize the government’s dual goals in its objective function in three steps. First, the government weighs the trade policy preferences of citizens (given by their utility function) by the parameter β ∈ [0,1.] Second, it simultaneously weighs the preferences of owners in its utility function via (1− β ). Third, it is also interested to extract rent from the 28

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  39

payments provided by the owners. Therefore, the government (i.e., the ruling party A) has the following weighted objective function, defined as (2.5)

u A = β u l + (1 − β )uc + r (t J ,φ , dJ ) l l c 1 2  = β [u ( p, I )] + (1 − β )[u (t A ,φ )] + φ  t A + dA  2 

where β ∈[0 ,1] is an exogenous measure of the relative weight of the interests attached to labor relative to those of the owners’ in the government’s objective function. The parameter β that influences the relative weight mentioned above is, in turn, determined by the institutional context, described below. Similar to the government, the opposition is interested to extract rent from contributions but is also interested in obtaining political support from citizens and the owners. Thus, by symmetry, the opposition party’s objective function is (2.6)

1 2  u B = β [u l ( p,I l ) ] + (1 − β ) [u c (t B ,φ ) ]+ φ  t B + d B  2  

The government and the opposition’s goal is to optimally set their tariff policy, given by the pair (tA , t B ), from their respective objective function. Since (tA , t B )∈ Γ × Γ and Γ ⊂ ℜ, each party’s preference from their objective function is represented by u J : Γ → ℜ;[ J ∈ {A, B}]. Note that the choice of tariff policy is, however, influenced by the political—­that is, institutional—­environment in which these parties operate. To understand how the institutional context matters for trade policy, I follow Mansfield, Milner, and Rosendorff (2002) and assume that the parameter β ∈[0 ,1]—­which is incorporated in the government and the opposition’s objective function—­is determined by the institutional context in which these parties operate.30 More broadly, β captures the shift from autocratic rule to a new democratic regime. Specifically, I assume that when the polity shifts to a full-­fledged new democracy, which “ensures a credible transfer of political power to the majority of citizens,”31 then limβ →1.32 When β = 1, the parties operate in a new democracy and behave like parties in a Downsian model in which their objective is to set tariff policy so as to maximize their likelihood of winning the first posttransition election.33 I formalize electoral competition in the model of trade policy in a new democracy as follows. First, I define the difference in the share of votes for tariff policy tA against policy tB as the share of votes for tA minus the share of votes for tB. This is defined more formally as D : Γ × Γ → ℜ, where

40 / Chapter Two

(2.7)

D(t A , tB ) = ∫γ

A ( t A , tB )

f( z) d z − ∫γ

B (t A ,t B )

f (z)dz

If  tA = tB = tˆ, then γA = γB = {tˆ}, and so D( t,t) = 0 for all t ∈ Γ, which implies that both parties receive the same fraction of the vote. Following the standard Downsian framework, the party that obtains more than 50% of the vote wins the first postdemocratic transition election and thus implements its optimal tariff policy proposal. When both parties receive the same fraction of the votes, a fair coin decides the winner. The winning probability of tariff policy tA against tB is thus defined as (2.8)

1  π (t A , tB ) = 1 / 2 0 

if D( tA , tB ) > 0 if D( tA , tB ) = 0 if D( tA , tB ) < 0

Each party’s strategic objective is to optimally set tariff policy to win the election. The payoff functions are therefore their respective probabilities of winning: Π A (t A , t B ) = π (t A , t B) D

(2.9)

Π DB (t A, t B ) = 1− π (tA , t B)

Given the structure of political competition in the model, the equilibrium concept is technically defined as follows: Let Γ ⊂ ℜ be a set of tariff policies and let F : ℜ → [0,1] be a distribution func-

* tion. The Downsian equilibrium for (Γ, F) is a pair of policies (t* A ,tB ) ∈ Γ × Γ such that

(2.10)

* π (t*A , t* B ) ≥ π (t A ,t B )

∀t A ∈ Γ

* (1 − π (t *A ,t * B )) ≥ (1 − π (t A ,t B ))

∀t B ∈ Γ

Stated in less technical language, the structure of political competition formalized in the model (via equations 2.7 to 2.10) explicitly captures the political parties’ ex ante uncertainty about whether or not they will win the election during the immediate postdemocratic transition period. As shown in the next section, the parties’ uncertainty affects their strategic behavior and also their optimal choice of trade policy in equilibrium; it also (as described below) affects the strategic behavior of industry owners in the model. Finally, it was mentioned earlier that a critical feature of new democracies is that incumbents in these states tend to rig the election (i.e., engage in electoral fraud) to survive in office (Schedler 2002; Way and Levitsky 2006). This feature of new democracies is formalized in the model

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  41

by introducing the possibility that the government in the new democratic regime—­after observing the electoral outcome—­decides whether or not to rig the election result with probability α ∈[0 ,1]. The likelihood with which the government rigs the election is assumed to be inversely related to its probability of reelection—­that is, α = π (tA , t B ) and (1 − α ) = (1 − π (tA , t B))—­ which is intuitive. The sequence of moves in the model is as follows. The owners of industries, capital, move first, deciding the optimal level of contributions that they may offer to the parties in order to influence the opposition and the government’s choice of tariff policy. The contributions that are offered by the owners are contingent on the degree of asset specificity of their industry and their anticipation of whether or not the parties will implement some tariff policy that is compatible with their (the owners’) trade policy interests. The government and the opposition move next by proposing their respective optimal tariff policy in the context of electoral competition, given β and (1 − β ). The government’s tariff policy determines domestic prices as well as the income of voters (that they observe), which determine the voters’ welfare. Once the government chooses the tariff policy, the voters examine the utility that they obtain from the tariff policy. The voters, including the median voter, then choose to reelect or remove the government and install the opposition in office, which depends on whether the observed tariff policy maximizes their utility. Their voting behavior affects the government’s probability of reelection. Finally, based on its anticipation of the electoral outcome, the government decides ex ante that it may rescind or manipulate the electoral result ex post and thus commit electoral fraud with probability α. I solve the model described by using the subgame-­perfect Nash equilibrium solution concept. Equilibrium Tariffs and Monetary Payments in New Democracies The model’s equilibrium result is mathematically derived and characterized in the appendix to this chapter. The equilibrium and comparative static results that generate substantive theoretical predictions are presented below as nontechnically as possible and also with the aid of some illustrations extracted from the model. The first part of the equilibrium result (described in the appendix) formally characterizes the following outcomes in equilibrium in a new democratic regime: (i) the optimal tariff policy set by the political parties (t*J ) [ J ∈ {A, B}], (ii) the tariff policy preference of the median mv * voter (tdem ), and (iii) the monetary payments offered by the industry owners * ˆ (r (tJ ,φ , d J)). The second part of the equilibrium result is as follows:

42 / Chapter Two Lemma 1 (a) The transition to democracy in a developing country induces the parties to propose tariff policy that is biased toward the group in society that constitutes the electoral majority. This implies that the tariff policy proposed by the parties and implemented by the government in office *

P (tdem ) in a new democratic regime is biased toward the trade policy prefer-

ences of labor (i.e., workers). (b)  In a new democratic regime in the developing world, the tariff  policy proposed by the parties and implemented by the government in equilibrium converges to the median voter’s optimal tariff policy preference. Proof: See appendix.

The proof of these two results is provided in the appendix of this book. I discuss here the causal intuition underlying these results. The transition to democracy has two key political effects that influence the optimal trade policy choice of parties when the trade policy preference of societal actors is determined by their factor mobility. The first effect of democratization is that it engenders an expansion of  the selectorate and the winning coalition. As such, the expansion of the selectorate leads to enfranchisement of the workers who constitute the demographic majority in the newly democratized state.34 The enfranchisement of  the workers enhances their political leverage, given their majority status in the polity. The second effect of  democratization is the introduction of “free and fair” elections that generate political competition for office among the parties in a new democracy (Diamond and Partner 1993; Diamond 1999). As stated by Milner and Kubota (2005: 115), “Democratization means a movement toward majority rule with universal suffrage in contested elections.” The model reveals that the expansion of  the selectorate and the introduction of a credible election process induce political parties in newly democratized regimes to propose trade policies that are biased toward the trade policy preferences of  workers. This happens for several reasons. Increased political contestation among the parties in a new democracy generates uncertainty about the ruling party’s ability to retain office and the opposition parties’ ability to defeat the incumbent party and assume office. Such uncertainty drives the ruling party to enlarge its support in the selectorate. It also encourages the party in office to concurrently reduce the political support that the opposition may receive in the selectorate. This particular political dynamic in new democracies is emphasized by Verdier (1998: 589), who posits that “elites in a new democracy maximize profits by raiding one

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  43

another and promising tax cuts to masses in exchange for electoral (and other) support. More voters won to one’s side means greater capacity to raid the other side.” The model suggests that the ruling party in the immediate posttransition period expands the party’s political support group by building mass loyalty among workers, since workers constitute the vast majority of  the electorate in the developing economy. To build mass loyalty among workers, the ruling party has strong political incentives to adopt a trade policy position that is biased toward the preferences of  labor. Doing so allows the party in office to politically co-­opt the workers, induce the workers’ loyalty toward the executive, and discourage them from supporting the opposition. The opposition parties in the new democracy, whose incentives are similar to the ruling party, rationally respond to the ruling party’s strategy—­as suggested in lemma 1—­by proposing a trade policy position that is also biased toward the preferences of labor. The reason for doing so is not difficult to discern. In particular, if the opposition party fails to optimally respond to the government’s strategy by not co-­opting workers, then their own political future in the nascent democracy may be seriously compromised. Indeed, the failure to co-­opt workers will politically weaken the opposition while increasing the ruling party’s political power, particularly if the executive uses trade policy to successfully mobilize a majority of the workers into its political camp. Thus the opposition has no incentive to deviate from the strategy of proposing a trade policy position that is biased toward the interests of workers. The main consequence of the parties’ prolabor bias in the posttransition period is that it drives them to shift the equilibrium trade policy position to the median voter’s trade policy preference. This occurs because elections completely bind the executive and the opposition after democratization. When elections bind the parties, the ruling party’s ability to politically survive in office and the opposition’s goal to win office will depend on the median voter’s decision to reelect or vote the executive out of office. Indeed, once elections credibly constrain the trade policy decisions of the parties, the parties can maximize their vote share and their likelihood of winning the election only if they shift their proposed trade policy to the median voter’s preference. The second reason that drives the parties lies in the fact that this action complements their strategy of shifting trade policy toward the preference of  labor. Specifically, recall that the median voter in a developing economy is a worker that belongs to the labor force. Since the parties in a newly democratized state use trade policy as a tool to solicit the support of

44 / Chapter Two

t

(immobile)

Government’s tariff policy choice

(t A , t B ) [0,1]

(t A , t B )

0

(t A , t B ) 1

t mv

t β

0

1

*

*

A t dem

Median Voters’ tariff policy preference

tJ

(mobile)

Figure 2.1  Convergence to the median voter’s optimal tariff policy preference

the workers (the majority), owing to reasons delineated earlier and because the median voter is a worker, it therefore gives them further incentives to shift the equilibrium trade policy to the median voter’s preference. This dynamic from the model is illustrated for the ruling party (for example) in figure 2.1. In this figure, the x-­axis captures the shift from autocratic rule (β = 0 ) to a new democracy (β = 1), the y-­axis on the left-­hand mv* side illustrates the median voter’s optimal tariff policy preference t dem (at which his utility function is maximized), and the y-­axis on the right-­hand side shows the government’s rent-­maximizing tariff tˆ. The figure also illustrates the government’s probability of reelection (p (tA , t B )) from choosing tariff policy tA in response to the opposition’s choice of tA; as noted earlier, the government’s reelection probability depends on the median voter’s voting behavior after observing the ruling party and the opposition’s proposed tariff policy. As shown in figure 2.1, when β = 0, which implies a situation where the government is not constrained by electoral institutions and is autocratic, the government’s dominant strategy is to act as a pure rent seeker. It does so by implementing the rent-­maximizing tariff rate of tˆ. Once a democratic transition occurs and the regime shifts to a1 new democracy—­that is, β = 1—­ the incumbent’s strategic choice between the rent-­maximizing tariff and the median voter’s optimal tariff depends on the median voter’s voting behavior. If the government sets its equilibrium tariff policy (t* A ) at the median mv* voter’s optimal tariff preference (tdem ), then it maximizes the median voter’s utility, and this induces the median voter to vote for the government. This

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  45

in turn maximizes the ruling party’s likelihood of winning the election (π (tA , t B ) = 1). If the government fails to do so, however, the median voter switches to the opposition, and this increases the government’s probability of losing the election. Thus, as shown formally in the appendix, when β = 1, the government has no incentives to deviate from its best response of setting the tariff policy in equilibrium at the median voter’s optimal tariff policy preference. Once the incumbent locates the equilibrium tariff policy at the median voter’s optimal preference, the median voter has no incentives to mv* deviate from the strategy of voting for the incumbent, as t* A = tdem maximizes his or her utility. The logic described above also holds for the opposition party, therein explaining the result in lemma 1. Examples from the immediate postdemocratic transition period across various developing countries support the causal claims posited above. For instance, Weyland (2002) shows that democratization in Latin American countries weakened the effect of vested protectionist interests but increased the political power of the electorate, composed mainly of workers and the poor. He states that in Latin America “democratization reduces the political clout of the vested interests that benefited the most from the old development model, such as protectionist business sectors and the military. At the same time, it enhances the role of the electorate, including the large mass of poor people who received meager benefits under the old development model” (2002: 60). The “enhanced role of the electorate” following democratization in Latin American states encouraged policy makers in these countries to shift the agenda of domestic economic and trade policy to the preference of the expanded selectorate, which mainly includes workers and the poor (Weyland 2002: 60–­61; Armijo 2007). The overthrow of Marcos’ military dictatorship and the shift to democracy in the Philippines in 1986 enhanced the power of workers and the poor and their ability to influence policies during Corazon Aquino’s tenure in office. As suggested by Miller, “Marcos’ overthrow and the inauguration of the Aquino government created a political environment that offered hope and change . . . New policies were proposed that drew on the rhetoric of people’s participation and justice . . . the democratic opening allowed (them) to gain access and operate in the arenas of Congress, the Executive and government agencies with some potential for impact. The existence of special members of Congress appointed by the president to represent the urban and rural poor also helped some of the coalitions gain an inside voice in legislative deliberations” (1994: 6–­7). The result in lemma 1 and the examples discussed above are both interesting and important. But they also raise the following question: What is the

46 / Chapter Two

effect of the parties’ decision to shift the equilibrium tariff policy to the median voter’s optimal preference on the prospects for trade liberalization in newly democratized states? The first set of comparative statics derived from the equilibrium result provides an answer to this question. These comparative statics suggest that the prospects for trade reform in newly democratized regimes depends on the degree of labor mobility, as this determines the trade policy preference of workers, which includes the median voter. More specifically, the comparative static result—­the formal proof of this result is described in the appendix—­that directly answers the aforementioned question can be stated as follows. Proposition 1 (a)  In a new democratic regime, the government (i.e., the ruling party A) optimally reduces tariffs when the degree of the interindustry occupational mobility of the median voter and thus workers is sufficiently high. (b) The reduction of tariffs by the government under the conditions mentioned above increases its likelihood of reelection in the new democratic state. Proof: See appendix.

The causal intuition that leads to this prediction in proposition 1 is as follows. To begin with, note that when interindustry occupational mobility of labor is sufficiently high, the median voter’s level of occupational will also be reasonably high, as the median voter is a worker in a developing economy. The model shows that trade liberalization (i.e., reduction in tariffs) not only intensifies product-­market competition but also induces domestic industries that benefit from more trade openness to raise real output (for convenience, I label domestic industries that produce more output in the wake of trade reforms as “rising” industries). Labor-­market demand for workers thus increases in rising industries in order to utilize additional labor for producing more output, and such increased demand generates higher real income for workers that can relocate to these industries. Note that occupationally mobile workers (this includes the occupationally mobile median voter) find it much easier to relocate to these rising industries (because of lower adjustment costs), which implies that the real income of the occupationally mobile worker will—­by virtue of shifting to and working in rising industries—­increase because of trade liberalization.35 Thus high interindustry labor mobility means that the mobile median voter

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  47

gains from trade reforms (i.e., reduction of tariffs) and thus supports such reforms. This also holds more generally for workers (the demographic majority in the electorate) in the economy when labor mobility is sufficiently high. In equilibrium, the ruling and the opposition parties will observe that the median voter’s (and thus the workers’) level of interindustry occupational mobility is fairly high. They will therefore learn that the occupationally mobile median voter favors trade reforms. This will induce the parties to propose trade liberalization in equilibrium in the posttransition period because of the following reasons. First, the posttransition election binds the ruling and the opposition party in the newly democratized state. Since the election makes the two parties accountable to voters, the occupationally mobile median voter (who prefers trade reform) can credibly threaten to vote the ruling party out of office if it does not implement trade liberalization. The “mobile-­type” median voter can also credibly threaten to not vote for the opposition if it fails to propose trade reforms. Indeed, given that the occupationally mobile median voter can credibly use its electoral leverage to remove or reward the parties in the immediate posttransition period, the parties’ Nash strategy will be to propose and implement trade reforms by significantly reducing trade barriers. Second, the posttransition election constrains the rent-­seeking behavior of both parties. It particularly reduces their incentives to derive rent from trade protection when labor mobility is high, as the median voter in this case can use his or her electoral leverage to punish the parties if they extract rent by raising trade barriers. Lower incentives to extract rent from tariffs influences the parties to be more concerned with maximizing the occupationally mobile median voter’s wage from trade policy; this is achieved by curtailing tariffs. It is worth emphasizing here that parties in a newly democratized regime do not have incentives to deviate from adopting trade liberalization when interindustry labor mobility is sufficiently high. Trade reforms, in fact, allow each party to reap electoral dividends. For example, reducing tariffs in this case allows the ruling party it to secure the electoral support of the mobile median voter, since the “mobile-­type” median voter optimally chooses to vote for the incumbent party if it liberalizes trade policy. This allows the government to maximize its vote share. The same logic described above also holds for the opposition parties in the newly democratized polity. The Nash strategy of  the parties after democratization thus ensures that they optimally choose to reduce tariffs when labor mobility is sufficiently high.

48 / Chapter Two

*

A t dem

Equilibrium Tariff

0

1 2

4

6

8

10

12

mv (Simulated values)

Figure 2.2  Equilibrium tariffs for variation in b  and the median voter’s occupational mobility

A simple simulation derived from the model illustrates not only the causal intuition discussed above but also the result in part (a) of proposition 1 (see figure 2.2). This figure indicates the direction of the optimal tariff level set by the government (see y-­axis) as the interindustry occupational momobile bility of labor (and thus the median voter) shifts from low to high (θ l ) for two different levels of β : when β = 0 and when β = 1. Figure 2.2 shows quite interestingly that the optimal tariff set by the government in the mobile model increases in an autocracy (where β = 0), even when θ l is high. In sharp contrast, the optimal tariff implemented by the government in a new democracy (β = 1) in equilibrium strictly decreases as the interindustry 1 occupational mobility of workers shifts from low to high. The figure thus indicates that new democratic regimes have a negative effect on tariffs when labor mobility is sufficiently high. The causal story discussed so far and figure 2.2 thus generate the following testable hypothesis (which, as shown later, also applies to NTBs): Hypothesis 2.1. Governments in new democracies in the developing world will reduce tariffs during the initial posttransition years when the level of interindustry labor mobility is sufficiently high.

The model’s central theoretical result that the interaction between political parties and workers generates a prolabor bias in tariff policy in newly democratized states has important consequences for the behavior of capital owners of asset-­specific industries (who favor trade protection) in these states. This effect is summarized in the following comparative static result, which is proved formally in the appendix:

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  49 Proposition 2 The total monetary payment (legal campaign contributions and illegal donations) provided by the owners in equilibrium (ˆr (t* J , φ, d J )) declines in a new democracy—­even if the degree of asset specificity of their industries is high—­ when the interindustry occupational mobility of workers is sufficiently high. Proof: See appendix.

The causal logic that leads to proposition 2 (the proof of the proposition is in the book’s appendix) is straightforward. Specifically, note that in equilibrium, the capital owners observe that political competition unleashed by the transition to democracy drives the parties in the new democracy to propose trade policy that is biased toward the preferences of the median voter and labor. The posttransition election also places a strong binding constraint on the parties. I argue that this binding constraint encourages the parties to shift the equilibrium trade policy away from the preference of capital owners. Historical evidence broadly corroborates the causal claim that democratization induces politicians to shift trade policy away from the preference of capital owners. For instance, when the Philippines experienced a transition to democracy in 1986, the new democratic incumbent Corazon Aquino started disassociating herself and her political party from business groups that had close relationships with dictator Ferdinand Marcos, who was deposed from office in 1985. This is suggested by the following report from the Wall Street Journal in 1986: “Economic recovery still depends, however, on whether the Philippines can throw off decades of import-­substitution and cronyism, and start selling its wares in the world marketplace. For months, Mrs. Aquino has been gamely chipping away at monumental domestic trade barriers . . . Predictably, this liberalization is drawing sharp protests from entrenched, protected industries . . . Mrs. Aquino is not giving up on this grueling liberalization process . . . and is also maintaining a distance between herself and the entrenched but organized protected industries.”36 Corazon Aquino’s decision to distance herself from protectionist business groups had some critical consequences for business–­government relations during her reign in office. Owners of heavy industries such as motor vehicles, petrochemicals, and transport equipment that routinely lobbied Marcos’s authoritarian government for more protection did not exert pressure on Aquino’s government to raise trade barriers (Medalla 2002, 2004). This is because these industrialists were aware that Aquino was in “no mood” to “entertain” their protectionist demands.37 Funding for Aquino’s

50 / Chapter Two

political party in the country’s new democracy was also reportedly cut by industrialists, as they perceived her administration as a clear threat to their protectionist interests (Cororaton 1998). Similarly, leaders in Brazil and Peru distanced themselves from vested protectionist business groups during the immediate posttransition period (Weyland 2002: 141–­42). Democratization in Latin American states “weakened the political power of business associations, which were once dominated by protectionist sectors” (Weyland 2002: 60). Policy makers in the new democratic regimes of Latin America during the 1980s also “appealed to the will of the people and the common good and attacked special interests . . . to justify structural reforms that entailed diffuse . . . benefits and clear, concentrated costs” (Weyland 2002). Appealing to “the will of the people” and attacking special interests weakened financial ties between posttransition governments and protectionist interest groups in the new democracies of Latin America (Weyland 2002). Many industry-­based groups that favored trade protection were reluctant to “open their checkbook” and finance parties after democratization in Brazil and Peru, as they felt that populist leaders from these parties would ignore their protectionist interests.38 In fact, industrialists who favored trade protection in Brazil’s new democracy were hesitant to lobby and provide contributions to the posttransition government under Sarney, as they felt that the Sarney government ignored their pledges (Payne 1992). Thus democratization clearly dampened the incentives for capital owners to build or solidify financial ties with parties in new democratic regimes across Latin America. Note that the main consequence of the parties shifting the equilibrium trade policy away from the preference of capital owners is that neither the ruling party nor the opposition can credibly commit themselves ex ante to pursue the interests of the owners. The fact that the ruling and the opposition party cannot credibly commit themselves to pursue the owners’ interests engenders ex ante uncertainty among the owners about their ability to influence the political parties (and thus policy makers) ex post after providing monetary payments (both legal contributions and illegal donations) to the parties. This uncertainty makes it both prohibitively costly and risky for the owners to make contributions and illegal donations to the parties. Such uncertainty is also likely to generate a belief among capital owners that the parties are unlikely to respond positively to their (the owners’) trade policy preference. Thus capital owners in a new democracy are more likely to anticipate that their monetary payments will not influence trade policies implemented by the parties when labor mobility is sufficiently high. Consequently, the owners in the new democracy will rationally respond to

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  51

the commitment problem described above by reducing the legal and illegal payments that they offer to the parties in equilibrium. In short, the prediction in proposition 2 and the preceding discussion leads to the following hypothesis: Hypothesis 2.2. Campaign contributions offered to politicians by the owners of industries (that favor trade protection) in new democracies to influence trade policy decrease when interindustry labor mobility is reasonably high.

Trade Reforms and Electoral Fraud in New Democracies It was mentioned earlier that democratic institutions during the immediate posttransition years may be weak in developing countries and that policy makers in these states may be tempted to rig the election if they anticipate losing the election (Schedler 2002; Levitsky and Way 2006). Existing studies in this issue-­area suggest, for example, the intensity of electoral competition and the weakness of state bureaucracies in new democratic regimes may influence political leaders in these countries to manipulate the election result (Lehoucq and Molina 2002; Schedler 2009). While I acknowledge that the two factors mentioned above may encourage policy makers in new democracies to rig the election, I explore here whether and how trade liberalization in a new democratic regime in the developing world affects the likelihood of electoral fraud. The model also provides the following result, which allows one to answer this question: Proposition 3 When political parties in new democracies in the developing world adopt trade reforms during the immediate posttransition years, this reduces the likelihood of electoral fraud in these states. Proof: See appendix.

The formal proof of this proposition is provided in the appendix of this book. The causal intuition that explains the result in proposition 3 is as follows. To start with, note that when the government in a new democracy adopts trade reform by reducing tariffs (which occurs when θ lmobile is reasonably high), it reduces the volume of tariff revenue that is available from imposition of tariffs. A fall in tariff revenue reduces the opportunity for rent seeking and the amount of rent extracted in equilibrium by the government. This claim partly reflects Verdier’s argument stating that “tariffs are particularistic

52 / Chapter Two

rents, and new democracies simply must curtail particularistic rents out of political . . . necessity” (1998: 605; emphasis added). The fact that the government in the model extracts lower rent in equilibrium by reducing tariffs has the following effect. Specifically, the occupationally mobile median voter in the model learns that lower tariffs curtail rent seeking and rent extraction by the government. This also induces the median voter to vote for the executive, therein maximizing the government’s likelihood of winning the election ex post. Moving up the game tree, the government in the new democracy rationally anticipates the median voter’s support in this case and thus expects to win the election with a relatively high probability by liberalizing trade policies. Since the government believes ex ante that trade reforms will increase its likelihood of winning the election ex post, it reduces the credibility of the political threat to its survival in office from the opposition. It also increases the government’s time horizons in office. Reduced political threats to the government’s political survival in office, as well as longer time horizons that result from trade reforms, unambiguously decrease the government’s incentives to rig the election as predicted in proposition 3. The third proposition thus leads to the following testable claim: Hypothesis 2.3. The adoption of trade liberalization in new democracies in the developing world reduces the likelihood of electoral fraud.

Extending the Analysis to NTBs In this section, I extend the model described so far—­which focuses on tariffs and financial payments made by capital in new democracies—­to explore how trade politics in new democratic regimes affect NTBs. To the best of my knowledge, there is no conventional or standard approach for modeling NTBs in the kind of political economy models of trade policy described in the previous section. That said, I build on the formal model of trade protection presented in the preceding section to study the political economy of NTBs in new democracies across the developing world. The overall structure of the model for NTBs is identical to the model for tariffs and contributions (legal and illegal donations) discussed in this chapter. In particular, the economy; the trade policy preferences of labor and capital, which are determined by their relative factor mobility; and the main political features of a new democracy (expansion of the selectorate, introduction of elections, etc.) are the same as before. However, we extend the model presented earlier in the following three main ways to analyze the

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  53

effect of trade politics in new democratic regimes on NTBs. The appendix of the book contains the full formal description of the NTBs model and the proofs to save space here. First, following computable general equilibrium models that examine the effect of NTBs on prices and welfare,39 I assume that the trade policy choice adopted by the parties J ∈{A, B} is given by tJ. This is common knowledge to all the players in the game. tJ consists of two components—­ that is, tJ = tJ + tJ,n, where tJ is the respective tariff policy choice, while tJ, n, which denotes NTBs, is the additional policy instrument that the parties employ for trade protection.40 The domestic price of the imported good (in terms of the exported good) in this case is thus p = p* + t J = p* + tJ + tJ,n, where p* is the fixed world price, while tJ and tJ, n were each defined above. The level of NTBs proposed by the parties and implemented by the government if it retains office directly influences the domestic price of goods produced. This consequently affects the income of voters and the profits obtained by capital (the owners of industries).41 To keep the extended model technically tractable—­and to maintain consistency with the structure of the model on tariffs presented above—­I simply substitute tJ with tJ in the model for democracy and tariffs described earlier. Substituting tJ with tJ in the owners’ utility function in (2.4) leads to (2.11)

c 1 2  u (t J ,φ ) = k(t J) + ( c / C ) [s(t J) + t J m (t J )] − φ  t J + d J  2 

where m′(tJ ) = m′(t J ) = m′(tJ + tJ,n ). I let m′(tJ + tJ,n ) < 0. Since p = p* + tJ in the NTBs case, the workers’ utility function in equation 2.1 in the model for NTBs can be defined as follows: u l (p* + t J , I l ) = I l − ( p*+ t J )d( p*+ t J ) + U [d(p*+ t J )]. Finally, substituting tJ with t J in the ruling and opposition parties’ objective functions lead respectively to (2.12)

l l c   β [u ( p*+ t A ,I )]+ (1− β ) [ u (t A ,φ )] +φ  1 t A2 + d A   2

(2.13)

l c 1  β [u l (p*+ t B , I )] + (1− β )[u (t B ,φ)] + φ  t 2B + d B  2 

Second, as before, I assume that the owners also attempt to influence the level of NTBs in the new democratic regime conditional on the degree of asset specificity of the industries they own.42 They do so by providing both legal contributions and illegal donations to the parties. Third, after observing the NTBs proposed by the parties, the voters (which include workers and the median voter) evaluate the effect that the

54 / Chapter Two

proposed NTBs have on their utility and then choose to reelect the government or install the opposition into office. However, unlike voting over tariff policy, voters in the model are uncertain to some extent about NTBs in that they know that NTBs are being proposed and implemented by the ruling party (for example) but they are not completely certain about the exact level of NTBs being implemented. This is because certain types of NTBs (e.g., sanitary and phytosanitary measures) are less transparent than other more observable NTBS, therein making it difficult to gauge the precise level of NTBs. Hence I incorporate this uncertainty by assuming that the exact level of NTBs proposed by the parties, as perceived by the voters, is uniformly distributed over a finite interval. More formally, this means that NTBs in the model (as perceived by the voters) is uniformly distributed such that tJ,n ∼ U [ t n , t n ] and t J ,n ∈R +. After extending the model along the lines described above, I solve for and formally characterize the equilibrium level of NTBs that is proposed by the parties and implemented by the government in a new democratic regime. The formal derivation of the equilibrium result for NTBs in a new democracy is described in the appendix. The first part of the equilibrium result formally characterizes the political parties’ optimal NTBs policy in equilibrium in the new democracy, the median voter’s optimal NTBs policy preference, and the optimal monetary contributions offered in equilibrium by the industry owners in the new democracy. The second part of the equilibrium result is as follows: Lemma 2 (a) In the immediate postdemocratic transition period, the policy on NTBs proposed by the parties and implemented by the government in the new democracy is biased toward the trade policy preferences of workers. (b)  In a new democratic regime, the equilibrium level of NTBs proposed by the parties and implemented by the government converges to the median voter’s optimal preference for NTBs. Proof: See appendix.

The theoretical results in lemma 2 are proved formally in the appendix. The results of this lemma are not described here, as they are similar to the predictions in lemma 1. Yet it is interesting to note here that although the median voter (like other workers) is partly uncertain about the exact level of NTBs being proposed, the parties rationally choose to not exploit this uncertainty by shifting the equilibrium NTBs policy away from the median voter’s optimal choice. Furthermore, departing from recent studies on democracies

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  55

and NTBs (see Kono 2006, 2007), the parties in the model will also not provide biased information about NTBs in the newly democratized polity. Instead, the NTBs policy that they propose converges to the median voter’s optimal preference for NTBs. Two reasons explain these claims. First, the expansion of the selectorate during the posttransition years makes it extremely risky for the parties—­who are interested in politically co-­opting the electoral majority, the workers—­to deviate from the median voter’s (who is also a worker) optimal choice when setting the optimal level of NTBs. If one party (say the ruling party) deviates from the median voter’s optimal preference, then the other party (the opposition) will respond by locating its proposed NTBs policy at the median voter’s optimal choice. This increases the probability of losing the election for the party that deviates from the median voter’s preference. Since both parties prefer to win the election, they have no incentives to exploit the median voter’s uncertainty by shifting the equilibrium NTBs policy away from the latter’s optimal choice. Second, unlike recent accounts of NTBs in democracies (Kono 2006), the model presented here reveals that parties in a new democracy have little or no political incentives to provide biased information about NTBs. This is because in equilibrium the rational median voter can observe and gauge the effect that NTBs have on domestic prices and income. The median voter can use this “informational cue” of the price and income effect of NTBs to evaluate the level of NTBs proposed by the parties in equilibrium. He or she can also take this informational cue and either vote for the government if the NTBs policy is in accordance with the median voter’s optimal choice or vote the government out of office if the NTBs policy is not in accordance with the median voter’s optimal choice. Because the parties know that the median voter can analyze the price and income effect of NTBs to determine his or her voting decision, they cannot credibly “fool” the median voter in equilibrium by providing biased information about NTBs. This in turn gives the parties incentives to truthfully reveal their NTBs policy and situate the equilibrium NTBs policy at the median voter’s optimal preference. The following two comparative statics are derived from the equilibrium result in lemma 2: Proposition 4 (a)  The equilibrium policy on NTBs implemented by the government in a newly democratic developing country is dependent on the interindustry occupational mobility of labor and thus the median voter. Hence, in a new democratic regime, the government optimally reduces NTBs

56 / Chapter Two when the degree of the interindustry occupational mobility of labor is sufficiently high. (b)  In a new democratic regime, the total monetary payments offered to the parties by the owners (i.e., capital) to influence NTBs decreases in equilibrium when labor mobility is high. Proof: See appendix.

These results are largely similar to propositions 1 and 2. Stated briefly, the model predicts that the opposition and, more important, the ruling party will rationally respond to the “mobile-­type” median voter’s preference elucidated earlier by lowering NTBs. Doing so reduces nominal price rigidities. This consequently generates real income gains for the occupationally median voter in this case.43 Lowering NTBs also reduces protectionist rents that can be extracted by the government. Thus when the occupationally mobile median voter observes a decrease in NTBs adopted by the government, he or she politically supports the incumbent and votes for the government in equilibrium. This maximizes the government’s probability of reelection and induces it to adhere to the optimal strategy of reducing NTBs, even though a lower level of NTBs reduces the incumbent’s protectionist rents. Second, the incumbent and the opposition’s prolabor policy bias with respect to designing NTBs reduces their incentives to pursue ex post the promises that they may have made ex ante to the owners about policies regarding NTBs. Moreover, note that the owners do not have the electoral leverage to induce either the government or the opposition to shift their policy bias from labor to the owners’ interests. The inability of the political parties to credibly commit themselves to pursue the owners’ interests in a new democracy induces the owners’ to believe ex ante that the financial payments that they offer to the parties is unlikely to influence the parties to alter NTBs policy to suit their industries’ interests. This encourages them to reduce their monetary payments (legal contributions and illegal donations) along the equilibrium path, even when the asset specificity of the industries run by the owners is sufficiently high.

Conclusion This chapter presents a model that explores how expansion of the selectorate, electoral competition, and the potential for electoral fraud in new democracies shape the strategic response of parties in these states to the trade policy preferences of two key groups in society: labor (workers) and capital (owners of industries). The trade policy preference of the workers and the

Trade Protection and Electoral Malpractice in New Democratic Regimes  /  57

owners, which is determined by their degree of factor mobility. affects the strategic choices that the political parties adopt in the model. However, the ability of each of these two groups to translate their preferences to trade policy outcomes in a new democracy depends on the political context of their interaction with multiple political parties: the ruling party (the government) and the opposition parties. Indeed, it is the institutional context of a new democratic regime in which strategic interaction occurs among the key players to determine which group’s “voice” in society is more predominant in the issue-­area of trade policy. The central result that emerges from the model presented here is that during the immediate years following a democratic transition, the level of tariffs and NTBs that are proposed by the parties and implemented by the government in office is weighted toward the trade policy preferences of labor (and thus the median voter), rather than toward the preferences of capital. Consequently, new democratic regimes have a negative effect on tariffs and NTBs when the level of labor mobility is sufficiently high. Moreover, given that the parties favor labor over capital, protectionist industry owners often anticipate that the financial payments that they offer to the parties are unlikely to influence trade policy outcomes. This dissuades them from providing such “payments.” Finally, the model shows that trade liberalization in a new democracy decreases the probability of electoral fraud in these states. The theoretical arguments proposed in this chapter agree with several existing studies that hypothesize that the transition to democracy influences trade barriers in developing countries (e.g., Milner and Kubota 2005; Yu 2007; Tavares 2008). Yet the predictions derived from the model depart to some extent from existing studies, suggesting that the effect of politics on tariffs and NTBs in the immediate years following a democratic transition is not direct—­rather, it is conditional on the degree of the interindustry occupational mobility of workers. Additionally, recent studies on NTBs suggest that democracies are more likely to favor higher NTBs to protect their industries from import competition (Drope 2007; Kono 2007, 2008). While it is possible that full-­fledged, well-­established democracies may increase NTBs, this chapter shows that new democracies have a negative effect on NTBs when the level of labor mobility is reasonably high. Put together, the theoretical results derived from the model lead to numerous testable hypotheses, stated earlier. In chapter 3, I present the results derived from systematically testing these hypotheses. The causal mechanisms underlying these hypotheses are evaluated in chapters 6 and 7 (the case-­study chapters of this book).

Three

Trade Protection and Electoral Fraud in New Democracies: The Empirical Evidence The first hypothesis (hypothesis 2.1) from the formal model on new democracies and trade protection is as follows: governments in new democracies in the developing world reduce tariffs and N T Bs when the level of interindustry labor mobility is sufficiently high. The second hypothesis (hypothesis 2.2) suggests that campaign contributions offered to politicians by the owners of industries that favor trade protection decrease in new democracies when interindustry labor mobility is reasonably high. The third hypothesis (hypothesis 2.3) is that the adoption of trade liberalization in new democracies reduces the likelihood of electoral fraud. Does the data provide statistical support for these hypotheses? In this chapter, I statistically test the hypotheses stated above on a comprehensive time-­series-­cross-­sectional (TSCS) data set of tariffs and N T Bs across several developing countries from 1972 to 2008. Unfortunately, cross-­national and pooled data on the amount of campaign contributions offered by the owners of industries to political parties in developing countries is not available. This makes it impossible to directly assess the prediction about campaign contributions in hypothesis 2.2. Yet, as discussed below, the World Economic Forum’s (WEF ’s) firm-­level survey-­response data set permits me to test the main causal claim that directly leads to the prediction in the aforementioned hypothesis. Recall that this causal claim posits that industry owners in new democratic regimes—­where labor mobility is sufficiently high—­are more likely to believe that the campaign contributions that they offer will not (owing to the reasons discussed in chapter 2) influence economic (including trade) policies that are implemented by the ruling party. This claim is carefully evaluated using the relevant data compiled by the WEF; the WEF ’s data are based on responses to survey questionnaires provided by firm owners from several developed and developing countries.

60 / Chapter Three

The rest of this chapter is organized into three sections. The first section, “Tariffs and N T Bs in New Democracies,” focuses on testing hypothesis 2.1. The next section, “Campaign Contributions in New Democratized Regimes,” evaluates hypothesis 2.2, and the third section, “Trade Liberalization in New Democracies and Electoral Fraud,” discusses the tests for hypothesis 2.3. For each of these three parts, I describe the sample, the statistical methodology, the operationalization of the dependent variable, and independent plus control variables that are employed for testing the relevant hypothesis; I then present the results from testing this hypothesis. The main findings are summarized in the conclusion of this chapter.

Tariffs and NTBs in New Democracies Sample and Dependent Variable Since hypothesis 2.1 applies to developing countries, I conduct the statistical analysis by using a TSCS data set of 129 developing countries from 1972 to 2008. The 129 developing countries in the sample are listed in table 3.1. The size and temporal range of the sample is comprehensive, as it includes all the developing countries observed during the 1972 to 2008 period for which data to operationalize the dependent variable(s) (described below) are available. There are two dependent variables of interest in hypothesis in 2.1: tariffs and N T Bs. I first describe the operationalization of the dependent variable, tariffs, for the empirical tests and then discuss the operationalization of the second dependent variable, N T Bs. The main indicator for measuring tariffs (the first dependent variable) is the import duty coverage ratio, which is labeled as import duties. For each country, the import duty coverage ratio is defined as the total value of a country’s import duties divided by the total value of its imports in a given year and is expressed as a percentage. Data for import duties are drawn from the following sources: the World Bank’s World Development Indicators (various years); the United Nations Conference on Trade and Development’s (UNCTAD’s) Trade Analysis and Information System database (TRAINS; var­ ious years); the World Trade Organization’s (W TO’s) Integrated Data Base (1998) and Trade Policy Review Series (various years); the UNCTAD’s Hand­­ book of Trade Control Measures of Developing Countries (1987) and the Directory of Import Regimes (1994);1 and, finally, the Global Trade Analysis Project’s (GTAP; 2011) data package. A key advantage of the import duties measure is that it is widely accepted as an accurate measure of trade protection via tar­ iffs by economists and political scientists (see, for example, Irwin 1998;

Trade Protection and Electoral Fraud in New Democracies  /  61 Table 3.1  List of developing countries in the data Albania Algeria Argentina Armenia Bahamas Bahrain Bangladesh Barbados Belize Benin Bhutan Bolivia Botswana Brazil Brunei Bulgaria Burkina Faso Burundi Cameroon Central African Republic Chad Chile China Colombia Comoros Congo, Democratic Republic of Congo, Republic of Costa Rica Cote d’Ivoire Cyprus Czech Republic Djibouti Dominica Dominican Republic Ecuador Egypt El Salvador Estonia Ethiopia Fiji Gabon Gambia Georgia

Ghana Grenada Guatemala Guinea Guinea Bissau Guyana Haiti Honduras Hungary India Indonesia Iran Jamaica Jordan Kazakhstan Kenya Korea Kuwait Kyrgyzstan Latvia Lebanon Lesotho Liberia Lithuania Madagascar Malawi Malaysia Maldives Mali Malta Mauritius Mexico Moldova Mongolia Morocco Myanmar Namibia Nepal Nicaragua Niger Nigeria Oman Pakistan

Panama Papua New Guinea Paraguay Peru Philippines Poland Romania Russia Rwanda Senegal Seychelles Sierra Leone Singapore Slovakia Slovenia Somalia South Africa Sri Lanka St. Kitts St. Lucia St. Vincent Sudan Suriname Swaziland Syria Tajikistan Tanzania Thailand Togo Tonga Trinidad and Tobago Tunisia Turkey Uganda Ukraine United Arab Emirates Uruguay Vanuatu Venezuela Vietnam Yemen Zambia Zimbabwe

Clemens and Williamson 2004; Hankla 2006; Dutt and Mitra 2006; Muk­ herjee et al. 2009). Furthermore, the import duties variable is available during the 1972 to 2008 period for the developing countries in the sample. The summary statistics for import duties from the full developing countries sample reveals that the mean of this measure is equal to 12.49%, while

62 / Chapter Three

the standard deviation is 8.7%. This indicates that there exists substantial variation for the import duties measure in the sample. While the main mea­ sure of tariff barriers is import duties, I also use statutory tariff as an alternative measure of tariffs for robustness tests. The data for statutory tariff are drawn from the sources mentioned above. The mean of this measure in the full developing countries sample is equal to 16.54%, while the standard deviation of this dependent variable is 12.19%. Thus there exists sufficient variation in the statutory tariff measure in the sample. The second dependent variable of interest in hypothesis 2.1 is N T Bs. Measuring N T Bs for developing countries is notoriously difficult, given the paucity of data on N T Bs for several developing states. Following extant research,2 I operationalize N T Bs as a single category core N T B coverage ratio (denoted as core N T Bs) for each country-­year. This is measured by calculating the percentage of imports subjected to core N T Bs. Note that core N T Bs include price-­control measures (excluding antidumping), quantitative restrictions (e.g., voluntary export restraints), finance control measures, monopoly measures, and technical regulation. I employ core N T B coverage ratios because they are the most commonly employed measure of N T Bs3 and have the maximum country-­year coverage compared to other mea­ sures of N T Bs.4 As reported below, I obtain similar results when employing frequency ratios—­the proportion of tariff lines covered by N T Bs (denoted as N T Bs ratio)—­to measure N T Bs. Data for core N T Bs are taken from the following sources: UNCTAD’s Trade Analysis and Information System database (TRAINS; various years),5 the World Bank’s (2006a) Trade and Production database and World Integrated Trade Solution (WITS) package (2007b), the WTO’s Integrated Data Base (1998) and Trade Policy Review Series (various years), the UNCTAD’s Handbook of Trade Control Measures of Develop­ ing Countries (1987) and Directory of Import Regimes (1994), and the GTAP (2011) data package. Although the core N T Bs measure described above is commonly used by scholars to operationalize N T Bs, a key drawback with this measure and all other available measures of N T Bs is that it is missing for several developing countries in the sample. In fact, the core N T Bs measure that I employ is available only from 1990 onward for the developing countries in my data. Data on core N T Bs exist for only two to three years for approximately one-­third of the developing countries in the sample.6 This leads to a highly unbalanced panel and also reduces temporal variation in the level of N T Bs within countries. Furthermore, the temporal range of the sample reduces considerably, given that data for N T Bs are only available from 1990 onward for developing states. Despite these shortcomings, however, I use the data on core

Trade Protection and Electoral Fraud in New Democracies  /  63

N T Bs to operationalize N T Bs because, as mentioned above, they (i) provide maximum country-­year coverage on N T Bs for developing countries that is available from primary and secondary sources and (ii) are generally considered an accurate measure of N T Bs by scholars.7 The mean of core N T Bs (operationalized as a coverage ratio) in the sample is 0.28, and the standard deviation is 0.17, which indicates that there is sufficient variation for this measure as well. The descriptive statistics for the three trade protection dependent variables (import duties, statutory tariff, and core N T Bs) as well as the independent and control variables that are described below are provided in table A in the book’s online appendix.8 Independent Variables for Testing Hypothesis 2.1 Hypothesis 2.1 posits that in developing countries, new democratic regimes have a negative effect on tariffs and N T Bs, conditional on the degree of the interindustry occupational mobility of workers in the economy. One thus needs to interact two key variables to test this hypothesis: an indicator variable for new democratic regimes in the developing world and a continuous measure that operationalizes the interindustry occupational mobility for each country-­year. I begin with a description of the procedure that we employed to operationalize “new democratic regimes” and then describe the formula for operationalizing the labor mobility measure for each country. Following Rodrik and Wacziarg (2005), I constructed the measure of new democratic regimes in two steps. First, I identified the exact year in which de­­ mocratic transitions occurred in the developing countries in the sample dur­­ ing the 1972 to 2008 period. To this end, I relied on different historical (i.e., primary) and secondary sources to precisely identify the year in which de­ mocratic transitions occurred in the countries in our data. These sources in­­ clude the following: the Freedom House (2005) and the Polity9 project for each country, Przeworski et al. (2000), United States Library of Congress (2015), Central Intelligence Agency (2008), and the United States Department of  State. (2015). More specifically, following recent research on democratic transitions,10 I identify each episode of democratic transition in the data according to the timing of the first international deemed “free and fair elections” and the adoption of a new democratic constitution after a prolonged period of autocratic rule.11 Based on the aforementioned criteria and the sources mentioned above, I find that during the 1972 to 2008 time period, there have been sixty-­four episodes of democratic transition across the developing countries in the sample for which data on trade barriers are available. Table 3.2

Table 3.2  Episodes of democratic transition in data set

Country

Democratic transition year

Albania

1992

Argentina

1983

Bangladesh

1991

Benin

1991

Bolivia

1982

Brazil

1985

Bulgaria

1990

Cape Verde

1991

Central African Republic

1993

Chile

1990

Comoros

1990

Croatia

2000

Czech Republic

1993

Djibouti

1999

Dominican Republic Ecuador

1978

El Salvador

1994

Estonia

1991

1979

Brief description of democratic transition episode Subsequent (in 1991 and 1992) elections marked the ending of the Communist rule. Coded as new democracy from 1992 to 1996. First free and fair elections after a prolonged period of military dictatorship. Coded as new democracy from 1983 to 1987. First postindependence (1971) free and fair elections. Coded as new democracy from 1991 to 1995. Adoption of new democratic constitution; free and fair elections. Coded as new democracy from 1991 to 1995. Return to civilian rule. Military steps down. Reconvention of 1980 democratic constitution. Coded as new democracy from 1982 to 1986. Return to civilian rule. Military steps down. Reconvention of 1980 democratic constitution. Coded as new democracy from 1985 to 1989. First internationally deemed fair and free elections after more than two decades of military rule. Coded as new democracy from 1990 to 1994. First post-­Communism free and fair general assembly elections. Coded as new democracy from 1991 to 1995. First free election after the oppressive Bokassa rule. The 1996 presidential elections were deemed free but marked by fraud allegations. Huge political instability is still present. Coded as new democracy from 1993 to 1997. First free and fair presidential elections, ending two decades of military rule by Augusto Pinochet. Coded as new democracy from 1990 to 1994. First postindependence (1975) free and fair elections. Coded as new democracy from 1990 to 1994. First free and fair legislative and presidential elections since independence and the ending of the Bosnian War. Coded as new democracy from 2000 to 2004. Independence from Czechoslovakia. First post-­ Communism constitution came into effect. Coded as new democracy from 1993 to 1997. First postindependence internationally declared free and fair elections. Coded as new democracy from 1999 to 2003. Return to civilian rule. Free and fair presidential elections. Coded as new democracy from 1978 to 1982. First internationally deemed free and fair presidential elections after a prolonged period of military rule. Coded as new democracy from 1979 to 1983. First free and fair elections after the end of a long-­lasting civil war and military rule. Coded as new democracy from 1994 to 1998. Independence from the USSR. Ratification of old (1938) constitution. Coded as new democracy from 1991 to 1995.

Table 3.2  (continued )

Country

Democratic transition year

Ethiopia

1995

Ghana

1996

Guatemala

1996

Guyana

1992

Haiti

1994

Honduras

1982

Hungary

1990

Indonesia

1999

Korea, Republic of

1988

Latvia

1991

Lesotho

1993

Lithuania

1991

Macedonia

1991

Madagascar

1993

Malawi

1994

Mali

1992

Mexico

1997

Mongolia

1992

Brief description of democratic transition episode First multiparty elections after a long-­lived Communist era. Coded as new democracy from 1995 to 1999. Internationally deemed free and fair elections. Coded as new democracy from 1996 to 2000. End of civil war; return to civilian rule. Coded as new democracy from 1996 to 2000. First postindependence (1966) free and fair elections. Coded as new democracy from 1992 to 1996. A US intervention placed the winner of the 1990 elections in office. Coded as new democracy from 1994 to 1999. Adoption of new democratic constitution after a prolonged period of oppressive rule. First democratically elected president takes office. Coded as new democracy from 1982 to 1986. First post-­Communism free and fair presidential elections. Coded as new democracy from 1990 to 1994. First multiparty elections after the collapse of the Suharto regime. Coded as new democracy from 1999 to 2003. Democratically elected government resumes office. Adoption of new democratic constitution. Coded as new democracy from 1988 to 1992. Independence from USSR. Ratification of old (1922) democratic constitution. Coded as new democracy from 1991 to 1995. Military abandons power and internationally deemed free and fair elections mark the return to civilian rule. Coded as new democracy from 1993 to 1997. Independence from the USSR. Coded as new democracy from 1991 to 1995. Independence from former Yugoslavia. First constitution approved. National Unity government formed. New democracy from 1991 to 1995. Presidential elections after a twenty-­year-­long military junta. Coded as new democracy from 1993 to 1997. First postindependence (1961) free and fair parliamentary and presidential elections. Coded as new democracy from 1994 to 1998. New democratic constitution established a multiparty system. Fair and free legislative and presidential elections followed. Coded as new democracy from 1992 to 1996. For the first time since 1929, the Institutional Revolutionary Party (PRI) lost absolute power in the lower house after the 1997 legislative elections. Coded as new democracy from 1997 to 2001. New democratic constitution established a multiparty system. Coded as new democracy from 1992 to 1996.

Table 3.2  (continued )

Country

Democratic transition year

Mozambique

1994

Nepal

1991

Nicaragua

1990

Nigeria

1999

Pakistan

1988

Panama

1994

Paraguay

1993

Peru

1980

Philippines

1987

Poland

1990

Russia

1993

Senegal

2000

Slovak Republic

1993

Slovenia

1992

South Africa

1994

Brief description of democratic transition episode First postindependence (1975) parliamentary and presidential elections. Coded as new democracy from 1994 to 1998. First free and fair elections since the early sixties. Coded as new democracy from 1991 to 1995. First free and fair elections after the Somosa dictatorship and the Santinistas revolution. Coded as new democracy from 1990 to 1994. After consecutive coups and military interventions, internationally declared free and fair elections mark the return to civilian rule. Coded as new democracy from 1999 to 2003. Legislative elections were held; restoration of the 1985 democratic constitution. In spite of fair and free elections in the nineties, the military coup of 1999 blocked democratization. Coded as new democracy from 1988 to 1992. First free and fair presidential and legislative elections after the US intervention. Coded as new democracy from 1994 to 1998. First presidential elections after decades of military rule. Coded as new democracy from 1993 to 1997. Internationally declared fair and free legislative and presidential elections. Coded as new democracy from 1980 to 1984. Adoption of new democratic constitution; free and fair elections led to the overthrow of Marcos’s regime. Coded as new democracy from 1987 to 1991. First post-­Communist free and fair presidential, legislative, and local elections. Adoption of new democratic constitution. Coded as new democracy from 1990 to 1994. Adoption of first post-­Communist constitution; free and fair Duma elections. Coded as new democracy from 1993 to 1997. First postindependence (1960) internationally deemed fair and free elections. Coded as new democracy from 2000 to 2004. Independence from Czechoslovakia. First post-­ Communism elections; a new democratic constitution came into effect. Coded as new democracy from 1993 to 1997. First free presidential and legislative elections since gaining independence from Yugoslavia,. Adoption of a new democratic constitution. Coded as new democracy from 1992 to 1996. First free elections with universal participation brought in power Nelson Mandela and ended the apartheid regime. Coded as new democracy from 1994 to 1998.

Trade Protection and Electoral Fraud in New Democracies  /  67 Table 3.2  (continued )

Country

Democratic transition year

Suriname

1991

Tanzania

1995

Thailand

1992

Turkey

1983

Ukraine

1991

Uruguay

1985

Zambia

1991

Brief description of democratic transition episode Return to civilian government after a one-­party regime; free and fair elections. Coded as new democracy from 1991 to 1995. First postindependence internationally deemed free and fair elections. Coded as new democracy from 1995 to 1999. Military was forced to step down. Free legislative elections followed. Coded as new democracy from 1992 to 1996. First free and fair legislative elections after a military dictatorship. Coded as new democracy from 1983 to 1987. Independence from the USSR. Legislative elections followed. Coded as new democracy from 1991 to 1995. Army returned power to the democratically elected president. Coded as new democracy from 1985 to 1989. First postindependence free and fair elections. New democratic constitution came into effect. Coded as new democracy from 1991 to 1995.

Note: This table reports the country, timing, and a brief description of the democratization event in each case. It lists full incidents of democratic transitions—­that is, in these countries, democratic institutions became fully consolidated (according both to the Polity and Freedom House indicators) within five years after making a transition to democracy.

lists each of these sixty-­four episodes of democratic transitions in the developing countries in the data, the year in which a democratic transition occurred in these countries, and a brief description of each transition event. Second, after identifying the democratic transition episodes in the data, I coded the dummy variable new democracy as 1 in the year t, in which a democratic transition occurred in each country (according to the list in table 3.2), and the subsequent four posttransition years (t + 4) for each country that experienced a democratic transition; it is coded as 0 for all other years. Put differently, new democracy is coded as 1 in the year of democratic transition and for the immediate four posttransition years—­which leads to a total of five years—­for each country in the data that experienced a democratic transition.12 Note that the results remain robust when new democracy is coded as 1 in the year in which a democratic transition occurs and (i) for three, five, or six posttransition years and (ii) for all postdemocratic transition years.13 I adopt a five-­year rule to code new democracy, where new democracy is coded as 1 at t (democratic transition year) and for each year in t + 4, because it helps me closely test the theoretical claim that focuses on how politics in a new democratic regime during the immediate years after a democratic

68 / Chapter Three Table 3.3  Economy-­wide employment categories in ILO, UNIDO, and GTAP data sets: Industries classified according to the three-­digit ISIC code Agriculture products, not processed Beverages Electrical machinery, appliances Fabricated metal products Finance, insurance, real estate Food products Footwear Furniture, fixtures, except primary metal Glass and glass products Iron and steel Leather products Mining and quarrying Miscellaneous petroleum and coal products Miscellaneous manufactures Nonelectrical machinery Nonferrous metal basic industries

Other chemicals Other nonmetallic mineral products Paper and paper products Petroleum refineries Pharmaceuticals Plastic products Pottery, china, and earthenware Printing, publishing, and allied industries Professional and scientific equipment Rubber products Textiles Tobacco Transport equipment Wearing apparel, except footwear Wholesale and retail trade, repair Wood, wood and cork, except furniture

Note: “Agricultural products, not processed” includes raw food materials such as rice, wheat, soybeans, corn, and sugar.

transition affect trade protection. This five-­year rule has also been used by scholars who have assessed the effect of new democracies on economic outcomes (see, for example, Rodrik and Wacziarg 2005). The third column in table 3.2, which lists the countries in the sample that experienced a democratic consolidation, not only describes each transition episode but also provides the years in which each country in the table is coded as a new democracy. Next, I operationalize the second key independent variable—­the interindustry occupational mobility of labor (in short, interindustry labor mobility) for each country-­year. Measuring interindustry labor mobility is notoriously difficult. To be as comprehensive as possible, I use two measures that operationalize the degree of interindustry occupational mobility of labor for each country-­year. The first measure, which is drawn from Wacziarg and Wallack (2004) and Hiscox and Rickard (2002), focuses strictly on the occupational mobility of labor from industry to industry across thirty-­two industries at the three-­digit International Standard of Industrial Classification (ISIC) level, as classified by the International Labor Organization (ILO) and United Nations Industrial Development Organization (UNIDO). I focus on the interindustry occupational mobility of workers across these thirty-­two industries—­which are listed in table 3.3—­because comprehensive data that are required to operationalize the interindustry labor mobility measure are available only for these thirty-­two industries at the three-­digit ISIC level.14

Trade Protection and Electoral Fraud in New Democracies  /  69

More formally, note that the first measure of interindustry labor mobility, labeled labor mobility, focuses strictly on the occupational mobility of labor from industry to industry within the economy. The variable labor mobility is based on operationalizing a well-­known measure of interin­dus­ try labor mobility (see Wacziarg and Wallack 2004; Hiscox and Rickard 2002). This measure is computed for each country-­year by isolating the fraction of jobs that move from one industry to another, independent of over­­ all employment gains or losses. Let E tj ,i denote employment in industry j in country i at time t. Hence N

(3.1)

labor mobility =

∑E

t j ,i

t −δ

− E j, i −

N

N

∑E − ∑ E t j ,i

j =1

j =1

1 2

N

∑( E

t j, i

t−δ j, i

j =1

t −δ

+ E j, i )

j =1

N

where the summation

S is over all N = 32 industries at the three-­digit ISIC j =1

level listed in table 3.3. The difference between the term on the left and the term on the right in the numerator in equation 3.1 gives the employment changes that result from the pure shifts of jobs across different industries.15 The denominator in equation 3.1 computes the average of total employment for the industries in consideration between t and t – d. I let d equal two years to capture meaningful shifts in interindustry labor mobility and to minimize the effects of business cycle shocks. Setting d  equal to one or three years did not alter the results reported for this measure below. The labor mobility measure is a continuous variable that ranges from low to high interindustry occupational mobility. Hence when labor mobility shifts from a lower to higher value, it implies that there has been an increase in interindustry labor mobility. Figure 3.1 illustrates the distribution of the labor mobility measure for two groups of developing countries in the data. The first group includes the sixty-­four “new democracies” in table 3.2 that are coded as 1 in the initial postdemocratic transition years. The second group consists of established democracies (e.g., Costa Rica and India) and autocracies in the developing countries sample. Note that countries in the second group are not coded as 1 in for the new democracy measure. Turning to figure 3.1, one finds that that the distribution of  labor mobility is similar across the two groups of countries in the data. It also shows that the distribution of labor mobility within each of these groups resembles a log-­normal distribution, which shows that there is significant variation in this measure for both sets of countries. Descriptive

70 / Chapter Three 3.5

3.0

New democracy = 1 New democracy = 0

2.5

Density

2.0 1.5 1.0 0.5 0.0 -60

10

70

Figure 3.1  Distribution of labor mobility measure

statistics reveal that the mean of the labor mobility measure in the developing countries sample is 10.14, the standard deviation is 9.25, and the minimum and maximum values are –­40.05 and 106.78, respectively. This further confirms that there is substantial variation for this measure in the developing countries sample. This provides more empirical leverage to test the first hypothesis. Data to compute the labor mobility measure are drawn from the International Labor Organization (ILO; 2007, 2008) LABORSTA 1 databases, the UNIDO (2007) database, the UNIDO (2009) Industrial De­ velopment Report, GTAP (2006, 2011), and the UN (various years) National Account Statistics Database. For robustness tests, I use an alternative measure of the interindustry occupational mobility of workers. This alternative measure is an elasticity measure of interindustry labor mobility. As suggested by extant theoretical and empirical studies on interindustry labor mobility, the elasticity mea­ sure (described below) conceptualizes and operationalizes labor mobility as the ability of workers to respond to wage incentives by moving across industries (Jones 1965; Hill and Mendez 1983; Hiscox 2002). In other words, the elasticity measure captures labor mobility by operationalizing the actual responsiveness of workers’ employment choices to interindustry wage differentials in each country (Hill and Mendez 1983; Hiscox 2002). To develop this second measure, I need to estimate the elasticity measure of interindustry labor mobility at the country-­year level. This elasticity measure is based

Trade Protection and Electoral Fraud in New Democracies  /  71

on the assumption that in competitive labor markets, the differential between the wage of industry i and the average wage of the whole economy is the primary determinant of labor movement into or out of industry i. Hence the change of the wage differential at time t –1 should explain the change of relative employment of industry i at time t. Following extant studies,16 I thus develop the elasticity measure of interindustry labor mobility in two steps. For each country-­year, I estimate the following equation: (3.2)



ln(E it / Et ) = γ t + β t ∗ ln(wi t −1 / wt −1 ) + ε it

where Eit is the employment at time t in each ISIC-­3 industry i listed in table 3.3, Et is total employment across all the ISIC-­3 industries in the economy (listed in table 3.3) at time t, w it -1 is the average wage in each ISIC-­3 industry i at t  – 1, and wi -1 is the economy-­wide average wage of all the ISIC-­3 industries at time t  – 1. Note that I lag the average industry wage and economy-­wide average wage by one period to empirically capture the idea that workers move their productive assets only after they perceive the wage differentials in the economy. Furthermore, in equation 3.2, e it is the error term that is assumed to be normally distributed and independent and identically distributed and gt is the constant term. After estimating 3.2 via ordinary least squares (OLS) with Newey-­West standard errors that are cor­ rected for heteroscedasticity and first-­order serial correlation, I obtain the es­ timated elasticity measure of interindustry labor mobility for each country-­ year, which is given by (3.3)

βˆ t =

d ln(E it / E t ) d ln(w it −1 /w t −1)

The estimated elasticity measure of interindustry labor mobility for each country-­year from equation 3.3 is labeled as ILM-­elasticity. Higher values of this measure imply higher levels of the interindustry occupational mobility of labor. The mean of this measure is 9.13 and the standard deviation is 13.79. This indicates that there is substantial variation in this measure in the sample. Data to operationalize the ILM-­elasticity measure are drawn from the same sources that are used to operationalize the labor mobility measure. To test the interactive effect posited in hypothesis 2.1, I interact the new democracy variable with labor mobility and introduce new democracy × labor mobility as well as the individual components of this interaction term in the empirical model where the dependent variable is the level of trade protection (tariffs and N T Bs). I repeat this exercise with the interaction term new

72 / Chapter Three

democracy × ILM-­elasticity in a separate model to test the robustness of the results. From hypothesis 1, I expect that the coefficient of new democracy × labor mobility and new democracy × ILM-­elasticity will be negative in the specification. Control Variables and Statistical Methodology I include several control variables that have been identified by scholars as important determinants of trade barriers in developing countries (e.g., Guisinger 2001; Milner and Kubota 2005; Ozden and Reinhardt 2005; Henisz and Mansfield 2006; Kono 2006). To begin with, extant research suggests that two sets of factors should be included. First, it is often argued that small countries tend to be more open than large ones (e.g., Katzenstein 1985; Easterly and Rebelo 1993; Rodrik 1997). I thus include two commonly used measures to account for a country’s size: log population and log GDP. A country’s level of economic development is also likely to affect its trade policy; more developed countries tend to have lower trade barriers (Easterly and Rebelo 1993; Rodrik 1995). Hence I control for log GDP per capita. Second, scholars suggest that economic crises tend to generate trade reforms in some developing countries but not others (e.g., Dornbusch 1992; La Ferrara 1996; Tornell 1998; Bruno 1996). For instance, Tornell (1998) claims that developing countries are likely to liberalize their trade regimes after an economic crisis because a crisis generates conflict among vested in­­ terests and weakens their political power. He defines a country as being in crisis if either its inflation rate is increasing rapidly or its real income is declining sharply. I thus use two different notions of economic crisis, both of which stress that crises are unusual and extreme shocks. The first one, which is similar to Tornell’s, deems an economic crisis to exist if one of two conditions holds: either the country’s inflation rate was 40% or more and it increased by 25% or more from the year before, or per capita GDP fell by 15% or more from the previous year. I label this variable economic crisis, which is coded as 1 if either of the two conditions mentioned above is satisfied and is coded as 0 otherwise.17 The second form of crisis involves the balance of payments. I include the dummy bop crisis (where bop denotes “balance of payments”), which is coded as 1 if a country’s level of international reserves falls to less than the equivalent of three months’ worth of imports and is coded as 0 otherwise.18 This second notion of crisis relates to a country’s debt and capital flight problems. The correlation between economic crisis and bop crisis is weak (r =.109) and statistically insignificant. Since researchers suggest that both

Trade Protection and Electoral Fraud in New Democracies  /  73

these types of crises can induce trade policy reform in developing countries, I include both these variables in the empirical model.19 Third, scholars suggest that international financial institutions such as the International Monetary Fund (IMF) and the World Bank may influence the trade policies of developing countries (Guisinger 2001; Milner and Ku­­ bota 2001, 2005; Subramanian and Wei 2007). I include a dummy variable for participation in IMF stabilization programs, which is labeled as IMF program. I also examine the impact of  the General Agreement on Tariffs and Trade (GAT  T) and W  TO on these countries. Joining GAT  T/W  TO may induce countries to lower their trade barriers (Gowa and Kim 2005; Tomz et al. 2005).20 I thus add the dummy variable GAT T/W  TO in the model, which is coded as 1 when a country is in the WTO/GAT  T and is coded as 0 otherwise. Researchers have also hypothesized that financial openness and a reduction in trade barriers are complementary (Garrett et al. 2000; Quinn 2001). To control for financial openness, I include Chinn and Ito’s (2006) capital ac­ count openness index in the specification. In the model in chapter 2, I assumed that the higher the degree of asset specificity of owners’ industries, the greater their incentives to lobby for trade protection. Furthermore, existing studies claim that the asset spec­ ificity of industries indeed affects lobbying and trade policy (Alt et al. 1996, Zahariadis 2002). To explore this claim empirically, researchers measure the asset specificity of industries by examining their research and development (R&D) expenditures because, according to Alt et al. (1999: 108), the intensity of R&D spending is “associated with either physical or capital specificity if the R&D produces learning effects.” Unfortunately, R&D spending for each industry within developing countries is not available. Thus to control for asset specificity, I compute the geographic concentration of industries in each country, which acts as a proxy for asset specificity.21 To this end, I develop from Spiezia (2002) a 0–­1 index of the adjusted geographic concentration of production by industries in each country in the sample; I label this measure the AGC index (where AGC denotes “adjusted geographic concentration”). The formula for the AGC index AGC index = GC / GC Max. N

In this formula, GC =

S y -a j

j

, where y j is the production share of  region j

j =0

Max

in a country, N stands for number of regions, and GC is the area of region j expressed as a percentage of  the country’s entire geographic area. Moreover, in the denominator is given by (3.4)

GC

Max

=

∑ a +1− a j

j ≠ min

min

= 2 (1 − amin)

74 / Chapter Three

where a min is the relative area of the smallest region.22 The 0–­1 AGC in­ dex ranges from low to high geographic concentration of production and is expected from existing studies to have a positive sign (e.g., Busch and Reinhardt 1999). Existing studies also suggest that relative factor endowments (specifically the capital-­labor ratio) may matter for trade policy in developing countries (Dutt and Mitra 2002; Milner and Kubota 2005). Hence I control for log capital-labor ratio in the empirical model.23 I also conducted a series of spec­ ification robustness tests by including additional control variables in the model that may potentially affect trade barriers in developing countries. The specification robustness tests are described later in this chapter. I turn to discuss the statistical methodology employed to test hypothe­­ sis 2.1. To this end, note that a TSCS data set as well as continuous measures for the dependent variables (import duties and core NTBs) are employed for the tests. Hypothesis 2.1 is thus tested by estimating TSCS regression models with panel-corrected standard errors (PCSEs) that are adjusted to correct for heteroscedasticity and contemporaneous correlation (Beck and Katz 1995). The models in which the elasticity measure of labor mobility (ILM-elasticity) is included as one of the independent variables are, however, estimated with bootstrapped standard errors. This is because the ILM-elasticity measure is itself estimated from equations 3.1 and 3.2. Therefore, I use bootstrapped standard errors to account for the estimation uncertainty that occurs when the estimated ILM-elasticity measure is used as an independent variable.24 I also include country-specific fixed effects in each empirical model to account for country-specific heterogeneity. I initially included dummies for each year in the specification but dropped these because F-tests indicate that the temporal dummies are jointly insignificant. To correct for serial cor­­ relation, I include the lag of the relevant dependent variable in the empirical models (Beck and Katz 1995). I conduct a battery of specification and econometric robustness tests as well as diagnostic checks to assess the consistency of my results. The results from these extensive robustness tests are reported below. A First Look at New Democracies, Labor Mobility, and Trade Barriers I discuss two sets of examples derived from the developing countries sample to broadly illustrate the relationship among newly democratized regimes, labor mobility, and tariffs. To this end, consider the following two examples: (i) the Philippines and Vietnam in Southeast Asia and (ii) Ghana and Uganda in Africa. The Philippines and Vietnam differ on numerous

Trade Protection and Electoral Fraud in New Democracies  /  75 60

Democratic Transition

60

40

45

20

30

0

15

Tariff Rate

Labor Mobility

Import duties (%)

75

-20

Labor Mobility 0

-40 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

75

60

60

40

45

1

30

15

0

20

0

Tariff Rate

Labor Mobility

Import duties (%)

Figure 3.2  Democratization, labor mobility, and import duties in the Philippines

-20

Labor Mobility

-40

1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

Figure 3.3  Authoritarian regime, labor mobility, and import duties in Vietnam

dimensions. Yet the Philippines and Vietnam are both low-­income developing countries, and the size and composition of the manufacturing and agricultural sectors (as a percentage of GDP) of these two Southeast Asian countries are similar.25 Further, one learns from the data used to construct figures 3.2 (the Philippines) and 3.3 (Vietnam) that the mean level of interindustry labor mobility in these two countries is roughly the same and the marginal rate of growth in the level of labor mobility in both these countries is virtually identical from 1982 to 1999. However, there is one key political difference—­the Philippines experienced a transition to democracy in 1986–­87, while Vietnam is observed as an autocratic regime from the 1970s onward. This difference is crucial, as figure 3.2 broadly shows that the 1

76 / Chapter Three

transition to democracy combined with growing labor mobility facilitated a sharp reduction in import duties in the Philippines. In Vietnam, however, the level of import duties did not decline but remained rather high, even though labor mobility increased in the country’s economy (see figure 3.3). The second example involves Ghana and Uganda. These two African nations are both low-­income developing countries, and the share of the manufacturing, service, and agricultural sector (as a percentage of GDP) is similar across these two countries (Hutchful 2002). The mean level and marginal increase in the level of interindustry labor mobility is also strikingly similar in these two countries. Despite these similarities, one finds that import duties decreased sharply in Ghana in the context of growing labor mobility only after the country experienced a transition to democracy in 1996 (see figure 3.4). Import duties, however, did not decrease appreciably and have consistently remained at a fairly high level over time in Uganda, even though labor mobility has increased in the country since the 1980s (figure 3.5). These two sets of examples therefore illustrate the theoretical claim that a negative correlation between higher labor mobility and tariffs is likely to exist in the posttransition period in newly democratized regimes. This phenomenon is less likely to occur in authoritarian regimes. I also analyzed the bivariate correlation between import duties and interindustry labor mobility and between the level of the main measure of labor mobility in the sixty-­four new democracies listed in table 3.2 and import duties across these countries during the immediate posttransition period. This exercise shows that higher labor mobility is indeed associated with a dramatic decrease in import duties in newly democratized regimes during

75

45

Import duties (%)

60

30

45

15

30

0

15

Tariff Rate

Labor Mobility

Democratic Transition

-15

Labor Mobility 0

-30 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

Figure 3.4  Democratization, labor mobility, and import duties in Ghana

75

45

60

30

45

15

30

0

15

Tariff Rate

Labor Mobility

Import duties (%)

Trade Protection and Electoral Fraud in New Democracies  /  77

-15

Labor Mobility 0

-30 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

Figure 3.5  Autocracy, labor mobility, and import duties in Uganda

the initial posttransition years. This is illustrated in figure A of the book’s online appendix, where one can observe that for both a linear regression line and a loss smoothing line, higher labor mobility (x-­axis) is negatively associated with statutory tariffs (y-­axis) in the sixty-­four new democracies for the posttransition period.26 A Spearman’s rho test indicates that this negative association is both strong (–­0.65) and statistically significant (p < 0.01). This figure further confirms the theoretical prediction of a negative correlation between higher labor mobility and tariffs is likely to exist in the posttransition period in newly democratized regimes. Results for Tariffs and NTBs Models 1, 2, and 3 in table 3.4 present the results from some baseline spec­ ifications (a.k.a. “stripped-­down”1models) that include the interaction term new democracy × labor mobility, the individual components of  this interaction term, log GDP per capita, and the lag of the dependent variable. The estimated effect of new democracy × labor mobility on each of the following dependent variables is negative and highly significant in the stripped-­down models: import duties (model 1), statutory tariff (model 2), and core N T Bs (model 3).27 Similarly, the statistical impact of new democracy × ILM-­elasticity on import duties (model 4), statutory tariff (model 5), and core N T Bs (model 6) is negative and significant in the stripped-­down specification.28 These results are encouraging but insufficient, as the baseline specifications do not include the remaining control variables. Therefore, table 3.5 first reports the results from two fully specified models (which include all the controls listed earlier) that test hypothesis 2.1 by

–­1.06 (2.02) –­1.02*** (0.328)

7.14*** (1.88) 1703

8.29*** (3.23) 1781

–­1.12*** (0.411) –­0.051*** (0.019) –­0.528 (0.607)

–­1.14 (2.16) –­1.19*** (0.424)

–1.08*** (0.32) –­0.047*** (0.015) –­0.715 (0.784)

Model 2

Model 1

1.87** (0.95) 764

–­1.51 (1.35) –­0.649*** (0.211)

–­1.74*** (0.66) –­0.04* (0.022) –­0.439 (0.577)

Model 3

Core NTBs

–­1.36*** (0.451) 9.85** (5.15) 1781

–­0.697 (0.624) –­1.68 (2.0)

–­1.14*** (0.329) –­0.038*** (0.015)

Model 4

Import duties

–­1.12*** (0.409) 6.05*** (0.611) 1703

–­0.563 (0.489) –­1.52 (1.7)

–­1.08*** (0.274) –­0.031*** (0.012)

Model 5

Statutory tariffs

–­0.833*** (0.326) 1.41** (0.673) 764

–­0.324 (0.509) –­2.0 (3.1)

–­0.978** (0.463) –­0.053* (0.031)

Model 6

Core NTBs

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. PCSEs are reported in parentheses for models 1, 2, and 3. Bootstrapped standard errors are reported in parentheses for models 4, 5, and 6. Each model in the table is estimated with country-­specific fixed effects that are not reported to save space.

N

new democracy × labor mobility new democracy × ILM-­elasticity Constant

new democracy

ILM-­elasticity

labor mobility

lag dependent variable log GDP per capita

Statutory tariffs

Import duties

Table 3.4  Baseline specifications for trade barriers

Trade Protection and Electoral Fraud in New Democracies  /  79

using the two main tariff measures as the dependent variables: import duties and statutory tariffs. These models are each estimated with a lagged depen­ dent variable, PCSEs, and fixed effects. The estimated effect of the interaction term new democracy × labor mobility on import duties (model 7, table 3.5) and statutory tariffs (model 8, table 3.5) is negative and statistically significant at the 1% level. This result statistically corroborates the prediction for tariffs in hypothesis 2.1. With respect to the individual components of this interaction term, one finds that the estimate of the individual labor mobility measure and the new democracy variable is statistically insignificant. Hence it is indeed the interaction of the two independent variables—­rather than each variable individually—­that has a statistically negative effect on tariffs in developing states. To gain a full appreciation of the impact of new democracy × labor mobility on import duties, I derive and analyze the substantive effect of  this interaction term in two steps. First, I plot in figure 3.6 (derived from model 1) the effect of new democracy on import duties across the observed range of labor mobility in the sample. The solid line in this figure plots the effect of new democracy on import duties at different levels of labor mobility, while the dashed lines in the figure indicate the 95% confidence intervals. The figure shows that new democracy does not have a significant impact on import duties for low levels of labor mobility. However, once the labor mobility measure rises beyond its mean value of 10.14 in the sample, the negative effect of new democracy on import duties becomes statistically significant. Additionally, one finds from the estimated interaction term new democracy × labor mobility that new de­ mocracy reduces import duties by approximately 17.3% when its companion variable labor mobility is increased by one standard deviation above its mean while holding other variables in the specification at their respective means in the sample. The marginal (and thus substantive) effect of new democracy on statutory tariffs across the observed range of labor mobility in the sample is similar to the substantive effect described above and is thus not described here to save space. Thus figure 3.6 and the percentage effect reported above suggest that there exists strong substantive support for hypothesis 2.1. Second, I compared the substantive effect of labor mobility on tariffs between the group of countries in the sample that are coded as 1 for new democracies and authoritarian regime country-­years in the data. The authoritarian regime country-­years consist of country-­years that are not coded as 1 for new democracy, since these autocracies did not experience a transition to democracy during the 1972 to 2008 period. Comparing the substantive effect of labor mobility on tariffs between the two sets of developing countries mentioned above further evaluates whether the influence of interindustry

IMF program

new democracy × ILM-­elasticity

new democracy × labor mobility

new democracy

economic crisis

bop crisis

ILM-­elasticity

labor mobility

log GDP per capita

lag dependent variable

–­0.033 (0.068)

–­0.030 (0.042) –­0.037 (0.034) –­0.714 (0.659) –­0.526*** (0.17)

–­0.058 (0.046)

–­0.016 (0.029) –­0.028 (0.033) –­0.632 (0.781) –­0.481*** (0.154)

0.798*** (0.13) –­0.004** (0.002) –­0.416 (0.399)

Model 8

Model 7 0.712*** (0.119) –­0.006*** (0.002) –­0.372 (0.512)

Statutory tariffs

Import duties

Table 3.5  Main results for trade barriers

–­0.530*** (0.191) –­0.047 (0.043)

–­0.301 (0.265) –­0.039 (0.077) –­0.027 (0.023) –­0.693 (0.687)

0.659*** (0.124) –­0.047*** (0.015)

Model 9

Import duties

–­0.404*** (0.159) –­0.068 (0.112)

–­0.118 (0.149) –­0.022 (0.055) –­0.036 (0.075) –­0.601 (0.775)

0.628*** (0.091) –­0.051** (0.027)

Model 10

Statutory tariffs

–­0.044 (0.095)

0.021* (0.012) 0.032 (0.04) –­0.437 (0.64) –­0.611*** (0.142)

0.884*** (0.239) –­0.005* (0.003) –­0.147 (0.203)

Model 11

–­0.417*** (0.112) –­0.078 (0.073)

–­0.079 (0.165) 0.012 (0.08) 0.036 (0.032) –­0.355 (0.617)

0.502*** (0.093) –­0.041* (0.022)

Model 12

Core NTBs

1.95 (1.79) –­0.060 (0.091) 0.030 (0.042) –­0.025 (0.079) 0.114 (0.885) 7.40*** (2.02) 1619

0.311*** (0.06)

1.3 (1.41) –­0.02 (0.039) 0.016 (0.029) –­0.055 (0.102) 0.133 (0.352) 7.90*** (2.18) 1619

0.427*** (0.081)

0.098 (0.077) –­0.010 (0.011) 0.023 (0.02) –­0.031 (0.03) 0.059 (0.05) 1.13** (0.554) 1574

0.104* (0.061)

0.065 (0.092) –­0.009 (0.012) 0.025 (0.016) –­0.035 (0.034) 0.044 (0.031) 1.82** (0.729) 1574

0.121** (0.058)

–­0.023** (0.011) 0.147 (0.134) –­0.061 (0.098) 0.235 (0.631) 11.31*** (2.72) 731

–­1.82** (0.093)

–­0.029*** (0.007) 0.035 (0.037) 0.001 (0.003) 0.055 (0.041) 1.66*** (0.45) 688

–­0.339*** (0.072)

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. PCSEs are reported in parentheses for models 7, 8, and 11. Bootstrapped standard errors are reported in parentheses for models 9, 10, and 12. Each model in the table is estimated with country-­specific fixed effects that are not reported to save space.

N

Constant

AGC index

GATT/WTO

log capital-­labor ratio

log GDP

log population

real effective exchange rate (REER)

capital account open

Marginal effect of new democracy on import duties

82 / Chapter Three 40

0

-40

-80 -50

0

50

100

Labor Mobility

Marginal effect

95% Confidence Interval

Figure 3.6  Marginal effect of new democracy × labor mobility on import duties

labor mobility on trade barriers differs between democratizing regimes and nondemocratizing states that are observed as autocracies in the 1972–­2008 period. To conduct this comparison, I first interact the labor mobility measure with a dummy variable for authoritarian regimes (labeled autocracy) and estimated the effect of autocracy × labor mobility on trade barriers.29 Results from this exercise, which are reported in the book’s online appendix (see table D), show that the estimated effect of autocracy × labor mobility on statu­ tory tariff and import duties in stripped-­down and fully specified models is positive but insignificant.30 I also extracted figure 3.7 from the model in which I test the effect of autocracy × labor mobility on statutory tariff. Observe in this figure that the marginal effect of labor mobility on tariffs in the sample of autocracies is pos­ itive but insignificant. This is in sharp contrast to the strong negative effect that labor mobility has on tariffs in the set of countries that made a transition to democracy. Hence when I compare the effect of labor mobility on tariffs across the two groups of countries mentioned above, I find that the in­­ stitutional context—­in this case, democratization—­plays a critical role in in­­ 1 fluencing the negative effect that interindustry labor mobility has on trade protection in developing countries. As an initial test of robustness, I replaced the labor mobility measure with the elasticity measure of labor mobility, which is labeled as ILM-­elasticity. Specifically, I dropped labor mobility from the model and instead incorporated new democracy × ILM-­elasticity as well as the individual components of this interaction term in the specification. The effect of new democracy ×

Trade Protection and Electoral Fraud in New Democracies  /  83

ILM-­elasticity on import duties (model 9, table 3.5) and statutory tariffs (model 10, table 3.5) is negative and highly significant in the specifications that are estimated with bootstrapped standard errors and country-­specific fixed effects. To complete reporting the results for hypothesis 2.1, I turn to discuss the estimates in models 11 and 12 that are obtained from statistically testing this hypothesis for N T Bs, operationalized as core N T Bs. Existing empirical studies typically control for each country’s real effective exchange rate (REER) rather than capital account openness and log population when testing the determinants of N T Bs (e.g., Mansfield and Busch 1995; Kono 2006). Thus in models 11 and 12, I include a measure of the real effective exchange rate (REER) but exclude capital account openness and log population.31 The remaining controls in these two models are identical to the other controls in models 7–­10. The impact of new democracy × labor mobility and new democracy × ILM-­elasticity on core N T Bs is negative and significant at the 5% level (see models 11 and 12 in table 3.5).32 These results statistically corroborate the latter half of the prediction in hypothesis 2.1, which posits that the effect of new democratic regimes on N T Bs in developing states will be negative, conditional on the level of interindustry labor mobility. To analyze the marginal effect of new democracy × labor mobility on core N T Bs, I first plot in figure 3.8 the marginal effect of new democracy on core N T Bs across the observed range of labor mobility in the sample. The solid line in this figure plots the marginal effect of new democracy on core N T Bs at different levels of labor mobility, while the dashed lines in the figure indicate the 95% confidence intervals. The figure shows that new democracy does not

5

Effect on Import duties

0 -5 -10

-15 -20 -50

0

50 Labor mobility

100

Effect when autocracy=1

Figure 3.7  Marginal effect of autocracy × labor mobility on import duties

150

Marginal effect of new democracy on core NTBs

84 / Chapter Three 40

0

-40

-80

-50

0 Marginal effect

50 Labor Mobility

100

95% Confidence Interval

Figure 3.8  Marginal effect of new democracy × labor mobility on core NTBs

have a significant impact on core N T Bs for low levels of labor mobility. However, once the labor mobility measure rises beyond its mean value of 10.14 in the sample, the negative effect of new democracy on core N T Bs becomes statistically significant at the 5% level. We also learn from the marginal effect discussed above that new democracy reduces core N T Bs by approximately 10% when its companion variable labor mobility is increased by one standard deviation above its mean while other variables in the specification are held at their respective mean in the sample. Unlike the strong statistical support for hypothesis 2.1, I find weak support for the remaining political and economic control variables. For example, the economic control variables, IMF program, bop crisis, economic crisis, log population, log capital-­labor ratio, and AGC index are each consistently in­­ significant in the specifications where tariffs is the dependent variable. The REER measure is insignificant in the models in which core NTBs is the dependent variable. Log GDP per capita has the predicted negative sign but 1 is significant at the 10% level in the first two models in table 3.5. Capital account openness is, however, positive and significant in the models where import duties and statutory tariff are the dependent variables. The GAT  T/W  TO dummy is insignificant. This supports the contention by Rose (2004) and Milner and Kubota (2005), who find that that the influence of the GAT  T/ W  TO on the trade policies of developing countries is weak. However, the lag of the dependent variable is positive and highly significant in all the specifications.

Trade Protection and Electoral Fraud in New Democracies  /  85

Robustness Tests for Tariffs and NTBs I conduct some specification and econometric robustness tests. For the first specification robustness test, I include the following additional controls: veto players and welfare spending (as a percentage of GDP). I include veto  play­­ ers using data from the World Bank’s Database of Political Institutions (DPI; 2012), even though the literature is divided over the issue of whether more veto players in government leads to higher trade protection (Henisz and Mansfield 2006; Milner et al. 2007). I control for welfare spending (as a percentage of GDP), since welfare expenditures by the government may be channeled toward unemployment benefits available to workers, which in turn may influence the workers’ incentives to remain occupationally specific or mobile (see, for example, Estevez-­Abe et al. 2002). In the augmented empirical specification with the additional controls mentioned above, the estimated effect of new democracy × labor mobility on import duties (model 13, table 3.6) and core N T Bs (model 14, table 3.6) remains negative and significant. I also find, but do not report to save space, that the coefficient of  new democracy × labor mobility is negative and significant in the augmented specification when statutory tariff is the dependent variable.33 Some more controls were added to the specifications; these controls include government partisanship, the percentage of the population over age fifteen that has completed secondary-­level education for each country as a proxy for educational attainment (education), and a linear time trend in the model.34 I do not report the results obtained after including these additional controls to save space, but the main results were unchanged. I also checked whether the main estimates remain robust after accounting for three potential econometric problems. First, it is possible that the independent variables of interest (new democracy, labor mobility, and ILM-­ elasticity) may be endogenous to the measures of tariffs. Hence I implemented Hurlin and Venet’s (2003) granger causality test for panel data to assess the potential endogenous relationship of the dependent variables (import duties, statutory tariff, and core N T Bs) and each of the three indepen­ dent variables (new democracy, labor mobility, and ILM-­elasticity). F-­statistics from the Hurlin and Venet (2003) tests conducted in the developing countries sample indicate that import duties, core N T Bs, and statutory tariff do not statistically influence new democracy, labor mobility, and ILM-­elasticity. This indicates that the dependent variables that I employ are not endogenous to the independent variables. Yet, out of an abundance of caution, I address the possibility of endogeneity by testing hypothesis 2.1 via a “system–­generalized method of

REER

capital account open

IMF program

new democracy × labor mobility

new democracy

economic crisis

bop crisis

labor mobility

log GDP per capita

Spatial-­autoregressive (AR)

lag dependent variable

Table 3.6  Robustness test results

–­0.003*** (0.001) –­0.247 (0.371) –­0.01 (0.072) –­0.021 (0.073) –­0.520 (0.396) –­0.367*** (0.08) –­0.004 (0.005) 0.15*** (0.049) –­0.384 (0.298)

–­0.002** (0.00) –­0.115 (0.153) –­0.022 (0.035) –­0.026 (0.06) –­0.371 (0.509) –­0.484*** (0.064) –­0.0 (0.0) 0.125*** (0.037) –­0.416 (0.343)

0.623*** (0.019)

Model 14

Model 13

0.454*** (0.045)

Core NTBs

Import duties

xtpcse models

–­0.01*** (0.003) –­0.251 (0.377) –­0.047 (0.099) –­0.036 (0.097) –­0.406 (0.822) –­0.372*** (0.099) –­0.012 (0.018) 0.188*** (0.064) –­0.168 (0.177)

0.322*** (0.051)

Model 15

Import duties

–­0.002** (0.001) –­0.095 (0.102) –­0.005 (0.057) –­0.024 (0.039) –­0.211 (0.159) –­0.265*** (0.072) –­0.0 (0.0) 0.122*** (0.036) –­0.165 (0.192)

0.419*** (0.125)

Model 16

Core NTBs

System–­generalized method of m oments (system-­GMM) models

–­0.003** (0.001) –­0.411 (0.827) –­0.021 (0.09) –­0.028 (0.03) –­0.714 (0.852) 0.293*** (0.088) –­0.0 (0.0) 0.109*** (0.025) –­0.153 (0.184)

0.67*** (0.078)

Model 17

Import duties

Heckman outcome equation

0.346*** (0.052) 0.084 (0.079) –­0.006*** (0.001) –­0.309 (0.265) –­0.018 (0.078) –­0.045 (0.033) –­0.387 (0.55) 0.438*** (0.04) –­0.001 (0.004) 0.152*** (0.04) –­0.207 (0.279)

Model 18

Import duties

Spatial-­AR error

569

2.9*** (0.721)

–­0.038 (0.059) 0.563 (0.979) 0.049 (0.033) –­0.039 (0.033) 0.227 (0.57) 0.051 (0.035) –­0.145 (0.812)

0.155 (0.643) –­2.51** –­0.384 1477

1.59*** (0.425)

0.062 (0.098) –­0.091 (0.198) 0.124 (0.877) 0.046 (0.034) –­0.237 (1.12)

–­0.029 (0.053)

0.129 (0.602) –­1.74** –­0.29 569

1.25*** (0.356)

–­0.039 (0.117) 0.434 (0.825) 0.031 (0.02) –­0.071 (0.12) 0.162 (0.387) 0.033 (0.029) –­0.264 (0.955)

1477

0.042 (0.067) –­0.042 (0.067) 0.13 (0.126) 0.066** (0.032) –­0.193 (0.43) –­0.029* (0.016) 2.12*** (0.24) –­1056.31

–­0.018** (0.009)

1477

1.84*** (0.603) –­893.14

0.055 (0.047) –­0.106 (0.144) 0.199 (0.721) 0.049 (0.036) –­0.381 (1.09)

–­0.016 (0.008)

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. The results reported for the system-­GMM models are from the one-­step estimation, except the Sargan test and the AR(1) and AR(2) tests, which are taken from the second-­step estimation. Note from Blundell and Bond (1998) that a negative and statistically significant AR(1) term plus a statistically insignificant AR(2) term indicates no serial correlation.

Log likelihood Hansen J test (p-­value) AR(1) AR(2) N

1477

1.33*** (0.219)

Constant

λˆ

welfare spending

veto players

AGC index

GATT/WTO

0.058 (0.045) –­0.058 (0.045) 0.211 (0.612) 0.029 (0.028) –­0.362 (0.274)

–­0.047 (0.075)

log capital-­labor ratio

log population

log GDP

88 / Chapter Three

moments” (system-­GMM) model that combines a regression in first differences and a regression in levels; estimating the two equations (levels and differences) in a single system leads to consistent and efficient estimates (Blundell and Bond 1998). This approach corrects for potential endogeneity by using moment conditions to derive a set of valid instruments for the potentially endogenous explanatory variables. It also corrects for serial correlation, controls for country-­specific fixed effects, and accounts for heteroscedasticity via White’s heteroscedasticity robust standard errors. I follow Blundell and Bond’s (1998) advice and estimate what they term a system-­ GMM model that involves estimation of a single system that combines a re­­ gression in first differences and a regression in levels. The instruments for the regression in first differences are lagged levels (dated t –­2) of the endogenous explanatory variables, while the instruments for the regression in levels are the lagged differences of the endogenous explanatory variables.35 Results from the system-­GMM models confirm the main statistically signifi­ cant findings about the interactive effect of new democracy and labor mobility on (i) import duties (model 15, table 3.6)36 and (ii) core N T Bs (model 16, table 3.6).37 One needs to address two more potential econometric challenges. First, it is plausible that factors that influence a nonrandom phenomenon such as the likelihood of democratic regimes may also explain the policy choice, including trade policies, of governments in emerging democracies.38 If one does not account for the nonrandomness of democratic regimes in the estimation process when testing their effect on variables such as trade protec­ tion, then the estimated results may suffer from selection bias. A typical strat­­ egy for controlling for selection bias is to use a Heckman selection model (Przeworski et al. 2000). I thus estimate a Heckman selection model. Spe­ cifically, I explicitly account for factors that influence the likelihood that democracy may emerge in the selection equation,39 while in the outcome equation of the Heckman model, I separately estimate the effect of new democracy × labor mobility and several control variables (listed earlier) on each of the three dependent variables: import duties, statutory tariff, and core N T Bs. I do not report the results from the selection model, as it is not directly relevant to the empirical analysis.40 However, the effect of new democracy × labor mobility in the outcome equation of the Heckman model is negative and significant at the 5% level when the dependent variable is import duties (see model 17, table 3.6) and statutory tariff and core N T Bs (not reported to save space). Second, some scholars suggest that economic policies, including trade policy, may be influenced by international diffusion in which developing

Trade Protection and Electoral Fraud in New Democracies  /  89

nations are, for example, driven to reduce trade barriers when they observe geographically neighboring countries—­which compete for the same pool of international capital—­decrease their level of trade protection (e.g., Milner and Kubota 2005; Guisinger 2001). This implies that the level of trade protection may be influenced by geographic proximity and thus characterized by spatial dependence, since governments may choose tariff rates based on the trade policy choices of neighboring countries. Interestingly, tests reveal that none of the three measures of trade protection that I employ (import duties, statutory tariff, and core NTBs) suffer from spatial dependence.41 Yet I evaluate the impact of new democracy × labor mobility in a spatial-­ autoregressive error model (hereafter spatial-­AR error model) that accounts for spatial dependence in the data in the estimation process. The spatial-­AR error model and the procedure used to estimate this model are briefly described at the end of the book.42 Note that the estimated coefficient of new democracy × labor mobility is negative and statistically significant in the spatial-­ AR error model where the dependent variable is (i) import duties (model 18, table 3.6) and (ii) statutory tariff and core N T Bs (not reported to save space).43 Finally, diagnostic tests reveal that none of the models in tables 3.5 and 3.6 suffer from severe multicollinearity, serial correlation, or omitted variable bias, and that the residuals are normally distributed.44 Exploring the Link between New Democracies and Labor Mobility Another statistical challenge that has not been addressed so far is the possibility that the transition to democracy engenders higher levels of labor mobility, which in turn leads to trade reforms. After all, some newly democratized regimes in eastern Europe, such as Poland, concurrently eased labor market restrictions (which promoted labor mobility) and adopted trade reforms during the initial posttransition period. If democratization first leads to higher labor mobility and then subsequently facilitates trade liberalization, then the validity of my conditional causal hypothesis and the supporting empirical evidence reported above—­namely, that newly democratized states reduce trade barriers when labor mobility is sufficiently high—­is questionable. Hence, to remove doubts about the validity of the reported empirical results, I conduct three empirical exercises to specifically check whether newly democratized regimes foster—­and are thus statistically associated with—­higher interindustry labor mobility in developing economies. These exercises are also employed to evaluate the statistical association of labor mobility and authoritarian regimes in the 1972–­2008 time period. Checking the link between labor mobility and authoritarian regimes

log inflation

trade openness

real GDP growth

log GDP per capita

unemployment rate

population over sixty-­five

partisanship

lag labor mobility

0.044*** (0.011)

0.145** (0.072) –­0.017 (0.051) –­0.088 (0.077) –­0.028** (0.014) 0.029** (0.01) 0.02* (0.012) 0.030 (0.085) –­0.0 (0.0)

Model 20

Model 19

0.197*** (0.066)

OLS

OLS

Dependent variable: labor mobility

Table 3.7  New democracies, autocracies, and labor mobility

0.407*** (0.134) –­0.024 (0.065) –­0.092 (0.085) –­0.035** (0.017) 0.032*** (0.011) 0.034* (0.019) 0.024 (0.097) –­0.0 (0.0)

Model 21

Double-­censored Tobit

0.033 (0.09)

0.03*** (0.011)

0.242*** (0.062)

Model 22

OLS

0.132*** (0.043) –­0.026 (0.057) –­0.091 (0.095) –­0.021** (0.01) 0.026*** (0.009) 0.039* (0.024) 0.026 (0.11) –­0.0 (0.0)

Model 23

OLS

0.108** (0.052) 0.023 (0.052) –­0.067 (0.059) –­0.019** (0.009) 0.022** (0.008) 0.015* (0.009) 0.021 (0.109) –­0.0 (0.0)

Model 24

Double-­censored Tobit

1634

0.001*** (0.003) 0.643*** (0.175)

0.033 (0.057)

1019

0.031* (0.018) 0.021 (0.017) –­1.66 (2.02) –­0.058 (0.046) 0.008*** (0.002) 0.517*** (0.165)

0.018 (0.029)

384.61** 1019

0.029* (0.017) 0.054 (0.082) –­2.0 (1.65) –­0.074 (0.059) 0.011*** (0.004) 0.729*** (0.197) 1077.91

0.021 (0.056)

1634

0.0017*** (0.004) 0.584*** (0.166)

0.026 (0.06)

259.74** 1019

0.014 (0.027) 0.03 (0.027) 0.049 (0.08) –­1.52 (1.7) –­0.044 (0.095) 0.015*** (0.003) 0.815*** (0.17) 1123.44

1019

0.011 (0.023) 0.026 (0.018) 0.015 (0.067) –­1.68 (2.0) –­0.067 (0.054) 0.01*** (0.002) 0.834*** (0.249)

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. Each model is estimated with country-­clustered standard errors and country-­specific fixed effects.

Log likelihood R2 c 2 N

Constant

time trend

union density

compensation

polity

labor deregulation

autocracy

new democracy

92 / Chapter Three 75

Labor Mobility

50

25

0

Labor Mobility in Autocracies

-25

Labor Mobility in New Democracies -50 1978

1982

1986

1990

1994

1998

2002

2006

Figure 3.9.  Labor mobility in autocracies and new democracies, 1972–­2008

helps one further assess if  it is just the transition to democracy that facilitates and is thus statistically associated with higher levels of labor mobility.45 For the first empirical exercise, I compared the mean level of the main measure of labor mobility described in equation 3.1 for autocracies and newly democratized states (i.e., countries that are coded as 1 for the new de­ mocracy measure). The mean level of labor mobility in newly democratized regimes during the initial posttransition years is approximately 7.45. The two-­sample t-­test and the Mann-­Whitney Wilcoxon difference-­of-­means test show that this mean score of 7.45 is largely similar to the mean labor mobil­ ity  level of 7.89 for authoritarian regime country-­years. The difference in the mean of labor mobility—­and the elasticity measure of interindustry labor mo­­ bility in equation 3.246—­for autocracies versus new democracies is also statistically insignificant. These are simple descriptive results, but they indicate that it is unlikely that the transition to democracy is associated with higher levels of labor mobility in comparison to autocracies. For the second empirical exercise, consider figure 3.9, derived from the 1 average of the labor mobility mea­ data. This figure illustrates the moving sure during the 1972 to 2008 period for autocracies and new democracies. It is clear from this figure that the degree of labor mobility has increased over time in both types of political regimes mentioned above. The marginal rate at which labor mobility has increased across these two types of political regimes for the 1972–­2008 period is also similar.47 This broadly suggests that democratization by itself does not precede higher labor mobility, as autocracies are also positively associated with increasing labor mobility in developing economies.

Trade Protection and Electoral Fraud in New Democracies  /  93

For the third empirical exercise, I estimated some statistical models to evaluate whether newly democratized regimes have a statistically positive effect on labor mobility in equation 3.1. The estimators employed are an OLS model (estimated with country-­clustered standard errors and fixed effects) and a double-­censored Tobit model (estimated with country-­clustered standard errors and fixed effects). The double-­censored Tobit model is used as the labor mobility measure (which is the dependent variable in this test) and is bounded from above and below. The effect of the new democracy dummy variable on labor mobility is positive but statistically insignificant in the stripped-­down OLS specification in model 19 (table 3.7), where no controls are included. The impact of new democracy on labor mobility remains positive but insignificant in the fully specified OLS (model 20, table 3.7) and double-­censored Tobit models (model 21, table 3.7). I include the following controls, which—­as suggested by extant research on nominal labor market rigidities—­influence labor mobility: log GDP per capita, population over sixty-­five, partisanship, unemployment rate, real GDP growth, trade openness, log inflation, labor deregulation, compensation, union density, and a linear time trend.48 The insignificant coefficient for new democracy suggests that newly democratized regimes do not statistically have a significant positive effect on interindustry labor mobility. Put more bluntly, the transition to democracy does not lead to higher levels of labor mobility. Finally, the effect of the au­ tocracy dummy on labor mobility in the stripped-­down and fully specified OLS and double-­censored Tobit models is positive but statistically insignificant as well (see models 22–­24, table 3.7). Thus neither newly democratized regimes nor autocracies have a statistically significant positive effect on the level of interindustry labor mobility.

Campaign Contributions in New Democratic Regimes As mentioned earlier, it is impossible to test hypothesis 2.2 because neither cross-­national nor pooled data on the actual amount of campaign contributions provided by the owners of industries is available for developing countries. Rather I focus on evaluating the key causal claim that leads to this hypothesis. Recall that this causal claim (labeled here as claim 1) posits that industry owners in new democracies where labor mobility is sufficiently high are more likely to believe that the campaign contributions that they offer will not influence policies, including trade policies that are implemented by the ruling party. I test this claim by using the World Economic Forum’s (WEF ’s) data on survey responses provided by firm owners from several

94 / Chapter Three

developed and developing countries. As described below, the WEF has gathered data on the extent to which industry executives within countries believe that the legal donations that they provide to parties (e.g., during elections through campaign finance) influences economic policies, including trade policies. These data have been gathered from responses to survey questionnaires completed by industry executives within several developed and developing countries across the globe. To be more precise, in the following years—­2001, 2002, 2003, 2004, and 2006—­the WEF (2004b) conducted a mass survey of executives (called the Executive Opinion Survey [EOS]) from 18 different industries49 in as many as 104 countries. The survey that was given to the executives in each country contains many questions. However, one particular question (described below) in the EOS survey is relevant for assessing claim 1. This question asks industry executives in each country to rank on a 1–­7 ordinal scale their perception of the extent to which they believe that legal donations (i.e., campaign contributions) provided to parties/politicians directly influence policies in their country. This question is worded as follows: ·· Question A. To what extent do the legal donations (you provide) to political parties have a direct influence on policies?50

Responses to this survey question were placed on an ordinal scale that ranges from a score of 1 (“there is a direct relation between donations and policies”) to 7 (“donations have no impact of on influencing policies”). The highest response score of  7 on this ordinal scale implies that industry executives essentially believe the legal contributions that they provide have no impact on the policies designed and implemented by policy makers. Note that the WEF put together the data for the industry executives’ response to this survey question for each of the 104 countries. It then aggregated the survey-­response data for each country and reported a score on the 1 to 7 or­ dinal scale listed above for each country for the following years: 2001, 2002, 2003, 2004, and 2006. This score has been annually reported and published for each country by the WEF in the Global Competitiveness Report in 2001, 2002, 2003, 2004, and 2006. I estimated some statistical models to evaluate claim 1 by using the WEF ’s (2004b) EOS survey-­response data. To this end, I first exclude the twenty-­two advanced industrial Organisation for Economic Co-­ordination and Development (OECD) countries from the EOS database, as these countries are not relevant for the analysis.51 This led to a total of eighty-­two developing countries in the sample that are listed in table 3.8. Next I use the

Trade Protection and Electoral Fraud in New Democracies  /  95 Table 3.8  WEF ’s Executive Opinion Survey data set Panel A. List of all developing countries in the WEF EOS sample Algeria, Angola, Argentina, Bahrain, Bangladesh, Bolivia, Bosnia-­Herzegovina, Botswana, Brazil, Bulgaria, Chile, China, Colombia, Costa-­Rica, Croatia, Czech Republic, Dominican Republic, Ecuador, Egypt, El Salvador, Estonia, Ethiopia, Gambia, Georgia, Ghana, Guatemala, Honduras, Hong Kong Hungary, India, Indonesia, Israel, Jamaica, Jordan, Kenya, Latvia, Lithuania, Macedonia, Madagascar, Malawi, Malaysia, Mali, Malta, Mauritius, Mexico, Morocco, Mozambique, Namibia, Nicaragua, Nigeria, Pakistan, Panama, Paraguay, Peru, Philippines, Poland, Romania, Russia, Serbia-­Montenegro, Singapore, Slovak Republic, Slovenia, South Africa, South Korea, Sri Lanka, Taiwan, Tanzania, Thailand, Trinidad and Tobago, Tunisia, Turkey, Uganda, Ukraine, United Arab Emirates, Uruguay, Venezuela, Vietnam, Zambia, Zimbabwe Panel B. List of new democracies in the WEF EOS sample Bangladesh, Bosnia, Bulgaria, Chile, Croatia, Czech Republic, El Salvador, Estonia, Georgia, Guatemala, Hungary, Latvia, Lithuania, Macedonia, Madagascar, Malawi, Mali, Mexico, Mozambique, Namibia, Nicaragua, Nigeria, Panama, Paraguay, Poland, Romania, Serbia-­ Montenegro, Slovak Republic, Slovenia, South Africa, Tanzania, Thailand, Ukraine, Zambia

WEF’s reported country-­year 1–­7 ordinal response score to question A for these 112 countries as the dependent variable for testing claim 1; this ordinal dependent variable is labeled as contributions influence. Because contribu­ tions influence is an ordinal variable that ranges from 1 to 7, I employ an ordered probit (OP) model to test this claim. One needs to interact the following two independent variables to test the causal claim that industry owners in new democratic regimes are more likely to believe that the campaign contributions that they offer will not influence po­­licies when labor mobility is sufficiently high: a dummy variable for new democratic regimes and a continuous measure for interindustry labor mobility. For the latter independent variable, I employ the measure of labor mobility in equation 3.1 that operationalizes the occupational mobility of workers across industries. With respect to operationalizing new democratic regimes, note that because the EOS was conducted from 2001 onward,52 countries in the EOS data are classified as new democracies if they experienced a transition to democracy in 1990 and any other year following 1990. The year 1990 is used as the threshold, since the most recent episodes of democratic transition in the developing world have occurred from 1990 onward. Moreover, given that the survey to compile the EOS data set was conducted from 2001 to 2006, countries that were relatively new democracies in the time period indicated above are precisely those states that experienced a transition in the immediate decade preceding 2001. The dummy new democratic regime is thus coded as 1 for the developing countries in the WEF EOS data if they experienced a transition to democracy in 1990 and

96 / Chapter Three

any other year following 1990; it is coded as 0 otherwise. Note, however, that the results do not alter substantively or significantly in the statistical sense if countries are classified as new democracies and they experienced a democratic transition in 1985 and in any year following 1985. To test claim 1, I interact new democratic regime with labor mobility and introduce the new democratic regime × labor mobility interaction term and the individual components of this interaction term in the OP specification. Based on the causal claim, I anticipate that the estimated coefficient of this interaction term will be positive in the OP specification. To identify the control variables for the OP specification, I turn to the study by Kaufman et al. (2008) that explores, among other issues, the extent to which industry executives believe that the legal donations that they offer to political parties/candidates influences policies in developed and developing countries. Their study suggests that the following three main variables may generate cross-­national variation in contributions influence: log GDP per capita, government fractionalization, and the degree of media freedom. Following this study, I therefore control for log GDP per capita, a 0–­1 index of gov­ ernment fractionalization that is drawn from the World Bank’s Database of Political Institutions (DPI; 2012), and a 0–­100 index of media freedom that is taken from the Freedom House (2012) database. The results from the OP specification are reported in model 25 (estimated without random effects) and model 26 (estimated with random effects) in table 3.9. The estimate of the new democratic regime × labor mobility in models 25 and 26 is positive and highly significant. This statistically corroborates the prediction in causal claim 1. The control variable log GDP per capita is negative and significant, but the remaining two controls, government fractionalization and media freedom, are each insignificant in the OP models. The substantive effect of new democratic regime × labor mobility in the OP specification is derived as follows. To begin with, I use the estimates from the full OP specification in model 25 and parametric bootstraps53 to compute the marginal effect of new democratic regime on the predicted probability of the maximum score in the ordinal contributions influence measure across the entire range of labor mobility; the maximum score of the ordinal contribu­ tions influence measure is equal to 7, and this score captures the following response: “Donations have no impact on influencing policies.” (I focus on assessing the predicted probability of this maximum ordinal level of 7, as it directly allows me to test the argument posited in claim 1.)54 The results in this case (illustrated in figure 3.10) show that new democratic regime does not have a significant impact on the predicted probability of contributions

Trade Protection and Electoral Fraud in New Democracies  /  97 Table 3.9  Ordered probit results for dependent variable contributions influence Model 25 log GDP per capita new democratic regime new democratic regime × labor mobility labor mobility government fractionalization media freedom µ1 µ2 µ3 µ4 µ5 µ6 Constant Random effects Log likelihood N

Model 26

0.012 (0.019) 0.057 (0.046) 0.081** (0.039) 0.026 (0.095) 0.016 (0.018) –­0.036** (0.019) –­0.995 (1.376) –­0.046 (1.450) 0.743 (1.284) 1.534 (1.294) 0.817*** (0.189) 0.743 (1.284) 2.087** (1.062) No –­2102.72 311

0.015 (0.023) 0.051 (0.042) 0.075** (0.034) 0.021 (0.112) 0.022 (0.034) –­0.03** (0.014) –­0.452 (0.924) 0.332 (0.923) 1.485 (0.920) 2.241** (0.92) –­2.96*** (0.86) 1.485 (0.92) 2.339** (1.175) Yes –­1967.85 311

Note: ***, **, * indicate 1%, 5%, and 10% levels of significance, respectively.

influence being equal to 7 for low levels of labor mobility. However, once the labor mobility measure increases above its mean in the sample, the positive effect of new democratic regime on the predicted probability of contribu­ tions influence (7) becomes statistically significant. New democratic regime increases the predicted probability of contributions influence being equal to 7 by 12.3% when its companion variable labor mobility is increased by one standard deviation above its mean while holding other variables in the OP specification at their mean. This effect is statistically significant at the 95% confidence level. Thus there exists strong statistical and substantive support for the prediction in causal claim 1 (summarized earlier), which directly leads to the prediction in hypothesis 2.2.

Marginal effect of new democracy on ∆prob of contribution influence = 7

98 / Chapter Three

0.6

0.4

0.2

0.0

-10.0 -30

0

30

60

90

120

Labor mobility Marginal effect

95% Confidence Interval

Figure 3.10 Effect of new democracy on contribution influence conditional on labor mobility

Trade Liberalization in New Democracies and Electoral Fraud: What Do We Learn from the Data? Sample, Dependent Variable, and Statistical Model The TSCS sample of 129 developing countries listed in table 3.1 is employed to test hypothesis 2.3, which posits that the adoption of trade liberalization in new democracies across the developing world reduces the likelihood of electoral fraud. Note that in the formal model in chapter 2 that produces the prediction mentioned above, the dependent variable is electoral fraud as an “event” that may occur in new democratic regimes with some probability. Hence the electoral fraud dependent variable is operationalized as a dichotomous measure to carefully test hypothesis 2.3. This dependent variable is operationalized as follows. Specifically, recall that the dummy new democracy is coded as 1 in the year of democratic transition and for the immediate four posttransition years 1 for each developing country in the data that experienced a democratic transition (and is coded as 0 otherwise). I use this new democracy measure to identify the countries in the data that are classified as new democracies and the years in which they are observed as new democracies (see table 3.2 for a list of these countries). I then code whether or not electoral fraud occurred in these countries. To do so, I follow the World Bank’s (2012) Domestic

Trade Protection and Electoral Fraud in New Democracies  /  99

Political Institutions (DPI) data and code the dependent variable in hypothesis 2.3—­electoral fraud—­as 1 when (i) electoral results are rigged by the incumbent or (ii) opposition parties are officially and constitutionally prohibited (or suppressed) in elections held in countries during the years in which they are observed as new democracies. If neither of these two conditions is satisfied, then the electoral fraud variable is coded as 0.55 Since electoral fraud is a dichotomous dependent variable in a TSCS data set, I employ the panel-­probit model estimated with random effects as the main statistical model to test the hypothesis about the effect of trade liberalization on electoral fraud. The main results remain robust when I use an alternative estimator—­the Markov-­transition probit model—­to test the hypothesis. Independent and Control Variables To derive the comparative static results in the formal model (see proposition 3 in chapter 2) that led to hypothesis 2.3, I conceptualized trade liberalization as a policy choice adopted by governments in new democracies with some probability under certain conditions. Thus, for the independent variable, I need a measure of whether or not governments in new democratic regimes adopted trade liberalization to test the hypothesis on electoral fraud. This measure is developed in two steps. For the first step, the dummy new democracy is coded as 1 in the year of democratic transition and for the immediate four posttransition years for each developing country in the data that experienced a democratic transition (and is coded as 0 other­ wise). I use this new democracy measure to identify the countries in the data that are classified as new democracies and the years in which they are observed as new democracies (listed in table 3.2). I then measured whether or not governments in these new democratic regimes in the sample adopted trade liberalization by using the Wacziarg and Welch (2004) dichotomous measure of trade liberalization that has been updated and partially edited from the Sachs and Warner (1995) measure of trade reform. Specifically, according to the original Sachs and Warner (1995) measure and the updated Wacziarg and Welch (2004) measure, a country is deemed to have adopted trade liberalization and is thus open if any one of the following criteria is true: NTBs cover less than 40% of trade, average tariff rates are strictly less than 40%, the black market exchange rate depreciated by less than 20% relative to the official exchange rate during the 1970s or 1980s, a state monopoly did not exist, and a socialist economy did not exist. This

100 / Chapter Three

measure is very useful because it considers many forms of protectionism and broadly captures the conceptualization of trade liberalization stated in hypothesis 2.3. The Wacziarg and Welch (2004) measure of trade liberalization also precisely identifies the years (i.e., when) trade liberalization was implemented by governments in several developed and developing countries. I used the Wacziarg and Welch (2004) trade liberalization measure to identify whether and when governments in new democratic regimes in my data adopted trade liberalization (as per the Wacziarg and Wallack criteria) during the immedi­ ate postdemocratic transition years. Thus the dummy independent variable new democracy trade is coded as 1 if governments in the new democracies in the sample adopted trade liberalization (according to the Wacziarg and Wallack criteria) during the immediate years following a democratic transition; it is coded as 0 otherwise. Numerous control variables are included in the panel-­probit model. To this end, I rely on some existing theoretical claims about the determinants of electoral fraud to identify a litany of control variables for the models where electoral fraud is the dependent variable.56 To begin with, I include log GDP per capita, as several studies suggest that higher levels of development make it harder for political parties to buy votes and this consequently discourages electoral fraud (Schedler 2002; Auyero 1999). I control for the new democ­ racy dummy variable as well, since scholars have argued that fragile political institutions in newly democratic regimes in the developing world make it easier for incumbents to manipulate elections in these countries (Brusco 2004; Schaffer and Schedler 2007; Bratton 2008). Some scholars suggest that electoral fraud is the consequence of concentrated economic interests. This was the case in imperial Germany, for example, where junkers and coal mine owners made sure the votes of their employees and subordinates were used to support conservative parties (Anderson 2000). I add the AGC index variable to the empirical model that was described earlier. This variable acts as a proxy for concentrated economic interests in the specification. Some studies of electoral fraud suggest that developing countries that were British colonies have inherited a robust legal system; hence it makes it more difficult for governments in former British colonies to rig the election (see, for example, Dominguez and McCann 1996). I therefore control for the dummy variable British colony, which is coded as 1 for countries in the sample that were former British colonies. Various researchers have also hypothesized that higher levels of ethnolinguistic fractionalization engenders political instability that may encourage political parties to manipulate elections in developing countries (Bunce and Wolchik 2006). I account for

Trade Protection and Electoral Fraud in New Democracies  /  101

this phenomenon by adding ethnolinguistic fractionalization (ELF) to the panel-­probit model. I also control for the Wacziarg and Welch (2003) trade liberalization dummy (described above) for all other developing countries in the sample that are not new democratic regimes.57 This is because it is plausible that trade reforms and increased exposure to the international economic system may influence the behavior of incumbents when opting to engage in electoral fraud. Following the aforementioned logic, I control for capital account openness, since higher levels of financial openness tend to consolidate democratic institutions, which in turn may dissuade leaders from manipulating elections.58 A handful of scholars suggest that governments in the developing world are more tempted to rig elections when economic crises occur and domestic economic conditions deteriorate considerably (Klesner 1993; Pacek and Radcliff 1995). I thus add economic crisis and bop crisis (two variables that were described earlier) to account for this possibility. Finally, higher levels of democracy may translate to more formal institutional constraints, therein making it harder for the executive to rig elections (Schedler 2002; Alston and Gallo 2009). I incorporate the polity measure of democracy to account for this possibility. Results for Electoral Fraud The effect of new democracy trade on electoral fraud in the full panel-­probit spe­ cification (which is estimated with random effects) in model 27 in table 3.10 is negative and significant at the 1% level. I conduct two empirical exercises to derive and fully explore the substantive effect of this result. For the first exercise, I use the estimates from the full panel-­probit specification in model 27 and parametric bootstraps59 to compute the marginal effect of a 0 to 1 change in the binary trade liberalization measure on the predicted probability of electoral fraud for two types of country-­years in the sample: country-­years in which the new democracy dummy is equal to 0 and country-­ years in which the new democracy dummy is equal to 1. The resultant first differences in expected values are reported—­via box plots of their distributions—­in figure 3.11. Figure 3.11 first indicates that in countries that are not classified as new democratic regimes (i.e., developing states in which the new democracy dummy is equal to 0), the effect of trade liberalization on the predicted probability of electoral fraud is not only statistically insignificant but also substantively negligible (almost 0). By contrast, and in support of hypothe­­sis 2.3, I find that in new democratic country-­years (in which the new democracy

102 / Chapter Three Table 3.10. Panel probit results for electoral fraud Model 27 lag electoral fraud log GDP per capita new democracy trade reform new democracy trade capital account openness AGC index polity

0.504*** (0.106) –­0.01*** (0.003) 0.068 (0.122) –­0.081 (0.136) –­0.163*** (0.051) 0.102*** (0.043) 0.075** (0.031) –­0.011 (0.024)

log population election monitor 0.064** (0.029)

ELF IMF program economic crisis bop crisis de facto judicial independence (DFJI) Constant Wald c  Log likelihood N 2

–­1.051*** (0.302) 57.43 –­987.21 1465

Model 28 0.497*** (0.123) –­0.009*** (0.003) 0.054 (0.124) –­0.072 (0.129) –­0.138*** (0.043) 0.092** (0.045) 0.069** (0.031) –­0.014 (0.031) 0.165 (0.388) –­0.021** (0.011) 0.053** (0.027) –­0.0 (0.0) –­0.025 (0.114) –­0.034 (0.099) –­0.012 (0.039) –­1.118*** (0.21) 59.26 –­949.73 1012

Note: ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively. The panel-­probit models in this table are each estimated with random effects.

dummy is equal to 1), trade liberalization has a sizable negative effect on electoral fraud. In particular, a 0 to 1 increase of the dummy trade liberaliza­ tion variable yields a 19.7% decrease in the predicted probability of elec­ toral fraud in new democratic regimes. This marginal effect is statistically significant, at the 95% confidence level. Hence there exists strong statistical and substantive support for hypothesis 2.3. For the second empirical exercise, I again used the estimates from the panel-­probit specification in model 27 (table 3.10) and parametric

Trade Protection and Electoral Fraud in New Democracies  /  103

bootstraps60 to compute the marginal effect of a 0 to 1 change in the binary new democracy measure on the predicted probability of electoral fraud when the trade liberalization dummy is (i) equal to 0 or (ii) equal to 1. The result from this latter exercise, which is illustrated in figure 3.12, further confirms that the effect of the new democracy dummy on the likelihood of electoral fraud is statistically insignificant when trade liberalization is not adopted. But, as predicted in hypothesis 2.3, the new democracy dummy has a statistically negative effect on change in the predicted probability of electoral fraud when trade liberalization is adopted.

∆probability electoral fraud

0.4

0.2

Trade reform = 1

0.0

-0.2

Trade reform = 0

Figure 3.11 Effect of new democracy on electoral fraud for each level of trade reform

∆probability electoral fraud

0.4

0.2

New democracy = 1 0.0

-0.2

New democracy = 0

Figure 3.12 Effect of trade reform on electoral fraud for each level of new democracy 

104 / Chapter Three

I find mixed empirical support for the remaining variables in the spec­ ification. For instance, log GDP per capita has the predicted negative sign and is significant in both the models in table 3.10. The variable ELF is in the predicted positive direction and is significant as well. Capital account openness is significant but (surprisingly) positive. The coefficient of economic crisis and bop crisis are each insignificant, while the estimate of polity is insignificant. However, the estimate of AGC index is negative and significant. The estimate of  the trade liberalization dummy for countries that are not new democracies in the sample is statistically insignificant. I conducted two robustness tests. For the first robustness test, I include the following additional controls in the specification, which may affect the likelihood of electoral fraud—­log population, a measure of de facto judicial independence (labeled DFJI), and the dummy variable IMF that is coded as 1 for countries observed under IMF programs.61 I also add the dummy variable election monitor, which is coded as 1 if the national election is monitored by international observers; it is coded as 0 otherwise. This variable is included in the specification as scholars debate whether or not the presence of “outside election monitors” (i.e., international observers) makes it difficult for incumbents to engage in electoral fraud (Hyde 2011; Kelley 2012; Simpser and Donno 2012). The data to code election monitor are drawn from the data on international electoral monitoring compiled by Kelley (2012).62 The main findings discussed above do not alter statistically or substantively after incorporating these additional controls in the panel-­probit model (see model 28, table 3.10). Log population, DFJI, and IMF are each in­­ significant, while election monitor is negative and significant at the 5% level. The effect of new democracy trade on electoral fraud remains negative and sig­ nificant in the Markov-­transition probit model (not reported to save space) that accounts more explicitly for temporal dependence in TSCS data sets when the dependent variable is dichotomous (Beck et al. 2002).

Conclusion The model in chapter 2 predicts that new democratic regimes have a negative effect on trade barriers (tariffs and N  T Bs) when the interindustry occupational mobility of workers is sufficiently high. The extensive statistical analysis presented earlier in this chapter statistically corroborates this claim. The results reported above also corroborate the causal argument that industry owners in new democracies where labor mobility is high tend to believe

Trade Protection and Electoral Fraud in New Democracies  /  105

that the campaign contributions that they offer will not influence economic policies (including trade policies) that are implemented by the government. Finally, there exists statistical support for the hypothesis that the adoption of trade liberalization in new democracies reduces the likelihood of electoral fraud. Two important substantive conclusions emerge from this chapter. First, much of the democracy and trade literature focuses exclusively on how the level of democracy affects trade protection in developing countries.63 The empirical analysis presented here, however, suggests that it is not the level of democracy per se but rather trade politics in new democracies during the democratic transition year and in the immediate posttransition years that influence trade protection in developing states. Moreover, unlike extant research, it is the interaction between the supply side (i.e., new democratic regimes) and the demand side (i.e., the trade policy preferences of workers, which are determined by their degree of interindustry occupational mobility) that has a statistically negative effect on both tariffs and NTBs in developing countries. Second, some analysts have recently claimed that the phenomenon of providing donations to political parties is rampant in new democratic regimes across the developing world and that this may hinder efficient policy making in these countries.64 There is no doubt that “special interests” provide contributions to political parties in developing countries that have recently experienced a transition to democracy. Yet the empirical evidence discussed in this chapter broadly supports the theoretical claim in the issue-­ area of trade policy that new democratic regimes have a negative effect on contributions offered to politicians (as perceived by industry executives). This arguably suggests that the phenomenon of providing contributions to politicians in new democracies may not be as rampant as assumed and that the impact of these contributions (when offered) on policy may potentially be weak. The link between campaign contributions from the owners of industries and trade policy in developing countries, including new democratic regimes in the developing world, requires more empirical research. The absence of any data on actual levels of campaign finance in most developing countries makes this a daunting task. But such a research agenda is worth pursuing. Indeed, in the analysis of especially the Brazil case study later in this book, I use some within-­country data to evaluate the claims about campaign contributions in new democracies. Before I turn to the case-­study analysis, it is worth reemphasizing here that most new democracies in the developing world evolve into full-­fledged

106 / Chapter Three

democracies or, in other words, into “newly consolidated democracies.” I also mentioned earlier that the degree of trade restrictions varies signifi­­ cantly across not only these newly consolidated democracies but also established democracies in the developing world. In the next chapter, I present a formal model that provides a theoretical framework that accounts for the variation in trade barriers across newly consolidated and established democracies.

Four

Political Particularism and Trade Policy in Developing Democracies

To develop a complete picture of democracy’s influence on trade policy in developing states, students of IPE also need to theoretically understand the dynamics of trade politics in established developing democracies1 and in new democracies across the developing world that have evolved into consolidated democracies.2 This is necessary because the level of trade protection varies significantly across both established developing democracies and developing states that have recently evolved into consolidated democracies (see figure 1.4). Indeed, theoretical analysis of the effect of democracy on trade policy in developing countries will be thorough and comprehensive only after one theoretically accounts for variation in trade barriers across established and recently consolidated developing country democracies. I therefore develop a model in this chapter that explores how certain electoral institutions influence trade protection and contributions (legal and illegal) by industry owners, given the relative factor mobility of labor and capital. Similar to my theoretical account of new democratic regimes and trade policy, I assume in the model presented in this chapter that the trade policy preferences of labor and capital are determined by their relative factor mobility. However, in this model, I study the impact of the level of political particularism3 on strategic interaction among labor, capital, and the political parties in the polity. I then analyze the effect of this interaction on trade policy and contributions in stable developing country democracies. The focus in this chapter on the link among particularism, factor mobility, and trade protection in developing democracies should not be taken to mean that other “supply-­side” (political) variables do not matter for trade policy in democracies. To be sure, a large literature has highlighted the effect of  veto players, partisanship, presidential systems, electoral rules, and the number of  parties

108 / Chapter Four

on trade barriers (Rogowski 1987; Nielson 2003; McGillivray 2004; Milner and Judkins 2004).4 I submit that these political variables matter for trade policy and thus explicitly control for these variables in the empirical analysis. Yet I focus on the link between particularism and trade protection partly because a growing body of research suggests that particularism affects fiscal policies and foreign direct investment in developing democracies (Shugart and Haggard 1997; Eaton 2002; Hallerberg and Marier 2004; Hicken and Simmons 2008; Garland and Biglaiser 2009). Note that some (albeit few) studies in IPE have assessed how political particularism affects trade policy.5 However, these studies focus on either a few advanced industrial democracies6 or just middle-­income democracies7 but not on the entire set of democratic developing countries as done here. Additionally, scholars that analyze the impact of particularism on trade barriers also solely focus on the supply side by analyzing how party discipline, for example, shapes the trade policy decisions of governments. But they pay less attention to the possibility that political particularism may interact with “demand-­side” factors, such as the preferences of societal groups, to influence trade protection. The model in this chapter goes substantially beyond these studies here by developing a more holistic theory that explores how demand-­side factors associated with factor mobility—­particularly the interindustry occupational mobility of workers and the asset specificity of industries owned by capital—­interact with the level of political particularism to influence tariffs, N TBs, and the financial payments provided by protectionist industry owners in developing democracies. This chapter is organized into three sections. The first section, “A Model of Particularism, Trade Policy, and ‘Contributions,’” is divided into three parts. The first part discusses the model’s players and economy and then provides a definition of political particularism. This definition is employed to formalize the main features of particularism in the model. The second part has a more technical focus on the economy, the players’ utility functions, the formalization of particularism, the sequence of moves, and the solution concept. The third part begins with a detailed discussion of the results that explain how higher levels of particularism in developing democracies—­that is, particularistic democracies—­affect tariffs and the payments provided by capital to politicians (legal contributions and illegal donations). It then presents how lower levels of particularism—­that is, in party-­centered democracies—­influence tariffs and financial payments provided by capital. The second section of this chapter, “Particularism and N TBs in Developing Democracies,” extends the model to study the effect of

Political Particularism and Trade Policy in Developing Democracies  /  109

political particularism on N TBs in established and “newly consolidated” developing democracies. The chapter concludes with a summary of the main findings and the substantive implications of these findings.

A Model of Particularism, Trade Policy, and “Contributions” This section is divided into three parts. The first section, “Defining Political Particularism,” provides a descriptive background of the two main features of particularism formalized in the model presented below. The second section, “The Economy, the Players, Particularism, and Sequence of Moves,” presents the details of the model. The third section, “Particularism, Equilibrium Trade Protection, and Contributions,” discusses the equilibrium result and comparative statics derived from the model. Defining Political Particularism The term political particularism has been interpreted in numerous ways in the comparative politics literature on party system organization in developing democracies.8 For the model described in the next part, I adopt a more general definition of particularism that is drawn from key studies of electoral institutions in developing country democracies by scholars such as Carey and Shugart (1995), Mainwaring and Scully (1995), Hix (2004), and Hicken (2006). These scholars suggest that particularism ranges from a low to a high level.9 At one end of this particularism continuum we find particularistic democracies, or in other words, countries that exhibit high levels of particularism. According to scholars, in particularistic democracies, “there are significant incentives for candidates to cultivate personal identification and support among the electorate and to compete with candidates from their own party in addition to candidates from other parties” (Hix 2004, 198).10 Party leaders in particularistic democracies thus find it difficult to exert control over rank-­and-­file party members, as these members (as well as the leaders) have incentives to cultivate their personal vote (Carey and Shugart 1995; Hix 2004; Hicken and Simmons 2008). Particularistic democracies are usually labeled as “candidate-­centered systems,”11 and researchers argue that numerous institutional factors and electoral rules influence the extent to which democracies are candidate centered.12 A key political feature that results from personal vote cultivation is that politicians in particularistic democracies tend to individually seek financial contributions from industry owners or other donors to finance

110 / Chapter Four

their election campaigns, and as such, this induces them to strongly account for the preferences of “special interests” (Chang 2005; Hicken 2006; Hicken and Simmons 2008). At the other end of the particularism continuum we find party-­centered systems—­democracies with low levels of  particularism—­where “parties pres­ ent lists of candidates, and voters cannot influence the order of the candidates on them. . . . [These] systems consequently allow party leaders to exert a high degree of control over their legislators. Without the strategic need to appeal directly to the electorate, candidates have no incentives to break ranks with the party line. In fact, an individual candidate has a positive incentive to go along with the party line to improve her position on the party list” (Hix 2004, 196). Party-­centered democracies in the model are thus conceptualized as countries where party leaders and members do not cultivate their personal vote (Hix 2004; Hicken 2006). This in turn reduces their incentives to individually seek contributions from potential donors to finance their election campaign (Hix 2004; Hicken and Simmons 2008). The model described below explicitly formalizes this first key feature of  particularism—­ namely, the degree to which politicians cultivate their personal vote that may drive them to seek financial support from capital. Apart from the feature described above, studies also suggest that party leaders in particularistic democracies are more likely to divert financial resources (including contributions) for personal gain—­that is, corruption (see, for example, Samuels 1995; Chang 2005; Hicken 2006, 2009; Hicken and Simmons 2008). Conversely, leaders in party-­centered democracies where the level of particularism is low tend to invest financial resources to build and sustain the party they lead.13 This second feature is also explicitly formalized in the model. I next turn to present this model. The Economy, the Players, Particularism, and Sequence of Moves The structure of the small open economy of a developing country democracy in the model presented in this part—­specifically the domestic price and supply of the imported good, the tariff level, tariff revenue, and the factors of production—­is exactly similar to the model in chapter 2. The description of the economy in the model is therefore not repeated here to save space. The societal actors in the model are labor and capital (industry owners). Their trade policy preference is determined by their relative factor mobility. mobile The factor mobility of labor and capital is given by the parameters θ l (for workers) and φ (for capital), which were defined in the previous chapter. Further, as discussed in chapter 2, occupationally mobile (occupationally

Political Particularism and Trade Policy in Developing Democracies  /  111

specific) workers favor trade liberalization, while owners who own asset-­ specific industries do not. The utility function of the workers in the model is given by equa­tion 2.1 and their income is defined in equation 2.3 in chapter 2. The workers constitute the electoral majority in the polity, and as voters, they choose to re­ elect or remove the incumbent from office after observing the tariff policy proposed by the incumbent. The utility function of the owners (i.e., capital) is given by equation 2.4 in the previous chapter, and the owners’ attempt to influence the level of trade protection by providing both legal contributions and illegal donations to politicians. There are two parties in the model, each with party leaders: the ruling party, or in other words, the government (denoted as A), and the opposition party (denoted as B). Leaders from party A propose tariff policy tA, while leaders from party B propose tariff policy tB. The share of votes for tariff policy tA against tariff policy tB is defined by equation 2.7 in chapter 2. The winning probability of tariff policy tA against tB is defined as equation 2.8. Thus the formal structure of electoral competition in this chapter’s model is similar to the model described in chapter 2. I next turn to describe the objective function of the party leaders. The objective function of the party leaders from the ruling and opposition parties in this chapter’s model are similar to the ruling and opposition party’s objective function in the previous chapter. There are, however, some key differences in the structure as well as critical parameters in the party leaders’ respective objective function in this chapter versus the parties’ objective function in chapter 2. As described below, these differences emerge from the formalization of the two key features of particularism in this chapter’s model (which was obviously not done in the model from the preceding chapter). To see this in more detail, first note that the party leaders from the ruling party (i.e., the government) in the model are concerned with political support garnered from the two productive groups in society: the owners (capital) and the workers (labor). However, leaders from each party are also interested to extract rent from tariffs and the contributions (legal and illegal donations) offered by the owners. Their objective is to therefore balance the desire for political support with the gains from rent extraction. I formalize this in the objective function of the leaders from both parties in two steps. First, I assume that in their objective function, the party leaders weigh the trade policy preferences of the owners (given by their utility function u c (t J ,φ )) by the parameter λ∈[0 ,1], while simultaneously weighing the preferences of workers (given by their utility function ul(p,Il )) via (1 − λ ). The

112 / Chapter Four

parameter λ formalizes the first key feature of political particularism discussed earlier—­the extent to which politicians (i.e., party leaders) cultivate their personal vote that influences them to seek financial support from capital. The greater (lesser) the party leaders cultivate their personal vote, the greater (lower) the value of λ in the [0,1] interval.14 Second, it was mentioned earlier that leaders in particularistic democracies (where the level of particularism is high) tend to divert financial resources such as contributions and illegal donations for personal use, while leaders in party-­centered democracies (where particularism is low) distribute a larger share of such resources to party members. I formalize this particular dynamic in the party leaders’ objective function by assuming that 1 the share of the owners’ total monetary payments (r (t J , φ , dJ) = φ tJ + φ d J) 2 that is diverted for personal gain is given by ∆ ∈[0,1]. The greater (lesser) the financial payments diverted for personal use, the higher (lower) the value of ∆ in the [0,1] interval.15 Based on the description in the preceding paragraphs, the party leaders from the ruling party have the following weighted objective function: (4.1)



uA = λ uc + (1 − λ) ul + ∆ (r (t j ,φ , dJ ) )  1 2  c l l = λ[u (t A , φ )] + (1 − λ ) [u ( p , I ) ] + ∆  φ  t A + d A   2   

By symmetry, the opposition party leaders’ objective function is (4.2)

 1  u B = λ[ u c (t B , φ)] + (1 − λ )[u l (p,I l ) ] + ∆ φ  t B2 + dB  2   

The respective party leaders’ goal is to optimally set their tariff policy, given by the pair , from their respective objective function to win the election. The payoff functions are therefore their respective probabilities of winning: D

(4.3)

Π A (t A , t B ) = π (t A , t B ) Π DB (t A , t B ) = 1 − π (t A , t B )

The sequence of moves in the model is as follows. Capital (i.e., the owners of industries) move first, deciding the optimal level of contributions that they will offer to leaders from the ruling and opposition parties to influence tariff policy. After receiving the contributions, party leaders choose to apportion some share of the contributions for personal use and the remainder to rank-­and-­file party members. The leaders from the ruling party and the opposition then propose their respective tariff policy in the context of electoral

Political Particularism and Trade Policy in Developing Democracies  /  113

competition. When doing so, they account for the trade policy preferences of owners and labor, which includes the median voter. The tariff policy proposed by the ruling party leaders determines domestic prices as well as the income of voters (which they observe). Once the ruling party chooses the tariff policy, the voters examine the utility that they obtain from the tariff policy. The voters, including the median voter, then choose to reelect or remove the ruling party and install the opposition in office. This depends on whether the observed tariff policy maximizes the voters’ utility. Their voting behavior determines the incumbent’s probability of reelection and his or her probability of losing office. I solve the model described by using the subgame-­perfect Nash equilibrium solution concept. The model’s results and substantive theoretical predictions are presented below, with illustrations and simulations derived from the model. Particularism, Equilibrium Trade Protection, and Contributions The mathematical derivation and characterization of the model’s equilibrium result is presented in the appendix of this book. The equilibrium and comparative static results are presented below in nontechnical language with the aid of illustrations that are derived from the model. The first part of the equilibrium result described in the appendix formally characterizes the following for developing democracies with low levels of particularism: (i) the optimal tariff policy set by the party leaders (tˆJ), (ii) the optimal tariff policy preference of the median voter (tˆ mv), and (iii) the optimal financial payments offered by the industry owners (r*(tˆJ , φ ,dJ)). The second part of the equilibrium result is stated as follows: Lemma 1 (a) A higher level of political particularism—­that is, in particularistic democracies—­induces party leaders to propose tariff policy in equilibrium that converges to the median voter’s optimal tariff policy preference. (b)  The tariff policy proposed by party leaders in equilibrium in particularistic democracies is biased toward the trade policy preference of labor (workers). Proof: See appendix.

The formal proofs of these two results are provided in the appendix. The theoretical rationale for the results in lemma 1 is as follows. To begin with, recall that the electoral calculus of  individual party members in democracies with high levels of particularism is dominated by the desire to cultivate

114 / Chapter Four

their personal vote (Carey and Shugart 1995; Shugart 2001; Hicken 2006). This implies that they focus more on “shoring up their own power than carrying out plans to broaden the party’s appeal” (Scarrow 2005: 18). The loyalty of individual party members to their party is suspect. There is thus little or no guarantee that rank-­and-­file members will help party leaders win the election by promoting the party’s agenda in the electorate (Carey 1998; Hix 2004; Hicken 2006). In the model, the leaders from each party in particularistic democracies (i.e., countries where the level of particularism is high) realize that they cannot rely on rank-­and-­file members to win the election, owing to the reasons summarized above.16 The model reveals that these office-­seeking leaders—­ who are interested in maximizing their vote share to win the election17—­ rationally respond to their inability to rely on rank-­and-­file party members by directly appealing to the median voter.18 Doing so allows them to maximize their own (and their party’s) electoral prospects without relying on individual party members to rally support for their respective party. More formally, in the model, the leaders from each party directly appeal to the median voter by locating the equilibrium tariff policy at the median voter’s optimal tariff preference.19 Note that when the ruling party leaders (for example) set the equilibrium tariff at the median voter’s optimal tariff preference, it maximizes the median voter’s utility.20 This induces the median voter to vote for the ruling party and consequently maximizes the likelihood of winning the election for leaders from the ruling party. It also minimizes threats to their political survival in office from the opposition, even though rank-­and-­file party members do not promote the ruling party’s agenda in the electorate. If, however, the ruling party fails to shift the equilibrium policy to the median voter’s optimal choice, then the median voter rationally switches his or her support to the leaders of the opposition party, and this increases the government’s probability of losing the election. Thus, as demonstrated formally in the appendix,21 when the level of particularism is high, ruling party leaders have no incentives to deviate from their best response of setting the tariff policy in equilibrium at the median voter’s optimal preference. Furthermore, the median voter has no incentive to deviate from his or her strategy of voting for the ruling party, given the latter’s best response as mentioned above.22 The political dynamics described above is illustrated in figure 4.1, which is derived from the model. The y-­axis in the right-­hand side of this figure indicates the party leaders’ optimal tariff policy choice. The x-­axis denotes the level of particularism given by λ . The figure indicates that, for instance,

Political Particularism and Trade Policy in Developing Democracies  /  115 tˆA = tˆmv

π (tˆA , t B ) = 1

Incumbent’s best response

Median voter’s best response

π (tˆA , t B ) [0,1]

0

λ

1

Particularism

Figure 4.1  Particularism and convergence to the median voter’s tariff policy preference

when the level of particularism λ becomes high, the ruling party leaders’ Nash response is to locate the equilibrium tariff policy (tˆJ) at the median voter’s optimal tariff policy preference (tˆ mv). The median voter’s best response to this Nash strategy is to vote for the ruling party leaders, since tˆJ = tˆ mv maximizes his or her utility. Thus when the level of particularism is high, the party leaders’ equilibrium tariff policy will converge to the median voter’s optimal tariff policy preference. Examples support the causal claims proposed above. For instance, Mainwaring (1995) states that party leaders in particularistic systems in Latin American countries such as Brazil “rely on direct linkages to the masses” by appealing to the “median citizen” and that “they are more likely than other to pursue policy measures with an eye toward publicity” (1995: 24). Such “publicity” is deemed necessary by party leaders to win the election, as they simply cannot depend on the possibility that rank-­and-­file party members will help them win the election (Mainwaring 1995; Weyland 2000). 1 out that in the Philippines’s Hutchcroft and Rocamora (2003) also point 23 particularistic democracy, Fidel Ramos did not rely on party members to mobilize electoral support for himself and the party. Instead he directly appealed to Filipino workers and the median voter to support his candidacy and “his” economic reform program while “blaming oligarchic groups for the country’s laggard economic status” (Hutchcroft and Rocamora 2003: 279). Similar to Ramos, “Estrada rode to overwhelming victory in 1998 with strong populist rhetoric and the enthusiastic support of millions of poor

116 / Chapter Four

Filipinos . . . It was, in the words of one Philippine political analyst, ‘the revenge of the masses’” (Hutchcroft and Rocamora 2003: 280). The first result in lemma 1 leads to the second result in the lemma. This second result posits that the equilibrium tariff policy will be biased toward the trade policy preferences of workers (labor) when the level of particularism is high. The causal logic that explains this result is straightforward. First, note that the median voter in a developing democracy is a worker. Since the equilibrium tariff policy converges to the median voter’s optimal tariff policy preference when particularism is high (see part [a] of lemma 1), it follows logically that the equilibrium tariff policy will be biased toward labor rather than capital. Second, in particularistic democracies, party leaders have political incentives to mobilize mass support and enhance their electoral prospects by directly soliciting the political support of the electoral majority—­the workers. Indeed, leaders in particularistic systems “base their rule on appeals to the mass of the populace”24 and stress mass-­based support in the electorate “as the only legitimate criterion for considering citizens’ interests” (Weyland 2000: 6). This is because they cannot, as mentioned earlier, rely on individual party members to mobilize mass support in this case. Because party leaders have incentives to solicit the support of labor for their political survival, they thus place greater weight on the trade policy preferences of workers when λ is high. This in turn encourages them to propose tariff policy in equilibrium that is biased toward the preference of workers. Given that labor’s trade policy preference is weighted more strongly by the party, the level of tariffs proposed by these leaders in a particularistic democracy will therefore be influenced by the level of interindustry occupational mobility of labor. The party leaders’ tactic of directly appealing to the electoral majority (the workers) has three main consequences for the politics of trade policy in particularistic systems. These consequences are summarized in the following comparative statics derived from the equilibrium result: Proposition 1 (a)  For higher levels of particularism, the party leaders optimally reduce tariffs when the degree of the interindustry occupational mobility of workers is sufficiently high. (b) The party leaders’ decision to optimally reduce tariffs under the political (higher particularism) and economic (sufficiently high interindustry labor mobility) conditions described above maximizes their probability of reelection.

Political Particularism and Trade Policy in Developing Democracies  /  117 (c)  Higher levels of political particularism reduce lobbying for protection by the owners of asset-­specific industries, even when the degree of labor mobility is sufficiently high. As a result, the financial payment (legal and illegal donations) offered by capital (the owners) to influ­ence trade policy decreases in equilibrium in particularistic democracies. Proof: See appendix.

The formal proofs of these comparative static results are provided in the appendix. The causal intuition that accounts for the comparative statics in part (a) and (b) of this proposition is as follows. First, when the interindustry occupational mobility of workers is sufficiently high, the median voter’s—­who is a worker—­level of occupational mobility is also high. As discussed in chapter 2, the occupationally mobile median voter supports a reduction in trade barriers, since his or her net gain from more trade openness is positive. In equilibrium, the party leaders, including the incumbent, observe in this case that the occupationally mobile median voter is more likely to support trade liberalization and a reduction in tariffs. Because high levels of particularism induce party leaders in particularistic democracies to locate equilibrium tariff policy at the median voter’s optimal tariff choice (as shown in lemma 1), the incumbent will therefore optimally respond to the mobile-­type median voter’s trade policy preference by reducing tariffs. Second, the model shows that when the median voter’s level of interindustry occupational mobility is sufficiently high, the reelection concerns of party leaders in particularistic democracies become more pronounced than their incentives for rent extraction from tariffs. In fact, in particularistic democracies, the party leaders sacrifice the opportunity to extract rent by not raising tariffs in this case; instead, their dominant strategy is to reduce tariffs when labor mobility is high, as they rationally expect that doing so will induce the occupationally mobile median voter to vote for them. Observe that the strategy of reducing tariffs pays off, as it helps the ruling party leaders, for example, to secure the electoral support of the occupationally median voter. The “mobile-­type” median voter in fact optimally chooses to vote for the incumbent in equilibrium when the government cuts tariffs. This allows the incumbent and therefore the ruling party to maximize its vote share and thus increase the party’s likelihood of winning the election, as stated in part (b) of proposition 1. Moreover, neither the ruling party leaders nor the median voter have incentives to deviate from their respective equilibrium strategy mentioned above.

118 / Chapter Four

10

Tariff (%)

8 6 4 2

2

10

4 20

6

Particularism

8

30 10

Labor mobility

40

Figure 4.2.  Particularism, labor mobility, and trade protection

A simulation derived from the model illustrates the comparative static result in part (a) of proposition 1 (see figure 4.2). The hyperplane in this figure indicates that as the level of particularism (λ) combined with the degree of the workers’ interindustry occupational mobility reaches a high level, the optimal tariff set by the incumbent in equilibrium decreases sharply. Hence this figure shows that the effect of higher levels of particularism on tariffs in developing democracies is negative, conditional on interindustry labor mobility being sufficiently high. The model further suggests that the political incentives for party leaders (including leaders from the ruling party) to1reduce tariffs when labor mobility is high also induces them to successfully implement trade reforms, even though there is no guarantee ex ante that rank-­and-­file party members will support trade liberalization policies on the legislative floor when labor mobility is reasonably high. Indeed, party leaders that operate in particularistic institutions can (and do) adopt at least three different strategies that facilitate implementation of  trade reforms when workers favor higher levels of trade openness.25 First, ruling party leaders may employ executive decrees instead of attempting to secure support from rank-­and-­file party members to implement trade reforms, as they may be unable to pressure these members to approve the reform agenda. This is common in particularistic systems in the developing world, as trade policy is often designed and implemented at the executive level rather than by the legislature (see, for example, Power 1998). Case studies of structural reforms by leaders in particularistic systems across

Political Particularism and Trade Policy in Developing Democracies  /  119

Latin America reveal that these leaders “indeed enacted crucial parts of their neoliberal agenda,”26 especially trade liberalization measures, “via decree” (Ferreira-­Rubio and Goretti 1998; Power 1998; Weyland 2000).27 Ruling party leaders in particularistic democracies may use a second strategy to pass trade reforms, especially when the majority in the electorate (i.e., workers) supports trade liberalization (which occurs when labor mobility is sufficiently high). This alternative strategy involves the use of “rhetorical attacks” to denounce rank-­and-­file members as “defenders of special interests” if they oppose the incumbent’s initiative to restrict trade restrictions (Durand 1998; Weyland 2000, 2002). Leaders in candidate-­centered systems have incentives to initiate such attacks, since they do not have sufficient control over party members and thus cannot pressure party members to support their liberalization effort. These attacks may affect rank-­and-­file members’ reputation among workers (the majority in the electorate) if they resist reforms, as workers favor trade liberalization when labor mobility is fairly high.28 This in turn may dampen the members’ incentives to oppose reforms, since these costs could hurt their electoral prospects. Denouncing party members who resist trade liberalization in a preemptive manner may thus provide party leaders in particularistic systems with more maneuvering room to facilitate implementation of trade reforms when labor mobility is high.29 As a third strategy, party leaders in particularistic systems may form trade policy coalitions with individual legislators from various parties in the legislative floor. I argue that ruling party leaders in particularistic systems may form such coalitions precisely because they lack sufficient leverage over their party members. Insufficient leverage over party members may make leaders uncertain ex ante about whether or not these members will support their reform effort. Forming trade policy coalitions may therefore provide political insurance against such uncertainty and help the ruling party leaders garner sufficient legislative support to pass trade reforms when labor mobility is sufficiently high. Note that all the three strategies discussed above that leaders in particularistic democracies can adopt to obtain legislative support for their proposed trade reform policies are “publicly” observable by voters (including workers) in the polity. This allows party leaders in particularistic democracies to credibly signal to voters and thus workers (when labor mobility is sufficiently high) that they have committed themselves to liberalize trade policies. In short, the comparative statics in parts (a) and (b) in proposition 1 and the preceding discussion leads to the following prediction: Hypothesis 4.1. Incumbents in developing democracies that are character­ ized by higher levels of particularism (i.e., particularistic democracies) will

120 / Chapter Four reduce tariffs when the degree of interindustry labor mobility is sufficiently high.

I next turn to explain the third result (part [c]) in proposition 1, which predicts that the financial payments (legal contributions and illegal donations) offered by industry owners decreases in particularistic democracies when the degree of labor mobility is sufficiently high. The intuition underlying this result is twofold. First, given that a higher level of particularism drives party leaders to favor labor and thus appeal to the median voter, there will be less need for them to obtain the political support of the owners of industries that prefer more trade protection when designing trade policy. In fact, as party leaders rely more on directly soliciting the support of the median voter to obtain electoral dividends, they will be more reluctant to associate themselves with “special interests,” including the owners. In equilibrium, the owners will observe the party leaders favoring the interests of labor by shifting tariff policy toward the median voter’s optimal tariff choice. As a result, the owners will learn ex ante in particularistic systems that their financial leverage may fail to elicit a favorable trade policy response (i.e., higher protection) from party leaders (including the incumbent), even after providing payments to politicians. This dissuades them from providing contributions to the political candidates to influence tariff policy in such systems. Second, party leaders in democracies with higher levels of particularism (i.e., candidate-­centered systems) cannot credibly commit themselves ex ante to pursue the trade policy interests of the owners, because of their political incentive to favor labor over the owners. Further, Weyland (2000, 2002) suggests that leaders in particularistic developing democracies will maintain their distance from many organized business groups, as personalistic leaders perceive such groups as rent-­seeking special interests who seek to interfere with the market and their political autonomy. Because party leaders (including the incumbent) in particularistic systems maintain a safe distance from special interests and cannot credibly commit to pursue the interests of the owners, the owners have no incentives to lobby the candidates to influence tariff policy. Indeed, the aforementioned commitment problem also generates uncertainty ex ante among the owners about their ability to influence trade policy ex post, even after providing contributions to the political candidates. This makes it risky ex ante for the owners to offer contributions to the candidates. Hence the owners will rationally respond to the commitment problem discussed above by reducing their lobbying effort and the contributions that they provide to the candidates in systems

Political Particularism and Trade Policy in Developing Democracies  /  121

Equilibrium Payments

0 .5

1 Asset-specificity

Figure 4.3.  Equilibrium payments for different levels of l as f increases

that are more candidate centered. Moreover, as demonstrated formally in the appendix30 and as illustrated in figure 3.3, the contributions from the owners will decline in particularistic democracies (i.e., l = 1), even when the degree of asset specificity of their respective industries ( φ ) is high. A good example of  the phenomenon described above is President Alberto Fujimori in candidate-­centered Peru. During his reign in office, “Fujimori kept domestically oriented, protectionist sectors and associations at bay” (Weyland 2000: 14). Similarly, Prime Minister Chuan Leepkai in Thailand also kept a safe distance away from owners of industries that opposed trade reforms (Phongpaichit and Baker 1996). In Brazil, President Fernando Collor de Mello built up an economic team that “offered internal opposition to the established leadership of the powerful FIESP (Federação das Indústrias do Estado de São Paulo), which encompassed many import-­substituting industrial sectors that were therefore afraid of . . . neoliberalism” (Weyland 2000: 12).31 Lastly, in the Philippines, “[President Fidel V.] Ramos and his chief theoretician, former general Jose Almonte, blamed oligarchic groups for the country’s laggard economic status and combined measures of economic liberalization, privatization, and infrastructure development with 1 concerted attacks on ‘cartels and monopolies.’ . . . They asserted the need to build a more capable state and free the state of oligarchic influence” (Hutchcroft and Rocamora 2003: 279–­80). Put together, the result in part (c) of proposition 2 and the preceding examples lead to the following hypothesis: Hypothesis 4.2. Campaign contributions provided by the owners to influence trade policy in developing democracies where the level of particularism is

122 / Chapter Four high (i.e., in particularistic democracies) decrease when labor mobility is reasonably high.

The comparative static results presented so far have focused on the interactive effect of high levels of particularism and high interindustry labor mobility on tariffs implemented by party leaders and financial payments (legal contributions and illegal donations) offered by industry owners in developing democracies. What happens to the politics of trade policy and the optimal level of payments offered by the owners when the level of particularism is low? Comparative statics from the model provide the following result that addresses this question. Proposition 2 When the level of political particularism in developing democracies is low—­ that is, the system becomes more party centered—­financial payments (contributions and illegal donations) provided by the owners (capital) increases and consequently trade barriers are increased as well. Proof: See appendix.

The formal proof of this result is provided in the appendix. The causal logic that explains the results in proposition 2 is as follows. To start with, recall that developing democracies with low levels of particularism are party-­centered systems that have “centralized” parties (Carey 2003; Hix 2004; Hicken 2006). Party leaders in these centralized party systems32 often control the rank of candidates on their party’s electoral list (Carey 2003; Hix 2004).33 This allows the leaders to occupy the top ranks of their party’s electoral list by virtue of nominating themselves to these top positions (Carey 2003; Carey and Reynolds 2007). Occupying the top ranks of their party’s electoral list reduces the leaderships’ susceptibility to electoral punishment by voters in party-­centered democracies. At the same time, it places the burden (and the political costs) of electoral defeat solely on the shoulders of rank-­and-­file party members (Carey 2003; Carey and Reynolds 2007). This is emphasized by Carey (2003) who points out that in party-­centered systems, the party leaders’ access to power and perks increases dramatically, but their electoral vulnerability decreases in a corresponding manner because leaders occupy the top positions on party electoral lists. The leadership’s susceptibility to electoral punishment is mitigated, even if the party as a whole loses electoral ground. Therefore, the leaders who stand to gain the most from violating

Political Particularism and Trade Policy in Developing Democracies  /  123 public trust and pillaging state resources stand to suffer the least electorally if their party is punished. Rank-­and-­file politicians, whose heads are the first to roll in any partisan electoral setback, might object to being relegated to the marginal list positions that buffer their leaders, but would-­be rebels . . . are simply removed from the lists or demoted to perilous or even hopeless list positions by the leadership. (2003: 195)

Additionally, leaders in countries where the level of particularism is low (i.e., party-­centered systems) realize that they can protect themselves from electoral punishment if they occupy the top ranks of their party’s electoral list (Shugart 2001; Carey 2003). This realization provides them with strong incentives “to campaign for a higher position on a party list rather than seek the approbation of voters” (Shugart 2001: 176). Their incentive to invest substantial effort toward campaigning within their party for a higher position on the party’s electoral list ensures that leaders in party-­centered systems “are not really responsive to constituents at all, but rather to an internal constituency within their own party” (Shugart 2001: 176). This tends to “break the connection” between voters and party leaders in systems with centralized parties (Coppedge 1994; Shugart 2001: 176; Carey 2003). As shown in the model, the fact that party leaders in systems where the level of particularism is low can insulate themselves from punishment by voters drives them to shift trade policy away from the interests of the median voter (and thus workers) and toward the capital owners’ preference.34 This is because the decreased vulnerability of these leaders to electoral punishment reduces their incentives to be accountable to large, broad constituencies such as workers (Shugart 2001; Carey and Reynolds 2007). It also makes it difficult for voters, including workers, in party-­centered democracies “to monitor and punish” party leaders for poor responsiveness to the voters’ interest (Shugart 2001: 177). Low electoral accountability thus makes it easier for leaders in centralized party systems to adopt policies that are not consistent with the preferences of a broad group of citizens, such as workers. As stated by Carey and Reynolds (2007), in party-­centered developing democracies, “parties may be strong internally but be vacuous and fickle when it comes to policy content. When this is the case, parties fail to deliver programs that respond to citizen preferences in the manner depicted by the strong party ideal, and do not advance the cause of accountable government” (Carey and Reynolds 2007: 271).35 Another adverse consequence of decreased susceptibility to electoral punishment is that it makes it easier for party leaders to engage in rent seeking in party-­centered systems. This implies—­as predicted by the

124 / Chapter Four

model—­that leaders in democracies where the level of  particularism is low will have greater incentives to implement higher trade barriers. Doing so will help them maximize the rents that they can extract from trade protection. More important, since these party leaders will be interested to extract rent from trade barriers, it will encourage them to shift trade policy toward the preference of capital owners, as the owners favor more trade restrictions. Lastly, note that leaders of political parties in party-­centered systems in the developing world tend to invest significant financial resources to not only build and sustain their respective party but also campaign for a higher position on their party’s electoral list (Kavanagh 1995; Farrell 1996). They have incentives to invest resources for such intraparty campaigns because occupying the top ranks of their party’s electoral list (as explained earlier) reduces their susceptibility to punishment by voters. But their decision to invest resources for this purpose also raises the following question: Where do party leaders obtain the monetary payments to finance resources that are employed for intraparty campaigns? The model suggests that leaders in party-­centered developing democracies will strategically opt to finance resources for intraparty campaigns by soliciting and obtaining monetary contributions from the capital owners. Note that the leaders can successfully extract contributions from the owners only if they implement tariff policies that are compatible with the owners’ trade policy preferences. The owners will have little or no incentives to provide contributions if the leaders propose trade policies that favor the interests of labor. Therefore, to obtain contributions and donations from the protectionist industry owners, party leaders in party-­centered democracies will increase tariffs. Doing so is compatible with the owners’ interests. This leads to a symbiotic relationship in which capital owners increase the financial support that they provide to party leaders in party-­centered systems and the leaders return this favor by not only implementing but also maintaining high trade barriers that suit the owners’ protectionist interests. I empirically evaluate the hypotheses and claims that emerge from propositions 1 and 2 in chapters 5, 6, and 7. At this stage, I turn to analyze the consequences that particularism has on NTBs in developing democracies.

Particularism and NTBs in Developing Democracies The model in this chapter has shown that the degree of particularism combines with the level of labor mobility to influence tariffs and financial

Political Particularism and Trade Policy in Developing Democracies  /  125

payments offered by capital to politicians in developing democracies. But can differences in particularism explain variation in N TBs across developing democracies as well? To answer this question, I adopt the formal model of political particularism and trade policy that was described in the preceding sections of this chapter. However, as done in chapter 2, I also extend the formal model presented here in three main ways to more closely examine the effect of particularism on N TBs in developing democracies. I simply list these three extensions below, as these extensions were described in detail in chapter 2. The technical details and proofs from the model on particularism and N TBs are presented in the appendix. The three extensions to the model in this chapter are as follows. First, I assume that in addition to tariffs tJ , the party leaders can use N T Bs, denoted as tJ ,n[ J∈{A,B}], to protect domestic industries from import competition. The N TBs that they propose affect the domestic price of goods, which influences the income of workers and the returns from capital invested by the owners.36 Second, I assume that the owners of industries attempt to influence the level of N T Bs by lobbying for more protection.37 To this end, they provide campaign contributions and illegal donations to the political candidates. Their incentive to do so is conditional on the degree of asset specificity of the industries that they own. Third, after observing the N T Bs proposed by the party leaders, the voters (including the median voter) evaluate the effect that the proposed N TBs have on their utility. They then choose to reelect the incumbent or install the opposition into office. However, as mentioned in chapter 2, voters in the model are uncertain to some extent about N TBs in that they know that N TBs are being proposed and implemented but are not completely certain about the exact level of N TBs being implemented. This is because certain types of N TBs are less transparent than other more observable N TBs, which makes it difficult to gauge the precise level of N TBs. I incorporate this uncertainty by assuming formally that the exact level of NTBs proposed by the parties, as perceived by the voters, is uniformly distributed over a finite interval—­that is, tJ,n ∼ U [t n ,t n ] and tJ ,n∈ℜ+. After extending the model along the lines described above, I solve for and formally characterize the equilibrium level of N TBs. The formal derivation of this equilibrium result is described in the appendix. This equilibrium result for N TBs leads to the following lemma: Lemma 2 When the level of political particularism is high in developing democra­ cies—­that is, in particularistic democracies—­the equilibrium policy on N TBs

126 / Chapter Four proposed by the party leaders converges to the median voter’s optimal preference for N TBs and is biased toward the trade policy preferences of  labor. Proof: See appendix.

The proof of lemma 2 is provided in the book’s appendix. This lemma is consistent with the equilibrium result in lemma 1 on political particularism and tariff barriers in developing democracies, as the causal mechanisms that generate the results in lemmas 1 and 2 are similar. I briefly discuss below two explanations for the result in lemma 2. First, as mentioned earlier, party leaders in democracies characterized by high levels of particularism can no longer rely on individual party members to build the party’s support base in the electorate or to increase party leaders’ electoral prospects, as party members primarily have incentives to cultivate their personal vote. This encourages leaders from the political parties to prioritize their reelection concerns over and above their incentives to extract rent from N TBs. Second, it also discourages party leaders, including the incumbent, from exploiting the uncertainty that the median voter and other voters have about the exact level of N TBs being proposed. In other words, politicians in particularistic democracies do not exploit the median voter’s uncertainty by (i) shifting the N TBs policy away from the median voter’s optimal choice and (ii) providing biased information about N TBs to voters. Instead, the party leaders (including the incumbent) directly appeal to the median voter by locating the equilibrium N TBs policy at the median voter’s optimal choice for N TBs. Locating the equilibrium N TBs at the median voter’s optimal N TBs choice maximizes the median voter’s utility. This induces the median voter to reelect the incumbent. Neither the incumbent nor the median voter has incentives to deviate from their respective strategies in equilibrium as described above. Additionally, as mentioned in lemma 2, convergence to the median voter’s optimal choice ensures that the equilibrium policy on N TBs is biased toward the preferences of workers, since the median voter is a worker in the developing democracy. Two comparative static results are derived from the N TBs version of the model: Proposition 3 (a)  When the level of political particularism is high in developing democracies (which implies particularistic democracies), the equilibrium policy on N TBs depends on the interindustry occupational mobility of workers, including the median voter. Higher levels of particularism there-

Political Particularism and Trade Policy in Developing Democracies  /  127 fore have a negative impact on N TBs when the degree of the interindustry occupational mobility of workers is sufficiently high. (b)  When the level of particularism is low in developing democracies (which implies party-­centered democracies), the equilibrium policy on NTBs is biased toward the trade policy preferences of the owners. Proof: See appendix.

Similar to proposition 1, I find that the level of N TBs proposed by the party leaders critically depends on the interindustry occupational mobility of workers, including the median voter. If the workers’ and the median voter’s level of occupational mobility is sufficiently high, then the median voter will favor a reduction in N TBs for reasons discussed earlier. As predicted by the model, the party leaders, including the incumbent, will rationally respond to the “mobile-­type” median voter’s preference elucidated earlier by lowering N TBs. Doing so reduces nominal price rigidities, which generates real income gains for the median voter in this case.38 Lowering N TBs also reduces protectionist rents that can be extracted by party leaders. Thus when the occupationally mobile median voter observes that the government reduces N TBs, he or she supports the incumbent and votes for the government in equilibrium. I show in the appendix that this maximizes the reelection probability of the ruling party leaders39 and induces them to not deviate from the optimal strategy of reducing N TBs, even though a lower level of N TBs reduces the protectionist rents that they can extract. In contrast, I find that when the level of particularism is low, the equilibrium level of N TBs proposed by the party leaders is biased toward the optimal level of N TBs preferred by the owners (capital). The rationale for this outcome lies in the fact that in developing democracies that are party centered, party leaders have incentives to solicit contributions and “donations” from the industry owners to expand their financial resources. The reasons underlying this claim were discussed earlier. In equilibrium, the owners recognize that the party leaders are rationally interested to solicit contributions and donations (legal or illegal). This encourages the owners, particularly the owners of industries with higher levels of asset specificity, to offer more financial payments to the party leaders. Indeed, the demand for more contributions by the leaders opens a window of opportunity for the owners to influence trade policy. Providing more contributions also increases the bargaining leverage of the owners vis-­à-­vis the candidates. This allows them to credibly threaten to not provide contributions to the party leaders if the leaders choose to ignore the trade policy interests of the owners. Since party leaders in systems that are more party

128 / Chapter Four

centered are interested in extracting significant contributions and donations, and because the owners can credibly threaten to curtail these contributions, the leaders respond to the owners’ contributions by ensuring that the equilibrium N TBs policy is biased toward the preferences of capital.

Conclusion The model in this chapter explored how political particularism combines with relative factor mobility of labor and capital to influence trade barriers and contributions (legal as well as illegal) in established and consolidated developing democracies. One learns from the model that higher levels of particularism drive party leaders to strategically locate the equilibrium policy for tariffs and N TBs at the median voter’s optimal trade policy preference. As a result, their trade policy decisions are therefore critically determined by the extent of the interindustry occupational mobility of workers, since the median voter in a developing democracy is, by definition, a worker. Hence when the interindustry occupational mobility of workers (including the median voter) is sufficiently high, higher levels of particularism induce the party leaders to reduce tariffs and N TBs. This discourages the owners from providing contributions and donations to politicians in particularistic democracies, even when the asset specificity of their industries is high. Finally, the model suggests that the politics of trade policy in party-­centered democracies (i.e., democracies that exhibit low levels of particularism) are in fact exactly the opposite of particularistic democracies in the developing world. Put together, the model in this chapter generates testable hypotheses and causal claims that that are evaluated in the later chapters of this book. The theoretical results reported here indicate that the impact of political particularism on trade protection in developing democracies is more nuanced than the claims suggested in extant studies that posit a positive correlation between particularism and trade protection in advanced industrial or middle-­income democracies (e.g., McGillivray 1997, 2004; Nielson 2003; Park and Jensen 2007). Whether or not the predictions from the model discussed above are empirically valid is carefully evaluated in chapters 5, 6, and 7. To this end, the next chapter tests the hypotheses derived from the model on particularism, trade protection, and payments provided by capital (legal and illegal donations) in developing democracies.

Five

Empirical Tests for Political Particularism, Trade Protection, and Contributions

The formal model in chapter 4 examined how the political dynamics generated by the level of political particularism in newly consolidated and established developing country democracies (hereafter developing democracies) affect strategic interaction among party leaders and the two productive groups in society—­labor and capital. Building on this, the model analyzes the impact of this political interaction on trade barriers and campaign contributions. Recall that the first hypothesis (hypothesis 4.1) derived from this model posits that incumbents in developing democracies that exhibit higher levels of particularism will reduce tariffs and N T Bs when the level of interindustry occupational mobility of workers is sufficiently high. The second hypothesis (hypothesis 4.2) predicts that the amount of “legal” campaign contributions provided by protectionist industry owners to party leaders in particularistic democracies will decrease when labor mobility in these countries is high. In this chapter, I statistically test these two hypotheses on a comprehensive time-­series-­cross-­sectional (TSCS) sample of democracies from the developing world. The rest of this chapter is organized into two sections. In section one, “Particularism, Tariffs, and N T Bs,” I provide a detailed description of the sample, the statistical methodology, and the operationalization of the dependent variable, independent variable, as well as control variables that are employed for testing hypothesis 4.1. I then present the statistical results from this sample. In section two, “Particularism, Labor Mobility, and Contributions,” it is first acknowledged that it is impossible to directly test hypothesis 4.2, as there is no publicly available cross-­national or pooled data on the amount of campaign contributions offered by industry owners to politicians in developing democracies. I therefore analyze the World Economic Forum’s (WEF’s) firm-­level survey-­response data set. This data set

130 / Chapter Five

allows me to statistically evaluate the causal claim that industry owners in particularistic democracies—­where labor mobility is sufficiently high—­are more likely to believe that the campaign contributions that they offer will not (because of the reasons discussed in chapter 4) influence the economic (including trade) policies implemented by the incumbent. The chapter ends with a conclusion in which I summarize the results and the main contributions to the literature utilized by the chapter’s empirical analysis.

Particularism, Tariffs, and NTBs Sample, Dependent Variable, and Statistical Methodology A sample of democratic country-­years from the developing world is needed to test hypothesis 4.1, as this hypothesis focuses on developing country democracies. I compile a TSCS sample of ninety-­one developing countries that are observed as democracies—­according to the Przeworski et al. (2000) criteria (described below)—­during the 1972 to 2008 period to test this hypothesis. The democracies in the sample satisfy Przeworski et al.’s (2000) criteria for a democracy, which are (i) the chief executive and legislature must be directly elected, (ii) there must be more than one party in the legislature, and (iii) incumbents must allow a lawful alternation of office if defeated in elections. The list of ninety-­one developing countries in the sample that are observed as democracies and the years in which they are observed as democracies are listed in table 5.1. The size and temporal range of the sample is comprehensive, as it includes all democracies in the developing world observed during the 1972 to 2008 period (based on the Przeworski et al. criteria) for which data to operationalize the dependent and independent variables (described below) are available. The results reported below remain robust if countries are coded as democracies when their Polity score is greater than or equal to +6 in a –­10 (full autocracy) to +10 (full democracy) scale.1 To conserve space, however, I focus on reporting the results obtained from the country-­year sample of developing democracies that satisfy Przeworski et al.’s (2000) criteria for a democracy. There are two dependent variables of interest in hypothesis 4.1: tariffs and N T Bs. The main indicator used to measure the first dependent variable (tariffs) is the import duty coverage ratio, which is labeled as import duties and is expressed as a percentage. The operationalization of this variable and the data sources from which I derived import duties were described in chap­­ ter 3. I also use statutory tariff as an alternative measure of trade protection

Table 5.1  List of  democratic country-­years in sample Country

Period

Country

Period

Country

Period

Albania Argentina

Fiji Georgia

2003–­2008 2003–­2008

Niger Nigeria

Armenia

1992–­2008 1973–­1975, 1983–­2008 1990–­2008

Ghana

Bahamas Bangladesh

1972–­2008 1991–­2008

Grenada Guatemala

Barbados

1972–­2008

Guinea Bissau

Belarus Belize Benin Bolivia

1990 1972–­2008 1991–­2008 1982–­2008

Guyana Haiti Honduras Hungary

Botswana Brazil Bulgaria Central African Republic Chile

2003–­2008 1985–­2008 1990–­2008 1993–­2008

India Indonesia Jamaica Kenya

1979–­1980, Pakistan 1996–­2008 1972–­2005 Panama 1972–­1973, Papua 1986–­2008 New Guinea 1972–­1973, Paraguay 2000–­2008 1992–­2008 Peru 1994–­2008 Poland 1982–­2008 Romania 1990–­2008 Russian Federation 1972–­2008 Senegal 1999–­2008 Sierra Leone 1972–­2008 Slovak Republic 2002–­2008 Slovenia

1993–­2008 1979–­1982, 1999–­2008 1972–­1976, 1988–­1998 1989–­2008 1972–­2008

2000–­2008 1996–­2008 1990–­2008 1990–­2008

1972, 1990–­2008 1972–­2008 1972–­1974, 1990–­1994, 2003–­2008 1992–­1996 2003–­2008

Korea, South

1988–­2008

South Africa

1994–­2008

Latvia Lebanon

1990–­2008 1972–­1974, 2003–­2008

Sri Lanka Sudan

1972–­2008 1986–­2008

Lesotho Lithuania

1993–­2008 1990–­2008

Sudan Suriname

1986–­1988 1972–­2008

1972–­2008

Madagascar

1993–­2008

Thailand

Malawi Mali

1994–­2008 1992–­2008

Cyprus

2000–­2008 1990, 1999–­2008 1983–­2008

Malta

1972–­2008

Tonga Trinidad and Tobago Turkey

1975, 1983–­2008 2003–­2008 1972–­2008

Czech Republic Djibouti Dominica

1990–­2008 1972–­1976 1972–­2008

Mauritius Mexico Mongolia

1972–­2008 2000–­2008 1992–­2008

Uganda Ukraine Uruguay

Dominican Republic Ecuador El Salvador

1978–­2008

Moldova

Vanuatu

1979–­2008 1984–­2008

Namibia Nepal

Venezuela Zambia

1972–­2008 1991–­2008

Estonia

1990–­2008

Nicaragua

1990, 1996–­2008 1990–­2008 1991–­2001, 2008 1990–­2008

1973–­1979, 1983–­2008 1980–­1984 1991–­2008 1972, 1985–­2008 1972–­2008

Colombia Comoros

Congo Congo, Democratic Republic Costa Rica Cote d’Ivoire Croatia

2003–­2008 1980–­2008 1989–­2008 1990–­2008 1990–­2008

Note: The time period in the columns indicate the years in which each country is observed as a democracy according to the Przeworski et al. (2000) criteria for a democratic regime.

132 / Chapter Five

for robustness tests; the data sources for this measure are provided in chapter 3. In the sample of developing country democracies, the mean and standard deviation of import duties are, respectively, 10.2 and 16.2, and the mean and standard deviation of statutory tariff are, respectively, 15.1 and 12.6. Thus there exists substantial variation in each of these two dependent variables in the sample. The second dependent variable of interest in hypothesis 4.1 is N T Bs. As described in chapter 4, I operationalize N T Bs as a single category core NTB coverage ratio (denoted as core N T Bs) for each democratic country-­year, which is measured by calculating the percentage of imports subjected to core N T Bs.2 The sources from which I gathered data on N T Bs for developing democracies are largely available from 1990 onward (and these data sources were listed in chapter 3).3 As a result, the size of the sample decreases considerably when I use core N T Bs for the empirical analysis. Yet the mean and standard deviation of core N T Bs (operationalized as a coverage ratio) are 0.22 and 0.19, respectively. With respect to statistical methodology, I test hypothesis 4.1 by estimating TSCS regression models with panel-­corrected standard errors (PCSEs) that are adjusted to correct for heteroscedasticity and contemporaneous correlation (Beck and Katz 1995). The models in which the elasticity mea­ sure of labor mobility (ILM-­elasticity) is included as one of the independent variables are, however, estimated with bootstrapped standard errors.4 I also include country-­specific fixed effects in each empirical model to account for country-­specific heterogeneity. I initially included dummies for each year in the specification but dropped these, as F-­tests indicate that the temporal dummies are jointly insignificant. To correct for serial correlation, I also include the lag of the relevant dependent variable in the empirical models (Beck and Katz 1995). I conduct a battery of robustness tests and diagnostic checks as well. The results from these robustness tests are reported below. Independent and Control Variables for Testing Hypothesis 4.1 The prediction in hypothesis 4.1 posits an interactive effect since it suggests that leaders in developing democracies that are characterized by higher levels of particularism will reduce tariffs when the interindustry occupational mobility of workers is reasonably high. I therefore need to interact two variables to test this hypothesis: an index for political (i.e., electoral) particularism for developing democracies and a continuous measure that operationalizes the interindustry occupational mobility of  workers for each developing democratic country-­year. I describe below the operationalization

Empirical Tests for Particularism, Trade Protection, and Contributions  /  133

of the index of electoral particularism. I then briefly define the measures of interindustry labor mobility, as these measures were described in chapter 3. I rely on the data and coding scheme developed by Wallack et al. (2003) and Johnson and Wallack (2006) to operationalize the index of political particularism. Based on Carey and Shugart’s (1995) definition of electoral particularism, the Wallack et al. (2003) index of particularism operationalizes the extent to which the political setting places a premium on cultivating a personal vote by politicians and is thus more candidate centered. The index has three components: ballot, pool, and vote. Each component is coded on a 0, 1, or 2 scale, where higher values denote greater incentives to cultivate a personal vote and therefore higher levels of particularism. More specifically, ballot describes the ease with which someone could get his or her name on the ballot in a position that makes winning a seat likely. Closed-­list systems where parties determine the candidates as well as their order on the ballot make this access difficult and are therefore scored as 0.5 Systems where party nominations are required for a viable candidacy but where voters can determine the order of candidates on the party’s list are scored as 1. Electoral systems where party nomination is not required for a successful campaign thus making access the easiest are scored as 2. The variable pool measures the extent to which a candidate can benefit from the votes of other candidates from his or her own party. The assumption here is that candidates who do not expect to receive “spillover” votes from copartisans will try harder to build personal reputations. A score of 0 means that “a vote for any candidate of a given party is counted first as a vote for the whole party list for the purpose of determining how many seats are to be allocated to the list” (Carey and Shugart 1995: 421). A score of 1 indicates that votes are pooled across candidates or factions in the same party, and systems where votes accrue only to individual candidates and no pooling occurs at all are scored as 2. Lastly, vote measures whether voters cast votes primarily for candidates or parties. Systems where voters can only choose among parties are scored as 0. Systems where voters can express preferences for multiple candidates within party lists, across parties, or through a two-­stage election (i.e., primaries or run-­offs) are scored as 1. Finally, systems where voters cast only one vote, for either a candidate or a party faction, are scored as 2.6 Wallack et al. (2003) and Johnson and Wallack (2006) average the scores of these three variables to create a summary index of particularism for several country-­years.7 In unicameral systems, this summary index corresponds to the whole legislature in bicameral systems to each house and in mixed systems to each subset of legislators. Each house is given a weight

134 / Chapter Five 30

Percent of Total

25

20

15

10

5

0 0

2

4

6

8

10

12

Particularism index

Figure 5.1  Frequency distribution of particularism across developing democracies 

of 0.5, regardless of the relative numbers of seats. Within each house, each group of legislators chosen under similar rules is given a weight according to its proportion of total legislators in that 1 house. The Johnson and Wallack (2006) index of particularism ranges from 0 (low particularism) to 13 (high particularism). The frequency histogram in figure 5.1 indicates that there is substantial variation in this particularism measure across the developing democracies in the sample. This provides one with ample leverage to test hypothesis 4.1. In addition to particularism, I require a measure for the interindustry occupational mobility of workers in developing democracies to test the prediction in hypothesis 4.1. From chapter 3, I use two measures that op­ erationalize the level of interindustry occupational mobility of workers for each democratic country-­year in the developing world that are included in the sample. The first of these two measures is the Wacziarg and Wallack (2004) measure of the interindustry occupational mobility of workers, which is labeled as labor mobility. I operationalize this labor mobility measure for the developing democracies in the data across the thirty-­two industries listed in table 3.3 via the procedure and the relevant formula (see equa­tion 3.1) employed to calculate this measure, which was described in chapter 4.

Empirical Tests for Particularism, Trade Protection, and Contributions  /  135

Figure 5.2 illustrates the distribution of  the labor mobility measure for two sets of developing democracies in the data. The first set include particularistic (i.e., candidate-­centered) developing democracies in which the Johnson and Wallack (2006) particularism score is 7 or higher on the 0–­13 scale. The second set includes developing democracies with low levels of particularism (i.e., party-­centered democracies) that have a particularism score of 6 or lower. This figure reveals that the distribution of labor mobility is similar across these two set of countries. It also shows that the distribution of labor mobility within each of these two groups resembles a log-­normal distribution. Descriptive statistics reveal that the mean of the labor mobility measure in the developing countries sample is 16.05, the standard deviation is 8.91, and the minimum and maximum values are –­47.5 and 104.3, respectively. This further confirms that there is substantial variation for this measure in the developing countries sample. The second measure is the elasticity measure of the interindustry occupational mobility of workers, which is denoted as ILM-­elasticity. I operationalize the ILM-­elasticity measure for the developing democracies in the data by using the procedure described in chapter 4. This procedure includes the estimation of equation 3.2 that allows one to derive the elasticity mea­ sure of interindustry labor mobility for each democratic country-­year in the

0.15

High particularism Low particularism

Density

0.10

0.05

0.00 -40

-20

0

20

40

60

80

100

Labor mobility

Figure 5.2  Distribution of  labor mobility across developing democracies

136 / Chapter Five

sample; this elasticity measure is given by the estimated formula of βˆt, which is defined in equation 3.3 in chapter 3.8 Since hypothesis 4.1 predicts that the effect of particularism on trade protection in developing democracies is conditional on the degree of the interindustry occupational mobility of workers, I first interact the particularism variable with labor mobility. I then introduce particularism × labor mobility, as well as the individual components of this interaction term in separate empirical models, where the dependent variable is one of the following three measures: import duties, statutory tariffs, or core N T Bs. I repeat this exercise by including the interaction term particularism × ILM-­elasticity and the individual components of this interaction term in a separate model to test the robustness of my results. Based on the prediction in hypothesis 4.1, I expect that the coefficient of particularism × labor mobility and particularism × ILM-­ elasticity will be negative in the specification. Numerous economic and political control variables are included in the models. The economic controls incorporated in the empirical models are common in the IPE literature on trade protection and are also similar to the controls that were included in the empirical models discussed in chapter 3. Hence, for the sake of  brevity, I merely list these control variables below. The economic controls in the specification where the tariff measures serve as the dependent variable are as follows: log GDP per capita, log population, economic crisis, balance-­of-­payments (bop) crisis, capital account openness, log GDP, AGC index, log capital-­labor ratio, and the GATT/WTO dummy. In the models in which core N T Bs is the dependent variable, I include all the economic control variables but add real effective exchange rate (REER) and exclude log population and capital account openness. Four political control variables are included in the specification. First, I follow Milner and Judkins (2004) and include the presidential democracy dummy in the specification for countries with a presidential system. I also include government partisanship that is coded on a 0–­2 scale, with 0 representing right governments and 2 left governments. Data for these three variables are from the World Bank’s Database of  Political Institutions (DPI; 2010). Milner and Judkins (2004) claim that the presidential system is more favorable to free trade, while left governments are more protectionist. Presidential democracy is expected to be negative, but government partisanship will be positive. I mentioned in the previous chapter that the literature is divided over the issue of whether or not more veto players in government leads to higher trade protection (see, for example, Henisz and Mansfield 2006; Milner et al. 2007; Mukherjee et al. 2009).

Empirical Tests for Particularism, Trade Protection, and Contributions  /  137

Veto players is thus added to the specification using data from the World Bank’s (2010) DPI. I include a dummy for countries in the sample that use the proportional representation (PR) electoral rule in the specification, as scholars have hypothesized that democracies that employ the PR rule are more open to free trade (Rogowski 1987, 1995; Nielson 2003). Last but not the least, numerous researchers suggest that district magnitude may affect the level of trade protection in democracies (Rogowski 1989; Hankla 2006). I therefore control for average district magnitude in the models, taken from the World Bank’s (2008) DPI. The descriptive statistics for the main trade protection dependent variables as well as the independent and control variables that are described above are provided in table B in the book’s online appendix.9 Preliminary Analysis for Particularism and Tariffs I discuss some examples derived from the developing-­country democracies data as well as examine some relevant correlations in this data before discussing the results from the estimated statistical models. To start, consider, for instance, two democracies from Latin America: Costa Rica and Ecuador. These two Latin American democracies share some economic similarities: they are both middle-­income developing countries, and the share of the manufacturing and service sector (as a percentage of GDP) is similar in these countries.10 Notwithstanding these similarities, import duties have decreased sharply in candidate-­centered (and thus highly particularistic) Ecuador—­which employs the open-­list PR electoral system—­as labor mobility increased over time in the country (see figure 5.3). In sharp contrast, import duties have not appreciably decreased and have largely remained at the same level over time in Costa Rica, even though labor mobility has increased in the country’s economy during the last 2–­3 decades (figure 5.4). These two examples illustrate—­if not empirically establish—­the theoretical prediction that we are likely to observe a negative correlation between labor mobility and tariffs in candidate-­centered but not party-­centered developing democracies in the developing world. Additionally, a bivariate comparison of the level of the main measure of labor mobility in forty-­one candidate-­centered (i.e., highly particularistic) democracies listed and import duties in these candidate-­centered democracies indicates that high labor mobility is indeed associated with a dramatic decrease in tariffs in these countries. This is best evidenced in fig­­ ure B in the book’s online appendix, where one can observe that for both

138 / Chapter Five 75

Import duties (%)

60

60 Open-list PR

Democratic Transition

40

45

20

30

0

15

-20

Import duties Labor Mobility

0

-40 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

Figure 5.3  Labor mobility and import duties in candidate-­centered Ecuador

75

60 Closed-list PR

45

20

30

15

0

Import duties

Labor Mobility

40

Import duties (%)

60

-20

Labor Mobility 0

-40 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

Figure 5.4  Labor mobility and1import duties in party-­centered Costa Rica

a linear regression line and a loss smoothing line, higher labor mobility (x-­axis) is negatively associated with statutory tariffs (y-­axis) in the forty-­ one candidate-­centered democracies.11 A Spearman’s rho test indicates that this negative association is both strong (–­0.65) and statistically significant (p < 0.01). However, one finds that the association between higher labor mobility and import duties in thirty-­seven party-­centered democracies (i.e., democracies that have low levels of particularism) is substantively negligible (almost 0) and statistically insignificant.12 These simple results are encouraging but not sufficient. I therefore turn to report the results from the estimated statistical models.

1

Empirical Tests for Particularism, Trade Protection, and Contributions  /  139

Results for Particularism and Trade Barriers The models in table 5.2 present the results from some baseline specifications in which I include the main interaction term of interest (e.g., particular­­ ism × labor mobility), the individual components of this interaction term, log GDP per capita, and the lag of the dependent variable. The estimated effect of particularism × labor mobility on each of the following dependent variables is negative and highly significant in the baseline specification: import duties (model 1), statutory tariff (model 2), and core N T Bs (model 3).13 Likewise, in the stripped-­down models, the statistical impact of particularism × ILM-­ elasticity on import duties (model 4), statutory tariff (model 5), and core N T Bs (model 6) has the predicted negative sign and is highly significant.14 These baseline specification results are thus encouraging. Models 7–­9 in table 5.3 report the estimates from fully specified models (which include all the controls) that are used for testing hypothesis 4.1. These models are each estimated with a lagged dependent variable, PCSEs, and fixed effects. The coefficient of the interaction term particularism × labor mobility is negative and statistically significant at the 1% level in the models in which the dependent variable is import duties (model 7), statutory tariffs (model 8), or core N T Bs (model 9). This statistically corroborates the prediction in hypothesis 4.1. With respect to the individual components of this interaction term, one finds that the estimate of the individual labor mobility measure and the particularism variable is statistically insignificant. It is thus the interaction of the two independent variables—­rather than each variable individually—­that has a statistically negative effect on trade protection in developing democracies. To gain a better sense of how particularism × labor mobility influences import duties, I derive and analyze the substantive effect of this interaction term from model 1 in two steps. First, I plot in figure 5.5 the effect of particularism on the level of import duties across the observed range of labor mobility in the developing democracies sample. The solid line in this figure plots the marginal effect of particularism on import duties at different levels of labor  mo­­ bility, while the dashed lines in the figure indicate the 95% confidence intervals. The figure shows that particularism does not have a significant impact on import duties for low levels of labor mobility. However, once the labor mobility measure increases well above its mean of 16.05 in the sample, the negative effect of particularism on import duties becomes statistically signifi­ cant. Second, I find from the estimate of particularism × labor mobility that particularism reduces import duties by approximately 12.1% when its com­ panion variable labor mobility is increased by one standard deviation above

1.721*** (0.106) 1367

–­0.41*** (0.083)

0.891*** (0.166) –­3.73 (2.64) –­0.343 (0.412) –­0.139 (0.255)

1.508*** (0.102) 1367

–­0.422*** (0.095)

0.903*** (0.183) –­3.65 (3.05) –­0.357 (0.422) –­0.112 (0.347)

Model 2

Statutory tariff

1.43*** (0.118) 512

–­0.265*** (0.054)

0.823*** (0.156) –­4.58* (2.71) –­0.385 (0.456) –­0.095 (0.103)

Model 3

Core NTBs

–­0.372*** (0.099) 1.394*** (0.269) 1254

–­0.375 (0.701)

0.416*** (0.081) –­2.04 (1.83) –­0.603 (0.51)

Model 4

Import duties

–­0.324*** (0.085) 1.455*** (0.323) 1254

–­0.214 (0.429)

0.46*** (0.064) –­2.68 (1.29) –­0.524 (0.57)

Model 5

Statutory tariff

–­0.202*** (0.073) 1.616*** (0.178) 491

–­0.383 (0.506)

0.319** (0.09) –­3.19 (2.77) –­0.624 (0.59)

Model 6

Core NTBs

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. PCSEs are reported in parentheses for all models. Each model in the table is estimated with country-­specific fixed effects that are not reported to save space.

N

Constant

particularism × ILM-­elasticity

particularism × labor mobility

ILM-­elasticity

labor mobility

particularism

log GDP per capita

lag dependent variable

Model 1

Import duties

Table 5.2  Particularism, labor mobility, and trade restrictions: Baseline specifications

REER

capital acct openness

economic crisis

bop crisis

GATT/WTO

particularism × ILM-­elasticity

particularism × labor mobility

ILM-­elasticity

labor mobility

particularism

log GDP per capita

lag dependent variable

–­1.66*** (0.607) 0.083 (0.517) 0.66 (0.949) 0.001 (0.32)

–­0.384*** (0.075)

0.755*** (0.053) –­3.19 (1.95) –­0.262 (0.309) –­0.116 (0.214)

Model 7

Import duties

Statutory tariff

–­1.64*** (0.606) 0.18 (0.535) 1.29 (0.939) –­0.015 (0.32)

–­0.319*** (0.064)

0.759*** (0.053) –­2.65 (2.13) –­0.249 (0.353) –­0.099 (0.247)

Model 8

Table 5.3  Main results for particularism and trade barriers

–­0.046 (0.23)

–­1.66*** (0.575) –­0.008 (0.503) 0.732 (0.953)

–­0.197*** (0.061)

0.758*** (0.054) –­3.61* (2.11) –­0.124 (0.156) –­0.078 (0.092)

Model 9

Core NTBs

–­0.184*** (0.047) –­0.063 (0.092) –­0.028 (0.051) 0.086 (0.09) 0.018 (0.056)

–­0.094 (0.103)

0.324*** (0.064) –­1.07 (2.15) –­0.427 (0.548)

Model 10

Import duties

–­0.145*** (0.056) –­0.065 (0.094) –­0.039 (0.054) 0.077 (0.065) 0.021 (0.065)

–­0.085 (0.064)

0.317*** (0.065) –­1.22 (1.41) –­0.453 (0.625)

Model 11

Statutory tariff

–­0.035 (0.041)

–­0.109*** (0.037) 0.045 (0.096) 0.073 (0.082) –­0.033 (0.03)

–­.059 (0.075)

0.161*** (0.049) –­1.97 (1.7) –­0.341 (0.297)

Model 12

Core NTBs

Model 8 4.45 (10.5) 0.06 (0.103) 0.02 (0.224) 1.27 (0.996) –­0.18 (0.259) 0.662 (0.925) –­12.88 (14.0) 1.072** (0.04) 1123

4.63 (10.54) 0.088 (0.111) 0.078 (0.216) 1.89** (0.79) –­0.165 (0.259) 0.774 (0.899) –­15.38** (7.0) 1.08*** (0.03) 1123

Statutory tariff

Model 7

Import duties

29.37* (17.2) 1.052** (0.026) 401

8.16 (10.97) 0.149 (0.115) 0.083 (0.215) 2.06** (0.801) –­0.16 (0.256)

Model 9

Core NTBs

2.082 (3.079) 0.118 (0.177) 0.074 (0.149) 1.08 (0.709) –­0.138 (0.144) 0.504 (0.681) –­3.22 (4.16) –­1.231*** (0.488) 1077

Model 10

Import duties

3.077 (3.096) 0.105 (0.15) 0.081 (0.152) 1.73 (0.953) –­0.123 (0.184) 0.617 (0.723) –­9.26 (12.01) –­1.146*** (0.344) 1077

Model 11

Statutory tariff

4.11 (3.65) –­1.714*** (0.254) 380

1.091 (0.983) 0.093 (0.298) 0.045 (0.087) 0.836 (0.74) 0.051 (0.077)

Model 12

Core NTBs

Note: ***, **, and * denotes significance at the 1%, 5%, and 10% level, respectively. PCSEs are reported in parentheses for all models. Each model in the table is estimated with country-­specific fixed effects that are not reported to save space.

N

Constant

PR

log population

veto players

presidential

partisanship

log capital labor ratio

AGC index

Table 5.3  (continued )

Marginal effect of particularism on import duties

Empirical Tests for Particularism, Trade Protection, and Contributions  /  143 20

0

-20

-40 -50

0

50 Labor Mobility

Marginal effect

100

95% Confidence Interval

Marginal effect of particularism on core NTBs

Figure 5.5 Effect of  particularism × labor mobility on import duties

10

0

-10

-20 -50

0

50 Labor Mobility

Marginal effect

100

95% Confidence Interval

Figure 5.6 Effect of particularism × labor mobility on core NTBs 1

its mean while holding other variables in the specification at their respective mean. This effect is statistically significant at the 95% confidence level. The marginal effect of  particularism on statutory tariffs across the observed range of labor mobility in the sample is similar to the substantive effect described above. To analyze the marginal effect of particularism × labor mobility on core N T Bs, I first plot, with 95% confidence intervals, the marginal effect of particularism on core N T Bs across the observed range of labor mobility in the sample (see figure 5.6). The figure shows that particularism does not have 1

144 / Chapter Five

a significant impact on core N T Bs for low levels of labor mobility. However, once the labor mobility measure increases substantially above its mean in the sample, the negative effect of particularism on core N T Bs becomes statistically significant at the 5% level. One also learns from this marginal effect that particularism reduces core N T Bs by 9.2% when its companion variable labor mobility is increased by one standard deviation above its mean while other variables in the specification are held at their respective mean in the sample. This substantive effect is significant at the 95% confidence level. Thus the substantive effects illustrated in the figures strongly support the prediction posited in hypothesis 4.1. As an initial test of robustness, I replaced the labor mobility measure with the elasticity measure of labor mobility, which is labeled as ILM-­elasticity. Specifically, I dropped labor mobility from the model and instead incor­­ porated new democracy × ILM-­elasticity as well as the individual compo­­ nents of this interaction term in the specification. The effect of new democ­ racy × ILM-­elasticity on import duties (model 10, table 5.3), statutory tariffs (model 11, table 5.3), and core N T Bs (model 12, table 5.3) is negative and highly significant. The results for the different control variables in the empirical models in table 5.3 vary significantly. For instance, the coefficients for the GATT/WTO dummy, AGC index, IMF program, capital account openness, economic crisis, and log GDP are each consistently insignificant in the models where tariffs (import duties and statutory tariffs) is the dependent variable. These economic controls are also insignificant in the specifications in which core N T Bs is the dependent variable.15 The estimate of log GDP per capita is positive and significant in all the models, indicating quite surprisingly that across developing democracies, higher per capita income is associated with more trade restrictions. The estimate of log capital-­labor ratio is negative and significant in some specifications but statistically insignificant in the other models. Likewise, the coefficient for bop crisis is significant in some (but not all) the models in table 5.3, even though the sign of this estimated coefficient is inconsistent. The estimated results for the political controls vary as well across the model. For example, the coefficients of  average district magnitude, veto players, and partisanship are consistently insignificant. The estimate of  the presidential democracy dummy is positive and significant in the models in which tariffs is the dependent variable but not in the models where N T Bs is the depen­ dent variable. The lag of the dependent variable is positive and significant at the 1% level in the models in which it is included, which is perhaps not

Empirical Tests for Particularism, Trade Protection, and Contributions  /  145

surprising. The results reported in the models in table 5.3 also pass all diagnostic checks.16 Robustness Tests I conduct a battery of robustness tests and diagnostic checks. I first evaluate whether the main results hold after including the following additional controls to the specification: effective number of legislative parties (enlp) and a dummy variable for federal democracies ( federal) in the sample. Taagepera and Shugart’s (1989) measure of effective number of legislative parties (enlp) is included, since Nielson (2003: 475) claims that greater party fragmentation in the legislature leads to increased trade protection. The federal dummy is added to the specification, as Milner and Judkins (2004) find that federal democracies tend to maintain lower trade barriers. The estimated effect of  particularism × labor mobility is negative and significant at the 5% level in the augmented empirical specification (with the additional controls) in which the dependent variables are (i) import duties (model 7, table 5.4), (ii) core N T Bs (model 8, table 5.4), and (iii) statutory tariffs (not reported in table 5.4 to save space).17 Some more controls are added to the specification. These controls include the percentage of the population over age fifteen that has completed secondary-­level education for each country as a proxy for educational attainment (education), welfare spending (% GDP), and a linear time trend. I do not report the results obtained after including these additional controls to save space, but the main results presented in this chapter do not alter. As done in the previous chapter, I implemented Hurlin and Venet’s (2003) granger causality test for panel data to assess the potential endogenous relationship between each of the dependent variables (import duties, statutory tariff, and core N T Bs) and each of the three independent variables (particularism, labor mobility, and ILM-­elasticity). F-­statistics from the Hurlin and Venet (2003) tests indicate that there is no endogenous relationship between each of the dependent variables and each of the independent variables listed above. To be safe, however, I further address the possibility of endogeneity by testing hypothesis 4.1 via a “system–­generalized method of moments” (“system-­GMM”) model that combines a regression in first differences and a regression in levels; estimating the two equations (levels and differences) in a single system leads to consistent and efficient estimates (Blundell and Bond 1998). The system-­GMM model was described in chapter 3. The

REER

capital acct openness

econ crisis

bop crisis

GATT/WTO

particularism × labor mobility

labor mobility

particularism

log GDP per capita

spatial AR

lag dependent variable

Table 5.4  Robustness tests

–­2.50 (3.44) –­0.223 (0.317) –­0.065 (0.106) –­0.289*** (0.073) –­2.4*** (0.805) 0.774 (0.899) 0.802 (1.1) –­0.287 (0.451) –­0.103 (0.274)

–­2.014 (3.71) –­0.104 (0.153) –­0.041 (0.132) –­0.099*** (0.031) –­2.23*** (0.812) 0.662 (0.925) 1.94* (1.0)

0.734*** (0.073)

Model 8

Model 7

0.728*** (0.075)

Core NTBs

Import duties

xtpcse models

0.156*** (0.042) –­0.19 (0.256) –­0.07 (0.116) –­0.224*** (0.06) –­0.022 (0.016) 0.008 (0.011) 0.071*** (0.021) –­0.11 (0.505)

0.671*** (0.051)

Model 9

Import duties

–­0.0 (0.003)

0.138*** (0.046) –­0.096 (0.167) –­0.037 (0.124) –­0.087*** (0.023) –­0.001 (0.015) 0.023 (0.019) 0.112*** (0.02)

0.533*** (0.06)

Model 10

Core NTBs

System-­GMM models

0.472*** (0.065) 0.058 (0.163) –­2.50 (3.44) –­0.188 (0.254) –­0.075 (0.131) –­0.211*** (0.058) –­0.012 (0.019) 0.018 (0.025) 0.06** (0.03) 0.109 (0.203)

Model 11

Import duties

Spatial-­AR error model

–­2.014 (3.71) –­0.151 (0.228) –­0.089 (0.166) –­0.198*** (0.064) –­0.01 (0.015) 0.022 (0.019) 0.082** (0.043) 0.102 (0.106)

0.409*** (0.061)

Model 12

Import duties

Heckman outcome equation

Constant

federal

enlp

veto players

PR

partisanship

log population

Presidential

log capital labor ratio

AGC index

1036

3.612 (10.075) –­0.079 (0.178) 0.539 (0.828) 0.353 (0.263) 0.008 (0.011) 0.071*** (0.021) –­0.127 (0.186) –­0.0 (0.001) 1.43*** (0.279) 1.22* (0.077)

343

0.023* (0.013) 0.112*** (0.02) –­0.149 (0.19) –­0.0 (0.001) 1.69*** (0.308) 1.98** (0.866)

7.53 (16.33) 0.004 (0.187) 0.636 (0.654)

–­2.93** –­0.411 1036

0.18 (0.611)

1.43*** (0.279) 0.0 (0.0) 0.055 (0.032) 0.381 (0.306) 0.009** (0.004) 0.168** (0.084) –­0.12 (0.14) –­0.0 (0.0) 1.04** (0.512) 2.55** (0.923)

0.157 (0.7) –­1.89** –­0.34 343

0.008** (0.004) 0.131 (0.077) –­0.118 (0.136) –­0.0 (0.002) 1.29** (0.308) 1.1* (0.055)

1.69*** (0.308) 0.0 (0.003) 0.254 (0.431)

952

–­901.16

1.91 (0.249) 0.0 (0.0) 0.611 (0.224) 0.293 (0.748) 0.009 (0.01) 0.122 (0.164) –­0.115 (0.17) –­0.0 (0.0) 2.04 (1.91) 1.4** (0.062)

952

1.53 (0.388) 0.0 (0.003) 0.354 (0.402) 0.25 (0.622) 0.002 (0.007) 0.101 (0.097) –­0.109 (0.298) –­0.0 (0.001) 1.93 (1.382) 2.45** (1.37) –­0.049 (0.064) –­1549.22

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. PCSEs are reported in parentheses for all models. Each model in the table is estimated with country-­specific fixed effects that are not reported to save space.

AR(1) AR(2) N

Log likelihood Hansen J test (p-­value)

l

˘

148 / Chapter Five

results indicate that the interactive effect of particularism and labor mobility is negative and highly significant in the system-­GMM model in which the dependent variable is import duties (model 9, table 5.4), core N T Bs (model 10, table 5.4),18 and statutory tariffs (not reported to save space).19 It is also plausible that the trade policies adopted by democratic governments in the developing world may be influenced by international diffusion. Indeed, Milner and Kubota (2005) and Guisinger (2001) suggest that governments may be driven to reduce trade barriers when they observe geographically neighboring countries—­that compete for the same pool of international capital—­decrease their level of trade protection. This implies that the level of trade protection may be influenced by geographic proximity and thus characterized by spatial dependence, since governments may choose tariff rates based on the trade policy choices of neighboring countries. Diagnostic tests reveal that none of the three measures of trade protection that we employ (import duties, statutory tariff, and core N T Bs) suffer from spatial dependence.20 Yet, to be cautious, I tested the impact of particularism × labor mobility on each of the three aforementioned measures of trade protection in a spatial-­autoregressive error model (hereafter spatial-­AR error model) that accounts for spatial dependence in the data in the estimation process.21 The estimated coefficient of particularism × labor mobility is negative and statistically significant in the spatial-­AR error model when the dependent variable is import duties (model 11, table 5.4) or statutory tariffs and core N T Bs (not reported because of space constraints).22 Finally, I check whether the results remain robust after accounting for a potential selection bias problem that may occur from estimating the impact of an institutional variable like particularism on trade barriers in developing democracies. After all, it is plausible that factors that influence a nonrandom phenomenon such as the choice of electoral systems that exhibit a higher (or lower) degree of particularism by political elites in developing democracies may also potentially affect their optimal choice of trade policy. If one does not account for the nonrandom occurrence of electoral particularism in developing democracies when testing the interactive effect of particularism and labor mobility on trade barriers, then the obtained results may suffer from selection bias. To address this potential selection bias problem, I estimate a Heckman selection model. I first converted the particularism measure into a binary variable labeled as particularistic system. Recall that the ordinal particularism index ranges from 1 to 13, where higher scores on this index (more specifically, a score of 7 or higher) operationalize the presence of  particularistic democracies, while a score of 6 or lower operationalizes party-­centered

Empirical Tests for Particularism, Trade Protection, and Contributions  /  149

systems (see Johnson and Wallack 2005; Wallack et al. 2003). The particularistic system binary variable is thus coded as 1 for countries in the sample whose score on the 1–­13 particularism scale is equal to or greater than 7; it is coded as 0 otherwise. The Heckman selection model results reported below remain robust if  particularistic system is coded as 1 for countries whose score on the 1–­13 particularism scale is equal to or greater than 5 or 6. From the preceding information, it follows that the binary dependent variable in the first stage selection equation of the Heckman model is the particularistic system dummy, while the continuous dependent variable in the second stage outcome equation is a continuous measure of trade protection (e.g., import duties) described earlier. In the selection equation of the Heckman model in this case, I thus explicitly account for the factors that influence the likelihood that developing democracies in the sample have a particularistic system. To this end, I turn to some extant theoretical studies that have suggested a litany of factors that influence when democratic states are more (or less) likely to adopt a particularistic electoral system.23 Some of these studies suggest that countries with higher (lower) income24 and democracies that use the closed-­list PR electoral rule  25 are less likely to have particularistic systems. Hence I include log GDP per capita and a dummy variable for developing democracies that have a closed-­list PR system26 in the selection equation. Conversely, scholars suggest that the following electoral institutions generate particularistic democracies: developing country democracies that use the single-­transferable vote system (labeled as stv) and single-­member district plurality systems with open ballots (denoted as smd ballot).27 I thus control for the aforementioned variables in the selection equation as well. Some researchers hypothesize that developing democracies that have experienced a civil war are less likely to have particularistic systems;28 thus I include the lag of the civil war dummy in the selection equation, which is coded as 1 when countries in the sample suffer from a civil war and is coded as 0 otherwise. Lastly, I add the lag of the binary particularistic system mea­ sure in the selection equation. The second stage of the Heckman model is the outcome equation in which I estimate the effect of the particularism × labor mobility, the constitutive components of this interaction term, and the control variables listed earlier on the continuous dependent variable, import duties. The effect of particularism × labor mobility in the outcome equation of the Heckman model is negative and highly significant when the dependent variable is import duties (model 12, table 5.4) or statutory tariff, and core N T Bs (not reported to save space). The results from the selection equation of the Heckman model in

150 / Chapter Five

which the binary particularistic system dummy is the dependent variable are reported in table E in the book’s online appendix.

Particularism, Labor Mobility, and Contributions I focus on evaluating the main causal claim (labeled as claim 2) that directly lead to hypothesis 4.2, since data for the actual amount of campaign contributions provided by industry owners is not available for developing country democracies. This causal claim posits that industry owners in countries with higher levels of particularism will be more likely to believe that the campaign contributions that they offer will not influence policies (e.g., trade policies) when the degree of  labor mobility is sufficiently high. This claim is evaluated using the World Economic Forum’s (WEF 2004b) survey-­response data that tracks the extent to which industry executives in countries believe that the legal donations (i.e., campaign finance) they provide to politicians in parties during elections influences policies (including trade policies). As described in chapter 3, this data has been gathered for 2001, 2002, 2003, 2004, and 2006 by the WEF from the response of executives from 18 different industries in 104 countries to the following question: ·· Question A. To what extent do the legal donations [you provide] to political parties have a direct influence on policies?

Responses to this survey question are placed on an ordinal scale that ranges from a score of 1 (“there is a direct relation between donations and policies”) to 7 (“donations have no impact of on influencing policies”). The highest response score of 7 on this ordinal scale thus implies that industry executives believe that the legal campaign contributions that they provide have no impact on the policies designed and implemented by the incumbent. The WEF put together the data on responses to question A from industry executives for each of the 104 countries. It then aggregated the survey-­ response data for each country and reported for each country-­year (2001, 2002, 2003, 2004, 2006) a score on the 1 to 7 ordinal scale listed above. I estimate a set of statistical models to evaluate claim 2 by using the WEF’s survey-­response data described above. As done in the previous chapter, I first exclude the twenty-­two advanced industrial Organisation for Economic Co-­ operation and Development (OECD) countries from this survey-­response data set, as these countries are not relevant for the analysis. This led to a total of eighty-­two developing countries. After doing so, I used the Przeworski

Empirical Tests for Particularism, Trade Protection, and Contributions  /  151 Table 5.5  Developing democracies in WEF’s Executive Opinion Survey data set Argentina, Bangladesh, Bolivia, Bosnia-­Herzegovina, Botswana, Brazil, Bulgaria, Chile, Colombia, Costa-­Rica, Czech Republic, Dominican Republic, Ecuador, El Salvador, Estonia, Georgia, Ghana, Guatemala, Honduras, Hungary, India, Indonesia, Israel, Jamaica, Kenya, Latvia, Lithuania, Macedonia, Madagascar, Malawi, Malta, Mauritius, Mexico, Namibia, Nicaragua, Nigeria, Panama, Paraguay, Peru, Philippines, Poland, Romania, Slovak Republic, Slovenia, South Africa, South Korea, Sri Lanka, Taiwan, Tanzania, Thailand, Trinidad and Tobago, Turkey, Uganda, Ukraine, Uruguay, Zambia

et al. (2000) criteria for democracy (described earlier) to identify the states that are democracies in this group of eighty-­two developing countries. This exercise led to a total of fifty-­six democracies in the set of eighty-­two developing states. These fifty-­six developing country democracies are listed in table 5.5. Next, I use the WEF’s reported country-­year 1–­7 ordinal response score to question 1 for these fifty-­six developing democracies as the depen­ dent variable for testing claim 1. This ordinal dependent variable is labeled as contributions influence, and as described in chapter 3, it ranges on a scale of 1 to 7. I employ an ordered probit (OP) model to test this claim, as contributions influence is an ordinal variable. The following two independent variables need to be interacted to test causal claim 2: an index of electoral particularism and a continuous mea­ sure for interindustry labor mobility. For the second independent variable listed above, I employ the measure of labor mobility in equation 4.1 (see chapter 4) that operationalizes the interindustry occupational mobility of workers. For the first independent variable, I employ Wallack et al.’s (2003) and Johnson and Wallack’s (2006) index of political particularism that was described earlier. To test claim 2, I interact particularism with labor mobility and introduce the particularism × labor mobility interaction term and the individual components of this interaction term in the OP specification. Based on this claim, I anticipate that the estimated coefficient of this interaction term will be positive in the OP specification. Finally, as discussed in chapter 3, I follow Kaufman et al.’s (2008) study and control for the following variables in the OP specification: log GDP per capita, a 0–­1 index of government fractionalization drawn from the World Bank’s Database of Political Institutions (DPI; 2012), and a 0–­100 index of media freedom that is taken from the Freedom House (2012). The results from the OP specification are reported in models 13 (OP model estimated without random effects) and 14 (estimated with random effects) in table 5.6. The estimate of the particularism × labor mobility in models 13 and 14 is positive and statistically significant at the 5% level. This

152 / Chapter Five

statistically corroborates the prediction in claim 2. The control variable log GDP per capita is negative but weakly significant at the 10% level, while the remaining controls, government fractionalization and media freedom, are each insignificant in the OP models. I conduct the following exercise to derive the substantive effect of particularism × labor mobility on contributions influence. For this exercise, I use the estimates from the full OP specification in model 13 and parametric bootstraps29 to extract the marginal effect of particularism on the predicted probability of the maximum score in the ordinal contributions influence measure across the entire range of labor mobility in the developing democracies sample. The maximum score of the ordinal contributions influence mea­ sure is equal to 7, and this score captures the industry-­executive response

Table 5.6  Results from OP models for dependent variable contributions influence

log GDP per capita particularism × labor mobility labor mobility Particularism media freedom government fractionalization µ1 µ2 µ3 µ4 µ5 µ6 Constant Log likelihood N Random effects

Model 13

Model 14

0.086 (0.093) 0.021** (0.011) 0.032 (0.049) –­0.014 (0.017) –­0.073* (0.041) 0.032 (0.029) –­2.272 (1.325) –­1.472 (1.328) 0.220 (1.319) 0.551 (1.32) –­1.78*** (1.151) –­2.542*** (1.152) 2.015*** (0.753) –­970.21 309 No

0.051 (0.087) 0.02** (0.01) 0.033 (0.061) –­0.011 (0.025) –­0.052 (0.068) 0.037 (0.03) –­0.445 (1.199) 0.335 (1.199) 1.574 (1.195) –­4.662*** (1.153) –­3.847*** (1.153) 0.628 (1.714) 1.819*** (0.665) –­825.47 309 Yes

Note: ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively.

Marginal effect of particularim on ∆prob of contribution influence = 7

Empirical Tests for Particularism, Trade Protection, and Contributions  /  153

0.6

0.4

0.2

0.0 -0.2 -30

0

30

60

90

120

Labor mobility Marginal effect

95% Confidence Interval

Figure 5.7 Effect of  particularism on contribution influence conditional on labor mobility

that campaign contributions have no impact on influencing policies.30 I also report the 95% confidence intervals of each of these marginal effects. The result from this exercise is illustrated in figure 5.7. This figure shows that particularism does not have a significant impact on the predicted probability of contributions influence being equal to 7 for low levels of  labor mobility. However, once the labor mobility measure increases above its mean in the sample, the positive effect of particularism on the predicted probability of contributions influence equaling 7 becomes statistically significant. In fact, particularism increases the predicted probability of contributions influence being equal to 7 by 14.6% when its companion variable labor mobility is in­ creased by one standard deviation above its mean while holding other variables in the specification at their respective mean. This effect is statistically significant at the 95% confidence level. Thus there exists strong statistical and substantive support for the prediction in claim 2 that leads to the prediction in hypothesis 4.2.

Conclusion 1

This chapter has presented a cumulative body of evidence that corroborates the hypotheses about trade barriers and campaign contributions from the formal model on particularism and trade policy described in chapter 4. The empirical results reported in this chapter contribute to the literature in three main ways. First and foremost, the results presented in the first half of this chapter suggest that the degree of political particularism does not have a

154 / Chapter Five

statistically independent effect on trade protection in democratic developing countries. Rather, the effect of particularism on trade barriers depends on the degree of interindustry labor mobility. I also found that governments in particularistic developing democracies implement trade reforms when the level of interindustry occupational mobility of workers is reasonably high. To the best of my knowledge, this empirical result has not been suggested in previous research. Second, numerous studies claim that politicians in particularistic systems are more receptive to the policy preferences of interest groups such as the owners of industries (Shin et al. 2005; Hicken and Simmons 2008; Webb and White 2007). Consequently, these studies predict that campaign contributions from capital tend to be high in these systems. In contrast, the empirical analysis in this chapter reveals that in the specific context of trade policy, industry owners tend to reduce the contributions that they offer when labor mobility is high. At the very least, this suggests that it is difficult to reach generalizable inferences about the susceptibility of  politicians to campaign finance in particularistic democracies across all policy issue-­areas. This is because their incentives to be responsive to “special interests” may vary dramatically across different economic policy issue-­areas. While the empirical analysis presented here and in chapter 4 corroborate the testable hypotheses, they do not adequately asses the underlying causal mechanisms that lead to the predictions in the hypotheses posited in chapters 2 and 4. Numerous questions still remain unanswered. For instance, can empirical evidence from case studies support the causal claim that political leaders in particularistic developing democracies typically place greater weight on the trade policy preferences of workers rather than the interests of capital (the owners) when designing trade policies? Is it plausible that party leaders in party-­centered democracies tend to be more receptive to the protectionist interests of  the owners in order to obtain contributions, which in turn are invested by them to build their party? I turn to answer these (and other) questions that emerge from the theoretical logic in the formal models by carefully analyzing three case studies in detail in chapters 6 and 7.

Six

Democracy, Political Particularism, and Trade Liberalization in Brazil

This chapter moves away from the large econometric tests undertaken in chapters 3 and 5. Instead, it presents supporting evidence for the theoretical arguments presented in chapters 2 and 4 by analyzing in depth trade liberalization and campaign contributions in a prominent democracy in the developing world, Brazil.1 Beyond this, chapter 7 offers a parallel account of the politics of trade reforms and campaign contributions in two other democratic developing countries, India and South Africa. The Brazil case study is used to evaluate the book’s main causal claims for numerous methodological and substantive reasons. First, the Brazil case is examined mainly to determine the interests and strategies of different political actors in the design of trade policies and therefore to overcome the limitations of pure statistical work. As such, the Brazil case allows one to evaluate the causal claims in chapters 2 and 4, which posit that leaders in new democracies and candidate-­centered systems will design trade policies that favor the interests of labor rather than protectionist industry owners when labor mobility is high. Second, following standard scientific research (Cook and Campbell 1979), the Brazil case study, as well as the India and South Africa cases, adds external validity to the statistical results. Whereas the econometric tests reveal that the formal models are accurate at the national level, the comparative analysis of Brazil, India, and South Africa shows that some of the causal mechanisms underlying the theoretical predictions in chapters 2 and 3 find empirical support. Third, studying Brazil and comparing it to the other two cases introduces variation in the dependent and independent variables for the case-­study analysis. As discussed below, Brazil made a formal transition to full-­fledged democracy in 1985 and is hence observed as a new democratic regime during the latter half of the 1980s. During the initial postdemocratic transition

156 / Chapter Six

years (1986–­90) in Brazil, the country’s electoral system was (as discussed later) unstable and “unsettled” (Mainwaring 1997; Samuels 2000). After the consolidation of democracy in Brazil by the early 1990s, a stable electoral system was established in the country. This stable electoral system is the open-­list proportional representation (PR) system and, as I will show below, ensures that Brazil’s democracy is (since the early 1990s) candidate centered (Mainwaring 1997; Ames 2001).2 South Africa made a transition to democracy (with universal suffrage on a nonracial basis) in 1994. Much like Brazil during the 1986–­90 period, South Africa is thus observed as new democratic regime from 1995 to around 1998–­99. Moreover, during the first three to four initial posttransition years, South Africa was ruled by a “government of national unity” that included representatives from all three parties: the African National Congress, the National Party, and the Inkhata Freedom Party (Jacobs and Calland 2002; Hirsch 2005). One consequence of this national unity government was that the country’s electoral system mattered little in terms of shaping or influencing political competition during the immediate posttransition period in South Africa. However, once democracy became consolidated by the late 1990s, the country’s newly established electoral system—­the closed-­ list PR electoral system—­came to the forefront. In fact, by the late 1990s, competition in South Africa’s political landscape was completely driven by the closed-­list PR system, which in essence gave rise to a party-­centered democracy (Southall 2004). Thus although Brazil and South Africa are institutionally “similar” in that both these countries are observed as “new democracies” for a certain time period, there is clear variation in the level of particularism of these two countries. This variation provides an opportunity to assess the book’s claim about how particularism affects trade policy in developing country democracies under certain economic conditions. This book focuses on India for the third case study. Unlike Brazil and South Africa, India has remained an established democracy from 1947 till today, except from late 1975 to early 1977, when then Prime Minister Indira Gandhi temporarily declared a state of political emergency.3 Yet, like South Africa, India is a party-­centered democracy.4 The India case is therefore useful from a comparative case-­study perspective, as it permits one to (i) compare trade policies of new democracies (Brazil and South Africa during the initial posttransition years) versus an established democracy (India) and (ii) assess whether and how the level of trade protection varies between candidate-­centered and party-­centered democracies. As we will see in this chapter and the subsequent chapter, different experiences with democracy and differences in particularism across the three cases help us

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  157 Brazil India

Trade share of GDP %

90

South-Africa 70

50

30

10 1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

Figure 6.1  Trade in goods as percentage of  GDP: Brazil, India, and South Africa

further investigate the causal logic lying between the explanatory variables of interest and the outcome(s) of interest—­trade protection and campaign contributions. Fourth, although there are obvious differences between the three cases5—­as elucidated above—­the cases also share certain key similarities. The sizes of Brazil’s and India’s economies are largely similar in nominal terms,6 while Brazil and South Africa are both middle-­income countries.7 Moreover, the degree of industrial diversification and technological development across these three cases are largely similar. Finally, as indicated in figure 6.1, the volume of trade flows as a share of GDP for each of these countries steadily increased from 1988 to 2002.8 Put together, the reasons delineated above permit one to use Mill’s method-­of-­difference when comparing these three cases to assess the causal mechanisms underlying the pre­ dictions from the formal models presented earlier. 1 The remainder of this chapter is organized into the following sections. The first section, “A Brief Background of Brazil’s Economy and Institutions,” is a primer on Brazil’s economy and the country’s political institutions. The next section, “A Concise History of Trade Reforms in Brazil,” briefly describes the history of trade policy and reforms in Brazil. In section three, I explore how the level of interindustry labor mobility played a critical role in inducing successive governments in Brazil to adopt trade reforms during the immediate postdemocratic transition years and when the country shifted to a candidate-­centered system. The third section, “Political Institutions, Labor Mobility, and Trade Politics in Brazil,” also presents some historical evidence showing that trade policies in Brazil shifted away from the interests of industry owners that favored trade protection during the years in which

158 / Chapter Six

the country is observed as a (i) new democracy and (ii) candidate-­centered system. Furthermore, I employ some industry-­year tariffs to statistically assess the claims about the effect of the interindustry occupational mobility of labor on trade protection in Brazil when the country is first observed as a new democracy and then a candidate-­centered system. In the section “Politicians and Campaign Contributions in Brazil,” I analyze some descriptive data on the amount of campaign contributions given to politicians in Brazil from industries. Analyses of these data are used to evaluate claims about contributions to political actors in new democracies and candidate-­centered systems. In the section “Trade Reforms and Electoral Fraud in a New Democratic Regime,” I study whether and how trade liberalization has helped to reduce the likelihood of electoral fraud in Brazil. The chapter ends with a brief conclusion.

A Brief Background of Brazil’s Economy and Institutions Brazil occupies an area covering about 47% of the South American continent, and with respect to nominal GDP, it is the tenth largest economy in the world.9 It possesses a diversified industrial and agricultural sector. In 2008, for example, the contribution of agricultural value added to GDP was 7.8%, while for industry and services, it was 32.1% and 57.9%, respectively.10 Brazil is a major exporter of mineral ores, soybeans, frozen orange juice, and processed meat, in all of which the country has a relatively natural comparative advantage. The country is also one of the two leading exporters of small/medium aircraft, the other being Canada. However, overall, in terms of world competitiveness, Brazil’s exports to world markets suggest a relatively poor performance.11 Although Brazil’s export performance has not lived up to its potential, its social indicators have been improving. According to the World Bank (2003), the indicators of gross school enrollment suggest a progressive improvement from 1980 to 2000, especially in primary and secondary enrolment. As shown below, the improvement (decline) in Brazil’s literacy (illiteracy) rates had profound consequences for the occupational mobility of workers in the labor force. With respect to its political institutions, Brazil experienced several years of autocratic rule in the postwar decades (Fausto 1999; Levine 1999). Brazil’s military launched a successful coup in 1964. This coup led the mili­tary to seize power and govern directly from 1964 to 1985. During the later years of military rule (the late 1970s and early 1980s), the military had left the economy in shambles, with soaring inequality and national debt, and thousands of Brazilians were deported, imprisoned, tortured, or murdered

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  159

(Burns 1993; Levine 1999). Given the political and economic failures of Brazil’s military dictatorship, it is perhaps not surprising that Brazil made a formal transition to democracy in 1985 in which Tancredo Neves emerged as the country’s civilian-­elected president (Rochon and Mitchell 1989; Stepan 1989; Payne 1991). Neves died before being sworn in, and the elected vice president, José Sarney, was sworn in as president in his place. President José Sarney was the first incumbent to undertake substantive trade liberalization measures (Rajapatirana et al. 1997; Muendler 2007). This first wave of liberalization occurred, as emphasized below, during the years in which Brazil was a new democratic regime. Sarney was succeeded by Fernando Collor de Mello. Following Sarney, President Collor and his administration took several steps to increase both the depth and pace of reduction of trade barriers in Brazil, which is also discussed in detail below (De Souza 1999; Almeida 2004). President Collor did not remain in office for long. He was eventually impeached from office in 1994, owing to political scandals that rocked his party (Levine 1999). Following Collor’s impeachment, the acting president, Itamar Franco, was sworn in as president on a temporary basis (Fausto 1999; Castro and Carvalho 2002). In elections held on October 3, 1994, however, Fernando Henrique Cardoso, Collor‘s finance minister, defeated the left-­ wing candidate Lula da Silva and assumed the presidency till 2002. Similar to Collor, Cardoso maintained the momentum of trade reforms (Castro and Carvalho 2002; Weyland 2002; Pereira 2004). He also successfully tackled Brazil’s debt problems and guided Brazil. Yet, in 2002, Cardoso lost the presidential election to Luiz Inácio Lula da Silva, who in turn has served as Brazil’s president till 2011. While Brazil emerged as a democracy in 1985, it has since remained a robust democracy (Ames 1995a, 2001; Mainwaring 1999). During the initial posttransition period, the conspicuous absence of strong and disciplined parties and, conversely, the presence of strong political actors (e.g., Sarney) made it difficult for the political elite to converge on a particular electoral system (Mainwaring 1997; Ames 2001). The immediate consequence of this was that Brazil’s political elite openly discussed and to some extent experimented with different types of proportional representation systems with various formulas (Ames 1995a, 1995b; Geddes 2004; Mainwaring 1999). As a result, Brazil’s electoral system initially appeared checkered and unsettled. The presence of weak political parties but strong individual political figures had one more effect—­it arguably induced Brazil’s political elite during the posttransition years to adopt an open-­list (d’Hondt) proportional representation system that promotes candidate-­centered (i.e., individual-­centered)

160 / Chapter Six

rather than party-­centered competition (Carey and Shugart 1995; Samuels 1999; Norris 2002). In fact, as emphasized by research on electoral systems in political science, rules like primaries or open-­list PR deny party leaders control over the ballot and reduce their leverage in managing the political behavior of individual legislators.12 Consequently, open-­list PR electoral rules (as used by Brazil) engender a candidate-­centered system where party unity is low, party leaders are weak, and individual politicians have incentives to cultivate a personal vote. Since Brazil uses an open-­list PR system, it is not surprising that many scholars have emphasized the highly candidate-­centered nature of Brazil’s electoral rules and argue that these rules lead to extreme levels of individualist behavior among political actors (Geddes 1994; Carey and Shugart 1995; Mainwaring 1999; Ames 2001; Cox and McCubbins 2001). Researchers point to a variety of evidence to support the causal connection between electoral law and pervasive personal electioneering in Brazil, including pork barrel politics (Ames 1995a, 1995b), parties with only minimal extraparliamentary organization, high levels of party switching (Mainwaring 1999), and low levels of unity in legislative voting (Mainwaring and Perez-­Liñan 1997). Carey and Shugart (1995) developed a rationale for the choice for such extreme candidate-­centered electoral law, which they dubbed the “inefficient secret.” According to this view, the extreme individualist incentives created by Brazil’s rules ensure that the structure of the country’s party and electoral system evolves into a “pure” candidate-­centered system. As stated by Lyne (2008: 293), “Brazilian party leaders lack the necessary powers to influence rank-­and-­file behavior sufficient to counteract the incentives of the highly personalist open-­list proportional representation rules that govern federal legislative elections. Brazilian party leaders (also) similarly have limited control over the ballot and do not have robust direct powers for controlling floor voting.” To summarize, the country’s electoral system was initially unsettled immediately after Brazil’s transition to a full-­fledged democracy in 1985. However, after the consolidation of democracy in Brazil, the country’s open-­list PR rule established candidate-­centered electoral competition. I next turn to briefly discuss the history of trade policies in Brazil.

A Concise History of  Trade Reforms in Brazil In the immediate years following World War II, Brazil followed several other developing nations in Asia and adopted import-­substitution-­ industrialization (ISI) policies, which meant that high tariffs were imposed

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  161

to protect several domestic sectors and industries in the economy. Brazil’s highly diversified and closed economy yielded inward-­looking trade policies that focused primarily on the domestic market. Even the most competitive industrial sectors, such as pulp and paper, depended heavily on the domestic market (Kingstone 1999, 2001). Key industries in the manufacturing sector such as the auto industry also relied heavily on protection of the domestic market from import competition (Kingstone 2001; Cardoso 2009). As a result, neither industrial groups nor successive incumbents in Brazil favored trade liberalization; instead, industries and politicians in Brazil consistently supported trade protection (Marconini 2005; Cardoso 2009). Moreover, firms and industries maintained close relationships to the state through cartorios, a dense network of formal and informal links (Schneider 1997; Kingstone 1999, 2001) that facilitated unproductive rent seeking. In short, between the 1950s to the mid-­1980s, Brazil’s economy was characterized by high levels of protection, low levels of competitiveness, domestic market orientation, and ample rent-­seeking opportunities from trade policy (Payne 1994; Cardoso 2009). Although ISI helped protect Brazil’s domestic industries, it had a disastrous effect on the economy, particularly from the 1970s. For example, the current-­account deficit increased from US$1.7 billion in 1973 to US$12.8 bil­ lion in 1980, owing to poor export performance. The foreign debt rose from US$6.4 billion in 1963 to nearly US$54 billion in 1980.13 The growth rate in Brazil slowed down dramatically, and real GDP per capita also remained relatively stagnant, rising from $4,527 in 1978 to just $4,568 in 1989. Additionally, as pointed out by Weyland (2002: 73), in Brazil, “ISI also had some inherent flaws, . . . which caused ever graver constraints as industrialization advanced. Oligopolistic ownership, protected from foreign competition by a host of trade barriers, led to considerable inefficiency, which—­contrary to the implications of infant industry arguments—­did not seem to diminish over time. The resulting lack of international competitiveness in sectors where Latin America’s (including Brazil’s) low labor costs did not yield a decisive comparative advantage limited industrial exports.” Notwithstanding the failure of ISI as a development strategy from the 1970s itself, successive governments in Brazil neither dismantled ISI nor adopted any substantive trade reform measures (Weyland 2002; Abreu 2004; Pereira 2004). In fact, it was not until the mid-­1980s that Brazil gradually made a shift toward dismantling ISI and embracing trade reforms (Weyland 1996, 2002; Abreu 2004; Cardoso 2009). The new industrial policy introduced in 1988 by the Sarney administration represented a strategy reversal and the beginning of a process of

162 / Chapter Six 49

Trade Barriers (%)

42 35 28 21 14 Unweighted tariff 7

Core NTBs

0 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004

Figure 6.2  Moving average of unweighted tariffs and core NTBs in Brazil

progressive trade liberalization. The approach was slightly slow at first and involved only the elimination of tariffs. As stated by Cardoso (2009: 8), “In the first phase, corresponding to the period 1988–­1989, two tariff reforms were undertaken seeking to eliminate the redundant share of the nominal tariff without significantly impacting import volumes.” However, from 1989–­1990 onward, the pace of reform accelerated, and the newly sworn-­in Collor administration actively embraced trade liberalization as a long-­term development strategy (Weyland 2002). The reforms introduced involved the complete removal of quantitative restrictions and the introduction of a timetable for tariff reductions that was implemented in five steps from November 1988 to July 1993 (Edwards 1997; Abreu 2004). The first two steps emphasized reduction in tariffs on capital and intermediate goods, while the latter three steps involved reduction in the protection granted to final (consumer) goods. Moreover, “the special tax regimes followed the withdrawal of several NTBs, in the period 1991–­1993, and were in turn sub1 tariff reductions, which had been stituted by a schedule of gradual import announced in advance” (Cardoso 2009: 8). Finally, the third phase of trade liberalization took place in 1994. This third phase was associated with tariff reductions that were promoted at the beginning of the Plano Real, with the intention to discipline domestic prices through greater external competition (Weyland 2002; Cardoso 2009). Figure 6.2 illustrates the moving average of the unweighted tariff rate14 from 1971 to 2004 (solid line) and the core NTBs ratio (in percentage terms)15 from 1987 to 2004 (dashed line) in Brazil. Note that the moving

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  163

average of the unweighted tariff rate declined considerably from 1986 to around 1990 and then again decreased sharply from 1992 to 1999. Core NTBs declined from 1988 to 1998 as well. Hence Brazil tariffs and NTBs were quite effectively reduced from 1986–­87 to 1990 and then during the 1990s. What explains this change from ISI-­inspired policies to trade liberalization from the mid-­1980s and then during the 1990s? We turn to address this question in detail in the following section.

Political Institutions, Labor Mobility, and Trade Politics in Brazil Democratization, Labor Mobility, and Trade Policy The causal story from the formal model in chapter 2 suggests that leaders in a new democratic regime will rationally weigh the trade policy interests of workers over the protectionist preferences of industry owners (capital) when labor mobility is high. They will also reduce trade restrictions when interindustry labor mobility is high. Curtailing trade barriers increases the incumbent and the ruling party’s likelihood of reelection. We find support for these causal claims during the immediate posttransition years (i.e., 1986 to 1990) in Brazil. To see why, first consider figure 6.3. The dotted line in the figure illustrates the annual series of the level of interindustry occupational mobility of labor across all Brazilian industries at the three-­digit International Standard Industrial Classification (ISIC) level (thirty-­three industries total) Democratic Transition Year

49

Labor mobility

35

4.50 3.75

28

Unweighted tariff

3.00

21

Core NTBs

2.25

14

1.50

7

0.75

0

0.00 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004

Figure 6.3  Interindustry labor mobility and trade protection in Brazil

Labor Mobility

Trade Barriers (%)

42

164 / Chapter Six Table 6.1  List of industries (three-­digit ISIC level) in tradable sectors in Brazil Agriculture products, not processed Airplane production Automobiles, trucks, and buses Beverages Coffee products Electrical machinery Electronic equipment Footwear Glass and glass products Industrial chemicals Iron and steel Leather products Manufacturing Meat products Mining and quarrying Nonferrous metal basic industries Nonmetallic mineral products Nonmetallic products Office and computing machinery Paper, pulp, and cardboard Perfumery products Petrochemicals Pharmaceuticals Plastic Processed foods Professional and scientific goods Rubber products Textiles Tobacco Transportation, storage, communication Vehicle parts Wearing apparel Wood and furniture Note: “Not processed agricultural products” include raw food materials such as rice, wheat, soybeans, corn, and sugar.

from 1977 to 2004, for which data are available to operationalize this series. These thirty-­three industries are listed in table 6.1. I used the formula described in equation 4.1 in chapter 4 to calculate the degree of occupational mobility of workers in thirty-­three different industries on an annual basis from 1977 to 2004 in Brazil;16 the dotted line in the figure is derived from this measure.17 The solid line in figure 6.3 plots the moving average of unweighted tariffs from 1971 to 2004, while the dashed line plots the moving average of the core NTBs ratio from 1987 to 2004. Last but not the least, the vertical line in the figure indicates the year (1985) in which Brazil made a formal transition to a full-­fledged democracy.

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  165

Figure 6.3 reveals two interesting but sharply different patterns. On the one hand, it shows that the interindustry occupational mobility of workers in Brazil increased at a steady rate from 1986 to 1990–­91 and (more im­ portant) became sufficiently high in this time period.18 On the other hand, however, the figure indicates that tariffs in Brazil started declining only from 1986—­that is, the year following the democratic transition of 1985—­and continued to decline steadily from 1986 to 1990 and thereafter. Similarly, NTBs decreased from 1988 onward and remained low thereafter. Hence these patterns mentioned above reveal that the first wave of trade liberalization and therefore the first major reduction of trade barriers in Brazil occurred from 1986 to 1990 when the country is observed as a “new democracy” and when the interindustry occupational mobility of labor increased in the country. One can thus arguably infer from this figure, as suggested in chapter 2, that it was the combination of Brazil being a new democratic regime and the increase in labor mobility in the country’s economy from the mid-­to late 1980s that contributed to the first wave of trade liberalization in the country. Did government officials in a newly democratic Brazil respond to growing labor mobility in the country by weighing the trade policy preferences of workers more than the interests of industry owners, as suggested by the model in chapter 2? And did Brazilian officials reduce trade barriers during the immediate postdemocratic transition years—­as predicted by the model in chapter 2—­because they felt that workers in the country could adjust to trade liberalization, owing to increased occupational mobility in the labor force? A close look at the historical record shows that the first two incumbents in Brazil’s new democratic regime, President Sarney and President Collor, indeed made a decisive shift away from the interests of “special interest groups”—­especially industrialists who favored trade protection—­to the preferences of workers when designing trade policies. This is emphasized by Weyland (2002: 134), who suggests that Sarney and Collor appealed to “large numbers of unorganized citizens” and “used support from a heterogeneous following” to “facilitate the launching of profound structural reforms that dismantled the inward-­looking, state-­interventionist development model and gave market forces free rein . . . In a systematic mobilization of bias, they invoked the will of the people against special interests that defended their own ‘privileges’ at the expense of the common good” (Weyland 2002: 134). Other researchers have also suggested that during the immediate postdemocratic transition years, Brazil’s politicians recognized that trade policy

166 / Chapter Six

could be used as a tool to enhance the income of the median voter and that this would generate electoral dividends. Consequently, this encouraged politicians, including the incumbent, to design trade policies that were weighted toward the trade policy preference of the median voter in Brazil who was (and is) a worker (Armijo 2007; Armijo and Kearney 2008). As stated by Armijo (2007: 2024), government leaders in newly democratic Brazil adopted trade policies that were biased toward the trade policy preferences of the median voter since they recognized that “the only way to protect the income of the median member of the set of included political actors (in this case, the median resident) is to pursue (trade) policies that generate at least a rough approximation of stable . . . growth.” While the transition to democracy induced Brazilian leaders to develop a trade policy bias toward the median voter’s (a worker’s) preference, it also encouraged them to design trade policies that were not in alignment with the interests of owners, particularly those that favored trade protection (Weyland 2002; Armijo and Kearney 2008). This is indicated by a prominent newspaper in Brazil in 1987, which emphasized that the Sarney administration in a newly democratic Brazil ensured that trade policies were “no longer hostage to the vested interests of business groups and large industrial houses.”19 Not surprisingly, Brazil’s industry reacted sharply against the lack of attention paid to their trade policy interests by the Sarney administration (Payne 1992; Nelson 1995; Weyland 1997, 2002). They also opposed the implementation of trade reforms by Sarney because this hurt industry profits (Nelson 1995; Gereffi 1996), as suggested in an interview of a director of the São Paulo Federation of Industries (FIESP) conducted by Payne (1992: 2): “Industry in particular suffered and reacted by blaming the government. One director from [FIESP] stated that Sarney had failed to guarantee the minimum needs of industry: profit, a market, affordable credit, stable rules.” Brazilian officials paid substantial attention to the preferences of workers and particularly their ability to adjust to globalization when designing trade policies. For instance, there was a growing recognition in the late 1980s among officials in President Sarney’s government that the majority of workers in the Brazilian economy could adjust to trade liberalization because of the growth in occupational mobility in the labor force (Muendler 2002; Lopez-­Cordova and Moreira 2003). This was emphasized in a report by the Ministry of Labor and Employment in 1987: “Brazil has witnessed a remarkable increase in the productivity and skills of workers in recent years . . . unlike before, the average Brazilian worker today has greater ability

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  167

to change jobs between firms to capture the gains from trade globalization if required.”20 Furthermore, the PMDB (Partido do Movimento Democrático Brasileiro) headed by President Sarney also realized by the late 1980s that Brazilian workers were disillusioned with ISI and were, in fact, increasingly predisposed toward more trade liberalization. PMDB party members indeed recognized that adopting trade reforms is necessary from a political perspective as it may yield substantial electoral dividends for the party. This sentiment was candidly expressed in the PMDB’s party manifesto in 1986: “It is no secret that Brazilian workers are tired of trade protection . . . They realize, just as we do, that more import competition will help to lower prices, increase efficiency, generate growth, and above all, create more jobs . . . It is therefore important for the PMDB to pay serious attention to the wishes of the workers and relax import restrictions since this will allow the party to build a solid electoral base among workers.”21 The aforementioned statement in the PMDB’s party manifesto further corroborates my causal claim that after the transition to democracy, ruling party members (i.e., the PMDB in Brazil after 1985) indeed start paying more attention to the trade policy preferences of workers. The evidence also shows that the PMDB believed that workers could cope with globalization and that PMDB may gain electoral dividends by reducing trade barriers. But did this belief induce the incumbent party to reduce trade reforms? The historical record provides an affirmative answer to this question. For instance, Weyland (2002) suggests that by the late 1980s, expectations of “strong popular support” from workers “for drastic adjustment” of economic (including trade) policies in Brazil “helped neopopulist leaders silence opponents of market reform . . . (and) facilitated the assault on the innumerable regulations, subsidies, and protections that shielded these sectors from the rigors of the market” (Weyland 2002: 141). As a result, “trade liberalization—­the neoliberal reform most costly for protectionist sectors—­ advanced at a rapid rate” (Weyland 2002: 157). Furthermore, a report prepared by the ruling PMDB party on economic reforms recommended “the implementation of trade reforms . . . It will lead to much needed economic growth in Brazil and will help the party win the electoral support from the formal sector urban workers and the salaried middle class that will gain from the opportunities created by trade reform.”22 Thus the historical evidence on trade politics from Brazil in the late 1980s, the simple analysis of the relevant Brazilian time-­series data (see figure 6.3), and Weyland’s (2002) insights corroborate the causal claim

168 / Chapter Six

about the nature of strategic interaction between incumbents and workers as well as the owners (capital) in new democracies in the developing world. It also supports my causal arguments about how interindustry labor mobility affects the political calculations of policy makers in new democracies and consequently the trade policies that they adopt. I next turn to address how Brazil’s particularistic (i.e., candidate-­centered) system affected the country’s trade restrictions in the 1990s to assess the causal claims from the model in chapter 4. Recall that the central causal claim from the model in chapter 3 is that a higher level of particularism in developing democracies—­that is, the shift to a candidate-­centered/particularistic system—­will induce the incumbent to weigh the trade policy preferences of  workers over capital when interindustry labor flows is substantial. Incumbents in candidate-­centered systems will choose to reduce trade restrictions when interindustry mobility is high. Does the empirical evidence from Brazil (during the years in which it is observed as a particularistic system) support this argument? To address this question, let us return to figure 6.3 in order to check the slope of the dotted, solid, and dashed lines in the figure from 1991 to 2004. I focus on the 1991 to 2004 time period because this includes the years in which Brazil is observed as a candidate-­centered system. The dotted, solid, and dashed lines in the figure during the 1991 to 2004 time period reveal two patterns. First, the dotted line indicates that that the interindustry occupational mobility of labor in Brazil increased once again during the 1990s, and moreover, the level of interindustry labor mobility remained fairly high in Brazil during the last decade of the twentieth century. Second, the figure shows that tariffs and NTBs declined sharply from 1992 to 2002 under the Collor, Franco, and Cardoso administrations. This implies that the governments in Brazil that succeeded the Sarney administration did not reverse the process of trade reforms. Rather, Presidents Collor, Franco, and Cardoso initiated a second wave of trade reforms. The temporal pattern of interindustry labor mobility and trade protection in Brazil shows, as predicted by my model, that it was a combination of the candidate-­centered nature of Brazil’s electoral system and the increase in labor mobility in the country’s economy during the 1990s that contributed to the decline in both tariffs and NTBs. But did policy makers in Brazil prioritize the trade policy preferences of workers and distance themselves from the protectionist interests of some industry owners when designing trade policies in the 1990s? An examination of the historical evidence reveals that the political influence of industrialists (i.e., capital) decreased considerably under the Collor administration in the 1990s when Brazil was firmly established as a candidate-­centered system. Moreover, President Collor (and his successors)

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  169

shifted the direction of trade policy away from the economic interests of industry owners and reduced trade barriers, even though these owners openly requested for more trade protection. As stated by Payne, Industrialists perceived a loss of influence during the Collor government. This was partially apparent in his appointment of Zélia Cardoso de Melo, a former member of the Communist Party, to the Finance Ministry. Collor ignored both business elites’ alternative recommendations for, and protests against, the Cardoso appointment. Another example of business elites’ lack of influence over Collor was the Collor Plan, an economic program designed without business input and announced shortly after he took office in March 1990 . . . the plan threatened businesses by. cracking down on capital flight, reducing import barriers,. and increasing taxes. (1992: 23; emphasis added)

Further, Payne (1992: 20) points out that under Collor, business elites from manufacturing industries that supported protection, “lost some of their influence . . . Collor proved unwilling to negotiate with them, openly attacked them, appointed cabinet ministers whom they vehemently opposed, and adopted policies inimical to their interests.” President Collor’s tactic of openly attacking especially the owners of industries that favored trade protection reached a peak during the early 1990s in candidate-­centered Brazil. Indeed, Collor publicly denounced Brazil’s domestic automobile industry, for example, which was using the FIESP to officially lobby the government for more trade protection. As pointed out by Weyland (2002: 143), “Collor went even further by publicly snubbing FIESP on several occasions, disqualifying it as a ‘den of the retrograde,’ and engaging in angry exchanges with FIESP President Mario Amato. And although the automobile industry was widely seen as one of Brazil’s leading sectors, the President branded cars produced in Brazil as outdated carocas, in order to justify his trade liberalization, which sought to improve productivity, efficiency and competitiveness” (2002: 143). Apart from Collor, incumbents from successive ruling political parties in the 1990s—­Itamar Franco and Fernando H. Cardoso—­overtly chose to distance themselves away from industry owners that not only were highly protected from import competition in the past but also had close relationships with the political elite via cartorios (Harrison et al. 2004; Moreira 2008; Oli­ veira 2009). By doing this, successive incumbents signaled to voters that they were more interested to respond with alacrity to the demands of the common man, including workers, instead of pandering to the interests of previously protected industries (Weyland 2002; Moreira 2008). Because

170 / Chapter Six

successive incumbents chose to distance themselves from the owners of asset-­specific industries once Brazil evolved into first a new democracy and then a candidate-­centered system, the political leverage that these industries had with respect to influencing trade barriers declined (Payne 1992; Weyland 2002; Abreu 2004; Cardoso 2009). It is therefore clear that trade policies implemented by policy makers in candidate-­centered Brazil diverged particularly from the preferences of owners (i.e., capital) who favored trade protection. But did government leaders and officials design trade policies in candidate-­centered Brazil during the 1990s that were biased toward the economic interests of workers? To answer this question, first observe that the candidate-­centered nature of Brazil’s electoral system encouraged President Collor, his successor President Franco from 1991 to 1995, and President Cardoso from 1995 to resort to a “per­sonalist type of leadership” (Weyland 2002: 141). Collor, Franco, and Cardoso’s personalist type of leadership, in turn, encouraged them to appeal directly to the interests of large groups like labor when setting the agenda for trade policy reform, and this bolstered their trade reform effort (Payne 1994; Pereira 2004; Abreu 2004; Weyland 2002;). As suggested by Weyland (2002: 143), “Collor also courted a moderate trade union confederation and labor in rural and urban Brazil in order to promote his reform agenda, which included trade policy reforms.” Furthermore, “Collor used his strong mass support to face down potential resistance (to economic reform). The thirty-­five million presidential votes he won in the presidential election and the high approval for his first stabilization plan were among the main arguments that the president and his aides used to coax a reluctant Congress into passing the . . . adjustment program” (Weyland 2002: 143). Collor and Cardoso’s direct appeal to groups like labor when setting the agenda for neoliberal economic (including trade) reforms had two important consequences. First, similar to their predecessor Sarney, Collor and his advisers (which included his successor Franco) as well as Cardoso and his associates recognized that the degree of interindustry labor mobility in Brazil was fairly high and that Brazilian workers could thus potentially gain from more trade openness. This is indicated in an interview given by an official from Brazil’s ministry of finance during the days of the Cardoso administration: “Under President Cardoso, we assessed and found that Brazil’ workers in the formal and informal sectors were mobile enough to either move out of declining industries or move into rising industries that gain from lower trade bar­ riers. . . . our understanding about the ability of workers to adjust to glob­ alization coupled with President Cardoso’s promise to promote economic

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  171

growth encouraged us to pursue the path of trade liberalization.”23 Second, this belief held by Collor, Franco, and Cardoso induced them to introduce a second wave of trade reforms, since they perceived that doing so would enhance their political support and stature in society. This is emphasized by Weyland (2002: 141), who suggests that Brazilian political leaders in the 1990s adopted economic and trade policy reforms since it “boosted neopopulist leadership . . . and gave personalistic leaders a technocratic mantle of legitimation and external support . . . in sum, neopopulist politics and neoliberal reform reinforced each other.” In addition to Weyland’s (2002) insight, Pio (2000), Abreu (2006), and Pereira (2004) suggest that the Brazilian government voluntarily committed itself to reduce tariffs with the MERCOSUR countries24 in 1992 because it was confident that workers in the labor force in Brazil were mobile enough to easily adjust to import flows from other nations. In fact, in 1994, the Franco administration committed itself to reduce Brazil’s average nominal tariff rate to as low as 12%, since it did not fear a political backlash from labor even if Brazilian firms faced intense import competition from its trading partners in MERCOSUR (Pio 2000; Ribeiro et al. 2003; Abreu 2004). This trend was also continued by the Cardoso administration from 1998, which implemented several trade reform measures (Kume et al. 2000; Abreu 2004; Cardoso 2009). In summary, the qualitative analysis of the political economy of Brazil’s trade policy since 1986 corroborates my causal arguments that if interindustry labor mobility in the economy is sufficiently high, then leaders in new democratic regimes and in consolidate candidate-­centered democracies cater to the trade policy preferences of labor more than protectionist capital owners. But is there quantitative support for the implications of the aforementioned causal claim? For instance, if governments in a new democratic regime and in a candidate-­centered system favor the trade policy interests of occupationally mobile workers over the preferences of industry owners—­as predicted by the model in chapter 2 and as shown above in Brazil’s case—­ then it is likely that interindustry labor mobility will have a negative effect on trade barriers in Brazil during the years in which it observed as a new democracy and then (later) as a candidate-­centered system. Conversely, since governments in a new democratic regime and in a candidate-­centered system pay less attention to the preferences of owners, it is plausible that the protectionist interests of industry owners may have a negligible impact on trade restrictions in a country like Brazil. Are these two causal predictions statistically valid in the case of Brazil? The next section addresses this question.

172 / Chapter Six

Analyzing Industry-­Year Tariffs To evaluate the quantitative validity of the claims in the preceding paragraph, I gathered a panel data set of industry-­year tariffs from 1977 to 2006 for Brazilian industries at the three-­digit ISIC level (specifically thirty-­three industries) for which data are available to operationalize industry-­year tariffs. These thirty-­three different industries are listed in table 6.1. The industry-­year tariff rate, which is simply labeled as tariff, serves as the dependent variable for the statistical analysis.25 Unfortunately, data for NTBs is not available for Brazil at the industry-­year level for the previous three decades.26 Hence the within-­country statistical analysis of trade protection in Brazil focuses on evaluating the causal implications from the models in chapter 2 and 3 on industry-­year tariff data from Brazil. There are two main explanatory variables of interest in the specification. The first is an annual measure of interindustry labor mobility (labeled labor mobility) for Brazil. The formula in equation 4.1 in chapter 4 is used to calculate the occupational mobility of workers in the thirty-­three different industries listed in table 6.1 on an annual basis from 1977 to 2006 in Brazil. Second, one needs to develop a measure that operationalizes the “protectionist interests” of owners of industries in Brazil to assess if tariff policies implemented by successive governments in Brazil respond to the protectionist preferences of industry. Following the lead of economists who have statistically analyzed within-­country data on trade barriers, I use the import penetration ratio across thirty-­three industries in Brazil (listed in table 6.1) as a proxy to capture the protectionist interests of industry owners. This is because the import penetration ratio provides an “intuitively appealing way to categorize industries facing significant foreign competition” (Kletzer 2002: 21). Moreover, as import penetration increases and domestic industries face more foreign competition, they are more likely to lobby their government for more protection of their respective industry from import competition (Baldwin 1995; Lee and Swagel 1997). I thus calculate the annual series of the import penetration ratio for industries in Brazil for which the data are available to compute this measure. Specifically, the import penetration ratio is calculated per year by dividing industry imports by the sum of industry outputs plus imports (the denominator is industry supply) across the thirty-­ three industries in Brazil listed in table 6.1.27 This measure is labeled as import penetration ratio. Following extant quantitative research on trade protection in Brazil by economists,28 the following control variables are included in the model:

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  173

the annual series of Brazil’s capital-­output ratio, investment-­output ratio, log GDP,29 the lag of the relevant dependent variable, and a time trend in the specification. After gathering the data for the dependent, independent, and control variables together, I divided the entire 1977 to 2006 industry-­ year panel into three separate samples that correspond to three different time periods: 1977–­85, 1986–­92, 1993–­2004. The first time period (1977–­ 85) corresponds to the years in which Brazil is observed as a dictatorship. The second period (1986–­92) largely captures the years in which Brazil is observed as a new democratic regime, while the third time interval (1993–­ 2006) captures the years in which the country is quite firmly a candidate-­ centered system. Breaking up the full sample into three separate samples for the three time periods mentioned above allows one to separately analyze the statistical effect of import penetration ratio and labor mobility on tariff  in Brazil during the years in which it is observed as a dictatorship (model 1, table 6.2), new democracy (model 2, table 6.2), and candidate-­centered system (model 3, table 6.2). Since labor mobility and import penetration ratio may be endoge­ nous to the dependent variable (tariff ), each of the three models mentioned above is estimated via the system–­generalized method of moments (system-­ GMM) approach. The system-­GMM estimator not only corrects for endogeneity but also accounts for heteroscedasticity, unobserved heterogeneity (at the industry-­year level), and serial correlation. Table 6.2 reports the results from the system-­GMM models. The estimate of import penetration ratio on tariff in Brazil for the 1977 to 1985 time period (see model 1, table 6.2)—­the years in which Brazil is observed as a dictatorship—­is positive and significant at the 1% level. This result suggests, as predicted by the model in chapter 2, that industry owners were able to successfully influence the military junta in Brazil to raise trade barriers between 1977 and 1985 to protect their industries from import competition. The coefficient of this variable is, however, statistically insignificant in model 2, where the sample ranges from 1986 to 1992 and thus includes the years in which Brazil is a new democratic regime, and model 3, where the panel ranges from 1993 to 2006 and therefore incorporates the years in which Brazil is observed as a candidate-­centered system. The insignificance of import penetration ratio is substantively interesting, as it confirms the theoretical expectation that the protectionist interests of industry owners—­which is proxied by the import penetration ratio mea­ sure—­is unlikely to statistically influence trade barriers in Brazil when it is observed as a new democratic regime (from 1986 to around 1992) and as a candidate-­centered system (from 1992 onward).

174 / Chapter Six Table 6.2  Industry concentration and trade protection in Brazil 1977–­85 Model 1 lag tariff (1977–­85)

1986–­91 Model 2

0.1321*** (0.0482)

lag tariff (1992–­2004)

log GDP capital-­output ratio investment-­output ratio labor mobility time trend Constant Sargan test (p-­value) AR(1) AR(2) N

Model 3

0.1746*** (0.0246)

lag tariff (1986–­91)

import penetration ratio

1992–­2006

0.0381*** (0.0071) –­0.0097 (0.0155) 0.0637 (0.0433) 0.0749 (0.1132) –­0.0114 (0.0135) –­0.032** (0.006) 0.6215*** (2765) 0.31 –­3.73** –­0.219 288

0.0452 (0.0750) –­0.0063 (0.0129) 0.0287 (0.0247) 0.0945 (0.1289) –­0.0103*** (0.0022) –­0.062*** (0.018) 0.8478*** (0.3155) 0.35 –­2.96** –­0.24 204

0.1515*** (0.0292) 0.0260 (0.0450) –0.0171**   (0.0063) 0.0119 (0.0264) 0.0807 (0.1560) –­0.0263*** (0.054) –­0.04*** (0.009) 0.6311*** (0.3125) 0.3 –­1.97** –­0.1116 351

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. The results reported for the system-­GMM models are from the one-­step estimation, except the Sargan test and the AR(1) as well as the AR(2) tests, which are taken from the second-­step estimation. Note from Blundell and Bond (1998) that a negative and statistically significant AR(1) term plus a statistically insignificant AR(2) term indicates no serial correlation.

The estimated coefficient of the interindustry labor mobility measure is insignificant in model 1, which indicates that the occupational mobility of workers mattered little to Brazil’s military dictators when designing trade policies. In contrast, the estimate of labor mobility is negative and highly significant in models 2 and 3. The negative and statistically significant estimate of labor mobility on tariffs in these two models indicates that leaders in Brazil’s new democracy, such as Sarney (based on the result for labor mobility in model 2, table 6.2), and leaders and candidates in its candidate-­centered system (from the result for labor mobility in model 3, table 6.2) accounted for the occupational mobility of workers in the labor force when choosing to adopt trade reforms. Finally, the annual series of the capital-­labor ratio and investment-­output ratio are each consistently insignificant in the

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  175

models. I next turn to analyze the validity of the theoretical claims about campaign contributions in candidate-­centered Brazil.

Politicians and Campaign Contributions in Brazil The formal models in chapters 2 and 3 provide two main causal predictions about campaign contributions in the context of trade politics in developing countries. The first is that industry owners who seek more trade protection have low incentives to provide contributions to politicians and leaders in new democracies and in particularistic systems, as these leaders cannot credibly commit themselves ex ante to implement the trade policies preferred by the owners, particularly when labor mobility is high (which, as seen earlier, holds true for Brazil since the mid-­1980s). This, in turn, leads to a decline in campaign contributions in these states. Second, the model in chapter 3 also implies that in developing democracies, contributions from the owners that favor trade protection in candidate-­centered systems will be lower than the contributions they would otherwise offer in party-­centered systems. In chapter 7, I empirically assess this second claim by comparing campaign contributions in candidate-­centered Brazil with the contributions given by industry in party-­centered systems such as India and South Africa. In this chapter, I focus on evaluating the first causal argument mentioned above, which predicts that contributions given by industry owners to influence policy are likely to decline in new democratic regimes and candidate-­centered systems. Brazil provides an opportunity to test this causal claim posited because it is observed as a new democracy (from 1986 to 1991) and as a candidate-­centered system (from 1991 onward). Existing studies also suggest that the industry often provides campaign finance to politicians in Brazil and that this breeds corruption (Fleisher 1997; Samuels 2001, 2002). This is suggested by an interesting story reported by The Economist: There is nothing new about Brazilian officials being caught on tape behaving badly. But when the official is the top aide of  Jose Dirceu, the president’s chief of staff, and the president is the saintly Luiz In cio Lula da Silva, Brazilians take notice. Such was the case when poca, a magazine, published portions of a videotape from 2002 showing Waldomiro Diniz, then head of Rio de Janeiro’s lottery, soliciting campaign contributions from a reputed boss of Brazil’s illegal numbers racket (known as jogo de bicho) . . . The system of party finance invites scandal. Brazilian candidates depend on large corporate contributions, notes David Samuels, a Brazil specialist at the University of  

176 / Chapter Six The Minnesota. PT, erstwhile scourge of business, has had less to spend than its rivals because of its dearth of corporate funding. As it has grown into a national force, says Mr. Samuels, the PT may have had to bend the legal guidelines as other parties have done.30

Brazil is in many ways a useful case for studying the phenomenon of campaign contributions to influence economic policy, since data on the amount of campaign contributions provided by Brazilian industries to politicians do exist. However, as described in the analysis to follow, data on campaign finance in Brazil are only available from 1994 but not earlier. Further, data on perceived levels of campaign contributions (analyzed in the previous chapters)—­which have been gathered by the World Economic Forum (WEF)—­are only available for 2004 and not earlier. This means that one cannot empirically assess the degree of campaign contributions given to politicians by Brazilian industry owners during the 1986 to 1990–­91 period, which includes the years in which Brazil is observed as a new democratic regime. Thus, given this data limitation, much (if not all) of my analysis below on contributions given by owners to influence trade policy focuses on the 1990s, when Brazil is observed as an established candidate-­centered system. To begin, I examine some qualitative and indirect evidence of campaign contributions offered by Brazilian industry to the ruling PMDB party—­ headed by President Sarney—­during the immediate years following Brazil’s transition to democracy in 1985. A careful analysis of this evidence shows that even though “Sarney personally attended meetings with industrialists and even occasionally solicited their opinions with regard to economic policies” (Payne 1992: 4), contributions from Brazilian industry (who favor more trade restrictions) to the PMDB and the opposition parties arguably decreased in newly democratic Brazil in the late 1980s. For example, in an interview conducted in 1987, the director of a Brazilian metalworking firm responded to a question about Sarney’s trade liberalization policies by stating that “the most terrible thing in the world for Brazilian business, including my company, is unfair foreign competition that has resulted from Sarney’s decision to lower tariffs. . . . Suddenly we have to live with tremendous economic risk because of import competition . . . This risk has forced us to cut back from investing and from providing finances to the PMDB and the state to secure our economic future.”31 The aforementioned statement is not the only example of business groups/industries cutting their contributions to parties such as the ruling

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  177

PMDB party in the immediate years after Brazil made a transition to democracy in 1985. For instance, when business associations observed the trade liberalization policies implemented by the Sarney administration in 1987–­88, they “repeatedly demanded that the government withdraw” trade reforms.32 Moreover, FIESP President Mario Amato not only “threatened to organize business groups to carry out civil disobedience” (Payne 1992: 5) against trade reforms but also made it clear to politicians from the PMDB, PDS (Democratic Social Party), and PP (Popular Party) that Brazilian industry would not “financially support” the parties if they persisted in ignoring the interests of “industries that are suffering from the flood of cheap imports resulting from trade reforms.”33 Finally, a report prepared by a business (interest) group Associação Brasileira da Indústria Eletro-­Eletrônica (ABINEE) in 1988 pointed out that the “sudden” trade reforms implemented by the Sarney administration in Brazil’s new democracy “severely eroded the profits and market share” of firms represented by ABINEE.34 Because of Sarney’s trade liberalization policies, the firms that constitute the ABINEE were advised to “voluntarily boycott the PMDB and other parties by not providing donations to these parties for their political campaigns.”35 Thus the preceding discussion, which provides some qualitative evidence about contributions by industry in Brazil in the late 1980s, indicates that (as predicted by the model in chapter 2) owners of domestic industries reduce contributions to politicians in new democratic regimes when they anticipate or observe the latter adopting trade reforms. What about campaign contributions provided by Brazilian capital to influence trade policy in the 1990s, in which Brazil is observed as an established candidate-­ centered system? This question is systematically addressed in two steps. First, from the Confederação Nacional da Indústria’s (hereafter CNI’s) survey of business opinion conducted in each of the three presidential election years (1994, 1998, and 2002), I identify the list of industries in Brazil that opposed trade liberalization and consequently attempted to influence trade policy by providing contributions to politicians. Second, using a novel data set on campaign finance from each of the three presidential election years mentioned above, I track the level of contributions given by these industries that opposed trade reforms across the three elections. To this end, I first examine the trade policy preferences expressed by industrialists across several different industries in Brazil. In particular, I assessed some descriptive statistics from the CNI survey of business opinion conducted in each of the three presidential election years (1994, 1998, and 2002).36 The CNI survey asks industrialists whether they oppose trade

178 / Chapter Six Table 6.3. Descriptive results from CNI survey Panel A: List of main industries surveyed by CNI (1994, 1998, and 2002) Aeronautical; automobiles; chemical industries; electrical machinery; iron and steel; leather and footwear; mineral extraction; nonmetallic minerals; petrochemicals; pharmaceuticals; transport equipment; wood and furniture Panel B Sample

Industrialists from twelve industries listed in panel A

Mean percentage opposed to trade reforms 1994 election (1) 72.4

1998 election (2) 73.7

2002 election (3) 71.5

liberalization and then reports descriptive data from the collected survey evidence. A key advantage of the CNI survey is that it surveys these industrialists from the twelve largest Brazilian industries, listed in panel A of table 6.3.37 Panel B of table 6.3 reports the descriptive results from the CNI survey of  industrialists from these twelve industries for the 1994 presidential election (column 1), 1998 election (column 2), and the 2000 election (column 3). The descriptive results in the columns provide information about the percentage of  industrialists surveyed across the twelve industries that op­ posed trade liberalization. As shown in the table, the mean percentage level of Brazilian industrialists across the twelve industries listed in panel A that opposed trade liberalization in 1994 is as high as 72.4% (see column 1). This figure is 73.7% in 1998 (see column 2) and 71.5% (see column 3) in 2002. Thus a clear majority of industrialists from the twelve key industries in Brazil opposed trade liberalization, which meant a continuation of ISI policies rather than trade reforms. The descriptive results in table 6.3 are not surprising. This is because research on the politics of trade policy in Brazil has consistently shown that business associations from industries such as iron and steel, automobiles, chemicals, mineral extraction, pharmaceuticals, and petrochemicals have often lobbied successive governments in Brazil for more trade protection (Costa Filho 1995; Kingstone 1999; Armijo and Kearney 2007). More important, based on analyses of contributions given by different industries in Brazil over time, recent studies have also suggested that the fourteen industries mentioned above actively provide financial (campaign) contributions to politicians in Brazil (Mello 1992; Samuels 1998, 2001). Furthermore, these industries provide contributions as a tool to influence economic poli-

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  179

cies such as tariffs in order to ensure that the implemented tariff policy is compatible with their protectionist interests (Mello 1992; Kingstone 1999). One builds on the existing studies mentioned above to address the following questions relevant to the analysis: Did opposition to trade liberalization by a majority of the industrialists from the twelve industries listed in table 6.3 induce these industries to provide contributions to parties to obtain trade protection from import competition? Or did contributions offered by these industries to influence trade policy decline in a candidate-­centered system like Brazil, as predicted by the model in chapter 4? I employ some campaign finance data from Brazil to answer these questions. With respect to this campaign finance data, observe that in 1993, Brazil’s congress passed a law requiring all candidates to submit a prestação de contas, or a registry of campaign contributions, to the national electoral court, the Tribunal Superior Eleitoral (TSE) in Brasília. This has resulted in a database that contains records of campaign contributions to political candidates, including federal deputies who constitute the bulk of the individual legislators on the legislative floor (Tribunal Superior Eleitoral [Brasil] 1995, 1999, 2003). Each entry includes the candidate’s party and electoral identification number, the date of the contribution, the contributor’s name and/or industrial affiliation, and the amount contributed. The contribution database also organizes contributions by source: corporate, private, political party, or unknown. I focus on corporate contributions and, more specifically, on the amount of contributions given by the twelve industries listed in panel A of table 6.3 in each of the three presidential election years (1994, 1998, and 2002). This is because these twelve key industries typically provide contributions to influence trade policy. Moreover, focusing on the contributions offered by these particular industries allows us to assess my causal claim about the level of contributions provided to politicians and parties in candidate-­centered Brazil to influence trade policy. Given rampant inflation in Brazil during the 1990s and our interest in comparing the extent of campaign contributions across three different presidential elections over time, I thus converted all entries to US dollars, based on real exchange rates during each relevant presidential election year examined below.38 Table 6.4 presents some descriptive statistics for the amount of contributions provided in the three presidential elections years (1994, 1998, and 2002) by twelve industries, as reported in the Tribunal Superior Eleitoral (TSE) database. The table shows the total contribution amount provided to politicians to influence trade policy and the average amount contributed by the twelve listed industries for the 1994, 1998, and 2002 elections. Note that in

180 / Chapter Six Table 6.4  Descriptive statistics of campaign contributions in US dollars (real exchange rate)

Total amount of contributions Total number of contributing firms Average contribution per firm across the eleven industries Difference-­of-­means test for average contribution per firm between column (a) and column (b) (with p-­value) Difference-­of-­means test for average contribution per firm between column (b) and column (c) (with p-­value) Difference-­of-­means test for average contribution per firm between column (a) and column (c) (with p-­value)

(a) 1994 election

(b) 1998 election

(c) 2002 election

US$59,183,425 539

US$57,294,614 588

US$45,618, 828 524

US$109,802.30

US$97,439.82

US$87,058.83

–­11.26% (0.0)

–­10.65% (0.0)

–­20.7% (0.0)

Note: In 1994, 1 US dollar was equal to 0.845 Brazilian real. In 1998, 1 US dollar = 1.14 Brazilian real, while in 2002, 1 US dollar = 2.38 Brazilian real.

real US dollars, the average amount of contributions provided to politicians by the twelve industries that gave contributions to manipulate trade policy in the 1998 election is 11.26% lower than the average amount given by these industries in the 1994 election. In fact, the percentage difference between the 1998 and the 1994 average contribution amount reported above is statistically significant at the 1% level. Similarly, one finds that the average amount contributed by the twelve key Brazilian industries in the 2002 election is 10.65% lower than the mean amount given in the 1998 presidential election; this result is also statistically significant at the 1% level. Additionally, in table 6.4, the total amount of contributions provided by the twelve listed industries to politicians in the 1998 and 2002 election is lower than the total amount that they contributed in the 1994 election. The descriptive statistics in table 6.4 provide a broad picture of the amount of campaign contributions given by Brazilian industries that have protectionist interests and have provided contributions to political parties and candidates to influence the extent of trade protection. But it corroborates the causal claim from the model in chapter 3 suggesting that campaign contributions from industry owners with protectionist interests are likely to decrease in candidate-­centered systems in the developing world like Brazil. One can also conjecture from figure 6.3 that it was arguably the progressive

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  181

inability of industries with protectionist interests to influence trade policies in candidate-­centered Brazil that probably induced the owners of these industries to lower contributions to the political candidates. Three reasons from the model in chapter 4 as well as the qualitative analysis presented above for Brazil suggests that this conjecture is plausible. First, the anecdotal evidence presented in the previous section shows that the Cardoso and particularly the Collor administration in candidate-­ centered Brazil (in the 1990s) publicly denounced protectionist industries and shifted toward trade reforms that were inimical to the interests of industries that favored higher trade barriers (Payne 1992; Weyland 2002; Abreu 2004). Second, as discussed earlier, once Brazil evolved into an established candidate-­centered system, successive governments in Brazil displayed a clear bias toward the preferences of workers over capital when designing trade policies (Payne 1992; Weyland 2002; Cardoso 2009). Third, research on trade politics in Brazil also suggests that industrialists that favored trade protection recognized that their ability to influence the country’s trade policies had waned considerably during the 1990s (Abreu 2004; Pereira 2004). Given the erosion of their influence, it is not surprising that contributions provided by the owners with protectionist interests declined in candidate-­ centered Brazil in the previous decade, as predicted by the model and shown in table 6.4.

Trade Reforms and Electoral Fraud in a New Democratic Regime While the formal models in the theoretical chapters primarily focus on the interaction between political institutions and labor market conditions on trade protection, the model in chapter 2 also posits that the adoption of trade liberalization in new democratic regimes reduces the likelihood of electoral fraud. Brazil provides an interesting laboratory to assess this claim because it made a transition to democracy in 1985 and has therefore been a new democratic regime in the recent past. To assess the claim about trade reforms and electoral fraud in a new democracy by examining Brazil, I therefore begin with a brief historical background of electoral malpractices in the country. Interestingly and perhaps unfortunately, Brazil’s brief experience with democracy in the late nineteenth century was marred with periodic bouts of violence and systematic electoral fraud (Graham 1989). For instance, in his study of nineteenth century Brazil, Graham (1989) shows that political parties in the country missed no opportunity to impugn the behavior of

182 / Chapter Six

their rivals, manipulate electoral results, and use violence to win municipal, state, and central state elections. Democracy did not last long in Brazil, and the country was ruled by military dictators during much of the postwar de­ cades. In the closing decades of the twentieth century, however, the story of electoral fraud and malpractice in a newly democratic Brazil is dramatically different. To see why, consider the year 1985, in which Brazil held its first democratic election since 1962. There were some allegations of fraud in the 1985 election. In particular, some analysts in 1986 accused the military and the ruling party at the time (the PMDB) of orchestrating the 1985 election results to ensure that the leader of the PMDB, Tancredo Neves, would become president (Schneider 1991; Bowman 2003). These allegations were eventually withdrawn, even though there were some lingering doubts about electoral malpractices in the 1985 election. In contrast to the election in 1985, the presidential election of 1990, 1994, 1998, and 2002 in Brazil have been universally recognized as free and fair and completely devoid of fraud (Bowman 2003). To be sure, there have been some allegations and concerns about “vote buying” in the 1998 election. For example, an article published online by CNN on October 2, 1998, presented a simple story emphasizing that vote buying occurred in Brazil during the 1998 presidential election: Luciene Ramos dos Santos knew a year ago whom she would vote for in Sunday’s election—­whoever would buy her a refrigerator. Like many voters in Brazil’s poor Northeast, dos Santos doesn’t expect anything from the politicians who win national elections once they take office. So, she makes the most of their need for her now. “Candidates or their friends pass by and ask if we need anything. They write it down, and then we wait and see who comes through,” she said, standing outside her clapboard shack plastered with campaign posters. While Brazil has made strides in curtailing election fraud, vote-­buying is a time-­hallowed practice in the poor interior. Candidates offer everything from cash to coffins and vasectomies for votes. One popular gift is a new pair of shoes—­one shoe is given before the election and the other after, if the votes are delivered.39

In addition to “buying” votes, Brazilian politicians have also often been accused of taking bribes during presidential elections (Brusco et al. 2004). However, apart from some incidences of vote buying and corruption scandals during elections, few researchers doubt the fact that since 1990, election

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  183

results and elections in general have not been tampered with or subjected to fraud by political parties, including the ruling party (Bowman 2003). In fact, after the well-­known problems associated with vote counting in the 2000 US presidential election, the South Florida Sentinel Times published an article on November 5, 2002, extolling the complete absence of voting irregularities and electoral fraud in the 2002 presidential election in Brazil: “After the polls closed in an Oct. 27 runoff vote for Brazil’s president and many state governors, more than 90 percent of the ballots cast had been tabulated within a few hours, and the entire vote had been tallied less than 10 hours later. . . . There were no ‘dimpled chads’ and few accusations of electoral glitches. . . . ‘The reason the system works so well is because there is little room for error,’ said Armando Cardoso, a spokesman for Brazil’s Superior Electoral Tribunal” (Jones 2002). Therefore, put together, it is clear from Brazil’s recent political history that allegations and actual concerns about the practice of electoral fraud have decreased dramatically after the country made a transition to democracy in the mid-­1980s. Additionally, scholars of Latin American politics also emphasize the decline in electoral fraud and malpractices in Brazil compared to other new Latin American democracies (Lehoucq 2003; Brusco et al. 2004). What explains this decline in the frequency and thus likelihood of electoral fraud in a new democratic regime like Brazil? Brazilian policy analysts have often suggested that the introduction of automated voting machines—­particularly the IBM PC 80386 compatible, which is adapted to perform as a voting machine and is known as UE96—­ has prevented the occurrence of electoral fraud and vote rigging in Brazil.40 Although intuitive, this argument overlooks the fact that automated voting machines were only introduced as late as 1998. However, there were no reported incidences of electoral fraud or vote rigging in the two presidential elections prior to 1998—­namely, the 1990 and 1994 elections. Thus electoral fraud did not occur in the two elections prior to 1998, even in the absence of automated voting machines in Brazil, which raises doubts about any claim that focuses on the role of automated voting machines. Given the drawback of the claim summarized above, I propose here—­as predicted from the model in chapter 2—­that the adoption of trade liberalization in a new democracy like Brazil during the late 1980s and early 1990s was one of the key factors that dampened the likelihood of electoral fraud in the country. The historical evidence from Brazil during the immediate postdemocratic transition period indeed supports this claim. To see this in some detail, first note that analysts have suggested that contrary to conventional wisdom, the trade reforms—­which (as described earlier) was

184 / Chapter Six

the central part of the “heterodox” Cruzado economic plan—­adopted by the Sarney administration did generate economic growth and jobs (Maneschy et al. 2002). More important, this put an end to the economic stagnation, low growth, and high inflation that persisted prior to Brazil’s transition to democracy in the mid-­1980s. As a result, the political popularity of the Sarney-­led government soared significantly, and it won the first posttransition elections with relative ease (Ames 2001). As suggested by Pang (1989: 92), “Sarney secured the hugest electoral win in Brazilian history; the party he had just joined, Brazilian Democratic Movement Party (PMDB), won in 26 out of 27 states and in more than 3,000 municipalities.” The fact that Sarney’s party, the PMDB, won the posttransition election with ease (which resulted from the successful trade liberalization program under the Cruzado plan) was a critical factor that induced the ruling political elite in Brazil to not resort to voter intimidation, violence, and electoral fraud to win elections. Indeed, Brazil’s leaders at the time recognized that they did not need to resort to electoral fraud, given that they had “respected the rules of the game and had secured their power in a fully democratic electoral process” (Perreira et al. 2006: 53). Apart from the futility of resorting to electoral fraud, Brazilian political leaders from the PMDB during the immediate posttransition period were concerned that manipulating electoral results may jeopardize the consolidation of democracy in Brazil. This in turn would hurt their electoral prospects and would have potentially caused voters/citizens (who generally gained from trade reforms) to oppose the PMDB. In short, as predicted by the model in chapter 2, trade liberalization—­which generated benefits for Brazilian citizens and encouraged them to vote for the PMDB—­indeed induced politicians in Brazil’s new democracy to curtail electoral malpractices.

Conclusion An analysis of the relevant record plus data on trade politics and policy in Brazil from 1986 to 1990 supports the causal prediction that the level of trade barriers implemented by incumbents in new democratic regimes are biased toward the preferences of labor over capital and are also reduced if interindustry labor mobility is sufficiently high. Examination of the political economy of trade policy in Brazil since 1991 also reveals that the candidate-­ centered nature of Brazil’s democracy and growing labor market mobility encouraged its political leaders to further liberalize trade policies that favored the interests of workers rather than protectionist industry owners. Additionally, the Brazil case study provides some empirical bite to my causal

Democracy, Political Particularism, and Trade Liberalization in Brazil  /  185

argument that contributions from industry owners that prefer higher import barriers tend to reduce in new democracies and in candidate-­centered systems, particularly if they believe that politicians cannot credibly promise to change trade policies in ways that suit their (the owners’) preference. Finally, evidence garnered from primary and secondary sources shows that trade liberalization has helped decrease the likelihood of electoral fraud in Brazil. While the Brazil case study corroborates the book’s theoretical claims, it is worth mentioning here that numerous scholars suggest that the debt crisis in Brazil during the 1980s and country’s subsequent participation in International Monetary Fund (IMF) programs led to trade reforms in the 1980s (Amman and Baer 2002). A more careful examination of the timing of trade reforms in Brazil indicates that the debt crisis did not necessarily encourage Brazil’s leaders to reduce trade barriers. To see why, first note that the debt crisis that affected Brazil (and other Latin American countries) began around 1979 and reached its peak in 1982 (Amman and Baer 2002). Yet, as discussed in detail earlier, Brazil’s leaders began to earnestly reduce trade barriers only after the country’s transition to democracy—­that is, in 1986. Trade reforms in Brazil thus occurred almost four years after the country suffered from a debt crisis, which is substantial. Furthermore, the debt crisis in Brazil was resolved during the years—­specifically from 1986 to 1990—­in which ambitious trade reforms were launched in Brazil. It is therefore unlikely that Brazil’s debt crisis engendered trade liberalization in the country. Did participation in IMF programs lead to trade liberalization in Brazil? It is indeed true that Brazil chose to partake in and participated in a series of IMF programs from 1983 to 1985 (Galano 1994). It is also true that these IMF-­supported programs contained several conditions that Brazil had to implement to receive the IMF’s financial assistance. However, none of these conditions included liberalization of trade barriers. Rather, as summarized by Galano (1994: 339–­340), “ Brazil and the IMF agreed to several credit tranche purchases between 1983 and 1985.117 The IMF agreed to these credit tranche purchases only when the Brazilian government agreed to: (1) continue to devalue their currency, the Cruzeiro; (2) initiate massive cuts in domestic spending; (3) freeze all wages; (4) reduce most of government subsidized credit; and (5) curb state industries’ voracious appetite for foreign borrowing.” In other words, the IMF program conditions for Brazil (as indicated above) were almost entirely focused on fiscal and monetary policy and to a lesser extent financial sector conditions. Such conditions had little or no association with the Brazil’s trade policies. This suggests that it is unlikely that participation in IMF programs induced leaders in Brazil’s new democracy to reduce trade barriers in the late 1980s.

186 / Chapter Six

In sum, the extensive analysis of Brazil’s case is extremely useful for my empirical analysis but not sufficient. This is because the causal arguments from my models also have cross-­national implications. For instance, the formal model in chapter 4 predicts that across developing democracies, trade barriers and campaign contributions provided by industry owners to influence policy will be higher in party-­centered systems relative to candidate-­centered systems. To assess these (and other) cross-­national implications from the formal models, I examine the political economy of trade reforms in two party-­centered democracies in the developing world, India and South Africa, in the next chapter.

Seven

Trade Politics and Contributions in India and South Africa

The analysis of the Brazil case in chapter 6 showed how politics in a new democratic regime (Brazil between 1986 and 1991–­92) and in a candidate-­ centered system (Brazil from 1991–­92 onward) affects trade policies as well as campaign contributions. Yet focusing on just Brazil to assess the theoretical arguments in this book is insufficient. This is because the theoretical models presented in the earlier chapters of this book not only focus on how new democratic regimes affect trade barriers in developing states (chapter 2) but also explore the impact of party-­centered systems on trade restrictions and contributions by the owners in developing states. More specifically, the causal story presented from the model in chapter 4 predicts that in developing democracies, party-­centered systems encourage lobbying for protection by the owners of industries who have incentives to do so. Consequently, the contributions that the owners (capital) provide to politicians in party-­centered systems increase and are also higher than contributions in candidate-­centered systems. The model also posits that in party-­centered democracies, the trade policy decisions of party leaders are biased toward the protectionist interests of the owners rather than the policy preferences of  workers. Are the causal predictions mentioned above empirically valid? In this chapter, we attempt to answer this question by examining the political economy of trade protection and campaign contributions in two prominent democracies in the developing world: India and South Africa. The substantive and methodological rationale for selecting the India and South Africa cases was described in the previous chapter and is hence not repeated here. The remainder of this chapter is organized as follows. The first section provides a brief background of India’s economy, the country’s political institutions, and the history of trade protection in India. I employ both historical

188 / Chapter Seven

evidence and quantitative analysis of within-­country data from India to assess the effects that party-­centered systems have on tariffs and contributions by owners that favor trade protection. After analyzing the Indian case, I turn to examine South Africa. The analysis of the South African case begins by exploring the country’s economy and political institutions and its history of trade protection. Historical evidence and quantitative analysis of within-­ country data from South Africa is then employed to evaluate how the transition to democracy combined with interindustry labor mobility to influence trade policy in South Africa and how the country’s party-­centered system affects trade politics and contributions by protectionist interests (i.e., the owners). While doing so, I examine how trade liberalization has helped reduce the likelihood of electoral fraud in South Africa. The chapter concludes with a brief summary of the main findings presented from the two cases.

Political Parties, Trade Politics, and Contributions in India Economy and Political Institutions India (which obtained independence in 1947) is the world’s most populous democracy and is one of the most stable democracies in the developing world (Lijphart 1996; Dasgupta 2001; Kohli 2001). The economy of India is the twelfth largest economy in the world in nominal GDP terms and the fourth largest by purchasing power parity (CIA 2009). However, in 2013, India’s per capita income in PPP terms was $4,077, while its per capita income in nominal terms was US$1,504 (IMF 2013). From 1947 to 1991, India was a closed economy “with high tariff walls, strict rules for trade flows, and low trade volumes” (Sinha 2007: 1187). Ma­ jor industries in India’s economy include telecommunications, textiles, chemicals, food processing, steel, transportation equipment, cement, mining, petroleum, and machinery, as well as agricultural products such as rice, wheat, oilseed, cotton, jute, tea, sugarcane, potatoes, cattle, poultry, and fish (Library of Congress 2004). In 2009, the labor force in the country had a total of half a billion workers, therein making it one of the largest labor markets in the world (IMF 2009).1 India’s key democratic institutions, including separation of powers as well as the occurrence of regular elections at both the national and state level, is firmly entrenched (Kohli 2001; Ganguly et al. 2007). Hence, notwithstanding the authoritarian episode of 1975–­77, the prospects of a democratic reversal in India are remote. In short, India is an established democ­ racy. Further, in the post–­World War II era, India’s democratic institutions

Trade Politics and Contributions in India and South Africa  /  189

predate those of Brazil and South Africa, as the latter two countries recently experienced a transition to democracy. Unlike Brazil, the level of political particularism in India is low; in addition, India is a party-­centered democracy (Sridharan 1999, 2009; Yadav 2005, 2011; Chibber and Kollman 2009). Three institutional features of India’s democracy ensure that the country is a party-­centered rather than a candidate-­centered system. First, representatives in the Indian parliament are elected via a single member district (SMD) plurality electoral system. As stated by Hix (2004: 197), in SMD systems that employ plurality rule (which is prevalent in India), rank-­and-­file party members have no incentives to cultivate their personal vote since these systems “do not allow candidates to make direct appeals against rival candidates from their own party. (rather) the personalities of party leaderships has a significant impact on the electoral fortunes of the candidates in each constituency.(thus) candidates have incentives to support their parties’ positions.” The role and behavior of party leaders in SMD plurality systems have a significant impact on the electoral fortunes of the candidates in each constituency, which promotes intraparty unity and allows party leaders to exert a high degree of control over their rank-­and-­file party members (Carey and Shugart 1995; Hix 2004; Carey 2007). Therefore, the SMD plurality system that is used in India generates a party-­centered democracy. Second, legislative rules in India regarding the introduction of bills and amendments, committee recommendations for bills, and questions and answers on the legislative floor provide party leaders with additional leverage to control individual members within their own party. This is pointed out by Yadav (2005: 23), who states that in India, “the procedural flow of legislation on the floor and through committees, recognition of members for introducing bills, for urgent and zero hour questions, are all at the discretion of the leadership of the party controlling this chamber. Parties not in the ruling coalition have no formal say in the flow of legislative business. Any consultations are conducted at the level of party leaders and rank and file members have no say in these consultations. Individual members must go through party leaders to gain any personal considerations on policy or legislation.” Third, in February 1985, the government of India implemented the Anti-­ Defection Act. This act constitutionally bans defections of individual party members from one party to another during elections and immediately after elections (Manor 1988; Sridharan 1999). This “sharpened party bound­ aries, . . . strengthened party leadership and . . . the centralization of candidate selection for elections by party leaders” (Sridharan 1999: 1206). The

190 / Chapter Seven

Anti-­Defection Act in India thus also fosters a party-­centered system, as it deters individual party members from cultivating their personal vote and allows leaders to more firmly control their party. Trade Policy in India From 1951 to 1991, Indian policy makers followed an import-­substitution-­ industrialization (ISI) model of development, which was accompanied by extensive regulatory controls over the economy (Wadhva 1994; Sinha 2007). Successive governments in India assiduously implemented ISI policies, and as a result, the level of trade protection in India was one of the highest in the developing world. This high level of trade protection is illustrated in figure 7.1; the solid line in this figure shows the moving average of the unweighted tariff rate in India from 1978 to 2003. We find in this figure that from 1978 to 1990, the moving average of the unweighted tariff rate in India was an astoundingly high at 130%.2 Some steps were taken toward trade liberalization by Prime Minister Indira Gandhi in 1979 and later by Prime Minister Rajiv Gandhi in 1985 (Wadhva 1994; Dutta 1998). But unlike the initial posttransition years in Brazil and South Africa (described later), these attempts at trade reform were abruptly and rapidly reversed in India’s established democracy (Jalan 1991; Sachs et al. 1991). After the 1990 Gulf War, however, the Congress-­led government in India implemented some far-­reaching trade reform measures that led to a decline in tariff rates. This is indicated in figure 7.1, which shows that India’s moving

140 120

Percentage

100 80 60 40 Unweighted tariff 20

Core NTBs

0

1978

1981

1984

1987

1990

1993

1996

1999

Figure 7.1  Moving average of unweighted tariffs and core NTBs in India

2002

Trade Politics and Contributions in India and South Africa  /  191

average unweighted tariff rate decreased significantly from 128% in 1990 to less than 56% by 1995. Similarly, the dashed line in figure 7.1, which illustrates the moving average of the core NTBs ratio for India (in percentage terms) from 1987 to 2003, shows that core NTBs declined from 68% in 1989 to around 35% by 1995.3 This wave of dramatic trade liberalization and reduction of trade barriers did not last for long, however. Indeed, fig­ ure 7.1 shows that during the second half of the 1990s, the moving average of the unweighted tariff rate and the core NTBs ratio increased, although this increase was not dramatic. India, therefore, entered a “protectionist phase between 1996 and 1998 (that) seemed to reverse some of the directional shifts of the early 1990s” (Sinha 2007: 1192). Thus the pattern of trade protection in India from 1978 to 2003 can be divided into three periods. In the first period (between 1975 and 1990), India was the archetypical import substituting regime, with the highest average tariff rates in the world. However, the second period (between 1991 and 1996) was characterized by some trade liberalization. Finally, in the third period (from 1997 to 2002), India again entered a “protectionist phase” as trade restrictions were increased. What explains this peculiar trend in trade policy in India from the late 1970s to the early years of this decade? We turn to address this question in some detail in the following section. The Politics of  Trade Protection in India A key causal claim in the model in chapter 3 is that policy makers in party-­ centered democracies are biased toward the interests of owners that favor trade protection when designing and implementing trade policies. Interestingly, a careful analysis of the political determinants of trade protection in India provides empirical support for the aforementioned causal claim in the model. To see why, first consider the 1978 to 1990 time period, when the moving average tariff level was as high as 130%. During this period, the degree of interindustry occupational mobility of labor in India increased steadily and was fairly high. This is shown in figure 7.2, which illustrates the annual series of the level of interindustry occupational mobility of labor and the elasticity of interindustry labor mobility across twenty-­nine industries listed in table 7.1 at the three-­digit International Standard Industrial Classification (ISIC) level in India from 1978 to 2003.4 Although relatively high interindustry labor mobility may have potentially given policy makers in India more “political maneuvering room” to relax trade restrictions—­since occupationally mobile workers in developing countries typically prefer trade reforms5—­we find that the moving average

ILM-elasticity

12

4.5

10

3.0

8

2.5

6

2.0

4

1.0 ILM ILM-elasticity

2 0

0.5

Inter-industry labor mobility (ILM)

Trade reforms (1991-92)

14

0 1978

1981

1984

1987

1990

1993

1996

1999

2002

Figure 7.2  Labor mobility in India

Table 7.1. List of industries (three-­digit ISIC level) in India and South Africa Agriculture products, not processed Beverages Electrical machinery, appliances Fabricated metal products Finance, insurance, real estate Food products Footwear Furniture, fixtures, except primary metal Glass and glass products Iron and steel Leather products Mining and quarrying Miscellaneous petroleum and coal products 1 Nonelectrical machinery Nonferrous metal basic industries Other chemicals Other nonmetallic mineral products Paper and paper products Pharmaceuticals Plastic products Pottery, china, and earthenware Professional and scientific equipment Rubber products Textiles Tobacco Transport Equipment Wearing apparel, except footwear Wholesale and retail trade, repair Wood, wood and cork, except furniture Note: “Not processed agricultural products” include raw food materials such as rice, wheat, soybeans, corn, and sugar.

Trade Politics and Contributions in India and South Africa  /  193

of unweighted tariffs was extremely high in India between 1978 and 1990. Why is this so? A systematic analysis of the relevant historical evidence shows that the party-­centered nature of India’s democracy encouraged Indian policy makers to favor the interests of industry owners that preferred trade protection between 1978 and 1990 when designing trade policies. To begin, recall my argument that in a party-­centered system like India, party leaders need to provide financial side payments to rank-­and-­file party members to “buy” their loyalty and maintain intraparty unity. Providing these financial side payments is, however, not costless. Moreover, the daily political expenses borne by party leaders that involve offering such side payments require a substantial amount of financial resources. This is emphasized by a well-­known Indian political economist, Arun Kumar (2002: 228), who states that “state funding of elections will not eliminate the need for money, black or white, to run Indian politics in its present form. In the interviews with politicians it became clear that they need funds everyday and not just when elections come . . . Since state funding cannot match the needs of the political parties. . . . (the) political parties will continue to supplement funds through other means.” In fact, to finance the side payments mentioned above, party leaders actively solicit contributions particularly from the owners of industries that have access to large amounts of capital. As emphasized by Kumar (2002: 144), “By and large, political leaders admitted to having a nexus with big business, local industry and trade associations. They get money and help in kind from them not only for elections but also for day to day running. They admit to a quid-­pro-­quo for the help.” A key question emerges from Kumar’s (2002) insightful analysis: Which industries constitute the “big business” groups that have a mutually symbiotic relationship with the political parties in India? As an answer to this question, Mazumdar (2008) suggests that large business houses, which in­ cludes owners of manufacturing industries such as automobiles, motorcy­ cles, chemicals, and iron and steel—­who have consistently favored trade pro­ tection since the 1960s—­constitute the most influential “big business” group that shares a mutually compatible relationship with political parties. Based on a report on the influence business groups that was (ironically) conducted by the government of India, Mazumdar (2008: 18) concludes that during the 1970s and 1980s, “a narrow group of large businesses have acquired significant influence with the State, and are able to secure major benefits through such influence.” He also points out that the symbiotic relationship between large (industry-­based) business houses and successive governments in India was “the direct result of cronyism in the economic domain—­the control and influence of large business houses . . . and the

194 / Chapter Seven

deliberate manipulation by large business houses of the licensing mechanism and policies related to trade restrictions” (Mazumdar 2008: 18). What were the consequences of this crony-­capitalist relationship between big business groups and political parties, particularly with respect to trade policy? An immediate consequence was that the protectionist interests of big business groups were prioritized over and above the interests and welfare of the public, particularly in the area of trade policy. This has been suggested in an early study of business-­government relations in India by Goyal (1979: 19), who suggests that “the primary factor responsible for the high levels of trade protection” in India during the 1970s “has been the exercise of ‘economic power’ by influential big business houses of India. The relationship between the business houses and the ruling political parties has been close; and the associations have been obviously to mutual advantage—­at the cost of public interest and in disregard of the constitutional obligations and contrary to oft repeated public pronouncements by the leadership.” More remarkably, an internal report prepared by the Planning Commission (a government organization in India) in 1983 also noted that trade policies in India “favored the protectionist interests of big business houses instead of promoting the welfare of the public at large . . . there is a growing concern that trade and capital control policies are dictated by vested protectionist interests . . . this has come at the expense of implementing policies that enhance growth that benefits the citizens.”6 The Planning Commission report mentioned above and Goyal’s (1979) finding is important. This is because it supports the causal prediction from the model in chapter 4, which suggests that political parties in party-­centered systems will be more concerned about the protectionist interests of the owners than the preferences of the general public, including workers. The second consequence of the crony-­capitalist relationship alluded to above is that it not only allowed business groups to directly intervene in the policy-­making process but also gave them an opportunity to substantially influence the actual content of trade policies. In fact, in an interview held in 1992, the director of ASSOCHAM (the Associated Chambers of Commerce and Industry of India) candidly admitted that ASSOCHAM directly intervenes in the design and implementation of trade policies in India: We have consistently had a policy of working with the civil service, including the ministry of Commerce, that is there at the centre. We find that they are the administration, they are the prominent people there. And, therefore, we feel that while we must address the political level . . . we also provide feedback for

Trade Politics and Contributions in India and South Africa  /  195 developing trade and other external economic policies . . . and we do that . . . on a continuous basis . . . There is a continuous dialogue and interaction with the bureaucracy at all levels . . . We look at it this way: development is a partnership process between government and industry. And we are the junior partner of the government.7

The fact that government leaders in a party-­centered system like India prioritized the protectionist interests of certain business groups and gave these groups unfettered access to the process of trade policy making had a dual impact. For one, as discussed in detail below, it not only encouraged successive governments in India during the 1970s and 1980s to solicit contributions from these probusiness groups but also induced these groups to provide substantial contributions to parties to influence policy (Kochanek 1985, 1987; Jenkins 1999; Sinha 2005). Second, despite some halfhearted steps taken toward trade liberalization by Prime Ministers Indira Gandhi and Rajiv Gandhi in the 1980s,8 policy makers acquiesced to the protectionist demands of business groups in both the 1970s and the 1980s (Harris 1987; Wadhva 1994). This is suggested in an interview with a retired official from the Ministry of Commerce who revealed that in the 1970s and 1980s, the administration in New Delhi responded favorably to the protectionist demands of vested interest groups and that this helped sustain high levels of trade protection in India: “Before 1991, the Congress and the Janata-­led government were often sympathetic to the protectionist sentiments of large industrial houses in India. Much, if not all, of this sympathy stemmed from a mutually dependent relationship between big manufacturing industries and the government. . . . in exchange for financial help from industry, the government in New Delhi agreed to protect industry from import competition and to not open the domestic market to foreign goods.”9 Thus there is little ambiguity in the fact that the party-­centered nature of Indian democracy helped foster close ties between protectionist business groups and successive governments in New Delhi before 1991. This in turn ensured that the level of trade protection remained consistently high from 1978 to 1991, as illustrated in figure 7.1. However, figure 7.1 also shows that both tariffs and NTBs were reduced from 1991 to 1995. The “rush to trade reforms” in India in the early 1990s is surprising, given that political parties in a party-­centered democracy like India are, as discussed above, susceptible to protectionist pressures from industry-­based interest groups. Why did Indian policy makers adopt trade re­forms by drastically cutting trade barriers in 1991? In India’s case, it is important to note that it was neither political factors related to its domestic

196 / Chapter Seven

institutions nor economic factors such as the interindustry occupational mobility of labor that led to trade liberalization. For example, throughout 1980s, 1990s, and beyond, India remained an established democracy with a party-­centered system (Kohli 2001; Ganguly et al. 2007). Therefore, it is not plausible to suggest that a change in domestic political institutions engendered trade reforms in India. Moreover, as illustrated in figure 7.2, the level of interindustry labor mobility in India increased in the 1980s but remained virtually flat from 1990 to 1993. Thus we cannot attribute the adoption of trade reforms in India to rising labor mobility either. What, then, explains India’s sudden enthusiasm for trade reforms in 1991? The answer to this question lies in the fact that India launched its trade reforms in 1991 under the pressure of a twin economic crisis (Jalan 1991; Sachs et al. 1991; Wadhva 1994; Joshi and Little 1997). The twin crisis involved an unmanageable balance of payments and a socially intolerably high rate of inflation that was building up in the 1980s and reached its peak in 1990–­91. The current account deficit as a percentage of GDP peaked at a high of 3.1%, compared to an average of 1.4% during the 1980s. The inflation rate (as measured by point-­to-­point changes in the wholesale price index) had also climbed to the politically dangerous double-­digit level, hitting 12.1% in 1990–­91. The central government’s fiscal deficit alone peaked at 7.9% as a percentage of GDP in 1989–­90, which was unsustainable. Last but not the least, foreign-­exchange reserves dwindled to a low of US$2.2 billion, with less than fifteen days’ cover against annual imports. India stared bankruptcy in the face as it struggled to meet external debt obligations.10 With the help of a balance of payments loan facility from the International Monetary Fund (IMF), the Indian government in 1991 chose to solve the twin economic crises by launching economic, including trade, reforms (Sachs et al. 1991; Joshi and Little 1997). The Rao government did so because it recognized in 1991 that it could promote investment and growth in India by reducing trade barriers (Joshi and Little 1997; Rajan and Sen 2002; Sally 2008). Hence the Indian government reduced trade barriers between 1991 and 1994, as illustrated in figure 7.1. After launching the relatively rapid trade reforms, however, the Rao government was soon confronted with the political reality of raising funds to finance the ruling Congress party’s campaign for the 1994 state elections (Kochanek 1995, 1996; Pedersen 2000). To raise the funds for the state-­ level elections and to sustain the loyalty of rank-­and-­file party members, the Rao government in a party-­centered system like India—­as predicted by the

Trade Politics and Contributions in India and South Africa  /  197

model in chapter 4—­turned to business groups, particularly to large industrial houses, for financial support (Kochanek 1995, 1996; Jenkins 1999; Hardgrave and Kochanek 2007). Business groups in India that largely included the owners of industries such as automobiles, motorcycles, and chemicals reacted to the government’s request for financial support in two ways. First, they formed an organization called the Bombay Club to voice their concerns against trade reforms (Kochanek 1995, 1996; Jenkins 1999; Hardgrave and Kochanek 2007). The Bombay Club consisted of a group of prominent Indian industries that includes Bajaj industries, the Birla group, the Tatas, and Reliance industries. Second, the Bombay Club bluntly asked the Rao government for a reversal of trade reforms and more protection of their industries from import competition. As stated by Hardgrave and Kochanek (2007: 463), The reforms have not enjoyed the unified support of the business elite or India’s apex business associations. While most industrialists and associations welcomes the move toward deregulations and deregulation of the domestic private sector, they were far less enthusiastic about reducing tariff protection for Indian industry, the opening of the Indian economy to foreign trade, and investment and globalization of the Indian economy. Business resistance to trade reform crystallized in late 1993 as the immediate economic crisis began to ease. The initial attack came from members of the Bombay Club, which demanded a level playing field. While each of India’s major apex associations applauded the Bombay’s club for a level playing field, the strongest endorsement of the Club’s policy critique came from the FICCI.

Hardgrave and Kochanek’s (2007) claim is reflected in an interview of a prominent Indian industrialist (Gaurav Swarup) by an Indian magazine Frontline later in 1999. In the interview, Swarup emphasized that in 1994 he personally requested the Rao government to stop trade reforms and prevent the entry of multinational corporations (hereafter MNCs) into the Indian domestic market: “It is necessary to regulate the entry of MNCs in not just industrial sectors, but also nonindustrial sectors. If the entry of MNCs is not regulated . . . then Indian industry will not be allowed to grow. There are some sectors that are relatively new, but have a lot of potential. If MNCs operating in these sectors enter, the Indian companies in those sectors will simply be lost. They have to be given time to become global players.”11 How did the Rao government respond to the Bombay Club’s request for trade

198 / Chapter Seven

protection and a reversal of trade reforms? Did it ignore or belittle the concerns of the Bombay Club as President Collor had done to the São Paulo Federation of Industries (FIESP) in 1993 in a candidate-­centered system like Brazil? Interestingly, as suggested by my formal model in chapter 4, the government in a party-­centered democracy such as India responded favorably to the preferences and demands of the owners of industries that were resisting trade reforms. It did so in three ways. First, both the Rao government and its successor (the Deve Gowda and the Bharatiya Janata Party government) consistently raised tariffs from 1995 to 1999 for several industries, including those that requested for more trade protection. Not surprisingly, as shown in figure 7.1, tariffs and NTBs increased from 1995 to 2000. Second, Indian diplomats at the World Trade Organization (WTO) were reportedly instructed by the government to not concede to further demands for trade liberalization by other countries (i.e., the advanced democracies) at Geneva in order to protect Indian industry from import competition.12 Third, the government at the time officially declared that slowing down trade reforms and increasing trade barriers was necessary to give Indian industry a “level playing field” to compete against imported goods. This was suggested in a report on economic reforms brought out by the Indian Ministry of Finance in 1995: “The fundamental objective of economic reforms is to bring about rapid and sustained improvement in the quality of the people of India. . . . This can be best achieved by allowing Indian industry to grow so that it promotes private investment. To promote Indian industry, it must be protected from excessive import competition and should also be provided with a level playing field” (Ministry of Finance 1995: 43). Thus the historical evidence from India presented above supports the causal claim from chapter 3, which suggests that parties in party-­centered democracies (e.g., India) tend to be biased toward the protectionist interests of the owners that seek trade protection. But is there quantitative support for the implications of the aforementioned causal claim? Specifically, if it is indeed the case that governments in party-­centered democracies are more receptive to the protectionist interests of certain business groups—­as suggested by my model and as shown in the case analysis above—­then it is plausible that the protectionist demands of these business groups (if successful) may positively influence tariffs in a party-­centered system like India. Likewise, if governments in party-­centered systems pay less attention to the preferences of labor (which is also suggested by my model), then it follows that the interindustry occupational mobility of workers that determines their trade policy preferences may have a negligible impact on

Trade Politics and Contributions in India and South Africa  /  199

trade protection in a party-­centered system such as India. Are these two implications empirically valid? Analyzing Within-­Country Data from India To answer the question in the preceding paragraph, I gathered a panel data set of industry-­year tariffs on an annual basis from 1978 to 2005 for the twenty-­nine industries for which tariff data are available at the three-­digit ISIC level in India. These twenty-­nine different industries are listed in ta­ ble 7.1. The industry-­year tariff rate for India, which is simply labeled as tariff-­India, serves as the dependent variable for the statistical analysis. 13 Since data for NTBs are not available for India at the industry-­year level, the statistical analysis focuses on assessing the implications mentioned above in the within-­country data for industry-­year tariffs from India. The first key explanatory variable in the model is an annual measure of interindustry labor mobility (labeled as ILM) for India.14 I used the formula in equation 4.1 in chapter 4 to calculate the occupational mobility of workers in twenty-­nine different industries (listed in table 7.1) on an annual basis from 1978 to 2005 in India; this calculated measure is illustrated in the solid line in figure 7.2. Second, as done in the Brazil case in chapter 6, I used the import penetration ratio measure for India to assess quantitatively whether tariff policies implemented by the government are receptive to the protectionist interests of the owners of some industries. More specifically, I use the annual series of the import penetration ratio across twenty-­four industries at the three-­digit ISIC level in India as a proxy to capture the protectionist interests of the owners of these industries. The import penetration ratio is calculated per year by dividing industry imports by the sum of industry outputs plus imports (the denominator is industry supply) across the twenty-­nine industries in India listed in table 7.1.15 This measure is calculated for twenty-­nine industries, as data to operationalize the import penetration ratio at the three-­digit ISIC level are only available for these industries. The measure is labeled as import penetration ratio. Apart from these two explanatory variables, I follow extant studies on trade protection in India16 and add the following variables to the specification: capital-­output ratio, investment-­output ratio, log GDP17, and the lag of the dependent variable. The dummy variable, bop crisis (where bop denotes “balance of payments”), is added to the specification and coded as 1 if the level of forex reserves (in India) falls to less than the equivalent of three months’ worth of imports; it is coded as 0 otherwise. Finally, a time trend is included in the specification. The specification is estimated via the system–­generalized

200 / Chapter Seven Table 7.2  System-­GMM results for tariffs in India Model 1 lag dep variable ILM import penetration ratio log GDP capital-­output ratio investment-­output ratio bop crisis time trend Constant Sargan test AR (1) AR (2) N

0.634*** (0.072) –­0.122 (0.927) 0.250* (0.075) –­0.048 (0.156) 0.162 (0.189) 0.322 (0.475) –­0.07*** (0.025) –­0.009 (0.022) –­1.22*** (0.233) 0.21 –­1.43** –­0.515 579

Note: ***,**, and * denote significance at the 1%, 5%, and 10% levels, respectively. The results reported for the system-­GMM models are from the one-­step estimation, except the Sargan test and the AR(1) as well as the AR(2) tests, which are taken from the second-­step estimation. A negative and statistically significant AR(1) term plus a statistically insignificant AR(2) term indicates no serial correlation (Blundell and Bond 1998).

methods of moment (system-­GMM) approach, which helps correct for potential endogeneity problems. Model 1 in table 7.2 reports the results from estimating the effect of ILM and import penetration ratio on the dependent var­ iable, tariff-­India. The estimated effect of bop crisis is negative and highly significant at the 1% level in model 1. This is not surprising, given that a serious balance-­of-­ payments crisis engendered a dramatic reduction in trade barriers in India in 1991. In contrast, the estimated effect of ILM on tariff-­India is negative and statistically insignificant.18 This insignificant effect broadly suggests—­as indicated in my discussion of the India case and as implied by my model (stated above)—­that in a party-­centered system like India, incumbents are less likely to be receptive to the trade policy preferences of labor.

Trade Politics and Contributions in India and South Africa  /  201

In sharp contrast, the estimate of import penetration ratio is positive and highly significant in model 1. The positive and significant coefficient of import penetration ratio statistically corroborates a key implication from my causal argument, suggesting that in a party-­centered system such as India, trade barriers will be strongly influenced by the interests of protectionist industry owners. In short, the qualitative evidence discussed earlier shows that the party-­centered nature of India’s democracy affects business-­government relationships and thus influences trade politics. The quantitative evidence from within-­country data on tariffs from India, in turn, reveals that the protectionist interests of the owners of industries also influences trade policy outcomes. I turn to analyze the empirical validity of my causal claims about contributions in party-­centered democracies. Political Parties and Industry Contributions in India In the previous section, we presented some evidence showing that the party-­ centered nature of India’s democracy drives political parties in the country to solicit contributions from industry and that this encourages the parties to implement trade policies that are biased toward the preferences of industry owners that favor trade protection. But is there empirical support in the India case for the causal claim that the owners of industries provide substantial contributions to parties/politicians in party-­centered systems because they rationally anticipate that party leaders are likely to be more receptive to their (the owners’) trade policy interests in these systems? Furthermore, are the contributions provided by industries to influence policy in party-­centered systems higher than the contributions offered in candidate-­ centered systems? Answering these questions by conducting a direct empirical test on the size of contributions provided by different industries in India is not possible. This is because data on the amount of contributions offered by industries, business groups, or corporate houses in India are not available, unlike in Brazil. I therefore use qualitative evidence and data on the perceived level of contributions from the World Economic Forum’s (WEF; 2004b) Executive Opinion Survey (EOS) database for India to empirically assess the second causal claim mentioned above. With respect to the historical evidence, existing studies on business-­ government relations in India indicate that India’s party-­centered system facilitates a quid-­pro-­quo relationship between the owners of industries (e.g., business groups and corporate houses) and the political parties. In

202 / Chapter Seven

this quid-­pro-­quo relationship, the latter (the parties) “deliver” the policies requested by the owners in return for their contributions (Kochanek 1995; Sinha 2005; Yadav 2005, 2011). This is emphasized by Sinha (2005: 20), who posits that “India’s regulatory system and the business-­politics relationship were analogous to a political exchange: in exchange for specific favors, business actors supported governments and politicians. The government’s role was overly regulative and the business associations sought only particularistic benefits for their firms or sectors abandoning any governance or developmental activities. Thus, India came to be invoked as a classic ‘rent-­seeking society’ and its business-­state pattern characterized by ‘embedded particularism. ’” In a recent study on campaign contributions in India, Yadav (2011) conducted a clustered, randomized survey of industry owners to assess the extent to which contributions were solicited by political parties and (subsequently) offered by corporate houses in the country. To this end, she conducted several interviews with industrialists and executives in India. Her interviews revealed that party leaders in India’s party-­ centered democracy frequently requested the owners of industries for contributions to finance expenditures for their party. The industrialists, in turn, voluntarily provided the contributions in order to influence policy, including trade policy outcomes. As emphasized by Yadav (2011: 126), In India, the strategic rationale for lobbying parties at both the legislative and electoral stages was articulated very clearly in interviews. For example, the expectation of a legislative quid pro quo was emphasized by the director of a manufacturing sector chamber, “At election time, the party leaders come or they call and ask for so much. How can you say no? The local MPs are of no use unless they are party leaders, you have to make the parties happy.” . . . The chairman of another influential chamber in the service sector stated that “policy issues, bureaucratic hurdles, legislation . . . this is all handled by parties. You cannot expect something for nothing, so you must pay when they ask. Election campaigns are important but donations are needed at all times . . . it helps to influence economic, trade and tax policies.”

Finally, a law report released by the Supreme Court of India in 1999 bluntly pointed out that business houses and corporate groups frequently gave large donations to political parties in India and that this was done to influence both tax and trade policies: In the very scheme of things and as pointed out by the Supreme Court in its various decisions, substantial funds contributed to political parties would

Trade Politics and Contributions in India and South Africa  /  203 come only from business houses, corporate groups and companies. Such a situation sends a clear message from the political parties to big business houses and to powerful corporations that their future financial well being will depend upon the extent to which they extend financial support to the political party. Indeed most business houses already know where their interest lies and they make their contributions accordingly to that political party which is likely to advance their interest . . . the contributions that are provided by business houses to parties are done to influence a whole range of policies pertaining to excise taxes, tariffs, and industry subsidies. (Law Commission of India, 1999)

In addition to the qualitative evidence described above, I use the WEF’s EOS data (2004b) to broadly assess the causal predictions about contributions in party-­centered systems. To this end, I conduct two empirical exercises. First, the WEF’s EOS database for Brazil and India is employed to extract data on the degree of contributions/donations given to politicians, as perceived by industry executives, in both these countries. Second, I use the WEF EOS data for Brazil and India to analyze the extent to which donations to politicians influence policy—­including trade policy—­outcomes, as perceived by industry executives in each of these two countries. Doing so allows one to analyze whether contributions offered to influence policy in a party-­centered democracy like India are higher than the contributions offered in a candidate-­centered system like Brazil, as predicted by the model in chapter 3. In chapter 4, I pointed out that the WEF has gathered data on perceptions of donations/contributions given to political parties and candidates based on responses to survey questionnaires completed by industry executives within several developed and developing countries. Specifically, in the previous decade, the WEF conducted a mass survey of executives from 18 different industries19 in as many as 104 countries. The survey that was given to the executives in each country contained several questions. However, two questions in the EOS survey are relevant for the comparative analysis of contributions in Brazil (candidate-­centered system) and India (party-­centered system). These two questions are as follows: ·· Question 1. EOS Q4.13, “How common are donations to political parties and candidates in your country?” Each response to this question is ranked on a 1–­7 ordinal scale, where 1 denotes that donations never occur and 7 indicates that providing donations to political parties and candidates is common.

204 / Chapter Seven ·· Question 2. EOS Q4.14, “To what extent do legal contributions to politi­ cal parties and candidates have a direct influence on public policy out­ comes?” Each response to this latter question is also ranked on a 1–­7 ordi­ nal scale, where 1 denotes “little direct influence on policy” and 7 indicates “very close link between donations and policy.”

As indicated in the aforementioned questions, executives from eighteen different industries in each country were asked to rank on a 1–­7 scale their perception of the extent to which providing donations to political parties and candidates are common in their country. They were also asked to rank on a 1–­7 scale whether they believe that the donations/contributions that they provide have a direct influence on policy outcomes. Note that the reference to policy outcomes in question EOS Q4.14 includes trade policy as well. The responses given by industry executives in each country for the two aforementioned (and other) questions in the EOS survey were reported and published for each country by the WEF in the Global Competitiveness Report in 2004. A close examination of the responses for each of the two questions posited above by industry executives in Brazil and India shows that in the 2004 WEF sample, there exists a total of eighty-­nine responses by industry executives to questions EOS Q4.13 and EOS Q4.14 in Brazil and eighty-­one responses by industry executives to these two questions in India. The sizes of the WEF samples of survey responses in Brazil and India are thus similar. Using the WEF (2004b) EOS data for Brazil and India, I first calculated for each country the mean score on the 1–­7 scale for both the questions mentioned above. The mean score on the 1–­7 scale for both questions in each country is reported in table 7.3. Panel A in table 7.3 shows that the mean score for question EOS Q 4.13, which deals with the frequency of donations given to political actors as perceived by industry executives, is 3.8 (on a 1–­7 scale) in Brazil but as high as 5.4 in India. Additionally, as reported in panel A, the difference-­of-­ means test shows that the average score for question EOS Q 4.13 for India is statistically (at the 1% level) higher than the mean score for the same question for Brazil. Panel B of table 7.3 reports that the mean score for question EOS Q 4.14, which is concerned with perceptions about the extent to which donations influence policy outcomes, is 3.4 in Brazil and a substantially higher 4.9 in India. A difference-­of-­means test reveals that the average score for question EOS Q 4.14 for India is also statistically (at the 1% level) higher than the mean score for the same question for Brazil.

Trade Politics and Contributions in India and South Africa  /  205 Table 7.3  Contributions in Brazil and India

Question

Mean for Brazil

Mean for India

Difference of means in column (a) and (b); (p-­value)

Panel A

(a)

(b)

(c)

EOS question 4.13: How common are donations to politicians in your country? 1 (never occurs) to 7 (common)

3.8

5.4

1.6 (0.0)

3.4

4.9

1.5 (0.0)

Panel B EOS question 4.14: To what extent do legal contributions to politicians have a direct influence on specific public policy outcomes? 1 (low influence) to 7 (high influence)

The descriptive statistical results reported above for Brazil and India are encouraging. At a more general level, these results suggest, as anticipated by the theoretical story in chapter 4, that contributions offered by the owners of industries to political actors in party-­centered democracies like India are higher than the contributions provided in candidate-­centered systems like Brazil. Moreover, the results reveal that unlike their counterparts in Brazil, industry executives in India perceive that the contributions they offer are more likely to substantially influence economic policy, including trade policy, outcomes. This latter finding provides some confirmation of the claim from the model in chapter 4 that the owners of industries in party-­centered democracies are more likely to believe that their contributions to political actors will indeed have the effect of influencing policy in a way that is compatible with their interests. Yet it is important to observe that the descriptive statistics presented in table 7.3 are not based on data on the amount of contributions given to candidates from parties in order to influence trade policy. Rather, the results from the WEF (2004b) sample are extracted from survey responses about the perceived frequency of donations given to political actors by industry executives in Brazil and India. Notwithstanding these limitations, however,

206 / Chapter Seven

when combined with the anecdotal analysis presented earlier in this section, the analysis of the WEF EOS data (2004b) for Brazil and India provides some evidence for my causal arguments derived from the model in chapter 4. I turn to examine the South Africa case.

Analyzing the South African Case South Africa’s Economy and Political Institutions The size of South Africa’s economy in 2013 was equal to $350 billion in nominal GDP terms. Thus South Africa’s economy is smaller than that of Brazil and India. However, as mentioned earlier, South Africa is a middle-­ income country like Brazil. In terms of its economic structure, South Africa exhibits many characteristics associated with developing countries. For example, its economy is characterized by a division of labor in formal and informal sectors and an uneven distribution of wealth and income (Fedderke 2002). The country is the world’s largest producer and exporter of gold and platinum and also exports a significant amount of coal. The processing of minerals to produce ferroalloys, stainless steels, and similar products is a major industry (Feinstein 2005; Faulkner and Loewald 2008). The country’s diverse industrial base includes motor vehicles and parts, railway rolling stock, synthetic fuels, and mining equipment and machinery (Feinstein 2005; Kularatne 2006). Finally, as described below, South Africa’s previous economic system was based on import substitution, high tariffs and subsidies, anticompetitive behavior, and extensive government intervention in the economy. But in the mid-­1990s, the South African government moved to reduce tariffs and exchange controls (Fedderke and Vaze 2001; Edwards 2005; Edwards and Alves 2006) The history of political institutions in South Africa has been marred by colonialism, an official policy of racism, and a lack of democracy (Worden 2000; Thompson 2001). After WWII, the country emerged as a one-­party dictatorial state with limited enfranchisement for a small section of the population (Deegan 2000; Worden 2000). Specifically, in 1948, the National Party (NP) won the all-­white elections and began passing legislation codifying and enforcing an even stricter policy of white domination and racial separation known as “apartheid” (Deegan 2000; Beinart 2001). Moreover, in the early 1960s, the African National Congress (ANC) and the Pan-­African Congress (PAC) were banned. Popular uprisings in black and colored townships in 1976 and later in 1985 convinced key NP members of the need for change by the late 1980s

Trade Politics and Contributions in India and South Africa  /  207

(Deegan 2000; Worden 2000; Thompson 2001). Consequently, in February 1990, State President F. W. de Klerk, who had come to power in September 1989, renounced the banning of the ANC, the PAC, and other antiapartheid groups. He also started abolishing the last “pillars of apartheid,” including various segregation acts from the apartheid era (Thompson 2001). A long series of negotiations between the NP and the ANC resulted in a new constitution, which was promulgated into law in December 1993. The country’s first nonracial elections were held on April 26–­28, 1994, and it formally made a transition to a full-­fledged democracy by May 1994, with Nelson Mandela as president (Sadie 1998; Boraine 2001; Thompson 2001). After the transition to democracy, South Africa’s leaders did adopt a closed-­list proportional representation (PR) system. This is hardly surprising, because postdemocratic transition South Africa (unlike posttransition Brazil) had strong, centralized, and highly disciplined political parties like the ANC. Thus parties and party leaders in South Africa opted to adopt an electoral system (i.e., the closed-­list PR system) that both promotes and sustains strong and disciplined parties over time (Southall 2004; Hirsch 2005). Yet, as suggested in the previous chapter, this electoral system during the initial posttransition years did not shape or influence political competition in South Africa. One reason for this is because such competition was in essence negated by the presence of a “national unity government” that ruled South Africa during the posttransition years. The second reason is that the new parties (and a whole new class of voters) in South Africa had “almost no experience with the country’s party-­list system” (Sadie 1998: 14), and as a result, during the immediate posttransition period, South Africa’s electoral system was for the for the most “unsettled,” “unstable,” and certainly not fully consolidated (Saide 1998; Brooks 2004). After the consolidation of democracy (by the late 1990s), how­ever, South Africa clearly emerged as a strong party-­centered system where the level of particularism is low (Sadie 1998; Plasser and Plasser 2002; Brooks 2004). Two institutional features of South Africa’s democracy make the country party centered rather than candidate centered. First, as mentioned above, South Africa uses the closed-­list PR system to elect representatives to the national South African legislature (Sadie 1998; Plasser and Plasser 2002). A closed-­list PR system promotes party discipline and centralization of the decision-­making process within parties, which allows party leaders to exert a high degree of control over rank-­and-­file party members (Carey and Shugart 1995; Hicken 2004; Hix 2004). Consequently, individual party members have no incentives to cultivate their personal vote in closed-­list PR systems. As stated by Hix (2004: 196),

208 / Chapter Seven At one end of the spectrum, closed-­list proportional representation (PR) systems represent the most party-­centered settings. In these systems, parties present lists of candidates, and voters cannot influence the order of the candidates on them. Closed-­list systems consequently allow party leaders to exert a high degree of control over their legislators. Without the strategic need to appeal directly to the electorate, candidates have no incentives to break ranks with the party line. In fact, an individual candidate has a positive incentive to go along with the party line—­to improve her position on the party list.

The fact that the closed-­list PR electoral rule in South Africa fosters party discipline and reduces the incentives for individual party members to cultivate their personal vote is also stressed by specialists of South African politics. For instance, Sadie (1998: 280) points out that individual candidates in South Africa “have a difficult time getting elected without a party label and party support.” Additionally, Plasser and Plasser (2002: 143) claim that “generally, campaign frameworks in India as well as in South Africa tend to stimulate more party-­centered campaigning styles refraining or discouraging independent candidates without nomination by or affiliation with a political party . . . strong intraparty discipline and detailed legal regulations and institutional restrictions of campaign practices in both countries (India and South Africa) tend to strengthen the influential role of strong party organizations.” Apart from the closed-­list PR electoral system, there exists an antidefection law in South Africa that was implemented in 1997 (Van der Merwe and Du Plessis 2004). The antidefection law in South Africa effectively deters individual party members from cultivating their personal vote and fosters control over individual party members by party leaders, which are the two main features of party-­centered systems (Van der Merwe and Du Plessis 2004; Spiess 2009). Thus the use of a closed-­list PR electoral system and the presence of the antidefection law make South Africa a party-­centered system. The Evolution of Trade Protection in South Africa Until the mid-­1990s, South Africa’s trade regime was characterized by a high and complex tariff structure and extensive import controls. Some halfhearted attempts were made to liberalize trade policy in 1983. But these were halted and reversed by the mid-­1980s. In fact, the solid line in fig­ure 7.3—­ which illustrates the moving average of simple (unweighted) tariff rate in South Africa from 1979 to 2004—­shows that by the end of the 1980s, South

Trade Politics and Contributions in India and South Africa  /  209 Democratic transition year

49 42

Percentage

35 28 21 14 Unweighted tariff 7 0 1980

Core NTBs

1983

1986

1989

1992

1995

1998

2001

Figure 7.3  Moving average of unweighted tariffs and core NTBs in South Africa

Africa had a very restrictive trade regime, with a moving average unweighted tariff rate of 33%.20 Between 1990 and 1995, the moving average of unweighted tariffs also remained fairly high in South Africa. Furthermore, the dashed line in figure 7.3, which illustrates the moving average of the core NTBs ratio in South Africa from 1988 to 2003, reveals that from 1988 to 1995, the core NTBs ratio was (in percentage terms) as high as 26%.21 Not surprisingly, then, South Africa’s trade regime was considered highly complex, with the highest number of tariff lines and the widest range of tariff rates in the developing world (Fedderke and Vaize 2001; Du Pleiss et al. 2007). After the country experienced a transition to democracy in 1994, the new government in 1995 announced its intention to rationalize tariff policy aimed at reducing and simplifying the tariff structure. In particular, the liberalization of the trade regime consisted of the replacement of quantitative restrictions with ad valorem tariff lines, a simplification of the tariff regime, accompanied by a gradual but significant reduction in tariff rates and a phasing out of a substantial export 1subsidization scheme. As indicated by the solid line in figure 7.3, the moving average of unweighted tariffs was reduced by 12% from 1995 to 1999. The dashed line in the figure shows that the core NTBs ratio in the country was cut by 11% from 1995 to 1998. The wave of trade liberalization in South Africa thus lasted from 1995 to around 1999. However, as shown in figure 7.3, both the moving average of unweighted tariffs and the core NTBs ratio increased again from 1999 to 2002. In other words, South Africa made a turn toward higher trade protection from 1999 to 2002. Thus, similar to India (but unlike Brazil), the

210 / Chapter Seven

pattern of trade protection in South Africa from 1979 to 2003 can be clas­ sified into three periods. During the first period (between 1978 and 1994), South Africa employed import-­substitution-­industrialization (ISI) policies and thus had high average levels of trade protection. In the second period (from 1995 to 1999), tariffs and NTBs in the country decreased consider­ ably. Finally, in the third period (from 1999 to 2004), trade restrictions again increased in South Africa but did not revert back to the peak tariff levels that were observed in the country during the 1970s and 1980s. What explains the three phases of trade policy in South Africa delineated above? Further, did trade liberalization in South Africa between 1995 and 1999 decrease the likelihood of electoral fraud, as predicted by the model in chap­ter 2? I address this question below. Politics of Trade Policy in South Africa While the level of trade protection peaked in South Africa particularly during the 1970s and 1980s, the degree of interindustry occupational mobility of labor increased steadily and became quite high in the country from the late 1970s to the 1990s. This is shown in figure 7.4, which illustrates the annual series of the level of interindustry occupational mobility of labor (the solid line) and the elasticity of interindustry labor mobility (the dotted line) across twenty-­nine industries (listed in table 7.1) at the three-­digit ISIC level from 1978 to 2003.22 Note that data to operationalize interindustry labor mobility at the three-­digit ISIC level are only available for these industries in South Africa. Although the increase in interindustry labor mobility in South Africa from the late 1970s arguably provided an opportunity for South African policy makers to adopt trade reforms—­since occupationally mobile workers rationally favor trade liberalization—­they did not reduce trade barriers. Instead, they continued with ISI policies and maintained high trade barriers from the late 1970s to 1994–­95. Why? To answer this question, first note that the single-­party dictatorial regime of South Africa from the 1970s to 1994—­dominated largely by the National Party (NP)—­had strong business and political ties with two groups in society who supported high levels of trade protection: “upper-­ income white groups and import-­substituting manufacturers” (Waldmeier 1997: 58). Since the dictatorial South African state from the 1970s to the early 1990s (and before) was essentially run as an “all-­white Afrikaner ethnic corporate body” (Kotze and Du Toit 1995: 31), with little or no interest in promoting the welfare of workers (who were predominantly black and Asian), policy makers from the ruling NP government acquiesced to the

Trade Politics and Contributions in India and South Africa  /  211

6.0

15

5.0

12

4.0

9

3.0

ILM-elasticity

18

6

2.0 ILM ILM-elasticity

3

1.0

Inter-industry labor mobility (ILM)

Democratic transition year

21

0

0

1978

1981

1984

1987

1990

1993

1996

1999

2002

Figure 7.4  Labor mobility in South Africa

protectionist demands of these import-­substituting manufacturers (Hudson and Sarakinsky 1986; Adam et al. 1997). Moreover, the NP appeased import-­substituting manufacturers in South Africa’s dictatorial polity in the 1970s, 1980s, and early 1990s, since officials from the NP obtained rent from implementing the trade restrictions demanded by the manufacturers (Adam et al 1997; Waldmeier 1997). Indeed, Bienen (1992: 76–­77) suggests that in dictatorial regimes in South Africa (from the 1970s to the early 1990s) and other African countries, “trade liberalization policies were often extremely hard to formulate and implement . . . precisely because . . . powerful officials (civilian and military) benefit from the controls that have been established over imports and exports. It is government officials who ration and distribute scarce imports, including foreign exchange. They realize the rents that accrue from the systems they construct and control. Of course, officials have allies—­(mainly) import-­substituting manufacturers.” Bienen’s (1992) insight is important. This is because it suggests, as posited by the model in chapter 2, that policy makers in dictatorial regimes (i.e., when ββ = 0 1 in the model in chapter 2) such as South Africa in the 1970s and 1980s have incentives to prioritize the trade policy preferences of protectionist industrial groups (the owners) over the interests of labor when designing trade policies. This consequently led to higher trade barriers, as observed in fig­ ure 7.3, for South Africa during the 1978 to 1994 period when the country was an authoritarian regime. However, South Africa adopted trade reforms after the country made a formal transition to a full-­fledged democracy in 1994. Indeed, as indicated in figure 7.3, tariffs and NTBs declined considerably in 1995–­96, or in other

212 / Chapter Seven

words, within a year or two after the country made a transition to democracy. What explains this “rush to free trade” in South Africa in 1995–­96? The model in chapter 2 predicts that the ruling party in a new democratic regime has political incentives to reduce trade restrictions when the interindustry labor mobility is sufficiently high. The case of South Africa provides support for this causal claim. To understand why this is so, first consider figure 7.4. The two lines in this figure indicate that the level of interindustry occupational mobility of labor and the elasticity of interindustry labor mobility across twenty-­nine industries in South Africa increased sharply and thus became high from 1992 to 1997. Hence the sharp increase in interindustry labor mobility from 1992 preceded the drastic cuts in trade barriers in a “newly democratic” South Africa that started in 1995–­96. Did high labor mobility combine with South Africa’s transition to democracy in 1994 to help foster trade liberalization and thus a reduction of trade barriers, as de­ scribed above? A careful examination of the historical record provides an affirmative answer to this question. The historical record, in fact, reveals that South Africa’s transition to democracy—­which led to the enfranchisement of the majority black community (who were predominantly workers) and the emergence of ANC as the ruling party—­encouraged policy makers to design trade policies that favored the welfare and preferences of labor over the protectionist interests of industry owners who were primarily white Afrikaners (Louw 2000; Lodge 2002). This is suggested by Hirsch (2005: 117) who states that “the democratic transition in South Africa encouraged the ANC to adopt fiscal and trade policies such that it would benefit the majority black community who mainly worked in manufacturing industries in urban areas and who were the main source of electoral support for the ANC . . . the ANC also distanced itself away from white Afrikaner business groups that enjoyed special privileges under the Botha and the De Clerk government.” The prioritization of the welfare of workers over the preferences of the “white capitalists of the old order” (Southall 2004: 315) is also evident in the recommendations made by the government’s investigation into South Africa’s tariff protection policy in early 1995: “Tariff policy, like other economic policies, needs to be used as a tool to promote and sustain growth which is vital for raising the living standard of workers in South Africa . . . it is important not to hold trade policy hostage to the vested interests of business groups that were protected from import competition in the past.”23 The recommendation in the government’s official report in 1995 (quoted above) and Hirsch’s (2005) claim makes it clear that policy makers in newly democratic South Africa, as suggested by my model, were determined to

Trade Politics and Contributions in India and South Africa  /  213

implement trade policies that would suit the interests of labor rather than capital. This is hardly surprising, given that workers in the economy constituted the electoral majority in South Africa and, by far, the largest support base for the ruling party, the ANC (Jacobs and Calland 2002; Southall 2004; Hirsch 2005). Key officials and ministers in South Africa’s government in the mid-­ 1990s were cognizant of the fact that labor mobility and productivity had accelerated in the economy and that workers could therefore both adjust to as well as benefit from trade openness. For instance, in an interview, a retired official from South Africa’s Department of Trade and Industry (DTI) pointed out that “we were aware in the Ministry of Finance and the DTI in 1995 that the productivity, mobility and ability of South African workers especially in the manufacturing sector had grown rapidly had grown rapidly in the last 10–­15 years . . . Therefore we felt at the time that the reduction of import tariffs is an integral part of a process of improving the living standards of an increasingly productive and mobile workforce since workers would gain from a more competitive outward-­oriented economy” (DTI 1996: 12). More important, the report released by the South African DTI in 1995 emphasized that reduction of “import tariffs” would generate more jobs for workers, since they have the “skills to shift their occupation to rising industries” that gain from globalization: “Trade policy should be characterized by a reduction in tariffs to contain input prices . . . opening the economy to import competition will create opportunities for growth and improve competitiveness . . . This will generate more employment for workers in South Africa who in recent years have developed the necessary skills to shift their occupation to rising industries that grow rapidly when trade restrictions are removed” (DTI 1995: 25). Both the DTI and the Ministry of Finance developed and circulated successive assessments about the benefits of trade liberalization for workers to party officials from the ANC during the 1990s (Edwards 2001; Fedderke and Vaze 2001; Fourie 1996). This not only provided useful information about trade reforms to key party officials from the ANC but also convinced them (the party officials) that bringing an end to ISI would promote growth and employment for the increasingly mobile labor force in the country (Bell 1997; Tsikata 1999; Edwards 2001). Since the ANC is ideologically committed to enhance the welfare of workers and because officials from the ANC recognized that generating more employment for workers via trade reforms would potentially translate into more electoral dividends for the party, the party officials took the calculated step of reducing trade restrictions (Fourie

214 / Chapter Seven

1996; Tsikata 1999; Fedderke and Vaze 2001). This is suggested, for instance, in the ANC’s manifesto on economic policy in late 1994: “Increasing South Africa’s role in the global economy by lowering tariffs and opening its borders to international goods and services will provide a boost to the economy . . . It will also generate employment and empower workers that will ultimately benefit the ANC politically and economically” (African National Congress 1994: 82). Put together, the brief historical analysis discussed above shows that the transition to democracy in 1994 encouraged South African policy makers from the ruling ANC party to reduce tariffs, as they believed that rising labor mobility would allow workers to benefit from trade liberalization. They also recognized that trade liberalization could increase the ANC’s electoral prospects. Did trade liberalization in South Africa between 1995 and 1999 reduce the likelihood of electoral fraud, as predicted by the model in chapter 2? Three reasons demonstrate that the adoption of trade reforms by the government in 1995 was an important, if not the only, factor that helped reduce the likelihood of electoral fraud. First, some studies show that growing trade relations between advanced democracies and South Africa during the immediate post–­democratic transition years—­which resulted from the adoption of trade reforms—­sensitized leaders from the ruling ANC that economic sanctions may result if party officials manipulate electoral outcomes (Waldmeier 1997; Handley 2005). As stated by Brooks (2004: 18–­19), As such, South Africa’s reliance on capital and foreign trade, as well as its leading role in the African Renaissance and as (an) exemplar of economic development and governance on the continent, places its democracy far too directly in the global eye for the government to risk stepping out of line . . . the government can . . . neither afford to abuse the position and authority that has been conferred on it by the electorate, and hence fully risk losing their support; nor can it afford to deter investors and international actors by creating a climate of political instability.

Second, the leaders in South Africa’s new democracy also recognized that the growth realized from trade reforms would increase the ANC’s electoral support and its political popularity (Handley 2005; Hirsch 2005). Since trade (and other economic reforms) could potentially generate electoral dividends for the ANC, key policy makers felt that they must take action to deter electoral fraud, as the ANC could “rely on its economic performance instead of electoral manipulation to strengthen its political base in

Trade Politics and Contributions in India and South Africa  /  215

society.”24 Given the ANC’s expectation that trade reforms would be politically beneficial for the party—­as suggested by the party’s economic report in 1994, cited above—­and that vote rigging could engender political instability, the ANC actively took steps to prevent electoral fraud in South Africa. Specifically, the government in South Africa “created the Independent Electoral Commission (IEC), a national election agency . . . in charge of administering elections and prevent acts of voter intimidation and electoral fraud.”25 Interestingly, the ANC justified the creation of IEC in 1996 by suggesting that “the creation of the IEC will send a message to the world and to the citizens of South Africa that democracy in the country, while relatively young, puts absolutely no one above the sacred process of election, and that leaders are chosen only by the genuine will of the people . . . sending this message is vital since it will demonstrate that the ANC can win elections by its economic performance and the policies that it adopts rather than by manipulating elections.”26 Thus leaders in South Africa were deterred from engaging in electoral fraud, owing to the reasons delineated above. Although trade barriers were consistently reduced in South Africa from 1996 to around 2000, the process of trade liberalization came to an abrupt halt in 2000. Instead, tariffs and NTBs in the country increased from 2000 to 2004, even though this increase was not dramatic. What explains this reversal of trade liberalization in South Africa at the turn of the century? While some scholars suggest that a balance of payments problem led to the surge in trade protection mentioned above,27 a closer look at trade politics in the country reveals that the consolidation of the party-­centered nature of South Africa’s democracy by 1999–­2000 fostered close political ties between protectionist business groups and the ANC (the ruling party). This, in turn, contributed to the sudden reversal of trade reforms. More specifically, observe that in 1995 the newly democratic South African state adopted the closed-­list PR electoral rule. As discussed earlier, this particular electoral rule ensured that South Africa evolved into a party-­ centered rather than a candidate-­centered system by the late 1990s (Giliomee and Simkins 1999; Lanegran 2000; Brooks 2004). Not surprisingly, in South Africa’s newly developed party-­centered system, party leaders of the major political parties—­the ANC (the ruling party), Inkatha Freedom Party (IFP), and the New National Party (NNP)—­started exerting strong control of their rank-­and-­file party members (Lanegran 2000; Brooks 2004). They did so through a combination of “threatening expulsion for defection to the opposition . . . and . . . promising financial rewards and promotion within the party to party members that remained loyal to the party” (Pottie and Lodge 2000: 127).

216 / Chapter Seven

Providing financial side-­payments to rank-­and-­file party members to maintain intraparty unity is costly. Hence to raise funds for elections and maintaining party cohesion, the main political parties in South Africa, particularly the ruling ANC party, started soliciting contributions from South African business groups and corporate houses. As stated by Sachikonye (2000: 18–­19), in South Africa, “political campaigns do not come cheap. Similarly, maintaining a party between elections is not inexpensive . . . since access to resources is a decisive factor in electoral contests and party exis­ tence, parties have to raise funding from private sources that includes industry.” To solicit contributions, parties in South Africa also actively co-­opted the economic, including trade, policy interests of these business groups in their respective party manifesto (Southall 1999, 2004; Brooks 2004; Lodge 2001). Indeed, as stated by Brooks (2004: 11), by 1998–­99, the ruling ANC party started taking a “more ‘probusiness’ stance” and were “more likely to court the interests and favor of capital.” Moreover, in the ANC’s party manifesto in 1999, party leaders and officials from the ANC “justified the existence, expansion . . . and the need to co-­opt . . . a black bourgeoisie, so long as it plays by the rules laid down by the party; and . . . endorsed the need for close co-­operation with white capitalists.”28 The ANC’s new bias toward capital in party-­centered South Africa and its decision to solicit contributions from the owners corroborates the causal claim from chapter 3, which posits that parties in party-­centered systems have political incentives to build financial ties with industry owners. However, the ANC’s shift from being a more prolabor party to a more procapital stance came as a pleasant surprise to business groups and corporate houses in South Africa during the late 1990s. These business groups and corporate houses, in fact, reacted to the ANC’s effort to co-­opt them in four main ways. First, as stated by Brooks (2004: 11), business groups “entered into an agreement with the ANC,” since they realized that in a party-­centered democracy like South Africa, the ruling ANC party could credibly promise to implement policies that are compatible with their (the business group’s) interests. Second, capital responded to the ANC’s “request” for donations by providing the ANC with substantial contributions (Southall 1999, 2004; Habib and Taylor 2001). Third, business groups with vested protectionist interests made a serious effort to cultivate political influence within the ANC by forming business ventures with party members from the ANC. As noted by Tangri and Southall (2008: 13), “To exercise political influence within ruling circles, it (business groups) wanted to attract prominent black former politicians turned businessmen/women with good political connections.

Trade Politics and Contributions in India and South Africa  /  217

Many white companies have sold their stakes to black businessmen/women who serve in the highest decision-­making structures of the ANC, and who can push for policies to the benefit of capital.” Fourth, business groups and corporate houses in South Africa recognized that the ANC’s effort to co-­opt capital provided them with a “window of opportunity” to credibly voice their demands (Southall 2004). Among the numerous demands made by business groups to the ANC in 1999 was the “urgent need to halt trade reforms, protect South African industry from unfair price competition stemming from cheap imports and provide subsidies to domestic industries.”29 How did the ruling ANC party respond to the strategies of domestic business groups? As predicted by the model in chapter 4, the ANC in a party-­centered democracy like South Africa reacted favorably to the interests of the industry owners who requested for more trade protection. For instance, in a report released by the economic advisory group within the ANC in late 1999, it was emphasized that “an important objective for the government is to prevent the decline of South African industry and to promote the expansion of domestic industries . . . as a first step, this can be achieved by not exposing South African industry to unfair price competition generated by cheaper imported goods.”30 Further, key officials from the ANC reportedly gave instructions to the DTI to slow down the pace of trade reforms and even raise trade barriers in certain areas to prevent domestic industries from being “ravaged” by import competition. The “rush to protect” domestic industries from import competition is unsurprising, since in party-­centered South Africa, political parties like the ANC (as suggested by the model) are susceptible to the protectionist interests of  business groups with whom they share a symbiotic relationship. This symbiotic relationship of the ANC and business groups was bluntly stated in an interview with the minister of minerals and energy affairs in South Africa, Phumzile Mlambo-­Ngcuka, in 2001, who pointed out that “government and business groups are like Siamese twins. We always have to look at each other’s interests” (Cameron 2002: 7). Last but not least, the political pressure exerted by domestic industries in South Africa in 1999 clearly had the desired the policy effect on trade restrictions. After all, the figure shows that the moving average of unweighted tariffs and NTBs increased during the 2000–­2003 time period as a response to the protectionist demands of domestic industries in South Africa. Put together, the historical evidence presented above supports the theoretical claim from chapter 3 suggesting that parties in party-­centered democracies like South Africa tend to be biased toward the protectionist interests of the

218 / Chapter Seven

owners that seek trade protection. But does quantitative support exist for the implications of this theoretical claim in within-­country data for tariffs in South Africa? To address this question, I gathered a panel data set of industry-­year tariffs on an annual basis from 1979 to 2004 for twenty-­nine different industries (at the three-­digit ISIC level) in South Africa, which are listed in table 7.1; tariff data at the three-­digit ISIC industry-­year level are only available for these twenty-­nine industries. The industry-­year tariff rate for South Africa, which is labeled as tariff, serves as the dependent variable for the statistical analysis. Similar to Brazil and India, data for NTBs are not available for South Africa at the industry-­year level. Hence the quantitative analysis focuses on just the within-­country data for industry-­year tariffs from South Africa. The first key explanatory variable in the model where tariff is the depen­ dent variable is an annual measure of interindustry labor mobility (labeled as labor mobility) for South Africa.31 For the second explanatory variable, I use the annual series of the import penetration ratio across twenty-­nine South African industries (listed in table 7.1) at the three-­digit ISIC level as a proxy to capture the protectionist interests of the owners of these industries in South Africa. This import penetration ratio (labeled as import ratio) measure is calculated per year by dividing industry imports by the sum of industry outputs plus imports (the denominator is industry supply) across the twenty-­nine industries in South Africa; data to operationalize the import penetration ratio variable at the three-­digit ISIC level is only available for these twenty-­nine industries in South Africa.32 The following controls are added to the specification, drawn from time-­series economic data from South Africa: capital-­output ratio, investment-­output ratio, log GDP 33, and the lag of the dependent variable. I also include a time trend in the specification. After gathering the data for the dependent, independent, and control variables, I divided the entire 1979 to 2003 industry-­year panel into three separate samples that correspond to three different time periods: 1979–­1993, 1994–­1999, 2000–­2004. Breaking up the full sample into three separate samples for the three time periods mentioned above allows us to separately analyze the statistical effect of import ratio and labor mobility on tariff  in South Africa during the years in which it is observed as a dictatorship (model 2, for the panel years 1979–­1993); new democratic regime (model 3, for the years 1994–­1999); and party-­centered system (model 4, for the 2000–­2003 time period). I estimated each of the three different models listed above via the system-­GMM approach, which corrects for potential endogeneity among

Trade Politics and Contributions in India and South Africa  /  219 Table 7.4  System-­GMM results for tariffs in South Africa

lag tariff (1978–­93)

1978–­93

1994–­99

2000–­2004

Model 2

Model 3

Model 4

0.831*** (0.079) 0.767*** (0.058)

lag tariff (1994–­99) lag tariff (2000–­2004) import ratio log GDP capital-­output ratio investment-­output ratio labor mobility time trend Constant Sargan test AR(1) AR(2) N

0.294*** (0.075) –­0.054 (0.256) 0.118 (0.25) 0.2 (0.13) –­0.319 (0.406) –­0.024 (0.09) 1.86*** (0.3) 0.31 –­2.29** –­0.305 387

0.178 (0.151) –­0.058 (0.259) 0.103 (0.202) 0.188 (0.131) –­0.552*** (0.17) –­0.071 (0.128) 2.67*** (1.06) 0.26 –­1.95** –­0.323 159

0.728*** (0.035) 0.382*** (0.115) –­0.044 (0.253) 0.126 (0.193) 0.176 (0.134) –­0.18 (0.292) –­0.075 (0.089) 1.95*** (0.51) 0.29 –­2.03** –­0.277 138

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. The results reported for the system-­GMM models are from the one-­step estimation, except the Sargan test and the AR(1) as well as the AR(2) tests, which are taken from the second-­step estimation. A negative and statistically significant AR(1) term plus a statistically insignificant AR(2) term indicates no serial correlation (Blundell and Bond 1998).

the explanatory variables and the dependent variable. The results from these system-­GMM models are reported in table 7.4. The estimate of import ratio on tariff in South Africa for the 1978 to 1993 time period (see model 2, table 7.4)—­the years in which South Africa is observed as a dictatorship—­is positive and significant at the 1% level. It is also positive and significant at the 1% level for the 2000 to 2004 time period (see model 4, table 7.4), which is estimated for the years in which South Africa is a consolidated party-­centered system. These results therefore offer empirical support for the theoretical argument that in developing states, governments are more responsive to protectionist interest when the country is dictatorial or a party-­centered system. In contrast, import ratio is statistically

220 / Chapter Seven

insignificant in model 3, where the sample ranges from 1994 to 1999 and thus includes the years in which South Africa is a new democratic regime. This confirms my expectation that protectionist business groups may not have had sufficient leverage to statistically influence trade barriers in South Africa when the country was a new democracy. The coefficient of labor mobility is statistically insignificant in models 1 and 3. As suggested by the formal models in chapters 2 and 3, this indicates that the occupational mobility of workers mattered less to South African policy makers when the country was a dictatorial state and when it is observed as a party-­centered system. However, labor mobility is negative and highly significant in model 3. This supports the causal prediction in chap­ ter 2 positing that leaders in newly democratic states such as South Africa be­ tween 1994 and 1999 are more likely to reduce trade barriers conditional on the level of interindustry labor mobility. In the following section, I analyze the validity of the causal predictions about campaign contributions in party-­centered democracies by focusing on contributions in South Africa. Contributions in Party-­Centered South Africa Is there empirical support for the anticipated theoretical claim that contributions from industry owners with protectionist preferences increased in South Africa during the years in which its government is observed as a party-­ centered system? Conducting such an empirical analysis is not easy. This is because even though parties in South Africa are required by the Public Funding of Represented Political Parties Act to disclose public funding sources (i.e., from the state), they are not required by law to reveal private funding by industries. As stated by Lodge and Scheidegger (2005: 17), “There are no legal . . . requirements for disclosure of the sources or amounts of private donations.” Hence data on the size of contributions offered by industries and business groups in South Africa are not available. Therefore I use both qualitative evidence and data on perceived level of contributions from the WEF’s (2004b) EOS database for South Africa to assess the causal predictions mentioned above. In terms of qualitative evidence, I showed earlier that once South Africa developed into a full-­fledged party-­centered democracy by the late 1990s, the ruling ANC party started actively soliciting contributions and co-­opting the policy interests of business groups into the government agenda. Did business groups respond to this co-­optation by providing more contributions? A report published in 2003 by the Institute for Democracy in South Africa (IDSA 2003)—­an organization that tracks funding to South African

Trade Politics and Contributions in India and South Africa  /  221

political parties—­emphasized that the private funding of parties by business groups, for example, was quite large and ubiquitous during the 1998 to 2003 period when the country is observed as an established party-­centered democracy: South Africa’s 20 represented political parties will receive a total of R66.6 million for general expenditure . . . If one unpacks the total amount, the governing party (ANC) will receive R42.5 million with the main opposition party (DA) receiving about R7.1 million. However, considering that these state funds are mainly used to cover routine expenditure, parties will still require greater amounts to finance the 2004 election campaign. The ANC reportedly spent at least R100 million on its 1999 campaign and the DA reportedly spent in the region of R20 million on its campaign. The vast disparity between state funding and campaign expenses indicates that all parties will be wooing the corporate community . . . to raise funds for multi-­million rand media campaigns, leaving ample space for influence peddling. (IDSA 2003: 19)

The 2003 report by the IDSA also points out that contributions/donations to parties, especially the ANC, has grown exponentially in South Africa since 1998. Indeed the IDSA (2003: 7) report estimates that between 1998 and 2003, the private funds received by the ANC grew as follows: “1998/ 99—­R52.1 million; 1999/2000—­R54.7 million; 2000/01—­R57.8 million; 2001/2002—­R62.8 million; and 2002/2003—­R67.4 million.”34 Moreover, during the late 1990s, even party officials often implicitly admitted that their respective party received large financial contributions from business groups and/or businesspersons. For instance, as stated by Lodge, “the Treasurer-­General of the ANC, Makhenkesi Stofile, told an audience at Mafikeng in 1997 that R2 million donations to the ANC had now become customary among black business people. In return for this generosity ‘we opted for the role of facilitators for black business in the country,’ he noted. The Inkatha Freedom party (IFP) accepted donations from illegal casino operators while the provincial government began preparing legislation to regulate the casino industry” (2002: 134). Finally, scholars that study funding of parties in South Africa have also found that the discrepancy between stated elected expenditures by the parties and the amount of funds that they receive from public sources (state funding) has increased substantially in recent years. Scholars have inferred that the growing gap between stated expenditures and public funding clearly indicates that private funding from industry to parties has grown dramatically since 1998/99. As emphasized by Mbaya (2004: 83), in South Africa,

222 / Chapter Seven

“political parties spent an estimated R300–­500 million on the 1999 elections. However, only a fraction (R53 million) of this total expenditure was acquired from the public fund . . . the corrupting power of money provides a good reason to worry that, if not properly and effectively regulated, political funding in terms of private donations can do a disservice to the very rationale of political participation.” The evidence discussed above provides useful information but is in­ sufficient. Hence we use the WEF’s (2004b) EOS data for South Africa to broadly assess the causal predictions about contributions in party-­centered systems. Specifically, I conduct two empirical exercises. First, I employ the WEF’s EOS database for Brazil and South Africa to extract data on the levels of contributions/donations given to political actors, as perceived by industry executives, in these two countries. The data for the perceived levels of contributions are based on responses by industry executives to question EOS Q4.13, which was described earlier; recall that responses to this question are coded by the WEF on a 1 (donations never occur) to a 7 (donations are common) scale. Second, the WEF EOS data for Brazil and South Africa are used to obtain information about the extent to which industry executives perceive that the donations they provide to political actors actually influence policy outcomes (including trade protection) in these two countries. The data to compile this information are drawn from responses by industry-­executives to question EOS Q4.13, which was described earlier; responses to this question are coded by the WEF on a 1 (no influence) to a 7 (high influence) scale. One finds in the WEF’s (2004) EOS sample that there exists a total of eighty-­five responses by industry executives to questions EOS Q4.13 and EOS Q4.14 in South Africa. This is similar to the total of eighty-­nine responses by industry executives to these two questions in Brazil. We calculated for the mean score on the 1–­7 scale for both the questions mentioned above for South Africa by using the WEF (2004) EOS data. The mean score on the 1–­7 scale for both questions in Brazil and South Africa is reported in table 7.5. Panel A in table 7.5 shows that the mean score for question EOS Q 4.13 that deals with the frequency of donations given to political actors, as perceived by industry executives, is 3.8 (on a 1–­7 scale) in Brazil but a substantially higher 5.5 in South Africa. The difference-­of-­means test reported in panel A shows that the average score for question EOS Q 4.13 for South Africa is statistically (at the 1% level) higher than the mean score for the same question for Brazil. Panel B of table 7.5 reveals that the mean score for question EOS Q 4.14, which captures perceptions about the extent to which

Trade Politics and Contributions in India and South Africa  /  223 Table 7.5  Contributions in Brazil and South Africa

Question

Mean for Brazil

Mean for South Africa

Difference of means in column (a) and (b) (p-­value)

Panel A

(a)

(b)

(c)

EOS question 4.13: How common are donations to politicians in your country? 1 (never occurs) to 7 (common)

3.8

5.5

1.7 (0.0)

3.4

5.0

1.6 (0.0)

Panel B EOS question 4.14: To what extent do legal contributions to politicians have a direct influence on specific public policy outcomes? 1 (low influence) to 7 (high influence)

donations influence policy outcomes, is 3.4 in Brazil but as high as 5.0 in South Africa. A difference-­of-­means test in panel B reveals that the average score for question EOS Q 4.14 for South Africa is statistically (at the 1% level) higher than the mean score for the same question for Brazil. Thus the descriptive results reported above suggest, as anticipated by the model in chapter 4, that the levels of contributions offered by industries to politicians in party-­centered democracies like South Africa are generally higher than the levels of contributions provided in candidate-­centered systems like Brazil. As predicted by our theoretical story, the results also show that, unlike in Brazil, industry executives in party-­centered South Africa perceive that the contributions that they offer are more likely to substantially influence economic policy, including trade policy, outcomes.

Conclusion In this chapter, we used both qualitative evidence and statistical analysis of within-­country data from India and South Africa, two developing countries with party-­centered systems, to assess two main causal claims about the link between party-­centered systems and trade politics in developing democracies. The first causal argument posits that the trade policy decisions

224 / Chapter Seven

of parties in party-­centered systems will be biased toward the trade policy interests of industry owners (who favor trade protection) rather than the preferences of workers. The second key causal argument suggests that the bias in trade policy toward owners will induce owners to provide substantial contributions to parties in party-­centered systems. Building on this, the model in chapter 3 predicts that the contributions provided to politicians in party-­centered systems will be greater than the contributions offered to politicians in candidate-­centered systems. A detailed analysis of the India and South Africa cases provides compelling support for the two causal arguments stated above. Additionally, since South Africa made a transition to a full-­fledged democracy in 1994, I used this case to study whether trade politics in South Africa during the years in which it is observed as a new democracy (i.e., between 1994 and 1999) confirms the causal claims that emerge from the model in chapter 2. As predicted by the model in chapter 2, during the immediate postdemocratic transition years in South Africa, the country’s policy makers designed trade policies that favored the interests of workers over those of various industry owners who were staunchly resisting trade reforms. The evidence from this case also shows that the adoption of trade reforms in South Africa certainly provided officials from the ruling ANC party incentives to not resort to electoral fraud. In fact, concerns about preventing electoral fraud also encouraged political leaders in South Africa to design institutions, specifically the Independent Election Commission, to increase the transparency and legitimacy of the country’s election process. Thus, put together, the anecdotal evidence as well as within-­country data analysis from Brazil, India, and South Africa support the key causal claims in the models presented in chapters 2 and 4.

E i g ht

Conclusion

The collapse of the Soviet Union helped precipitate the third wave of de­ mocratization across the developing world. This process of democratization started in Latin America, spread to eastern Europe, and then spread to parts of Asia and eventually to Africa. The global movement toward democracy was also both preceded and followed by economic reforms, including trade liberalization, across many developing countries.1 The concurrent spread of democracy and trade liberalization across the developing world was thus remarkable in terms of its geographical scope. Yet political scientists were initially pessimistic about the possibility that democratization fosters eco­ nomic (including trade) reforms in developing states.2 Indeed, early studies on democracy and economic reform often suggest that incumbents in new democracies in the developing world would not be able to cope with the mass political backlash that results from the hardships engendered by economic reform (Przeworski 1991; Nelson 1993; Armijo et al. 1994; Haggard and Kaufman 1995). A handful of prominent scholars claimed in the early 1990s that democracy and economic reform were in­ herently incompatible in developing states (Walton and Ragin 1990; Nelson 1993). Some scholars also argued that the pursuit of economic liberaliza­ tion would generate economic instability or jeopardize democracy in these states (Przeworski 1991; Nelson 1993). A new wave of research by international political economy (IPE) schol­ ars, however, challenged this conventional wisdom about the inherent in­­ compatibility of democracy and economic reform in developing countries. Students of IPE instead suggested and found that the emergence of democ­ racy promotes trade liberalization in not just the developed but also the de­ veloping world (Frye and Mansfield 2004; Milner and Kubota 2005; Eichen­ green and Leblang 2008; Tavares 2008). This finding has encouraged some

226 / Chapter Eight

researchers to infer that democracy and economic globalization go “hand in hand” (Giavazzi and Tabellini 2005; Eichengreen and Leblang 2008; Tavares 2008).3 But the idea that democracy and globalization go “hand in hand” has been at least partly questioned by Kono (2006). He suggested that demo­ cratic incumbents may reduce tariffs. But democratic leaders also actively employ nontransparent policy tools such as nontariff barriers (NTBs) to protect their domestic markets from import competition. Despite Kono’s (2006) claim, the idea that the introduction of democratic institutions fos­ ters trade reforms has been widely accepted by scholars in political science and economics (see, for example, Giavazzi and Tabellini 2005; Milner and Kubota 2005; Eichengreen and Leblang 2008; Milner and Mukherjee 2009; Tavares 2008). This book also began with the widely accepted claim in IPE that the transition to democracy leads to a reduction in trade barriers in develop­ ing states. However, a more systematic analysis of the relationship between democratization and trade protection revealed that the pattern of trade pro­ tection in developing states is more complex than suggested by extant re­ search. For instance, even though trade restrictions did fall in the immediate postdemocratic transition years for many new democracies, they increased or remained consistently high in several other newly democratized develop­ ing states. Trade barriers also vary significantly across not just established democracies but also “newly consolidated” democracies across the develop­ ing world. The intriguing patterns associated with trade protection in devel­ oping countries thus raised two general questions that constitute the core puzzles in this book. First, why do trade barriers decrease in the immediate years following a democratic transition in some new democratic regimes across the devel­ oping world but not others? Second, why do trade barriers vary substan­ tially across established and newly consolidated democracies in developing states? The primary purpose of this book has been to provide a set of theo­ retical models that answer the questions posited above and generate test­ able hypotheses. A broader objective of this book is to understand how do­ mestic political institutions and economic conditions interact to influence trade policy outcomes as well as the relationship between business (indus­ try owners) and policy makers across developing countries. I addressed these two core puzzles by following a two-­pronged strategy. First, I used political economy models of trade protection4 to develop game-­ theoretic models of this interaction. To this end, I posited a world pop­ ulated with three main political actors, endowed with different types and

Conclusion / 227

quantities of assets and intent on converging on a menu of trade policies that maximize their income or rent. The models helped me explain how po­ litical contestation in new democratic regimes interacts with factor mobility to affect the optimal design of trade policies. The models also provided an opportunity to conceptually explore how higher levels of electoral particu­ larism combine with relative factor mobility of especially labor in devel­ oping country democracies to influence trade policies in these states. But the models do not merely offer predictions about the political economy of trade policy in developing states. Rather, they also provide insights on how the institutional context influences the behavior of “capital” (e.g., owners of import-­competing industries) and their interaction with policy makers in developing economies. For the second strategy, I amassed a body of empirical evidence to evalu­ ate the predictions that emerge from the theoretical models. I employed pooled statistical data on trade protection in developing countries, within-­ country data on trade barriers, cross-­national survey data on perceptions of contributions given to political actors, and finally a detailed comparison of three critical cases. The analysis of multiple forms of empirical evidence is necessary to evaluate the testable hypotheses and their underlying causal logic, which is derived from the game-­theoretic models. In the next section of this concluding chapter, I summarize the book’s central arguments and findings. The chapter then focuses on answering the following three questions: What are the main theoretical and empirical contributions of the findings presented in this project? What lessons can scholars extract from the book’s analyses to understand trade politics in the international economic system? How does trade politics in the domestic arena affect other outcomes, such as economic growth? Finally, I end the chapter by identifying some questions that still remain to be answered on the issue of democratic institutions and trade policy in developing states.

Summary and Contributions The model on democratization and trade policy in chapter 2 starts with the premise that the introduction of elections in a new democratic regime unleashes intense political competition between the political parties (Mans­ field and Snyder 1995a, 2002; Diamond 1999; Corrales 2001). Increased po­­ litical contestation induces the opposition and the ruling party to situate the equilibrium level of tariffs and NTBs closer to the median voter’s optimal trade policy preference to maximize their likelihood of winning the election. Incumbents in new democracies in the developing world will reduce trade

228 / Chapter Eight

restrictions the most when labor mobility is sufficiently high, as higher la­ bor mobility produces an occupationally mobile median voter who benefits from trade reforms. Political competition among parties in a new dem­ ocratic regime also dissuades the owners of asset-­specific industries from providing financial payments (including campaign contributions) to the par­­ ties to influence trade policy when labor mobility is high. An additional re­ sult that emerges from the model on democratization and trade protection is that trade liberalization reduces the likelihood of electoral fraud in new democracies. This facilitates the consolidation of democracy in developing states. Yet it was emphasized earlier that, even after the consolidation of de­ mocracy, trade barriers vary substantially across “newly consolidated” and established democracies. I constructed an additional model in chapter 4 to account for this variation. The main prediction from this latter model is that higher levels of particularism induce party leaders to locate the equilibrium level of trade restrictions closer to the median voter’s optimal trade policy preference. Consequently, party leaders in particularistic (i.e., candidate-­ centered) systems optimally choose to reduce trade barriers most when la­ bor mobility is sufficiently high. Moreover, the model predicts that industry owners in democracies that exhibit higher levels of particularism reduce the monetary payments that they offer to politicians to alter trade policy when labor mobility is reasonably high. Low levels of electoral particularism (i.e., party-­centered systems), however, have the opposite effect. A multimethodological approach has been employed in this manuscript to test the comparative static predictions derived from the game-­theoretic models and the causal stories that lead to these predictions. Results from large-­n data sets of trade protection (tariffs and NTBs) in developing states statistically corroborate the theoretical claims about the interactive effect of new democratic regimes and labor mobility on trade barriers in developing economies. Estimates from a variety of statistical models also support the theoretical predictions about political particularism and labor mobility on trade restrictions in developing country democracies. Furthermore, statistical results obtained from the World Economic Fo­ rum’s (WEF’s) Executive Opinion Survey (EOS; 2004b) data set corroborate the causal claim that industry owners in new democracies and particular­ istic democracies perceive that their contributions to politicians will not influence trade policies when labor mobility is high. Finally, the three case studies discussed in this book—­Brazil, India, and South Africa—­help “process-­trace” the key causal claims derived from the political economy models of trade policy. The Brazil and South Africa cases reveal that my

Conclusion / 229

theoretical arguments about trade politics in new democracies are empiri­ cally plausible. The three case studies also show that the theoretical claims about how and when electoral particularism affects the design of trade pol­ icies are plausible. What are the main contributions that emerge from the theoretical and empirical results that were summarized in the preceding paragraphs? The key conceptual contribution of the two game-­theoretic models presented in the earlier chapters is that they provide precise causal mechanisms that explain how democratic institutions aggregate the trade policy preferences of capital and labor in developing economies. As such, these mechanisms explain when the trade policy preferences of labor and capital translate to trade policy outcomes—­and the variations in these outcomes—­that we ob­ serve across the developing world. More important, the specific theoretical contributions that emerge from the model’s mechanisms are threefold. First, the model reveals that greater occupational mobility of labor helps promote trade reforms, which is a standard result. But the model on democ­ ratization and trade politics also suggests that greater labor mobility leads to trade liberalization in developing economies in the institutional context of newly democratized regimes. The second contribution stems from the lat­­ ter model of electoral particularism and trade policy in developing country democracies. In particular, the main result from this model challenges pre­­ vious research hypothesizing that candidate-­centered systems are asso­­ciated with higher trade barriers.5 Instead, the model suggests that candidate-­ centered systems—­that is, democracies that exhibit higher levels of partic­ ularism—­promote trade reforms particularly when interindustry labor mobility is high. The results further suggest that party-­centered systems in developing de­ mocracies are, in contrast to previous claims,6 not receptive to trade reforms. The third theoretical contribution is the prediction that trade liberalization reduces the likelihood of electoral fraud in newly democratized states. This prediction departs from existing explanations of electoral manipulation, which primarily focus on the effects of international election observers or other “outside monitors” on the probability of electoral fraud.7 The book’s theoretical analysis provides some interesting insights about the impact of domestic political institutions on campaign contributions by capital in developing country democracies, which to my knowledge has not been systematically assessed in the IPE literature on trade.8 Some (al­ beit few) scholars in comparative political economy have suggested that the emergence of democracy engenders close ties between business groups (i.e., capital) and politicians or political parties (Hellman 1994; Kitschelt

230 / Chapter Eight

1995, 2000; Keefer 2007). As a result, young democracies in developing countries are usually characterized by patron-­client relations between capi­ tal (business groups) and the state. This results in more corruption, crony capitalism, and higher levels of contributions to political actors (Kang 2002; Keefer 2007; Kitschelt and Wilkinson 2007). While I introduced a “corrup­ tion mechanism” in the models, with industry owners offering both legal contributions and illegal donations, the models of trade politics in the ear­ lier chapters did not analyze in any detail the impact of democratization on corruption or crony capitalism. However, the first model in the book does suggest that extant claims about close ties between business groups and political actors in new democratic regimes may not be fully accurate in the specific context of trade policy. For instance, the theoretical analysis presented in chapter 2 suggests that the inability of parties to credibly commit themselves ex ante to implement ex post the trade policies preferred by capital weakens (does not strengthen) the relationship between business groups or industry owners and the state in new democracies. The theoretical chapters also predict that higher levels of electoral particularism in developing democracies weaken political and financial ties between capital and politicians in the issue-­area of trade pol­ icy. This insight departs from existing studies positing that highly particu­ laristic democracies are characterized by the strongest ties between business and political actors (Kitschelt and Wilkinson 2007; Lyne 2008). This book’s contributions are not just limited to theoretical insights and some nuanced conceptual innovations in game-­theoretic models of trade politics. Rather, the empirical results reported in this manuscript also con­ tribute to existing studies in three main ways. First, extant research on the impact of democratization on trade protection focuses on estimating the effect of either the continuous Polity index of democracy or the dichoto­ mous Przeworski et al. (2000) measure of democracy on trade barriers (Frye and Mansfield 2004; Milner and Kubota 2005; Eichengreen and Leblang 2008; Tavares 2008). Unlike these studies, I develop an indicator variable for new democratic regimes that operationalizes the years—­specifically, the immediate postdemocratic transition period—­in which some developing countries are observed as new democracies in the sample. Doing so allowed me to closely test the first set of hypotheses about trade protection in new democracies. It also allows researchers to analyze how political dynamics within new democratic regimes affects trade policies. This is important for IPE scholars who want to understand more deeply the empirical relation­ ship between democratization and trade policy outcomes (Garrett 2000; Milner and Kubota 2005; Eichengreen and Leblang 2008).

Conclusion / 231

Second, I carefully operationalized a well-­known measure of the inter­ industry occupational mobility of labor for the developing countries in the sample.9 While the measures of interindustry labor mobility that I use for the tests are not novel per se, the empirical analysis is among the first to sta­ tistically evaluate the interactive effect of domestic political institutions and interindustry labor mobility on trade policy outcomes in a comprehensive sample of developing countries. This is important because the empirical results reveal that the institutional context matters significantly (in the sta­ tistical sense) when determining whether or not the trade policy preferences of workers actually result in a change in trade policy. This is a step forward from existing studies that analyze the effects of factor mobility on trade politics10 but do not systematically test how political institutions moderate the impact that relative factor mobility has on trade policy outcomes. One also learns from the empirical results that factor mobility, in par­ ticular the interindustry occupational mobility (or lack thereof ) of work­ ers, arguably influences trade policy outcomes more than relative factor endowments (i.e., log capital-­labor ratio) in not just new democracies but also newly consolidated and established democracies across the developing world. This should not be interpreted to mean that factor endowments do not matter for trade policy outcomes. I submit that relative factor endow­ ments are especially useful for understanding differences in international trade policies among developed and developing countries (see Spilimbergo et al. 1999; Dutt and Mitra 2002). However, given that the temporal varia­tion in interindustry labor mobility is significantly more than that of factor en­ dowments within developing countries, I believe that on an annual (or year-­ to-­year) basis, it is not factor endowments but instead temporal changes in labor mobility that play a vital role in influencing the trade policy decisions of governments in developing democracies. Whether or not the aforemen­ tioned conjecture is valid requires further empirical research that is beyond the scope of this book. This book’s third major contribution is as follows: the paucity of cross-­ national data on the amount of campaign contributions given by capital (industry owners) to politicians in developing states made it impossible to evaluate claims about how political institutions influence contributions. Yet the analysis of the WEF’s EOS data and three cases takes a critical step for­ ward in this direction by assessing how domestic political institutions can affect campaign contributions to politicians, given the relative factor mobil­ ity of labor. I found that domestic political institutions (new democracies and the level of particularism) combine with the relative factor mobility of labor to shape the beliefs that industry owners hold with respect to the

232 / Chapter Eight

influence of their contributions on policies. The case studies also provide insights about the links between domestic institutions and cross-­national variation in contributions given to parties. These findings add some inter­ esting empirical insights to the IPE literature on trade. This is substantively important, considering that evaluations of the impact of different demo­ cratic political institutions on contributions are generally sparse in the IPE literature.

General Implications and Policy Lessons This book almost entirely focuses on how domestic politics affects the stra­ tegic behavior of policy makers, capital, and labor in the context of trade policy. Yet the analysis presented in the previous chapters has implications for the politics of trade in the international economic system. To understand these implications, first note from the “policy world” that key democratic developing countries such as India, Indonesia, and South Africa blocked multilateral trade reform efforts at the Doha round and subsequent Can­ cun meeting on trade negotiations conducted under the auspices of the World Trade Organization (WTO; Narlikar and Wilkinson 2004). Numer­ ous policy analysts and journalists have suggested that the Doha round and Cancun meeting on multilateral trade negotiations collapsed owing to the “intransigence” of trade negotiators from some large developing democ­ racies, including Brazil, India, and South Africa (see, for example, Dunoff 2001; Beattie and Williams 2005; Waddington 2005; Kennedy 2008). These analysts and reporters, however, do not systematically explain why certain developing democracies consistently blocked multilateral trade reform ef­ forts at the Doha round and Cancun meeting. The various theoretical insights and empirical results presented in this manuscript provide some answers to the question posited in the previous paragraph. Specifically, the theoretical story presented in chapter 3 suggests that domestic institutional characteristics of developing states—­especially their level of political particularism—­has arguably had a crucial impact, in­­ fluencing the governments of certain developing democracies to adopt a more rigid bargaining position when negotiating a reduction in trade barri­ ers with developed nations at the WTO. In particular, a closer examination of the Doha round of trade talks (and subsequent Cancun meeting) reveals that two main party-­centered democracies in the developing world—­namely, India and South Africa11—­had been at the forefront of blocking multilat­ eral trade reform efforts and encouraging other nations to oppose the trade liberalization goals set by the WTO (Cline 2004; Narlikar 2006). Moreover,

Conclusion / 233

a preliminary investigation of the trade negotiation strategies of the three party-­centered democracies mentioned previously indicates that protection­ ist business groups in each of these three democracies successfully lobbied and convinced their respective governments to not accept the WTO’s pro­ posal of mutual reduction of trade barriers on each occasion (Page and Kleen 2005; Narlikar 2006; Jensen and Gibbon 2007). For example, the Electronic Industries Association of India (ELCINA), a lobbying association, success­ fully lobbied the government in New Delhi to block attempts by the WTO to formally propose the liberalization of tariffs in the electronics sector by large developing countries such as India. The bargaining tactics adopted by the three party-­centered democracies (India, Indonesia, and South Africa) is not surprising from the perspective presented in this manuscript. This is because the results reported earlier indicate that governments in party-­centered democracies, which are inher­ ently characterized by low levels of particularism, have political incentives to weigh more strongly the protectionist interests of capital (i.e., owners of asset-­specific industries) in their polity. Consequently, the domestic politi­ cal incentives of these governments will induce them to obfuscate multi­ lateral trade liberalization efforts at the WTO. Thus, with respect to policy making, one learns from the book’s analysis that the WTO and negotiators from advanced industrial democracies should take into account the insti­ tutional characteristics (e.g., electoral particularism) of developing democ­ racies when designing, for example, the trade liberalization objectives of multilateral trade agreements. The WTO should also avoid seeking exces­ sive trade policy concessions, especially from party-­centered developing de­ mocracies, and guard against setting overambitious goals of trade liberali­ zation in multilateral trade agreements. Slowing down (but not stopping) the pace of trade liberalization will facilitate multilateral negotiations be­ tween advanced industrial and developing democracies at the WTO. This in turn may help prevent the collapse of multilateral trade negotiations in the future. Research on the collapse of the recent WTO-­based multilateral trade talks by journalists and policy analysts neither examine nor identify whether there is a common causal pattern or a set of specific domestic political factors that drive certain developing states to consistently stall multilateral trade nego­ tiations over time.12 This is unfortunate because identifying which particu­ lar political or institutional factors induce developing countries to hinder multilateral trade efforts may help the WTO devise solutions that effectively address the domestic problems faced by these states; this in turn could fa­ cilitate the smooth progress of multilateral trade reform efforts in the future.

234 / Chapter Eight

Hence to broadly explore which specific domestic institutional factors encourage democracies in the developing world to stall progress in multi­ lateral trade talks, I conducted a preliminary examination of the data on the moving average of  the annual number of  (i) antidumping petitions initiated by developing democracies from 1980 to 2008 and (ii) safeguard investiga­ tions filed by developing democracies from 1996 to 2010 at the WTO.13 I did so because recent studies in international economics in this issue-­area sug­ gest that countries often initiate safeguard and antidumping petitions cases at the WTO in order to “legally” protect their domestic markets from import competition (Krueger 1996; Blonigen and Prusa 2001; Prusa 2005; Bown 2009, 2010; Hoekman and Kostecki 2009).14 In other words, safeguard and antidumping petitions are often used as tools for trade protection. The preliminary examination of the relevant data shows that among democratic developing countries, party-­centered democracies initiate safe­ guard and antidumping petitions far more frequently than candidate-­ centered democracies. In fact, the annual average number of safeguard and antidumping petitions initiated by party-­centered democracies is statistically higher than that initiated by particularistic democracies across the develop­ ing world from 1980–­2008 (for antidumping petitions) and 1995–­2009 (for safeguard investigations). Furthermore, figures 8.1 and 8.2 respectively indicate that across developing democracies, the moving average of the an­ nual number of antidumping petitions (figure 8.1) and safeguard investi­ gations (figure 8.2) initiated since the 1990s by party-­centered developing democracies has steadily increased and the annual number initiated by par­ ticularistic developing democracies has decreased. The descriptive results discussed above are interesting because, contrary to conventional wisdom, they indicate that developing democracies do not employ protectionist instruments such as antidumping petitions in a ho­ mogenous or uniform manner, as suggested by some policy-­based studies in this issue-­area (Malhotra and Kassam 2006). Rather, there is substantial var­ iation in the use of such protectionist instruments, since party-­centered de­ mocracies statistically (in the sense of descriptive statistics) initiate a higher number of antidumping and safeguard petitions than candidate-­centered democracies in the developing world. Stated more broadly, this indicates that differences in domestic political institutions can explain which devel­ oping democracies are more likely, on average, to legally use protection­ ist instruments and consequently stall multilateral trade reform efforts at the WTO. Given that I have identified which developing democracies (this largely includes party-­centered democracies) tend to often employ protec­ tionist instruments that are legally allowed by the WTO but stall multilateral

Average of anti-dumping petitions

Conclusion / 235

80

60

40

20

1980

1984

1988

1992

1996

2000

2004

2008

Candidate-centered

Party-centered

Average of safeguard investigation cases

Figure 8.1  Antidumping petitions in developing democracies

40

30

1

20

10

1996

1998

2000

Party-centered

2002

2004

2006

2008

2010

Particularistic democracies

Figure 8.2  Safeguard investigations in developing democracies

trade liberalization efforts, I can potentially use the theoretical analysis dis­­ cussed earlier in this book to explain why these countries consistently choose to do so. Understanding why some developing countries have a tendency to delay or block multilateral trade talks will perhaps help the W TO design a lim­­ited yet optimal multilateral trade agreement that is more likely to be po­­ liti­cally acceptable in a wide spectrum of developing states. Moreover, trade negotiators from advanced industrial democracies can use the information from the preliminary empirical finding discussed above to separately negotiate trade agreements with governments from

236 / Chapter Eight

Real GDP growth rate (%)

10 8 6 4

2 0 -30

0

30 Labor mobility (structural 95% CI

60

90

Marginal effect

Figure 8.3  Effect of labor mobility on economic growth in developing democracies

candidate-­centered developing democracies, instead of negotiating trade agreements simultaneously with all developing democracies or a large set of developing democracies. Trade negotiations with candidate-­centered democracies could lead to more efficient bargaining outcomes, as govern­ ments from these democracies in the developing world are more receptive to trade reforms and initiate a lower number of antidumping and escape-­ clause petitions, compared to party-­centered systems. If advanced democra­ cies adopt the bargaining tactic of separately negotiating trade agreements with candidate-­centered developing democracies, it may also increase the likelihood of successful multilateral trade negotiations between developed countries and a nontrivial set of developing countries, which will promote global trade flows. While the main findings presented here clearly have implications for in­ ternational trade politics, the book’s focus on the relative factor mobility 1 of labor is important in that labor mobility also affects other economic outcomes and policy issue-­areas as well. To make sense of this claim, note that a cursory examination of my data of developing democracies listed in table 5.1 (in chapter 5) reveals that higher occupational mobility of labor has a strong positive and statistically significant effect on economic growth in the sample of developing democratic country-­years.15 Indeed, estimates from system-­GMM models—­which account for potential endogeneity be­ tween interindustry labor mobility and economic growth in the empirical specification—­show that a one standard deviation increase in the interin­ dustry occupational mobility of labor from its mean in the sample has the

Conclusion / 237

effect of statistically increasing the real GDP growth rate by approximately 2.1% in developing democracies. This marginal effect, which is substantial, is statistically significant and illustrated in figure 8.3. Since greater interindustry occupational mobility of workers fosters trade openness in many developing democracies and increases economic growth in developing democracies as a whole, it is important that governments in these states adopt policies that promote the occupational mobility of workers in the labor force. To do so, democratic governments in develop­ ing countries will need to increase public investment in education of their workers, as extant research on advanced democracies by economists sug­ gests that investment in education increases the occupational mobility of labor under certain conditions (Nunez 2003; Adnett et al. 2004; Gielen 2007). Unfortunately, the data reveal that between 1975 and 2005, the aver­ age amount of  government expenditure on tertiary and secondary education as a percentage of GDP is as low as 3% across all democracies in the devel­ oping world.16 This indicates that governments in developing democracies should act swiftly and substantially increase investment in public education to increase the occupational mobility of workers in their economy. Doing so will help promote trade liberalization and economic growth, which as suggested by economists, increases economic welfare.17 It may also politi­ cally benefit governments, as voters in developing democracies are likely to reelect incumbents into office when the economic growth rate increases.

Extensions and Areas for Future Research Similar to extant social-­science theories that employ game-­theoretic mod­ els, the book’s parsimonious theoretical framework is highly simplified. To focus on the causal mechanisms that are important for trade policy in de­ veloping democracies, I abstracted from many details as well as other po­ tentially important causal stories. This means that certain questions have been left unanswered and some issues have also been left unresolved, which present ample opportunities for further research. Therefore, I briefly discuss below five key issue-­areas in which more theoretical and empirical work can and should be done to make more progress in the study of democracy and trade policy in the developing world. With respect to the first issue-­area, note that the theoretical framework here has exclusively concentrated on analyzing trade policy making in the domestic arena, which is determined by domestic political institutions and the preferences of the main productive groups in society. Yet, in reality, we know that the trade policy choices of governments in developing democracies are

238 / Chapter Eight

also influenced by strategic interaction with other states at the international system level as well as in international institutions such as the WTO. Thus, to develop a more comprehensive theory of the political determinants of trade policy in new and established democracies, one needs to extend the models in this manuscript into a “two-­level” game-­theoretic framework (Putnam 1988; Milner 1997). The two-­level game-­theoretic framework will allow one and other researchers to explore in a unified theoretical model how domestic political institutions and international politics at the system level to shape the trade policy decisions of policy makers in developing democracies. Scholars have developed sophisticated two-­level game models of trade politics.18 Replicating this two-­level framework in the context of my models will be extremely challenging. However, it may be worthwhile to pursue this research agenda, as this may yield richer theoretical insights about the political economy of trade protection in developing countries. In terms of the second issue-­area, it is worth mentioning here that future literature on trade politics in developing democracies must provide richer models of the workings of the economy than those presented in this book. Specifically, I only examined situations in which the degree of interindustry labor mobility is given exogenously. While this is a common assumption in models of trade policy that focus on labor market conditions,19 we do know that, in reality, capital accumulates and technology changes over time, which in turn may affect the temporal dynamics of interindustry labor mo­ bility. Incorporating these dynamics of  technological change and accumula­ tion into the theoretical framework will be an important step forward. Such extensions may also help account for temporal variation in trade barriers within developing democracies. Third, research about trade politics in new democracies in this manu­ script focuses on how expansion of the selectorate and electoral transition after the transition to democracy affects trade politics. Exploring the effect of the expansion of the selectorate on trade politics in new democracies is an important but just one of other potential causal accounts that may explain trade policies in new democratic regimes. For instance, it is plausible that the nature of the selectorate prior to democratization or even the type of authoritarian regime in place (e.g., military or single party) may affect the type of political parties and leaders—­and the trade policy preferences of these actors—­that emerge after the transition to democracy.20 This may sub­ sequently affect the nature and content of trade policies, including the level of trade restrictions. It is also plausible that the nature of the selectorate prior to the democratic transition may influence the kind of “trade policy coalitions” that emerge after democratization. Whether or not these claims

Conclusion / 239

are valid requires a substantial amount of further research. The details of these alternative causal accounts need to be developed in much more depth, which is beyond the scope of this manuscript but deserves to be examined in the future. The fourth issue-­area that needs attention is the inescapable fact that many autocratic states continue to exist in the developing world. Further, the trade policies that are implemented by some autocracies such as China substantially affect global trade flows. Hence, building on the foundation of the models presented in the earlier chapters, it may be important for researchers in IPE to develop theoretical models that explain how relative factor mobility of labor (i.e., interindustry labor mobility) and capital may affect the trade policy choices of leaders in autocratic regimes. In chapter 2, I conducted a “bare-­bones” formal analysis of the influence of labor mobility in autocratic states by briefly examining what happens to the equilibrium level of tariffs and financial contributions by capital when the asset specificity of capital increases while the degree of interindustry labor mobility also increases. The simulations in some of the figures in chapter 2 illustrate the results from this particular conceptual exercise. Al­ though the results from the simulations in chapter 2 are interesting, they are clearly not sufficient. To develop a richer theory and comprehensive models of trade protection in developing countries, one should study more system­ atically how relative factor mobility determines the strategic behavior and policy choices of dictators in the issue-­area of trade policy. Finally, notwithstanding the exhaustive statistical tests and comparative case-­study research conducted in this book, the empirical research presented can be developed further into three directions. First, the temporal range of the sample of developing countries, which dates from 1972 to 2008, needs to be extended further back to the 1960s or earlier. This may be difficult to do, given the paucity of publicly available data on trade barriers across de­­ veloping countries before 1972. But it may be fruitful to substantially ex­ tend the temporal range of the sample used here (if possible). Doing so may enhance the empirical generalizability of the reported results if the esti­ mates derived from the expanded sample corroborate the theoretical claims. Second, statistical evaluation of the theoretical claims about the impact of domestic institutions on campaign contributions has been severely limited by the absence of cross-­national data on the amount of campaign contribu­ tions given to political parties in developing countries. That said, it might be feasible to compile or develop more extensive panel (i.e., over time) cross-­national survey data sets that contain infor­ mation about either the approximate amount of contributions given by

240 / Chapter Eight

industry-­executives or perceptions of the contributions that they offered to political actors in developing democracies. This survey data set can then be used to at least indirectly test the theoretical propositions about how do­ mestic political institutions affect the dynamics of campaign contributions in the context of trade policy making in democratic developing countries. Whatever future direction the study of democracy and trade policy might take, one hopes that this book has succeeded in providing an important stepping-­stone toward understanding the dynamics of trade politics and campaign contributions in developing country democracies.

Appendix

Mathematical Proofs

The appendix presents the mathematical proofs of the game-­theoretic model provided in chapter 2. The second section then formally describes the mathematical proofs of the game-­theoretic model presented in chapter 4.

Proofs for the Model on Democratization, Trade Policy, and Campaign Contributions This section first presents the equilibrium and comparative static results derived from the model presented in chapter 2. I start with the equilibrium result, which is presented in two parts. The first part characterizes the optimal tariff policy preference of the median voter in a new democracy, the optimal tariff policy choice of the parties in the newly democratized state, and the optimal monetary payments offered by the owners to the parties in the new democracy. The second part formally presents the proof of lemma 1 stated in the text. I then present the proofs for each proposition stated in the text and various “claims” mentioned in the chapter. Equilibrium Result I There exists a subgame-­perfect Nash equilibrium of the trade policy game in the new democratic regime described above, where (i) The median voter’s optimal tariff policy preference in the new democratic regime is given by

(A2.1)

*

mv

mv t dem = (1 − θ )

∂v H ∂p m′(t J )

242 / Appendix (ii)

The parties’ [ J ∈ {A , B}] optimal tariff policy preference is

t *J =

(A2.2)

β (∂v / ∂p) (H / L)[1 − θ mv ] β [φ + m′(t J ) / L] + (1 − β ) [m′(t J)]

(iii) The owners’ optimal monetary payment amount, given t* J , is characterized by

  β (∂ v / ∂p) (H / L)[1 − θ mv ] rˆ(t*J , φ , d J) = φ   β [φ + m′(t J ) / L] + (1 − β ) [m′(t J )]   

(A2.3)

Proof

I mentioned earlier that in a new democracy within a developing economy, the median voter is a worker in the labor force. Thus the median voter’s utility function is given as umv = u l. Differentiating umv with respect to tJ leads to mv

du dI mv = − d( p) + dt J dt J = (θ − 1) mv

(A2.4)

∂ v H  y ′0 ( p) + py ′( p) + y (p)  1 dT + − d( p)  + ∂p L  L  L dt J

∂v H  y( p )  1 dT + − d( p) + ∂p L  L  L dt J ∂v H t J mv + m′(t J ) = (θ − 1) ∂p L L = (θ mv − 1)

The first line in (A2.4) uses Roy’s identity, the second line employs equation (2.3) in the text, the third line invokes the MRT-­equals-­price condition y ′0 ( p) + py ′(p) = 0, where MRT is the marginal rate of transformation, and the fourth line is based on the fact that tariff revenue can be written as T = tJ [d (p)L − y (p)] = tJ m( p), which implies that dT / dt J = m(p) + t J m ′(p). Note that in (A2.4), (∂ v /∂ p) > 0 for countries that are relatively scarce in human capital (which implies developing countries) and thus import skill-­intensive mv* goods. Setting dumv/ dtJ = 0 and solving for tdem leads to1 (A2.5)

mv*

mv

t dem = (1 − θ )

∂v H ∂p m′(t J )

Recall that the government and the opposition’s objective function is 1  uJ = β u1 + (1− β ) u c + φ  t J2 + d J . I thus need to solve for t*J from duJ / dtJ. 2 

Mathematical Proofs / 243

Since u l = umv in uJ, and given that the median voter’s (a worker’s) utility is weighted by in each party’s objective function, it follows from (A2.4) that   ∂v H tJ β (θ mv − 1) + m′(t J ) ∂ p L L  

(A2.6)

Differentiating each component of uc in (A2.4) with respect to tJ and recalling that c / C is a constant, I first obtain

d(t J m(tJ )) = m(t J ) + tJ m′(tJ). dt J

Market clearing requires y(t J ) = d(t J ) + m(t J ), which ⇒

d(s (tJ)) ≡ − y(t J ) = dt J

−d(t J) − m(t J). The rewards from capital invested in y(t J ) is Finally,

dk(tJ ) = d(t J ). dt J

dr (tJ , φ ) = φtJ. Putting the above information together, and given dt J

that u c is weighted by (1 − β ), leads to c

du = (1 − β )[t J m ′(t J ) − φ t J]) dt J

(A2.7)

Putting (A2.6) and (A2.7) together and keeping in mind that ∂r (tJ , φ )/∂φ = φtJ, one gets the full functional form of  du J / dt J, which is (A2.8)

  ∂v H t J β  (θ mv − 1) + m′(t J ) + (1 − β ) [t J m ′(t J) − φ t J ] + φ t J ∂ p L L  

Setting du J / dt J = 0 in (A2.8) and solving for t*J leads, after some algebra, to

t *J =

(A2.9)

β (∂v / ∂p) (H / L)[1 − θ mv ] β [φ + m′(t J ) / L] + (1 − β ) [m′(t J )]

The owners know (by backward induction) that the party’s optimal tariff policy schedule is given by t*J in (A2.9). Substituting t*J in r (tJ , φ , dJ) leads to 1  2 r (t*J , φ, dJ ) = φ  (t*J ) + d J . This implies that 2  (A2.10) QED

∂r (t*J , φ, d J ) ∂t *J

 β (∂ v / ∂p) (H / L)[1 − θ mv ]  = rˆ(t*J , φ, d J) = φt *J = φ   β [φ + m′(t J ) / L] + (1 − β ) [m′(t J)]   

244 / Appendix

Proof of Lemma 1

I first prove part (b) of lemma 1 and then prove part (a) of this lemma. (b) To complete the proof of this part, I need to show that after the transition to democracy (i.e., when β = 1), the parties’ tariff policy proposal in a Downsian equilibrium will converge to the median voter’s optimal tariff policy preference. We then demonstrate that neither party will deviate from the median voter’s optimal tariff policy preference in a new demo*

*

mv * mv cratic regime. Suppose that t* A ≠ tdem and tB = tdem. Then the median voter * votes for party B, which, from (2.7) and (2.8), means that π (t* A , t B ) = 0 and *

* * * mv * (1 − π (t* A ,t B )) = 1 as D (t A , t B ) < 0 in this case. If, however, tA = t dem , then party mv* A can increase its chances of winning to 50%, or ½. Therefore, t* A = tdem , * mv* * * where π (t*A ,t * B ) = 1 / 2, strictly dominates tA ≠ t dem , where π (tA , tB ) = 0. * mv* mv * * Similarly, t* B = tdem , where 1 − π (tA , t B ) = 1 / 2, strictly dominates t* B t dem , where mv*

* * * 1−π (t* A , tB ) = 0. Thus in a Downsian equilibrium, tA = t B = t dem . Addition­ ally, let b = 1 (which captures the shift to a new democracy) in the expresH mv* mv ∂v , which ⇒ t*J = tdem sion for t*J . When b = 1, then t* for J = (1− θ ) ∂p m′(t J ) b = 1. * mv* Next suppose that t* A ≠ tB = tdem —­that is, suppose that party A chooses to deviate from the median voter’s optimal tariff policy preference when b = 1. * * * Moreover, assume without loss of generality that t* A < tB (i.e., tA ≠ t B ). * * From (2.7) and (2.8), D( t*A , t* B ) = 0 when tA = t B . Because D is differentiable * , t ), at (t* the conditions for equilibrium require that A B * * ∂D( t* ∂D (t* A , tB ) A ,tB ) = =0 ∂ tA ∂ tB

(A2.11)

From (2.6) and given that f is a single-­peaked density function, the condition in (A2.11) implies that (A2.12)

0=

* * ∂D( t* ∂D (t* A , tB ) A ,tB ) = = ( f (t Ab ) − f (t Aa ) )− ( f (t Bb ) − f (tt Ba ) ) ∂tA ∂tB

That is, (A2.13)

f (t Ab ) − f (t Aa ) = f (t Bb ) − f (t Ba )

Mathematical Proofs / 245

If f (tAb ) > f (t Bb), then the left-­hand side of (A2.13) is nonnegative. Since t Ba ∈ (t Ab , t Bb), the strict quasi concavity implies that D( t*A , t* B ) = 0. But because b b f ( t ) < f ( t ) ) = 0, D( t*A , t* . Thus the right-­ h and side of (A2.13) it follows that A B B is strictly negative. We found a contradiction. Symmetrically, if f (tBa ) ≥ f (tAa ), then f (tAb ) > f (t Bb), and so (A2.13) does not hold either. Therefore, we have * * mv* * proved that (A2.12) cannot be satisfied by any (t* A , tB ) with tA ≠ t B = t dem. mv* * * * Similarly (A2.12) cannot be satisfied by any (t* A , tB ) with t dem = tA ≠ t B . This

implies that there cannot exist an outcome where either party deviates from the median voter’s optimal tariff policy preference. QED (a) The proof of part (i) of this lemma shows that t*J = t mv* when b = 1. Since the median voter is a worker in a developing economy, the fact that t*J = t mv* when β = 1 ⇒ the parties’ equilibrium tariff policy in a new democracy is biased toward labor, who constitute the electoral majority. QED Claim 1 When the median voter’s level of interindustry occupational mobility is sufficiently high, he or she prefers a reduction in tariffs. Because the median voter is a worker, this implies that workers (i.e., labor) in a new democratic state prefer a reduction in tariffs when their occupational mobility is reasonably high. Proof

Suppose that the median voter’s degree of interindustry occupational mobility increases from a low of q mv to a high of q mv. Then $ q mv and q mv such that q mv > q mv with θ mv ∈[0 ,1] and θ mv ∈ [0 ,1]. Substituting q mv for mv mv* = (1 − θ mv )(∂ v / ∂p)(H / m′(t J)). Substituting q mv for q mv in t dem leads to tdem mv* q mv in t dem leads to t mv = (1 − θ mv )(∂ v / ∂p)(H / m′(t J)). Since θ mv > θ mv, it dem mv follows that tdem as (1 − θ mv ) < (1 − θ mv ) ∀θ mv, θ mv ∈ [0,1]. Further, < t mv dem

dI mv dI mv dI mv ⇒ > > 0, which when dθ mv dθ mv dθ mv mv* t*J = tdem → 0. Hence an occupationally mobile voter prefers reduction in

note from equation (2.3) in the text,

tariffs, as this directly increases his or her income and (in turn) generates more utility from the lower tariff. Note that because u mv = u l, one can conduct the exercise described above to show that workers will prefer lower tariffs when their occupational mobility is sufficiently high. QED

246 / Appendix

Proof of Proposition 1 (a) One needs to show that β →1 (A2.14)

∂t* A < 0. First, observe that ∂θ mv

∂t *A −[β (∂ v / ∂p)(H / L)] = mv β [φ + m′(t J ) / L] + (1− β ) (m ′(t J )) ∂θ

After a democratic transition, b shifts from b = 0 to b = 1. Since the optimal tariff schedule is continuous, we let lim β →1 in (A2.14); this leads ∂t* −[(∂v / ∂p )(H / L)] A = . In this expression, φ + m′(t J )/ L > 0 as to lim β →1 mv (φ + m′ (t J ) / L) ∂θ

f > 1 and −1 < (m′ (t J) / L) ≤ 0. Further, −[(∂ v / ∂ p)(H / L)] < 0 since (∂v / ∂p) > 0. ∂t* This implies that lim β →1 Amv < 0. QED ∂θ (b) From the proof of claim 1, we know that the occupationally mobile median voter and occupationally mobile workers prefer low tariffs. If the government (party A) reduces tariffs t*A in this case, then for t*A < tB*, the median voter and workers vote for A, and it follows from (2.6) that D( t*A , t* B ) > 0. * * p ( t , t ) = 1, ) > 0 From (2.7), D (t*A , t* implies that which is the upper bound A B B of π (t*A , t*B ) ∈ [0,1]. If the government does not reduce t*A in this case, then * * for t*A > tB*, D( t*A , t* B ) < 0 and p (t A ,t B ) = 0. Thus the government increases its likelihood of reelection if it reduces t* A when the median voter’s (and thus the workers’) degree of interindustry occupational mobility increases. QED Proof of Proposition 2 * The owners’ marginal contribution amount, given t* A = t J , is defined by ˆr (t* J , φ, dJ ) in equation (A2.10). From (A2.10), we get after some algebra and collecting terms: (A2.15)

∂rˆ(t* β (∂ v / ∂p) (H / L)[(1 − θ mv ) + φ (θ mv − 1)] A , φ, dJ ) = (φ + (m′(t J )/ L) ∂φ

After the country makes a transition to a new democracy, b shifts from b = 0 to b = 1. Further, if the interindustry occupational mobility of the median voter (and thus workers) is high, then limθmv →1. Letting lim β →1 and limθmv →1 in (A2.15) leads to

lim mv

β →1, θ

→1

∂rˆ(tA* , φ , dJ) ∂φ

→ 0. QED

Mathematical Proofs / 247

Proof of Proposition 3 Recall that α = (1 − π (tA , tB )) and (1 − α ) = π (tA , t B ). From the proof of parts * mv* * (a) and (b) in proposition 1, π (t* A , tB ) = 1 and (1 − π (tA , t B )) = 0, when t A = t dem ∂t* A < 0. Substituting π (t* A , t B) = 1 into the expression for a ∂θ mv ∂t* leads to a = 0. Thus a strictly decreases when lim β →1 Amv < 0. ∂θ and lim β →1

Equilibrium Result for NTBs In the model for NTBs, there exists a subgame-­perfect Nash equilibrium in the new democratic regime where (i)

The median voter’s optimal NTBs policy preference is

(A2.16) (ii)

*

∂v H ′ ∂p m (t J + tJ,n )

The parties’ [ J ∈ {A , B}] optimal NTBs policy preference is

(A2.17)

    

mv t mv ) J , dem = (1 − θ

t *J ,n =

β (∂v / ∂p)(H / L)[1 − θ mv ] − t J [(m′(t J + t J, n )(1 − β ) + β (φ + (m′(t J + t J, n ) / L))] [β (m′(t J + t J ,n )(1 / L − 1) + m′(t J + t J , n ) + βφ]

(iii) The owners’ optimal monetary payment amount given t*J,n is characterized by

(A2.18)

rˆ(t *J ,n ,φ , d J ) = φ

 β (∂v / ∂p)(H / L)[1 − θ mv ] − t J [(m′(t J + t J ,n )(1 − β ) + β (φ + (m′(t J + t J ,n ) / L))]      [β (m′(t J + t J, n )(1 / L − 1) + m′(t J + t J, n ) + βφ]  

Proof

Substituting tJ with tJ in (A2.4) leads to (A2.19)

du mv ∂v H t J = (θ mv − 1) + m′(t J) dt J ∂p L L mv*

*

mv Setting du mv / dt J = 0 and solving for t J ,dem leads to t Jmv , dem = (1 − θ )

∂v ∂p

248 / Appendix

H . After substituting tJ with t J in (2.4) in the text, the owners’ utility m′(tJ + t J,n ) function is (A2.20)

1  u c (t J ,φ ) = k(t J ) + (c / C)[s(t J ) + t J m(t J )] − φ  t J2 + d J 2 

while m′( tJ ) = m′(t J ) = m′(t J + t J , n ); I let m′(tJ + t J , n) < 0. Recall that dumv ∂v H t J = (θ mv − 1) + m′(t J) ∂p L L dt J

(A2.21)

Further, after substituting tJ with t J throughout the entire model, I obtain (A2.22)

duJ  mv  ∂v H t J + m′(t J ) + (1 − β )[t J m′(t J ) − φ t J ] + φ t J = β  (θ − 1) ∂ dt J p L L  

which is equivalent to

(A2.29)

du J  mv  ∂v H t J + t J, n = β (θ − 1) + m′(t J + t J, n ) + ∂p L L dt J   (1 − β )[ tJ m′(t J + tJ,n ) − φ(t J + t J,n ) ] + φ(t J + tJ,n)

From (A2.29), we obtain after some algebra (A2.30)   t *J ,n =

β (∂v / ∂p)(H / L)[1 − θ mv ] − t J [(m′(t J + t J ,n )(1 − β ) + β (φ + (m′(t J + t J ,n ) / L))] [β (m′(t J + t J ,n )(1 / L − 1) + m′(t J + t J , n ) + βφ] Using the proof of equilibrium result I and equation (A2.30), the optimal payment amount provided by the owners for the case of NTBs is given by ∂r (t*J , n ,φ , d J ) (A2.31)   = φ t*J , n ∂t*J , n  β (∂v / ∂p)(H / L)[1 − θ mv ] − t J [(m′(t J + t J ,n )(1 − β ) + β (φ + (m′(t J + t J, n ) / L))]  =φ    [β (m′(t J + t J ,n )(1 / L − 1) + m′(t J + t J ,n ) + βφ]   = rˆ(t*J , n ,φ , d J ) QED

Mathematical Proofs / 249

Proof of Lemma 2 (a)

Follows directly from the proof of part (a) in lemma 1.

(b)

Follows directly from the proof of part (b) in lemma 1. QED

Proof of Proposition 4 (a) Note that ∂t *J, n

(A2.32)

∂θ

mv

=

−[β (∂ v / ∂ p)(H / L)] ′ [β (m (t J + t J, n ) (1 / L − 1) + m′(t J + t J ,n) + βφ]

Since b shifts from b = 0 to b = 1, I let lim β →1 in (A2.32), which leads to (A2.33)

lim β →1

∂t *J ,n ∂θ

mv

=

−[(∂ v / ∂ p)(H / L)] [(m′(t J + t J ,n)(1 / L − 1) + m′(t J + t J ,n) + φ]

In (A2.33), [(m′(t J + t J,n )(1 / L − 1) + m′(t J + t J,n ) + φ ] > 0 as f > 1, (m′(tJ + t Jn, ) (1/L −1) > 0, and −1 < (m′(tJ + t J,n) ≤ 0. Further, −[(∂ v / ∂p)(H / L) ] < 0 since ∂t* J,n (∂v / ∂p) > 0. This implies that lim β →1 mv > 0. QED ∂θ (b) Note that (A2.34)

∂rˆ(t * J ,n , φ, dJ ) ∂θ

mv

− β (∂ v / ∂ p)(H / L)   =φ  ′ ′ β βφ m ( t + t )( 1 / L − 1 m ( t + t ) + ] [ ( ) + J J ,n J J ,n  

It was shown above that for lim β →1, [(m′(tJ + t J,n )(1/L − 1) + m′(t J + t J, n ) + φ]> 0 ∂rˆ(t*J, n , φ, dJ ) < 0. QED and − [(∂v/ ∂p)(H / L)] < 0. Hence it follows that lim β →1,φ →1 ∂θ mv

Proofs for the Model on Particularism, Trade Policy, and Campaign Contributions This section first presents the equilibrium and comparative static results derived from the model presented in chapter 4. I start with the equilibrium result, which characterizes the optimal tariff policy preference of the median voter in a consolidated democracy (where the level of particularism is held fixed), the optimal tariff policy choice of the parties in the consolidated democracy, and the optimal monetary payments offered by the owners to the parties in the consolidated democracy. The second part formally presents the proof of lemma 1 that is in the text of chapter 4. I then present

250 / Appendix

the proofs for each proposition stated in the text of chapter 4 and various “claims” mentioned in the chapter. Equilibrium Result There exists a subgame-­perfect Nash equilibrium in the model on particularism and trade policy where (i)

The median voter’s optimal tariff policy preference is given by

∂v H tˆ mv = (1 − θ mv ) ∂p m′(t J )

(A4.1) (ii)

The optimal tariff policy preference of leaders from parties J ∈{A , B} is

(A4.2)

tˆ J =

(λ − 1)(∂v / ∂p)(H / L)[θ mv − 1] λ(φ − m′(t J / L) + ∆φ + (1 − λ )(m′(t J ))

(iii) The owners’ optimal monetary payment amount, given tˆJ , is characterized by

(A4.3)

  (λ − 1)(∂v / ∂p)(H / L)[θ mv − 1] rˆ(tˆJ ,φ , dJ ) = φ   λ(φ − m′(t J ) / L) + ∆φ + (1 − λ )(m′(t J ))   

2

Proof

The derivation of tˆ mv in equation (A4.1) is described in the proof of the equilibrium tariffs result in the democratization and trade policy model. It is thus not repeated here. I turn to solve for tˆ J, set by the leaders from the ruling party since, by symmetry, the opposition party leader’s optimal tariff choice in equilibrium is similar. I then formally characterize rˆ(tˆJ, φ, dJ). Substituting u l, u c, and r (tJ , f , dJ) in (4.3) leads to the full form of the ruling party leaders’ objective function: 1  (A4.4)   λ[k(t A ) + (c / C)[s(t A ) + t A m(t A )] − φ  t 2A + d A )] + 2   1 2  l (1 − λ )[I − pd(p) + U[d(p)]] + ∆  φ  t A + d A     2 To solve for tˆJ from (A4.4), I need to first derive the full functional form of  du J / dt J. To this end, first note from the median voter theorem that party leaders from the ruling (and opposition) party will set tariff policy to maximize the median voter’s utility to increase their chances of winning office.

Mathematical Proofs / 251

From equation (A2.4), and given the median voter’s (a worker’s) utility is weighted by (1 − λ ), it follows that   ∂v H t J (1 − λ )  (θ mv − 1) + m′(t J ) ∂ p L L  

(A4.5)

From the proof of the derivation of the equilibrium tariff level in chapter 2, ∂u c = [tJ m ′(tJ) − φ t J]. Since the leader of the ruling party we also know that ∂tJ weighs u c by l , we get

λ

(A4.6)

∂u c = λ[t J m′(tJ ) − ∆φ tJ] ∂t J

Differentiating ∆φ[(1 / 2)tJ2 + d J ] with respect to tJ leads to DftJ. Putting (A4.5), (A4.6), and DftJ together, one gets the following full functional form of  du J / dt J: (A4.7)

  ∂v H t J λ[t J m′(t J) − ∆φ t J] + (1 − λ ) (θ mv − 1) + m′(t J )  + ∆φtJ p L L ∂  

Setting duJ / dt J = 0 in (A4.7) and solving for the equilibrium tariff policy leads to (A4.8)

tˆ J =

(λ − 1)(∂v / ∂p)(H / L)[θ mv − 1] λ(φ + m′(t J ) / L) + ∆φ + (1 − λ )(m′(tJ ))

By backward induction, the owners must offer the party leaders a finan­ cial payment of at least r (t J , φ, d J) ≥ tˆJ to inauence tariff policy. Since r (t J , φ , dJ) ≥ tˆJ is sufficient, I substitute tˆJ in r (tJ , φ , dJ), which leads to 1  r (tˆJ , φ , d J) = φ  (tˆ J)2 + d J: 2    ∂r (tˆ J ,φ ,d J ) (λ − 1)(∂v /∂p ) (H / L)[θ mv − 1] = φtˆ J = φ    ˆ ∂t J  λ(φ + m′ (t J ) / L) + ∆φ + (1 − λ ) (m′(t J ))  = rˆ(tˆ J ,φ ,dJ)

(A4.9)

As shown in the proof of equilibrium result I in chapter 2, the equilibrium offer of rˆ(t*J , φ, dJ) is optimal for the owners. QED Proof of Lemma 1

(a) The median voter’s optimal tariff policy preference from the previous proof is given by

252 / Appendix

H mv mv ∂v tˆ = (1 − θ ) ∂p m′(tJ)

(A4.10)

From the median voter theorem, the leaders from the ruling party will set tariff policy to maximize the median voter’s utility with respect to tJ in order to maximize their probability of reelection. Recall from chapter 2 that 1  π (tA , tB ) = 1 / 2 0 

(A4.11) where D( tA , t B ) =

∫

γ A (t A, t B )

f (z)dz −

∫

γ B (tA , t B )

if D(t A , t B ) > 0 if D( t A , t B ) = 0 if D( t A ,t B ) < 0 f ( z)dz. Given this expression and

(A4.11), we know from the median voter theorem that ∂π (tA, tB) / ∂t = 0 only when the following condition holds: (A4.12)

mv

∂π ∂u ≡ =0 ∂t J ∂t J

where (and hereafter) ∂π denotes ∂π (tA ,t B ). Because the ruling party weighs + the median voter’s (who is a worker) utility function (u mv ) by (1 − λ ), the expression in (A4.12) can be rewritten as mv

(A4.13)

∂π ∂u ≡ (1 − λ) =0 ∂t J ∂t J

We now need to show that the condition in equation (A4.12) holds when the level of particularism is high—­that is, when lim λ → 1. From the right-­ hand side of (A4.13), one obtains (A4.14)

 ∂ u mv ∂π ∂u mv ≡ −λ   ∂ tJ ∂ tJ ∂ tJ 

  = 0 

Observe that ∂π / ∂ t J = 0 in (A4.14) when ∂u mv / ∂t J = λ (∂u mv / ∂t J ). Further, ∂u mv / ∂t J = λ(∂u mv / ∂t J ) only when lim λ → 1. This implies that when lim λ → 1, (A4.15)

mv  ∂u mv  ∂π ∂u = lim λ →1  ≡  ∂t J  ∂t J ∂t J  

Equation (A4.15) implies that ∂π / ∂t J = ∂u mv / ∂t J, which is identical to ∂v H the expression in (A4.12). Since ∂u mv/∂tJ = 0 when tˆ mv = (1 − θ mv ) , ∂p m′(t J ) it follows that ∂π / ∂t J = ∂u mv / ∂t J when ∂π / ∂t J = tˆ mv. In other words, the ruling party leaders’ tariff policy in equilibrium converges to the median

Mathematical Proofs / 253

voter’s optimal tariff policy preference when λ → 1, as this maximizes their reelection probability. Since the tariff policy proposed by the ruling party leaders in equilibrium (tˆ J) converges to tˆ mv, the median voter’s realized utility is maximized from the tariff policy tˆ mv proposed by the leaders during elections. Hence the median voter votes for the ruling party leaders when tˆJ = tˆ mv, as claimed. QED (b) From the proof of part (a) of this lemma, we know that the ruling party leaders in a particularistic democracy maximize their likelihood of winning the election after shifting the equilibrium tariff policy to the median voters’ tariff policy preference. Since the median voter in a developing democracy is a worker, this implies that when the level of particularism is high, the ruling party maximizes its reelection probability by setting tariff policy that is biased toward workers, who constitute the electoral majority. QED Proof of Proposition 1 (a) First note that ∂tˆJ (A4.16)

∂θ

mv

=

[(λ − 1) (∂ v / ∂ p)(H / L)][λ (φ + m′(t J ) / L) + ∆φ + (1 − λ)(m′ (tJ ))] 2 [λ (φ + m′(t J ) / L) + ∆φ + (1 − λ)(m′ (tJ))]

Higher levels of political particularism imply that lim ∆ → 1. Letting lim ∆ → 1 in (A4.8) and collecting terms leads to (A4.17)

lim ∆→1

∂tˆ J ∂θ mv

=

[(λ − 1)(∂v / ∂p)(H / L)] [λ(φ + m′(t J ) / L) + φ + (1 − λ )(m′(t J ))]

An increase in particularism also implies that lim λ → 1. One can easily check from (A4.17) that if lim λ → 1, then lim λ →1

∂tˆ J → 0, which implies ∂θ mv

that tˆ J < 0 when lim λ → 1, lim ∆ → 1, and θ mv is high. QED (b) From the proof of part (a) in lemma 1, we know that  tˆJ = tˆmv when λ → 1. Further, from the proof of part (a) of this proposition, we know that that ∂tˆ J lim λ →1 mv → 0. The occupationally mobile median voter will vote for the ∂θ ruling party leaders with probability 1, as its utility gets maximized when

254 / Appendix

tˆ J < 0. Consequently, when tˆ J < 0, the ruling party leaders’ probability of reelection gets maximized. QED (c) Note that (A4.18)

∂rˆ(tˆJ , φ, d J ) mv ∂θ

(λ − 1)(∂v / ∂p)(H / L)   = 2φ   ′ ′ λ φ φ λ + + + − ∆ 1 ( m ( t ) / L ) ( ) ( m ( t )) J J  

When the particularism is high, then lim λ → 1 and lim ∆ → 1. When ˆ ˆ lim λ → 1 and lim ∆ → 1, then ∂r (t J ,φ ,d J ) strictly decreases. QED ∂θ mv Proof of Proposition 2 When the level of particularism is low, it follows that lim λ → 0 and lim ∆ → 0. Letting lim λ → 0 and lim ∆ → 0 in the expression for rˆ(tˆ J ,φ ,d J) leads to (A4.19) From (A4.19),

 − (∂v / ∂p)(H / L)[θ mv − 1]  rˆ(t *J , φ, dJ ) = φ    (m′(t J ))   ∂rˆ(tˆ J ,φ ,d J )  − (∂v /∂p ) (H / L)  . Note that ∂rˆ(tˆ J ,φ ,d J) >0 =φ mv  ∂θ mv (m′ (t J) ) ∂θ  

strictly increases for lim φ → 1, since − (∂v /∂p ) (H / L)θ mv < 0, (m′ (t J)) < 0, and

φ > 0. Letting lim λ → 0 and lim ∆ → 0 in the expression tˆ J in (A4.16) leads to (A4.20)

tˆ J =

(λ − 1)(∂v / ∂p)(H / L)[θ mv − 1] λ(φ + m′(t J ) / L) + ∆φ + (1 − λ )(m′(t J))

From (A4.20), (A4.21)

∂tˆJ ∂θ

mv

=

− (∂v / ∂p)(H / L)](m′(t J )) [(m′(t J))]

2

when lim λ → 0 and lim ∆ → 0. In (A4.21), − (∂v /∂p ) (H / L)](m′(t J)) > 0 in the numerator, since m′(t J ) < 0 and [(m′(t J) ) ]2 > 0 in the denominator. Thus ∂tˆ J > 0 for lim λ → 0 and lim ∆ → 0. QED ∂θ mv

Mathematical Proofs / 255

Proof of Claim 1 Recall that ∂π / ∂t J = 0 in (A3.14) when ∂u mv / ∂t J = λ (∂u mv / ∂t J ). When lim λ → 0, then ∂u mv / ∂t J ≠ λ (∂u mv / ∂t J ) as λ(∂u mv / ∂t J ) = 0 for lim λ → 0. Note that ∂u mv / ∂t J ≠ λ (∂u mv / ∂t J ) implies that ∂π / ∂ t J ≠ 0 for lim λ → 0. In other words, party leaders in party-­centered democracies fail to maximize their probability of reelection by locating their equilibrium at the median voter’s optimal tariff policy choice. This therefore drives them to shift the equilibrium policy away from the median voter’s tariff policy preference and (thus) toward capital’s tariff policy preference. Extending the Model of Particularism and Trade Politics to NTBs Similar to chapter 2, I assume that the NTB policy choice adopted by the respective leaders from J ∈{A ,B} is given by t J. This is common knowledge to all the players in the game. t J = t J + t J , n, where t J is the respective tariff policy choice and t J ,n (that denotes NTBs) is the additional policy instrument that the party leaders employ for trade protection. The domestic price of the imported good (in terms of the exported good) is thus  p = p * + t J = p * + t J + t J ,n, where p* is the fixed world price, while and were defined above. Substituting t J with t J in the owners’ utility function in (2.4) in chapter 2 leads to 1  u c (t J ,φ ) = k(t J ) + (c / C)[s(t J ) + t J m(t J )] − φ  t 2J + d J  2  Each worker’s utility is given by u l (p * + t J , I l ) = I l − (p * + t J )d(p * + t J ) + U[d(p * + t J )] in the NTBs case. Moreover, as mentioned in the text, t J ,n ∼ U[t n , tn ] and t J,n ∈ ℜ +. Finally, the full form of the objective from the leaders of each party is given by

(A4.22)

1  λ[k(t J ) + (c / C)[s(t J ) + t J m(t J ) − φ  t 2A + d A )] + (1 − λ )[I l − pd(p) + U[d(p)]] + ∆ 2   1 2 1 2   l t A + d A )] + (1 − λ )[I − pd(p) + U[d(p)]] + ∆  φ  t A + d A   J ) + t J m(t J ) − φ  2    2 (A4.23)

where p = p * + t J = p * + t J + t J , n. NTBs Equilibrium Result

I use the same procedure described in the proof of lemma 1 to solve for the optimal level of NTBs that the party leaders set in equilibrium. After doing so, we find that in the subgame-­perfect Nash equilibrium, the level of NTBs is characterized by

256 / Appendix mv

(A4.24)

tˆ J ,n =

λ t J m′(tJ + tJ, n ) + (∂v / ∂p)(H / L)[θ (1 − λ ) − (1 + λ )] [λφ + ∆φ + m′(t J + t J ,n ) / L − λ m′(t J + t J , n )]

By backward induction and the solution in (A4.24), the optimal payment (legal contributions and illegal donations) that the owners offer to the candidates in equilibrium is given by rˆ(tˆ J , n , φ, d J ) = (A4.25)

 λ t m′(t J + t J ,n ) + (∂v / ∂p)(H / L)[θ mv (1 − λ ) − (1 + λ )]  φ J    [λφ + ∆φ + m′(t J + t J ,n) / L − λ m′(tJ + t J , n )]   Proof of Lemma 2

The proof of this lemma follows directly ∀t J ,n ∼ U[t n , t n] from the proof of parts (a) and (b) in lemma 1. QED Proof of Proposition 3 (a) To start with, (A4.26)

∂ tˆ J mv ∂θ

=

[(∂v / ∂p)(H / L)(1 − λ )] 2 [λφ + φ + m′(t J + t J ,n ) / L) + ∆φ + (1 − λ )(m′(t J + t J ,n ))]

Higher levels of particularism imply that lim λ → 1 and lim ∆ → 1. One can ∂tˆ J easily check that for lim λ → 1 and lim ∆ → 1, strictly decreases. QED ∂θ mv (b) When political particularism is low, then lim λ → 0 and lim ∆ → 0. One ˆ can easily check that for lim λ → 0 and lim ∆ → 0, ∂t J strictly increases. ∂θ mv For lim λ → 0 and lim ∆ → 0, (A4.27)

 (∂v / ∂p)(H / L)[θ mv − 1]  rˆ(tˆJ , n , φ, dJ ) = φ   [m′(t J + t J ,n) / L ] < 0   

In (A3.27), [m′(t J + t J, n ) / L] < 0 and (∂v /∂p ) (H / L)[θ mv − 1] < 0, which ⇒ rˆ(tˆ J ,n ,φ , d J) strictly increases for lim λ → 0 and lim ∆ → 0. QED

Notes

C h a p t e r On e

1.

Cited from Beres and Burnetko (2009: 103); also see Wałęsa (1991) and Millard (2000). 2. Wałęsa (1991); Beres and Burnetko (2009: 104–­5). 3. See Cooper (2008). 4. Lind (1997: 97–­98). Several policy pundits who subscribed to the “Washington Consensus” in the 1990s often publicly argued that economic (e.g., trade) liberalization in new democracies discourages leaders in these states from rigging or manipulating elections owing to reputational concerns. Examples of these pundits include Lind (1997) and Luttwak (1999). 5. For the case of Argentina and Brazil, see Weyland (2002); for Ghana, see Ayine (2004); and for the Philippines, see Balisacan and Hill (2003). 6. For the case of Bolivia and Peru, see Weyland (2002); for Nigeria, Pakistan, Nepal, and Turkey, see Kliot and Newman (2013). 7. The data for operationalizing import duty coverage ratio are listed in chapter 3. 8. The term “newly consolidated” democracies refers to new democratic regimes that did not relapse back into autocratic rule within three years after the formal transition to a full-­fledged democracy. 9. This is captured in the data used to illustrate figure 1.1. 10. Przeworski et al.’s (2000) definition of democratic countries is described in chapters 3 and 5 of this book. 11. These demand-­side theories of trade protection are discussed in detail in the “Politics of Trade in Current Research” section of this chapter. Prominent works on demand-­ side explanations of trade policy include, for instance, Busch and Reinhardt (1999, 2000), Hiscox (1999, 2002), Rogowski (1987), Grossman and Helpman (1994, 2002), Baldwin and Magee (2000), and Scheve and Slaughter (2001). 12. Existing supply-­side studies of trade protection are also discussed in the third section of this chapter. Prominent supply-­side studies of trade politics include Mansfield et al. (2000); Kono (2006, 2008); Hankla (2006); McGillivray (2004); Mansfield, Milner, and Pevehouse (2007, 2008); Park and Jensen (2007); and Grossman and Helpman (2005). 13. Henisz and Mansfield (2006); O’Rourke et al. (2007); Kono (2008).

258  /  Notes to Chapter One 14. Studies that examine the effect of factor endowments on cross-­sectional variation in trade restrictions include, for example, Rogowski (1987, 1990), Krueger (1990), Davis (1996), and Dutt and Mitra (2002). 15. Here are some examples of studies that explore the impact of particularism in developing countries: (i) Hallerberg and Marier (2004) and Hicken and Simmons (2008) explore particularism’s impact on fiscal policies (e.g., public spending); (ii) Chang (2008) explores particularism’s impact on corruption; (iii) Garland and Biglaiser (2009) explores particularism’s impact on foreign direct investment; and (iv) McGillivray (2004), Hankla (2006), and Kono (2009) explore its impact on trade policies. 16. Studies on trade policy preferences and outcomes in Latin America include Baker (2005), Nielsen (2003), and Brooks and Kurtz (2007). Studies on these preferences and outcomes in central and eastern Europe include Frye and Mansfield (2004). 17. Theories about trade policy preferences held by individuals have been extensively tested by using survey data drawn, for example, from the National Election Studies (NES), National Annenberg Election Study (NAES), International Social Survey Pro­ gram (ISSP), and the Americas Barometer Survey. 18. Henisz and Mansfield (2006); O’Rourke et al. (2007); Kono (2008, 2009). 19. Endogenous models of trade policy that examine lobbying in trade policy but not the specific institutional context in which lobbying occurs include Trefler (1993), Baldwin (1999), and Mitra (1999). 20. The model that I construct builds on but also substantially expands Feenstra’s (1984) and Grossman and Helpman’s (1994, 2002) work. In particular, unlike the models presented by these economists, the models presented in this book formalize key political and institutional features of (i) new democratic regimes (see chapter 2) and (ii) particularistic democracies (see chapter 3). 21. Since workers constitute the vast majority of the electorate in developing countries, the “median voter” is likely to be a worker rather an owner of industries. This is also assumed in formal models of trade politics in developing countries; see, for example, Rama and Tabellini (1998) and Dutt and Mitra (2002). 22. See Facchini and Willmann (2001); Rama (1994); Matschke (2004); Cameron et al. (2007). 23. See Hamermesh (1987, 1993), Moscarini (2001), Kletzer (2004), and Cameron et al. (2007). 24. Also see Alt et al. (1996, 1999) for a more detailed exposition of this claim. 25. Hammermesh (1987, 1993); Facchini and Willmann (2001); Rama (1994); Matschke (2004). 26. It is important to emphasize here, however, that this model attempts to neither identify the determinants of particularism in developing countries nor explain why particularism varies substantially across developing democracies. Doing so is beyond the scope of this project, and moreover, it has also been studied extensively in the comparative politics literature. For this literature, see Mainwaring and Scully (1995), Mainwaring and Shugart (1997), Carey and Shugart (1995, 1998), Wallack et al. (2003), and Gorfman (2007). 27. The main features of particularism that are incorporated in my model are drawn from Carey and Shugart (1995, 1998), Wallack et al. (2003), Hicken (2009), and Cox and McCubbins (1993, 2001). 28. These data are taken from World Economic Forum (2004b).

Notes to Chapter Two  /  259 29. Import duties in India have been approximately 12% higher than in Brazil during the last three decades. 30. I show in the case-­study chapters of this book that governments in Brazil have successively cut trade barriers in the last two decades more radically than governments in India and South Africa. 31. Brazil’s transition to democracy occurred in 1985. South Africa’s democratic transition occurred in 1994. 32. The following studies show that Brazil is a candidate-­centered system: Ames (2001), Mainwaring (1999), Shugart and Carey (1992), Geddes (1994), Samuels (2000), and Mainwaring and Perez-­Liñan (1997). 33. Since South Africa uses a closed-­list proportional representation (PR) electoral system, it is considered a party-­centered democracy (Sadie 1998; Brooks 2004). India is also a party-­centered democracy where party leaders have strong control over rank-­ and-­file members from their own party (see Chhibber 1999; Sridharan 2009). 34. As estimated by the International Monetary Fund (IMF) in 2012, the size of Brazil’s economy in nominal GDP terms is approximately $2.2 trillion, while the size of In­ dia’s economy is roughly $1.9 trillion. 35. The nominal per capita income in 2012 in Brazil was US$11,139, while it was US$7,508 in South Africa. Brazil and South Africa are therefore middle-­income countries. Chapter two

1.

In other words, I focus on how the relative factor mobility of labor (i.e., workers) and capital (i.e., the capital owners) affects trade policy preferences and consequently prospects for trade reforms. I focus on labor and capital because these are the two main factors of production in existing general equilibrium and political economy models of trade (see, for example, Mussa 1982; Grossman 1984; Goldberg and Maggie 1999; Dutt and Mitra 2002; Milner and Kubota 2001, 2005). There are additional sound theoretical and empirical reasons for focusing on the role of labor and capital; I briefly discuss these reasons later in this chapter. 2. Focusing on the degree of factor mobility of labor and capital to derive their respective trade policy preferences has been done earlier in the IPE trade policy literature. For instance, several studies assess theoretically the impact that the level of asset specificity of industries has on trade politics. For this, see Alt et al. (1996), Hathaway (1998), Gilligan (1997), Alt and Gilligan (1994), Hiscox (2001a), and Zahariadis (2002). Additionally, recent work by Hainmueller and Hiscox (2008), Hiscox and Rickard (2002), as well as economists such as Winters (2002), Matschke (2004), and Cassing (1996), explores how the interindustry occupational mobility of workers determines their trade policy preferences and trade policy outcomes. 3. Scholars have debated whether the Heckscher-­Ohlin or the specific factors model provides a better account of the political economy of trade policy and thus the trade policy preferences of individuals in developing countries. Edwards (1995), Yang (1995), Dutt and Mitra (2002), and Milner and Kubota (2005) subscribe to the view that the Heckscher-­Ohlin model provides a more compelling theoretical account of trade politics and trade policy attitudes of individuals in developing states. Economists such as Deardorff (1984), Xu (2001), Topalova (2004), and Porto (2004) argue that the specific factors model (or variations of this model) provides an accurate approximation of trade politics and trade policy attitudes in developing states.

260  /  Notes to Chapter Two 4. Milner and Kubota (2005: 116) also emphasize that “developing countries by definition possess relatively less capital than labor.” This implies from the Stolper-­ Samuelson theorem that the scarce factor in developing states (capital) will favor trade protection while the abundant factor (labor) will favor trade liberalization. 5. General equilibrium models of trade by Mussa (1982) and Grossman (1984) were among the first to relax the assumption in the Heckscher-­Ohlin framework that factors (labor and capital) are perfectly mobile. Mussa (1982) introduced imperfect labor mobility in his model, while Grossman (1984) characterized the economy’s general equilibrium in a model where capital is imperfectly mobile. A new generation of models of trade policy by economists that examine the impact of the degree of factor mobility of labor and/or capital build on the models developed by Mussa (1982) and Grossman (1984); for these new models, see, for instance, Moscarini (2001), Cameron et al. (2007), Matschke (2006), and Costinot (2009). 6. Scholars have documented that domestic industry owners who seek protection from import competition in developing democracies (including “new democracies”) both lobby and offer campaign contributions to influence trade policy so that it suits their interests (Payne 1992; Gawande and Krishna 2003). For instance, shortly after Brazil’s transition to democracy, the main capital owners in the country included owners of heavy asset-­specific industries such as automobile and truck production, metal works, and petroleum processing (Payne 1992). These owners unambiguously preferred trade protection. They therefore financed the electoral campaigns of politicians to influence politicians to maintain high trade barriers. Campaign contributions were also offered by industry owners to influence trade policy in the posttransition period in Peru (see Durand 1992) and Turkey (Celasun and Arslan 1997). 7. Quasi rents constitute the difference between the rate of return in a factor’s current (best) use and the opportunity cost—­that is, the best alternative use. 8. This claim is drawn from recent theoretical work in IPE suggesting that a crucial determinant of the trade policy preferences of capital owners is the degree of asset specificity of the industries they own (Alt et al. 1996; Hathaway 1998; Hiscox 2001, 2002; Zahariadis 2002). These scholars theorize that industries or firms characterized by higher levels of asset specificity face greater adjustment costs when threatened by import competition (Alt et al. 1996; Gilligan 1997; Zahariadis 2002). Owners of asset-­specific industries consequently have greater incentives to lobby for more trade protection. 9. Studies have also documented that industry owners provide illegal donations (i.e., bribes) to influence a variety of economic policies, including trade protection. See, for example, Kaufmann and Vicente (2005) and Kaufmann et al. (2006). 10. The economist Harrison (2007: 16) suggests that occupationally mobile workers in developing economies can “easily relocate away from contracting toward expanding sectors in the context of  trade reforms.” This allows mobile workers to directly ben­ efit from trade liberalization. 11. For studies that examine the main political characteristics of new democratic and democratizing regimes—­the characteristics that are employed in the formal analysis—­ see Przeworski (1991), O’Donnell and Schmitter (1986), O’Donnell (1992), Diamond (1999), Di Palma (1990), and Horowitz (1992). 12. A common assumption in existing models of democratic transitions is that the transition to and hence emergence of a new democracy entails the move toward majority rule and the introduction of elections. For this, see Acemoglu and Robinson (2006), Boix (2003), Milner and Kubota (2005), and Bueno De Mesquita et al. (2002, 2003).

Notes to Chapter Two  /  261 Also see the comparative politics literature on democratization by Linz and Stepan (1996) and Bunce (2003). 13. There exists, to the best of my knowledge, two papers that present a formal model to study the effect of democratization (i.e., newly democratic states) on trade protection in developing countries (Milner and Kubota 2001) and developed countries in the nineteenth century (O’ Rourke and Taylor 2006). These scholars formally examine, as I do here, how extension of the franchise and the movement toward majority rule in newly democratic states affects tariff policies in developing countries (Milner and Kubota 2001) and developed countries (O’Rourke and Taylor 2006). I substantially extend their work by studying not only the impact that the move toward majority rule has on trade policy but also the effect of other critical features of new democratic regimes states (e.g., the presence of weak political parties in new democracies) on trade protection in the model. 14. Existing studies that employ formal models to study trade politics in developing countries also assume (as we do) that the median voter is a worker; see Rama and Tabellini (1998), Milner and Kubota (2001), and Dutt and Mitra (2002, 2005). 15. Defining tariff revenue as T = t J m( p ) is standard practice in existing models of trade policy; see Mayer (1984), Feenstra (2004), and Hillman (2002). 16. This assumption is also drawn from Grossman and Helpman (1994, 2002). 17. See Alt et al. (1996), Gilligan (1997), and Zahariadis (2002) for the claim that owners of asset-­specific industries have incentives to lobby for higher trade barriers. 18. Models of trade often directly or indirectly use some variant of the utility function for citizens as employed by Feenstra (2004). See, for example, Laussel and Riezman (2005) and O’ Rourke and Taylor (2006). 19. Several labor and development economists assume that the occupational mobility of workers is driven by their level of human capital or skill endowment or their investment in human capital; see, for instance, Alvarez (1999), Bandhyopadhyay (1997), and Moscarini and Vella (2003). This is intuitive, considering that human capital is equivalent to an individual’s education level and/or work experience, which, in turn, can drive his or her ability to be occupationally mobile. The results that I report do not alter substantively if we assume that occupational mobility is driven by skill endowment or investment in human capital. 20. We obtain similar results if q lmobile = (H l / L l ) / (H / L), where L l = 1, and θ lmobile = H l/ H. We assume the functional form for q lmobile defined above (and in this footnote) because it is plausible that the higher (lower) the individual’s degree of  human capital relative to the human capital level of other individuals (who he or she competes with in the labor market), the greater (lesser) his or her ability to shift jobs. 21. I can expand the total GDP equation to y 0 ( p )+ py( p) = wL + vH + l k, where k is the owners’ physical capital and l is the rent from this capital. If one assumes that the marginal rate of transformation (MRT)–­equals–­price condition holds in this case as well (i.e., y 0¢ ( p)+ py ¢( p) = 0), then the results reported below will not alter significantly or substantively. 22. For this result, see, for example, Hamermesh (1993), Cameron et al. (2007), Rama (2003), and Papageorgiou et al. (1992). 23. See proof of claim 2 and proposition 1 in the appendix. 24. Since the voters are distributed according to F and because the set of all voters who have preferences over tariff policy is g (tA , tB), one can define the turnout rate as

∫

dF.

γ (tA, tB )

262  /  Notes to Chapter Two 25. See Alt et al. (1996), Gilligan (1997), and Zahariadis (2002). 26. I assume that φ ≥ 1 (and φ ≠ 0) to ensure that the degree of asset specificity of the owners’ industries is not negligible or close (or converging to) measure 0. 1 f tJ + f dJ is rewritten as r(tJ ,f , dJ) = f [(1/2) t J + d J ]. 2 28. See, for example, Grossman and Helpman (1994, 2002), Goldberg and Maggie (1999), and Matschke (2004). 29. For examples of this literature, see O’Donnell and Schmitter (1986), O’Donnell (1992), Przeworski (1991), and Diamond (1999). 30. In other words, the institutional context influences the relative weight that the parties place on the preferences of workers and the owners. This has important consequences that are explored below. 31. Acemoglu and Robinson (2006: 175). 32. A country that has made a transition to a new democracy is one where (i) free and fair elections are held, (ii) parties respect the rules of the democratic game, and (iii) the incumbent party voluntarily leaves office if it loses the posttransition election. For this definition, see Przeworski (1991), Przeworski et al. (2000), and O’Donnell and Schmitter (1994). 33. It is worth noting here that the results derived from the model do not change if one assumes that the parties are ideological (i.e., partisan) and not merely Downsian. 34. For this, see Milner and Kubota (2005: 115). The term “selectorate” is taken from Bueno De Mesquita et al. (2003). 35. This claim is proved formally in the proof of claim 1 in the book’s appendix. Note that this particular claim posited in the text (which is derived from the model) is a standard result in decision-­theoretic models or trade policy preferences with or without “search frictions”; for examples, see Moscarini (2001) and Costinot (2009). 36. Wall Street Journal, September 18, 1986. 37. See Cororaton (1998). 38. For the Brazil case, see Payne (1992). For the Peru case, see Durand (1992). 39. Herberg (1990). 40. In the sparse formal literature on NTBs, scholars often assume that NTBs are a strategic complement to tariff policy. See, for example, the computable general equilibrium models of NTBs by Fugazza (2008) and Herberg (1990) mentioned above. The few existing decision-­theoretic or game-­theoretic models on NTBs also assume that NTBs are complementary to tariff policy and hence used for trade protection along with tariffs; for this, see Ederington and McCalman (2006) and Yu (2000). For these reasons, I formalize NTBs as a policy instrument that is complementary to tariffs and can thus be used for protection in addition to tariffs. 41. Before NTBs are implemented, political competition among the ruling party and opposition parties induces the political parties to also campaign on the issue of NTBs during elections. 42. Similar to tariffs, there exists an inverse relationship between NTBs and the asset specificity of industries in the model—­that is, owners of industries that produce the importable good g and are characterized by a high (low) degree of asset specificity are more (less) likely to lobby for higher NTBs to protect their industries from import competition. 27. After collecting terms, r(tJ , f, d J ) =

Notes to Chapter Three  /  263 43. Numerous theoretical and empirical studies by economists show that lowering NTBs reduces price rigidities. This increases the welfare of consumers. For this, see Herberg (1990), Beghin (2008), Bradford (2000), and Anderson and Wincoop (2001). Chapter three

1. Other sources that were also employed to operationalize tariffs include World Bank (1994, 1996, 1998–­2000), OECD (1996), and World Trade Organization (1998). 2. See, for example, Mansfield and Busch (1995), Gawande and Hansen (1999), Kono (2006), and Nicita and Olarreaga (2001). 3. For instance, the core NTB coverage ratio is used by Gawande and Hansen (1999) and Kono (2006) to operationalize NTBs. 4. Other measures of NTBs include, for example, the proportion of tariff lines covered by NTBs that I use for robustness tests. Some scholars also employ just quantitative restrictions as a proxy for NTBs (for this, see Deardorff and Stern 1985; Laird and Yeats 1990). While useful, I believe that quantitative restrictions are one component of NTBs and hence too limited for the analysis. 5. More specifically, they are taken from Haveman’s extracts from the TRAINS database, versions 2–­8. 6. In addition, the observations of core NTBs are unevenly spaced in that in some cases, observations are separated by only one year, while in others, they are separated by as many as six to seven years. 7. For this, see Mansfield and Busch (1995), Gawande and Hansen (1999), and Nicita and Olarreaga (2001). 8. The book’s online appendix is available at http://press.uchicago.edu/books/mukher jee/index.html. 9. This information is drawn from the Polity IV data set that contains coded annual information on regime authority characteristics and transitions for all independent states in the global state system from 1800 to 2006. 10. For this see, Boix (2003), Boix and Stokes (2003), and Przeworski et al. (2000). 11. In most cases of democratic transition, the adoption of a new constitution and elections occur within one or two years. 12. To give a more concrete example of how I operationalized new democracy, consider the case of Poland. Given that Poland made a transition to democracy in 1990, the new democracy variable is coded as 1 for Poland in 1990–­1994; however, it is coded as 0 for Poland for the years preceding 1990 and the years following 1994. 13. The results from the models in which the new democracy dummy is coded as 1 for all postdemocratic transition years are reported in table C in the book’s online appendix at http://press.uchicago.edu/books/mukherjee/index.html. Specifically, the interactive effect of new democracy × labor mobility on import duties, statutory tariffs, and core NTBs remain negative and significant when the new democracy dummy is coded as 1 for all postdemocratic transition years. 14. This implies that the measure of labor mobility provides the maximum coverage possible for operationalizing the interindustry occupational mobility of workers in each country. 15. The term on the left of labor mobility refers to the number of  job changes between t and t−δ, while the term on its right refers to the number of  job losses or gains not offset by a gain or loss in other sectors. 16. See, for example, Hill and Mendez (1983) and Langhammer and Foders (2006).

264  /  Notes to Chapter Three 17. This measure of economic crisis is not new. Indeed, the economic crisis measure is drawn from Milner and Kubota (2005) and is updated in my data till 2005. 18. This bop crisis is also drawn from Milner and Kubota (2005) and is updated in my data till 2005. 19. See Dornbusch (1992), Lustzig (2004), Tornell (1998), and Milner and Kubota (2005). 20. Some scholars do not agree with the claim that the GATT/WTO successfully promoted trade liberalization in developing countries. For this, see Rose (2004). 21. Although I proxy for asset specificity by using a measure of geographic concentration of industries, scholars have studied the effect of geographic concentration on trade barriers for particularly advanced industrial democracies; for this, see McGillivray (2004) and Busch and Reinhardt (1999). 22. Data to construct the AGC index are drawn from Spiezia (2002), GTAP (2006), and UNIDO (2008). 23. This measure is developed in two steps. I first divide for each country-­year the total capital stock by the size of the total labor force in the economy to obtain the capital-­ labor ratio. The data for the capital stock and the labor force are drawn from Heston, Summers, and Aten (2009) and Przeworski et al. (2000). I then log the capital-­labor ratio to derive the log capita-­labor ratio measure. 24. I thank an anonymous reviewer for this suggestion. 25. For this claim and a more general comparison of the economies of  Vietnam and the Philippines, see Boyce (1993) and Dean et al. (1994). 26. This figure is available at http://press.uchicago.edu/books/mukherjee/index.html. 27. Models 1–­3 are estimated with PCSEs and country-­fixed effects. 28. Models 4–­6 are estimated with bootstrapped standard errors and country-­fixed effects. 29. The autocracy dummy variable is coded by using the Cheibub, Gandhi, and Vreeland (2010) criteria for defining authoritarian states. According to these scholars, democracy is defined as a state where executives and legislatures are chosen by competitive elections. Further, the winners of elections must not be known in advance, the elected candidates must assume office following the election, and elections must take place regularly. States not meeting these criteria are coded as 1 for the autocracy dummy; it is coded as 0 otherwise. 30. See table D at http://press.uchicago.edu/books/mukherjee/index.html. 31. Data for exchange rate are drawn from IMF (2011) and World Bank (2011). The exchange rate variable is operationalized as the amount of local currency units per US dollar for each country-­year. 32. Estimated with a lagged dependent variable, PCSEs, and country-­fixed effects. 33. New democracy × ILM-­elasticity is negative and significant in the specification with the additional controls when the dependent variable is (i) import duties, (ii), statutory tariffs, or (iii) core NTBs. 34. I add government partisanship coded on a 0–­2 scale, where 0 denotes right governments and 2 left governments, since Milner and Judkins (2004) claim that left-­ leaning governments are more protectionist. Data for government partisanship are from the World Bank (2008). Data to operationalize education are from Easterly and Levine (2003) and World Bank (2005). 35. Blundell and Bond (1998) show that estimating the two equations (levels and differences) in a single system reduces the potential bias and imprecision associated with just the first-­difference GMM estimator.

Notes to Chapter Three  /  265 36. The Hansen J-­statistics in the system-­GMM models are consistent with the null hypothesis that the model’s overidentifying restrictions are valid. Diagnostic statistics from the system-­GMM models also reveal that the disturbances show no sign of serial correlation. 37. System-­GMM models estimated on the sample also confirm the main findings about the interactive effect of new democracy and labor mobility on statutory tariffs. 38. For details on this particular selection bias problem, see, for example, Przeworski et al. (2000) and Boix (2003). 39. The dichotomous dependent variable in the selection equation is coded as 1 for a democratic regime and is 0 otherwise. Based on the literature that identifies various factors that influence the likelihood that democracies may occur (see Przeworski et al. 2000 and Boix 2003), I incorporate the following variables in the selection equation of the statistical model where democratic regime is the dependent variable: real GDP per capita; the index of religious fractionalization; ethno-­linguistic fractionalization; the percentage of Catholics, Protestants, and Muslims in the population for each country-­year; a dummy for former British colonies (colony); and the lag of the democratic regime dummy. 40. Results from the selection equation of the Heckman model are available on request. 41. Baltagi et al.’s (2007) Lagrange multiplier test for spatial autocorrelation fails to reject the null of no spatial autocorrelation at the 1% level for all the three measures of trade protection in the data. 42. Because spatial dependence in the data may be a nuisance parameter that needs to be accounted for when testing hypothesis 2.1, I estimate a spatial AR error model. 43. I also found, but do not report to save space, that the coefficient of democracy × ILM-­ elasticity is negative and statistically significant in the system-­GMM model, spatial-­AR error model, and the outcome equation of the Heckman selection model for each of the three dependent variables of interest: import duties, statutory tariff, and core NTBs. 44. The largest and mean variance inflation factor value in the models is less than 10 and greater than 1, respectively; thus multicollinearity is not a problem. To assess for serial correlation, I first plotted the correlograms and partial correlograms of the residuals from each estimated model; I also checked the p-­values of the autocorrelation function and partial autocorrelation function for first lag as well as several additional lags of the residuals from of each estimated model. Results from this exercise reveal that the residuals do not suffer from serial correlation. Further, the Breusch-­Godfrey LM test failed to reject the null of no serial correlation in the outcome and selection equations, respectively. The reset test shows that there is no omitted variable bias problem; the Jarque-­Bera test shows that the residuals are distributed normally. 45. If one finds that only newly democratized states are statistically associated with higher labor mobility, then it is more plausible that the transition to democracy pro­ motes labor mobility. 46. The mean of ILM-­elasticity is respectively 6.21 for new democracies and 6.77 for autocracies. 47. During the 1972–­2008 period, labor mobility increased by 11.4% for new democracies and 12% for autocracies. 48. Studies of the determinants of labor mobility that incorporate these controls include Boerri and Terrell (2002), Jolivet et al. (2006), and Devine and Kiefer (1991). 49. These eighteen industries include iron and basic steel, office and computing machinery, plastic products, electrical industrial machinery, electrical appliances and houseware, beverage industries, tobacco manufacturing, textiles, transportation and

266  /  Notes to Chapter Three storage and communication, shipbuilding and repairing, agriculture, fishing, finan­ cial intermediation, retail trade, restaurants, construction, mining manufacturing, and real estate. 50. WEF EOS question 4.14 taken from World Economic Forum (2004b). 51. The advanced OECD democracies that are excluded from the EOS sample are as follows: Australia, Austria, Belgium, Canada, Finland, France, Germany, Greece, Ireland, Iceland, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and the United States. 52. As indicated in World Economic Forum (2004b)—­from which I gathered the descriptive data to analyze hypothesis 2.3—­the survey of industry executive was done from 2001 to 2006 (with the exception of one year, 2005). 53. For our bootstraps, m = 1,000. All control variables were held to their means or modes. 54. I also computed and illustrated the marginal effect of new democratic regime on the predicted probability of each of the remaining scores (i.e., 1, 2, 3, 4, 5, and 6) in the ordinal contributions influence measure across the entire range of labor mobility. These are not reported here to save space. But it is worth noting that these results are consistent with the theoretical expectation that is posited in causal claim 1. 55. In World Bank (2008), electoral fraud is also coded as 1 when opposition parties boycott the election because of intimidation by the government. Researchers at the World Bank have operationalized this measure based (partly) on reports of electoral malfeasance or fraud provided by the US State department, Human Rights Watch, OSCE (Europe), the United Nations, and the Organization of American States. 56. A recent empirical study by Kelley (2008) evaluates how norms and organizational interests of international election monitoring organizations influence the assessment of elections by these institutions, while Hyde and Beaulieu (2009) assess empirically why monitoring by international organizations induces opposition parties to boycott elections in countries that have recently experienced a transition to democracy. This is different from my empirical focus that examines how the adoption of trade reforms in new democracies and other domestic determinants of electoral malpractice (that I control for) affect the likelihood of electoral fraud. 57. Note that the correlation between the trade liberalization dummy and new democracy trade is weak (0.031) and statistically insignificant. 58. The possibility that more financial openness fosters higher levels of democracy and democratic consolidation has been studied by Dailami (2004), Eichengreen and Leblang (2008), and Milner and Mukherjee (2009). 59. For the bootstraps, m = 1,000. All control variables were held to their means or modes. 60. Again, for the bootstraps, m = 1,000. All control variables were held to their means or modes. 61. For this, see Lehoucq (2003). 62. Results from additional robustness tests conducted for the electoral fraud measure are reported in table E in the book’s online appendix at http://press.uchicago.edu /books/mukherjee/index.html. 63. See, for example, Mansfield et al (2000), Milner and Kubota (2005), Kono (2006), and Tavares (2008). 64. For this, see Kaufmann (2003) and Shah and Schacter (2004).

Notes to Chapter Four  /  267 Chapter FOUR

1. Examples of established developing (country) democracies include Costa Rica and India, for instance. These two countries are observed as stable and established democracies for much of the post–­World War II decade. 2. Developing countries that have recently experienced a transition to democracy and have successfully evolved into consolidated democracies include, for example, Argen­ tina, Brazil, Poland, and South Africa. 3. Scholars often use the term electoral (instead of political) particularism (e.g., Carey and Shugart 1995, Shugart and Haggard 1997, and Eaton 2002). I employ the word political (rather than electoral) particularism, as I examine more broadly the political ramifications of particularism—­affecting the electoral incentives of politicians in developing democracies—­for trade policy. 4. In a recent paper, Hankla (2006) empirically examines the impact of electoral particularism on trade protection in a mixed sample of developed and developing countries. Unlike Hankla’s (2006) work, my study focuses on how particularism combines with the relative factor mobility of  labor to affect trade policy in just established and consolidated developing country democracies. I also theoretically address the impact of particularism on campaign contributions provided by capital owners in the context of trade policy in developing democracies—­an issue-­area that is not examined systematically in Hankla (2006). 5. Nielson (2003) studies the effect of the number of political parties on trade policy, while Hankla (2006) examines how party centralization and the stability of “electoral linkages” matter for trade policy. McGillivray (1997, 2004) analyzes the interactive effect of the majoritarian system and party discipline on tariff barriers in some advanced industrial democracies. 6. For example, McGillivray (1997, 2004) focuses on just the United States and Canada. 7. Nielson’s (2003) work on party systems and trade barriers is confined to eighteen middle-­income presidential democracies. Hankla (2006) analyzes the impact of party centralization and stability of electoral linkages on tariffs in developed and developing democracies. However, as mentioned above, his work does not really provide a unified theoretical model of how trade policy preferences of societal groups (the demand side) may interact with particularism (the supply side) to influence tariffs and NTBs in developing countries. 8. The literature on particularism is fairly extensive. For some of the earlier literature on particularism, see Carey and Shugart (1995), Shugart and Haggard (1997), Mainwaring and Shugart (1997), and Carey (1998). 9. The level of particularism is theoretically conceptualized and empirically operationalized as an ordinal index that ranges from a low to a high value (see Carey and Shugart 1995, Wallack et al. 2005, and Johnson and Wallack 2007). 10. Carey (1998) also defines particularistic democracies as systems where politicians have “direct electoral incentives to build personal support among voters” (Carey 1998: 196). 11. For this terminology, see Carey (1998), Mainwaring and Shugart (1997), Hix (2004), Hicken (2006), and Simmons and Hicken (2008). 12. This includes rules about candidate nomination and the order of election (open versus closed lists). Researchers debate about the type of electoral rules that influence the degree to which democracies are candidate centered. Most scholars, however, agree that the open-­list PR rule and single-­transferable vote systems engender

268  /  Notes to Chapter Four candidate-­centered systems (Hix 2004; Hicken 2006). The closed-­list PR rule in contrast tends to generate party-­centered democracies (Hix 2004; Hicken 2006). Scholars also suggest that whether or not votes from various party candidates are pooled to assign seats to the party as a whole strongly influences the level of particularism. For this, see Mainwaring and Scully (1995), Carey and Shugart (1995), and Rius and Van De Walle (2003). 13. The fact that political parties in democracies tend to use some share of the contributions that they receive to buy and sustain the loyalty of party members has been particularly studied both formally and empirically in the American context. For this, see Ansolabehere (2000), Pearson (2008), and Leyden and Borrelli (1990). A more recent study by Claessens et al. (2008) reveals that contributions are also used by party leaders to “buy off” party members even in developing democracies like Brazil. Following these studies, I assume that party leaders in the model may choose to distribute some share of the total payments provided by the owners r (tJ , φ, dJ) to rank-­ and-­file party members or divert it for personal use. 14. Hence limλ → 1 implies a particularistic democracy, while limλ → 0 implies a party-­centered democracy. 15. Thus lim ∆ → 1 implies a “pure” particularistic democracy in which the bulk of r (tJ , φ, dJ) is diverted by the leaders for personal gain, while implies a party-­centered democracy where is not diverted for personal gain but rather distributed to party members. 16. This dynamic also applies to the leaders from the opposition party or parties. 17. As emphasized by Levitsky (2003: 10), “Winning public office is a primary goal” for party leaders in developing country democracies. 18. I provide a formal proof of this claim in the proof of part (a) of lemma 1 in the appendix. Note that I do not suggest that high levels of  particularism always drive party leaders to appeal to the median voter. Rather, according to the model, it is only in the context of electoral competition over trade policy that one expects to observe the party leaders to cater to the median voter’s tariff policy preferences. Whether or not this causal claim from the model is empirically valid is evaluated in the later chapters of this book. 19. I derive and formally characterize the median voter’s optimal tariff policy preference in the proof of part (a) of lemma 1 in the appendix. 20. Because the median voter’s utility function is strictly concave, his optimal tariff policy preference is the point at which his utility function reaches a global maximum. 21. This claim is demonstrated more formally in the proof of part (b) of proposition 1. 22. The model’s logic explained in this paragraph also applies to the opposition party leaders. 23. See Hicken (2006), who describes why Philippines is a particularistic democracy. 24. Weyland (2000: 6). 25. This implies in the context of  my theoretical framework that personalist leaders will adopt these three strategies when labor mobility is high, as workers favor more trade openness under this condition. 26. Weyland (2000: 13). 27. Extant analyses of trade policy making in particularistic democracies in Asia, such as the Philippines and Thailand (from 1994–­97), also indicate that incumbents used various executive instruments to liberalize trade policies in these countries where labor mobility is sufficiently high. They did so to “circumvent potential intra-­party opposition to globalization” and to increase the likelihood of quickly implementing trade reforms.

Notes to Chapter Five  /  269 28. An interesting study by Estache et al. (2000) suggests that leaders in Latin America’s personalistic democracies used the media to attack politicians who opposed economic reforms to help voters identify the political candidates that were resisting reforms that they (the voters) could benefit from. In other words, the rhetorical attacks provided “information” to voters about politicians who were more susceptible to protectionist interests. 29. The strategy described above has been used by leaders in particularistic countries as diverse as Brazil, Peru, and the Philippines, and it has helped leaders in these states implement trade reforms (Durand 1998; Weyland 2000, 2002; Hutchcroft and Rocamora 2003). Weyland (2000), for instance, suggests that attacks launched by party leaders in particularistic systems against party members who oppose reforms has delegitimized these opposing actors as “selfish defenders of special privileges . . . and [has] thus facilitated the enactment and implementation of reforms” (2000: 12). 30. This claim is shown formally in the proof of part (c) of proposition 1 in the appendix. 31. For this, see also Kingstone (1999: 136–­37). 32. The terms party-­centered systems and systems with centralized parties (i.e., centralized party systems) are used interchangeably, since political parties are highly centralized in party-­centered systems. 33. Control over nomination of candidates to the party’s electoral list is centralized among party bosses in party-­centered systems (Carey 2003; Carey and Reynolds 2007). This is especially (but not only) pronounced in closed-­list proportional representation (PR) systems—­which produces party-­centered democracies where party leaders decide the rank of candidates on the party’s electoral list (Carey 2003; Hix 2004). 34. See proof of claim 1 in the appendix. 35. Carey and Reynolds (2007: 271) also state that “the strong party normative ideal prevalent in much academic work rooted in the experience of developed democracies is frequently inapplicable to how parties” operate in developing democracies. Thus they argue that strong, disciplined parties in developing states do not always adopt policies that maximize social welfare as suggested in some extant studies (e.g., Hankla 2006). 36. I assume in the extended model presented here that the parties also campaign on the issue of NTBs during elections, irrespective of  the extant degree of  political particularism in the polity. 37. There exists an inverse relationship between NTBs and the asset specificity of manufacturing industries in the model—­that is, the owners of  industries that are characterized by a high (low) degree of asset specificity are more (less) likely to lobby for higher NTBs to protect their industries from import competition. 38. Numerous theoretical and empirical studies by economists show that lowering NTBs reduces wage and price rigidities and thus increases the welfare of consumers. For this, see Herberg (1990), Beghin (2008), Bradford (2000), and Anderson and Wincoop (2001). 39. See proof of claim 1 in the appendix. Chapter FIve

1.

2.

Numerous scholars in IPE have used the aforementioned criterion from the Polity index to operationalize countries as democratic regimes. See, for example, Mansfield et al. (2000) and Hankla (2006). Core NTBs include price-­control measures (excluding antidumping), quantitative restrictions (e.g., voluntary export restraints), finance control measures, monopoly

270  /  Notes to Chapter Five measures, and technical regulation. The core NTBs measure that I employ is missing for several democratic country-­years in the sample. Yet I persist with using the core NTBs, as it provides us with the maximum country-­year coverage on NTBs for developing democracies. 3. Data for core NTBs are taken from the UNCTAD’s Trade Analysis and Information System (TRAINS), the World Bank’s (2006a) Trade and Production Database, the UNCTAD’s (1994) Directory of Import Regimes, the WTO’s Trade Policy Review Series, the World Integrated Trade Solution (WITS) package developed by the WTO and UNCTAD, and the GTAP (2006) data package. 4. The rationale for estimating the models that include the ILM-­elasticity measure with bootstrapped standard errors was discussed in chapter 3. 5. Following Hix (2004) and Carey and Shugart (1995), closed-­list systems include the following type of electoral systems: closed-­list proportional representation (PR); ordered-­list PR; single-­member, simple-­plurality (i.e., first-­past-­the-­post); and double­ballot (i.e., single-­member, alternative-­vote) systems. 6. Wallack et al. (2003) and Johnson and Wallack (2006) disagree with Carey and Shugart regarding how to code single-­member districts (SMDs). Carey and Shugart propose that SMDs are essentially closed-­list systems in particularly small districts, and they suggest coding ballot as 0 in the presence of SMD systems. They also propose coding pool as equal to 0 in the case of SMD systems. They argue that in the presence of a “list” of one candidate, votes are pooled across the entire list. Wallack and her coauthors disagree with both suggestions. Instead, they code SMD systems 0 for ballot only where the majority of districts are multimember, closed-­list, and proportional. Otherwise, they assign SMD systems a score of 1 in the ballot variable. With respect to pool, they code “single member districts as two on the Pool scale because they do not receive additional electoral support if  other candidates from their party are successful in other districts.” 7. Apart from calculating the average of ballot, pool, and vote to develop their index of particularism, Wallack et al. (2003) and Johnson and Wallack (2006) have also developed an alternative index of particularism by coding the ballot, pool, and vote variables as the “weighted averages . . . for each group of legislators that is elected under a different set of rules” (Wallack et al. 2003: 13). I obtain similar results if I use the particularism index described in the text or the alternative index of particularism described in this footnote. To save space, I only report here the results from using the particularism index described above in the text. 8. Data to compute the labor mobility measure are drawn from the International Labor Organization (2008) LABORSTA databases, the UNIDO (2007) database, the UNIDO (2009) Industrial Development Report, GTAP (2006), and the UN (various years) Na­ tional Account Statistics Database. 9. See table B at http://press.uchicago.edu/books/mukherjee/index.html. 10. For this, see Weyland (2002) and Cárdenas and Perry (2011). 11. See figure B at http://press.uchicago.edu/books/mukherjee/index.html. 12. See figure C at http://press.uchicago.edu/books/mukherjee/index.html. 13. Models 1–­3 are estimated with PCSEs and country-­fixed effects. 14. Models 4–­6 are estimated with bootstrapped standard errors and country-­fixed effects. 15. This does not include capital account openness that is excluded from the models in which core NTBs is the dependent variable.

Notes to Chapter Five  /  271 16. The largest and mean variance inflation factor value in the models is less than 10 and greater than 1, respectively; thus multicollinearity is not a problem. To assess for serial correlation, I first plotted the correlograms and partial correlograms of the residuals from each estimated model; I also checked the p-­values of the autocorrelation function and partial autocorrelation function for first lag as well as several additional lags of the residuals from of each estimated model. Results from this exercise reveal that the residuals do not suffer from serial correlation. Furthermore, the Breusch-­Godfrey LM test failed to reject the null of no serial correlation in the outcome and selection equations, respectively. The reset test shows that there is no omitted variable bias problem; the Jarque-­Bera test shows that the residuals are distributed normally. 17. I also found that the interactive effect of particularism × ILM-­elasticity on import duties, statutory tariffs, and core NTBs is negative and significant in the expanded empirical models with the additional controls. These results are available on request. 18. The Hansen J-­statistic in the system-­GMM models are consistent with the null hypothesis that the model’s overidentifying restrictions are valid. Diagnostic statistics from the system-­GMM models also reveal that the disturbances show no sign of serial correlation. 19. System-­GMM models estimated in the sample also confirm the findings about the interactive effects of particularism and ILM-­elasticis on each of the three dependent variables of interest. 20. Baltagi et al.’s (2007) Lagrange multiplier test for spatial autocorrelation fails to reject the null of no spatial autocorrelation at the 1% level for all the three measures of trade protection in the data. 21. The spatial-­AR model that I employ and the procedure used to estimate this model is described in the appendix of this book. 22. Results from the spatial-­AR model in which the dependent variable is statutory tariffs and core NTBs are available on request. The estimate of particularism × ILM-­elasticity remains negative and statistically significant in the spatial-­AR error model for each of the three measures of trade barriers used in this chapter. 23. See, for example, Carey and Shugart (1995), Shugart and Wattenberg (2003), Mainwaring and Shugart (1997), Hix (2004), and Hicken (2007). 24. Desposato (2006) and Shin (2010) claim that in less developed countries or districts where the majority of voters are poor and less educated, personalistic parties—­whose members are more likely to vote against party lines and switch parties in order to provide voters with those particularistic benefits—­are more likely to flourish. However, we are more likely to observe the opposite in countries with higher levels of per capita income. 25. For this, see Carey and Shugart (1995), Hicken (2002, 2007), and Hix (2004). 26. The reference category in this case includes democracies with an open-­list PR system. 27. See Hix (2004), Hicken (2007), and Shin (2010). 28. Shin (2010: 46). 29. For the bootstraps, m = 1,000. All control variables were held to their means or modes. 30. Similar to chapter 4, I focus in this chapter on assessing the predicted probability of this maximum ordinal level of 7, as it directly allows me to test the argument posited in claim 2.

272  /  Notes to Chapter Six C h a p t e r s ix

1. Apart from evaluating the validity of the causal arguments presented in the theoretical chapters, I also briefly examine whether alternative accounts (i.e., the occurrence of a debt crisis in Brazil in the early 1980s and the country’s subsequent participation in IMF programs) can account for trade reforms in Brazil, especially during the 1980s. 2. Scholars of Latin American politics suggest that Brazil is a candidate-­centered democracy; for this, see Ames (2001), Mainwaring (1999), Carey and Shugart (1995), Geddes (1994), Samuels (2000), and Mainwaring and Perez-­Liñan (1997). I follow the literature and subscribe to the view that Brazil is a candidate-­centered democracy. This is because Brazil uses the open-­list proportional representation (PR) electoral rule that generates a candidate-­centered electoral system. 3. During the emergency in India in 1975–­77, Prime Minister Indira Gandhi suspended elections and civil liberties. But this did not last long, and the emergency was lifted by March 1977. For an interesting and thoughtful discussion of the emergency in India, see Rudolph and Rudolph (1977). 4. South Africa uses a closed-­list PR electoral system and is therefore a party-­centered democracy (see, for example, Sadie 1998; Plasser and Plasser 2002). Although the party system in India has progressively fragmented in the last two to three decades, it is a party-­centered democracy where party leaders have strong control over their rank-­and-­file party members (see Chibber 1999; Sridharan 2009). 5. For example, India’s nominal per capita income is substantially lower than the nominal per capita income in Brazil and South Africa. The average growth rate in India in the last two decades has, however, been 2 to 2.5 times faster than Brazil and South Africa. 6. As estimated by the IMF in 2012, the size of  Brazil’s economy in nominal GDP terms is approximately $2.2 trillion, while the size of India’s economy is roughly $1.9 trillion. 7. The nominal per capita income in 2008 in Brazil was US$8,400, while it was US$5,685 in South Africa (World Bank 2008). Hence both Brazil and South Africa are middle-­income countries. 8. While Brazil, India, and South Africa’s exposure to global trade flows has steadily increased, we will see in this and the next chapter that the level of trade barriers over time across the three countries has been different. 9. As mentioned in an earlier footnote, the size of  Brazil’s nominal GDP in 2008 was $1.9 trillion, which makes it the tenth largest economy in the world. 10. These figures have been taken from the World Bank (2005). 11. As reported by Brazil’s Ministry of Development, Industry and Commerce (2003), Brazil’s exports from 1988 to 2002 grew by just 11% in real terms, which is relatively slow compared to the percentage growth in exports of  20% and above in other large emerging markets such as China, India, and Russia. That said, since 2003, the volume of Brazil’s exports has more than doubled. 12. A large body of  literature on electoral systems in political science claims and finds that the open-­list PR electoral rule promotes a candidate-­centered system. For this, see Carey and Shugart (1995), Cox and Rosenbluth (1995), and Samuels (1999, 2000, 2001). 13. The figures for Brazil’s debt, current account deficit, and growth rate have been computed based on the relevant data for Brazil, which are drawn from the IMF’s (2007) Government Financial Statistics.

Notes to Chapter Six  /  273 14. The data to compute the moving average of the unweighted tariff rate in Brazil are drawn from Kume et al. (2000), World Bank (2003, 2006b, 2007a), WTO (various years), and Instituto Brasileiro de Geografia e Estatísticas (IBGE; various years). 15. I operationalize NTBs in Brazil as a single category core NTB coverage ratio, which is measured by calculating the percentage of imports subjected to core NTBs; this mea­ sure is labeled as denoted as core NTBs. Note that core NTBs include price-­control measures (excluding antidumping), quantitative restrictions, finance control mea­ sures, monopoly measures, and technical regulation. Data for core NTBs for Brazil from 1987 are drawn from the following sources: Instituto Brasileiro de Geografia e Estatísticas (IBGE; various years); UNCTAD (various years), TRAINS database; Haveman and Thursby (2000); UN Comtrade (various years), Classification SITC Rev 3; World Bank (various years), WITS database; and Kee et al. (2006). 16. Prior to 1977, data to compute the occupational mobility of labor are available for only a limited set of industries at the three-­digit ISIC level. However, from 1977 onward, the data that exist to calculate the interindustry level of occupational mobility of labor are available for a much larger set of industries (i.e., thirty-­two). Hence I calculate the measure of interindustry labor mobility for Brazil starting from 1977. Data to calculate the annual level of interindustry occupational mobility of labor are taken from the UNIDO Industrial Statistics Database (various years for Brazil); Muendler et al. (2001); Instituto Brasileiro de Geografia e Estatísticas (IBGE; various years); Instituto de Pesquisa e Planejamento Econômico (IPEADATA); and Secretaria de Comércio Exterior/Ministério do Desenvolvimento Indústria e Comércio (SECEX/ MDIC). 17. I used the procedure described in chapter 4 and estimation of equations 4.2 and 4.3 to operationalize the elasticity of interindustry labor mobility across thirty-­two industries in Brazil from 1977 to 2004. The sources that are used to operationalize the elasticity of labor mobility in Brazil are listed in the previous footnote. I do not plot the elasticity of labor mobility in Brazil to save space in figure 6.3 and because the temporal variation of the elasticity measure is similar to the level of interindustry labor mobility illustrated in the figure. 18. As shown in figure 6.3, the degree of interindustry labor mobility also increased again from 1993–­94 to approximately 1999. 19. See “Brazil’s New Trade Policy” (1987: B1; translated from Portuguese to English). 20. See Relação Anual de Informações Sociais (1987: xxiv). 21. See Manifesto dos Mineiros (1986). 22. Taken from Acuna (1988). 23. Taken from Nicolau (1998: 106). 24. MERCOSUR is a subregional bloc comprising Argentina, Brazil, Paraguay, Uruguay, and Venezuela. Its associate countries are Chile, Bolivia, Peru, Colombia, and Ecuador. 25. The data to compute the industry-­year tariff from 1977 to 2004 are drawn from World Bank (2005), World Integrated Trade Solution (WITS); UN Comtrade (various years); Kume et al. (2000); Kee et al. (2006); and UNCTAD (various years), TRAINS database. 26. The only NTB measure that is available for Brazil is an annual average of core NTBs ratio for a set of industries from 1987. In effect, this generates just one observation per year for Brazil and a total of eighteen observations from 1987 to 2004. It is not feasible to estimate a full-­fledged model on such a small sample.

274  /  Notes to Chapter Seven 27. Data to compute this measure are from Moreira and Correa (1997), Ministério do Desenvolvimento (various years), Datamark (2002; Brazil), and Instituto Brasileiro de Geografia e Estatísticas (IBGE; 1977–­2006). 28. See, for example, Ferreira and Facchini (2004), Hay (2001), and Helfand (2000). 29. Data for Brazil’s capital-­output ratio, investment-­output ratio, and log GDP are taken from UNIDO Industrial Statistics Database (various years for Brazil); World Bank (2004, 2006a, 2006b), World Development Indicators; Instituto Brasileiro de Geografia e Estatísticas (IBGE; various years); and Instituto de Pesquisa e Planejamento Econômico (IPEADATA). 30. This quote is taken from “A Party-­Finance Scandal in Brazil,” The Economist, Feb­ ruary 19, 2004. 31. Interview of a director of a Brazilian metalworking firm cited in Meneguello (1989: 47). 32. Payne (1992: 5). 33. Taken from Lima and Lima (1987: 122). 34. See ABINEE (1988: 45). 35. See ABINEE (1988: 49). 36. The CNI research is conducted annually in the fall. The sample size in the survey includes anywhere between 800 and 920 industrialists who are randomly drawn from the Gazeta Mercantil register of businesses, which contains 5,800 addresses. In addition to being a large proportion of the total population, the sectoral breakdown of the respondents largely corresponds to the total population both in terms of the number of firms and in terms of the economic weight within the industrial economy, particularly the manufacturing sector. 37. The share of output to GDP produced by each of these fourteen industries is larger than the remaining industries in the economy. Thus these industries are the “largest” in terms of output in the economy. 38. In 1994, 1 US dollar = 0.845 Brazilian real. In 1998, 1 US dollar = 1.14 Brazilian real, while in 2002, 1 US dollar = 2.38 Brazilian real. 39. CNN (1998). 40. See, for example, Rodrigues-­Filho et al. (2006) and Maneschy (2002). Chapter seven

1. IMF (2009). 2. The data to compute the annual series of the moving average of unweighted tariffs for India are drawn from the WTO, Trade Policy Review: India (various years); World Bank (2003, 2007a, 2007b), World Development Indicators; IMF (various years), International Financial Statistics; UNCTAD (various years), TRAINS database; and Joshi and Little (1997). 3. I choose 1987 as the first year in figure 7.1, since data for NTBs for India are only avail­ able from 1987 onward. We operationalize NTBs in India as a single category core NTB coverage ratio, which is measured by calculating the percentage of imports subjected to core NTBs; this measure is denoted as core NTBs. Note that core NTBs include price-­ control measures (excluding antidumping), quantitative restrictions, finance control measures, monopoly measures, and technical regulation. Data to compute the core NTBs ratio for India from 1989 are drawn from the following sources: Government of India (various years), Economic Survey; Mehta (1999); UNCTAD (various years), TRAINS database; UN Comtrade (various years), Classification SITC Rev 3; National Council of Applied Economic Research (1999); and Kee et al. (2006).

Notes to Chapter Seven  /  275 4.

I focus on these industries, as data to operationalize interindustry labor mobility at the three-­digit ISIC level are only available for these industries. I used the formula described in equation 3.1 in chapter 3 to calculate the degree of occupational mobility of workers between twenty-­nine different industries on an annual basis from 1978 to 2003 in India. Likewise, we used the procedure described in chapter 4 and estimation of equations 4.2 and 4.3 to operationalize the annual series of the elasticity of interindustry labor mobility in India from 1978 to 2003. Prior to 1978, data to compute the occupational mobility of labor are available for only a limited set of industries. However, from 1978 onward, the data that exist to calculate the interindustry level of labor mobility are available for a much larger set of (twenty-­ nine) industries. Hence we calculate the measure of interindustry labor mobility for India starting from 1978. Data to calculate the annual level of interindustry mobility and the elasticity of labor mobility are taken from UNIDO (various years for India), Industrial Statistics Database; Central Statistical Organization (various years), Annual Survey of Industries; Ministry of Finance (various years), Economic Survey; and Ghose (1995). 5. For this claim, see Hamermesh (1987, 1993) and Winters (2002). 6. Planning Commission (1983: 27). 7. For this, see “Interview with Tarun Das” (1992). 8. Joshi and Little (1997), Sachs et al. (1991), and Jalan (1991). 9. Interview with official (name withheld because of request for anonymity) conducted in New Delhi on December 18, 2009. 10. The figures for the budget deficit, inflation rate, foreign exchange reserves, and the current account deficit for India, as cited above, are taken from the Ministry of Fi­ nance, Government of India, Economic Survey (various years). 11. Frontline, volume 16, issue 16, July 31 to August 13, 1999. At the time, Gaurav Swarup was deputy managing director of Paharpur Cooling Towers, Ltd., a Rs. 165 crore turn­ over company involved in the manufacture of process cooling equipment, cooling towers, and air-­cooled heat exchangers. 12. For this, see Joshi and Little (1997). 13. The data sources employed to compute this measure were listed in an earlier footnote. 14. The only NTB measure that is available for India is an annual average of the core NTBs for a set of industries from 1987. This generates just one observation for NTBs per year for India and a total of just seventeen observations till 2003. Thus it is not feasible to estimate a full-­fledged specification on such a small sample. 15. Data to operationalize the import penetration ratio variable for India are drawn from IMF (2008), International Financial Statistics; DGCIS (various years), Foreign Trade Sta­ tistics; Ministry of Finance (various years), Economic Survey; Kee et al. (2006); and Kapila (2001). 16. See Dutta and Ahmed (1997), Goldar and Kumari (2002), and Das (1998). 17. Data for India’s capital-­output ratio, investment-­output ratio, and log GDP are taken from UNIDO (various years for India), Industrial Statistics Database; World Bank (2004, 2006b), World Development Indicators; Ministry of Finance (various years), Economic Survey; and IMF (2008), International Finance Statistics. 18. I also found, but do not report to save space, the impact of the annual series of the elasticity measure of interindustry labor mobility. (This computed measure is the dashed line in figure 7.2.) 19. The eighteen industries include iron and basic steel, office and computing machinery, plastic products, electrical industrial machinery, electrical appliances and houseware,

276  /  Notes to Chapter Seven beverage industries, tobacco manufacturing, textiles, transportation and storage and communication, shipbuilding and repairing, agriculture, fishing, financial intermediation, retail trade, restaurants, construction, mining manufacturing, and real estate. 20. The data to compute the annual series of the moving average of unweighted tariffs for South Africa are drawn from the WTO, Trade Policy Review: South Africa (2004); IMF (various years), International Financial Statistics; South Africa Reserve Bank Quarterly Bulletin (various years); Edwards and Lawrence (2007); World Bank (various years), WITS database; and World Bank (2003, 2007a, 2007b), WDI. 21. We choose 1988 as the first year to operationalize the moving average of core NTBs, as data for NTBs for South Africa are only available from 1988 onward. We operationalize NTBs in South Africa as a single category core NTB coverage ratio, which is measured by calculating the percentage of imports subjected to core NTBs. The core NTBs include price-­control measures (excluding antidumping), quantitative restrictions, finance control measures, monopoly measures, and technical regulation. Data to compute the core NTBs ratio for South Africa from are drawn from UNCTAD (various years), TRAINS database; UN Comtrade (various years), Classification SITC Rev 3; Kee et al. (2006); Edwards and Lawrence (2007); Board of Tariffs and Trade, South Africa (1988–­2003), annual reports. 22. I used the formula in equation 3.1 in chapter 3 to calculate the degree of interindustry occupational mobility of workers across twenty-­nine different industries listed in table 7.1 on an annual basis from 1978 to 2003 for South Africa. I estimated equations 3.2 and 3.3 in chapter 3 to derive the elasticity measure of labor mobility across the same twenty-­nine industries for South Africa during the 1978 to 2003 time period. Data to calculate the annual level of interindustry mobility and the elasticity of interindustry labor mobility are taken from UNIDO (2009) and Klasen and Woodward (1998). 23. This quote is taken from Macoreconomic Research Group (MERG) for ANC (1995: 211). 24. This is taken from Black Economic Empowerment Commission (BEECOM; 2001: 115). 25. “Empowerment and the IEC” (1995). 26. This is stated in Independent Electoral Commission (IEC; 1997: 55). 27. See, for example, Strydom (1995) and Holden and Casale (2000). 28. Taken from Hirsch (2005: 187). 29. See the editorial entitled “Business and International Trade,” Business Day, April 6, 1999. 30. This is taken from the ANC resolution in 1999, cited in BEECOM (2001: 115). 31. We used the formula described in equation 4.1 in chapter 4 to calculate the level of occupational mobility of workers between all twenty-­nine industries (see table 7.1) on an annual basis from 1979 to 2003 in South Africa. The results reported below do not change if we employ the elasticity measure of interindustry labor mobility for South Africa. The data to operationalize the labor mobility measure are listed in an earlier footnote. 32. Data to operationalize the annual series of import ratio for South Africa are from Edwards and Lawrence (2007); Fedderke et al. (2003); South African Reserve Bank Quarterly Bulletin (various years); Department of Trade and Industry (DTI; various years), South African Trade Statistics; IMF (2008), International Financial Statistics. 33. Data for South Africa’s capital-­output ratio, investment-­output ratio, and log GDP are from UNIDO (various years for South Africa), Industrial Statistics Database; World

Notes to Chapter Eight  /  277 Bank (2004, 2006b), World Development Index; Department of Trade and Industry (DTI), Reserve Bank: Economic and Financial Data for South Africa; and IMF (2008), International Financial Statistics. 34. The letter R in this case refers to the South Africa’s national currency, the South African rand. C h a p t e r Eig h t

1. For this, see Milner and Kubota (2005) and Rodrik and Wacziarg (2005). 2. Other early studies on the political economy of economic reform by Haggard (1986, 1990) and Helleiner (1993), for example, focused on how debt crises, balance-­of-­ payments problems, and low economic growth induced governments in developing states to adopt economic reforms. These studies, however, did not explore how democratization or democracy in general influenced the prospects for economic reform in developing countries. 3. This does not mean that all political scientists accept that democracy and trade openness go hand in hand. Li and Reuveny (2003), for example, suggest that democracy and trade openness are not mutually compatible, while Garrett (2000) argues that democracy may promote trade liberalization in some—­but not all—­developing countries. 4. The technical foundations of the models have been inspired by political economy models of trade policy developed by Mansfield, Milner, and Rosendorff (2000), Grossman and Helpman (1994, 2002), and Feenstra (2003). 5. See, for example, McGillivray (2004), Hankla (2006), and Ehrlich (2007). 6. See Hankla (2006) and Ehrlich (2007). 7. See, for example, Hyde (2011) and Kelley (2012). 8. To be sure, the seminal political economy models of trade by Grossman and Helpman (1994, 2002) and Goldberg and Maggi (1999) formalize the phenomenon of campaign contributions in the context of trade policy making in advanced industrial democracies. Other studies by Gawande et al. (2006, 2009) evaluate the Grossman-­ Helpman model, especially in the context of lobbying over trade policy in the United States. However, these scholars do not explain how variation in domestic political institutions (e.g., the degree of electoral particularism) across developing country democracies can affect the equilibrium amount of campaign contributions offered by capital. Research on trade politics in IPE also recognizes that industrialists lobby politicians to influence trade policy in democracies (see McGillivray 2004). But these studies have not as yet developed systematic hypotheses of how domestic institutions can affect the amount of contributions given to influence trade policy in developing states. 9. This measure has been developed by Wacziarg and Wallack (2004) and Hiscox and Rickard (2002). 10. See, for example, Alt et al. (1996), Gilligan (1997), Hiscox (2001, 2002), and Zahariadis (2002). 11. In chapter 7, I explained why India and South Africa are party-­centered democracies. 12. For example, detailed policy analyses on the collapse of the Doha and Cancun trade talks by Dunoff (2001), Kennedy (2008), and Cho (2009) do not examine whether and how domestic political factors or institutions affect the trade negotiation strategies of developing democracies at the WTO. 13. The data on antidumping petitions (safeguard investigations) start in 1980 because systematic and comprehensive annual data on the number of antidumping (safeguard) petitions initiated by developing countries at the WTO are available from

278  /  Notes to Appendix 1980 onward. The data on the number of antidumping petitions filed by WTO-­ member developing democracies are drawn from WTO (2007), Zlate (2002), and Bown (2009). Data on the number of safeguard investigations filed by developing democracies that are WTO members are drawn from annual reports (from 1996 to 2010) of the WTO’s Committee on Safeguards, Zlate (2002), and Bown (2009, 2010). 14. Economists typically believe that countries usually initiate antidumping petitions and safeguard investigations to protect their domestic markets from import competition (see Prusa 2005; Bown 2009, 2010; Hoekman and Kostecki 2009). 15. Data on economic growth, which is operationalized as the real GDP growth rate, are drawn from the World Bank (2007a, 2007b) and IMF (2007). I estimated the effect of the structural and elasticity measure of interindustry labor mobility (described in detail in chapters 3 and 5) on the real GDP growth in the sample of developing democratic country-­years. The results obtained from each of these two measures of interindustry labor mobility are identical. 16. I calculated the mean of government spending on tertiary and secondary education as a percentage of GDP across the country-­years in the sample of developing democracies listed in table 5.1 (chapter 5). The data to calculate this mean are taken from the IMF (2007) and World Bank (2007a, 2007b). 17. Economists generally subscribe to the view that higher economic growth promotes economic welfare. For this, see, for example, Barro (1997), Barro and Sala-­i-­Martin (2003), and Bourguignon (2004). With respect to trade liberalization, some economists suggest that higher levels of trade openness generated by trade reforms engender more economic growth, which enhances social welfare (see, for example, Harrison 1996 and Frankel and Romer 1999). Other economists, however, suggest that trade reforms and thus more trade openness does not lead to higher growth and welfare in developing countries (Rodriguez and Rodrik 2001, Irwin and Tervio 2002, Rodrik et al. 2002). 18. For instance, see Milner (1997) and Mansfield, Milner, and Rosendorff (2002). 19. See Hamermesh (1993), Moscarini (2001), Winters (2002), and Matschke (2004). 20. I thank an anonymous reviewer for making this suggestion. a p p e ndix

1. Observe that θ mv ∈ [0,1], since θ lmobile ∈ [0,1 ].

R eferences

ABINEE. 1988. Abertura Comercial e Reestruturação Industrial no Brasil: Deve O Estado Intervir? Rio de Janeiro: Iuperj. Abreu, M. P. 2004. “Trade Liberalization and the Political Economy of  Protection in Brazil since 1987.” INTAL-­ITD Working Paper SITI 08B, Inter-­American Developing Bank, Buenos Aires. ————. 2006. “The FTAA and the Political Economy of Protection in Brazil and the US.” INTAL-­ITD Working Paper SITI 12, Inter-­American Developing Bank, Buenos Aires. Acemoglu, D., and J. A. Robinson. 2006. Economic Origins of Dictatorship and Democracy. New York: Cambridge University Press. Acuna, Carlos. 1988. “Empresarios y polftica (Parle I): La relaci6n de las organizaciones empresarias conregfmenes polfticos en América Latina.” Boleffn lnformativo Techint 255: 22. Adam, H., F. Van Zyl Slabbert, and K. Moodley. 1997. Comrades in Business: Post-­liberation Politics in South Africa. Cape Town: Tafelberg. Adnett, Nick, Spiros Bougheas, and Yannis Georgellis. 2004. “On the Trade-­Off between Work-­Related Training and Labor Mobility: The Role of  Firing and Exit Costs.” Journal of Economics 82(1): 49–­70. African National Congress. 1994. The Reconstruction and Development Programme: A Policy Framework. Johannesburg: Umanyano. Alesina, A., E. Spolaore., and R. Wacziarg. 2000. “Economic Integration and Political Disintegration.” American Economic Review 90(5): 1276–­96. Almeida, Maria H. T. 2004. “From Cardoso to Lula: The Politics and Policies of  Reform in Democratic Brazil.” Working Paper, University of  São Paulo. Alston, L. J., and A. A. Gallo. 2009. “Electoral Fraud, the Rise of Peron and Demise of Checks and Balances in Argentina.” NBER Working Paper 15209, National Bureau of Economic Research, Harvard University, Cambridge, MA. Alt, J., J. Frieden, M. J. Gilligan, D. Rodrik, and R. Rogowski. 1996. “The Political Economy of International Trade Enduring Puzzles and an Agenda for Inquiry.” Comparative Political Studies 29(6): 689–­717. Alt, J., and M. Gilligan. 1994. “The Political Economy of  Trading States.” Journal of Political Philosophy 2(2): 165–­92.

280 / References Alt, James E., F. Carlsen, P. Heum, and K. Johansen. 1999. “Asset Specificity and the Political Behavior of Firms: Lobbying for Subsidies in Norway.” International Organization 53(1): 99–­116. Alvarez, F. 1999. “Social Mobility: The Barro–­Becker Children Meet the Laitner–­Loury Dynasties.” Review of Economic Dynamics 2(1): 65–­103. Ames, Barry. 1995a. “Electoral Strategy under Open-­List PR.” American Journal of Political Science 39(2): 406–­33. ————. 1995b. “Electoral Rules, Constituency Pressures, and Pork Barrel: Bases of  Voting in the Brazilian Congress.” Journal of Politics 57(2): 324–­43. ————. 2001. The Deadlock of Democracy in Brazil. Ann Arbor: University of Michigan Press. Amman, E., and W. Baer. 2002. “The Development of Brazil’s Technology Capabilities in the Post War Period.” Latin American Business Review 13(2): 1–­29. Anderson, J. E., and Eric van Wincoop. 2001. “Borders, Trade and Welfare.” Boston College Working Papers in Economics 508, Boston College Department of  Economics. Anderson, K., and R. E. Baldwin. 1987. “The Political Market for Protection in Industrial Countries.” In Protection, Cooperation, Integration and Development, edited by A. M. El-­ Agraa, 20–­36. New York: Macmillan. Anderson, M. L. 2000. Practicing Democracy: Elections and Political Culture in Imperial Germany. Princeton, NJ: Princeton University Press. Ansolabehere, S., and J. M. Snyder. 2000. “Campaign War Chests and Congressional Elections.” Business and Politics 2(April): 9–­34. Armijo, Leslie. 2007. “The BRICs Countries (Brazil, Russia, India, and China) as Analytical Category: Insight or Mirage?” Asian Perspective 31(4): 1–­42. Armijo, Leslie, and Christine A. Kearney. 2008. “Does Democratization Alter the Policy Process? Trade Policymaking in Brazil.” Democratization 15(5; October): 991–­1017. Armijo, Leslie Elliott, Thomas J. Biersteker, and Abraham F. Lowenthal. 1994. “The Prob­ lems of Simultaneous Transitions.” Journal of Democracy 5(4): 161–­75. Athukorala, Prema-­Chandra. 2005. “Trade Policy Reforms and the Structure of  Protection in Vietnam.” Departmental Working Papers 2005–­6, Australian National University, Economics RSPAS. Auyero, Javier. 1999. “‘From the Client’s Point(s) of  View’: How Poor People Perceive and Evaluate Political Clientelism.” Theory and Society 28(2): 297–­334. Ayine, Dominic M. 2008. “Managing Trade Liberalization: Legal System Deficiencies and the Political Economy of Contingency Protection in Ghana.” Journal of African Law 48(2): 207–­38. Baker, A. 2003. “Why Is Trade Reform so Popular in Latin America? A Consumption-­Based Theory of  Trade Policy Preferences.” World Politics 55(April): 423–­55. ————. 2005. “Who Wants to Globalize: Consumer Tastes and Labor Markets in a Theory of  Trade Policy Beliefs.” American Journal of Political Science 49(4): 924–­38. Baldwin, R., and C. Magee. 2000. “Is Trade Policy for Sale? Congressional Voting on Recent Trade Bills.” Public Choice 105: 79–­101. Baldwin, Richard E. 1999. “Agglomeration and Endogenous Capital.” European Economic Review 43(2): 253–­80. Baldwin, Robert E. 1995. “The Effects of Trade and Foreign Direct Investment on Em­ ployment and Relative Wages.” NBER Working Paper No. w5037, National Bureau of  Economic Research, Harvard University, Cambridge, MA. Balisacan, Arsenio M., and Hal Hill (eds.). 2003. The Philippine Economy: Development, Policies, and Challenges. New York: Oxford University Press.

References / 281 Balistreri, E. J. 1997. “The Performance of  the Heckscher-­Ohlin-­Vanek Model in Predicting Endogenous Policy Forces at the Individual Level.” Canadian Journal of Economics 30(1): 1–­17. Baltagi, B., S. Song, B. Jung, and W. Koh. 2007. “Testing for Serial Correlation, Spatial Autocorrelation and Random Effects Using Panel Data.” Journal of Econometrics 140(1): 5–­51. Bandyopadhyay, S. 1997. “Demand Elasticities, Asymmetry and Strategic Trade Policy.” Journal of International Economics 42(1): 167–­77. Bandyopadhyay, S., and S. Roy. 2007. “Corruption and Trade Protection: Evidence from Panel Data.” Working Paper 2007-­022, Federal Reserve Bank of  St. Louis. Barro, Robert. 1997. “Determinants of  Economic Growth: A Cross-­Country Study.” NBER Working Paper 5698, National Bureau of Economic Research, Harvard University, Cambridge, MA. Barro, Robert J., and Xavier Sala-­i-­Martin. 2003. Economic Growth. 2nd edition. Cambridge: MIT Press. Bates, R., and A. Krueger. 1993. “Generalizations Arising from the Country Studies.” In Political and Economic Interactions in Economic Policy Reform, edited by Robert Bates and Anne Krueger. Cambridge: Blackwell. Bates, Robert, and Anne Krueger (eds.). 1993. Political and Economic Interactions in Economic Policy Reform: Evidence from Eight Countries. Oxford: Oxford University Press. Baunsgaard, T., and M. Keen. 2005. “Tax Revenue and (or?) Trade Liberalization.” IMF Working Paper 05/112. Beattie, Alan, and Frances Williams. 2005. “Prospects for WTO Gloomier as Talks End.” Financial Times, 27 July: 6. Beck, N., and J. Katz. 1995. “What to Do and Not to Do with Time-­Series-­Cross-­Sectional Data.” American Political Science Review 89(3): 634–­47. ————. 1995. “What to Do and Not to Do with Time-­Series-­Cross-­Sectional Data.” American Political Science Review 89(3): 634–­47. Beck, Nathaniel, Jonathan N. Katz, and Richard Tucker. 1998. “Taking Time Seriously: Time-­Series–­Cross-­Section Analysis with a Binary Dependent Variable.” American Jour­ nal of Political Science 42(4): 1260–­88. Beck, T., G. Clarke, A. Groff, P. Keefer, and P. Walsh. 2002. “New Tools in Comparative Political Economy: The Database of  Political Institutions.” World Bank Economic Review 15(1): 165–­76. Beghin J. 2008. “Non-­tariff Barriers.” In The New Palgrave Dictionary of Economics, 2nd edi­ tion, edited by S. N. Durlauf and L. E. Blume. New York: Palgrave Macmillan. Beghin, J., and M. Kherallah. 1994. “Political Institutions and International Patterns of Agricultural Protection.” The Review of Economics and Statistics 76(3): 482–­89. Beinart, W. 2001. Twentieth-­Century South Africa. Oxford: Oxford University Press. Bell, T. 1997. “Trade Policy.” In The Political Economy of South Africa’s Transition, edited by J. Michie and V. Padayachee. London: Dryden Press. Beres, W., and K. Burnetko. 2009. Nasza historia: 20 lat RP. Warsaw: Świat Ksiażki. Bergsten, C. Fred. 2008. China’s Rise: Challenges and Opportunities. Washington, DC: Pe­ terson Institute. Bergsten, F., and W. R. Cline. 1983. “Trade Policy in the 1980s: An Overview.” In Trade Policy in the 1980s, edited by William R. Cline. Washington, DC: Institute for International Economics. Bergsten, F., and J. Williamson. 1983. “Exchange Rates and  Trade Policy.” In Trade Policy in the 1980s, edited by  William R. Cline. Washington, DC: Institute for International Economics.

282 / References Bhagwati, J., and T. Srinivasan. 1999. “Outward-­Orientation and Development: Are the Revisionists Right?” Economic Growth Centre Discussion Paper no. 806, Yale University, New Haven, CT. ——— —. 2002. “Trade and Poverty in the Poor Countries.” American Economic Review 92(9): 180–­83. Bienen, Henry. 1991. “The Politics of  Trade Liberalization.” In Trade and Development in Sub-­Saharan Africa, edited by J. H. Frimpong-­Ansah, S. M. Ravi Kanbur, and Peter Svedberg. Manchester, UK: Manchester University Press. Black Economic Empowerment Commission (BEECOM). 2001. Black Economic Em­ powerment Report. Johannesburg: Skotaville. Blonigen, Bruce A., and Thomas J. Prusa. 2001. “Antidumping.” NBER Working Paper No. 8398, National Bureau of Economic Research, Harvard University, Cambridge, MA. Blundell, R. W., and S. R. Bond. 1998. “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models.” Journal of Econometrics 87: 115–­43. Board of Tariffs and Trade of South Africa. 1988–­2003. Annual Reports. Pretoria, South Africa: Pretoria Government Printing. Boeri, Tito, and Katherine Terrell. 2002. “Institutional Determinants of  Labor Reallocation in Transition.” Journal of Economic Perspectives 16(1): 51–­76. Boix, C., and S. Stokes. 2003. “Endogenous Democratization.” World Politics 55(4): 517–­49. Boix, Carles. 2003. Democracy and Redistribution. New York: Cambridge University Press. Bond, P. 2000. Elite Transition: From Apartheid to Neoliberalism in South Africa. London: Pluto Press. Boraine, Alex. 2001. A Country Unmasked: Inside South Africa’s Truth and Reconciliation Commission. New York: Oxford University Press. Bourguignon, François. 2004. “The Poverty-­Growth-­Inequality Triangle.” Working paper presented at the Indian Council for Research on International Economic Relations, New Delhi, India. Bowman, K. S. 2002. Militarization, Democracy, and Development: The Perils of Praetorianism in Latin America. State College, PA: Penn State University Press. Bown, Chad Philips. 2009. “The Global Resort to Antidumping, Safeguards, and Other Trade Remedies Amidst the Economic Crisis.” World Bank Policy Research Working Paper No. 5051, World Bank, Washington, DC. ————. 2010. Self-­Enforcing Trade: Developing Countries and WTO Dispute Settlement. Washington, DC: Brookings Institution Press. Boyce, James K. 1993. The Philippines: The Political Economy of Growth and Impoverishment in the Marcos Era. London: Macmillan Press. ————. 1994. “Inequality as a Cause of Environmental Degradation.” Ecological Economics 11(3): 169–­78. Boylan, Delia. 1999. “Bureaucratic Reform in Developing Countries: A Comparison of Presidential and Parliamentary Rule.” Working Papers 9914, Harris School of  Public Policy Studies, University of Chicago. Bradford, Scott. 2000. “Rents, Votes, and Protection: Explaining the Structure of  Trade Barriers Across Industries.” Paper presented at the Econometric Society, Seattle, WA. Brambor, Thomas, William Roberts Clark, and Matt Golder. 2005. Understanding Interaction Models: Improving Empirical Analyses. Political Analysis 14(1; January): 63–­82.

References / 283 Bratton, Michael. 2008. “Vote Buying and Violence in Nigerian Election Campaigns.” Electoral Studies 27(4): 621–­32. Brazil Ministry of Development, Industry, and Commerce. Various years, including 2003. Outras estatísticas brasileiras de comércio exterior. Brasília: Government of Brazil. “Brazil’s New Trade Policy.” 1987. Agora São Paulo, April 10: B1. Brooks, H. 2004. “The Dominant Party System: Challenges for South Africa’s Second Decade of Democracy.” The Electoral Institute of Southern Africa, Occasional Paper No. 25, Johannesburg, South Africa. Brooks, Sarah M., and Marcus J. Kurtz. 2007. “Capital, Trade, and the Political Economies of Reform.” American Journal of Political Science 51(4): 703–­20. Bruno, M. 1996. Deep Crises and Reform. What Have We Learned? Washington, DC: World Bank. Brusco, Valeria, Marcelo Nazareno, and Susan Stokes. 2004. “Vote Buying in Argentina.” Latin American Research Review 39(2): 66–­88. Bueno de Mesquita, B., A. Smith, R. R. Siverson, and J. D. Morrow. 2002. “Political Institutions, Policy Choice and the Survival of  Leaders.” British Journal of Political Science 32(4): 559–­90. ————. 2003. The Logic of Political Survival. Cambridge: MIT Press. Buiter, Willem. 2003. “The Failure of  Cancun.” Retrieved December 10, 2012, from http:// www.nber.org/~wbuiter/cancun.pdf. Bullock, D. S. 1992. “Welfare Implications of  Equilibrium Supply and Demand Curves in an Open Economy.” American Journal of Agricultural Economics 75(February): 52–­58. Bunce, V., and S. L. Wolchik. 2006. “Favorable Conditions and Electoral Revolutions.” Journal of Democracy 17(4): 5–­18. Bunce, Valerie. 2003. “Rethinking Recent Democratization: Lessons from the Post-­ communist Experience.” World Politics 55: 167–­92. Burbidge, J. B., Magee, L., and Robb, A. L. 1988. “Alternative Transformations to Handle Extreme Values of  the Dependent Variable.” Journal of the American Statistical Association 83(401): 123–­27. Burns, E. D. 1993. A History of Brazil. New York: Columbia University Press. Busch, M., and E. Reinhardt. 1999. “Industrial Location and Protection: The Political and Economic Geography of U.S. Nontariff Barriers.” American Journal of Political Science 43(4): 1028–­50. ————. 2000. “Geography, International Trade, and Political Mobilization in U.S. In­ dustries.” American Journal of Political Science 44(4): 703–­19. Busch, Marc L., and Eric Reinhardt. 2006. “Three’s a Crowd: Third Parties and WTO Dispute Settlement.” World Politics 58(3): 446–­77. “Business and International Trade.” 1999. Business Day, April 6, p. 7. Cable News Network (CNN). 1998. “In Rural Brazil, Vote Buying a Time-­Honored Practice.” Retrieved from http://www.latinamericanstudies.org/brazil/vote.htm. Cadot, O., J. De Melo, and M. Olarreaga. 2004. “Lobbying, Counterlobbying, and the Structure of  Tariff Protection in Rich and Poor Countries.” World Bank Economic Review 18(3): 345–­66. Cadot, Olivier, Céline Carrère, and Vanessa Strauss-­Kahn. 2011. “Export Diversification: What’s behind the Hump?” Review of Economics and Statistics 93(2): 590–­96. Cain, B., J. Ferejohn, and M. Fiorina. 1987. The Personal Vote: Constituency Service and Electoral Independence. Cambridge: Harvard University Press. Cameron, J. 2002. “South Africa: SA Gov ‘Astonished’ by Market Chaos.” Retrieved August 15, 2002, from http://allafrica.com/stories/200208160394.html.

284 / References Cameron, S., Chaudhuri, S., and J. McLaren. 2007. “Trade Shocks and Labor Adjustment: Theory.” NBER Working Paper No. 13463, National Bureau of Economic Research, Harvard University, Cambridge, MA. Cárdenas, Mauricio, and Guillermo Perry. 2011. “Fiscal Policy in Latin America.” In The Oxford Handbook of Latin American Economics, edited by Jose Antonio Ocampo and Jaime Ros, 266–­92. New York: Oxford University Press. Cardoso, Eliana. 2009. “A Brief History of Trade Policies in Brazil: From Isi, Export Promotion and Import Liberalization to Multilateral and Regional Agreements.” Paper presented at The Political Economy of  Trade Policy in the BRICS Conference, New Orleans, LA, March 27–­28. Carey, J., and M. S. Shugart. 1995. “Incentives to Cultivate a Personal Vote: A Rank-­ Ordering of  Electoral Systems.” Electoral Studies 14(4): 417–­39. ————. 1998. Executive Decree Authority. New York: Cambridge University Press. Carey, J. M. 1998. Term Limits and Legislative Representation. Cambridge: Cambridge Uni­ versity Press. Carey, John. 2003. “Discipline, Accountability, and Legislative Voting in Latin America.” Comparative Politics 35(2): 191–­211. ————. 2007. “Competing Principals, Political Institutions, and Party Unity in Legislative Voting.” American Journal of Political Science 51(1): 92–­107. Carey, John, and Andrew Reynolds. 2007. “Parties and Accountable Government in New Democracies.” Party Politics 13(2): 255–­74. Cassing, James H. 1996. “Protectionist Mutual Funds.” European Journal of Political Economy 12(1): 1–­18. Castro, M. F., and M. I. V. Carvalho. 2002. “Globalização e transformações políticas recentes no Brasil: Os anos 1990.” Revista de Sociologia Política 18: 109–­29. Celasun, Merih, and Ismail Arslan. 1997. “State-­Owned Enterprises and Privatization in Turkey: Policy, Performance and Reform Experience, 1985–­1995.” Working Paper 9709, Economic Research Forum. Central Intelligence Agency. 2008. The World Factbook 2008. Washington, DC: Author. ——— —. 2009. “The World Factbook: Rank Order—­India GDP (Purchasing Power Parity).” Retrieved from https://www.cia.gov/library/publications/the-world-factbook/rank order/2001rank.html. Chang, Eric. 2005. “Electoral Incentives for Political Corruption under Open-­List Pro­ portional Representation.” Journal of Politics 67(3): 716–­30. ————. 2008 “Electoral Incentives and Budgetary Spending: Rethinking the Role of Political Institutions.” Journal of Politics 70(4): 1086–­97. Chang, Eric, and Miriam Golden. 2007. “Electoral Systems, District Magnitude and Corruption.” British Journal of Political Science 37(1): 115–­37. Cheibub, J. A., J. Gandhi, and J. R. Vreeland. 2010. “Democracy and Dictatorship Re­ visited.” Public Choice  143(1–­2): 67–­101. Chhibber, Pradeep. 1999. Democracy without Associations: Transformation of the Party System and Social Cleavages in India. Ann Arbor: University of Michigan Press. Chhibber, Pradeep, and Ken Kollman. 2009. The Formation of National Party Systems: Federalism and Party Competition in Canada, Great Britain, India, and the United States. Princeton, NJ: Princeton University Press. Chinn, M., and H. Ito. 2006. “What Matters for Financial Development? Capital Controls, Institutions, and Interactions.” Journal of Development Economics 81(1; October): 163–­92.

References / 285 Claessens, Stijn, Erik Feijen, and Luc Laeven. 2008. “Political Connections and Preferential Access to Finance: The Role of  Campaign Contributions.” Journal of Financial Economics 88(3): 554–­80. Clemens, M. A., and J. G. Williamson. 2001. “A Tariff-­Growth Paradox? Protection’s Impact the World Around 1875–­1997.” NBER Working Paper No. 8459, National Bureau of Economic Research, Harvard University, Cambridge, MA. Cline, William Richard. 2004. Trade Policy and Global Poverty. Washington, DC: Peterson Institute. Clinton, Bill. 1996. Between Hope and History: Meeting America’s Challenges for the 21st Cen­ tury. New York: Crown Publishing. Collier, D., Brady, H. E., and J. Seawright. 2004. “Sources of Leverage in Causal Inference: Toward an Alternative View of  Methodology.” In Rethinking Social Inquiry: Diverse Tools, Shared Standards, edited by Henry Brady and David Collier. Lanham, MD: Rowman and Littlefield. Confederação Nacional da Indústria. 1994. Abertura comercial e estratégia tecnológica: A visão de líderes industriais brasileiros [Commercial opening and technology strategy: The view of Brazilian industrial leaders]. Rio de Janeiro: Author. ————. 1998. Abertura comercial e estratégia tecnológica: A visão de líderes industriais brasileiros [Commercial opening and technology strategy: The view of Brazilian industrial leaders]. Rio de Janeiro: Author. ————. 2002. Abertura comercial e estratégia tecnológica: A visão de líderes industriais brasileiros [Commercial opening and technology strategy: The view of Brazilian industrial leaders]. Rio de Janeiro: Author. Cook, Thomas D., and Donald T. Campbell. 1979. Quasi-­experimentation: Design and Analysis Issues for Field Settings. Chicago: Rand McNally. Cooper, John Milton. 2008. Reconsidering Woodrow Wilson: Progressivism, Internationalism, War, and Peace. Baltimore: Johns Hopkins University Press. Coppedge, Michael. 1994. “Prospects for Democratic Governability in Venezuela.” Journal of Inter-­American Studies and World Affairs 36(2): 39–­64. Cororaton, C. B. 1998. “Total Factor Productivity of  the Philippine Manufacturing Sector: 1956–­1995.” Unpublished manuscript. Philippine Institute for Development Studies. Manila: Philippines. Corrales, Javier. 2001. “Strong Societies, Weak Parties: Regime Change in Cuba and Venezuela in the Late 1950s and Today.” Latin American Politics and Society 43(2): 81–­113. Costa Filho, C. R. P. 1995. “Liberalização do Comércio: Padrões de Interação entre Elites Burocráticas e Atores Sociais.” Paper presented at the nineteenth annual meeting of ANPOCS, Caxambu, Brazil. Costinot, A. 2009. “On the Origins of Comparative Advantage.” Journal of International Eco­ nomics 77(2): 255–­64. Cox, Gary W., and M. D. McCubbins. 1993. Legislative Leviathan: Party Government in the House. Berkeley: University of  California Press. ————. 2001. “The Institutional Determinants of Economic Policy Outcomes.” In Pres­ idents, Parliaments and Policy, edited by Stephen Haggard and Mathew D. McCubbins. Cambridge: Cambridge University Press. Cox, Gary W., and Francis Rosenbluth. 1995. “The Anatomy of a Split: The Liberal Democrats of  Japan.” Electoral Studies 14(4): 355–­76. Cox, Gary W., and Michael F. Thies. 2000. “How Much Does Money Matter? ‘Buying Votes’ in Japan, 1967–­1990.” Comparative Political Studies 33(1): 37–­57.

286 / References Dailami, M. 2004. “Financial Openness, Democracy and Redistributive Policy.” World Bank Policy Research Working Paper No. 2372. Das, D. K. 1998. “Trade Liberalisation and Productivity Growth: A Disaggregated Analysis of  Indian Manufacturing Sectors.” Working Paper, No. E/200/98, Institute of  Economic Growth, New Delhi. Dasgupta, Partha. 2001. Human Well-­Being and the Natural Environment. Oxford: Oxford University Press. Datamark (Brazil). 2004. Foreign Trade Database. Sao Paulo, Brazil: Author. Davidson, C., and Matusz, S. J. 2000. “Globalization and Labour-­Market Adjustment: How Fast and at What Cost?” Oxford Review of Economic Policy 16(3): 42–­56. ————. 2001. Trade and Turnover: Theory and Evidence. East Lansing: Michigan State University. Davis, Christina L., and Sarah Blodgett Bermeo. 2009. “Who Files? Developing Country Participation in GATT/WTO Adjudication.” The Journal of Politics 71(3): 1033–­49. Davis, Donald R. 1996. “Trade Liberalization and Income Distribution.” NBER Working Paper No. W5693, National Bureau of Economic Research, Harvard University, Cam­ bridge, MA. Dean, J. M., S. Desai, and J. Riedel. 1994. “Trade Policy Reform in Developing Countries since 1985: A Review of the Evidence.” World Bank Discussion Paper No. 267, World Bank Publications, Washington, DC. Deardorff, A. 1998. “Determinants of  Bilateral Trade: Does Gravity Work in a Neoclassical World?” In The Regionalization of the World Economy, edited by J. Frankel, 7–­32. Chicago: University of Chicago Press. Deardorff, A. V., and R. M. Stern. 1985. “The Structure of Tariff Protection: Effects of Foreign Tariffs and Existing NTBs.” The Review of Economics and Statistics 67(4): 539–­48. Deegan, H. 2000. Politics of the New South Africa: Apartheid and After. Cranbury, NJ: Addison-­Wesley. Del Castillo, Pilar, and Daniel Zovatto. (eds.) 1998. La Financiación de la Política en Democracy. Baltimore: Johns Hopkins University Press. Department of Trade and Industry (DTI). Various years. South African Trade Statistics. Pre­ toria: Government of South Africa. ————. 1994. Reserve Bank: Economic and Financial Data for South Africa. Pretoria: Gov­ ernment of South Africa. ——— —. 1995. The Labor Market and Adjustment to Trade Liberalization. Pretoria: Government of South Africa. ————. 1996. South Africa’s Economic Transformation: A Strategy for Broad-­Based Economic Development. Pretoria: Government of South Africa. De Souza, A. 1999. Cardoso and the Struggle for Reform in Brazil. Journal of Democracy 10(3): 49–­63. Desposato, Scott W. 2006. “The Impact of  Electoral Rules on Legislative Parties: Lessons from the Brazilian Senate and Chamber of Deputies.” Journal of Politics 68(4): 1018–­30. Devine, T. J., and N. M. Kiefer. 1991. Empirical Labor Economics: The Search Approach. New York: Oxford University Press. DGCIS. (various years). Foreign Trade Statistics. New Delhi: Government of India. Dhingra, S. 2009. “Reconciling Observed Tariffs and the Median Voter Model.” Mimeo, Department of Economics, Columbia University. Diamond, Larry. 1999. Developing Democracy: Toward Consolidation. Baltimore: Johns Hopkins University Press.

References / 287 Diamond, Larry, and Marc Plattner (eds.). 1993. The Global Resurgence of Democracy. Baltimore: Johns Hopkins University Press. Di Palma, Giuseppe. 1990. To Craft Democracies: An Essay on Democratic Transitions. Berkeley: University of  California Press. Dominguez, J. 1998. “Free Politics and Free Markets in Latin America.” Journal of Democracy 9(4): 70–­84. Dominguez, J. I., and James A. McCann. 1996. Democratizing Mexico: Public Opinion and Electoral Choices. Baltimore: Johns Hopkins University Press. Dornbusch, R. 1992. “Monetary Policy of Post-­Communism: Lessons from the End of the Austro-­Hungarian Empire.” Weltwirsch Archives 128: 391–­424. Drábek, Z., and J. Brada. 1998. “Exchange Rate Regimes and the Stability of  Trade. Policy in Transition Economies.” Journal of Comparative Economics 26(4): 642–­68. Drope, Jeffrey M. 2007. “The Political Economy of Non-­tariff Barriers in Emerging Econ­ omies.” Political Research Quarterly 60(3): 401–­14. Dunoff, J. L. 2001. “The WTO in Transition: Of  Constituents, Competence and Coherence.” The George Washington International Law Review 33(3/4): 979. Durand, F. 1992. The Political Formation and Consolidation of Peak Business Associations: The Case of Peru. Notre Dame, IN: Helen Kellogg Institute for International Studies, University of Notre Dame. Durand, T. 1998. “Forms of Incompetence.” Working paper presented at the Fourth International Conference on Competence-­Based Management, Norwegian School of Management, Oslo. Dutt, P., and D. Mitra. 2005. “Political Ideology and Endogenous Trade Policy: An Em­ pirical Investigation.” Review of Economics and Statistics 87(1): 59–­72. ——— —. 2006. “Labor versus Capital in Trade-­Policy: The Role of Ideology and Inequality.” Journal of International Economics 69(2): 310–­20. Easterly, W., and R. Levine. 2003. “Tropics, Germs, and Crops: How Endowments Influence Economic Development.” Journal of Monetary Economics 50(1): 3–­39. Easterly, William, and Sergio Rebelo. 1993. “Fiscal Policy and Economic Growth: An Empirical Investigation.” Journal of Monetary Economics 32: 417–­58. Eaton, Kent 2002. Politicians and Economic Reform in New Democracies: Argentina and the Philippines in the 1990s. State College, PA: Penn State University Press. The Economist. 2004. “A Party-­Finance Scandal in Brazil,” February 19. Ederington, Josh, and Phillip Mccalman. 2006. “The Impact of  Trade Liberalization on Productivity within and across Industries: Theory and Evidence.” Working Paper, Department of Economics, University of Kentucky. Edwards, L. 2001. “Trade and the Structure of South African Production, 1984–­97.” Devel­ opment Southern Africa 18(4): 471–­91. ————. 2005. “Has South Africa Liberalised Its Trade?” South African Journal of Econom­ ics 73(4): 754–­75. Edwards, L., and P. Alves. 2006. “South Africa’s Export Performance: Determinants of Export Supply.” South African Journal of Economics 74(3): 473–­500. Edwards, L., and R. Lawrence. 2007. “South African Trade Policy Matters: Trade Per­ formance and Trade Policy.” CID Working Paper No. 135, Center for International De­ velopment, Harvard University. Edwards, S. 1995. “Trade Policy, Exchange Rates, and Growth.” In Reform, Recovery, and Growth: Latin America and the Middle East, edited by Rudiger Dornbusch and Sebastian Edwards, 13–­52. Chicago: University of Chicago Press. Ehrlich, Sean D. 2007. “Access to Protection: Domestic Institutions and Trade Policy in

288 / References Democracies.” International Organization 61(3): 571–­605. Eichenrgreen, B., and D. Leblang. 2008. “Democracy and Globalization.” Economics and Politics 20(3): 289–­334. “Empowerment and the IEC.” 1995. Financial Mail, February 12, pp. 2. Estache, A., A. Gomez-­Lobo, and D. Leipziger. M. 2000. Utility Privatization and the Needs of the Poor in Latin America: Have We Learned Enough to Get It Right? Washington, DC: World Bank Publications. Estevez-­Abe, M., Torben Iverson, and David Soskice. 2001. “Social Protection and the Formation of Skills: A Reinterpretation of the Welfare State.” In Varieties of Capitalism: The Institutional Foundations of Comparative Advantage, edited by Peter Hall and David Soskice. London: Oxford University Press. Facchini, G. 2004. “The Political Economy of International Trade and Factor Mobility: A Survey.” Journal of Economic Surveys 18: 1–­31. Facchini, G., and G. Willmann. 2005. “The Political Economy of International Factor Mobility.” Journal of International Economics 67: 201–­19. Facchini, Giovanni, and Gerard Willmann. 2001. “Pareto Gains from Trade.” Economia Politica 18(2): 207–­15. Farrell, David M. 1996. “Campaign Strategies and Tactics.” In Comparing Democracies: Elections and Voting in Global Perspective, edited by Lawrence LeDuc, Richard G. Niemi, and Pippa Norris, 160–­83. London: Sage Publications. Fausto, Boris. 1999. A Concise History of Brazil. Cambridge: Cambridge University Press. Fedderke, J. 2002. “Technology, Human Capital and Growth: Evidence from a Middle Income Country Case Study Applying Dynamic Heterogeneous Panel Analysis.” Paper presented at the Royal Economic Society Annual Conference, Durban, South Africa. Fedderke, J. W., Y. Shin, and P. Vaze. 2003. “Trade and Labour Usage: An Examination of the South African Manufacturing Industry.” Unpublished Working Paper, School of Economics, University of Cape Town, Cape Town, South Africa. Feenstra, Robert C. 1984. “Voluntary Export Restraints in U.S. Autos, 1980–­1981: Quality, Employment and Welfare Effects.” In The Structure and Evolution of Recent U.S. Trade Policy, edited by R. Baldwin and A. Krueger. Chicago: University of Chicago Press for NBER. ————. 2004. Advanced International Trade: Theory and Evidence. Princeton, NJ: Princeton University Press. Feinstein, C. H. 2005. An Economic History of South Africa: Conquest, Discrimination and Development. Cambridge: Cambridge University Press. Fernandez, R., and D. Rodrik. 1993. “Status Quo Bias in the Presence of  Individual-­Specific Uncertainty.” American Economic Review 81: 1146–­55. Ferreira, P., and G. Facchini. 2005. “Trade Liberalization and Industrial Concentration: Evidence from Brazil.” Quarterly Review of Economics and Finance 45: 432–­46. Ferreira, Pedro C. G., and Giovanni Facchini. 2004. “Trade Liberalization and Industrial Concentration: Evidence from Brazil.” Working Paper, Department of Economics, Tinbergen Institute, Amsterdam, Netherlands. Ferreira-­Rubio, D., and M. Goretti. 1998. “When the President Governs Alone.” In Executive Decree Authority, edited by John M. Carey and Matthew Soberg Shugart, 33–­61. New York: Cambridge University Press. Figueiredo, Argelina, and Fernando Limongi. 1996. “Congresso Nacional: Organização, Finance and the Sources of the Personal Vote in Brazil.” Journal of Politics 64(3): 845–­63.

References / 289 Finger, J. M., M. D. Ingco, and U. Reincke. 1996. The Uruguay Round: Statistics on Tariff Concessions Given and Received. Washington, DC: World Bank Publications. Fleischer, D. 1997. “Political Corruption and Campaign Financing: Brazil’s Slow Shift towards Anti-­corruption Laws.” Paper presented at the Twentieth International Con­ gress of the Latin American Studies Association (LASA), Guadalajara, México. Fourie, F. C. van N. 1996. “Industrial Concentration Levels and Trends in South Africa: Completing the Picture.” South African Journal of Economics 64(1): 97–­121. Frankel, J. A., and D. Romer. 1999. “Does Trade Cause Growth?” American Economic Review 89(3): 379–­99. Franzese, R., and J. Hays. 2006. “Calculating and Presenting Spatio-­Temporal Patterns of Diffusion and Interdependence.” Paper presented at the General Conference of the European Consortium of  Political Research, Pisa, Italy. Freedom House. 2005. Freedom in the World: Political Rights and Liberties 1972–­2005. New York: Freedom House. Data set available at http://www.freedomhouse.org/ratings. ————. 2012. Freedom in the World Database. New York: Freedom House. Frieden, J., P. Ghezzi, and E. Stein. 2000. “Politics and Exchange Rates: A Cross-­Country Approach to Latin America.” IDB Working Paper R-­421, Inter-­American Development Bank, Washington, DC. Frieden, J., M. Pastor, and M. Tomz (eds.). 2000. Modern Political Economy and Latin America: Theory and Policy. Boulder, CO: Westview Press. Frieden, J. A. 1992. Debt, Development, and Democracy: Modern Political Economy and Latin America, 1965–­1985. Princeton, NJ: Princeton University Press. Frontline. 1999. “Stability Is Essential for the Economy.” Retrieved from http://www.front line.in/static/html/fl1616/16160140.htm. Frye, T., and E. D. Mansfield. 2004. “Timing Is Everything: Elections and Trade Liberalization in the Post-­Communist World.” Comparative Political Studies 37(4): 371–­98. Frye, Timothy, and Edward D. Mansfield. 2003. “Fragmenting Protection: The Political Economy of Trade Policy in the Post-­Communist World.” British Journal of Political Science 33(4): 635–­57. Galano III, Anthony. 1994. “International Monetary Fund Response to the Brazilian Debt Crisis: Whether the Effects of Conditionality Have Undermined Brazil’s National Sovereignty.” Peace International Law Review 6(2): 323–­51. Ganguly, Sumit, Larry Diamond, and Marc Plattner. 2007. The State of India’s Democracy. Baltimore: Johns Hopkins University Press. Garland, M., and G. Biglaiser. 2009. “Do Electoral Rules Matter? Political Institutions and Foreign Direct Investment in Latin America.” Comparative Political Studies 42(2): 224–­51. Garrett, G. 2000. “The Causes of Globalization.” Comparative Political Studies 33: 941–­91. Garrett, Geoffrey, Alexandra Guisinger, and Jason P. Sorens. 2000. “The Political Economy of Capital Account Liberalization.” Unpublished manuscript, Yale University. Gawande, K., and U. Bandyopadhyay. 2000. “Is Protection for Sale? Evidence on the Grossman-­Helpman Theory of Endogenous Protection.” Review of Economics and statistics 82(1): 139–­52. Gawande, K., and P. Krishna. 2003. “Lobbying and Counter-­Lobbying for  Trade Policy: An Empirical Investigation.” Mimeo, Princeton University. Gawande, Kishore, and Wendy L. Hansen. 1999. “Retaliation, Bargaining, and the Pursuit of ‘Free and Fair’ Trade.” International Organization 53: 117–­59.

290 / References Geddes, B. 1994. Politician’s Dilemma: Building State Capacity in Latin America. Series on Social Choice and Political Economy. Chicago: University of California Press. Gehlbarh, M., and S. Frandsen. 2002. “Trade Elasticities, Aggregation Bias and Welfare Effects in the GTAP Model.” Working Paper, Purdue University. Ghose, A. K. 1995. “Labour Market Flexibility and the Indian Economy.” The Indian Journal of Labour Economics 38(1): 55–­62. Giavazzi, Francesco, and Guido Tabellini. 2005. “Economic and Political Liberalizations.” Journal of Monetary Economics 52(7): 1297–­330. Gielen, Anne C. 2007. “Performance Pay, Training and Labor Mobility.” IZA Discussion Paper no. 2932, National Bureau of Economic Research, Cambridge, MA. Giliomee, H., and C. Simkins (eds.). 1999. The Awkward Embrace: One-­Party Domination and Democracy. Cape Town: Tafelberg. Gilligan, Michael, J. 1997. Empowering Exporters: Delegation, Reciprocity and Collective Action in Twentieth Century American Trade Policy. Ann Arbor: University of Michigan Press. Gleditsch, Kristian Skrede, and Michael D. Ward. 2006. “Diffusion and the International Context of Democratization.” International Organization 60(4): 911–­33. Global Trade Analysis Project. 2006. Global Trade, Assistance, and Production: The GTAP 6 Database. West Lafayette, IN: Center for Global Trade Analysis, Purdue University. ————. 2011. Global Trade, Assistance, and Production: The GTAP 9 Database. West Lafayette, IN: Center for Global Trade Analysis, Purdue University. Goldar, B. N., and Anita Kumari. 2002. “Import Liberalization and Productivity Growth in Indian Manufacturing Industries in the 1990s.” Working Paper No. E/219/2002, Institute of Economic Growth, New Delhi. Goldberg, O. J., and G. Maggi. 1999. “Protection for Sale: An Empirical Investigation.” American Economic Review 89(5): 1135–­55. Goldberg, O. K., and N. Pavcnik. 2007. “Distributional Effects of Globalization in De­ veloping Countries.” NBER Working Paper No. 12885, National Bureau of Economic Research, Harvard University, Cambridge, MA. Goldberg, P., and N. Pavcnik. 2004. “Trade, Inequality and Poverty: What Do We Know?” Paper presented at Brookings Trade Forum, Brookings Institute, Washington, DC. Goldberg, P., and N. Pavcnik. 2007. “Distributional Effects of Globalization in Developing Countries.” Journal of Economic Literature 45(1): 39–­82. Golden, M., and E. Chang. 2001. “Competitive Corruption: Factional Conflict and Polit­­ ical Malfeasance in Postwar Italian Christian Democracy.” World Politics 53(4): 588–­622. Government of India. Various years. Economic Survey. New Delhi: Ministry of  Finance. ————. Various years. Annual Survey of Industries. New Delhi: Central Statistical Organization. Gowa, Joanne, and Soo Yeon Kim. 2005. “An Exclusive Country Club: The Effects of the GATT on Trade, 1950–­1994.” World Politics 57(July): 453–­78. Goyal, S. K. 1979. Monopoly Capital and Public Policy. New Delhi: Allied Publishers. Graham, E. 1989. Politics and Patronage in Nineteenth Century Brazil. Stanford, CA: Stanford University Press. Grofman, Bernard. 2005. “Comparisons among Electoral Systems: Distinguishing between Localism and Candidate-­Centered Politics.” Electoral Studies 24(4): 735–­40. Grossman, G., and E. Helpman. 1994. “Protection for Sale.” American Economic Review 84(4): 833–­50. ————. 2002. Interest Groups and Trade Policy. Princeton, NJ: Princeton University Press.

References / 291 ————. 2006. “A Protectionist Bias in Majoritarian Politics.” The Quarterly Journal of Economics 120(4): 1239–­82. Grossman, G. M. 1984. “The Gains from International Factor Movements.” Journal of International Economics 17(1): 73–­83. Grossman, Gene M., and Elhanan Helpman. 2005. “Outsourcing in a Global Economy.” The Review of Economic Studies 72(1): 135–­59. Grossman, S. J., and O. D. Hart. 1986. “The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration.” Journal of Political Economy 94(4): 691–­718. Guisinger, Alexandra. 2001. “The Determinants of Trade Tax Liberalization.” Working Paper, Yale University. Habib, Adam, and Rupert Taylor. 2001. “Political Alliances and Parliamentary Opposition Post-­apartheid South Africa.” Democratization 8(1): 207–­26. Haggard, S., and R. Kauffman. 1995. The Political Economy of Democratic Transitions. Princeton, NJ: Princeton University Press. Haggard, Stephan. 1986. “The Newly Industrializing Countries in the International Sys­ tem.” World Politics 38(2): 343–­70. ————. 1990. Pathways from the Periphery: The Politics of Growth in the Newly Industrializing Countries. Ithaca, NY: Cornell University Press. Hainmueller, J., and M. J. Hiscox. 2006. “Learning to Love Globalization: The Effects of Education on Individual Attitudes toward International Trade.” International Orga­ nization 60(2): 469–­98. ——— —. 2007. “Educated Preferences: Explaining Individual Attitudes toward Immigration in Europe.” International Organization 61(2): 399–­442. Hainmueller, Jens, and Michael J. Hiscox. 2008. “Being Specific: Measuring Asset Specificity for Political Economy.” Working Paper, Harvard University. Hallerberg, M., and P. Marier. 2004. “Executive Authority, the Personal Vote, and Budget Discipline in Latin American and Caribbean Countries.” American Journal of Political Science 48(3): 571–­87. Hamermesh, D. 1993. Labor Demand. Princeton, NJ: Princeton University Press. Hamermesh, D. S. 1987. “The Costs of  Worker Displacement.” The Quarterly Journal of Economics 102(1): 55–­66. Hamermesh, D. S., and G. Pfann. 1996. “Adjustment Costs in Factor Demand.” Journal of Economic Literature 34(3): 1264–­92. Hankla, Charles. 2005. “Entering the Global Economy: Democracy, Party Strength, and the Political Economy of  International Trade.” PhD dissertation, Emory University. ————. 2006. “Party Strength and International Trade: A Cross National Analysis.” Com­ parative Political Studies 39(9): 1133–­56. Hardgrave, R. L., and S. A. Kochanek. 2007. India: Government and Politics in a Developing Nation. Oxford: Oxford University Press. Harris, John. 1987. “The State in Retreat? Why Has India Experienced Such Halfhearted Liberalization in the 80s?” Institute of Development Studies Bulletin 18(4): 3–­29. Harrison, A., and G. Hanson. 1999. “Who Gains from Trade Reform? Some Remaining Puzzles.” Journal of Development Economics 59(1): 125–­54. Harrison, Ann. 1996. “Openness and Growth: A Time-­Series, Cross-­Country Analysis for Developing Countries.” Journal of Development Economics 48(2): 419–­47. ————. 2007. Globalization and Poverty. Cambridge, MA: National Bureau of Economic Research. Harrison, Ann, and Ana Revenga. 1995. “Factor Markets and Trade Policy Reform.” Working Paper, World Bank, Washington, DC.

292 / References ————. 1998. “Labour Markets, Foreign Investment and Trade Policy Reform.” In Trade Policy Reform: Lessons and Implications, edited by J. Nash and W. Takacs. Washington, DC: World Bank. Harrison, Ann, and Andrés Rodríguez-­Clare. 2009. “Trade, Foreign Investment, and Industrial Policy for Developing Countries.” NBER Working Paper No. W15261, National Bureau of  Economic Research, Harvard University, Cambridge, MA. Harrison, Glen W., Thomas F. Rutherford, David G. Tarr, and Angelo Gurgel. 2004. “Trade Policy and Poverty Reduction in Brazil.” Working Paper No. 276, Central Bank of Chile, Santiago, Chile. Harrison, Peter. 2009. “Median Wages and Productivity Growth in Canada and the United States.” Working Paper, Center for the Standard of  Living Standards (CSLS), Research Note 2, Ottawa, Ontario, Canada. Hatfield, J. W., and Hauk, W. R. 2004. “The Effects of the Electoral Regime on Trade Policy.” Mimeo, Stanford Institute for Economic Policy Research. Hathaway, O. 1998. “Positive Feedback: The Impact of  Trade Liberalization on Industry Demands for Protection.” International Organization 52: 575–­612. Haveman, Jon, and Jerry G. Thursby. 2000. “The Impact of  Tariff and Non-­tariff Barriers to Trade in Agricultural Commodities: A Disaggregated Approach.” Purdue CIBER Working Papers, Paper 143, Purdue University, West Lafayette, LA. Hay, D. A. 2001. “The Post-­1990 Brazilian Trade Liberalization and the Performance of Large Manufacturing Firms: Productivity, Market Share, and Profits.” Economic Journal 111(473): 620–­41. Helleiner, Gerald K. 1993. The New Global Economy and the Developing Countries: Essays in International Economics and Development. Cheltenham, UK: Edward Elgar Publishing. Hellman, Judith Adler. 1994. “Mexican Popular Movements, Clientelism, and the Process of  Democratization.”  Latin American Perspectives 21(2): 124–­42. Henisz, Witold J., and Edward D. Mansfield. 2006. “Votes and Vetoes: The Political. Determinants of Commercial Openness.” International Studies Quarterly 50(1): 189–­211. Herberg, Horst. 1990. “Welfare Effects of Non-­tariff Barriers: A General Equilibrium Analysis.” Review of World Economics 126(3): 511–­22. Heston, Alan, Robert Summers, and Bettina Aten. 2009. Penn World Table. Version 6.3. Center for International Comparisons of Production, Income, and Prices at the Uni­ versity of  Pennsylvania. Retrieved from http://pwt.econ.upenn.edu. Hicken, A., and J. W. Simmons. 2008. “The Personal Vote and the Efficacy of Education Spending.” American Journal of Political Science 52(1): 109–­24. Hicken, Allen. 2004. “The Politics of Economic Reform in Thailand: Crisis and Compromise.” William Davidson Institute Working Paper No. 638, University of Michigan. ————. 2006. “Stuck in the Mud: Parties and Party Systems in Democratic Southeast Asia.” Taiwan Journal of Democracy 2(2): 23–­46. ————. 2009. Building Party Systems in Developing Democracies. New York: Cambridge University Press. Hicks, Raymond, Helen V. Milner, and Dustin Tingley. 2014. “Trade Policy, Economic Interests, and Party Politics in a Developing Country: The Political Economy of CAFTA-­DR.” International Studies Quarterly 58(1): 106–­17. Hill, J., and J. Mendez. 1983. “Factor Mobility and the General Equilibrium Model of Production.” Journal of International Economics 15(August): 19–­25.

References / 293 Hillman, Arye L. 2003. “Trade Liberalization and Globalization: A Survey.” CEPR Dis­ cussion Paper 3845. Hirsch, A. 2005. Season of Hope: Economic Reform under Mandela and Mbeki. Scotsville, South Africa: University of KwaZulu-­Natal Press. Hiscox M. J. 2001. “Class versus Industry Cleavages: Inter-­industry Factor Mobility and the Politics of Trade.” International Organization 55(1): 1–­46. Hiscox, M. J., and S. L. Kastner. 2002. “A General Measure of  Trade Policy Orientations: Gravity-­Model-­Based Estimates for 82 Nations, 1960 to 1992.” Unpublished Working Paper, Harvard University. Hiscox, M. J., and S. Rickard. 2002. “Birds of a Different Feather? Varieties of Capitalism, Factor Specificity and Interindustry Labor Movements.” Unpublished manuscript, Harvard University, Cambridge, MA. Hiscox, Michael J. 2002. “Commerce, Coalitions, and Factor Mobility: Evidence from Congressional Votes on Trade Legislation.” American Political Science Review 96(3): 593–­608. ————. 1999. “The Magic Bullet? The RTAA, Institutional Reform, and Trade Liberal­ ization.” International Organization 53(4): 669–­98. ————. 2001. “Inter-­industry Factor Mobility and the Politics of  Trade.” International Or­ ganization 55(1): 1–­46. Hix, Simon. 2004. “Electoral Institutions and Legislative Behavior: Explaining Voting Defection in the European Parliament.” World Politics 56(1): 194–­223. Hoekman, Bernard M., and Michel M. Kostecki. 2009. The Political Economy of the World Trading System. Oxford: Oxford University Press. Horowitz, Donald. 1992. A Democratic South Africa?: Constitutional Engineering in a Divided Society. Berkeley: University of  California Press. Hudson, P., and M. Sarakinsky. 1986. “Class Interests and Politics: The Case of the Urban African Bourgeoisie.” South African Review 3: 169–­85. Hudson, R. 1997. Brazil: A Country Study. Washington, DC: GPO for the Library of Congress. Hughes, H., and J. Waelbroeck. 1981. “Can Developing Country Exports Keep Growing in The 1980s?” The World Economy 4(2): 127–­48. Huntington, Samuel. 1991. The Third Wave: Democratization in the Late Twentieth Century. Norman: University of Oklahoma Press. Hurlin, C., and B. Venet. 2003. “Granger Causality Tests in Panel Data Models with Fixed Coefficients.” Mimeograph, Universitie Paris I, Paris, France. Hutchcroft, Paul D., and Joel Rocamora. 2003. “Strong Demands and Weak Institutions: The Origins and Evolution of  the Democratic Deficit in the Philippines.” Journal of East Asian Studies 3(2): 259–­92. Hutchful, Eboe. 2002. Ghana’s Adjustment Experience: The Paradox of Reform. Ghana: Woeli Publishing. Hyde, S., and E. Beaulieu. 2009. “In the Shadow of Democracy Promotion: Strategic Manipulation, International Observers, and Election Boycotts.” Comparative Political Studies 42(3): 392–­415. Hyde, S. D. 2011. “Catch Us If You Can: Election Monitoring and International Norm Diffusion.” American Journal of Political Science 55(2): 356–­69. Hyde, Susan D. 2011. The Pseudo-­Democrat’s Dilemma: Why Election Monitoring Became an International Norm. Ithaca, NY: Cornell University Press. IBGE. Various years. Instituto Brasileiro de Geografia e Estatísticas, http://www.ibge.gov.br.

294 / References Independent Electoral Commission (IEC). 1997. National and Provincial Elections Report. Johannesburg, South Africa: Electoral Commission of  South Africa. Institute for Democracy in South Africa (IDSA). 2003. “Regulation of Private Funding to Political Parties.” IDSA Position Paper, Political Information and Monitoring Service-­ South Africa (PIMS-­SA). Inter-­American Development Bank. 1998. “Statistics and Quantitative Analysis Data.” International Economics 10(2): 343–­60. International Labor Organization. 2007. Key Indicators of the Labor Market Database. New York: Author. ————. 2008. Key Indicators of the Labor Market Database? New York: Author. International Monetary Fund. Various years. International Financial Statistics. Washington, DC: Author. ————. 2006. Global Financial Statistics. CD-­ROM. Washington, DC: Author. ————. 2007. Global Financial Statistics. CD-­ROM. Washington, DC: Author. ————. 2008. Government Financial Statistics. CD-­ROM. Washington, DC: Author. ————. 2009. Report for Selected Countries and Subjects. Washington, DC: Author. Retrieved from http://www.imf.org/external/pubs/ft/weo/2009/01/weodata/weorept.aspx. ————. 2011. Government Financial Statistics. Washington, DC: Author. “Interview with Tarun Das: An Uncanny Ability to Anticipate Events.” 1992. Business Standard, 12 January, 6. Irwin, D. A. 1998. “Changes in U.S. Tariffs: The Role of Import Prices and Commercial Policies?” American Economic Review 88: 1015–­26. Irwin, Douglas A., and Marko Terviö. 2002. “Does Trade Raise Income?: Evidence from the Twentieth Century.” Journal of International Economics 58(1): 1–­18. Jacobs, S., and R. Calland (eds.). 2002. Thabo Mbeki’s World. London: Zed Books. Jalan, Bimal (ed.). 1991. The Indian Economy: Reforms and Prospects. New Delhi: Viking Publishers. Jenkins, Rob. 1999. Democratic Politics and Economic Reform in India. Cambridge: Cambridge University Press. Jensen, Michael Friis, and Peter Gibbon. 2007. “Africa and the WTO Doha Round: An Overview.” Development Policy Review 25(1): 5–­24. Johnson, Joel W., and Jessica S. Wallack. 2006. “Electoral Systems and the Personal Vote.” Retrieved from https://dataverse.harvard.edu/dataset.xhtml?persistentId=hdl :1902.1/17901. Jolivet, G., F. Postel-­Vinay, and J. M. Robin. 2006. “The Empirical Content of  the Job Search Model: Labor Mobility and Wage Distributions in Europe and the US.” European Economic Review 50(4): 877–­907. Jones, Mark. P. 2001. “Political Institutions and Public Policy in Argentina: An Overview of the Formation and Execution of the National Budget.” In Presidents, Parliaments, and Policy, edited by Stephan Haggard and Mathew D. McCubbins, 149–­82. Cambridge: Cambridge University Press. Jones, Patrice M. 2002. “Brazil’s ATM-­Style Balloting Draws Praise for Efficiency.” The Chicago Tribune. Retrieved from http://articles.chicagotribune.com/2002-11-01/news /0211010322_1_voting-brazilian-government-ballots. Joshi, Vijay, and I. M. D. Little. 1997. India’s Economic Reforms 1991–­2001. Oxford: Oxford University Press. Kalandadze, K., and M. A. Orenstein. 2009. “Electoral Protests and Democratization Be­ yond the Color Revolutions.” Comparative Political Studies 42(11): 1403–­25.

References / 295 Kambourov, G. 2009. “Labor Market Regulations and the Sectoral Reallocation of  Workers: The Case of  Trade Reforms.” Review of Economic Studies 76(4): 1321–­58. Kang, David C. 2002. Crony Capitalism: Corruption and Development in South Korea and the Philippines. New York: Cambridge University Press. Kapila, U. (ed.). 2001. Indian Economy Since Independence. New Delhi: Academic Foundation. Kassam, Shinan, and Nisha Malhotra. 2006. “Antidumping Duties in the Agriculture Sector: Trade Restricting or Trade Deflecting.” Paper presented at the American Agricultural Economics Association 2006 Annual Meeting, July 23–­26, Long Beach, CA. Katz, Lawrence F., and David H. Autor. 1999. “Changes in the Wage Structure and Earnings Inequality.” In Handbook of Labor Economics, vol. 3, edited by O. Ashenfelter and D. Card. Amsterdam: North-­Holland. Katzenstein, P. 1985. Small States in World Markets. Ithaca, NY: Cornell University Press. Kaufmann, D., J. Montoriol-­Garriga, and F. Recanatini. 2008. “How Does Bribery Affect Public Service Delivery? Micro-­evidence from Service Users and Public Officials in Peru.” Policy Research Working Paper Series 4492, World Bank. Kaufmann, Daniel. 2003. “Rethinking Governance: Empirical Lessons Challenge Orthodoxy.” Macroeconomics 0308007, EconWPA. Washington, DC: World Bank Institute. Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi. 2006. “Governance Matters V: Aggregate and Individual Governance Indicators for 1996–­2005.” Policy Research Working Paper Series 4012, World Bank. Kaufmann, Daniel, and P. Vicente. 2005. “Legal Corruption.” MPRA Paper 8186, University Library of Munich, Germany. Kavanagh, Dennis. 1995. Election Campaigning: The New Marketing of Politics. New Jersey: Blackwell. Kee, Hiau Looi, Alessandro Nicita, and Marcelo Olarreaga. 2006. “Estimating Trade Restrictiveness Indices.” World Bank Policy Research Working Paper No. 3840, World Bank, Washington, DC. Keefer, Philip. 2007. “Clientelism, Credibility, and the Policy Choices of  Young Democ­ racies.” American Journal of Political Science 51(4): 804–­21. Keele, Luke J., and Nathan J. Kelley. 2006 “Dynamic Models for Dynamic Theories: The Ins and Outs of LDVs.” Political Analysis 14(2): 186–­205. Kelley, Judith G. 2008. “Assessing the Complex Evolution of Norms: The Rise of Inter­ national Election Monitoring.” International Organization 62(2): 221–­55. ————. 2012. “International Influences on Elections in New Multiparty States.” Annual Review of Political Science 15: 203–­20. Kennedy, David. 2008. “The Mystery of  Global Governance.” Ohio Northern University Law Review 34: 827–­60. King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry. Princeton NJ: Princeton University Press. Kingstone, Peter. 1999. Crafting Coalitions for Reform: Business Preferences, Political Institutions and Neoliberal Reform in Brazil. State College, PA: Penn State University Press. ————. 2001. “Why Free Trade ‘Losers’ Support Free Trade: Industrialists and the Surpris­ ing Politics of  Trade Reform in Brazil.” Comparative Political Studies 9: 986–­1010. Kishore, Gawande, P. Krishna, and M. Olarreaga. 2009. “What Governments Maximize and Why: The View from Trade.” International Organization 63: 491–­532.

296 / References Kishore, Gawande, P. Krishna, and M. J. Robbins. 2006. “Foreign Lobbying and U.S. Trade Policy.” Review of Economics and Statistics 88(3): 563–­71. Kishore, Gawande, W. Maloney, and G. V. Montes-­Rojas. 2009. “Foreign Informational Lobbying Can Enhance Tourism: Evidence from the Caribbean.” Journal of Development Economics 90(2): 267–­75. Kitschelt, H. 2003. “Competitive Party Democracy and Political-­Economic Reform in Germany and Japan: Do Party Systems Make a Difference?” In The End of Diversity? Prospects for German and Japanese Capitalism, edited by Kozo Yamamura and Wolfgang Streeck, 334–­63. Ithaca, NY: Cornell University Press. Kitschelt, H., Z. Mansfeldova, R. Markowski, and G. Toka. 1999. Post-­Communist Party Systems: Competition, Representation, and Inter-­Party Collaboration. Cambridge: Cam­ bridge University Press. Kitschelt, Herbert. 1995. “Formation of Party Cleavages in Post-­Communist Democracies Theoretical Propositions.” Party Politics 1(4): 447–­72. ————. 2000. “Linkages between Citizens and Politicians in Democratic Polities.” Com­ parative Political Studies 33(6–­7): 845–­79. Kitschelt, Herbert, and Steven I. Wilkinson (eds.). 2007. Patrons, Clients and Policies: Patterns of Democratic Accountability and Political Competition. New York: Cambridge University Press. Klasen, Stephan, and Ingrid Woolard. 1998. “Surviving Unemployment without State Support: Unemployment and Household Formation in South Africa.” IZA Discussion Papers 237, Institute for the Study of  Labor. Klesner, Joseph. 1993. “Modernization, Economic Crisis, and Electoral Alignment in Mex­ ico.” Mexican Studies/Estudios Mexicanos 9(2): 187–­223. Kletzer, L. 2002. Imports, Exports, and Jobs: What Does Trade Mean for Employment and Job Loss? Kalamazoo, MI: W. E. Upjohn Institute for Employment Research. Kletzer, L. G. 2004. “Trade-­Related Job Loss and Wage Insurance: A Synthetic Review.” Paper submitted for the Globalization, Employment, and Economic Develop­ ment Workshop, Sloan Workshop Series in Industry Studies, Rockport, Massachusetts, June 14–­16. Kliot, Nurit, and David Newman (eds.). 2013. Geopolitics at the End of the Twentieth Century: The Changing World Political Map. New York: Routledge. Kochanek, Stanley A. 1985. “The Politics of Regulation: Rajiv’s New Mantras.” Journal of Commonwealth and Comparative Politics 23(3): 189–­211. ————. 1987. “Briefcase Politics in India: The Congress Party and the Business Elite.” Asian Survey 27(12): 1278–­1301. ————. 1995. “The Transformation of Interest Politics in India.” Pacific Affairs 68(4), 529–­50. ————. 1996. “Liberalisation and Business Lobbying in India.” Journal of Commonwealth and Comparative Politics 34(3): 155–­73. Kohli, Atul (ed.). 2001. The Success of India’s Democracy. Cambridge: Cambridge University Press. Kohli, D. N., B. N. Sharma, and P. Avasthi. 1983. Customs Tariff in India. New Delhi: CEN-­CUS. Kono, D. Y. 2007. “Who Liberalizes? Explaining Preferential Trade Liberalization.” Inter­ national Interactions 33(4): 401–­21. Kono, Daniel. 2006. “Optimal Obfuscation: Democracy and Trade Policy Transparency.” American Political Science Review 100: 369–­84. ————. 2008. “Democracy and Trade Discrimination.” Journal of Politics 70(4): 942–­55.

References / 297 Kono, Daniel Yuichi. 2009. “Market Structure, Electoral Institutions, and Trade Policy.” International Studies Quarterly 53(4): 885–­906. Kotin, Daniel. 1999. “The Political Economy of  Non-­tariff  Trade Barriers: A Test of  the Veto Players Theory of Policy Change.” Working Paper, Department of Political Science, University of California at Los Angeles. Kotze, Hennie, and Pierre Du Toit. 1995. “The State, Civil Society, and Democratic Transition in South Africa: A Survey of Elite Attitudes.” Journal of Conflict Resolution 39(1): 27–­48. Krueger, A. 1993. “How Computers Have Changed the Wage Structure: Evidence from Microdata, 1984–­1989.” Quarterly Journal of Economics 108: 33–­60. Krueger A. O. 1982. Trade and Employment in Developing Countries. Chicago: University of Chicago Press. Krueger, Anne O. 1990. “Asian Trade and Growth Lessons.” The American Economic Review 80(2): 108–­12. ————. 1996. “Threats to 21st Century Growth: The Challenge of the International Trading System.” In The Mosaic of Economic Growth, edited by Ralph Landau, Timothy Taylor, and Gavin Wright, 191–­214. Stanford CA: Stanford University Press. Kularatne, C. 2006. “Social and Economic Infrastructure Impacts on Economic Growth in South Africa.” Mimeo, National Treasury. Kumar, Arun. 2002. Black Economy in India. New Delhi: Penguin. Kume, Honorio, Guida Piani, and Carlos Frederico Braz de Souza. 2000. A Política Brasileira de Importações no período 1987–­98: Descrição e avaliação. Rio de Janeiro: IPEA. La Ferrara, E. 1996. “A Political Equilibrium Approach to Trade Liberalization in De­ veloping Countries.” Journal of International Trade and Economic Development 5(3): 287–­318. Laird, S., and A. Yeats. 1990. Quantitative Methods for Trade-­Barrier Analysis. New York: Macmillan. Lanegran, K. 2000. “South Africa’s 1999 Election: Consolidating a Dominant Party System.” Africa Today 48(2): 81–­102. Langhammer, R. J., and F. Foders. 2006. Labor Mobility and the World Economy. Berlin: Springer-­Verlag. Laussel, D., and R. Riezman. 2005. “The Sources of  Protectionist Drift in Representative Democracies.” European Economic Review 49(7): 1855–­76. Law Commission of India. 1999. “170th Report on Reform of Electoral Laws.” Retrieved from http://www.lawcommissionofindia.nic.in/lc170.htm. Lawrence, Robert Z. 2000. “Preparing for Seattle.” In Seattle, the WTO, and the Future of the Multilateral Trading System, edited by Roger B. Porter and Pierre Sauve. Cambridge: Harvard University and John F. Kennedy School of Government. Lee, J. W., and Robert J. Barro. 2001. “Schooling Quality in a Cross-­Section of Countries.” Economica 68(272): 465–­88. Lee., J. W., and P. Swagel. 1997. “Trade Barriers and Trade Flows across Countries and Industries.” Review of Economics and Statistics 74: 372–­82. Lehoucq, F., and I. Molina. 2002. Stuffing the Ballot Box: Fraud, Election Reform, and Democratization in Costa Rica. Cambridge: Cambridge University Press. Lehoucq, Fabrice. 2003. “Electoral Fraud: Causes, Types, and Consequences.” Annual Review of Political Science 6(1): 233–­56. Lehoucq, Fabrice, and David Wall. 1998. “Electoral Incentives under Proportional Representation: A Statistical Test.” Paper presented at the annual meeting of the Midwest Political Science Association, Chicago.

298 / References Levine, R. M. 1999. Brazilian Legacies. Armonk, NY: M. E. Sharpe. Levitsky, S., and L. Way. 2006 “Linkage versus Leverage: Rethinking the International Dimension of  Regime Change.” Comparative Politics 38(4; July): 379–­400. Levitsky, Steven. 2003. Transforming Labor-­Based Parties in Latin America: Argentine Peronism in Comparative Perspective. New York: Cambridge University Press. Lewis, Paul. 1994. “Political Institutionalization and Party Development in Post-­ Communist Poland.” Europe-­Asia Studies 45(5): 779–­99. Leyden, K. M., and S. A. Borrelli. 1990. “Party Contributions and Party Unity: Can Loyalty Be Bought.” Western Political Quarterly 43: 343–­65. Li, Q., and R. Reuveny. 2003. “Economic Globalization and Democracy: An Empirical Analysis.” British Journal of Political Science 33(1): 29–­54. Library of Congress. 2004. “Country Profile: India.” Library of Congress: Federal Research Division. Retrieved from http://lcweb2.loc.gov/frd/cs/profiles/India.pdf. Lijphart, A. 1996. “The Puzzle of Indian Democracy: A Consociational Interpretation.” The American Political Science Review 90(2): 258–­68. Lima, Olavo Brasil de, Jr., and Maria Regina Soares de Lima. 1987. “Autonomia Organiza­ cional e Policentrismo Decisório: A Política de Comércio Exterior.” In As Origens da Crise: Estado Autoritário e Planejamento no Brasil, edited by Olavo Brasil de Lima Jr. and Sérgio H. Abranches. São Paulo: Iuperj/Edições Vértice. Lind, Michael. 1997. Hamilton’s Republic: Readings in the American Democratic Nationalist Tradition. New York: Free Press. Linz, J., and A. Stepan. 1996. Problems of Democratic Transition and Consolidation: Southern Europe, South America and Post-­Communist Europe. Baltimore: Johns Hopkins University Press. Lodge, Tom. 2001. “How Political Parties Finance Electoral Campaigning in Southern Africa.” Journal of African Elections 1(1): 53–­60. ————. 2002. Politics in South Africa From Mandela to Mbeki. West Lafayette, IN: Purdue University. Lohmann, S., and S. O’Halloran. 1994. “Divided Government and US Trade Policy: Theory and Evidence.” International Organization 48: 595–­632. Lodge, Tom, and Ursula Sheidegger. 2006. Political Parties and Democratic Governance in South Africa: Electoral Institute of Southern Africa No. 25. Bloomington, IN: Indiana Press. Lomax, Bill. 1995. “Factions and Factionalism in Hungary’s New Party System.” Democ­ ratization 2(1): 125–­38. López-­Cordova, E., and M. M. Moreira. 2003. “Regional Integration and Productivity: The Experiences of Brazil and Mexico.” INTAL-­ITD Working Paper No. 14, Inter-­American Developing Bank, Buenos Aires. Lopez-­Cordova, J. E., and C. Missner. 2008. “The Impact of International Trade on Democracy: A Long-­Run Perspective.” World Politics 60(4): 539–­75. Louw, P. E. 2000. “South Africa’s Second Post-­apartheid Elections: A Reaffirmation of Racial Politics but a Loosening of Ethnic Bonds.” Australian Journal of International Affairs 54(2): 217–­38. Lusztig, M. 2004. The Limits of Protectionism: Building Coalitions for Free Trade. Pittsburgh: University of Pittsburgh Press. Luttwak, Edward. 1999. Turbo-­Capitalism: Winners and Losers in the Global Economy. New York: Weidenfeld and Nicolson. Lyne, Mona. 2008. “Proffering Pork: How Party Leaders Build Party Reputations in Brazil.” American Journal of Political Science 52(2): 290–­303.

References / 299 Macroeconomic Research Group (MERG) for ANC. 1995. Making Democracy Work: A Framework for Macroeconomic Policy in South Africa. 2nd edition. Cape Town: Center for Development Studies. Magee, Stephen. 1980. International Trade. Reading, MA: Addison-­Wesley. Mainwaring, S. 1995. “Introduction: Party Systems in Latin America.” In Building Demo­ cratic Institutions: Party Systems in Latin America, edited by Scott Mainwaring and Tim­ othy Scully. Stanford, CA: Stanford University Press. Mainwaring, S., and A. Perez-­Linan. 1997. “Party Discipline in the Brazilian Constitutional Congress.” Legislative Studies Quarterly 22(4): 453–­83. Mainwaring, S., and T. Scully. 1995. Building Democratic Institutions: Party Systems in Latin America. Stanford, CA: Stanford University Press. Mainwaring, S. and M. Shugart. 1997. Presidentialism and Democracy in Latin America. New York: Cambridge University Press. Mainwaring, Scott. 1997. “Multipartism, Robust Federalism, and Presidentialism: The Case of Brazil.” In Presidentialism and Democracy in Latin America, edited by Scott Mainwaring and Matthew Shugart, 55–­109. New York: Cambridge University Press. ————. 1998. “Party Systems in the Third Wave of Democratization.” The Journal of Democracy 9(3): 67–­81. ————. 1999. Rethinking Party Systems in the Third Wave of Democratization: The Case of Brazil. Stanford, CA: Stanford University Press. Makumbe, John. 2006. “Electoral Politics in Zimbabwe: Authoritarianism versus the People.” Africa Development 31(3): 45–­61. Manifesto dos Mineiros. 1986. Partido do Movimento Democrático Brasileiro, Brasilia, Brazil. Manor, James. 1988. “Parties and the Party System.” In India’s Democracy, edited by A. Kohli, 62–­99. Princeton, NJ: Princeton University Press. Mansfield, E., H. Milner, and J. Pevehouse. 2006. “Vetoing Cooperation: The Impact of Veto Players on International Trade Agreements.” British Journal of Political Science 36(4): 403–­32. ————. 2008. “Democracy, Veto Players and the Depth of Regional Integration.” The World Economy 31(1): 67–­96. Mansfield, E., H. Milner, and B. P. Rosendorff. 2000. “Free to Trade: Democracies, Autocracies and International Trade Negotiations.” American Political Science Review 94(2): 305–­21. ————. 2002. “Why Democracies Cooperate More: Electoral Control and International Trade Agreements.” International Organization 56(3): 477–­513. Mansfield, E. D., and M. Busch. 1995. “The Political Economy of Non-­tariff Barriers: A Cross-­National Analysis.” International Organization 49: 723–­49. Mansfield, E. D., and J. Snyder. 1995. “Democratization and the Danger of War.” Inter­ national Security 20(1): 1–­38. ——— —. 1996. “The Effects of  Democratization on War.” International Security 20(4): 196–­207. ————. 2005. “Democratic Transitions, Institutional Strength, and War.” International Organization 56(2): 297–­337. Mansfield, Edward D., Helen V. Milner, and Jon C. Pevehouse. 2007. “Vetoing Cooperation: The Impact of  Veto Players on Preferential Trading Arrangements.” British Journal of Political Science 37(3): 403–­32. Mansfield, Edward D., and Diana C. Mutz. 2009. “Support for Free Trade: Self-­Interest, Sociotropic Politics, and Out-­ Group Anxiety.” International Organization 63(3): 425–­57.

300 / References Mansfield, Edward D., and Jack Snyder. 2002. “Incomplete Democratization and the Outbreak of Military Disputes.” International Studies Quarterly 46(4): 529–­49. Marconini, M. 2005. “Brazil.” In The Political Economy of Trade Reforms in Emerging Markets—­Crisis or Opportunity?, edited by Razeen Sally, Peter Draper, and Phil Alves. London: Edward Elgar Publishing. Marshall, M. G., and K. Jaggers. 2002. Polity IV Project: Political Regime Characteristics and Transitions, 1800–­2002. College Park, MD: Center for International Development and Conflict Management, University of Maryland. Matschke, Xenia. 2004. “Labor Market Rigidities and the Political Economy of  Trade Pro­ tection.” Santa Cruz Center for International Economics, Working Paper. ————. 2006. “Do Labor Market Imperfections Increase Trade Protection? A Theoretical Investigation.” Working Paper 2006-­12, University of Connecticut, Department of Economics. Mayda A. M., and D. Rodrik. 2005. “Why Are Some People (and Countries) More Protectionist than Others?” European Economic Review 49(6): 1393–­1430. Mayer, Wolfgang. 1984. “Endogenous Tariff Formation.” American Economic Review 74: 970–­85. Mayer, Wolfgang, and Raymond Riezman. 1989. “Tariff Formation in a Multidimensional Voting Model.” Economics and Politics 1(1): 61–­79. Mazumdar, Surajit. 2008. “Crony Capitalism and India: Before and After Liberalization.” Unpublished manuscript, Institute for Studies in Industrial Development, New Delhi. Mazumder, Rabin. 2008. “Trade Liberalization and WTO: Impact on Developing Coun­ tries.” ICFAI Business School Research Centre-­Kolkata, Working Paper Series. Mbaya, Kennedy. 2004. “The Use of State Resources during Elections in South Africa.” In The Politics of State Resources: Party Funding in South Africa, edited by Khabele Matlosa. Johannesburg: Konrad-­Adenauer-­Stiftung Occasional Papers. McGillivray, F., and A Smith. 2004. “The Impact of Leadership Turnover on Trading Rela­ tions between States.” International Organization 58: 567–­600. McGillivray, Fiona. 1997. “Party Discipline as a Determinant of the Endogenous Formation of Tariffs.” American Journal of Political Science 41(2): 584–­607. ————. 2004. Privileging Industry: The Comparative Politics of Trade and Industrial Policy. Princeton, NJ: Princeton University Press. Medalla, E. 2002. “Fiscal Incentives Revisited.” Philippine Journal of Development 29(2): 1–­26. Mehta, Rajesh. 1999. Tariff and Non-­tariff Barriers of Indian Economy: A Profile. New Delhi: Non-­aligned and Other Developing Countries. Mello, Flavia de Campos. 1992. “O Brasil e o GATT: Análise da Posição Brasileira nas Negociações Comerciais Multilaterais.” Dissertação de Mestrado, IRI/PUC-­Rio. Millard, Frances. 2000. “Presidents and Democratization in Poland: The Roles of Lech Walesa and Aleksander Kwasniewski in Building a New Polity.” Journal of Communist Studies and Transition Politics 16(3): 39–­62. Miller, V. 1994. “NGOs and Grassroots Policy Influence: What Is Success?” IDR Reports 11(5): 2–­24. Milner, C., and P. Wright. 1998. “Modeling Labour Market Adjustment to Trade Liber­ alization in an Industrializing Economy.” Economic Journal 108. Milner, H., and K. Kubota. 2001. “Why the Rush to Free Trade? Democracy and Trade Policy in the Developing Countries.” Paper prepared to the annual meeting of the American Political Science Association, San Francisco.

References / 301 ————. 2005. “Why the Move to Free Trade? Democracy and Trade Policy in the Developing Countries.” International Organization 59: 107–­43. Milner, H. V., L. Andonova, and E. Mansfield. 2007. “International Trade and Environ­ mental Policy in the Postcommunist World.” Comparative Political Studies 40(7): 782–­807. Milner, H. V., and B. Judkins. 2004. “Partisanship, Trade Policy, and Globalization: Is There a Left-­Right Divide on Trade Policy?” International Studies Quarterly 48(1): 95–­120. Milner, H. V., E. Mansfield, and J. Pevehouse. 2006. “Vetoing Cooperation: The Impact of Veto Players on International Trade Agreements.” British Journal of Political Science 36(4): 403–­32. Milner, H. V., and B. Mukherjee. 2010. “Democracy and Skill-­Bias in Trade Protection in Developing Countries.” Working Paper, Department of  Politics, Princeton University. Milner, Helen. 1997. Interests, Institutions, and Information: Domestic Politics and International Relations. Princeton, NJ: Princeton University Press. Milner, Helen, and K. Kubota. 2001. “Why the Rush to Free Trade? Democracy and Trade Policy in the Developing Countries.” Working Paper, Columbia University. Milner, Helen, and Bumba Mukherjee. 2009. “Democratization and Economic Global­ ization.” Annual Review of Political Science 12: 163–­81. Ministério da Fazenda. Various years. IPEADATA. Instituto de Pesquisa e Planejamento. Brasilia: Government of Brazil. Ministério do Desenvolvimento. Various years. Indústria e Comércio Exterior Balança Comercial Brasileira. Brasilia: Government of Brazil. Ministério do Desenvolvimento Indústria eComércio. Various years. Ministério do Desenvolvimento/Indústria eComércio Exterior (SECEX/MDIC) Database. Brasilia: Government of Brazil. Ministry of Finance. 1995. Report of Task Force on Progress of Economic Reforms in India. New Delhi: Government of India. Mitra, Devashish. 1999. “Endogenous Lobby Formation and Endogenous Protection: A Long-­Run Model of Trade Policy Determination.” American Economic Review 89(5): 1116–­34. Molina, I., and F. Lehoucq. 1999. “Political Competition and Electoral Fraud: A Latin American Case Study.” Journal of Interdisciplinary History 30(2): 199–­234. Montinola, Gabriella R. 1999. “Parties and Accountability in the Philippines.” Journal of Democracy 10(1): 126–­40. Moreira, Mauricio M., and Paulo Guilherme Correa. 1997. “Abertura comercial e indústria: O que se pode esperar e o que se vem obtendo.” Revista de economía política 17(2): 61–­91. ————. 1998. “A First Look at the Impacts of  Trade Liberalization on Brazilian Manu­ facturing Industry.” World Development 26(10): 1859–­74. Moreira, Mauricio Mesquita, and Sheila Najberg. 2000. “Trade Liberalisation in Brazil: Creating or Exporting Jobs?” Journal of Development Studies 36: 78–­99. Moscarini, G. 2001. “Excess Worker Reallocation.” Review of Economic Studies 68(3): 593–­612. Moscarini, G., and F. Vella. 2002. “Aggregate Worker Relocation and Occupational Mobility in the United States: 1971–­2000.” IFS Working Papers W02/18, Institute for Fiscal Studies Institute. ————. 2003. “Occupational Mobility and Employment Reallocation: Evidence from the NLSY79.” Unpublished manuscript, Yale University.

302 / References Muendler, M. A. 1997 “Trade and Workforce Changeover in Brazil.” NBER Working Paper No. 12980, National Bureau of Economic Research, Harvard University, Cambridge, MA. ————. 2002. Trade, Technology and Productivity: A Study of Brazilian Manufactures 1986–­ 1988. Berkeley: University of California. Muendler, Marc, Luis Serven, and Claudia Sepulveda. 2001. “Productivity Growth in Brazilian Industry.” In Brazil: The New Growth Agenda, edited by Mark R. Thomas and Indermit Gill, vol. 2, 50–­83. Washington, DC: World Bank. Mukherjee, B., D. Smith, and Q. Li. 2009. “Labor (Im)mobility and the Politics of  Trade Protection in Majoritarian Democracies.” The Journal of Politics 71(1): 1–­18. Mussa, M. 1982. “Imperfect Factor Mobility and the Distribution of Income.” Journal of International Economics 12(1): 125–­41. Narlikar, Amrita, and Diana Tussie. 2004. “The G20 at the Cancun Ministerial: Devel­ oping Countries and their Evolving Coalitions in the WTO.” World Economy 27(7): 947–­66. Narlikar, Amrita, and Rorden Wilkinson. 2004. “Collapse at the WTO: A Cancun Post-­ mortem.” Third World Quarterly 25(3): 447–­60. National Council of Applied Economic Research. 1999. Report on Trade Protection in India. New Delhi: NCAER. Nelson, Daniel N. 1993. “Democracy, Markets and Security in Eastern Europe.” Survival 35(2): 156–­71. Nelson, Joan. 1995. “Linkage between Politics and Economics.” In Economic Reform and Democracy, edited by Larry Diamond and Marc F. Plattner, 44–­58. Baltimore: Johns Hopkins University Press. Nicita, Alessandro, and Marcelo Olarreaga. 2001. “Trade and Production, 1976–­99.” Policy Research Working Paper Series 2701, World Bank, Washington, DC. Nicolau, Ma R. 1998. “Un santuario ierorromano saguntino situado en la Montana Frontera (Sagunto, Valencia).” ACC 9: 25–­49. Nielson, Daniel L. 2003. “Supplying Trade Reform: Political Institutions and Liberalization in Middle-­Income Presidential Democracies.” American Journal of Political Science 47(3): 470–­91. Norris, Pippa. 2002. “Ballots not Bullets: Testing Consociational Theories of Ethnic Conflict, Electoral Systems, and Democratization.” In The Architecture of Democracy: Constitutional Design, Conflict Management, and Democracy, edited by Andrew Reynolds. Oxford: Oxford University Press. Nowzad, B. 1978. The Rise in Protectionism. Washington, DC: International Monetary Fund. Núñez, Clara Eugenia. 2003. “Within the European Periphery: Education and Labor Mobility in Twentieth-­Century Spain.” Paedagogica Historica 39(5): 621–­49. O’Donnell, G. 1992. “Transitions, Continuities, and Paradoxes.” In Issues in Democratic Consolidation: The New South American Democracies in Comparative Perspective, edited by Scott Mainwaring and G. O’Donnell. Notre Dame: Notre Dame Press. O’Donnell, G., and P. Schmitter. 1986. Transitions from Authoritarian Rule: Tentative Con­ clusions about Uncertain Democracies. Baltimore: Johns Hopkins University Press. O’Rourke, K., and A. Taylor. 2006. “Democracy and Protectionism.” Working Paper, Department of Economics, University of  California at Davis. O’Rourke, K. H., A. Rahman, and A. M. Taylor. 2007. “Trade, Knowledge and the Industrial Revolution.” NBER Working Paper No. 13057, National Bureau of  Economic Research, Harvard University, Cambridge, MA.

References / 303 O’Rourke, K. H., and R. Sinnott. 2001. “What Determines Attitudes toward Protection? Some Cross-­Country Evidence.” In Brookings Trade Forum 2001, edited by S. M. Collins and D. Rodrik, 157–­206. Washington, DC: Brookings Institute Press. OECD. 1996. Indicators of Tariff and Non-­tariff Barriers. Washington, DC: Author. Oliveira, Glauco Avelino Sampaio. 2009. “The Political Economy of Brazilian Trade Policy: On Trade, 1950–­1994.” World Politics 57(July): 453–­78. Ordeshook, P., and O. Shvetsova. 1994. “Ethnic Heterogeneity, District Magnitude, and the Number of Parties.” American Journal of Political Science 38: 100–­123. Ozden, Caglar, and Eric Reinhardt. 2005. “The Perversity of Preferences: GSP and Developing Country Trade Policies, 1976–­2000.” Journal of Development Economics 78(1): 1–­21. Pacek, A., and B. Radcliff. 1995. “The Political Economy of Competitive Elections in the Developing World.” American Journal of Political Science 57: 44–­46. Pahre, R., Fiona McGillivray, Iain McLean, and Cheryl Schonhardt-­Bailey. 2001. Inter­ national Trade and Political Institutions: Instituting Trade in the Long Nineteenth Century. London: Edward Elgar Publishing. Panagariya, Arvind. 1992. “Input Tariffs, Duty Drawbacks, and Tariff Reforms.” Journal of International Economics 32(1–­2): 131–­47. ————. 2008. India: The Emerging Giant. Oxford: Oxford University Press. Pang, Eul-­Soo. 1989. “Debt, Adjustment, and Democratic Cacophony in Brazil.” In Debt and Democracy in Latin America, edited by Barbara Stallings and Robert Kaufman, 127–­ 42. Boulder, CO: Westview Press. Papageorgiou, D., A. M. Choksi, and M. Michaely. 1992. Liberalizing Foreign Trade in Developing Countries: The Lessons of Experience. Washington, DC: World Bank Publications. Park, J. H., and N. Jensen. 2007. “Electoral Competition and Agricultural Support in OECD Countries.” American Journal of Political Science 51(2): 314–­29. Pavcnik, N., A. Blom, P. Goldberg, and N. Schady. 2004. “Trade Liberalization and Industry Wage Structure: Evidence from Brazil.” The World Bank Economic Review 18(3): 319–­44. Payne, Leigh A. 1991. “Working Class Strategies in the Transition to Democracy in Brazil.” Comparative Politics 23(2): 221–­38. ————. 1992. “Brazilian Business and the Democratic Transition: New Attitudes and Influence.” Kellogg Institute of International Studies Working Paper No. 179, Uni­ versity of Notre Dame. ————. 1994. Brazilian Industrialists and Democratic Change. Baltimore: Johns Hopkins University Press. Pearson, Kathryn L. 2008. “Party Loyalty and Discipline in the Individualistic Senate: Why Not Parties?” In Party Effects in the United States Senate, edited by Nathan Monroe, David Rohde, and Jason Roberts. Chicago: Chicago University Press. Pedersen, Jorgen Dige. 2000. “Explaining Economic Liberalization in India: State and Society Perspectives.” World Development 28(2): 265–­82. Pereira, Lia V. 2004. “Brazil Trade Liberalization Program.” Working Paper, Brasilia, Performance and Trade Policy, CID Working Paper No. 135, Centre for International Development, Harvard University. Phongpaichit, P., and C. Baker. 1996. Thailand’s Boom! Chiang Mai, Thailand: Silkworm Books. Pio, Carlos. 2000. “Liberalização do Comércio: Padrões de Interação entre Elites Burocráticas e Atores Sociais.” Retrieved from http://www.empreende.org.br/pdf /Estado/Padr%C3%B5es%20de%20Intera%C3%A7%C3%A3o%20entre%20 Elites%20Burocr%C3%A1ticas%20e%20Atores%20Soci.pdf.

304 / References Planning Commission. 1983. Report of the High Level Committee on Trade and Capital Account Policy. New Delhi: Government of  India. Plasser, F., and G. Plasser. 2002. Global Political Campaigning: A Worldwide Analysis of Campaign Professionals and Their Practices. Westport, CT: Praeger. Porto, G. G. 2004. “Informal Export Barriers and Poverty.” World Bank Policy Research Working Paper No. 3354, Washington, DC. Pottie, D., and T. Lodge. 2000. Electoral Management in Southern Africa. Johannesburg: EISA. Powell, G. Bingham. 1982. Contemporary Democracies: Participation, Stability and. Violence. Cambridge: Harvard University Press. Power, Timothy J. 1998. “Brazilian Politicians and Neoliberalism: Mapping Support for the Cardoso Reforms, 1995–­1997.” Journal of Interamerican Studies and World Affairs 40(4): 51–­72. Pritchett, L. 2006. Let Their People Come: Breaking the Gridlock on Global Labor Mobility. Washington, DC: Center for Global Development. Prusa, Thomas J. 2005. “Anti-­dumping: A Growing Problem in International Trade.” The World Economy 28(5): 683–­700. Przeworski, A., M. E. Alvarez, J. A. Cheibub, and F. Limongi. 2000. Democracy and Devel­ opment: Political Institutions and Well-­Being in the World, 1950–­90. New York: Cambridge University Press. Przeworski, Adam. 1991. Democracy and the Market: Political and Economic Reforms in East­ ern Europe and Latin America. New York: Cambridge University Press. Putnam, Robert D. 1988. “Diplomacy and Domestic Politics: The Logic of  Two-­Level Games.” International Organization 42(3): 427–­60. Quinn, Dennis P. 2001. “Political and International Financial Liberalization.” Paper Presented to the International Monetary Fund, Washington, DC. Ragin, Charles C. 1987. The Comparative Method. Berkeley: University of  California Press. Rajan, R., and R. Sen. 2002. “A Decade of  Trade Reforms in India: How It Compares with East Asia.” World Economics 3(4): 1–­14. Rama, M. 2003. “Globalization and Workers in Developing Countries.” Policy Research Working Paper 2958, World Bank, Washington, DC. Rama, M., and Tabellini, G. 1998. “Lobbying by Capital and Labor over Trade and Labor Market Policies.” European Economic Review 42: 1295–­316. Rama, Martin. 1994. “The Labor Market and Trade Reform in Manufacturing.” In The Effects of Protectionism on a Small Country: The Case of Uruguay, edited by M. Connolly and J. de Melo. Washington, DC: World Bank Regional and Sectoral Studies. Reidl, T. 2006. “The Dominant Party and Democratic Consolidation: South Africa’s ANC and India’s Congress Party’ Compared.” Seminar paper presented at the University of Cape Town, South Africa. Relação Anual de Informações Sociais. 1987. Features of the Second Formal Employment. Brasília: Ministério do Trabalho e Emprego, Government of Brazil. Revenga, A. 1997. “Employment and Wage Effects of  Trade Liberalization: The Case of Mexican Manufacturing.” Journal of Labor Economics 15(3): S20–­S43. Ribeiro, E. R., Carlos H. Corseuil, Daniel Santos, Paulo Furtado, Brunu Amorim, Luciana Servo, and André Souza. 2003. “Trade Liberalization, the Exchange Rate and Job Flows in Brazil.” Retrieved from http://www.iadb.org/res/files/RIBEIRO%20et%20al.pdf. Rickard, S. J. 2005. “The Costs of Risk: Examining the Missing Link between Globalization and Social Spending.” Working Paper, London School of  Economics.

References / 305 Rickard, Stephanie. 2002. “Asset Specificity, Domestic Institutions and Policy Choice: Explaining the Form of Compensatory Policies.” PhD dissertation, Department of Political Science, University of California at San Diego. Rigger, Shelley. 1999. Politics in Taiwan. New York: Routledge. Rigobon, R., and D. Rodrik. 2005. “Rule of Law, Democracy, Openness and Income: Es­ timating the Interrelationships.” Economics of Transition 13(3): 533–­64. Rius, Andrés, and Nicolas Van de Walle. 2003. “Political Institutions and Economic Policy Reform.” Unpublished manuscript. Rochon, Thomas R., and Michael J. Mitchell. 1989. “Social Bases of the Transition to Democracy in Brazil.” Comparative Politics 21(3): 307–­22. Rock, M. T. 1994. “Transitional Democracies and the Shift to Export-­Led Industrialization: Lessons from Thailand.” Studies in Comparative International Development 29(1): 18–­37. ————. 2007. “Corruption and Democracy.” DESA Working Paper 55. Rodrigues-­Filho, J., Cynthia J. Alexander, and Luciano C. Batista. 2006. “E-­Voting in Brazil: The Risks to Democracy.” Working Paper, Federal University of  Paraiba, Paraiba, Brazil. Rodríguez, F., and D. Rodrik. 2001. “Trade Policy and Economic Growth: A Skeptic’s Guide to the Cross-­National Evidence.” In NBER Macroeconomics Annual 2000, edited by Ben S. Bernanke and Kenneth Rogoff. Cambridge, MA: National Bureau of Economic Research. Rodrik, D., and R. Fernandez. 1993. “Why Is Trade Reform So Unpopular?” Discussion Paper No. 391, Centre for Economic Policy Research, London. Rodrik, D., and R. Wacziarg. 2005. “Do Democratic Transitions Produce Bad Economic Outcomes?” American Economic Review 95(2): 50–­55. Rodrik, Dani. 1988. “Imperfect Competition, Scale Economies, and Trade Policy in Developing Countries.” In Trade Policy Issues and Empirical Analysis, edited by Robert E. Baldwin. Chicago: University of  Chicago Press. ——— —. 1995. “Political Economy of  Trade Policy.” In Handbook of International Economics, vol. 3, edited by G. Grossman and K. Rogoff, 1457–­94. Amsterdam: North-­Holland. ————. 1997. Has Globalization Gone Too Far? Washington, DC: Institute for International Economics. ————. 1999. The New Global Economy and Developing Countries: Making Openness Work. Washington, DC: Overseas Development Council. Rogowski, R. (ed.). 1995. Comparative Politics and the International Political Economy. Vol. 8. London: Edward Elgar Publishing. Rogowski, Ronald. 1987. “Trade and the Variety of  Democratic Institutions.” International Organization 41(2): 203–­23. ————. 1989. Commerce and Coalitions: How Trade Affects Domestic Political Alignments. Princeton NJ: Princeton University Press. Rose, Andrew K. 2004. “Do We Really Know That the WTO Increases Trade?” American Economic Review 94(1): 98–­114. Rudolph, Lloyd I., and Susanne Hoeber Rudolph. 1977. “India’s Election: Backing into the Future.” Foreign Affairs 55(4): 836–­53. Rudra, Nita. 2008. Globalization and the Race to the Bottom in Developing Countries: Who Gets Hurt? Cambridge: Cambridge University Press. Sachikonye, L. 2000. “The Functioning and Funding of Political Parties (in the Sadc Region): Towards Sustainable Democratic Institutions in Southern Africa.” Paper presented at the International IDEA-­SADC Conference in Southern Africa, Gaborone, Botswana, May 8–­10.

306 / References Sachs, J. 1996. “Reforms in Eastern Europe and the Former Soviet Union in Light of the East Asian Experiences.” NBER Working Paper No. 5404, National Bureau of  Economic Research, Harvard University, Cambridge, MA. Sachs, J. D., and A. Warner 1995. “Economic Reform and the Process of  Global Integration.” Brookings Papers on Economic Activity 1: 1–­118. Sachs, Jeffrey D., Ashutosh Varshney, and Nirupam Bajpai (eds.). 1991. India in the Era of Economic Reforms. New Delhi: Oxford University Press. Sadie, Y. 1998. “Political Parties and Interest Groups.” In Government and Politics in the New South Africa, edited by A. Venter. Pretoria, South Africa: JL van Schaik Publishers. Sally, R. 2008. “The Political Economy of  Trade Policy Reform.” The Journal of International Trade and Diplomacy 2(2): 55–­96. Samuel, Vinay. 2005. “Business and Corruption.” Transformation: An International Journal of Holistic Mission Studies 12(1): 23–­27. Samuels, David. 2001. “Does Money Matter? Campaign Finance in Newly Democratic Countries: Theory and Evidence from Brazil.” Comparative Politics 34(3): 23–­42. ——— —. 2000 “Ambition and Competition: Explaining Turnover in the Brazilian Chamber of Deputies.” Legislative Studies Quarterly 25(3): 481–­97. ————. 2002. “Pork Barreling Is Not Credit Claiming or Advertising: Campaign Finance and the Sources of the Personal Vote in Brazil.” Journal of Politics 64(3): 845–­63. Samuels, David Julian. 1989. “Careerism and Its Consequences: Federalism, Elections, and Policy-­Making in Brazil.” PhD dissertation, University of California, San Diego. Sandberg, E. 2004. “South Africa’s Evolving Party System and the Dynamics of ANC Predominance” Paper presented at the annual meeting of the International Studies Association, Montreal, Canada. Scarrow, S. E. 2005. Political Parties and Democracy in Theoretical and Practical Perspectives: Implementing Intra-­party Democracy. Washington, DC: National Democratic Institute for International Affairs. Schaffer, Charles. 2002. “Disciplinary Reactions: Alienation and the Reform of  Vote Buying in the Philippines.” Paper presented at Trading Political Rights: The Comparative Pol­ itics of  Vote Buying Conference, Massachusetts Institute of  Technology. Schaffer, F. 2003. “Clean Elections and the Great Unwashed: Vote Buying and Voter Education in the Philippines.” Unpublished paper, Massachusetts Institute of Technology. Schaffer, F. C. 2002. “Might Cleaning Up Elections Keep People Away from the Polls? Historical and Comparative Perspectives.” International Political Science Review 23(1): 69–­84. ——— —. 2004. “Vote Buying in East Asia.” Unpublished manuscript, Cambridge Center for International Studies, Massachusetts Institute of  Technology. Schaffer, Frederic Charles, and Andreas Schedler. 2007. “What Is Vote Buying?” In Elections for Sale: The Causes and Consequences of Vote Buying, edited by Frederic Charles Shaffer, 24–­47. Schedler, Andreas. 2002. “Elections without Democracy: The Menu of Manipulation.” Journal of Democracy 13(2): 36–­50. ————. 2009. “The New Institutionalism in the Study of Authoritarian Regimes.” Paper presented at the 2009 Meetings of  the American Political Association, Toronto. Scheve, K., and M. Slaughter. 2001. “What Determines Trade Policy Preferences?” Journal Of International Economics 54(2): 267–­92. Schneider, B. R. (ed.). 1997. Business and the State in Developing Countries. Ithaca, NY: Cornell.

References / 307 Schneider, R. M. 1991. Order and Progress: A Political History of Brazil. Boulder, CO: Westview School of Economics and Political Science, University of Sydney. Shah, Anwar, and Mark Schacter. 2004. “Combating Corruption: Look before You Leap.” Finance and Development 41(4): 40–­43. Shapiro, Robert Y., and Benjamin Page. 1994. “Foreign Policy and Public Opinion.” In The New Politics of American Foreign Policy, edited by David Deese. New York: St. Martins. Shin, Jae H. 2010. “Why Personalistic Parties? The Choice of Candidate-­Centered Elec­ toral Systems in New Democracies.” Working Paper, Department of Political Science, UCLA. Shin, Myungsoon, Youngjae Jina, Donald A. Grossb, and Kihong Eom. 2005. “Money Matters in Party-­Centered Politics: Campaign Spending in Korean Congressional Elections.” Electoral Studies 24(1): 85–­101. Shugart, M., and S. Haggard. 1997. Institutions and Public Policy in Presidential Systems. San Diego: University of California at San Diego Press. Shugart. M. S., and M. P. Wattenberg. 2003. Mixed-­Member Electoral Systems: The Best of Both Worlds? Oxford: Oxford University Press. Shugart, Matthew S., and John M. Carey. 1992. Presidents and Assemblies: Constitutional Design and Electoral Dynamics. Cambridge: Cambridge University Press. Shugart, Matthew S., and John M. Carey. 1992. Presidents and Assemblies. New York: Cam­ bridge University Press. Shugart, Matthew S., Melody E. Valdini, and K. Suominen. 2005. “Looking for Locals: Voter Information Demands and Personal Vote-­Earning Attributes of  Legislators under Proportional Representation.” American Journal of Political Science 49(2): 437–­49. Shugart, Matthew Soberg. 2001. “Electoral ‘Efficiency’ and the Move to Mixed-­Member Systems.” Electoral Studies 20(2): 173–­93. Simmons, B. A., and Z. Elkins. 2004. “The Globalization of Liberalization: Policy Dif­ fusion in the International Political Economy.” American Political Science Review 98(1): 171–­89. Simpser, Alberto, and Daniela Donno. 2012. “Can International Election Monitoring Harm Governance?” The Journal of Politics 74(2): 501–­13. Sinha, Aseema. 2005. The Regional Roots of Developmental Politics in India: A Divided Le­ viathan. Bloomington: Indiana University Press. ————. 2007. “Global Linkages and Domestic Politics: Trade Reform and Institution Building in India in Comparative Perspective.” Comparative Political Studies 40(10): 1183–­210. South African Reserve Bank Quarterly Bulletin. Various years. Retrieved from https://www .resbank.co.za/Publications/QuarterlyBulletins/Pages/QuarterlyBulletins-Home.aspx. Southall, R. 1999. “South Africa 1999: The ANC and Democratic Consolidation.” A Jour­ nal of Opinion 27(2): 9–­16. ————. 2004. “The ANC and Black Capitalism in South Africa.” Review of African Political Economy 31(100): 313–­28. Spiezia, V. 2002. “Geographic Concentration of  Production and Unemployment in OECD Countries.” Cities and Regions, International Statistical Institute Journal December: 1–­37. Spilimbergo, Antonio, Juan Luis Londoño, and Miguel Székely. 1999. “Income Distri­ bution, Factor Endowments, and Trade Openness.” Journal of Development Economics 59(1): 77–­101. Sridharan, E. 1999. “Principles, Power and Coalition Politics in India: Lessons from Theory, Comparison and Recent History.” In Principles, Power and Politics, edited by D. D. Khanna and Gert W. Kueck, 270–­90. Delhi: Macmillan.

308 / References ————. 2009. “Party System Fragmentation, Intra-­party Democracy, and Opaque Political Finance.” Working Paper, Center for the Advanced Study of India, University of Pennsylvania. Stepan, Alfred C. (eds.). 1989. Democratizing Brazil: Problems of Transition and Consolidation. New York: Oxford University Press. Strydom, P. D. F. 1995. “International Trade and Economic Growth: The Opening-­Up of the South African Economy.” South African Journal of Economics 63(4): 306–­316. Su, S., and S. Yen. 1996. “Microeconometric Models of Infrequently Purchased Goods: An Application to Household Pork Consumption.” Empirical Economics 21: 513–­33. Subramanian, A., and S. Wei. 2007. “The WTO Promotes Trade, Strongly but Unevenly.” Journal of International Economics 72(1): 151–­75. Taagepera, Rein, and Matthew Soberg Shugart. 1989. Seats and Votes: The Effects and Determinants of Electoral Systems. New Haven, CT: Yale University Press. Tangri, R., and R. Southall. 2008. “The Politics of Black Economic Empowerment in South Africa.” Journal of Southern African Studies 34(3): 699–­716. Tavares, Jose. 2008. “Trade, Factor Proportions, and Political Rights.” The Review of Economics and Statistics 90(1): 63–­168. Thames, F., and M. Edwards. 2006. “Differentiating Mixed-­Member Electoral Systems: Member Proportional Systems and Mixed Member Majoritarian and Mixed-­ Government Expenditures.” Comparative Political Studies 39(7): 905–­27. Thompson, L. 2001. A History of South Africa. 3rd edition. New Haven, CT: Yale University Press. Tomz, Michael, Judith Goldstein, and Douglas Rivers 2005. “Membership Has Its Privileges: The Impact of GATT on International Trade.” Working Paper, Stanford University. ——— —. 2007. “Do We Really Know That the WTO Increases Trade?” The American Economic Review 97(5): 2005–­18. Topalova, P. 2004. Trade Liberalization and Firm Productivity: The Case of India. ePub No. 4-­28, International Monetary Fund. Tornell, Aaron. 1998. “Reform from Within.” NBER Working Paper No. W6497, National Bureau of Economic Research, Harvard University, Cambridge, MA. Trefler, D. 1993. “Trade Liberalization and the Theory of Endogenous Protection.” Journal of Political Economy 101: 138–­60. Tribunal Superior Eleitoral (Brasil). 1995. “Resultados das Eleições de 1994.” Computer archives, TSE0 Secretaria de Controle Interno, Brasília. ————. 1999. “Sistema de Controle de Recursos Arrecados-­SISCRA” [Accrued Campaign Funds Control System]. Computer archives, TSE0Secretaria de Controle Interno, Brasília. ————. 2003. “Sistema de Controle de Recursos Arrecados-­SISCRA” [Accrued Campaign Funds Control System]. Computer archives, TSE0Secretaria de Controle Interno, Brasília. Tsikata, Yvonne M. 1999. “Liberalization and Trade Performance in South Africa.” Informal Discussion Paper on Aspects of the South African Economy No. 13, World Bank, Washington, DC. United Nations. Various years. National Account Statistics Database. New York: United Nations. ————. 1994. Directory of Import Regimes. New York: Author. ——— —. 2006. Trade Analysis and Information Systems (TRAINS) Database. New York: Author.

References / 309 United Nations Comtrade. Various years. Classification SITC Rev 3. New York: UN Department of Economic and Social Affairs. ————. Various years. International Merchandise Trade Statistics (IMTS). New York: UN Department of Economic and Social Affairs. United Nations Conference on Trade and Development (UNCTAD). Various years. Trade Analysis and Information Systems (TRAINS). Geneva: Author. ————. 1987. Handbook of Trade Control Measures of Developing Countries. New York: United Nations. ————. 1994. Directory of Import Regimes. Geneva: Author. ————. 2007. Trade and Development Report: United Nations Conference on Trade and De­ velopment. CD-­ROM. Geneva: Author. United Nations Industrial Development Organization. Various years. UNIDO Industrial Statistics Database: 3-­and 4-­Digit Level of ISIC Code. Vienna: Author. ————. 2007. INDSTAT-­2 base. Vienna: Author. ————. 2008. INDSTAT-­2 base. Vienna: Author. ————. 2009. Industrial Development Report. New York: United Nations. United States Department of State. 2015. “Background Country Notes.” Retrieved from http://www.state.gov/r/pa/ei/bgn/. United States Library of Congress. 2015. “Country Studies/Area Handbook Series.” Federal Research Division, US Library of Congress. Retrieved from http://memory.loc.gov/frd /cs/. Van der Merwe, C. G., and Jacques E. du Plessis. 2004. Introduction to the Law of South Africa. London: Kluwer International. Verdier, Daniel. 1998. “Democratization and Trade Liberalization in Industrial Capitalist Countries, 1830s to 1930s.” Comparative Studies in Society and History 40(4): 587–­608. Vogel, David. 2003. “The Hare and the Tortoise Revisited: The New Politics of Consumer and Environmental Protection in Europe.” British Journal of Political Science 33(4). Wacziarg, R., and J. Wallack. 2004. “Trade Liberalization and Intersectoral Labor Move­ ments.” Journal of International Economics 64(2): 411–­39. Wacziarg, R., and K. H. Welch. 2003. “Trade Liberalization and Growth: New Evidence.” NBER Working Paper No. 10152, National Bureau of Economic Research, Harvard University, Cambridge, MA. Waddington, Richard. (2005). “WTO Seeks Face-­Saving Pact to Keep Talks Moving.” Reuters, December 13. Wadhva, Charan D. 1994. Economic Reforms in India and the Market Economy. New Delhi: Allied Publishers. Waldmeir, Patti. 1997. Anatomy of a Miracle: The End of Apartheid. New York: Norton. Wallack, J. S., A. Gaviria, H. Panizza, and E. Stein. 2003. “Political Particularism around the World.” The World Bank Economic Review 17(1): 133–­43. Walton, John, and Charles Ragin. 1990. “Global and National Sources of  Political Protest: Third World Responses to the Debt Crisis.” American Sociological Review 55(6): 876–­90. Way, L. A., and Levitsky, S. 2006. “The Dynamics of Autocratic Coercion after the Cold War.” Communist and Post-­Communist Studies 39(3): 387–­410. Webb, Paul D., and Stephen White (eds.). 2007. Party Politics in New Democracies. Oxford, New York: Oxford University Press. Weyland, Kurt. 1996. Democracy without Equity. Pittsburgh: University of Pittsburgh Press. ————. 1997. “The Brazilian State in the New Democracy.” Journal of Interamerican Studies and World Affairs 39(4): 63–­94.

310 / References ————. 2000. “A Paradox of Success? Determinants of  Political Support for President Fu­ jimori.” International Studies Quarterly 44(3): 481–­502. ————. 2002. The Politics of Market Reform in Fragile Democracies: Argentina, Brazil, Peru, and Venezuela. Princeton, NJ: Princeton University Press. Winters, L. A. 2002. “Trade Liberalisation and Poverty: What Are the Links?” The World Economy 25(9): 1339–­68. Wood, Adrian. 1997. “Wage Inequality in Developing Countries: The Latin American Challenge to East Asian Conventional Wisdom.” World Bank Economic Review 11(January): 33–­57. Worden, Nigel. 2000. “Making of Modern South Africa: Conquest, Segregation and Apartheid.” Working Paper 11, Inter-­American Development Bank. World Bank. Various years. World Integrated Trade Solution (WITS) Package. Washington, DC: Author. ————. 1994. “Trade Policy Reform in Developing Countries since 1985: A Review of the Evidence.” World Bank Discussion Paper No. 267, World Bank, Washington, DC. ————. 1996. The Uruguay Round: Statistics on Tariff Concessions Given and Received. Washington, DC: Author. ————. 1998–­2000. World Development Indicators. CD-­ROM. Washington, DC: Author. ————. 2002. World Development Indicators. CD-­ROM. Washington, DC: Author. ————. 2003. World Development Indicators. CD-­ROM. Washington, DC: Author. ————. 2004. World Development Indicators. CD-­ROM. Washington, DC: Author. ————. 2005. World Development Indicators. CD-­ROM. Washington, DC: Author. ————. 2006a. The World Bank Trade and Production Database, 1976–­1999. Washington, DC: Author. ————. 2006b. World Development Indicators. CD-­ROM. Washington, DC: Author. ————. 2007a. World Development Indicators. CD-­ROM. Washington, DC: Author. ————. 2007b. World Integrated Trade Solution (WITS) Package. Washington, DC: Author. ————. 2008. World Development Indicators. CD-­ROM. Washington, DC: Author. ————. 2010. Database of Political Institutions. Washington, DC: Author. ————. 2011. World Development Indicators. Washington, DC: Author. ————. 2012. Database of Political Institutions. Washington, DC: Author. World Economic Forum. Various years. Global Competitiveness Report. Geneva: Author. ————. 2004a. Executive Opinion Database. CD-­ROM. Geneva: Author. ————. 2004b. Global Competitiveness Forum. CD-­ROM. Geneva: Author. World Trade Organization. Various years. Annual Reports. Committee on Safeguards. Ge­ neva: Author. ————. Various years. Trade Policy Mechanism Review for India. Geneva: Author. ————. Various years. Trade Policy Mechanism Review for South Africa. Geneva: Author. ————. Various years. Trade Policy Review Series. Geneva: Author. ————. 2007. Antidumping Statistics from Reporting Countries. Geneva: Author. ————. 1998. Integrated Data Base (IDB). CD-­ROM. Washington, DC: Author. Xu, B. 2001. “Endogenous Technology Bias, International Trade and Relative Wages.” Unpublished manuscript, University of Florida. Yadav, Vineeta. 2005. “Interest Group Dynamics in Developing Countries: An Empirical Comparison of  Brazil and India.” Working Paper, Yale University. ————. 2011. Political Parties, Business Groups, and Corruption in Developing Countries. New York: Oxford University Press. Yang, C. C. 1995. “Endogenous Tariff Formation under Representative Democracy: A Probabilistic Voting Model.” The American Economic Review 85(5): 956–­63.

References / 311 Yen, S., H. Jensen, and A. Wang. 1996. “Cholesterol Information and Egg Consumption in the US: A Non-­normal and Heteroscedastic Double Hurdle Model.” European Review of Agricultural Economics 23(3): 343–­56. Yu, M. 2007. Trade, Democracy, and the Gravity Equation. Peking: China Center for Economic Research (CCER), Peking University. Yu, Miaojie. 2006. “Trade Protectionism and Electoral Competition: An Empirical Investigation.” Retrieved from http://papers.ssrn.com/sol3/papers.cfm?abstract_id =935334. Yu, Z. 2000. “A Model Of Substitution Of Non-­tariff Barriers for Tariffs.” Canadian Journal of Economics 33(4): 1069–­90. Zahariadis, Nikolaos. 2001. “Asset Specificity and State Subsidies in Industrialized Countries.” International Studies Quarterly 45(4): 603–­16. Zlate, Andrei. 2002. “The Role of Safeguards in Global Trade Protectionism.” American Enterprise Institute (AEI) Working Paper.

I nd e x

accountable, 15, 47, 123 African National Congress, 206 AGC index, 73, 74, 87, 100, 102, 104, 136, 142, 144, 147, 264n22 agriculture products, 68, 164, 168, 192 Argentina, 2, 4, 7, 61, 64, 95, 131, 151, 257, 273n24, 279, 283, 287, 294, 310 asset specificity, 20, 29, 32, 38, 41, 53, 56, 61, 62, 127, 128, 239, 262n26, 262n42, 264n21, 269n37 associations, 121, 138

dampen, 20, 21, 183 demand-­side, 6, 8, 10, 13, 14, 15, 16, 108, 257n11 democratic consolidation, 266n58 democratic developing countries, 10, 108, 234, 240 democratic transition 3, 4, 6, 9, 17, 29, 40, 45, 57, 64, 67, 95, 96, 100, 105, 156, 157, 165, 183, 263n11 Dirceu, Jose, 175 divergence, 3, 19

balance of payments, 72, 196, 215 Bharatiya Janata Party, 198 Bombay Club, 197, 198 bop crisis, 72, 80, 84, 86, 101, 102, 104, 141, 144, 146, 199, 200, 264n18 Brasilia, 179, 283, 299, 303, 304 BRICS, 23 Buiter, Willem, 12, 13, 283

economic crisis, 72, 80, 84, 101, 102, 141, 144, 196, 197, 264n17 economic geography, 283 economic reforms, 1, 3, 167, 198, 214, 225, 277n2 elasticity, 70, 71, 72, 78, 80, 82, 85, 92, 132, 135, 136, 139, 141, 144, 145, 191, 211, 212 electoral malpractice, 1, 29, 31, 33, 35, 37, 39, 41, 43, 45, 47, 49, 51, 53, 55, 181 Executive Opinion Survey, 23, 94, 95, 151, 228

campaign contributions, 93, 96, 97, 109, 113 Cardoso de Melo, Zélia, 169 case study, 105, 155, 184, 185 causal logic, 49, 122, 157 China, 12, 23, 61, 68, 95, 239, 272n11 coalition, 13, 15, 42, 45, 119, 189, 238 Communist, 63, 64, 65, 169 Comtrade, 273n15, 273n25, 274n3, 276n21 consolidated democracy, 249 contemporaneous correlation, 74

factor endowment, 8, 14, 15, 73, 231, 258n14 factor mobility, 9, 10, 14, 17, 20, 25, 26, 28, 29, 31, 32, 33, 42, 52, 57, 106, 108, 110, 127 financial payments, 17, 21, 26, 108, 112, 120, 122, 127, 228 foreign exchange, 211, 275 Freedom House, 63, 67, 96, 151

314 / Index free trade, 1, 2, 15, 136, 137, 212, 295, 298, 299, 300, 301 Fujimori, Alberto, 121 geographic goncentration, 73, 74, 264n21 goods, 31, 34, 125, 157, 159, 164, 198, 214, 217, 242 granger causality test, 85, 145 gross school enrollment, 158 heavy industries, 49, 93 Heckman Selection model, 88, 148, 149, 265n40, 265n43 Heckscher-­Ohlin model, 31, 259n3, 260n5 high tariffs, 160, 206 house(s), 133, 134, 166, 193, 194, 195, 197, 202, 216, 217 IMF program, 80, 84, 86, 102, 104, 144, 272n1 import-­substitution-­industrialization, 24, 160, 161, 190, 210 influence peddling, 221 informal sector, 170 institutional context, 9, 17, 39, 57, 224, 229, 231, 258n19, 262n30 interaction term, 71, 72, 77, 79, 82, 96, 136, 139, 144, 149, 151 ISIC (International Standard Industrial Classification), 68, 69, 71, 172, 191, 192, 199, 210, 218, 273n16, 275n4 Janata-­led government, 195 judicial independence, 104 junta, 65, 173 Kenya, 7, 61, 95, 131, 151 key industries, 161, 178, 179 knowledge, 5, 34, 36, 52, 53, 154, 229, 255 labor force, 19, 21, 43, 165, 166, 171, 173, 188, 213, 237, 242, 264n23 legal contributions, 32, 36, 38, 50, 53, 56, 94, 108, 111, 120, 122, 204, 205, 223, 230, 256 legitimacy, 224 lobbying, 37, 38, 73, 117, 120, 125, 202, 233, 258n19 Lula Da Silva, Luiz Inácio, 159, 175

manufacturing industries, 169, 193, 195, 212, 269n37 manufacturing sector, 75, 137, 161, 202, 213, 265n49, 274n36 Markov transition model, 99, 104 maximizes, 21, 41, 42, 44, 45, 114, 115, 127, 227, 253 median voter, 17, 19, 21, 37, 41, 43, 44, 45, 47, 48, 49, 52, 54, 55, 61, 227, 228, 241, 242, 243, 246, 253 mobilization, 165, 283 Morocco, 61, 95 Mozambique, 4, 6, 95 multiparty elections, 65 multiparty system, 65 National Party, 156, 206 neoliberal reform, 167, 171, 251 new democracy, 18, 19, 24, 25, 26, 30, 35, 39, 42, 43, 44, 48, 49, 50, 51, 52, 54, 55, 65 newly democratized regime, 1, 2, 4, 5, 6, 9, 17, 25, 32, 34, 39, 42, 55, 74, 76, 77, 89, 92, 93, 226, 228, 229, 241, 248 nontariff barriers, 3, 5, 29, 226 occupational mobility, 20, 21, 22, 24, 25, 26, 29, 33, 35, 36, 46, 47, 48, 49, 55, 56, 57, 63, 68, 69, 70, 71, 94, 104, 105, 108, 116, 117, 118, 126, 127, 132, 134, 154, 163, 191 oligarchic, 115, 121 oligopolistic ownership, 161 owners, industry, 9, 10, 17, 18, 19, 22, 25, 26, 31, 40, 55, 59, 93, 157, 163, 168, 169, 171, 187, 188, 193, 194, 198, 199, 260n6, 260nn8–­9, 262n30, 262n42 parametric bootstraps, 96, 101, 102, 152 patronage, 290 Philippines, 3, 4, 12, 13, 30, 45, 49, 61, 66, 74, 75, 76, 95, 121, 257n5, 264n25 268n23, 268n27, 269n29 physical capital, 119, 261 Pinochet, Augusto, 64 political competition, 18, 29, 34, 35, 40, 42, 49, 126, 157, 207, 228, 262n41 predicted probability, 96, 97, 101, 102, 103, 152, 153, 266n54, 271n30 privatization, 121

Index / 315 probit model, 99, 101, 102, 104 professional and scientific equipment, 68, 192 profits, 42, 53, 177 qualitative, 26, 171, 176, 177, 181, 201, 203, 220, 223 quantitative, 62, 162, 171, 172, 188, 198, 199, 201, 209, 218, 263n4, 269n2, 276n21 quasi concavity, 245 quasi-­rents, 32, 87, 260n7 rank-­and-­file, 109, 112, 114, 118, 119, 123, 160, 189, 196, 207, 215, 216, 259n33, 272n4 rent maximizing tariff, 44 rents, 32, 52, 56, 124, 127, 211, 260n7 Ricardo-­Viner model, 13, 31 ruling party, 17, 19, 20, 94, 111, 112, 113, 114, 115, 117, 118, 119, 120, 163, 182, 183, 212, 215, 227, 251, 252, 253, 254, 262n41 sample size, 274 selection bias, 88, 145, 265n38 socialist economy, 99 South Korea, 95, 151 spatial dependence, 89, 148, 265n42 state-­owned firms, 32 statutory tariffs, 62, 63, 77, 78, 79, 80, 82–­88, 132–­49, 263n13 substantive effect, 79, 96, 101, 139, 143, 144, 152 supply-­side, 9, 11, 16, 105, 107, 257n12, 267n7 system-­generalized method of moments. See System GMM system-­GMM, 86, 87, 88, 145, 146, 148, 173, 174, 200, 218, 219, 265nn36–­37, 265n43, 271nn18–­19

Thailand, 4, 7, 61, 95, 121, 131, 151 time horizons, 52, 210 time trend, 83, 91, 93, 145, 173–­74, 199, 200, 218, 219 trade liberalization, v, 3, 4, 12, 15, 19, 20, 21, 26, 27, 31, 32, 33, 46, 51–­57, 59, 60, 89, 98, 99–­101, 102–­11, 155, 162, 183, 198, 209–­11, 277n3, 278n17 transparency, 224 TSCS (time-­series-­cross-­sectional) data, 59, 60 twin economic crisis, 196 UNCTAD, 60, 62 UNIDO, 68, 70, 74, 264n22, 273n16 unweighted, 162, 163, 164, 190, 208, 209, 217, 273n14, 274n2 utility, 29, 34, 35, 37, 38, 41, 44, 45, 53, 54, 108, 111, 113, 114, 115, 125, 184, 243, 245, 248, 250, 251, 253 virtue, 42, 46, 122 voter, 17, 19, 21, 35, 36, 37, 41, 44, 46, 204, 207, 215, 227, 228, 237, 241, 242, 243 wages, 33, 185 weighted, 39, 57, 112, 116, 162, 166, 191, 193, 243, 251 welfare spending, 85, 87, 145 World Bank, 125, 127, 129, 130 World Economic Forum, 23, 25, 26, 59, 93, 201, 228, 258n28, 266n50, 266n52 WTO (World Trade Organization), 14, 62, 73, 81, 136, 141, 144, 198, 232, 233, 234, 238 Yugoslavia, 65, 66 Zambia, 4, 7, 61, 67, 95, 131, 151 Zimbabwe, 61, 95