Understanding Terrorism : A Socio-Economic Perspective 9781783508280, 9781783508273

Citation preview

UNDERSTANDING TERRORISM: A SOCIO-ECONOMIC PERSPECTIVE

Contributions to

Conflict Management, Peace Economics and Development Volume 22 SERIES EDITOR

MANAS CHATTERJI BOOKS IN THE SERIES

Military Missions and Their Implications Reconsidered: The Aftermath of September 11th, edited by G. Caforio and G. Ku¨mmel Managing Conflict in Economic Convergence of Regions in Greater Europe, by F. Carluer Cultural Differences between the Military and Parent Society in Democratic Countries, edited by G. Caforio Conflict and Peace in South Asia, edited by M. Chatterji and B. M. Jain War, Peace, and Security, edited by J. Fontanel and M. Chatterji Armed Forces and Conflict Resolution, edited by G. Caforio, G. Ku¨mmel and B. Purkayastha Regional Development and Conflict Management: A Case for Brazil, by R. Bar-El Crisis, Complexity and Conflict, by I. J. Azis Putting Teeth in the Tiger: Improving the Effectiveness of Arms Embargoes, edited by M. Brzoska and G. A. Lopez Peace Science: Theory and Cases, by P. Gangopadhyay and M. Chatterji Advances in Military Sociology: Essays in Honor of Charles C. Moskos (Two Volume Set), edited by Giuseppe Caforio Arms and Conflict in the Middle East, by Riad A. Attar Economics of War and Peace: Economic, Legal, and Political Perspectives, edited by Benjamin E. Goldsmith and Jurgen Brauer Conflict, Complexity and Mathematical Social Science, by Gordon Burt Frontiers of Peace Economics and Peace Science, edited by Manas Chatterji, Chen Bo and Rameshwar Misra Ethnic Conflict, Civil War and Cost of Conflict, edited by Raul Caruso Governance, Development and Conflict, edited by Manas Chatterji, Darvesh Gopal and Savita Singh New Wars, New Militaries, New Soldiers? Conflicts, the Armed Forces and the Soldierly Subject, edited by Gerhard Ku¨mmel and Joseph Soeters Cooperation for a Peaceful and Sustainable World, Part 1, edited by Chen Bo, Manas Chatterji and Hao Chaoyan Cooperation for a Peaceful and Sustainable World, Part 2, edited by Li Junsheng, Chen Bo and Hou Na Nuclear Disarmament: Regional Perspectives on Progress, edited by P. M. Kamath

Contributions to Conflict Management, Peace Economics and Development Volume 22

UNDERSTANDING TERRORISM: A SOCIO-ECONOMIC PERSPECTIVE EDITED BY

RAUL CARUSO Institute of Economic Policy, Catholic University of the Sacred Heart, Milan, Italy

ANDREA LOCATELLI Department of Political Science, Catholic University of the Sacred Heart, Milan, Italy

United Kingdom
North America
Japan India
Malaysia
China

Emerald Group Publishing Limited Howard House, Wagon Lane, Bingley BD16 1WA, UK First edition 2014 Copyright r 2014 Emerald Group Publishing Limited Reprints and permission service Contact: [email protected] No part of this book may be reproduced, stored in a retrieval system, transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without either the prior written permission of the publisher or a licence permitting restricted copying issued in the UK by The Copyright Licensing Agency and in the USA by The Copyright Clearance Center. Any opinions expressed in the chapters are those of the authors. Whilst Emerald makes every effort to ensure the quality and accuracy of its content, Emerald makes no representation implied or otherwise, as to the chapters’ suitability and application and disclaims any warranties, express or implied, to their use. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN: 978-1-78350-827-3 ISSN: 1572-8323 (Series)

ISOQAR certified Management System, awarded to Emerald for adherence to Environmental standard ISO 14001:2004. Certificate Number 1985 ISO 14001

CONTENTS LIST OF CONTRIBUTORS

vii

FOREWORD

ix

INTRODUCTION

xi

CHAPTER 1 WHAT IS TERRORISM? CONCEPTS, DEFINITIONS AND CLASSIFICATIONS Andrea Locatelli

1

CHAPTER 2 ECONOMIC DETERMINANTS OF TERRORISM Daniel Meierrieks

25

CHAPTER 3 MODELING TERROR ATTACKS: A CROSS-NATIONAL, OUT-OF-SAMPLE STUDY Ryan Bakker, Daniel W. Hill, Jr. and Will H. Moore

51

CHAPTER 4 MEASURING SECURITY Tilman Bru¨ck, Olaf J. de Groot and Neil T. N. Ferguson

69

CHAPTER 5 MEASURING TERRORISM WITH THE GLOBAL TERRORISM INDEX Daniel Hyslop and Thomas Morgan

97

CHAPTER 6 TERRORISM AND ECONOMIC SENTIMENT IN EUROPEAN COUNTRIES Christos Kollias and Stephanos Papadamou

115

CHAPTER 7 TERRORISM AS “A POLITICAL WORLD”: IDENTITY, STRATEGY, VALUES Damiano Palano 135 v

vi

CONTENTS

CHAPTER 8 THE ECONOMIC ANALYSIS OF LONE WOLF TERRORISM Peter Phillips

159

CHAPTER 9 A COMPARATIVE ANALYSIS OF HOMEGROWN TERRORISM Kaisa Hinkkainen

193

CHAPTER 10 CLOSE TO THE EDGE: CYBERTERRORISM TODAY Giampiero Giacomello

217

CHAPTER 11 SOME INSIGHTS ON THE LINK BETWEEN TERRORISM, ORGANISED CRIME AND ‘NEW WARS’ Carla Monteleone, Raul Caruso and Andrea Locatelli

237

LIST OF CONTRIBUTORS Ryan Bakker

Department of Political Science, University of Georgia, Athens, GA, USA

Tilman Bru¨ck

Stockholm International Peace Research Institute (SIPRI) and Households in Conflict Network (HiCN), Stockholm, Sweden

Raul Caruso

Institute of Economic Policy, Catholic University of the Sacred Heart, Milan, Italy

Olaf J. de Groot

Economic Commission for Latin America and the Caribbean (ECLAC), Santiago, Chile

Neil T. N. Ferguson

Stockholm International Peace Research Institute (SIPRI), Stockholm, Sweden

Giampiero Giacomello

Department of Political and Social Studies, University of Bologna, Bologna, Italy

Daniel W. Hill, Jr.

Department of International Affairs, University of Georgia, Athens, GA, USA

Kaisa Hinkkainen

School of Social and Political Sciences, University of Lincoln, Lincoln, UK

Daniel Hyslop

Institute for Economics and Peace, Sydney, Australia

Christos Kollias

Department of Economics, University of Thessaly, Volos, Greece

Andrea Locatelli

Department of Political Science, Catholic University of the Sacred Heart, Milan, Italy

vii

viii

LIST OF CONTRIBUTORS

Daniel Meierrieks

Department of Economics, University of Freiburg, Freiburg, Germany

Carla Monteleone

Department of European Studies and International Integration (DEMS), University of Palermo, Palermo, Italy

Will H. Moore

Department of Political Science, Florida State University, Tallahassee, FL, USA

Thomas Morgan

Institute for Economics and Peace, Sydney, Australia

Damiano Palano

Department of Political Science, Catholic University of the Sacred Heart, Milan, Italy

Stephanos Papadamou

Department of Economics, University of Thessaly, Volos, Greece

Peter Phillips

Faculty of Business and Law, University of Southern Queensland, Toowoomba, Australia

FOREWORD The phenomenon of Terrorism is nothing new. It has long been used by contesting parties in many conflict situations in the past not only in political arenas but also in industrial disputes, environmental, social issues, matters related to psychological characteristics of the protester as in school shootings in the United States, etc. It is true that it has come into prominence after 9/11 and subsequent events. But it is just not a political issue. It encompasses all aspects of social and physical sciences. Terrorism and technological development, particularly in information sciences and energy sources, and especially nuclear energy, are very much related. So we need to develop a general theory involving socioeconomic, political, technological, and physical variables. This is not an easy task. It needs to be emphasized that terrorism linked with political ideology common in pre-cold war and colonial times is still present. The Naxalite movement in India is a case in point. It is also entangled with matters of globalization and foreign direct investment. In many countries, examples of terrorism particularly by minority and tribal groups against multinational corporations can be cited. But this book with contributions from distinguished scholars deals with political terrorism. It starts with an attempt to define terrorism and a terrorist acknowledging that no single definition exists. No clear-cut link between terrorism and religious extremism can be established. However, multiple views are beneficial to develop a general system based on rational choice theory. Next, contributors look for the motives for terrorism, its measurements, its varieties (foreign and home grown), forecasting, cyber terrorism, organized crime, etc. The subject matter of this volume is very focused to the theme of the book series. Manas Chatterji Series Editor

ix

INTRODUCTION Conventional wisdom holds that the best way to present a book is to get straight to the point and let the reader know what it is about. Here we will indulge on the reader’s patience and start by discussing what this book is not about
that is what they cannot find in the following pages. So, it is worth emphasizing from the very beginning that, while this book is about terrorism, it is not aimed at developing a novel comprehensive theory. However, it aims to contribute to the huge recent literature on the topic with a specific approach. Needless to say, the issue has attracted widespread attention in the past decade, leading scholars from different fields and approaches to converge on this phenomenon (for comprehensive reviews, see Jackson & Sinclair, 2012; Martin, 2011; Schmid, 2011). The literature that has emerged is huge, and valuable analyses (along with controversial contributions) have provided both theoretical and empirical data to tackle the issue (Sandler, 2011). Admittedly, many problems still remain unsolved
most importantly, what turns a man into a terrorist, what makes contemporary terrorism so resilient to counterterrorism and so forth (these issues are treated, among others, in Caruso & Locatelli, 2008; Caruso & Schneider, 2013). Yet, we believe that the need to account for this phenomenon led to unintended consequences for the scholarly community. For one thing, the research path offered by current events
or, to put it bluntly, the opportunity to study an item on top of most states’ security agenda
gave many an incentive to apply their expertise to the study of terrorism (Ranstorp, 2007). Indeed, 9/11 just increased a well-established trend: as an even cursory look at the main academic journals on terrorism reveals, in the 1990s on average 83% of the articles were written by one-timers (Silke, 2004, p. 190). Secondly, and equally important, after the New York and Washington bombing, al Qaeda rose at the top of the research agenda, so colonizing the whole field of terrorist studies (Silke, 2007, p. 84). Thirdly, and partially related to the latter point, a great deal of studies focused exclusively on religious terrorism, with the twofold result of: (a) raising confusing questions on the allegedly ‘new’ nature of contemporary terrorism (Crenshaw, 2007; Kurtulus, 2011); (b) implicitly (and often unconsciously) xi

xii

INTRODUCTION

circumscribing the phenomenon under investigation to its contextualized contemporary form, even if it is true that religious terrorism has significantly shaped terrorist patterns worldwide. In this respect, Enders and Sandler (2000) empirically demonstrate how religious fundamentalists perpetrate fewer but more violent attacks than secular terrorist groups. In sum, such trend has brought about new perspectives and original thinking and a large set of empirical studies. On the other hand, as witnessed by the massive increase in publications on the topic,1 the burst of interest for the issue led to the growth of parallel, separate and impermeable bodies of literature. As an evidence of this, in their 2002 survey of the literature, Weinberg, Pedahzur, and Hirsch-Hoefler (2004) note how in a 55-cases sample, authors from different fields tend to frame the very definition of terrorism in different ways. As a result, economists, social and political scientists have developed theories, evidence and, in a sense, even a peculiar jargon of their own. Borrowing Andrew Silke’s words, we believe that ‘The study of terrorism is truly multi-disciplinary and no one discipline has been able to firmly take pre-eminence in research in the area. Researchers from fields such as political science, criminology, psychology, sociology, history, law, the military and communication sciences have all contributed to it’ (Silke, 2004, p. 188). Starting from this assumption, we got to the conclusion that the literature could gain more clarity (and hopefully better insight) if scholars from different fields engaged in an open and lively confrontation. So, this is what the volume is about: it is an attempt to gather analysts with different expertise and background around the same table, showing how their individual perspectives can contribute to a broader understanding of the issue at stake. As the reader will discover in the following pages, each contributor to this volume has analysed either a peculiar or a less intensely studied feature of terrorism. We did not pit them against each other, with a view to let them compete for the more comprehensive and/or sound explanation. By doing so, we believe, we avoided making this volume a classical three-cornered fight as suggested by Lakatos (1970)
that is a clash of theories, paradigms or disciplines, arguing for the best account of a given fact/phenomenon. Of course, disagreements and competing views will sometimes emerge on the way. However, rather than considering these differences as a problem, we will take them as a starting point for metatheoretical considerations: to put it bluntly, instead of wondering what is the best approach, we let scholars develop their own approach. If the reader finds some room of disagreement between them, we believe that these

Introduction

xiii

disagreements may actually turn to be beneficial to increase our knowledge of terrorism. To elaborate on this point, we might resort to a metaphor. In his defence of neorealism, John Mearsheimer equated theory (neorealist theory, in that case) to a ‘powerful flashlight in a dark room’ (2001, p. 9): it provides the analyst with a powerful tool to find a way in the darkness, but it cannot reach every corner. And, one may add, the more stuffed is the room, the more corners will be left in shadow. This is particularly true for terrorism, since it empirically shows itself under a variety of forms: different actions, motivations and even differently organized agents. All this calls for more than just a focused beam of light. In a nutshell, the message we aim to convey in this book is quite simple: as Ariel Merari forcefully argued, terrorism has so many facets that developing a general theory is virtually impossible; on the other hand, it can be fruitfully approached by a variety of angles (quoted in Silke, 2004, p. 190). Needless to say, the two disciplines that will hold central place hereafter are economics and political science. As concerns political science, for instance, much has been written on the social and political conditions conducive to political violence and terrorism; on the role of ideologies and religion as a spring for terrorism; on the recruitment process and in-group dynamics within terrorist organisations (see, among others, Davis & Cragin, 2009; Rosenfeld, 2011; Haunstrup Qvortrup, 2012). Likewise, economics has shed light on economic correlates of terrorism to highlight the emergence or not of opportunity costs, but also on productivity, dynamics of terrorist attacks and strategic interactions between terrorist groups and governmental actors (see Bru¨ck, 2007; Enders & Sandler, 2006; Llussa & Tavares, 2011). However, there are many reasons why economics and political science often seem to part company. In particular, we would highlight two main causes: (a) the ontological vision of terrorists; (b) the nature of the explanatory variables selected. With the term ontological vision of terrorists we refer to the underlying model of agent rationality assumed by the analyst. In other words, what kind of logic drives terrorist actions? Are his/her interests to be treated as endogenous or exogenous? What is the relationship between goals and means? The most widespread answer to these questions comes from rational choice theory applied in economics. In fact, economic analyses of terrorism share the basic assumption that terrorists are rational decision makers, namely homini economici. They are utility maximizers, whose goals can be measured in terms of self-interest. Accordingly, in choosing among

xiv

INTRODUCTION

the available means (violent/non-violent action, target selection, etc.) they follow the logic of consequences
that is they will opt for the one that most likely will lead to the preferred outcome (see among others Shugart, 2011; Van Um, 2011; Caplan, 2006; Anderton & Carter, 2005; Azam, 2005). This is epitomized by the famous definition developed by Sandler, Tschirhart, and Cauley (1983, p. 37) as the ‘premeditated, threatened or actual use of force or violence to attain a political goal through fear, coercion, or intimidation’. By reverse, political scientists are divided on this point. While some still consider terrorist acts as a form of strategic behaviour (Kydd & Walter, 2006), so mimicking the idea of homo economicus, a growing body of literature focuses on different mechanisms: some authors, like Abrahms (2006), see terrorists as solidarity seekers, whose goals are not necessarily the ones promoted by the organisation. In this view, interests are endogenous, in permanent evolution and, most importantly, the actions performed are not necessarily motivated by the consequences that follow. While few authors, as far as we know, have explicitly made this point, we can contrast the logic of consequences with the logic of appropriateness (March & Olsen, 1989, 1995): according to this view, actors’ behaviour (and terrorists make no exception) does not follow what they please, but rather their identity, or what they believe they are expected to do. These two approaches can complement each other
the former making sense of organisational, the latter of individual behaviour (for an in-depth discussion of these issue, see Chapter 7 in this volume). Moving to the second point, we need to discuss the variables usually considered as determinants of terrorism. As mentioned, terrorism is such a multifaceted phenomenon that no single variable can be realistically considered as the cause. That said, it comes to the surprise of no one that economists and political scientists tend to identify in their analyses different classes of factors. In line with the ontology discussed before, the economic approach has a penchant for the economic determinants of terrorism. Political scientists, on the other hand, are less committed to the study of economic variables. Even when they rely on the assumption of consequential linearity, they hardly relate terrorism to particular economic conditions (like poverty, deprivation and so forth). At most, they consider the social consequences of these conditions, since they argue that poor societies have always existed and only some of them served as a cradle of terrorism. Most frequently, and particularly after the rise of radical Islamic terrorism, political scientists focused on distinctive elements of the polities supposedly

Introduction

xv

nurturing terrorism, such as ideology, values and culture, but also the degree of freedom, civilian, political and social rights
in a word, the quality of democracy (Quan Li, 2005). Moving from these considerations, the book deals with a number of issues. In the first chapter Andrea Locatelli provides a broad overview of the complexities that prevent scholars from developing an agreed-on definition of what is terrorism. In fact, as some have argued, terrorism took place in so many forms and varieties that any definitional attempt is doomed to fail. On the other hand, its basic appearance is so violently spectacular that one could also say that ‘you recognize terrorism when you see it’. In the attempt to strike a balance between these two extreme positions, the chapter uses concepts and tools derived from political science to highlight the main features common to all cases of terrorism over time and space. In Chapter 2 Daniel Meierrieks reviews the literature on the economic determinants of terrorism vis-a`-vis terrorism’s non-economic determinants (politico-institutional conditions, demographic factors, foreign policy etc.). The chapter also highlights several problems that arise when analysing the role of economic factors in terrorism, namely (i) potential differences in the determinants of domestic and international terrorism and in the micro (individual) and macro (societal) foundations of terrorism, (ii) the complex interaction between economic development, state capacity and their respective relationship with terrorism, (iii) heterogeneity in the (economic) causes of terrorism, depending on the ideological affiliation of terrorist organizations, (iv) the relationship between terrorism and civil war and (v) the a priori unclear causal relationship between terrorism and economic variables. Meierrieks’ main point is that poverty is a ‘root cause’ of terrorism, although in a less straightforward manner than postulated by conventional wisdom. After discussing the forms and determinants of terrorism, Chapter 3 moves to the distribution of terrorist attacks. In particular, Ryan Bakker, Daniel W. Hill and Will H. Moore aim to develop a Bayesian, multilevel statistical model for forecasting. The relevance of this endeavour is both theoretical and empirical: from the first perspective, the academic literature is still quite limited, while in terms of policy-making being able to forecast terrorist attacks, even at an aggregated level of granularity such as the country-year, is in every statesman’s interest. Moore, Hill and Bakker’s analysis evaluates the sensitivity and specificity of the model’s performance, as well as the conditional distributions of positive and negative predicted values. Although the number of false positives and false negatives produced by the model suggests that it is not ready for use in policy planning, it can

xvi

INTRODUCTION

be used as a baseline against which to evaluate the performance of future forecast models. Chapters 4, 5 and 6 deal with measurement. In Chapter 4, Tilman Bru¨ck, Olaf J. de Groot and Neil T. N. Ferguson explore the difficulty of measuring security at the national level. While individual observations concerning attacks (or threats thereof) may exist, it is not immediately obvious how these can be aggregated and combined into an overall indicator. An important risk is that a measure is developed that is ad hoc, rather than scientifically grounded. In order to overcome this limit, the authors first discuss the merits of a number of existing ways of measuring security in order to develop a theoretical foundation of such an indicator. Secondly, they identify as a primary driver three main factors that determine the level of security
the level of threat from agents of insecurity, the provision of protection from these threats and the level of perceived threat. An important observation to add to these three elements is the focus on balance. The overprovision of protection is just as harmful to society as the underprovision thereof. Arguing along similar lines, Chapter 5 describes the rationale and methodology used to develop the Global Terrorism Index (GTI), a composite index which attempts to measure the unobserved latent impact of terrorism across 158 countries covering over 99% of the world’s population. The GTI attempts to capture the multidimensional direct impact of terrorist-related violence, in terms of its physical effect, as well as emotional wounds and fear, by attributing a single weighted average national level score. Based on the Global Terrorism Database (GTD), the GTI aims to be a research tool for analysts requiring a quantitative tool to comprehensively measure on an early basis the impact of terrorism, and to increase public awareness of key national trends on the empirical impact of terrorism. In their analysis, Daniel Hyslop and Thomas Morgan discuss the indicators selected to create the index, the methodological issues related to weighting and normalisation, as well as the potential future improvements to the GTI. In Chapter 6, Christos Kollias and Stephanos Papadamou consider the economic effect of terrorist attacks. By focusing on markets and market agents, they discuss the impact of terrorism-related incidents on the economic sentiment indicator
that is an index that weights the severity of each event. This index is then used to evaluate whether or not and to what extent the economic sentiment indicator in European countries has been affected. As confirmed by empirical evidence, although the threat of a terrorist attack is omnipresent, particularly in countries that are or have in the past been the victims of systematic and continuous terror campaigns,

Introduction

xvii

terrorist events are unforeseen and, depending among others on their magnitude in terms of victims and/or damages or target(s) hit, have the potential to shake and rattle markets and investors. Chapters 7
11 move from the issue of measurement and forecasting treated in previous chapters to provide an inward look at the main features of terrorism and varieties of terrorism. So, Damiano Palano reviews the existing literature on the rationality of terrorism in order to answer what is probably the main question on the topic: why terrorists act? The author compares three broad explanations: (a) the first focuses on the psychology of the individual and explains the action of the terrorist as a quest for social recognition; (b) the second explanation considers the terrorist’s action as a ‘rational’ action, so arguing that the use of terrorist violence is inscribed in a political strategy and in pursuit of medium- to long-term political goals; (c) the third concentrates on the role of cultural context and thus explains terrorist actions and the use of violence as the effect of norms, values, ideologies and behavioural codes. In his essay, the author tries to highlight the strengths and limitations of these three explanations, and, in particular, he seeks to clarify how to effectively explain different types of terrorist actions. In Chapter 8 Peter Phillips draws from economic theory of choice to account for the phenomenon of ‘lone wolf terrorism’. The analysis, based upon expected utility theory, allows for conclusions that are more directly operationally relevant for law enforcement. Moving to a different variety of terrorism, Chapter 9 investigates on the recent phenomenon of home-grown terrorism. Rather than focusing on the distinctiveness of the phenomenon, Kaisa Hinkkainen discusses analogies and differences between homegrown, international and domestic terrorism. Using a comparative method to analyse data from two main sources, ITERATE data on international and the TWEED data on domestic terrorism, she tests the similarities in various dimensions
target types, severity and the method of the attacks. The empirical results suggest that homegrown terrorism is inherently motivated by domestic issues. Moreover, variables of ethnic heterogeneity, political inclusiveness of fringe groups and problems in the democratisation process are good predictors of the occurrence of other forms of domestic and homegrown terrorism alike. In Chapter 10, Giampiero Giacomello discusses the nature and implications of cyberterrorism. In particular, he starts from the distinction between two types of cyberterrorism: ‘Hard-Rock’ and ‘Lite’. The former implies attacking a country’s critical infrastructures so disrupting critical societal functions, like water and energy distribution. Because of its complexity and costs, according to Giacomello, this type of cyberterrorism is far from

xviii

INTRODUCTION

frequent and actually almost non-existent. The latter type consists in using information technologies as a tool for organisational purposes
that is to communicate with affiliates, recruit new elements and, most importantly to publicize their deeds. By virtue of lower costs and high returns, this kind of cyberterrorism is much more common and, as a consequence, calls for more attention. Finally, Chapter 11 discusses one of the most controversial and obscure features of contemporary terrorist organisations: their connection with organized crime. Drawing from the existing literature on the topic as well as on the (limited) available evidence, Carla Monteleone, Raul Caruso and Andrea Locatelli take into consideration the main reasons why criminals and terrorists may find it convenient to join forces
or at least to mingle their practices. While in some cases this connection is occasional and opportunistic, based on ad hoc commonalities of interest, in other cases the relationship can be symbiotic. This is particularly evident in the context of the so-called ‘new wars’, where warring parties are typically private actors. Given the purpose of the book, the breath of topics treated and variety of contributions, we believe this volume can be addressed to several audiences. Obviously, as it should be evident from what we said before, our main addressee is the academic community. Within this group, we believe that graduate students may find it as a useful textbook. Secondly, the policy community may also infer knowledge to frame counterterrorism strategies. It should be stressed from the start that none of the following chapters has a policy prescriptive intent. And we are also aware of the inherent tension between the compelling needs of politics (getting precise solutions now) and the purpose of science (framing general theories). However, if counterterrorism is to be effective, policymakers need to be informed about the many controversial issues related to terrorism. We think that the contents of this volume contribute to this. Raul Caruso Andrea Locatelli Editors

NOTE 1. If searching on Google Scholar the keyword ‘terrorism’ in the period 2001
2013, the number of results amounts to about 40% of the whole scientific production indexed in that search engine.

xix

Introduction

REFERENCES Abrahms, M. (2006). What terrorists really want. International Security, 32(4), 78
105. Anderton, C., & Carter, J. (2005). On rational choice theory and the study of terrorism. Defence and Peace Economics, 16(4), 275
282. Azam, J. P. (2005). Suicide-bombing as inter-generational investment. Public Choice, 122(1
2), 177
198. Bru¨ck, T. (Ed.). (2007). Economic analysis of terrorism. London: Routledge. Caplan, B. (2006). Terrorism: The relevance of the rational choice model. Public Choice, 128(1), 91
107. Caruso, R., & Locatelli, A. (2008). Deadly contests: An economic note on al-Qaeda’s reward system. The Economics of Peace and security Journal, 3(2), 62
68. Caruso, R., & Schneider, F. (2013). Brutality of Jihadist terrorism. A contest theory perspective and empirical evidence in the period 2002
2010. Journal of Policy Modeling, 35(5), 685
696. Crenshaw, M. (2007). The debate over ‘new’ vs. ‘old’ terrorism. Paper prepared for the annual meeting of the American Political Science Association, Chicago, IL. Davis, P., & Cragin, K. (Eds.). (2009). Social science for counterterrorism: Putting the pieces together. Santa Monica, CA: RAND. Enders, W., & Sandler, T. (2000). Is transnational terrorism becoming more threatening? Journal of Conflict Resolution, 44, 307
332. Enders, W., & Sandler, T. (2006). The political economy of terrorism. Cambridge: Cambridge University Press. Haunstrup Qvortrup, M. (2012). Terrorism and political science. British Journal of Politics and International Relations, 14(4), 503
517. Jackson, R., & Sinclair, S. (Eds.). (2012). Contemporary debates on terrorism. New York, NY: Routledge. Kurtulus, E. (2011). The new terrorism and its critics. Studies in Conflict and Terrorism, 34(6), 476
500. Kydd, A., & Walter, B. (2006). The strategies of terrorism. International Security, 31(1), 49
80. Lakatos, I. (1970). Falsification and the methodology of scientific research programmes. In I. Lakatos & A. Musgrave (Eds.), Criticism and the growth of knowledge (pp. 91
196). Cambridge: Cambridge University Press. Llussa, F., & Tavares, J. (2011). The economics of terrorism: A (simple) taxonomy of the literature. Defence and Peace Economics, 22(2), 105
123. March, J., & Olsen, J. (1989). Rediscovering institutions. New York, NY: Free Press. March, J., & Olsen, J. (1995). Democratic governance. New York, NY: Free Press. Martin, G. (Ed.). (2011). The Sage encyclopedia of terrorism (2nd ed.). Los Angeles, CA: Sage. Mearsheimer, J. (2001). The tragedy of great power politics. New York, NY: Norton. Quan Li. (2005). Does democracy promote or reduce transnational terrorist incidents? Journal of Conflict Resolution, 49(2), 278
297. Ranstorp, M. (2007). Introduction: Mapping terrorism research
Challenges and priorities. In M. Ranstorp (Ed.), Mapping terrorism research: State of the art, gaps and future direction (pp. 1
24). London: Routledge. Rosenfeld, J. (2011). Terrorism, identity and legitimacy: The four waves theory and political violence. New York, NY: Routledge.

xx

INTRODUCTION

Sandler, T. (2011). New frontiers of terrorism research: An introduction. Journal of Peace Research, 48(3), 279
286. Sandler, T., Tschirhart, J. T., & Cauley, J. (1983). A theoretical analysis of transnational terrorism. The American Political Science Review, 77(1), 36
54. Schmid, A. (Ed.). (2011). The Routledge handbook of terrorism research. London: Routledge. Shugart II, W. F. (2011). Terrorism in a rational choice perspective. In C. J. Coyne & R. L. Mathers (Eds.), The handbook on the political economy of war. Cheltenham: Edward Elgar. Silke, A. (2004). The road less travelled: Recent trends in terrorism research. In A. Silke (Ed.), Research on terrorism: Trends, achievements and failures (pp. 186
213). London: Frank Cass. Silke, A. (2007). The impact of 9/11 on research on terrorism. In M. Ranstorp (Ed.), Mapping terrorism research: State of the art, gaps and future direction (pp. 76
93). London: Routledge. Van Um, E. (2011). Discussing concepts of terrorist rationality: Implications for counterterrorism policy. Defence and Peace Economics, 22(2), 161
179. Weinberg, L., Pedahzur, A., & Hirsch-Hoefler, S. (2004). The challenges of conceptualizing terrorism. Terrorism and Political Violence, 16(4), 777
794.

CHAPTER 1 WHAT IS TERRORISM? CONCEPTS, DEFINITIONS AND CLASSIFICATIONS Andrea Locatelli ABSTRACT Purpose
The aim of this chapter is to investigate the meaning of terrorism, with a view to highlighting the main hurdles in the way of creating a working definition, as well as the necessity of developing definitions and classifications of this phenomenon. Design methodology/approach
This chapter provides an overview of the literature on terrorism as a social/political phenomenon. It is therefore based on secondary sources. Findings
While most literature on the topic finds it pointless or impossible to define terrorism, here we argue just the opposite. Common critiques of current definitions may be overcome by using multiple definitions and classifications. Research limitations/implications
The chapter provides the methodological foundations for a comprehensive theoretical analysis of terrorism.

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 1
23 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022001

1

2

ANDREA LOCATELLI

Originality/value of the chapter
The chapter applies insights from methodology of social sciences to the problem of defining terrorism. Keywords: Terrorism; terrorists; concept definition; types of terrorism

INTRODUCTION In the past few decades, social sciences have benefited from impressive developments in terms of methodologies and techniques. Especially as far as quantitative methods are concerned, statistical analysis has provided analysts with a huge array of sophisticated tools. This is particularly helpful for those who study terrorism from an empirical perspective. Given the growing number of databases available (Bowie & Schmid, 2011), as well as the burgeoning quantitative literature on the topic, one may be tempted to believe that we currently know more about terrorism than we used to just a few decades ago. Alas, as many philosophers of science have forcefully claimed (in particular Khun, 1962; Lakatos, 1970), social sciences are hardly cumulative and studies on terrorism make no exception. We might have a comprehensive database of terrorist attacks on a global scale over the past 20 years, or even a better understanding of how al Qaeda works, but terrorism as a phenomenon still lacks an agreed-upon definition. Apparently, as a good amount of literature seems to show (see the excellent overview in McAllister & Schmid, 2011), there is no solution to this problem: terrorism is such a complex phenomenon that it is overdetermined (i.e. so many factors combine to cause it that it is just impossible to develop a theory). However, this should not be an excuse for giving up on this endeavour. In fact, social sciences may perform different heuristic functions. Developing nomological explanations is probably the most ambitious goal, but also the most elusive. A precondition for doing good research is defining and operationalising concepts properly (Della Porta & Keating, 2008, pp. 2
4; Sartori, 1970, 1984; Smelser, 1976). Or, to borrow Sartori’s warning, ‘concept formation stands prior to quantification’ (Sartori, 1970, p. 1038; emphasis original). As a consequence, the main purpose of this chapter is to sweep the conceptual minefield surrounding terrorism, first by speculating on the elements that define it, then by discussing a few criteria for classification. What, then, is terrorism? The term is used so commonly that one may wonder whether we really need to bother defining the concept. In fact, we

What is Terrorism? Concepts, Definitions and Classifications

3

could even come to the conclusion that ‘you know terrorism when you see it’ (Laqueur, 1977, p. 5). However, terrorist actions have taken place over such a long time span, and in so many parts of the world, that it is necessary to draw distinctions and investigate at greater length the different ways in which it is articulated. The motives driving terrorists, their modus operandi, target selection, organisational model
all are aspects of the same phenomenon that have made appearances in changing ways over centuries. A catch-all definition, in conclusion, proves extremely difficult to formulate, since on the one hand by broadening the definition one risks including phenomena other than terrorism, while on the other hand a narrower formulation may lead to the exclusion of some cases. In its attempt to come to terms with the chameleonic nature of terrorism, the chapter aims to provide the reader with a critical overview of the literature on the topic. In the first section, we will start by discussing why defining terrorism is so complicated and what the main analytical problems in this endeavour are. In the next section, we will first approach the target with a denotative definition
that is, a declaration of meaning aimed at setting the boundaries of a given phenomenon (Sartori, 1984, p. 30). With this goal in mind, we will conduct a critical overview of the main dimensions in which terrorism is articulated, with a view to drawing a line between terrorism and neighbouring concepts. In the third section, we will move closer to an operational definition
that is, one that spells out the main properties of terrorism and, according to them, provides indications on how to measure it. In light of the properties discussed in this section, the fourth paragraph will list down a few criteria for classification of terrorism aimed at developing sound comparative/quantitative analysis. The final paragraph will summarise and wrap up the argument presented in this chapter.

WHY BOTHER DEFINING TERRORISM? When it comes to defining terrorism, the analytical task is partially overwhelmed by a number of empirical difficulties. Due to the very relevance of the issue, the basic (methodological and deontological) assumption by which the researcher is detached from the phenomenon he/she is studying tends to fade. As noted by, among others, Smelser (2007, pp. 230
233), three main problems arise. The first one can be defined as a perspective bias. As discussed for more than 20 years by Alex Schmid and his

4

ANDREA LOCATELLI

colleagues (Paletz & Schmid, 1992; Schmid, 2004), what adds confusion is that different arenas (like governments, mass media, scholars and the broader public) tend to converge towards different biases. So, for example, governments prefer loose definitions that emphasise the private and criminal nature of terrorist groups;1 journalists, driven by the need to raise public attention, tend to stress elements of violence and anxiety; the debate in academia tends to be somehow more aseptic, while the public, although not homogeneous, is usually pulled one way or the other by the other arenas.2 The second problem outlined by Smelser has to do with what he calls recency and myopia. The underlying argument is that scholarly publications are somehow driven by not purely scientific considerations. In the case of terrorism, he claims that when crafting definitions scholars are biased by the most recent (or most shocking) events of their time. As a result, in highlighting the main properties of terrorism, authors (usually unconsciously) stress those features that seem constitutive of the phenomena at hand. So, just to give an example, in the late 40s and 50s, the main face of terrorism resembled guerrilla warfare (due to the frequent use of terrorist tactics by anti-colonial movements), while today the main aspect highlighted by scholars (Kurtulus, 2011; Laqueur, 1999) is the international dimension of terrorism, as well as links with religious ideologies. Finally, the third problem concerns the normative load associated with the term. This is actually quite common in Political Science
one may consider for instance the definition of democracy, imperialism, national interest and so on. Even more than these concepts, terrorism is associated with a clear stigma (Herbst, 2003, pp. 163
164). Although there are good reasons to agree with those
like Crenshaw (1983) and Wilkinson (1987)
who argue for the scientific foundations of normative definitions, this position is not widely accepted. In short, following this argument, labelling a given actor (or action) as terrorist would necessarily depend on a value judgement; also, it would serve the purpose of delegitimising either the actor or the action. The end result is that the debate on what terrorism is is hardly constructive, both in public opinion and at the governmental level. But there is also a substantial problem. First, terrorism is a recurrent phenomenon in history, but it has appeared in so many different forms that we may well borrow Clausewitz’ famous metaphor about war as ‘chameleon-like’. Second, no clear boundaries exist between terrorism and other forms of political violence (a good example being the current debate on the difference between terrorists and insurgents). All this considered, it should come as a surprise to no one that we still lack a broadly recognised

What is Terrorism? Concepts, Definitions and Classifications

5

definition of terrorism (Bogatyrenko, 2007; Laqueur, 2004, p. 232; Long, 1990, p. ix), and that even eminent scholars have pronounced the task of defining terrorism to be simply impossible (Laqueur, 1977, p. 79; 1999, p. 5), pointless (Malik, 2000) or even ‘infernal’ (Smelser, 2007, p. 229). Starting from these considerations, we won’t engage in the tiring and hardly productive task of reviewing the long list of definitions used by scholars (for a glimpse of how broad this literature has become see the appendix at the end of this chapter). We will certainly rely on some of the most common contributions to the field, but we will mostly focus on the criteria needed to circumscribe the borders and morphology of the concept.

BACKGROUND CONCEPT While aware of the limits that are intrinsic in any ambitious definitional effort, the next pages aim to circumscribe the range of empirical referents denoted by the concept of terrorism. In this endeavour we will follow the two-step approach developed by Adcock and Collier (2001, p. 530).3 As a result, we will first discuss what they call a background concept
that is, ‘the constellation of potentially diverse meanings associated with a given concept’ (Adcock & Collier, 2001, p. 530); second, we will frame a systematized concept
that is, a specific formulation which includes and spells out the essential properties of the concept. This two-step approach seems particularly suited to grasping the essence of terrorism, as it aims to balance the trade-off between complexity and simplification. Or, in other words, to solve the dilemma between the need to provide generalisations about complex phenomena that are as broad as possible, while making abstract concepts easily measurable. Such a synthesis is made possible thanks to the double level of conceptualisation: the background concept makes it possible to grasp the complexity of meanings and conceptions related to the term.4 The systematised concept, on the other hand, implies a specific and explicit definition
one that will certainly miss part of the broader constellation of meanings covered by the background concept, but will make operationalisation possible. As a first approximation to the background concept, we can start with a discussion of the connection between terrorism and crime. Although, as we shall see, it would be wrong to subsume terrorism as a mere criminal act, this perspective sheds some light on a number of aspects of the phenomenon under investigation. The idea underlying most terrorist actions is to

6

ANDREA LOCATELLI

inflict damage of some sort on a given target. This in turn implies an element of destruction and violence that is inevitably tied to the concept of illegality, as violence is deemed legal only as an exception, and under very strict conditions. Therefore, terrorism is something illegal, a breach of (domestic or international) law. These considerations, as trivial and incontestable as they may sound, tend to hide a particular configuration of terrorism: state terrorism. Since the state is by definition the holder of the legitimate monopoly of violence
or, in other terms, the institution in charge of implementing domestic law
it is extremely difficult to qualify state actions as terrorist. In other words, if being a criminal were a precondition to being a terrorist, state terrorism would hardly exist, since no state would recognise its behaviour as criminal. In other words, when a regime instils terror in its own population how can we define such actions as criminal? Borrowing from the centuryold distinction between mala prohibita and mala per se, Schmid (2004) argues that terrorism can be considered a crime only in this latter case. Put differently, the first term stands for actions that infringe conventional law, while the second refers to actions intrinsically criminal. By considering terrorism a public bad, it does not matter if it is impossible to appeal to a law or court. However, the sphere of criminality raises another problem. While few would doubt that terrorist actions are by definition criminal actions, not all crimes can be labelled as terrorism. For our purpose, then, it is critical to draw a clear boundary between terrorism and crime (and, likewise, between terrorist and criminal organizations). An example may help to clarify this problem: the Sicilian mafia is engaged in criminal activities. At the beginning of the past century, it spread terror among peasants using tactics similar to those used by terrorists. In other words, the mafia perpetrates crimes, it has even resorted to terrorism, but it does not qualify as a terrorist organization. Where, then, should we draw the line between crime and terrorism? In order to answer this question we must turn to the second dimension of the background concept: politics. Differently from criminal activities, terrorism is defined by a clear political orientation and/or relevance. Such a political dimension reminds us that terrorists usually (although not always) have a political goal. Borrowing again from Schmid, we can argue that ‘since terrorists generally challenge the monopoly of violence of the state and its ability to protect its citizens, terrorist acts obtain political significance even when the motivation for them is not primarily political’ (2004, p. 200). Considering the relationship between terrorism and politics, then,

What is Terrorism? Concepts, Definitions and Classifications

7

one might conclude that the use of terrorism represents one of the options available to both the state and social actors in the repertoire of political action. Similarly to our conclusion on the previous dimension, the relationship between terrorism and politics is symbiotic. The problem that arises, then, is once again how to find where lies the boundary between them. What makes terrorism different from other forms of political violence (e.g. assassination, riots or violent suppression of political dissent)? As before, to answer this question we have to turn to the third dimension considered here: terrorism as a form of communication. In particular, terrorism can be conceived in terms of costly signalling (Lapan & Sandler, 1993). According to this view, violence is used to gain credibility, and it becomes relevant as long as it implies an element of cost. Put differently, terrorist actions can be conceived as signs of resolve, lines in the sand to eliminate liars and bluffers from the game. In this view, then, violence is just a means, not an end. While other forms of political violence result in pure coercion (the target of violence also being the ultimate enemy), terrorism is only partly devoted to that purpose, the other part being forced compliance and/or deterrence.5 As a result, the answer to the question
what makes terrorism different from other forms of political violence?
can be traced down to the congenital practice of terrorism to engage in indirect strategies. As anticipated, this first-glance overview of the three main perspectives on terrorism clearly shows that we cannot rely on a single analytical framework, as none proves dominant or preferable in the literature. Quite the contrary, all perspectives benefit a lot from mutual contamination. Nonetheless, even in this broad and approximate form, this gives us the possibility of defining the genus of terrorism as a social activity with proper elements. In the following pages we will try to discuss at greater length the components of terrorism.

SYSTEMATIZED CONCEPT We can now move on to what Adcock and Collier define as the systematized concept
that is, ‘the specific formulation of a concept adopted by a particular researcher or group of researchers’ (Adcock & Collier, 2001, p. 530). Although they admit that this is usually summarized in a single definition, it also requires ‘a broader specification of the meaning and entailments’ (p. 532). This is particularly important when dealing with terrorism, since (as discussed in the previous pages), a parsimonious definition would miss

8

ANDREA LOCATELLI

central elements of the concept, while an overly denoted definition would squeeze excessively the range of empirical referents. In order to sort out these problems, we will try to balance these competing tensions by locating the main components of terrorism and discussing their relative weight as constitutive elements of the concept itself. The literature on terrorism has highlighted a number of elements. Although differences exist among scholars (Schmid, 2011; Schmid, 1984; Schmid & Jongman, 1988) and even across time (Weinberg et al., 2004) some of them have been met with broad consensus.6 Here we will focus on five elements: • The use of violence. The first constitutive element of terrorism is represented by the use of violence. Such a statement is apparently trivial and superfluous, but it is not. Or, at least, it deserves some qualification. First of all, while the main definition of violence implies physical destruction (of properties and, most commonly, human lives), we should also consider functional disruption
that is, those threats that endanger the capacity of a given society to function (for an early treatment see Buzan, 1983). In this sense, we can include in our definition cyber-terrorism,7 environmental terrorism etc., that otherwise would be lost. Second, as Schmid (2011, pp. 78
79) claimed, violence is a necessary but not sufficient condition for defining terrorism: what distinguishes terrorism from violence tout court is that terrorist actions are conceived to be exemplary: they include both an element of destruction (or disruption, as we have seen) and an element of threat aimed at coercing, intimidating or deterring a third party. • Political goal and/or effect. This condition marks a difference between public and private (which is usually, although not exclusively, confined to the realm of criminal activity) violence.8 The point that needs to be stressed here is that terrorism has a connection of some sort with the political sphere. Or, in Charles Tilly’s jargon, terrorism is part of the repertoire of political violence (Tilly, 2003, 2006). Most commonly, this takes the form of a political goal, like opposing (or sustaining, in the case of state terrorism) a given regime, claiming independence, calling for social revolution or fostering policy change (more on this in the next paragraph). But politics is part of terrorism even when the ultimate goal of the perpetrators is not political, but just ‘psychotic’, or criminal: this is the case when the effect of terror attacks is deemed so relevant by a given state (or group of states) that it is forced to take action in response to an existential threat (Addison, 2002, p. 73; Crenshaw, 1983, p. 25). What

What is Terrorism? Concepts, Definitions and Classifications

9

matters for our purposes is that one way or another terrorism is essentially political in nature. • Indirect approach
Terrorism is by definition an indirect approach. The term, originally coined by British strategist Sir Basil Liddell Hart (1974) as an antidote to wars of attrition like Word War I, implies a duplication of targets: one is the object of violence but is not the ultimate target; the other one is the enemy per se, the counterpart whose behaviour (or mere existence) we want to change. This strategy, aimed at weakening the enemy by hitting its soft belly, or even by proxy, is common to most terrorist activities.9 This aspect is recognized by the literature with different names
as we have seen, some refer to terrorism as a form of communication (Lapan & Sandler, 1993; Schmid & de Graaf, 1982), while some stress the difference between civilian victims and the ultimate audience, and others talk about intimidation (Sandler, Tschirhart, & Cauley, 1983, p. 37)
but the underlying principle does not change. To quote Brian Jenkins’ famous line, as the word terror reveals, most frequently the final purpose of terrorist organisations is to have ‘a lot of people watching, not a lot of people dead’. • Patent breach of accepted rules
Many authors have highlighted the immoral and/or illegal dimension of terrorist acts (Crenshaw, 1983; Falk, 2003; Wilkinson, 1987) and made it a central element of their definitions. This is obviously a controversial issue, which brings into the picture values and beliefs and makes any definitional attempt arbitrary by definition. Put differently, following this approach may well result in confirming the old saying ‘one man’s terrorist is another man’s freedom fighter’. But still, a central element of terrorism as an action is the infringement of some norm. In order to overcome this problem, we can refer here to accepted rules
usually, but not necessarily codified in domestic and international law. This element is particularly useful in wartime to discriminate between soldier and terrorist: while the former is subject to jus in bello and is legally bound to spare non-combatants (Colombo, 2006, pp. 129
134; Evangelista, 2008, pp. 37
43; Waltzer, 1977, pp. 144
146, 197
198), the latter has non-combatants as a primary target. In peacetime, the intrinsic proclivity of terrorism to breach accepted rules is witnessed by the many tactics used to perpetrate attacks (from kidnapping to mass murdering). • Relative strength of offence over defence
The final element that defines terrorism is frequently overlooked by scholars. However, it is probably the only feature that has remained constant over time and space. For a variety of reasons, but mostly as a consequence of its indirect approach,

10

ANDREA LOCATELLI

the balance between offence (i.e. perpetrating an attack) and defence (preventing or frustrating the effects of an attack) favours the former. In other words, planning an attack is often easier, less expensive and more likely to succeed than defending against it. Of course, this is not to say that terror attacks are always successful: indeed, as Jackson and Felinger (2009, pp. 4
5) recognise, it is impossible to gather reliable data on failed attacks, and even more to find correlations between failures and their likely causes. Yet the point is strategic, not tactical: single attacks may well be thwarted, but at the higher (strategic) level, the odds are in favour of terrorists. Summing up, we can conflate these elements into a systematised definition that would read as follows: terrorism is a peculiar form of political violence based on an indirect approach. It implies a patent breach of accepted rules and enjoys a tactical advantage over defence. Admittedly, this list of attributes also has some weak points: some may point to the lack of clear-cut standards that allow one to neatly discriminate between what terrorism is and what it is not (although the reader should know already that we find it simply impossible to identify such standards). Others could note that this list is just a starting point, but it is not enough, since it still includes many different varieties of terrorism (put differently, it does not represent a significant improvement on the background concept). Finally, others may take issue with the fact that these elements are not properly sorted out: They are all considered building blocks of terrorism, but some of them should be given priority (e.g. violence, political goal/effect) compared to others (breach to the rules, offence
defence balance). As mentioned, we find the first critique is based on an excessively ambitious epistemology: definitions are intended to shed light on phenomena and pave the ground for theoretical and empirical research. In other words, they should come to terms with reality, not vice versa, or they end up being straightjackets
maybe elegant ones, but still more of an impediment than a constructive tool. As concerns the second and third critiques, it is true that this formulation still contains elements of uncertainty. In order to address this issue and add clarity to the operational concept, we now discuss the usefulness of developing multiple classifications of the phenomenon.

CLASSIFICATIONS AND TYPOLOGIES After this brief journey through the constellations of meanings and defining attributes of terrorism, we have come to broadly define the range of

What is Terrorism? Concepts, Definitions and Classifications

11

empirical referents associated with the term. Within this range lie very different kinds of terrorists and terrorist acts. No doubt, al Qaeda, the Red Brigades and Unabomber
to name just a few examples
are different in terms of membership, political goals and ideological foundations. Likewise, attacks like those perpetrated by Aum Shinrikyo in Japan, Palestinian suicide bombers in Israel and anti-abortion movements in the United States (again, just to choose cases at random) hardly share common elements in target selection, tactics and modus operandi. Such variety calls for further clarification. In other words, it is necessary to define criteria for classification and, consequently, make comparison possible. This is true for both qualitative and quantitative research. This problem has been widely discussed in the literature and several competing criteria have been proposed to classify terrorism, terrorist groups and attacks (among others, see Dekmejian, 2007, p. 10; Kaplan, 2008; Lo¨ckinger, 2005; Parmentier & Weitekamp, 2007; Ganor, 2008; Schmid, Stohl, & Flemming, 1988 for a summary, see Schmid, 2011, pp. 158
201). Interestingly, it seems that most authors developed typologies with little concern for sound methodological criteria. In fact, the very same definition of typology is frequently stretched beyond its limits, and Mill’s criteria (exclusivity and exhaustivity) (Mill, 1843) have mostly been overlooked. To be clear, while it would be desirable to root analyses on terrorism in more solid methodological foundations, it is worth recalling the basic function of this scientific endeavour: ‘classification schemes, typologies, taxonomies do not make assertions and therefore cannot be judged true or false […] As concepts, they are tools for conferring organization and stability on our thoughts about reality. Like tools, they may be judged or found more or less useful for a particular purpose’ (Marradi, 1990, p. 149). The classification proposed here borrows elements from previous efforts. We do not have the ambition to develop something original. Rather, our intent is to pave the ground for the chapters that follow and highlight some of the main similarities and differences between various types of terrorism. As such, in order to balance the need for parsimonious categorisations and comprehensive classifications (i.e. one that includes as many types of terrorism as possible), we will consider three main criteria for classification: the nature of the actors involved, the ultimate goals that drive these actors and the type of perpetrators. As far as the first criterion is concerned, we can draw a distinction between state terrorism, state-sponsored terrorism and terrorism perpetrated by private actors. The first case is by definition an example of domestic terrorism.10 Examples include totalitarian regimes (Arendt, 1951; Friedrich & Brzezinki, 1956), like Nazi Germany, Stalin’s purges or Saddam

12

ANDREA LOCATELLI

Hussein’s atrocities against the Kurdish minority in Iraq. The second cell is represented by state-sponsored terrorism. Here we refer to terrorism used as a tool of foreign policy: regimes like Lybia under Muhammad Qaddafi made use of terrorism to weaken adversaries and foster foreign policy goals; likewise, Iran since the demise of the Shah nurtured terror organizations in other countries to spread the revolution. The third, and probably most common, type of terrorist actor is represented by private actors. Here we basically refer to transnational networks, groups, movements, or even individuals that use terror for a given cause. Contemporary examples of this wide and articulated group of perpetrators include al Qaeda, the Basque organization ETA and Anders Breivik. The second criterion is related to the final goal of the state/organization/ network/lone wolf. In a nutshell, what are terrorists fighting for? What is the ultimate objective they claim beyond the rhetoric surrounding their propaganda? Of course, this is a very complicated issue, one that implies taking into account the motivations of terrorists,11 the very idea of rationality (Anderton & Carter, 2006; Van Um, 2011) and even the difference between terrorists and terrorists groups (Abrahms, 2006). In order to reduce the complexity of such an issue, we will rely on Kydd and Walter’s (2006, pp. 52
55) contribution.12 By assuming that terrorist organizations are rational actors whose goals can be conceived in political terms, they find that five ideal
typical goals are recurrent over time: • Regime change
Usually the most common goal, it consists in the use of terrorism by a group representing a faction of the society that aims to subvert the internal political order or remove a ruler. Needless to say, this is typical of private perpetrators. Examples in this sense are Sendero Luminoso in Peru and al Qaeda in Iraq. • Policy change
It is the attempt to affect policies of a given state (or group of states) by means of terrorism. This goal is shared by both private actors and state-sponsored terrorism. As an example of the first kind, al Qaeda aims to force the United States to withdraw from Arab soil. A case of the second kind is Iran-sponsored Hezbollah and its ambitions in Lebanon. • Territorial change
It is the goal of a group that fights for the independence of a community. The enemy is a government that is considered illegitimate or oppressive. According to Rapoport (2004), this was the main drive for terrorist movements in the colonial era (the so-called second wave of terrorism). Nowadays relevant examples include the Irish IRA, Basque ETA and the Tamil Tigers.

What is Terrorism? Concepts, Definitions and Classifications

13

• Social control
This is another example of a mostly domestically oriented terrorist organization. It is best represented by the Ku Klux Clan. Its main goal is to impose or implement a given configuration of the society (in that particular case, discrimination between AfricanAmericans and white Americans). We might also contend that social control is also a possible goal of state terrorism. An example is given by the already mentioned confrontation between Saddam Hussein and the Kurdish minority; or the Nazi goal to wipe out the Jews in German society. • Status quo maintenance
This final goal is aimed at preserving a given internal order. This is, as Kydd and Walter contend, quite uncommon. The only example they find is the United Self-Defence Force of Colombia, a right-wing organization whose main goal was to promote the stability of the regime (Kydd & Walter, 2006, p. 53). On the other hand, we might also consider it as a goal of state terrorism, as long as it is aimed at perpetuating the control of a given elite (or even a single person) over the state machinery. Finally, the third criterion represents the type of actor. This label encompasses two aspects of the perpetrators: the membership of the terrorist organization (i.e. how many people are part of it, from one to many hundreds) and the structure of the organization (whether there are clear hierarchies, borders and divisions of labour). Most of the literature, especially in the wake of 9/11, focused on private actors, defining a variety of different types (see, among others, Arquilla & Ronfeldt, 2001; Jackson, 2006). They include: • Single terrorists, commonly referred to as Lone Wolves:13 in this case, terrorists act alone, usually have no contact with other groups and are motivated by personal (usually ideologically informed) interests. Examples are Theodore Kaczynski (Unabomber) in the United States and Anders Breivik in Norway. • A second type of terrorism is usually referred to as a clique. This refers to small groups of terrorists (in the order of four to five people), usually bound by strong social ties and independent of other organisations (Sageman, 2004, 2008). Examples are the perpetrators of the Madrid (2004) and London (2005) bombings (Caruso & Locatelli, 2008, 2009; see also Caruso & Schneider, 2013). • Then there are movements
that is, ‘a group of people with a common ideology who try together to achieve certain general goals’ (Jackson,

Type of actors

Ultimate goal

Criteria Nature of actors

Gov.+transnational Gov.+organization Trasnational Network Network Talibans and al Iran and Hezbollah al Qaeda Qaeda

Unabomber, Breivik

Leeds cell

ETA, IRA, Red Brigades

Anti-abortion movements

Clique Lone wolves

Organization Movement

Sendero Luminoso, ETA, al Qaeda in Iraq; Ku Klux Clan; United Self-Defense Force of Colombia

Lybia’s Muhammad Qaddafi; Iran’s Ayatollah Komehini

al Qaeda, ETA, IRA, Anders Breivik

Regime change, territorial change, policy change, social control, status quo maintenance

Lybia’s Muhammad Qaddafi; Iran’s Ayatollah Komehini

Revolutionary France, Stalin’s purges Social control; status quo maintenance Nazi Germany, Saddam Hussein’s Iraq Governmental agencies Death squads, secret police

Terrorism perpetrated by private actors

Policy change, territorial change, policy change, Status quo maintenance

State-sponsored terrorism

Criteria for a Classification of Terrorism.

State terrorism

Table 1.

14 ANDREA LOCATELLI

What is Terrorism? Concepts, Definitions and Classifications

15

2006, p. 248; see also Della Porta, 1992). Examples are the anti-abortion, environmentalist and animal rights movements (Davidson Smith, 1998). • Most common are organizations
that is, groups structured along hierarchical lines. Examples are the Irish IRA, Basque ETA and Italian Red Brigades. Finally, we have transnational networks
represented in their fullest form by al Qaeda. As far as state terrorism is concerned, the main type of actor is the death squad, but it can also be a government or paramilitary agency. Finally, moving to state-sponsored terrorism, here we have a connection between a governmental agency and either an organization or a transnational network. Examples of this sort are the connections between Iran and Hezbollah, and the Taliban and al Qaeda, respectively. As evident in Table 1, this endeavour is just a preliminary step for theoretical and empirical analysis of terrorism. The cells displayed may be used for different purposes: they can refine conceptualisation and measurement (Collier, LaPorte, & Seawright, 2012), support comparative studies (Sartori, 1970) or discern paths of causation (Elman, 2005). In the chapters that follow (see especially the second question), the authors have mostly pursued the third goal, as they used different approaches and tested competing theories to explain why and how people turn into terrorists. In order to sort out such a variety of contributions, we worked on the assumption that drawing lines among different types of terrorism is a necessary, although hardly sufficient, condition.

CONCLUDING REMARKS Terrorism is widely recognised as a textbook case of an ‘essentially contested concept’ (Gallie, 1956). The multifaceted nature of terrorism has deterred many scholars from engaging in definitional attempts. Indeed, many valuable contributions have openly admitted that generalisations are hindered by the complexity and contextual nature of the phenomenon. As a result, regardless of the few contributions to operationalising the concept and paving the ground for comparative analyses, just a tiny fraction of the current literature has developed sound methodologies for broad range theories. In this chapter we tried to come to terms with this theoretical gap. While admitting that an all-encompassing definition is beyond our reach (and probably of little use), we tried to solve the trade-off between analytical

16

ANDREA LOCATELLI

clarity and empirical precision (or, to put it differently, between intension and extension of the concept) with a three-stage strategy. The first step involved highlighting the main traits of a ‘background concept’; the second step was crafting a ‘systematized concept’; the third step consisted in developing a broad classification of types of terrorism. This exercise has led us to devise simple conceptual tools that ground empirical analysis in solid methodological foundations. In turn, future research will provide elements to refine and adapt these concepts to changing circumstances.

NOTES 1. It is also quite frustrating that despite a common interest in framing a widely accepted legal definition, no consensus has been reached so far. On this point see, among others, Zeidan (2004) and Saul (2006). 2. For this reason, Schmid’s lifetime conclusion is that for analytical purposes the best research path is to work in the academic arena, in the attempt to find an agreed-upon definition. Here we follow Schmid’s approach, since our work mostly addresses researchers. Yet, we do not find consensus to be an important requisite for framing a good definition. As we will argue later in the chapter, we hold that the way definitions are framed does not necessarily need to be consensual, rather useful for the purposes of the researcher. 3. Adcock and Collier actually discuss a four-stage process, but their goal (measurement validity) was different from ours. 4. This complexity, we should recall, is made even more problematic by the widespread tendency among scholars to study terrorism with few concerns for definition (Weinberg, Pedahzur, & Hirsch-Hoefler, 2004). 5. In his analysis of different dimensions of terrorism, Schmid (2004) finds two more (that we will skip here because irrelevant in this context): terrorism and warfare and terrorism as religion. 6. The broadest definition, supplied by Schmid (1984, pp. 76
77), contains 22 elements. In contrast, Weinberg et al. (2004, p. 786) are satisfied with just 5. 7. As concerns cyberterrorism, this point is also discussed in Giacomello’s chapter within this volume. 8. Of course, the very same idea of what is political should be discussed in depth, but this is clearly beyond the scope of our work. 9. Incidentally, indirect strategy is also inherent in guerrilla tactics and asymmetric conflicts as a weapon of the weak. This is one of the reasons why it is so difficult to draw a line between terrorists and insurgents. For valuable contributions to solve this controversy see Rapoport, 1977 and Merari, 1994. 10. We won’t follow the distinction used by other authors between domestic and international terrorism, since the dividing line is quite fuzzy. It is not therefore really a discriminatory criteria. For instance, many groups may have a narrow geographic scope and a merely domestic agenda, but still have connections with

What is Terrorism? Concepts, Definitions and Classifications

17

other international organizations (an example being al Qaeda in Iraq). An attempt to bypass this problem is Hinkkainen’s chapter within this volume. 11. This point is discussed at greater length in Palano’s chapter within the volume. 12. Differently from their approach, however, we think these goals can be applied to both state and non-state terrorism. 13. See on this type of terrorists Phillips’ chapter within the volume.

REFERENCES Abrahms, M. (2006). What terrorists really want. International Security, 32(4), 78
105. Adcock, R., & Collier, D. (2001). Measurement validity: A shared standard for qualitative and quantitative research. American Political Science Review, 95(3), 529
546. Addison, M. (2002). Violent politics: Strategies of internal conflict. New York, NY: Palgrave. Anderton, C., & Carter, J. (2006). On rational choice theory and the study of terrorism. Defence and Peace Economics, 16(4), 275
282. Arendt, H. (1951). The origins of totalitarianism. San Diego, CA: Harcourt. Aron, R. (1966). Peace and war. London: Weidenfeld & Nicolson. Arquilla, J., & Ronfeldt, D. (Eds.). (2001). Networks and netwars: The future of terror, crime and militancy. Santa Monica, CA: RAND. Bogatyrenko, O. (2007, February
March). Definitional analysis of terrorism: constructing concepts and populations for social science research. Paper presented at the 2007 meeting of the International Studies Association, Chicago. Bowie, N., & Schmid, A. (2011). Databases on terrorism. In A. Schmid (Ed.), The Routledge handbook of terrorism research (pp. 294
340). London: Routledge. Buzan, B. (1983). People, states and fear. The national security problem in international relations. Brighton: Wheatsheaf. Caruso, R., & Locatelli, A. (2008). Deadly contests: An economic note on al-Qaeda’s reward system. The Economics of Peace and security Journal, 3(2), 62
68. Caruso, R., & Locatelli, A. (2009). Una gara al massacro: Incentivi, premi e prestazioni dei gruppi terroristici legati ad al Qaeda. Il Politico, 221(2), 63
91. Caruso R., & Schneider, F. (2013). Brutality of jihadist terrorism. A contest theory perspective and empirical evidence in the period 2002
2010. Journal of Policy Modeling, 35(5), 685
696. Retrieved from http://www.sciencedirect.com/science/article/pii/S0161893813 000124 Collier, D., LaPorte, J., & Seawright, J. (2012). Putting typologies to work: Concept formation, measurement, and analytic rigor. Political Research Quarterly, 65(1), 217
232. Colombo, A. (2006). La guerra ineguale. Pace e violenza nel tramonto della societa` internazionale. Bologna: Il Mulino. Crenshaw, M. (1983). Introduction: Reflections on the effects of terrorism. In M. Crenshaw (Ed.), Terrorism, legitimacy and power: The consequences of political violence (pp. 1
37). Middletown, CT: Wesleyan University Press. Davidson Smith, G. (1998). Single issue terrorism. Commentary no. 74. Ottawa: Canadian Security Intelligence Service. Dekmejian, R. (2007). Spectrum of terror. Washington, DC: CQ Press.

18

ANDREA LOCATELLI

Della Porta, D. (Ed.). (1992). International social movement research: Participation in clandestine political organizations. Greenwich, CT: JAI Press. Della Porta, D., & Keating, M. (2008). Introduction. In D. Della Porta & M. Keating (Eds.), Approaches and methods in the social sciences: A pluralist perspective (pp. 1
16). Cambridge: Cambridge University Press. Elman, C. (2005). Explanatory typologies in qualitative studies of international politics. International Organization, 59(2), 293
326. Evangelista, M. (2008). Law, ethics and the war on terror. Cambridge: Polity Press. Falk, R. (2003). The great terror war. Northampton, MA: Olive Branch Press. Friedrich, C., & Brzezinski, Z. (1956). Totalitarian dictatorship and autocracy. Cambridge, MA: Harvard University Press. Gallie, W. B. (1956). Essentially contested concepts. Proceedings of the Aristotelian Society, 56(1), 167
198. Ganor, B. (2008). Terrorist organization typologies and the probability of a boomerang effect. Studies in Conflict and Terrorism, 31(4), 269
283. Herbst, P. (2003). Talking terrorism: A dictionary of the loaded language of political violence. Westport CT: Greenwood Press. Hoffman, B. (1997). The confluence of international and domestic trends in terrorism. Terrorism and Political Violence, 9(2), 1
15. Jackson, B. (2006). Groups, networks, or movements: A command-and-control-driven approach to classifying terrorist organizations and its application to Al Qaeda. Studies in Conflict and Terrorism, 29(3), 247
48. Jackson, B., & Felinger, D. (2009). Understanding why terrorist operations succeed of fail. RAND Occasional Paper. RAND, Santa Monica, CA. Retrieved from http://www.rand. org/content/dam/rand/pubs/occasional_papers/2009/RAND_OP257.pdf Kaplan, S. (2008). A typology of terrorism. Journal of Political Philosophy, 6(1), 1
38. Khun, T. (1962). The structure of scientific revolutions. Chicago, IL: The University of Chicago Press. Kurtulus, E. N. (2011). The ‘new terrorism’ and its critics. Studies in Conflict and Terrorism, 34(6), 476
500. Kydd, A., & Walter, B. (2006). The strategies of terrorism. International Security, 31(1), 49
80. Lakatos, I. (1970). Falsification and the methodology of scientific research programmes. In I. Lakatos & A. Musgrave (Eds.), Criticism and the growth of knowledge (pp. 91
196). Cambridge: Cambridge University Press. Lapan, H., & Sandler, T. (1993). Terrorism and signaling. European Journal of Political Economy, 9(3), 383
398. Laqueur, W. (1977). Terrorism. Boston, MA: Little, Brown. Laqueur, W. (1999). The new terrorism: Fanaticism and the arms of mass destruction. Oxford: Oxford University Press. Laqueur, W. (2004). No end to war: Terrorism in the twenty-first century. New York, NY: Continuum. Liddell Hart, B. (1974). Strategy. The indirect approach (2nd ed.). New York, NY: Signet Books. Lo¨ckinger, G. (2005). Terrorismus, Terrorismusabwehr, Terrorismusbekampfung. Vienna: Ministry of Defence. Retrieved from www.bmlv.gv.at/pdf_pool/publikationen/05_ttt_01_ ttt.pdf. Accessed on 21 January, 2013.

What is Terrorism? Concepts, Definitions and Classifications

19

Long, D. (1990). The anatomy of terrorism. New York, NY: Free Press. Malik, O. (2000). Enough of a definition of terrorism. London: Royal Institute of International Affairs. Marighela, C. (1971). For the liberation of Brazil. Harmondsworth: Penguin Books. Marradi, A. (1990). Classification, typology, taxonomy. Quality and Quantity, 24(2), 129
157. McAllister, B., & Schmid, A. (2011). Theories of terrorism. In A. Schmid (Ed.), The Routledge handbook of terrorism research (pp. 201
271). London: Routledge. Merari, A. (1994). Characteristics of terrorism, guerilla, and conventional war as modes of violent struggle. In V. S. Ramachandran (Ed.), Encyclopedia of human behaviour (Vol. 4). San Diego, CA: Academic Press. Mill, J. S. (1843). A System of logic, ratiocinative and inductive. New York, NY: Harper & Brothers. Paletz, D., & Schmid, A. (Eds.). (1992). Terrorism and the media: How researcher, terrorists, government, press, public and victims view and use the media. London: Sage. Parmentier, S., & Weitekamp, E. (2007). Political crimes and serious violations of human rights: Towards a criminology of international crimes. In S. Parmentier & E. Weitekamp (Eds.), Crime and human rights: Sociology of crime, law and deviance (Vol. 9). Amsterdam-Oxford: Elsevier. Rapoport, D. (1977). The politics of atrocity. In Y. Alexander & S. M. Finger (Eds.), Terrorism: Interdisciplinary perspectives. New York, NY: John Jay Press. Rapoport, D. (2004). Modern terror: The four waves. In A. Cronin & J. Ludes (Eds.), Attacking terrorism: Elements of a grand strategy (pp. 46
73). Washington, DC: Georgetown University Press. Sageman, M. (2004). Understanding terror networks. Philadelphia: University of Pennsylvania Press. Sageman, M. (2008). Leaderless jihad. Terror networks in the twenty-first century. Philadelphia: University of Pennsylvania Press. Sandler, T., Tschirhart, J. T., & Cauley, J. (1983). A theoretical analysis of transnational terrorism. The American Political Science Review, 77(1), 36
54. Sartori, G. (1970). Concept misformation in comparative politics. American Political Science Review, 64(4), 1033
1053. Sartori, G. (1984). Social science concepts. A systematic analysis. Beverly Hills: Sage. Saul, B. (2006). Defining terrorism in international law. Oxford: Oxford University Press. Schmid, A. (2004). Frameworks for conceptualising terrorism. Terrorism and Political Violence, 16(2), 197
221. Schmid, A. (Ed.). (2011). The Routledge handbook of terrorism research. London: Routledge. Schmid, A. (1984). Political terrorism: A research guide to concepts, theories, data bases and literature. Amsterdam: North-Holland. Schmid, A., & de Graaf, J. (1982). Violence as communication, insurgent terrorism and the western new media. London: Sage. Schmid, A. & Jongman, A. J. (1988). Political terrorism: A new guide to actors, authors, concepts, data bases, theories, and literature. Amsterdam: North Holland. Schmid, A., Stohl, M., & Flemming, P. (1988). Typologies. In A. Schmid & A. J. Jongman (Eds.), Political terrorism: A new guide to actors, authors, concepts, data bases, theories, and literature (pp. 39
59). Amsterdam: North-Holland. Smelser, N. (1976). Comparative methods in the social sciences. Englewood Cliffs, NJ: PrenticeHall.

20

ANDREA LOCATELLI

Smelser, N. (2007). The faces of terrorism, social and psychological dimensions. Princeton NJ: Princeton University Press. Tilly, C. (2003). The politics of collective violence. Cambridge: Cambridge University Press. Tilly, C. (2006). Regimes and repertoires. Chicago IL: University of Chicago Press. Van Um, E. (2011). Discussing concepts of terrorist rationality: Implications for counterterrorism policy. Defence and Peace Economics, 22(2), 161
179. Waltzer, M. (1977). Just and unjust wars. A moral argument with historical illustrations. New York, NY: Basic Books. Walzer, M. (2004). Arguing about war. New Haven, CT: Yale University Press. Weinberg, L., Pedahzur, A., & Hirsch-Hoefler, S. (2004). The challenges of conceptualizing terrorism. Terrorism and Political Violence. 16(4), 777
794. Wilkinson, P. (1974). Political terrorism. London: Macmillan. Wilkinson, P. (1987). Terrorism. An international research agenda? In P. Wilkinson & A. Stewart (Eds.), Contemporary research on terrorism (pp. xi
xx). Aberdeen: Aberdeen University Press. Wurth, P. (1941). La Re´pression internationale du terrorisme. Lausanne: Imprimerie la Concorde. Zeidan, S. (2004). Desperately seeking definition: The international community’s quest for identifying the specter of terrorism. 36 Cornell International Law Journal, 491
492.

What is Terrorism? Concepts, Definitions and Classifications

21

APPENDIX: A CONCISE LIST OF PREVIOUS DEFINITIONS OF TERRORISM (ADAPTED FROM SCHMID, 2011, PP. 99FF) 1794 Robespierre Terror is nothing else than immediate justice, severe, inflexible; it is therefore an outflow of virtue, it is not so much a specific principle than a consequence of the general principle of democracy applied to the most pressing needs of the motherland (Robespierre, quoted in Wurth, 1941, p. 12). 1966 Aron An action of violence is labelled ‘terrorist’ when the psychological effects are out of proportion to its purely physical result. In this sense, the so-called indiscriminate acts of revolutionaries are terrorist, as were the AngloAmerican zone bombings. The lack of discrimination helps to spread fear, for if no one in particular is a target, no one can be safe (Aron, 1966, p. 170). 1971 Marighela By terrorism I mean the use of bomb attacks (Marighela, 1971, p. 89). 1974 Wilkinson Our main concern is with political terror: that is to say with the use of coercive intimidation by revolutionary movements, regimes or individuals. … We have thus identified some of the key characteristics common to all forms of political terror: indiscriminateness, unpredictability, arbitrariness, ruthless destructiveness and the implicitly amoral and antinomian nature of a terrorist’s challenge. … Political terrorism, properly speaking, is a sustained policy involving the waging of organized terror either on the part of the state, a movement or faction, or by a small group of individuals. Systematic terrorism invariably entails some organizational structure, however rudimentary, and some kind of theory or ideology of terror (Wilkinson, 1974, p. 11). 1977 Laqueur Terrorism, interpreted here as the use of covert violence by a group for political ends, is usually directed against a government, less frequently against another group, class or party. The ends may vary from the redress of specific ‘grievances’ to the overthrow of a government and the taking of power, or to the liberation of a country from foreign rule. Terrorists seek to cause political, social and economic disruption, and for this purpose

22

ANDREA LOCATELLI

frequently engage in planned or indiscriminate murder … . Note: Any definition of political terrorism venturing beyond noting the systematic use of murder, injury and destruction or the threats of such acts toward achieving political ends is bound to lead to endless controversies (Laqueur, 1977, pp. 79, 79n). 1983 Sandler et al. Premeditated, threatened or actual use of force or violence to attain a political goal through fear, coercion or intimidation (Sandler et al., 1983, p. 37). 1984 Schmid Act of Terrorism = Peacetime Equivalent of War Crime 1. Deliberate attacks on civilians. 2. Taking of hostages. 3. Killing of prisoners/kidnapped persons (Schmid, 1984, pp. 109
110). 1988 Della Porta Terrorism is the action of clandestine political organizations, of small dimensions, which try to reach political aims through a continuous and almost exclusive use of violent forms of action (D. Della Porta, in answer to questionnaire, quoted in Schmid et al., 1988, p. 37). 1988 Schmid Academic Consensus Definition: Terrorism is an anxiety-inspiring method of repeated violent action, employed by (semi-)clandestine individual, group, or state actors, for idiosyncratic, criminal, or political reasons, whereby
in contrast to assassination
the direct targets of violence are not the main targets. The immediate human victims of violence are generally chosen randomly (targets of opportunity) or selectively (representative or symbolic targets) from a target population, and serve as message generators. Threat- and violence-based communication processes between terrorist (organization), (imperilled) victims, and main targets are used to manipulate the main target (audience(s)), turning it into a target of terror, a target of demands, or a target of attention, depending on whether intimidation, coercion, or propaganda is primarily sought, (Schmid et al., 1988, p. 28). 1997 Hoffman Religious terrorism tends to be more lethal than secular terrorism because of the radically different value systems, mechanisms of legitimization and

What is Terrorism? Concepts, Definitions and Classifications

23

justification concepts of morality. For religious terrorist violence first and foremost is a sacramental act of divine duty: executed in direct response to some theological demands or imperative and justified by scripture (Hoffman, 1997, p. 4). 2004 Walzer War terrorism: the effort to kill civilians in such large numbers that their government is forced to surrender. … State terrorism is commonly used by authoritarian and totalitarian government against their own people, to spread fear and make political opposition impossible (Walzer, 2004, p. 130).

CHAPTER 2 ECONOMIC DETERMINANTS OF TERRORISM Daniel Meierrieks ABSTRACT Purpose
The purpose of this contribution is to review the theoretical and empirical literature on the economic determinants of terrorism. Methodology/approach
Review of the relevant academic literature. Findings
This contribution shows that there is a theoretical foundation to the popular hypothesis that poor economic conditions are conducive to terrorism. A review of the empirical evidence on the economic determinants of terrorism, however, yields an inconclusive result. Some studies find that economic conditions (directly and indirectly) matter to terrorism, whereas a plurality of studies suggest that noneconomic factors are more important. Research limitations/implications
The findings of the survey indicate that it is unlikely that economic conditions are universal determinants of terrorism. By pointing at several avenues of future research (e.g., a focus on the role of ideology in terrorism), this contribution, however, argues that the opposite also does not need to be true. The influence of economic factors on terrorism should neither be overemphasized nor completely ruled out.

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 25
49 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022002

25

26

DANIEL MEIERRIEKS

Originality/value of chapter
The contribution offers a comprehensive overview of the economy
terrorism nexus and hints at promising areas of future research. Keywords: Terrorism; poverty; economic growth; causality; ideology

INTRODUCTION Interest in identifying the determinants of terrorism has gained renewed attention after the devastating terrorist attacks on New York and Washington, D.C., on September 11, 2001. Shortly after the attacks, a consensus among intellectuals, politicians, and the public emerged that socioeconomic underdevelopment crucially contributes to the emergence of terrorism (Jai, 2001). For instance, Kim Dae-jung, the Nobel Peace Prize recipient of 2000, argued that at “[…] the bottom of terrorism is poverty. That is the main cause” (cited in Jai, 2001). Similarly, U.S. President George W. Bush (2002) suggested that in underdeveloped countries “[…] persistent poverty and oppression can lead to hopelessness and despair,” leading to situations where these countries can ultimately become “havens for terror.” Indeed, there is a long tradition among the leaders and followers of terrorist movements to attribute their armed struggle
at least partly
to the prevalence of poor socioeconomic conditions. For instance, Vladimir I. Lenin (1906) argued that the “[…] old Russian terrorism was an affair of the intellectual conspirator; today as a general rule guerrilla warfare is waged by the worker combatant, or simply by the unemployed worker.” That is, for Lenin, terrorism has clear economic underpinnings, being rooted in economic disenfranchisement, unemployment, and
more broadly and consistent with the Marxist ideas of historical materialism
the class conflict between the proletariat (“the workers”) and the bourgeoisie who owns the means of production.1 Throughout the 20th century, left-wing revolutionary groups in the developed
for instance, the West German Rote Armee Fraktion (RAF) and the Italian Brigate Rosse (BR)
and in the developing world
for example, the Argentinian Eje´rcito Revolucionario del Pueblo (ERP) and the Philippine New People’s Army (NPA)
referred to similar lines of argument, styling their terrorist campaigns as armed struggle directed against social injustice and economic oppression on national and global levels (Shughart, 2006). Economic motivations have also been repeatedly

27

Economic Determinants of Terrorism

4

5000

3

4000

2

3000

1

2000

0

1000

–1

Fig. 1.

06

04

20

02

20

00

20

98

20

96

19

94

19

92

Terrorist Attacks

19

90

19

88

–2

19

19

84

82

19

80

19

19

86

r = –0.122

0

Economic Growth

6000

19

Terrorist Attacks

argued to fuel terrorist movements with ostensibly territorial or religious goals.2 For instance, Piazza (2011) suggests that economic discrimination of minority groups may have crucially fueled violent activities by separatist terrorist groups (e.g., Sri Lanka’s Liberation Tigers of Tamil Eelam, the LTTE). Mousseau (2011) argues that urban poverty has filled the ranks of and fueled popular support for Islamist terrorist groups such as the Turkish Islami Bu¨yu¨kdogu Akıncilar Cephesi (IBDA-C) and the Islamic Movement of Uzbekistan (IMU). To sum up, in the popular discourse and parts of the academic literature, socioeconomic underdevelopment has been repeatedly named as a root cause of terrorism. Indeed, the self-perception of many terrorist groups suggests that economic motivations matter strongly to their armed struggle. Fig. 1 plots the global number of terrorist attacks between 1980 and 2007 against the global growth of per capita income during this period of time. The data on terrorist activity are drawn from the Global Terrorism Database (the GTD) as introduced by LaFree and Dugan (2007).3 The growth series is extracted from the latest version of the World Development Indicators (WDI), as provided by the World Bank (2012). Fig.1 suggests that more terrorist activity negatively correlates with global economic activity (the correlation coefficient is
0.122). While Fig. 1 does not allow for making any causal arguments, it nevertheless illustrates the fact that, at least at first glance, popular perceptions about the socioeconomic roots of

Growth of GDP p.c.

Global Terrorist Activity and GDP Per Capita Growth, 1980
2007.

28

DANIEL MEIERRIEKS

terrorism do not seem to be unsubstantiated. For instance, the decline in global economic growth during the early 1990s was associated with a noticeable increase in world-wide terrorist incidents during the same time period. Given the role economic factors may play in fueling terrorist activity, this contribution provides an overview of the academic discussion of the nexus between them and terrorism. It shows that there is indeed a theoretical foundation to the popular claim that terrorism is more likely to originate in poor societies. However, it also summarizes the existing empirical research and finds that there is no empirical consensus with respect to this issue. Given the inconclusive evidence on the economic determinants of terrorism, this contribution discusses several possible explanations for it, particularly (i) heterogeneity in the causes of terrorism, (ii) differences in the causes of domestic and transnational terrorism, and (iii) the issue of causation. This contribution argues that these factors
while they potentially help to better understand the economy
terrorism nexus
remain under-analyzed and therefore open up interesting avenues of future research on the economic fundamentals of terrorism.

ECONOMIC DETERMINANTS OF TERRORISM: THEORY As shown in the introduction, popular wisdom has it that economic conditions matter to the emergence of terrorism. Intuitively, the poor
the likely would-be terrorists
ought to resort to violence to change the socioeconomic status quo in their favor. This section provides a more formal framework to model such interactions using economic theory. It first establishes that terrorists can be considered rational actors and then shows how economic factors influence the rational behavior of terrorists.4 Terrorists as Rational Actors Rational-choice theory can be used to provide a theoretical framework that mirrors the popular wisdom on the terrorism
economy relationship. Rational-choice models are the theoretical workhorse of most economic analyses of terrorist behavior (e.g., Caplan, 2006; Sandler & Enders, 2004). Analogous to the economic analysis of crime,5 rational-choice models

Economic Determinants of Terrorism

29

of terrorism assume terrorists to be rational actors
behaving more or less like homini economici
who try to maximize their utility, given the benefits, (opportunity) costs and economic constraints linked to these actions. In fact, while public perception tends to view terrorist behavior as “irrational,” psychological studies of individuals participating in terrorist groups
reviewed by Victoroff (2005)
do not suggest that terrorists routinely suffer from mental incapacities. Consequently, Caplan (2006), who provides an extensive analysis of terrorist irrationality, comes to the conclusion that the sympathizers of terrorism and most active terrorists act more or less rational.

The Terrorists’ Calculus and Economic Factors Using a rational-choice perspective to economically analyze terrorism, one can expect terrorists to choose the optimal (i.e., utility-maximizing) level of violence by considering the benefits from terrorism (e.g., from gaining government concessions) as well as its material costs (e.g., from government punishment) and opportunity costs (e.g., from foregone earnings). Arguably, a country’s long-run level of socioeconomic development (measured by, e.g., poverty levels) and short-run state of the economy (indicated by, e.g., economic growth) are anticipated to influence the terrorists’ calculus. In detail, one can expect an effect of economic factors on: (i) most importantly, the opportunity costs of terrorism. Similar to economic models of crime, the opportunity costs of terrorism refer to the foregone utility associated with non-terrorist activity. Typically, this utility comes from material reward (i.e., wages) associated with nonviolence, that is, participation in the ordinary economic life. Arguably, a poor structural level of economic development is likely to coincide with constrained means of economic participation. Unfavorable short-run economic conditions are expected to have the same effect. For instance, low levels of economic growth are usually accompanied by increases in unemployment or pressures on wages. This ought to mean lower opportunity costs of terrorism and thus (ceteris paribus) more terrorism; (ii) the material costs of terrorism. Usually, these costs are associated with the operations of a terrorist organization, that is, its offensive and defensive activities (acquiring/using financial resources, purchasing firearms and explosives, evading government punishment through the establishment and maintenance of safe houses, etc.). Crucially, the operating costs of terrorist groups tend to be lower with stronger popular support for terrorist

30

DANIEL MEIERRIEKS

groups (Paul, 2010). For instance, it ought to be easier for terrorists to find sanctuary when the population is sympathetic to their cause. Popular support for terrorism, in turn, also depends on economic factors (Paul, 2010). When economic conditions are unfavorable, popular support for terrorism is more likely. Economically speaking, the opportunity costs of supporting terrorism are lower when a country’s economic situation is poor (e.g., Freytag, Kru¨ger, Meierrieks, & Schneider, 2011). Consequently, this ought to lower the material (operating) costs of active terrorists, making terrorism (ceteris paribus) more likely to occur; and (iii) the benefits of terrorism. These benefits are linked to the tactical and strategic goals of terrorism. The tactical goals of terrorism include politicoeconomic destabilization (to weaken the enemy) and media attention (to communicate and publicize the terrorists’ cause), as argued by Schelling (1991). The strategic goals usually involve sweeping political, economic, religious, or social change (imposing a politico-economic system based on religious doctrine, gaining independence, etc.). Any success related to achieving the short-run and long-run goals of terrorism (through government concessions, gaining political influence, etc.) ought to constitute a benefit of terrorist activity. Clearly, poor economic conditions mean that the potential pay-off from terrorism (e.g., in the form of a redistribution of wealth or other economic concessions) is comparably more attractive. Then, terrorism ought to be (ceteris paribus) more likely due to comparatively higher benefits from terrorism during poor economic times. As shown above, rational-choice theory offers several pathways through which poor economic conditions are anticipated to coincide with more terrorist activity. Indeed, a number of academic contributions use this framework to develop more formal models of the interaction between economic variables and terrorism. For instance, Blomberg, Hess, and Weerapana (2004a, 2004b) introduce a model in which the poor state of the economy (i.e., slow economic growth) fosters terrorist activity by coinciding with especially low opportunity costs of terrorism as nonviolent opportunities of economic participation are limited and by raising the benefits of terrorism as terrorist success (i.e., a redistribution of wealth) means more access to an otherwise scarce resource base. Similarly, Freytag et al. (2011) argue that potential and active terrorists and terrorist supporters weigh up the mental rewards from terrorism (e.g., identity) against the material rewards from ordinary economic activity. Their analysis suggests that poor economic conditions mean reduced material rewards from nonviolence and thus an increased likelihood of terrorist activity, all else being equal.

Economic Determinants of Terrorism

31

ECONOMIC DETERMINANTS OF TERRORISM: EVIDENCE The previous section showed that there is a theoretical foundation to the popular perception that terrorism is rooted in poor economic conditions. Rational-choice theory suggests that unfavorable economic factors make terrorism more likely by lowering its opportunity and direct material costs and by raising its potential benefits. This section surveys various empirical examinations of this hypothesis. As shown below, the evidence regarding the economic roots of terrorism is far from being conclusive. Positive Evidence Empirical studies on the economic determinants of terrorism
that is, the main strand of empirical research
usually employ indicators of short-run economic performance (e.g., economic growth and inflation rates) and longrun (structural) economic conditions (e.g., per capita income and poverty and inequality levels) to test whether such variables exert a statistically significant effect on a measure of terrorist activity (usually, the number of terrorist incidents in a given country and year), net of the influence of noneconomic control variables (e.g., population size and regime type). Consistent with the rational-choice theory outlined in the previous section, some of these empirical studies indeed find that economic progress is detrimental to terrorism. Here, several studies find that higher levels of income and lower levels of poverty are obstacles to the production of terrorism (e.g., Blomberg & Hess, 2008a, 2008b; Caruso & Schneider, 2011; Freytag et al., 2011; Lai, 2007; Meierrieks, 2012; Santos Bravo & Mendes Dias, 2006). For instance, Meierrieks (2012) shows that urban poverty in developing countries may promote the emergence of terrorism, while Freytag et al. (2011) find that countries with higher per capita income levels are less likely to experience terrorism. Furthermore, there is evidence connecting solid short-run economic conditions (especially, sound economic growth) with less political violence (Blomberg et al., 2004a; Freytag et al., 2011; Muller & Weede, 1990). For instance, Blomberg et al. (2004a) find that economic downturns correlate with increased terrorist activity. Indeed, these empirical findings not only speak to popular perceptions about the causes of terrorism
“poverty causes terror”
and related expectations derived from rational-choice theory. They are also in line with the robust empirical evidence that economic success immunizes against the

32

DANIEL MEIERRIEKS

occurrence of civil wars as another
more severe
form of violent conflict (Collier & Hoeffler, 1998; Fearon & Laitin, 2003; Sambanis, 2002). For example, the findings of Collier and Hoeffler (1998) indicate that higher levels of economic development coincide with lower likelihoods of civil war, providing further evidence that economic success and conflict are diametrically opposed.

Indirect Linkages Another strand of the empirical literature tries to establish indirect linkages between socioeconomic conditions and the emergence of terrorist activity. These studies argue that socioeconomic progress is induced by prior policy change (the development of a welfare state, economic liberalization, etc.). Formally, these studies try to differentiate between intervening variables
through which the policy change ultimately influences terrorism
and confounding variables which may influence the policy change and the dependent variable (in this case, terrorism) at the same time and therefore need to be controlled for. This research design is outlined in more detail in, for example, Ray (2005). With respect to the study of the determinants of terrorism, studies that try to establish indirect links between economic conditions and terrorism generally argue that these very economic conditions are
as intervening variables
induced by some prior policy change, but influence the level of terrorist activity through their expected effect on the terrorists’ calculus (i.e., policy-induced economic improvements result in less terrorism by, e.g., increasing the opportunity costs of terrorism). For instance, Burgoon (2006) argues that social welfare policies reduce terrorist activity by ameliorating economic grievances (e.g., economic insecurity, inequality) that may have otherwise fueled terrorism. Given the full body of empirical research on potential indirect linkages between socioeconomic conditions and terrorist activity, there is evidence that (i) policies of economic integration and liberalization tend to reduce terrorism by fostering economic growth and development (Li & Schaub, 2004), (ii) social welfare policies and functioning social safety nets are associated with less terrorism by removing economic grievances (Burgoon, 2006; Krieger & Meierrieks, 2010), (iii) economic policies that counter economic discrimination of minorities lead to less terrorism as such discrimination may have otherwise motivated terrorism (Piazza, 2011), and (iv) the protection of economic rights (i.e., property rights protection) is negatively related

Economic Determinants of Terrorism

33

to terrorism by, on the one hand, fostering growth and development and, on the other hand, removing economic insecurity that may have otherwise constituted a breeding ground for terrorist groups due to comparatively low material and opportunity costs of terrorism (e.g., Basuchoudhary & Shughart, 2010; Gassebner & Luechinger, 2011).

Noneconomic Determinants of Terrorism While the aforementioned survey of the empirical literature is rather supportive of the idea that economic conditions
directly and indirectly
determine terrorist activity, a rather large number of empirical studies nevertheless come to another conclusion. In fact, recent literature reviews by Gassebner and Luechinger (2011), Kis-Katos, Liebert, and Schulze (2011), and Krieger and Meierrieks (2011) suggest that there is little academic
empirical consensus on the role of economic conditions in terrorism, with the majority of studies stressing the primacy of noneconomic over economic variables. For instance, the seminal study by Krueger and Maleckova (2003, p. 142) concludes that the “cross-country evidence that we have assembled suggests that, once civil liberties are taken into account, a country’s income level is unrelated to the number of terrorists who originate from that country […].” Another often-cited study by Abadie (2006, p. 55) similarly finds no “significant association between terrorism and economic variables such as income once the effect of other country characteristics is taken into account.” Instead of emphasizing the effect of poor economic conditions on terrorism, these empirical studies instead find that terrorism is rooted in (i) unfavorable political conditions which, for example, do not sufficiently protect civil liberties and human rights, do not provide a sound rule of law, and do not enable political participation (Abadie, 2006; Basuchoudhary & Shughart, 2010; Choi, 2010; Krueger & Laitin, 2008; Krueger & Maleckova, 2003; Kurrild-Klitgaard, Justesen, & Klemmensen, 2006; Piazza, 2006; Walsh & Piazza, 2010; Young & Dugan, 2011), (ii) demographic stress, for example, in the form of ethno-religious tensions, discrimination along ethno-religious lines, and the existence of a youth bulge (Basuchoudhary & Shughart, 2010; Urdal, 2006), (iii) trends in globalization that may generate resentment among traditionalists who fear that the inflow of foreign ideas threatens their local culture or religion (Lizardo, 2006; Zimmermann, 2011), and (iv) aggressive foreign policy behavior
in the form of, for example, military interventions, state sponsorship of terrorism, and overall

34

DANIEL MEIERRIEKS

politico-military dominance
which tends to coincide with more terrorist activity directed against the proponents of such policies (Bapat, 2007; Conrad, 2011; Dreher & Gassebner, 2008; O’Brien, 1996; Pape, 2003; Plu¨mper & Neumayer, 2010; Sobek & Braithwaite, 2005). In sum, the aforementioned plethora of empirical studies neglects, on the one hand, a prominent role of economic underdevelopment and disenfranchisement in terrorism. That is, although these studies control for the effect of economic factors on terrorism, they fail to establish a robust and statistically significant association. On the other hand, these studies hint at the role of noneconomic factors in terrorism. At a minimum, this strand of the empirical literature suggests that terrorism is not only economically determined.

DISCUSSION AND FUTURE RESEARCH The previous section has shown that the empirical evidence on the economic roots of terrorism is far from conclusive. While some studies find that economic factors matter, a majority of statistical analyses suggest that noneconomic factors are more important. Table 1 once more illustrates this finding by summarizing the findings of several empirical studies on this issue. This section discusses several possible explanations for the inconclusiveness of the empirical evidence. In so doing, it also provides some promising avenues of future research on this issue.

Heterogeneity in the Causes of Terrorism As a first explanation approach, the inconclusive evidence on the terrorism
economy nexus may be due to the possible heterogeneity in the causes of terrorism. In short, this line of reasoning suggests that economic factors may determine terrorist activity in some, but not all countries. Arguably, terrorist organizations differ with respect to their ideological motivations and ultimate goals. As outlined by Rapoport (2004) and Shughart (2006), the post-World War II period saw different waves of terrorist activity: first by anticolonialist and national liberation groups such as the Ethniki Organosis Kyprion Agoniston (EOKA) on Cyprus, then by left-wing groups, mainly motivated by the antagonisms of the Cold War, such as the Tupamaros in Uruguay, followed by a wave of Islamist

186 countries 2003
2004

179 countries 1968
2003

148 countries 1997
2002 112 countries 1975
1997

96 countries 1986
2002

172 countries 1970
2006

Blomberg and Hess (2008a)

Krueger and Maleckova (2003) Li and Schaub (2004)

Piazza (2006)

Piazza (2011)

Scope

Domestic terrorism

Unemployment; economic Growth; HDI; Gini index; Inflation; malnutrition GDP p.c.; Gini index; HDI; minority economic discrimination

GDP p.c.; trade openness; FDI; portfolio investment

Transnational terrorism

Transnational terrorism

GDP p.c.

GDP p.c.; trade openness

GDP p.c.; Gini index; Human Development Index (HDI)

Economic Determinants

Transnational terrorism

Transnational terrorism

Terrorism risk index

Type of Terrorism

No significant relationship between economic variables and terrorism High income/make a country less likely to produce terrorism No effect of per capita income on emergence of terrorism Terrorism negatively related to economic success (induced by economic integration) Economic variables do not matter substantially to incidences of terrorism Terrorism more likely in rich countries (indicated by GDP p.c./HDI) with inequality and minority discrimination

Main Findings

Assorted Empirical Studies on the Economic Causes of Terrorism.

Abadie (2006)

Study

Table 1.

Economic Determinants of Terrorism 35

36

DANIEL MEIERRIEKS

terrorism by groups such as Hezbollah, which originated in the Islamic Revolution in Iran, the Soviet invasion of Afghanistan and the Lebanese civil war. Given the diverse ideological profiles associated with these different terrorism waves, it seems intuitive that some terrorist groups have been more responsive to economic conditions than others. For instance, it seems plausible to assume that the armed struggle of the Guatemalan Eje´rcito Guerrillero de los Pobres (EGP) had economic roots, given that it emerged in response to poverty and socioeconomic discrimination and self-identified as a group that fought for a removal of socioeconomic grievances (Lopez, 1988). By contrast, the Islamist Izz ad-Din al-Qassam Brigades (i.e., the armed wing of Hamas) seems
at least at first glance
less likely to be motivated by economic factors, given their ideological focus and ultimate goal of establishing an independent Islamic Palestinian state. Indeed, its active members seem to be on average rather well-off and highly educated (Krueger & Maleckova, 2003). Given the history of distinct terrorism waves which have been associated with equally distinct ideological motivations and political goals, it seems more plausible to assume that economic grievances matter to some (e.g., the EGP), but not to all terrorist insurgencies. Intuitively, there ought to be a stronger connection between economic variables and left-wing (which is usually geared towards socioeconomic change) or separatist terrorism (which is often tied to economic discrimination; see Piazza, 2011) than between the economy and religious terrorism (which commonly strives for religious motives). Economically speaking, the calculus of terrorists may be intrinsically tied to their ideology and goals, where socioeconomic progress can be expected to have a stronger (terrorism-reducing) effect on the opportunity costs of left-wing and ethnic terrorists compared to its impact on the calculus of religious terrorists (e.g., Freytag et al., 2011). Bernholz (2004) even goes so far as to argue that Islamist terrorists are completely irresponsive to socioeconomic incentives due to their adherences to supreme values which they use to justify terrorism and which are preferred to all else. Nevertheless, almost all large-N studies on the causes of terrorism, which have been reviewed in the previous section, pool terrorism data. That is, they examine the relationship between economic (and other social) conditions and terrorism irrespective of the ideological motivations and goals of terrorist groups. Thus, implicitly, these studies work with the assumption that there is a unique terrorist calculus which economic variables enter the same way across all countries. Among the exceptions to this

Economic Determinants of Terrorism

37

rule are Robison, Crenshaw, and Jenkins (2006) and Kis-Katos, Liebert, and Schulze (2012). These studies carefully distinguish between different types of terrorism and come to the conclusion that they markedly differ with respect to their social foundations and the role economic factors play in them. For instance, Robison et al. (2006) find that left-wing and Islamist terrorism are driven by different determinants, with left-wing terrorism being more responsive to economic variables. Potentially, accounting for ideological bonds may improve the understanding of the economic (and noneconomic) determinants of distinct types of terrorism. This also ought to help to developed sounder counter-terrorism strategies that aim at alleviating specific grievances that fuel distinct types of terrorism. Arguably, however, such analyses necessitate careful data mining efforts, given that existing large-N datasets such as International Terrorism: Attributes of Terrorist Events (ITERATE; see, e.g., Mickolus, Sandler, Murdock, & Fleming, 1993) and the GTD do not provide information on the ideological profile of groups involved in terrorist incidents.6

Transnational and Domestic Terrorism Besides heterogeneity in the causes of terrorism, the inconclusive evidence regarding the role of economic factors in terrorism may also be due to a focus of previous research efforts (see Table 1) on the phenomenon of transnational terrorism and a neglect of the phenomenon of domestic terrorism. In short, this line of reasoning assumes that transnational terrorism has specific noneconomic roots, while domestic terrorism, which is far more common than transnational terrorism,7 may be more strongly connected to domestic economic conditions. By definition (e.g., Sandler & Enders, 2004, p. 302), terrorism is transnational when more than one country is involved, while domestic terrorism only affects one country. For example, an attack of the German RAF on a German politician in Germany (e.g., the assassination of the German Attorney General Buback and his companions in Karlsruhe in 1977) is a domestic terrorist attack as it only involves German perpetrators and victims inside Germany. By contrast, an attack by the RAF on a U.S. target in Germany (e.g., the bombing of an U.S. Army building in Frankfurt in 1972) is a transnational incident as it involves the targeting of nonGermans. As other examples of transnational terrorism, a terrorist group may “export” terrorism to another country (e.g., attacks by the Algerian Groupe Islamique Arme´, the GIA, in France), where its activity may also

38

DANIEL MEIERRIEKS

be sponsored by a third country (e.g., Iran-backed Hezbollah attacks against Israel). Empirical analyses of the determinants of terrorism
reviewed above
primarily examine the roots of transnational terrorism. Probably, this is simply due to data constraints. The oldest and most widely used dataset on terrorism, the ITERATE dataset, only provides information on transnational terrorist activity (Drakos, 2011). Only recently
that is, after 2008
the GTD has made available data on domestic and transnational terrorism (Drakos, 2011). Yet, domestic and transnational terrorism may have different roots. As shown above, the empirical mainstream so far suggests that economic variables do not matter strongly to the origins of transnational terrorism (Gassebner & Luechinger, 2011; Kis-Katos et al., 2011; Krieger & Meierrieks, 2011). Rather, there is considerable evidence that this kind of terrorism emerges in response to specific developments in the global cultural (e.g., spread of culture and “clash of civilizations”) and politico-military order (e.g., interventions and other forms of aggressive foreign policy behavior). For instance, Zimmermann (2011) argues that various aspects associated with the phenomenon of globalization (immigration, the global spread of ideologies and information, network effects, social polarization, etc.) are conducive to the emergence of transnational terrorist activity. Similarly, Bergesen and Lizardo (2004) suggest that transnational terrorism emerges in response to cultural and political globalization, particularly as a (cost-efficient) weapon of the disenfranchised against the political, military, and economic power of the West. As stressed above, the empirical findings of a number of empirical studies also suggest that a country’s foreign policy behavior is a crucial determinant of transnational terrorist activity directed against its interests. For instance, Dreher and Gassebner (2008) find that political proximity to the United States makes a country a likelier target of transnational attacks. Likewise, Conrad (2011) argues that there is a strong (positive) interaction between interstate rivalry, the sponsorship of terrorism, and the prevalence of transnational terrorism. However, it is unclear to what extent the global order matters to domestic terrorism. For example, Savun and Phillips (2009) find that democracies are vulnerable to transnational terrorism (due to specific foreign policies undertaken by democracies), while they are not particularly prone to domestic terrorism. This may indicate that domestic and transnational terrorism are
at least, partly
determined by different factors. In particular, one may speculate that domestic terrorism has stronger economic roots, while material conditions do not matter substantially to transnational

39

Economic Determinants of Terrorism

terrorism. Arguably, domestic economic factors (poverty, unemployment, etc.) are more likely to sway the terrorists’ calculus in the absence of international motivations. The internationalization of terrorism in turn may be more strongly fueled by foreign policy considerations and rather abstract trends in globalization. This line of reasoning calls for an analysis of the (economic) determinants of domestic terrorism, where related findings should then be contrasted with the existing evidence regarding the (economic) roots of transnational terrorism. Potentially, such an approach may show that internationalized terrorism is a distinct terrorism subset that is uniquely influenced by international cultural
economic and political factors (e.g., globalization, foreign policy behavior, foreign occupation), while domestic terrorism is more strongly connected to the domestic economic sphere. Newly available data on domestic terrorism provided by the GTD and Enders et al. (2011)8 is likely to aid such research. Such an approach should be helpful to make more general statements about the (economic) causes of terrorism. Yet, it remains to be seen whether such research can truly unravel all confusion associated with the evidence on the role of economic factors in terrorism. As argued by Sanchez-Cuenca and De La Calle (2009), the actors of domestic and transnational terrorism (i.e., terrorist organizations) are usually identical. Sanchez-Cuenca and De La Calle (2009) stress that most terrorist groups attack inside their “natural territory,” where they may target inhabitants of their own country (domestic terrorism) or citizens of other countries (transnational terrorism). They also argue that few terrorist groups systematically attack outside their “natural territory” for only international reasons. This argument calls for research on the determinants of “natural territory terrorism” and “systematically transnational terrorism” as an alternative avenue of future research.

Causality Another interesting aspect of the terrorism
economy relationship that has been largely disregarded by empirical research
and which may partly explain the inconclusiveness of this very evidence
is the issue of causation. Plausibly, and consistent with parts of the empirical findings reviewed above, economic conditions may determine terrorist activity. However, terrorism may also have a negative causal effect on these very conditions. For instance, Piazza (2009, p. 406
407) argues that “[…] scholars and analysts who link terrorism with poor levels of economic development fail

40

DANIEL MEIERRIEKS

to carefully distinguish cause from effect. In countries that have suffered long bouts of terrorist activity, it is possible that poverty is the result of terrorism rather than the cause.” As noted earlier, terrorists aim at economic destabilization (as an intermediate goal) to succeed in achieving ultimate sociopolitical goals (e.g., a redistribution of wealth or territory). Economic destabilization is expected to improve the terrorists’ bargaining position by inflicting costs upon the government. Here, the government has to weigh the cost of making political concessions against the present and anticipated costs of a prolonged terrorist campaign
for example, in the form of reduced economic activity
that result from the decision of not negotiating with terrorists (Sandler & Enders, 2008). Terrorism may impair economic activity directly by destroying an economy’s human and physical capital stock through destruction (loss of lives, property damage, etc.) and indirectly as markets and economic agents react to it. For instance, terrorism is anticipated to hurt certain sectors of an economy (e.g., tourism) that are particularly exposed to it (Bird, Blomberg, & Hess, 2008). Indeed, a number of empirical studies finds that terrorism is detrimental to capital and trade flows (Abadie & Gardeazabal, 2008; Nitsch & Schumacher, 2004) and further macroeconomic variables such as consumption, investment, and public spending (e.g., Crain & Crain, 2006; Gaibulloev & Sandler, 2008). Such effects help to explain why several studies find that terrorism is detrimental to economic growth on subnational, national, and global levels (e.g., Abadie & Gardeazabal, 2003; Crain & Crain, 2006; Gaibulloev & Sandler, 2008, 2011).9 Ultimately, these findings indicate that the terrorism
economy relationship may be more complex than commonly believed. For one, this calls for intensified research that specifically focuses on causality. Indeed, first empirical research on the causal relationship between economic variables and terrorism suggests that it may be rather complex (e.g., Araz-Takay, Arin, & Omay, 2009; Gries, Krieger, & Meierrieks, 2011; Meierrieks & Gries, 2012, 2013). For example, Araz-Takay et al. (2009) find that terrorism and economic activity in Turkey are bidirectionally and nonlinearly related. They find that causation runs in both directions and that terrorism is more harmful to economic activity when economic growth is strong, while economic activity affects terrorism only when economic growth is poor. For another, complex causal links between terrorism and the economy call for a more careful empirical design when the economic roots of terrorism are considered. For example, Sandler and Enders (2008, p. 25) note that “[…] economic conditions may be both a root cause of terrorism and

Economic Determinants of Terrorism

41

a consequence of terrorism,” so that “[…] a researcher must be prudent to test and/or correct for a potential endogeneity bias.” Indeed, econometrically reverse causation
as a source of endogeneity, that is, a situation where the explanatory variables are not independent of the error term, thereby violating one assumption of the classical linear regression model where the covariance of the explanatory variable and the error term ought to be equal to zero
may produce biased and inconsistent parameter estimates and affect hypotheses tests in settings where such feedback exists (e.g., Greene, 2012, pp. 259
263). Still, while the dangers of endogeneity are known, the potential influence of endogeneity on analyses of the economic roots of terrorism remains largely neglected. One exception is Abadie (2006), who shows that once the dependence of economic development on terrorism is accounted for there is no longer a statistically significant effect of economic development on the emergence of terrorism. He uses an instrumental variable approach, treating the economic variables in his model as endogenous and instrumenting them with landlock, arguing that variation in geographic landlock affects terrorism only through its effect on income levels (where landlocked countries are poorer due to higher transportation costs and lower levels of economic integration). While this approach seems to properly account for reverse causation, it is only feasible in a cross-national study (due to the time-invariance of the instrument) such as Abadie’s (2006). Yet, for cross-national time-series analyses other instruments and further analytical tools to successfully control for endogeneity are much-needed. This is yet another avenue of future research.

Further Issues Homogeneity in the causes of terrorism, differences in roots of domestic and transnational terrorism, and a disregard of the challenges of endogeneity/ reverse causation may (partly) explain why there is no empirical consensus on the role of economic variables in the emergence of terrorism. This list is, of course, by no means exhaustive. This section ends by briefly bringing up further issues that are relevant to the discussion of the terrorism
economy nexus and may also merit future research. First, future research may try to more thoroughly differentiate between terrorism and civil war. As hinted at above, there is robust empirical evidence that economic success (e.g., in the form of higher income levels) makes countries less likely to experience a civil war (e.g., Fearon & Laitin,

42

DANIEL MEIERRIEKS

2003; Sambanis, 2002). What is more, Findley and Young (2012) find that terrorism oftentimes overlaps with civil war. For instance, many insurgent groups (e.g., the Sendero Luminoso, the SL, in Peru) have used guerilla warfare in the hinterland, while resorting to terrorism in a country’s urban centres. Given these findings, it may be the case that terrorism and civil war have similar roots when terrorism is merely used as a strategy within a civil war. Clearly, this calls for research that examines more closely the interactions between distinct types of political violence. Alternatively, future research may try to clearly differentiate between terrorism and other forms of political violence. For instance, Sanchez-Cuenca and De La Calle (2012) argue that open civil wars always involve territorial control by insurgents, while terrorism is always a clandestine activity without territorial control. They contend that many insurgent groups in Latin America (e.g., the Fuerzas Armadas Revolucionarias de Colombia, the FARC) have been participants of a civil war, while violent groups in the West (e.g., the Action Directe, AD, in France) have operated as terrorist groups. Using this approach, Sanchez-Cuenca and De La Calle (2012) show that open insurgencies are more likely in poor countries, while clandestine terrorist activity is more likely in countries with higher levels of per capita income. Interestingly, Sanchez-Cuenca and De La Calle (2012) interpret per capita income as a measure of state capacity rather than socioeconomic development. Fearon and Laitin (2003, p. 80) also argue that higher per capita income corresponds to a higher level of state capacity (in terms of infrastructure, bureaucracy, etc.), where higher levels of state capacity make a civil war less likely. However, while state capacity deters large-scale conflicts (civil wars), it may actually be positively related to lower-scale violence (terrorism). When insurgents face a sufficiently strong enemy (in terms of state capacity), they do not challenge him openly (civil war) but resort to highly asymmetric warfare (terrorism) which is the costefficient means to violently voice dissent under such circumstances (Sanchez-Cuenca & De La Calle, 2012). This line of reasoning calls for future research that more thoroughly considers the interaction between economic development and state capacity. So far, economic development is measured by (real) GDP per capita (e.g., Abadie, 2006; Basuchoudhary & Shughart, 2010; Kurrild-Klitgaard et al., 2006; Krueger & Laitin, 2008) and usually interpreted as an indicator of foregone earnings that determines the opportunity costs of terrorism associated with the level of socioeconomic underdevelopment (e.g., Abadie, 2006, p. 51). Disregarding the potentially detrimental effects of socioeconomic underdevelopment and state capacity
which are usually measured by the same indicator (per capita income)
could

Economic Determinants of Terrorism

43

have made it rather difficult to identify the isolated effect of socioeconomic conditions on terrorism in previous empirical efforts. Finally, the large-N evidence reviewed above should be complemented by more research on individual (active) participation in terrorism and terrorist groups.10 Interestingly, several empirical studies on the micro determinants of terrorism suggest that a higher standard of living is positively associated with the individual decision to participate in terrorism (e.g., Berrebi, 2007; Krueger & Maleckova, 2003). For instance, Krueger and Maleckova (2003) find that terrorists active in the Arab
Israeli conflict are in fact fairly well-off and educated. Yet, the related evidence does not necessarily imply that national economic conditions do not matter (or are positively linked to the emergence of terrorism). For instance, as noted by Krueger and Maleckova (2003), well-off and educated individuals may nevertheless become terrorists
for example, as a “revolutionary vanguard” in a left-wing fashion
to voice dissent over a country’s general (unfavorable) socioeconomic situation. This calls for more cautious empirical approaches to differentiate between the relationship of individual and collective (economic) conditions with terrorism. In other words, it is not necessarily appropriate to draw conclusions on the interaction between national economic conditions and terrorism by using information and statistical analyses on the relationship between one’s individual socioeconomic status and terrorist participation.

CONCLUSION Is terrorism rooted in poor economic conditions? This contribution provided an overview of the academic discussion of the nexus between economic factors and terrorism. It showed that there is a theoretical (rational-action) foundation to the popular claim that terrorism is more common in poor countries. A review of the existing empirical research on the terrorism
economy nexus showed, however, that the empirical evidence on the economic determinants of terrorism remains inconclusive. Some studies find that economic conditions (directly and indirectly) matter to terrorism (e.g., Blomberg & Hess, 2008a, 2008b; Burgoon, 2006; Freytag et al., 2011). Yet, a plurality of studies suggest that noneconomic factors (e.g., demographic pressures, foreign policy, state failure) are more important to the emergence of terrorism (e.g., Abadie, 2006; Krueger & Maleckova, 2003; Pape, 2003; Piazza, 2006; Walsh & Piazza, 2010). What is more, these studies offer

44

DANIEL MEIERRIEKS

alternative (rational-choice) explanations that plausibly relate these noneconomic factors to the emergence of terrorism through their effect on the terrorists’ calculus. From a theoretical point of view, economic explanations of terrorism are therefore not superior to noneconomic ones. Empirically, these competing explanation approaches call for prudent statistical modeling. In sum, given the current state of research, it is unlikely that economic conditions are universal determinants of terrorism. However, it is similarly unlikely that unfavorable economic factors do not at all matter to terrorism. This contribution argued that future research may help to better understand the economy
terrorism nexus by, inter alia, more thoroughly considering (i) the (under-analyzed) influence of ideology, (ii) differences in the causes of domestic and transnational terrorism, and (iii) the issue of causation and endogeneity when examining the role of economic fundamentals in terrorism. Such approaches are likely to provide a fuller
probably, more complex
picture of terrorism
economy nexus that is likely to complement the existing research on the determinants of terrorism. Policywise, the results of this contribution suggest that economic development is no panacea for terrorism and should thereby not be prioritized. At the same time, however, economic underdevelopment, political grievances (repression, discrimination, etc.) and further demographic (identity conflict, youth burdens, etc.), and international (foreign policy, globalization) pressures tend to reinforce each other in vicious circles (Sen, 2008), thereby exacerbating the risk of terrorist conflict. This implies that the influence of economic development on the genesis of terrorism should not be completely disregarded.

NOTES 1. Cohen-Almagor (1991) also discusses the role of violence and terrorism in the writings of Marx, Engels, and Lenin. He similarly concludes that these writers considered terrorist violence to be an acceptable instrument to promote social change, where social conflict itself was attributed to economic antagonisms between societal classes. 2. Indeed, a number of separatist and religious terrorist groups have simultaneously embraced left-wing ideas geared toward socioeconomic redistribution. For instance, the Provisional Irish Republican Army (PIRA) fought for the unification of Northern Ireland with the rest of Ireland as a socialist republic (Shughart, 2006). As another example, the People’s Mujahedin of Iran (MEK) followed an ideology of Islamic Marxism. 3. The GTD defines terrorism as the intentional use of violence (or threat thereof) by a subnational (i.e., non-state) actor outside the context of legitimate

Economic Determinants of Terrorism

45

warfare activities (e.g., a civil war), where this act must be aimed at attaining a political, economic, religious, or social goal, with the intention to coerce, intimidate, or convey some other message to a larger audience than the immediate victims. 4. Note that this section in parts draws on material used in the survey of Schneider, Bru¨ck, and Meierrieks (2012). 5. Krueger and Maleckova (2003) show that rational-choice models of terrorism can be considered as an extension of the economic analysis of crime. Economic models of crime (e.g., Becker, 1968) suggest that criminals are rational actors who maximize their utility and are subject to a calculus that involves the costs of criminal activity (e.g., from the risk of punishment), its benefits (e.g., the “wage” a crime pays) and its opportunity costs (e.g., foregone earnings from noncriminal economic activity). 6. However, these datasets provide information
if known
on the name of involved terrorist groups, which makes it possible to backtrack their ideological profile. For instance, this is how Kis-Katos et al. (2012) develop their dataset. 7. Commonly, the relation between domestic and transnational terrorist attacks is assumed to be between ca. 3:1 to 4:1 (Enders, Sandler, & Gaibulloev, 2011; Sanchez-Cuenca & De La Calle, 2009). 8. Enders et al. (2011) process raw data from the GTD, which itself does not differentiate between domestic and transnational terrorism, to identify domestic terrorist incidents and minimize data inconsistencies associated with the GTD. 9. Reviews of the literature on the economic effects of terrorism are also given in Bird et al. (2008), Sandler and Enders (2008), and Schneider et al. (2012). 10. See, for example, Abrahms (2008), Sageman (2008), Mousseau (2011), and Caruso and Schneider (2013) as interesting examples of such research.

ACKNOWLEDGMENTS Parts of this contribution are based on my Ph.D. dissertation. I want to thank my scientific advisors Thomas Gries, Manfred Kraft, and Tim Krieger for generously supporting this research. I also want to thank Friedrich Schneider and Tilman Bru¨ck for their cooperation on a joint survey on the economics of counter-terrorism. This survey is also used for parts of this contribution. Finally, I want to thank Andrea Locatelli and Raul Caruso for further insightful comments on an earlier version of this contribution.

REFERENCES Abadie, A. (2006). Poverty, political freedom, and the roots of terrorism. American Economic Review, 96(2), 50
56. Abadie, A., & Gardeazabal, J. (2003). The economic costs of conflict: A case study of the Basque Country. American Economic Review, 93(1), 113–132.

46

DANIEL MEIERRIEKS

Abadie, A., & Gardeazabal, J. (2008). Terrorism and the world economy. European Economic Review, 52(1), 1
27. Abrahms, M. (2008). What terrorists really want: Terrorist motives and counterterrorism strategies. International Security, 32(4), 78
105. Araz-Takay, B., Arin, K. P., & Omay, T. (2009). The endogenous and non-linear relationship between terrorism and economic performance: Turkish evidence. Defence and Peace Economics, 20(1), 1
10. Bapat, N. A. (2007). The internationalization of terrorist campaigns. Conflict Management and Peace Science, 24(4), 265
280. Basuchoudhary, A., & Shughart, W. F. (2010). On ethnic conflict and the origins of transnational terrorism. Defence and Peace Economics, 21(1), 65
87. Becker, G. (1968). Crime and punishment: An economic approach. Journal of Political Economy, 76(2), 169
217. Bergesen, A. J., & Lizardo, O. (2004). International terrorism and the world-system. Sociological Theory, 22(1), 38
52. Bernholz, P. (2004). Supreme values as a basis for terror. European Journal of Political Economy, 20(2), 317
333. Berrebi, C. (2007). Evidence about the link between education, poverty and terrorism among Palestinians. Peace Economics, Peace Science and Public Policy, 13(1), 1
36. Bird, G., Blomberg, S. B., & Hess, G. D. (2008). International terrorism: Causes, consequences and cures. The World Economy, 31(2), 255
274. Blomberg, S. B., & Hess, G. D. (2008a). The lexus and the olive branch: Globalization, democratization and terrorism. In P. Keefer & N. Loayza (Eds.), Terrorism, economic development, and political openness (pp. 116
147). New York, NY: Cambridge University Press. Blomberg, S. B., & Hess, G. D. (2008b). From (no) butter to guns? Understanding the economic role in transnational terrorism. In P. Keefer & N. Loayza (Eds.), Terrorism, economic development, and political openness (pp. 83
115). New York, NY: Cambridge University Press. Blomberg, S. B., Hess, G. D., & Weerapana, A. (2004a). Economic conditions and terrorism. European Journal of Political Economy, 20(2), 463
478. Blomberg, S. B., Hess, G. D., & Weerapana, A. (2004b). An economic model of terrorism. Conflict Management and Peace Science, 21(1), 17
28. Burgoon, B. (2006). On welfare and terror: Social welfare policies and political
economic roots of terrorism. Journal of Conflict Resolution, 50(2), 176
203. Bush, G. W. (2002). Remarks by the President on global development. Speech at the InterAmerican Development Bank. March 14, 2002. Caplan, B. (2006). Terrorism: The relevance of the rational choice model. Public Choice, 128(1
2), 91
107. Caruso, R., & Schneider, F. (2011). The socio-economic determinants of terrorism and political violence in Western Europe (1994
2007). European Journal of Political Economy, 27(1), 37
49. Caruso R., & Schneider F. (2013). Brutality of Jihadist terrorism: A contest theory perspective and empirical evidence in the period 2002
2010. Journal of Policy Modeling, 35(5), 685
696. Retrieved from http://www.sciencedirect.com/science/article/pii/S0161893813000124 Choi, S.-W. (2010). Fighting terrorism through the rule of law? Journal of Conflict Resolution, 54(6), 940
966. Cohen-Almagor, R. (1991). Foundations of violence in the writings of Marx, Engels and Lenin. Terrorism and Political Violence, 3(2), 1
24.

Economic Determinants of Terrorism

47

Collier, P., & Hoeffler, A. (1998). On economic causes of civil war. Oxford Economic Papers, 50(4), 563
573. Conrad, J. (2011). Interstate rivalry and terrorism: An unprobed link. Journal of Conflict Resolution, 55(4), 529
555. Crain, N. V., & Crain, W. M. (2006). Terrorized economies. Public Choice, 128(1
2), 317
349. Drakos, K. (2011). Security economics: A guide for data availability and needs. Defence and Peace Economics, 22(2), 147
159. Dreher, A., & Gassebner, M. (2008). Does political proximity to the U.S. cause terror? Economics Letters, 99(1), 27
29. Enders, W., Sandler, T., & Gaibulloev, K. (2011). Domestic versus transnational terrorism: Data, decomposition, and dynamics. Journal of Peace Research, 48(3), 319
337. Fearon, J. D., & Laitin, D. D. (2003). Ethnicity, insurgency and civil war. American Political Science Review, 93(1), 75
90. Findley, M. G., & Young, J. K. (2012). Terrorism and civil war: A spatial and temporal approach to a conceptual problem. Perspectives on Politics, 10(2), 285
305. Freytag, A., Kru¨ger, J. J., Meierrieks, D., & Schneider, F. (2011). The origins of terrorism. Cross-country estimates of socio-economic determinants of terrorism. European Journal of Political Economy, 27(1), 5
16. Gaibulloev, K., & Sandler, T. (2008). Growth consequences of terrorism in Western Europe. Kyklos, 61(3), 411
424. Gaibulloev, K., & Sandler, T. (2011). The adverse effect of transnational and domestic terrorism on growth in Africa. Journal of Peace Research, 48(3), 355
371. Gassebner, M., & Luechinger, S. (2011). Lock, stock, and barrel: A comprehensive assessment of the determinants of terror. Public Choice, 149(3
4), 235
261. Greene, W. H. (2012). Econometric analysis. Harlow: Pearson Education Limited. Gries, T., Krieger, T., & Meierrieks, D. (2011). Causal linkages between domestic terrorism and economic growth. Defence and Peace Economics, 22(5), 493
508. Jai, L. L. (2001). Getting at the roots of terrorism. Christian Science Monitor, December 10, 2001. Kis-Katos, K., Liebert, H., & Schulze, G. G. (2011). On the origins of domestic and international terrorism. European Journal of Political Economy, 27(1), 17
38. Kis-Katos, K., Liebert, H., & Schulze, G. G. (2012). On the heterogeneity of terror. Mimeo. Krieger, T., & Meierrieks, D. (2010). Terrorism in the worlds of welfare capitalism. Journal of Conflict Resolution, 54(6), 902
939. Krieger, T., & Meierrieks, D. (2011). What causes terrorism? Public Choice, 147(1
2), 3
27. Krueger, A. B., & Laitin, D. D. (2008). Kto kogo? A cross-country study of the origins and targets of terrorism. In P. Keefer & N. Loayza (Eds.), Terrorism, economic development, and political openness (pp. 148
173). New York, NY: Cambridge University Press. Krueger, A. B., & Maleckova, J. (2003). Education, poverty and terrorism: Is there a causal connection? Journal of Economic Perspectives, 17(4), 119
144. Kurrild-Klitgaard, P., Justesen, M. K., & Klemmensen, R. (2006). The political economy of freedom, democracy and transnational terrorism. Public Choice, 128(1), 289
315. LaFree, G., & Dugan, L. (2007). Introducing the global terrorism database. Terrorism and Political Violence, 19(2), 181
204. Lai, B. (2007). Draining the swamp: An empirical investigation of the production of transnational terrorism, 1968
1998. Conflict Management and Peace Science, 24(4), 297
310.

48

DANIEL MEIERRIEKS

Lenin, V. I. (1906). Guerilla warfare. Proletary. September 30, 1906. Li, Q., & Schaub, D. (2004). Economic globalization and transnational terrorism: A pooled time-series analysis. Journal of Conflict Resolution, 48(2), 230
258. Lizardo, O. (2006). The effect of economic and cultural globalization on anti-U.S. transnational terrorism 1971
2000. Journal of World-Systems Research, 12(1), 149
186. Lopez, G. (1988). Terrorism in Latin America. In M. Stohl (Ed.), The politics of terrorism (pp. 497
524). New York, NY: Marcel Dekker. Meierrieks, D. (2012). Rooted in urban poverty? Failed modernization and terrorism. Peace Economics, Peace Science and Public Policy, 18(3), 1
11. Meierrieks, D., & Gries, T. (2012). Economic performance and terrorist activity in Latin America. Defence and Peace Economics, 23(5), 447
470. Meierrieks, D., & Gries, T. (2013). Causality between terrorism and economic growth. Journal of Peace Research, 50(1), 91
104. Mickolus, E. F., Sandler, T., Murdock, J. M., & Fleming, P. (1993). International Terrorism: Attributes of terrorist events 1988
1991 (ITERATE 4). Dunn Loring: Vineyard Software. Mousseau, M. (2011). Urban poverty and support for Islamist terror: Survey results of Muslims in fourteen countries. Journal of Peace Research, 48(1), 35
47. Muller, E. N., & Weede, E. (1990). Cross-national variation in political violence: A rational action approach. Journal of Conflict Resolution, 34(4), 624
651. Nitsch, V., & Schumacher, D. (2004). Terrorism and international trade: An empirical investigation. European Journal of Political Economy, 20(2), 423
433. O’Brien, S. P. (1996). Foreign policy crises and the resort to terrorism: A time-series analysis of conflict linkages. Journal of Conflict Resolution, 40(2), 320
335. Pape, R. A. (2003). The strategic role of suicide terrorism. American Political Science Review, 97(3), 343
361. Paul, C. (2010). As a fish swims in the sea: Relationships between factors contributing to support for terrorist or insurgent groups. Studies in Conflict & Terrorism, 33(6), 488
510. Piazza, J. A. (2006). Rooted in poverty? Terrorism, poor economic development, and social cleavages. Terrorism and Political Violence, 18(1), 159
177. Piazza, J. A. (2009). Economic development, poorly managed political conflict and terrorism in India. Studies in Conflict and Terrorism, 32(5), 406
419. Piazza, J. A. (2011). Poverty, minority discrimination, and domestic terrorism. Journal of Peace Research, 48(3), 339
353. Plu¨mper, T., & Neumeyer, E. (2010). The friend of my enemy is my enemy: International alliance and international terrorism. European Journal of Political Research, 49(1), 75
96. Rapoport, D. C. (2004). The four waves of modern terrorism. In A. K. Cronin & J. M. Ludes (Eds.), Attacking terrorism: Elements of a grand strategy (pp. 46
73). Washington, DC: Georgetown University Press. Ray, J. L. (2005). Constructing multivariate analysis (of dangerous dyads). Conflict Management and Peace Science, 22(4), 277
292. Robison, K. K., Crenshaw, E. M., & Jenkins, J. C. (2006). Ideologies of violence: The social origins of Islamist and Leftist transnational terrorism. Social Forces, 84(4), 2009
2026. Sageman, M. (2008). Leaderless Jihad: Terror networks in the twenty-first century. Philadelphia: The University of Pennsylvania Press. Sambanis, N. (2002). A review of recent advances and future directions in the quantitative literature on civil war. Defence and Peace Economics, 13(3), 215
243. Sanchez-Cuenca, I., & De La Calle, L. (2009). Domestic terrorism: The hidden side of political violence. Annual Review of Political Science, 12(1), 31
49.

Economic Determinants of Terrorism

49

Sanchez-Cuenca, I., & De La Calle, L. (2012). Rebels without a territory: An analysis of nonterritorial conflicts in the world, 1970
1997. Journal of Conflict Resolution, 56(4), 580
603. Sandler, T., & Enders, W. (2004). An economic perspective on transnational terrorism. European Journal of Political Economy, 20(2), 301
316. Sandler, T., & Enders, W. (2008). Economic consequences of terrorism in developed and developing countries: An overview. In P. Keefer & N. Loayza (Eds.), Terrorism, economic development, and political openness (pp. 17
47). New York, NY: Cambridge University Press. Santos Bravo, A. B., & Mendes Dias, C. M. (2006). An empirical analysis of terrorism: Deprivation, Islamism and geopolitical factors. Defence and Peace Economics, 17(4), 329
341. Savun, B., & Phillips, B. J. (2009). Democracy, foreign policy, and terrorism. Journal of Conflict Resolution, 53(6), 878
904. Schelling, T. C. (1991). What purposes can “international terrorism” serve? In R. G. Frey & C. C. Morris (Eds.), Violence, terrorism, and justice (pp. 18
32). Cambridge: Cambridge University Press. Schneider, F., Bru¨ck, T., & Meierrieks, D. (2012). The economics of counter-terrorism: A survey. Mimeo. Sen, A. (2008). Violence, identity and poverty. Journal of Peace Research, 45(1), 5
15. Shughart, W. F. (2006). An analytical history of terrorism, 1945
2000. Public Choice, 128(1
2), 7
39. Sobek, D., & Braithwaite, A. (2005). Victims of success: American dominance and terrorism. Conflict Management and Peace Science, 22(2), 135
148. Urdal, H. (2006). A clash of generations? Youth bulges and political violence. International Studies Quarterly, 50(3), 607
629. Victoroff, J. (2005). The mind of the terrorist: A review and critique of psychological approaches. Journal of Conflict Resolution, 49(1), 3
42. Walsh, J. I., & Piazza, J. A. (2010). Why respecting physical integrity rights reduces terrorism. Comparative Political Studies, 43(5), 551
577. World Bank. (2012). World Bank development indicators. Washington, DC: World Bank. Young, J. K., & Dugan, L. (2011). Veto players and terror. Journal of Peace Research, 48(1), 19
33. Zimmermann, E. (2011). Globalization and terrorism. European Journal of Political Economy, 27(1), 152
161.

CHAPTER 3 MODELING TERROR ATTACKS: A CROSS-NATIONAL, OUT-OF-SAMPLE STUDY Ryan Bakker, Daniel W. Hill, Jr. and Will H. Moore ABSTRACT Purpose
The purpose of this study is to assess the ability of a theoretically motivated statistical model to accurately forecast annual, national counts of terror attacks out-of-sample. Methodology/approach
Bayesian multi-level models, classification analysis, marginal calibration plots Findings
We find that the model forecasts reasonably well, but conclude that its overall performance suggests that it is not ready for use in policy planning. This is likely due to the coarse temporal and spatial aggregation of the data. Research limitations/implications
The implications of this study are that social scientists should devote more effort into evaluating the predictive power of their statistical models, and that annual, national data on

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 51
68 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022008

51

52

RYAN BAKKER ET AL.

violent conflict are probably too coarse to provide useful information for policy planning. Originality/value of paper
The primary value of our modeling effort is to provide a baseline against which to evaluate the performance of more region- and country-specific models to be developed in the future. Keywords: Dissent; events; forecasting; repression; terror; violence

INTRODUCTION There has been a recent surge in interest in forecasting conflict events (e.g., Bagozzi, 2011; Blair, Blattman, & Hartman, 2012; Brandt, Freeman, & Schrodt, 2011a, 2011b; Clauset, Young, & Gleditsch, 2007; Goldstone et al., 2010; Hegre, Karlsen, Nyga˚rd, Strand, & Urdal, 2013; Tikuisis, Carment, & Samy, 2013; Ulfelder, 2012; Ward, Greenhill, & Bakke, 2010; Ward et al., 2012; Zammit-Mangiona, Dewar, Kadirkamanat, & Sanguinettia, 2012). Yet there is a past as well: efforts to forecast conflict events such as domestic terror attacks has a considerable history (e.g., Andriole & Hopple, 1984; Andriole & Young, 1977; Gurr & Lichbach, 1979, 1986; Gurr & Moore, 1997),1 though interest waned rather considerably from the mid-1980s through the 1990s. This recent rise is almost certainly linked to the US government’s support of such research (e.g., O’Brien, 2010), and the decline from the 1980s through the early 1990s was likely due to the US government’s disinterest in such efforts. In this chapter we report a forecasting effort that uses a Bayesian, multilevel, Poisson-log normal mixture model of the number of terror attacks that occurred in each country in the world over the period from 1970 to 1996 (Moore, Bakker, & Hill, 2011). To specify the independent variables we developed a theoretical account that, unlike previous efforts, places the competition between dissidents and the state at the center of attention. Krieger and Meierrieks (2011) provide a thorough review of cross-national statistical studies of terror attacks and Gassebner and Luechinger (2011) take a “run a million regressions” approach to determine what independent variables have robust relationships with the number of terror attacks countries experience. Both studies produce the same list of structural characteristics of economies, polities, and societies that populate existing large-N studies. Neither reports variables that measure the behavior of the state or the non-terror behavior of dissident groups. In other words, all existing

Modeling Terror Attacks: A Cross-National, Out-of-Sample Study

53

cross-national research studies terror attacks as a function of structure alone. The primary contribution of our work is to bring behavior in. We argue that terror is a tactic, selected among several, by some dissident groups as they press their claims against the state. If we statistically model the annual number of such attacks per country, then we are modeling the outcome of decisions by dissident groups to use their resources to produce terror attacks rather than use protest demonstration tactics, or guerrilla tactics that seek to degrade the state’s coercive capacity (rather than terrorize citizens and state officials).2 Below we briefly describe the major claims of the theoretical argument, and also sketch both the empirical approach, as well as the findings of our previous study. There are at least two good reasons to explore a statistical model’s ability to forecast.3 First, the fit of a model to the data is important. In political science we have witnessed a valuable shift away from “maximizing explained variance” that dominated some analyses during the 1980s to a focus on the substantive impact of a variable upon the outcome. Further, hypothesis testing is the primary goal of most academic researchers who estimate statistical models. Thus, attention justifiably focuses upon hypothesis testing and the substantive interpretation of the size of effects. Yet, as Ward et al. (2010) explain, most research tends to overfocus on hypothesis testing to the exclusion of predictive power. That is, while the move away from maximizing measures of in-sample fit (e.g., “explained variance”) has been fruitful, a complete lack of attention to how well a model fits the data is lamentable. In particular, few political scientists appear to recognize the value of forecasting to assess the ability of one’s model to predict the outcome of interest. Analyzing the individual impact of each variable has value, but is incomplete; the accuracy of the model’s predictions remains outside of our view. Further, as shown in Ward et al. (2010), variables that produce statistically significant coefficient estimates do not necessarily improve model fit. In other work we tested our hypotheses and assessed the impact of individual variables (Moore et al., 2011). Had we stopped there we would have left important work undone. Here we extend the assessment of the impact of the variables in our model, shifting away from their individual effects on terror attacks to their ability to collectively predict which countries experience terror attacks. Second, forecasts can speak to policy makers who might use forecast output in decision support (e.g., Andriole & Hopple, 1984; O’Brien, 2010; Shellman & Moore, 2008). Yet, what policy makers might plausibly be interested in forecasts of the number of domestic terror attacks per country-year? To be able to forecast well the number of terror attacks that

54

RYAN BAKKER ET AL.

are expected to occur over a two-week period in a given city could be of obvious use to municipal officials, but at such a gross level of spatial and temporal aggregation as the country-year it is less obvious what type of decision support value exists. Yet, the US military, for example, has a number of units that engage in country level forecasting at the annual level, indeed using those forecasts to then aggregate further to both regional and semi-decade levels. In the US military those forecasts have primarily been made by expert human judgment without the support of statistical models, but that has begun to change in recent years. On the civilian side of the US military the J5 Strategic Plans and Policy; Deputy Undersecretary of Defense for Strategy Plans and Forces; the Guidance for the Employment of the Force (GEF); and Guidance for the Development of the Force (GDF) might plausibly find value in forecasts of expected terror attacks per country in the coming year. On the uniformed officer side the Combatant Commands, which coordinate planning across all of the services, would be the most likely consumer. Outside of the military the long range planning divisions of intelligence agencies as well as foreign policy bureaucracies could potentially have an interest. The remainder of the chapter proceeds as follows. We begin by sketching the empirical model in Moore et al. (2011): the theoretical account we developed, the statistical model we employed, and the results we obtained. The second section describes the approach we take to produce forecasts using the model in Moore et al. (2011). We present the results of the forecast in the third section, then wrap up with a conclusion.

A BRIEF TOUR OF THE EMPIRICAL MODEL The theory we use to specify the statistical model that we use to forecast out-of-sample below is described in Moore et al. (2011). That paper also provides details of our Bayesian, multi-level statistical model, the measurement models we estimated, and the data sources used. Here we briefly describe those, referring the reader to our other paper for details. Our knowledge of the set of concepts that influence the number of terror attacks experienced by different countries is rudimentary. Existing work on the incidence of terror focuses upon the structural characteristics of polities, economies, and societies, and fails to place competition between dissidents and states center stage. For example, Krieger and Meierrieks (2011) and Gassebner and Luechinger (2011) each scour the literature for studies

Modeling Terror Attacks: A Cross-National, Out-of-Sample Study

55

that examine the cross-national incidence of terror attacks. Krieger and Meierrieks offer a verbal summary of the various findings, whereas Gassebner and Luechinger identify all of the independent variables used in those studies, and include them in their own statistical meta-analysis which estimates regressions using every possible combination of variables from published studies. These two articles review effectively the same list of studies and thus consider the same lists of independent variables. As we describe below, we adopt a Bayesian measurement model approach to variable construction, and thus focus our attention on concepts. Details are available in Moore et al. (2011), and here we list the economic, political, and social structures included in our model, along with the expected sign. Only one of these variables is due to us: the coercive bureaucracy of the state (e.g., Gurr, 1988). It is interesting that the state’s coercive capacity (i.e., the size of the bureaucratic apparatus responsible for coercion) has gone missing from this literature. Its absence seems to underscore the extent to which the existing literature fails to conceptualize oppositional terror as a tactic selected by dissident groups in response to the structural circumstances they face. • Political Institutions Contestation (+) Participation (+) Veto players (+) Freedom of association, press, and speech (+) Coercive bureaucracy of the state (−) • Structural Characteristics of Economy and Society Strength of civil society (−) Macro-economic performance (−) Ethno-linguistic composition of society (+) Physical quality of life (−) Population size (+) As noted above, our project proposes a theoretical framework that places competition between dissident groups and government center stage. More specifically, we argue that the number of terror events one will observe in a given country-year will be a function of the behavior of states and dissidents as they compete for control of the policy agenda, and that the behavior of those actors will be influenced by the structural settings in which they are located. Details of that argument are fully developed in Moore et al. (2011). As we are interested here in whether the statistical model implied by the theory forecasts well out-of-sample, we list below

56

RYAN BAKKER ET AL.

the behavior we expected to have an impact on the annual number of terror attacks in a given country. • Behavior Coercive state behavior (+) Nonviolent behavior of dissidents (−) Violent (non-terror) behavior of dissidents (+) Violent international conflict (+) We measured most of the concepts above using estimated latent variables (e.g., Treier & Jackman, 2008).4 That is, for each country-year in our data, we collected a variety of observable indicators across a wide range of latent concepts. Given the variance in both the number of observable indicators and the number of years available per country in our data, the Bayesian strategy allows us to estimate latent variables for country-years that other studies treat as missing. Unlike the bulk of existing cross-national analyses of the incidence of terror, we study the sum of transnational and domestic terror attacks rather than only transnational ones. An absence of data has hampered the cross-national statistical study of domestic terror attacks: in their review of the large-N, cross-national literature Krieger and Meierrieks (2011, p. 19) could find only three studies that pool both domestic and transnational attacks, and two studies that examine only domestic terror attacks. An estimated 85% of terror attacks are domestic (LaFree & Dugan, 2007b), which means that what we know about the covariates of terror attacks is based on roughly 15% of the attacks that have taken place. The existence of the Global Terrorism Database (GTD) (LaFree & Dugan, 2007a) makes an analysis of the covariates of all terror attacks (i.e., both domestic and transnational) possible, and we use the total number of domestic and transnational terror attacks as the dependent variable in our study. Turning to estimation of the model’s parameters, it is conventional in cross-national statistical analyses of this type to assume a single baseline number of terror attacks for each country in the sample (i.e., estimate a single intercept for all countries). Of course, it is not necessary to impose this restrictive assumption, so we relaxed it. More specifically, we estimated a Poisson-lognormal mixture model with random effects for country-specific overdispersion (Cameron & Trivedi, 1998; Winkelmann, 2008). The Poisson-lognormal model allows us to estimate the expected number of counts per country-year as a multiplicative function of the covariates described above and a country-specific normally distributed random effect. We estimated the model’s parameters via Monte Carlo Markov Chain

Modeling Terror Attacks: A Cross-National, Out-of-Sample Study

57

(MCMC) in WinBUGS (Lunn, Thomas, Best, & Spiegelhalter, 2000).5 As noted above, details can be found in Moore et al. (2011). An analysis of the deviance information criterion statistic for (1) a null model, (2) a model of only the structural covariates, (3) a model of only the behavioral variables, and (4) the full model demonstrated that each provided a better in-sample fit than the model(s) listed before it. That provided strong evidence in favor of our claim that existing models
which include only structural variables
are missing most of the action by failing to theoretically and empirically model the impact of the behavior of both dissidents and the state upon the number of terror attacks a country experiences. The specific hypothesis tests provided further support for our theoretical approach. Three of the four behavior variables had the expected impact: violent (non-terror) dissent, government coercion, and international war each had a strong, positive association with the number of terror attacks. All five of the political institutions variables exhibited a substantial impact on the use of terror, though both Veto and Freedom of Association/Press/ Religion/Speech had considerably smaller impacts than Contestation, Participation, and the Coercive Capacity of the state. Finally, we found that three of the five socioeconomic structure variables exhibited a systematic relationship with the number of terror attacks occurring within a given country-year: strength of civil society and macroeconomic performance were negatively related to the number of terror attacks, and population had a positive relationship. Having briefly described the model and results reported in Moore et al. (2011), we turn our attention to assessing the model’s forecasting performance.

DESIGNING THE FORECAST Political scientists have yet to broadly embrace forecasting, and as a result much of what is done suffers weaknesses that are less prevalent in other fields (Brandt et al., 2011a). We wish to determine whether our model’s estimates are capable of being put to use to predict the number of terror attacks a given country will experience in a given year. We take as given that we will know the values for our X variables. This is frequently described as an out-of-sample forecast because the analyst first estimates the model parameters on a given sample (e.g., 1970
1996) and then plugs in the values of X for a year outside of the given sample (e.g., 1997), and

58

RYAN BAKKER ET AL.

multiplies those values against the coefficient estimates to obtain forecast values of the dependent variable for that year. A different, and considerably more challenging, forecasting task requires the analyst to also predict the values of X (e.g., Hegre et al., 2013). We do not undertake such an effort here. Some researchers use a model’s fit statistics, such as the root mean square error (RMSE), when conducting an out-of-sample forecast analysis. Brandt et al. (2011a), among others, criticize this. It is particularly unhelpful in our case because our dependent variable is an event count, which is an integer with a lower bound of zero. Due to this an error-based statistic such as RMSE is unhelpful because residuals are asymmetric: due to the presence of the lower bound under-predictions will be systematically smaller than overpredictions. Thus, measures like RMSE are biased in such settings. Researchers who assess the fit of binary models such as logit or probit are familiar with the type of approach that we adopt here. In those studies it is common to identify the percentage of correct predictions, and to further break down the analysis and examine the percentage of false negatives and false positives. This is sometimes called a classification analysis, and we employ it below. We used data from 1970 to 1996 to estimate the model parameter posterior densities, and then multiplied the values for the Xi1997 vector by those parameters to produce forecasts. We saved 1,000 draws from our estimated posterior densities, thus permitting us to construct credible intervals as a measure of the uncertainty of the forecast. That is, for each of the 153 countries we produced 1,000 forecasts, rounded them to their nearest integer value, and record both the median as well as the values at the .25 and .975 percentiles, thus recording both the central tendency of the forecast and our uncertainty about that value.6 That uncertainty is called the credible interval, and it is literally the range over which a given percentage (in our case, 95%) of the forecasts are distributed.7 We are interested in assessing the resulting 153 predictions in two ways: a binary prediction of whether there would be zero or more than zero attacks, and a count prediction which is the integer value. There are 153 countries in our dataset for 1997. Fig. 1 is a frequency distribution: 70 countries had zero attacks; and 83 had one or more, ranging from a single attack to over 350. To begin we evaluate the accuracy of binary predictions. For those we recode the predicted value of the count such that any predicted value greater than 0 is reassigned the value of 1. To elaborate, we first multiply the Xi1997 vector by the posterior parameter densities reported in Moore

Modeling Terror Attacks: A Cross-National, Out-of-Sample Study

59

150

Frequency

100

50

0 0

Fig. 1.

100

200 300 400 Number of Attacks

500

Terror Attacks in 153 Countries, 1997.

et al. (2011). We then calculate, for each observation in the 1997 data, the probability that the country will experience at least one terror attack.8 Using these probabilities we generate a binary prediction for each observation that is equal to one if the probability of experiencing at least one terror attack is >.5. We then report the following for our binary forecast: a classification analysis and a receiver operating characteristic (ROC) curve assessment. We also make use of the predicted count densities that we produced for each observation: to offer an intuitive visual account, we calculate (for each country) probabilities for each count in the empirical distribution, and produce a marginal calibration plot that records both the observed frequencies for each count and the predicted frequencies for each count. Finally, we examine both the mean and the 95% credible intervals of the density for each prediction, and evaluate whether the observed value from 1997 falls within the credible interval, as well as whether it equals the (rounded) median.

RESULTS We begin with the integer count forecasts, which are considerably more demanding than the binary ones. To begin, we produce a marginal calibration plot (Czado, Gneiting, & Held, 2009) to assess how well our count

60

RYAN BAKKER ET AL.

predictions map onto the observed distribution of terror attacks. The concept of marginal calibration is discussed in technical detail in Gneiting, Balabdaoui, and Raftery (2007), but the intuition behind the plot is straightforward: the frequency distribution of the predictions a count model produces should match fairly closely the observed frequency distribution. Fig. 2 presents a marginal calibration plot for our model’s count predictions. The plot displays the observed frequencies for 1997 and the predicted frequencies for each count value (to ease presentation, we bin higher counts).9 These predicted frequencies can then be compared to the actual number of cases that fall into each value/bin. Fig. 2 provides a rather positive assessment of the forecast, but as we see below, figures such as these can lead one to view a forecast more positively than one should. Table 1 reports both the proportion of observations where the predicted count fell within the 95% credible intervals,10 and the proportion of observations where the predicted count (rounded to the nearest integer) was equal to the observed count (precisely correct). The scores are 39%

50

Predicted Observed

30

20

10

5 (2 ] 5, 50 ] (5 0, 10 0 (1 00 ] ,2 (2 00] 00 ,3 (3 00] 00 ,5 11 ]

5]

5, 2 (1

5

(5 ,1

4

3

2

1

0 0

Frequency

40

Number of Attacks

Fig. 2.

Marginal Calibration Plot for Count Predictions.

61

Modeling Terror Attacks: A Cross-National, Out-of-Sample Study

Table 1. Count Prediction Performance for 1997. Metric

Proportion

Observed within CI Precisely Correct

Table 2. Obs = 0 Obs > 0 Total

.39 .22

Binary Prediction Classifications for 1997. Pred = 0

Pred > 0

Total

25 16 41

17 95 112

42 111 153

and 22%, respectively. This puts the forecast performance in a very different light. To be sure, a precisely correct integer count is a hyper-exacting standard, but note that the forecast does not improve terribly much when we relax it to include the 95% credible interval. That said, this is a pooled cross-national time-series model of the number of terror attacks that would occur in each country in the world, in 1997. Though the model is not ready for use in a policy making setting, we are frankly surprised that the model performs as well as it does. Note, for example, that we do not include a lagged value of the dependent variable in our model. Further, a different approach that would undoubtedly produce superior out-of-sample forecasts, would estimate each country individually, or perhaps estimate small groups of neighboring countries. That is, the purpose of our effort is not to identify the best out-of-sample forecasts of the number of terror attacks countries observed in 1997. Our goals are instead to (1) for the purpose of theory building add an assessment of the out-ofsample forecasting performance to the work reported in Moore et al. (2011), and (2) offer those forecasts as a baseline against which other efforts might be judged. In other words, both theoretical efforts and efforts that simply wish to produce useful forecasts of country’s expected number of terror attacks will be able to compare their efforts to ours, and will want to be sure that their models outperform this one. Let us turn our attention to the assessment of the binary forecasts. Table 2 reports a classification matrix that compares these binary predictions to a binary version of the value observed in 1997 (coded 1 if the count is > 0), with the row percentage reported below the frequency. Observations in the upper left cell correspond to accurate predictions of zero attacks,

62

RYAN BAKKER ET AL.

Table 3.

Binary Prediction Performance for 1997.

Metric Accuracy (correctly classified) Sensitivity (correct | obs > 0) Specificity (correct | obs = 0) Positive PV (correct | pred > 0) Negative PV (correct | pred = 0)

Proportion .78 .86 .60 .85 .61

while those in the lower right cell correspond to accurate positive predictions. The off diagonals are then false positives (upper right cell) and false negatives (lower left cell). We turn to Table 3 which shows several additional measures of predictive performance for the binary predictions. Accuracy is the proportion correctly classified; sensitivity is the proportion of ones correctly classified; and specificity the proportion of zeros correctly classified. Positive Predictive Value (PPV, sometimes called “precision”) is the proportion of positive predictions that actually experienced at least one attack, and Negative Predictive Value (NPV) is the proportion of 0 predictions that did not experience any attacks. Overall, our model correctly predicts 78% of the country-years in 1997. Note further that it does a better job forecasting countries that do experience an attack than it does those that do not: we correctly forecast 86% of the countries that experienced at least one terror attack in 1997, but only 60% of the countries that experienced no attacks. In any forecasting endeavor one needs to “choose one’s poison,” as it were, and decide whether to favor models that have greater sensitivity (i.e., produce relatively more false negatives) or specificity (i.e., those that produce relatively more false positives). Given politicians’ interest in retaining office, in a decision support setting we suspect that for the purpose of forecasting terror attacks models with greater sensitivity (correctly predicted ones) would be preferred. Our 86% sensitivity score is certainly respectable (e.g., 80% is sometimes used as a minimum performance threshold). However, in light of the fact that, in the 1997 data, far more countries experience terror attacks (73%) than do not (27%), the model’s ability to predict attacks is not as impressive. If one were to simply predict that all countries experience an attack this would produce an accurate prediction 73% of the time. We discuss these issues in detail below. Fig. 3 reports a ROC curve based on the binary predictions. The ROC curve plots the probability of accurately predicting a 1 (i.e., sensitivity) against the probability of inaccurately predicting a 0 (the false positive

63

0.6 0.4 0.2

Pr(Correct | Obs > 0)

0.8

1.0

Modeling Terror Attacks: A Cross-National, Out-of-Sample Study

Model (AUC = 0.80)

0.0

Random Prediction

0.0

0.2

0.4 0.6 0.8 Pr(Incorrect | Obs = 0)

Fig. 3.

ROC Plot for Binary Predictions.

1.0

rate, or 1 minus specificity) at different threshold values for a positive prediction. The benefit of the ROC curve is that it does not force one to (arbitrarily) choose a probability threshold for classifying positive predictions (which above, as in most studies, is .5). For a good model this threshold can be varied and the model loses very little predictive power (i.e., the curve stays toward the upper left corner of the graph, meaning the model does a good job of predicting both 0s and 1s). The area under the curve (AUC) is often used as a metric for comparing different models. For our model the AUC is .80, while the AUC for a random guess, represented by the dashed line, is .5 (Fig. 3). To summarize, we have shown that while this particular model forecasts binary outcomes well, it nonetheless is not a very good candidate for use in an applied, policy making setting. While its binary sensitivity is respectable for such a setting, this is largely a function of the relatively large number of countries that experienced at least one terror attack during the year for which we produce a forecast. Further, its specificity in inadequate, and its accuracy is not very impressive. Yet the model does have value as a global baseline that can be useful not only to scholars and practitioners developing and testing theory about the incidence of terror attacks, but also forecasts of that incidence. To this point we have concentrated our

64

RYAN BAKKER ET AL.

attention on annual incidence per country, but we hasten to observe that the baseline value of our model extends beyond the country-year. That is, should one be studying or forecasting using a more fine-grained temporal period, such as the quarter, month, or week, one’s model should be able to outperform ours. To the extent that it does not, the variables that we include serve as candidates that one should consider. Some readers might object that data availability may foreclose that possibility, and while that is certainly true, we must note that it would not for data that are only available at annual temporal aggregation: multi-level modeling makes it very attractive/feasible to collect/model data at multiple levels of temporal and/ or spatial aggregation (e.g., Gelman & Hill, 2006).

CONCLUSION During 2012 there were two high-profile attacks on political science, and each made its case in large part by arguing that political scientists could not predict anything.11 First, Arizona Congressman Jeff Flake introduced an amendment to the bill that funded the US National Science Foundation that would have eliminated the funding for the political science program.12 The amendment eventually failed, but it kicked off a buzz well beyond professional political science associations, attracting attention not only in national US magazines and newspapers, but even earning in response an editorial in Nature.13 Then in March 2013 Senator Thomas Coburn successfully attached an amendment to a continuing resolution bill that terminated financial support for the political science program through October of 2013.14 Second, in June of 2012 The New York Times published an Op-Ed written by a political science Professor at Northwestern University, Jacqueline Stevens.15 Professor Stevens argues that though she believes that statistical analyses in political science have become almost de rigueur in the field, the King is parading naked: unlike in other fields where statistical modelers produce useful forecasts, statistical political science models are so poor that few even try, and those that do fail miserably. Three things strike us about these episodes. First, they are critiques grounded in ignorance: neither Congressman Flake, who studied political science as an undergraduate at Brigham Young University, nor Professor Stevens has the training to properly evaluate the status of forecasting in political science. Indeed, in their public statements each points to unexpected events such as the Arab Spring and asserts that political scientists

Modeling Terror Attacks: A Cross-National, Out-of-Sample Study

65

failed to predict them. We do not bother to engage such silliness, other than to speculate that they likely rail against their local weather forecasts as well. Second, and more importantly, both Flake and Stevens are quite correct to wonder why, given the common use of statistical models in political science, they do not see more forecasts. As we noted in the introduction to this chapter, many researchers do not use any goodness-of-fit measures to evaluate their statistical models, and (out-of-sample) forecasts of violent conflict are exceedingly rare, the recent surge notwithstanding. What we did not note there is that the subfield of violent conflict is arguably the vanguard of political science forecasters. Put plainly, political scientists have paid inadequate attention to forecasting for far too long. Indeed, if one scours the World Wide Web for blog posts by political scientists responding to the attacks by Flake and Stevens one will encounter many that argue that the arguments fail because political scientists should not be forecasting as it is not part and parcel to scientific inquiry! Research presentations at recent meetings of the Political Methodology Society, and the articles that follow such gatherings, belie this opinion, but it remains widely enough held to be a concern. Progress is being made, but we need to recognize that there is a long way to go. Third, and most promisingly, this sort of backlash is the sort of thing one might anticipate as progress is made that changes the status quo. When a field of inquiry is in its youth it will grow in many different directions as it develops a proper identity. Though the unsuccessful Flake amendment of 2012 and the successful Coburn amendment of 2013 are in no small part to be understood through the lens of partisan budget battles, Stevens’s broadside was little more than an attempt by a member of a coalition that is losing influence to diminish the influence of one that has recently displaced it. Returning to the specifics of our study, to conclude their effort to forecast intranational violent crises of interest, Tikuisis et al. (2013, p. 1) write that “while model accuracy … usually exceeded 90 percent, the model did not generate sufficiently high and consistent precision … and [sensitivity] … for practical use.” We reach the same conclusion regarding our model’s ability to forecast: the precision is too low to recommend its use in a policy setting. Nevertheless, in addition to contributing to what we hope becomes the increasingly common use of out-of-sample forecasting among political scientists who use statistical models to test hypotheses, we contend that our model can serve as a useful baseline for those who wish to forecast terror incidence both at the country-year level of aggregation and lower levels.

66

RYAN BAKKER ET AL.

NOTES 1. Ward et al. (2012) provide a considerably more detailed discussion. 2. We adopt Schelling’s classic definition of terror as “violence intended to coerce the enemy rather than to weaken him militarily” (Schelling, 1960, p. 7). 3. See Ward et al. (2012) for a similar argument. 4. We were unable to identify enough variables to identify latent variable models for nonviolent and violent (non-terror) dissent, and thus create additive indices for both. These both come from Banks (2009). We use the number of antigovernment demonstrations and general strikes to measure nonviolent dissent, and the number of riots and acts of guerrilla warfare to measure violent, non-terror dissent. We also use single measures of ethno-linguistic composition of society and violent international conflict. 5. WinBUGS is available online at http://www.mrc-bsu.cam.ac.uk/bugs/ 6. We followed a similar procedure, described below, to generate the binary predictions. 7. A confidence interval is similar in that it is the range over which one expects a given percent of the distribution to lie. Put differently, a credible interval is the Bayesian paradigm’s version of a confidence interval. 8. For a Poisson model the probability of any particular count m can be calculated as: [exp(−μ) × μm]/m!, where μ is exp(xβ). For a zero count this is exp(−μ). The probability of a count larger than zero is then 1 − exp(−μ). 9. We calculated these predictions by creating a predicted probability for each value of the terror count, from 0 to the maximum observed count, for each observation in the 1997 data, and then calculated the mean of each probability across the observations. The mean probability for each count serves as the predicted proportion of the sample that takes on that particular value. This proportion is multiplied by the sample size to create a predicted frequency. 10. A credible interval is effectively the Bayesian equivalent of a confidence interval in frequentist statistics. 11. Both occurred prior to Nate Silver and others’ celebrated success forecasting the electoral college outcome in the 2012 US Presidential election. 12. H.AMDT.1058 (A025), http://thomas.loc.gov/cgi-bin/bdquery/D?d112:44:./ temp/bdeckr:: 13. Available online at http://www.nature.com/nature/journal/v487/n7407/full/ 487271a.html 14. The Economist “Political Science Research: Singled Out” online at http:// www.economist.com/blogs/democracyinamerica/2013/03/political-science-research 15. Available online at http://www.nytimes.com/2012/06/24/opinion/sunday/ political-scientists-are-lousy-forecasters.html

ACKNOWLEDGMENTS The project upon which this chapter draws has received helpful feedback from audiences at the Centre for the Study of Civil War Working Group at

Modeling Terror Attacks: A Cross-National, Out-of-Sample Study

67

Peace Research Institute, Oslo; the Departments of Political Science at Essex University, Florida State University, and University of North Texas; and the annual meetings of the International Studies Association (2010, New Orleans, LA) and Peace Science Society (2008, Claremont, CA). We appreciate comments and feedback on the project from John Ahlquist, Dave Armstrong, Jason Barabas, Bethany Barrett, Bill Berry, Halvard Buhaug, David Cunningham, Kathleen Cunningham, Scott Edwards, Mike Findley, James Forest, Scott Gates, Jeff Gill, Havard Hegre, Brian Lai, Jim Piazza, Chris Reenock, Dave Siegel, Havard Strand, and Joe Young.

REFERENCES Andriole, S. J., & Hopple, G. W. (1984). The rise and fall of event data: From basic research to applied use in the US department of defense. International Interactions, 10(3
4), 293
309. Andriole, S. J., & Young, R. A. (1977). Toward the development of an integrated crisis warning system. International Studies Quarterly, 21(1), 107
150. Bagozzi, B. (2011). Forecasting civil conflict with zero-inflated count models. Working Paper. Banks, A. S. (2009). Cross-national time-series data archive. Binghamton: Center for Social Analysis. Retrieved from http://www.databanks.sitehosting.net/ Blair, R., Blattman, C., & Hartman, A. (2012). Predicting local-level violence. Presented at the Northeast Political Methodology meeting, New York, NY. Brandt, P. T., Freeman, J. R., & Schrodt, P. A. (2011a). Racing horses: Constructing and evaluating forecasts in political science. Annual meeting of the Political Methodology Society. Retrieved from http://eventdata.psu.edu/papers.dir/RHMethods20110720.pdf Brandt, P. T., Freeman, J. R., & Schrodt, P. A. (2011b). Real time, time series forecasting of inter- and intra-state political conflict. Conflict Management and Peace Science, 28(1), 41
64. Cameron, A. C., & Trivedi, P. K. (1998). Regression analysis of count data. Cambridge: Cambridge University Press. Clauset, A., Young, M., & Gleditsch, K. S. (2007). On the frequency of severe terrorist events. Journal of Conflict Resolution, 51(1), 58
87. Czado, C., Gneiting, T., & Held, L. (2009). Predictive model assessment for count data. Biometrics, 65(4), 1254
1261. Gassebner, M., & Luechinger, S. (2011). Lock, stock, and barrel: A comprehensive assessment of the determinants of terror. Public Choice, 149(3), 235
261. Gelman, A., & Hill, J. (2006). Data analysis using regression and multilevel/hierarchical models. New York, NY: Cambridge University Press. Gneiting, T., Balabdaoui, F., & Raftery, A. E. (2007). Probabilistic forecasts, calibration and sharpness. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 69(2), 243
268. Goldstone, J. A., Bates, R. H., Epstein, D. L., Gurr, T. R., Lustik, M., Marshall, M. G., … Woodward, M. (2010). A global model for forecasting political instability. American Journal of Political Science, 54(1), 190
208.

68

RYAN BAKKER ET AL.

Gurr, T. R. (1988). War, revolution, and the growth of the coercive state. Comparative Political Studies, 21(1), 45
65. Gurr, T. R., & Lichbach, M. I. (1979). Forecasting domestic political conflict. In J. David Singer & M. D. Wallace (Eds.), To augur well: Early warning indicators in world politics (pp. 153
193). Beverly Hills, CA: Sage. Gurr, T. R., & Lichbach, M. I. (1986). Forecasting internal conflict. Comparative Political Studies, 19(1), 3
38. Gurr, T. R., & Moore, W. H. (1997). Ethnopolitical rebellion: A cross-sectional analysis of the 1980s with risk assessment for the 1990s. American Journal of Political Science, 41(4), 1079
1103. Hegre, H., Karlsen, J., Nyga˚rd, H. M., Strand, H., & Urdal, H. (2013). Predicting armed conflict, 2010
2050. International Studies Quarterly, 57(2), 250
270. June 2013. Krieger, T., & Meierrieks, D. (2011). What causes terrorism? Public Choice, 147, 3
27. Retrieved from http://dx.doi.org/10.1007/s11127-010-9601-1 LaFree, G., & Dugan, L. (2007a). Global terrorism database, 1971
1999: Codebook and data documentation. Ann Arbor, MI: Interuniversity Consortium for Political and Social Research. ICPSR Study # 04586. LaFree, G., & Dugan, L. (2007b). Introducing the global terrorism database. Terrorism and Political Violence, 19(2), 181
204. Lunn, D. J., Thomas, A., Best, N., & Spiegelhalter, D. (2000). WinBUGS
A Bayesian modelling framework: Concepts, structure, and extensibility. Statistics and Computing, 10(4), 325
337. Moore, W. H., Bakker, R., & Hill, D. W. (2011). How much terror? Dissidents, governments, institutions and the cross-national study of terror attacks. SSRN Working Paper. Retrieved from http://ssrn.com/abstract = 1977262 O’Brien, S. P. (2010). Crisis early warning and decision support: Contemporary approaches and thoughts on future research. International Studies Review, 12(1), 87
104. Schelling, T. C. (1960). The strategy of conflict. Cambridge: Harvard University Press. Shellman, S. M., & Moore, W. H. (2008). The new school of political conflict and terror: From attributes to behavior and why policymakers will care. Perspectives on Terrorism, 2(12), 3
10. Tikuisis, P., Carment, D., & Samy, Y. (2013). Prediction of intrastate conflict using state structural factors and events data. Journal of Conflict Resolution, 57(3), 410
444. Treier, S., & Jackman, S. (2008). Democracy as a latent variable. American Journal of Political Science, 52(1), 201
217. Ulfelder, J. (2012). Forecasting onsets of mass killing. Presented at the Northeast Political Methodology Meeting, New York, NY. Ward, M. D., Greenhill, B. D., & Bakke, K. M. (2010). The perils of policy by p-value: Predicting civil conflicts. Journal of Peace Research, 47(4), 363
375. Ward, M. D., Metternich, N. W., Carrington, C., Doref, C., Gallop, M., Hollenbach, F. M., … Weschle, S. (2012). Stepping into the future: The next generation of crisis forecasting models. Working Paper. Retrieved from http://mdwardlab.com/biblio/learning-past-and-steppingfuture-next-generation-crisis-prediction Winkelmann, R. (2008). Econometric analysis of count data. Berlin: Springer Verlag. Zammit-Mangiona, A., Dewar, M., Kadirkamanat, V., & Sanguinettia, G. (2012). Point process modelling of the Afghan War Diary. Proceedings of the National Academy of Sciences, 109(31), 12414
12419.

CHAPTER 4 MEASURING SECURITY Tilman Bru¨ck, Olaf J. de Groot and Neil T. N. Ferguson ABSTRACT Purpose
The purpose of this study is to define the interactions that determine how secure a society is from terrorism and to propose a method for measuring the threat of terrorism in an objective and spatio-temporally comparable manner. Methodology/approach
Game-theoretic analysis of the determinants of security and discussion of how to implement these interactions into a measure of security. Findings
We show that governments concerned with popularity have an incentive to over-invest in security and that, in certain situations, this leads to a deterioration in net security position. Our discussion provides an implementable means for measuring the levels of threat and protection, as well as individuals’ perceptions of both, which we propose can be combined into an objective and scientific measure of security. Research limitations/implications
The implication for researchers is the suggestion that efficiency, as well as scale of counter-terrorism, is important in determining a country’s overall security position.

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 69
95 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022009

69

TILMAN BRU¨CK ET AL.

70

Furthermore, we suggest that individuals’ perceptions are at least as important in determining suitable counter-terrorism policy as objective measures of protection and threat. The limitations of this research are found in the vast data requirements that any attempt to measure security will need. Originality/value of the chapter
We propose the first method for objectively measuring the net security position of a country, using economic and econometric means. Keywords: Transnational terrorism; domestic terrorism; measuring security; perceptions; indicator methodology; counter-terrorism policy

INTRODUCTION Are we secure? This basic question is one that policymakers and citizens face every day; however, it is also incredibly difficult to answer. Firstly, one must answer a whole range of preliminary questions; ‘Are we secure from what?’ ‘For whom are we providing security?’ and even ‘What is security?’ (Buzan & Hansen, 2009). Is a feeling of security a substitute for an absolute form of security or do they breed each other? In his Turing Award Lecture, Adi Shamir proposed three laws of security, the first of which is: ‘Absolutely secure systems do not exist.’1 If this is the case, we then must ask, ‘What is our benchmark level of security?’ Should one argue that the provision of the maximum possible amount of security is optimal (Bru¨ck, 2005)? Most existing terrorism indicators, such as the American ‘traffic light’ system, which we will discuss later, do not answer these questions. In this contribution, we aim to explore and answer these questions in order to propose a scientific method for the measurement of security. Accordingly, we must first define the terms we will use throughout this chapter. We define terrorist threats as the level of threat that would occur should the level of protection be set equal to zero. That is, the level of terrorist threat is the sum of all terrorist attacks that are realised, as well as those that are deterred and prevented. Protection, it follows, are the efforts that (successfully) deter or foil threats. We, therefore, deem security to be a trade-off between threat and protection. In other words, security can be considered as the proportion of potential threat that is realised, relative to the efforts made to prevent it from occurring. Accordingly, whilst security

Measuring Security

71

and insecurity are related, we do not treat them, simply, as opposites. The notion of the perception of security can be considered in a similar way, except that the focus here is on an individual’s perceptions of threat and protection that determine the outcome. Before further exploring the issue, however, it is important to answer the primary question: Secure from what? In the literature, many definitions of terrorism arise. For the purposes of our discussion, we build on that of Enders and Sandler (2005), who state, ‘Terrorism is the premeditated use or threat to use violence by individuals or subnational groups in order to gain a political or social objective through the intimidation of a large audience beyond that of their immediate victims.’ We extend this notion by requiring a clear political or social objective, however, which further differentiates terrorism from hate crime. This notion is particularly pertinent in the cases of Mohamed Merah and Wade Michael Page, who under this definition, committed hate crimes, as opposed to acts of terrorism. Despite such concerns, however, we present a flexible method of measuring security that can easily be extended to account for such definitional issues. Similarly, we also suggest this methodology can overcome the current debate about the coding of certain terrorist incidents. For example, both the ITERATE database of transnational terrorist incidents and TWEED database of domestic incidents include IRA attacks on mainland Great Britain and elsewhere in Europe. Our cost-based modelling suggests that, so long as all relevant incidents are included, but not double-counted, such definitional issues do not arise. A second notion that requires explicit focus is that our methodology focuses on threats to the security status of individuals, rather than threats to governments, private organisations or the economy. Our focus, therefore, is not on the economic cost of terrorism, for example. In this respect, our measure is designed as a scientific and robust alternative to governmental terrorism indicators. This is not to suggest that governments and other actors do not play an important role; simply that threats to them do not necessarily directly affect the overall level of security, as we aim to measure it. Rather, it is the implications for individuals of threats made to other actors that we identify as important. We consider this definition broadly, however: we are concerned about more than physical personal safety and also consider threats that could impact on an individual’s way of life. Throughout this work, we view terrorism as a triangular interaction of three groups of actors, which we refer to as Terrorists, the State and Society. We postulate that the combined actions of these three social groupings offer a relatively complete outline of the determination of

72

TILMAN BRU¨CK ET AL.

the levels of security and insecurity, which in turn provides a strong underpinning for the theoretical section and measurement discussions that follow. The role of the State we deem as self-evident; it provides protection from threat and is a potential target for attacks. We propose that civilians can be the victims of attacks primarily aimed at governments
Anders Breivik’s bomb in Oslo, before the shootings at Utøya, being a pertinent example. We deem a joint role of the State’s protection measures
through the direct deterrence of threats from Terrorists and in making Society feel safe. Accordingly, we suggest that the State is aware, not only of its desire to minimise terrorist threats but also of its desire to draw support from Society. Similarly, the State must also be aware that its security measures can change the direction or intensity of terrorism (Enders & Sandler, 2005). All of these interactions play important roles in determining the general level of security and, therefore, carry important notions for how security can be measured. We define Society as both private individuals and private-sector businesses. Society plays a role in the selection of the State, via elections, for example. In a hypothetical situation of two political parties with differing approaches to counter-terrorism, society therefore defines the security strategy by choosing which party to elect. Similarly, Society’s perception of its own security will relate, at least in part, to the actions of the State. We also propose that Society can play an important direct role in the provision of security
vigilance on aircraft, for example, is cited as a contributory factor in the failure of the attempted attacks by Richard Reid and Umar Abdulmutallab. Another component of society’s perception of its security, however, is the realised threat from Terrorists, which provides another key interaction. Perhaps surprisingly, Terrorists are actually the most difficult of our three players to define; following the modelling of the likes of Becker (1968) as extended by Landes (1978), we could essentially regard every individual in society as a potential Terrorist. This seems like an extreme assumption to make, however. Whilst a focus on those who engaged, or would be willing to, in terrorism may provide a stronger focus, this also requires certain definitional clarifications. For our purposes, we define Terrorists as those who do, or who would be willing to, engage in acts of terrorism, given a prevailing set of economic and social circumstances, should the level of protection be set equal to zero. Accordingly, these groups of actors play a number of key roles in the determination of the level of security and insecurity. We fully discuss these interactions in the theoretical model. To further clarify the parameters

73

Measuring Security

within which we ground our analysis, we start with an overview of the existing literature on (in)security measurement. The third section proves the extensive theoretical model that underpins our ideas of security measurement, followed by a discussion on the practical interpretation of this model in a real-life scenario. The fifth section then discusses the desirable properties that an indicator of security should have and finally, we present our conclusions in the sixth section.

LITERATURE REVIEW Until 2011, the United States used a colour-coded terrorist threat indicator known as the Homeland Security Advisory System (HSAS), which has since been replaced with the National Terrorism Advisory System (NTAS). The HSAS ‘traffic light’ system has received significant criticism (Department of Homeland Security, 2009), mainly because it was not based on objective criteria and could be politically manipulated (Shapiro & Cohen, 2007). Given that the security threat never went below Level 3 (Code Yellow) on a five-point scale, there is also suggestion of ‘threat inflation’, where the level of threat can rise but, rarely, decreases. The new NTAS system is more straightforward and issues only two types of warning; concerning threats that are either imminent or elevated. Importantly, warnings are temporary and have an expiry date, thus avoiding permanently elevated levels of warning (Department of Homeland Security, 2011). One problem that remains is that the new measure still lacks an underlying theoretical basis. As a result, it is difficult to compare security across time and space. The trustworthiness of the indicator also remains a concern, as recognised by Chandler (2011). Furthermore, Chandler also recognises that, even when an objective measure does exist, there is a significant disconnect between the objectively measured level of security and the perceived level. We draw here on the substantial literature that has been developed in the field of cybersecurity. Although seeming slightly abstract, a number of analogues can be drawn between attempts to secure computer networks and attempts to protect a country from terrorism. Both aim to create a ‘system’ that is impenetrable to malevolent outside influence, for example. Similarly, those responsible for outside threats may attempt to directly attack or, somehow, bypass or penetrate the provided security. The latter effect is noted in Enders and Sandler (2006) who suggest provision of metal

74

TILMAN BRU¨CK ET AL.

detectors at airports deflected attacks to other targets, such as trains, suggesting that terrorists aim to bypass aircraft protection, rather than penetrate it. Similarly, the dynamic response of protection, in situations where systems are penetrated or bypassed, also seems applicable to defence against adaptive terrorist threats. Jansen (2009) sees two major aspects in security measurement in the cybersecurity field. He distinguishes between correctness, which denotes whether security measures have been implemented correctly, and effectiveness, which implies that security measures work in the manner in which they are intended. He also points out that the transformation from qualitative to quantitative measures, and vice versa, is often obscure and without theoretical basis. Finally, he argues that existing indicators often do not coincide with the stated security objectives. In other words, they do not always measure what policymakers would like them to measure. Berinato (2005) gives examples of easy-to-use security measures, of which two are widely applicable in a more general security setting. Firstly, ‘baseline defences coverage’: the percentage of equipment that is covered by a basic level of protection and, secondly, ‘patch latency’: the degree to which units are updated with the latest security equipment. Both of these ideas carry obvious analogues to protection from terrorism. Baseline defences coverage basically proposes a cut-off for protection and tests the degree to which systems (or people) are compliant, which fits with the notion of Farrow (2007) who suggests it may actually be rational for governments to leave certain targets unprotected. The second looks at the effectiveness of protection, as defined by Jansen (2009), and tests whether the use of security measures is optimal, whilst also considering the dynamic responses to new threats, such as changes to airplane carry-on luggage in the wake of potential liquid explosive attacks. Torgerson (2007) distinguishes between different types of security metrics in the field of communication systems. First, systems can be measured according to their weakness, indicating that a system has a known number of weaknesses and looks at the degree to which these weaknesses can be mitigated. Second, a protection-based metric looks at the intensity of protection in the system at hand. Third, a threat-based metric takes into account the capacities and strategies of adversaries. A problem with this metric is that there may be a difference between the perceived threat level according to the system owner and the actual underlying threat level. He concludes that an optimal security metric combines these different inputs. Schroeder (2010) distinguishes different metrics for security across geographical levels: security from war/conflict at the international level,

75

Measuring Security

internal security/safety at the national level and fundamental rights at the individual level. However, none of these measures can truly be considered measures of security, but rather as potential inputs when it comes to the realised intensity of threats to security. A different, more quantitative approach is taken by Al-Rodhan, Stoudmann, and Finaud (2006), who describe what they refer to as a Security Matrix. They propose five primary variables: economic, environmental, military, political and societal. Next, for each of these variables, they propose a set of sub-indicators that are each ranked on an ordinal 1
3 scale. The major issue is that this index imports the underlying problems from other indices and requires subjective judgement to translate data into the proposed ordinal scale. Putting aside the specific choice of underlying variables, the distinction between this Security Matrix and an index such as the State Fragility or the Human Development Index seems to be limited. Another alternative method is introduced by Grigg (2005) when discussing the concept of Pareto Security (albeit in the field of cryptology). He concludes that Pareto Security is reached when a system is protected from known and practical threats and when security cannot be further improved without imposing significant costs. This seems suggestive of diminishing marginal returns from security provision.

MODELLING SECURITY In this section, we introduce a game-theoretic model of insecurity. We aim to use this model to highlight the nature of the interactions that can occur between the strategic players and the role these action profiles play in determining the level of security. We aim to directly implement a number of notions we present in our qualitative discussion and to discuss the potential outcomes of security and their policy recommendations. Accordingly, we introduce the three players we defined earlier in this chapter: Terrorists, the State, which we characterise as an Incumbent government, and Society. We begin our analysis with a simple game between Terrorists and the Incumbent, before extending the game to account for the role of Society. The simple game presented in the first section below is a two-player, single-shot, simultaneous game, whilst in the more complete model that follows, we extend to two periods, with Terrorists and the Incumbent simultaneously interacting in the first stage and Society choosing whether or not to support the Incumbent, or some Opposition, in the second.

TILMAN BRU¨CK ET AL.

76

Single-Stage Game We propose an Incumbent that is interested in its welfare spending, subject to the level of taxation it generates. We presume the level and direction of welfare spending reflects the political ideology of the governing party. Insecurity is assumed to impact, negatively, on the Incumbent’s outcomes. In this manner, we also propose that the Incumbent does not directly benefit from security provision, but instead, benefits only from the impact it has on reducing the level of realised insecurity. Following our earlier notion that insecurity is the sum total of both realised and deterred acts of insecurity, and following the understanding that no society experiences zero acts of insecurity, we assume a net negative impact on the government from insecurity. In other words, no government, no matter how much energy it devotes to security, can deter or prevent every act of insecurity. We assume that the net impact of realised insecurity exhibits diminishing returns to scale. By this, we suggest that in a relatively secure state, one extra incident has a greater impact than an extra incident in a less secure state. A roadside bomb, for example, is likely to have a significantly greater negative impact in London than, say, Kabul. We present a quasi-linear utility function, with welfare spending acting as the nume´raire. The implication of these notions is that the presence of insecurity has a joint role in the determination of the Incumbent’s outcomes. Directly, it has the impact of, ceteris paribus, decreasing the Incumbent’s payoff from any given action profile. Secondly, however, it also infers the requirement for spending on security. As the Incumbent is bound by a budget constraint, which we (naively) assume to be equal to tax revenues, this security spending can only be funded by a substitution away from welfare spending. We make a final assumption that the Incumbent’s spending on security is subject to some effectiveness factor. We elaborate on the particular role of this notion in the section ‘Measuring Security in Practice’. Following the notion that a Terrorist is anyone who would be willing to engage in acts of terrorism, should the probability of getting caught, or the act being prevented, equal zero, we set up their problem as a traditional labour market utility function. We show Terrorists gain from engagement in a terrorist ‘labour market’ and from a Cobb
Douglas style function of welfare and wages from legal labour. Terrorists gain from engaging in the legal labour market (wages) and from engagement in terrorism. In both markets, returns are subject to ability and constraints (costs). Furthermore, we note that those who engage in terrorism cannot be prevented from

77

Measuring Security

benefitting from the Incumbent’s welfare transfers. We note, finally, that any legal labour market income will be subject to taxation, which is not the case for engagement in terrorism. We also show, however, that the benefits to terrorism are discounted by the degree and effectiveness of security provision. As with the Incumbent, we present a quasi-linear utility function. Accordingly, we formalise the functions of the Incumbent and the Terrorist as follows: ΠI = w − r α ΠT =

tβ þ wl ηs

ð1Þ ð2Þ

where w and t are the proportions of resources devoted to welfare and terrorism by the Incumbent and the Terrorists, respectively, whilst l is the time devoted to legal labour market activity and s is the effort devoted to security spending. r is the level of realised threat, whilst η is the effectiveness factor of security spending. By noting the relationship between both the level and the effectiveness of security, we note that the level of realised threat is simply the total devotion to terrorism (t) minus the degree of protection afforded. We therefore rewrite Eq. (1) as ΠI = w − ðt − ηsÞα

ð3Þ

The Incumbent has a budget that is determined by its taxation policies. We add to this a full-utilisation of resources assumption, such that the government spends all of its resources within this game. As stated before, any increase in security provision by the Incumbent must be offset by a corresponding decrease in welfare spending. We normalise the cost of security provision to 1 and generate a relative cost of security policies denoted as cW. Similarly, a Terrorist cannot spend 25 hours per day engaged in any kind of activity. Terrorists also suffer from an ability constraint, however, which acts as a cost of engagement in either the labour market or terrorism. Here, we normalise the cost of engaging in terrorism to 1 and generate the relative cost of legal labour market participation, cL. For notational simplicity, we also normalise the budget constraints of both players to 1. This allows us to write the budget constraints for the Incumbent and the Terrorist, respectively BCI = 1 − s − cW w

ð4Þ

TILMAN BRU¨CK ET AL.

78

BCT = 1 − t − cL l

ð5Þ

In order to generate the best-response functions, we employ the equimarginal principle, which states that outcomes are maximised when the ratio of the marginal benefits of each choice variable is equal to the ratio of marginal costs. As such, we partially differentiate Eq. (3) with respect to the proportion of resources spent on security (s) and divide this by the first-order partial of Eq. (3) with respect to welfare spending (w). We similarly approach Eq. (2), dividing the first-order partial with respect to terrorism (t) by the first-order partial with respect to legal activity (l). We take the first-order partials of Eqs. (4) and (5) with respect to security spending and illegal activity, respectively, in order to generate the marginal costs. This yields ηαðt − ηsÞα − 1 =

1 cW

βtβ − 1 1 = cL wηs

ð6Þ ð7Þ

Having generated the equi-marginal outcomes of Eqs. (2) and (3), we rearrange Eqs. (6) and (7) for our choice variables, providing the bestresponse functions s=

t − [1=ηαcW ]1 − α η


t=

βcL wηs

1=ð1 − βÞ

ð8Þ

ð9Þ

Inspection of the best responses presented in Eqs. (8) and (9) reveals a sensible set of comparative statics. The Incumbent’s desire to engage in security measures increases in the elasticity of this spending and in the effectiveness of the spending. It is also shown to be increasing in the relative cost of welfare. Furthermore, the nature of strategic interaction with the choices of Terrorists can also be seen and shows that it is optimal for the Incumbent to linearly increase their devotion to security as Terrorists devote more efforts to violence. Eq. (9) also exhibits sensible comparative statics. The desire to engage in terrorism is increasing in the relative returns to terrorism and decreasing

Measuring Security

79

in the costs of engagement. The relationship between an individual’s decision to engage in the legal labour market or acts of terrorism and the Incumbent’s security policies, however, is more complex than in the Incumbent’s problem. The relationship is shown to be non-linear, with a desire to decrease acts of terrorism, in response to an increase in security spending, when the initial level of security spending is low. When the level of spending on security is high, however, the opposite incentive prevails. This ‘U-shaped’ best response seems entirely sensible. At low levels of security spending, the direct impact of the increase in spending reduces returns to terrorism, without having significant knock-on effects on legal labour market outcomes. Accordingly, we refer to this as the ‘direct deterrent effect’. When the level of spending is high, however, there is a much more pronounced impact on returns from the legal labour market. Due to the Incumbent’s budget constraint, available income remains constant but the level of welfare transfers decreases as spending on security increases. At high levels of spending, therefore, the Incumbent has the impact of deterring engagement in the legal labour market. An alternative way of thinking about this, more in line with our qualitative discussion, is that the maintenance of high taxation, combined with low welfare transfers, provides incentives for individuals to defect from the legal to the illegal labour markets. This fits with our notion of those who ‘would engage in criminal activity, given the prevailing economic and social conditions’. We now turn our attention to a discussion of the equilibrium outcomes of this model, which are presented graphically in Fig. 1. In Fig. 1, the dotted line represents the U-shaped best response of Terrorists to changes in Incumbent security spending, whilst the dashed line represents the linearly increasing devotion to security that defines the Incumbent’s best response to changes in Terrorists’ devotion to violence. The two thick, solid lines are the budget constraints. We see that this model, potentially, supports two internal equilibria. The first equilibrium suggests a comparatively low level of both security provision and threat, whilst the second shows both a high level of security provision and threat. A further outcome can prevail where there is no security or insecurity, but this would require no incentive for any individual to engage in illegal activity, which we believe not to be a realisable outcome. The intuition behind this model shows that the level of security is not, necessarily, improved by simply increasing the level of security spending. This model not only shows that such policies can be ineffective but that they can actually be detrimental. Large transfers away from social welfare incentivise defection from the legal to the illegal labour market. This

TILMAN BRU¨CK ET AL.

Insecurity Devotion

80

Security Devotion

Fig. 1.

Best Responses of Devotion of Resources to Security on Insecurity.

outcome could be exacerbated by considering taxation as an endogenous variable. The greater the number of individuals that defect to the illegal labour market, the lower the taxation revenues, which further reduces welfare spending.

Two-Stage Game We now consider an extension to this model. To this end, we introduce the role of the Society depicted in our qualitative discussion. We show Society to be interested in maximising its own outcomes. To this end, it selects the government that it feels is most likely to maximise these outcomes. We depict the Incumbent, in accord with the previous section, as a party with a policy of a subjective and interactive engagement strategy with Terrorists. We introduce the Opposition as an alternative political party that has a pre-stated and credible policy. For simplicity, we assume that society has full information on the outcomes of the Opposition’s policy, although we accept that in reality, the opposite may be true. The first stage of this game plays out much in the same way it did in the simple, one-stage, case. In the second stage of this game, however, Society faces a dichotomous choice between voting for the Incumbent or for the Opposition and its pre-stated security policy. We depict a Society that gains from its own efforts in the legal labour market and from government welfare transfers, but which is averse to the threat of insecurity. Accordingly,

81

Measuring Security

Society directly benefits from the net level of security spending but loses out due to the general prevalence of terrorism and/or fear of terrorism; however, these individuals must pay tax, which discounts their returns. We use the simplifying assumption, following the government’s budget constraint, that taxation equals the spending on welfare and security at the individual level. To model this, we show each individual in this Society to have a subjective and individual aversion to terrorism. For mathematical ease and notational simplicity, we assume this factor to be uniformly distributed on the interval [0, 1] but the logic of the analysis equally holds for other distributions. This aversion can be considered as a so-called ‘fear factor’. An individual with a fear factor of zero is entirely unaffected by any insecurity threat; an individual with a factor of one, however, is strongly affected by the same threat. We suggest this connection of preferences represents our notions of subjective security. We formalise Society’s preferences over voting for the Opposition or Incumbent as follows: ΠV|O = Δ

ð10Þ

ΠV|I = l þ w þ s − δt − T = l − δt

ð11Þ

where Δ is Society’s outcome from the Opposition’s pre-stated policy; δ is the subjective ‘fear factor’; T is taxation (which is assumed to equal w + s); and l, w, s and t are as stated before. Individuals in this society, therefore, will vote for the Opposition when their outcome in Eq. (10) is greater than that from Eq. (11), given their personal fear factor. This formally occurs when: δ≥

Δ−l t

ð12Þ

Thus, each individual in society for whom Eq. (12) holds will choose to vote for the Incumbent and those for whom it does not will give their support to the Opposition. Given our earlier distributional assumption, this infers that a proportion of Society, 1 − δ, will vote for the Incumbent, whilst the remaining proportion, δ, will vote for the Opposition. The Incumbent is now shown to be concerned not only with its own welfare priorities and the prevailing security position but also with its popularity. Terrorists’ incentives are shown to be unaffected by notions of popularity, although we accept that this may not be generally applicable.

TILMAN BRU¨CK ET AL.

82

As the Incumbent does not, expressly, generate utility from its security policies, it assumes that its popularity is Society’s judgement on its welfare policies. We proxy popularity through the vote share at the election. Accordingly, the Incumbent’s outcomes from the election are shown to discount their returns from welfare spending. We formalise this new outcome as follows:
 Δ−l ΠI = ð1 − δÞw − ðt − ηsÞ = 1 − w − ði − ηsÞα t α

ð13Þ

As in Section 3.1, we solve Eq. (13) for the first-order conditions and rearrange for the best response of security spending. This yields the result depicted in Eq. (14). As the Agent’s problem is unchanged, we plug the result shown in Eq. (9) and depicted in Fig. 1 back into this extended analysis. Accordingly, we add the outcome of Eq. (13) to Fig. 1 in order to generate Fig. 2, which is depicted as the solid line s=

t − [ð1 − ðd − lÞ=tÞ1=ð1 − αÞ =ηαcW ] η

ð14Þ

Insecurity Devotion

As before, we see that the Incumbent has an incentive to increase security spending as a result of increased insecurity by the Terrorists. In this depiction, however, Society’s preferences are shown to further discount the Incumbent’s perception of the effectiveness of its security policies.

Security Devotion

Fig. 2.

Best Responses of Devotion of Resources to Security on Insecurity Under Public Pressure.

Measuring Security

83

Accordingly, it creates an incentive to spend a greater proportion of resources on security, given any prevailing level of insecurity. It should be noted, however, that the more popular a government is, the less pronounced this impact is. That said, the role of Society and of popularity provides an incentive for the Incumbent to increase its devotion to security spending. We believe that result should be uncontroversial
a society that lives in fear of insecurity is likely to put pressure on governments to increase its devotion of resources to measures that combat (or, at least, devote resources to measures society believe will combat) insecurity. This last finding is, perhaps, the most profound outcome, in terms of the complexity of measuring security. It does not take a huge leap of imagination to believe that the policies which make society feel safer are those that are most visible. Accordingly, the role Society plays in providing government incentives to provide inefficient or ineffective measures of security, potentially at great expense, is suggested. We have already shown that large substitutions away from welfare towards security spending can, perversely, increase the net level of threat. This result adds to this finding a more disturbing fact: those societies that deeply fear insecurity are those who pressure governments to engage in exactly these kinds of substitutions. Indeed, under equilibrium conditions, we see how such an outcome can be possible. In Fig. 2, we once more depict an outcome that supports two internal solutions and, as before, the utopian no-insecurity corner solution. The differences we see between these outcomes and the outcomes in Fig. 1, however, reveal both the positive and negative roles that society can play in determining their own security. In the lower of the two equilibria, the increase in security spending increases the deterrent effect this spending has on the prevailing level of insecurity, leading to a lower level of insecurity and a safer society. At the upper equilibrium, however, we depict a situation where the pressure for higher security spending demanded by society actually increases the level of threat. In this situation, the deterrent effect of security spending is significantly smaller than the indirect impact that comes from reductions in welfare transfers, which makes criminal activity more lucrative. Accordingly, this model offers the suggestion that governments and civilians both have important but complex roles to play in determining the level of security. In societies we would broadly consider to be ‘safe’
those where there is some level of realised threat and a relatively low level of security
governments can raise the level of security by increasing security spending. On the other hand, societies that are already unstable can be further destabilised by increases in devotion to security spending.

84

TILMAN BRU¨CK ET AL.

In both our simple and extended cases, the argument presented calls for increases in the efficiency and effectiveness of security measures as the most effective method of reducing the prevailing level of insecurity. In many of the cases presented here, governments can increase the level of security by improving the efficiency of security policies. That said, however, such government policies may not make society feel safer, which we argue is just as important a part of security provision as the prevention of threat. A society that does not feel safe is unlikely to continue to elect a government that does not make it feel safe, suggesting the potential moral hazard of shorttermism. Accordingly, a further policy recommendation, aimed at making society feel safe (or, indeed, aimed towards making society feel safer than it actually is), can be extrapolated from this model. In these situations, the net security position in society can be increased in a threefold manner. Firstly, the security that is provided is more effective, decreasing the returns from engaging in insecurity. Secondly, this deterrent effect is reinforced by allowing governments to devote a greater level of resources to welfare spending, increasing the opportunity cost of engaging in criminal activity. Finally, by a transfer of this knowledge of effectiveness, governments can make society feel safer, reducing the desire in society for inefficiently high security spending which we have shown can further destabilise an insecure situation.

MEASURING SECURITY IN PRACTICE Based on the preceding analysis, and following some of the ideas set out in the introduction, we propose an interpretation of how (in)security can be measured. We start from the premise that security is a dynamic interplay of three opposing forces: threats delivered by terrorists, protection provided against these threats and individuals’ perceptions of their own security. This is an important distinction when it comes to security, since it makes clear that security can both be increased through a decrease in threat or through an increase in the quality or quantity of protection and should be focused both on making individuals safer and making those individuals feel safer. Considering protection first, it is necessary to look at efficiency and effectiveness: at the margin, increasing protection is only as valuable as the amount of amount of threat it deters. We thus argue that the optimal level of protection is achieved when the marginal cost of perceived protection

85

Measuring Security

equals the marginal benefit of the deterred threat. This has an important implication for our indicator of threat: it includes both realised and deterred insecurity. It is important to note that the optimal level of security may be reached at the break-even point of marginal benefits and costs of security, but that an optimal level of security is obviously associated with a lower level of threat as well as a higher amount of deterred threat. The equilibrium level of security then depends on the proportion of threat deterred through protection and the amount of threat realised, as well as societal beliefs about the scale of threats deterred and realised. An additional advantage of looking at security in this way is that both threat and protection can be expressed directly in a denomination that is easily comparable: money. We define threat as the total per capita value of all threats deterred and threats realised. Protection has two components as well; the first of which is also the total per capita value of threats deterred and the second as the marginal efficiency of protection. In other words, we define efficiency in the economic terms as E=

MBS MCS

ð15Þ

As an example, if one hires a security guard at a cost of h50,000 and she stops crime worth h60,000, but fails to prevent crime worth h5,000, then total threat = 60,000 + 5,000 = 65,000. Efficiency is 60,000/50,000 = 1.2. We suggest this efficiency notion closely matches the ‘effectiveness of security’ parameter introduced in the theoretical setting in section ‘Modelling Security’. One thing that is not addressed yet is the role of perceived threat. After all, if people do not feel they are safe (even though they are), terrorists still impact significantly on daily life. The proportion of perceived threat is thus an important intermediary for establishing the overall level of (in)security.

Data Requirements The indicator that we propose for objectively measuring the level of security is highly data intensive. In this section, we describe the data needs, as well as potential sources for such data. The section is (naturally) divided up between threat, protection and perceived threat. A continuing consideration is pointed out by Jansen (2009) who argues that one must carefully consider whether indicators are leading, lagging or concurrent, as each

86

TILMAN BRU¨CK ET AL.

should be treated very differently. In general, threat indicators are often lagging or concurrent, whereas protection indicators may be more likely to be leading. Perceived threat, in this regard, is difficult to categorise. Threat Data on security threats can be expected to be partially available. The problem is identifying which of the data is indeed available, and which data is held by public sources or away from the public. The more immediate problem is to select which of the data is relevant and which of the data is not. The first threat that many people think of when considering the issue of security is terrorism. Terrorists have some political goal and aim to use violence to intimidate a wider population, or governments, into generating concessions towards these aims. Frequently, terrorist attacks use private citizens as a medium to obtain these goals. Measuring the level of security therefore requires extensive data on threats made by terrorists, actual occurrences of terrorist attacks, both successful and unsuccessful, as well as information on terrorism networks and financing structures (Haigner, Schneider, & Wakolbinger, 2012), which measures potential (future) terrorism. The latter two are necessary to be able to predict the potential future productivity of terrorist groups. An important issue to note in this case is that it is important to analyse all different sources of terrorism, be it Islamists, left-wing extremists, right-wing extremists, ethnic/regional separatists, etc. It is necessary to generate a full picture of the network of terrorists and their capacity to produce threats to security. We also note that any attempt to measure and index security may be susceptible to definitional grey areas. A threat like indiscriminate violence, for example, may fall under terrorism (when thinking of lone wolf terrorism) or crime (when thinking of gang violence or school shootings). However, since indiscriminate violence is highly visible, this is one issue where one may expect individuals to be particularly worried, despite an inherently low level of threat. The greatest challenge in the measurement of threat is the aggregation of different threat types into a single measure. One cannot simply add up the number of transnational attack, the number of domestic attacks and so on. We therefore propose that we use the data available to us to first estimate the probability of an individual being a victim of a particular threat. Doing so yields a threat map describing the probability of each threat taking place, taking into account both the realised outcomes and the existing network of actors aiming to execute the threat. The second step is to estimate

Measuring Security

87

the average cost of each of the threats to an individual citizen. To do so, we aim to monetise the negative financial losses of fatalities, (permanent) injuries, lost work days and so forth. In this respect, we propose to follow the literature on daily-adjusted life years (Murray, 1994). Accordingly, we define our threat map as the probability that an individual will be the victim of a particular kind of terrorist event in a given year. We then combine this information with the average expected value of such an attack being realised. For example, say the probability of being victimised by an act of separatist terrorism is 1 in 10,000; this probability is the first entry into our threat map. Now, suppose the outcomes for different victims are distributed as follows: 95% of individuals are one hour late for work; 4% suffer a permanent injury and 1% die as a result. Finally, we need to know the number of individuals who are involved in an average attack. For illustrative purposes, let us say that it is 100 individuals. Let us suppose the average wage in this location is h50 per hour; this suggests that the cost of the 95% of victims who only suffer a delay is h4,750. Let us then suppose that the cost of permanent injuries is h20,000; in our example, four individuals suffer these injuries, adding a further h80,000 to the cost. Finally, following Bru¨ck, De Groot, and Schneider (2011), we define the statistical value of life to be h2,050,000. Therefore, the total negative financial value of this incident is h2,134,750, which we then interact with the 1 in 10,000 chance of being a victim of such an attack for a value of h2,134.75. This value is the total expected value of such an event. A further advantage of calculating potential threats in this way is that it enables the researcher to differentiate, at the individual level, an individual’s current risk levels. After all, an individual who never takes public transport will be less likely to be involved in the aforementioned public transport incident. Contingent on demanding data requirements, it is possible to include an individual’s entire behavioural pattern in this calculation, in order to generate an individual threat map. More feasible is the possibility of differentiating across sub-national regions and broad age groups or behavioural categories. Protection The level of protection can, in principle, be measured in a similar way to the level of threat. In this instance, we could, for example, look at the reduction in the probability of an individual being a victim of some act of terrorism that arises from protection expenditure. Using a similar conversion mechanism as discussed in the previous section, the net worth of this

88

TILMAN BRU¨CK ET AL.

protection can then be estimated. However, the data requirements are not trivial and the necessary data may not be publicly available. Additionally, the effectiveness as well as the scale of this protection must also be considered. Simply pouring money into programmes with no specific focus is less likely to increase protection than lower but well-directed spending. Whilst the likes of airport scanners and X-ray machines have been shown to be effective in reducing terrorist attacks against aircraft, for example (Enders & Sandler, 2006), it is less obvious just how much protection is offered by police officers on the beat. Accordingly, we consider not only the input of resources or effort in our data collection on the level of protection; where and how these resources are spent is also an important element of such a determination. To further discuss the data requirements of protection, therefore, we define three main providers of protection: the government, the private sector and individuals. Whilst there may be some blurring of the lines between these broad groupings, the combined actions of these three actors represent the full spectrum of protection action. In the rest of this section we consider the protection offered by each of these actors, and how to quantify their contributions, in turn. We consider protection offered by the government to be, at times, some of the most visible. Police officers, armed forces and intelligence services all play highly important and often visible roles in public security. Despite this visibility, however, the role of government in the prevention of terrorism runs significantly deeper. Accordingly, whilst we must concern ourselves with the state’s direct provision of counter-insecurity measures, other factors, like the strength of institutions, are also important. The value of such protection is, however, perhaps less immediately obvious than the number of law enforcement officers or intelligence. Despite this issue, however, governments maintain detailed spending accounts, which should, in principle, be usable as raw input data. The issue of quantifying government-level security spending, therefore, is found not in the absence of potential data sources but in the application of our ‘effectiveness’ notion (Jansen, 2009). In a number of cases, some inference can be drawn, however
data on crime ‘clearance rates’, for example, may suggest how effective spending on police forces is. Although less straightforward, the number of foiled terrorist attacks may also infer some degree of effectiveness, although we accept that this also has implications for the general underlying level of threat. We propose that individuals offer two different kinds of protection: the first we define as ‘deterrence’ and the second as ‘observation’. Deterrence is defined as actions individuals take to reduce the level of insecurity they

Measuring Security

89

personally face. It is our assumption that this kind of deterrence protection on a societal level should be relatively easy to quantify, in terms of expenditure
although the link to effectiveness is a familiar problem. Quantifying the impact of observation, on the other hand, requires significantly more thought. We define observation along the lines of vigilance
the willingness, for example, to report an unattended bag at an airport. Just how effective this as a method of deterrence is debated
the suggestion is made, for example, that vigilant passengers helped to prevent terrorist attacks by show-bomber Richard Reid (in 2001) and underpants-bomber Umar Abdulmutallab (in 2009) on transatlantic flights but represent only two examples amongst a series of poor or false reports. The impact of such poor reports may actually raise the cost of protection but not, necessarily, the level of protection itself. Finally, we also propose a significant role of the private sector. As with government-provided protection, much private-sector provision of security is visible
for example, security officers in airports
this area of protection is also much deeper than it, prima facie, appears. We propose, however, that the benefits from such actions can be readily quantified. In such cases, it seems obvious that the net reduction in insecurity that results from such spending is a function of the spending itself and technologies adopted. Accordingly, the level of spending to combat these threats can be used as a means to measure the level of protection provided by the private sector but it is unclear whether or not private-sector actors will be willing to share this raw information. Subjective Security Subjective security is, perhaps, the most esoteric component of measuring security, yet for a society to be secure, it must ‘feel’ secure. Accordingly, we include it as a third important factor in our efforts to measure security. Significant research has shown that the realised level of threat and the perceived level of threat do not always match. The German Socioeconomic Panel (GSOEP), for example, shows that some 85% of Germans consider themselves ‘somewhat concerned’ or ‘very concerned’ about transnational terrorism but inspection of this data reveals that those most concerned about the threat of terrorism are those who are, statistically, least likely to be affected by it (Bru¨ck & Mu¨ller, 2010). Whilst those living in rural areas are, on average, less likely to experience terrorism, they are disproportionately concerned about it. Those who live in urban areas and are, therefore, more likely to directly experience a terrorist attack are less concerned than the average.

TILMAN BRU¨CK ET AL.

90

Despite this relatively esoteric idea, the previous paragraph already alludes to a potential means of measuring perceived threat
namely, asking people. Whilst it is infeasible to ask everyone in a population about their view of insecurity, a number of representative surveys exist throughout Europe that deal with perceptions on a number of terrorismrelated issues. The data in the GSOEP, for example, includes questions on perceptions of both terrorism and crime. Similar questions are included in a number of domestic surveys, such the Italian FOCISV and the United Kingdom’s BSAS and also in the international Eurobarometer surveys. Precisely how to quantify data of this type into a measure that fits with the monetary-based suggestions of the two previous sections, however, is a more taxing problem. To answer this question, we turn our attention to research that has already taken place, beginning with Hess (2003), which works on a ‘willingness to pay’ idea. Building on microeconomic notions of indifference, this approach aims to model how much income individuals would be willing to forego, in order to face no-insecurity threats, whilst maintaining the same quality of life. In keeping with microeconomic nomenclature, the suggestion here is that an individual is indifferent between Level of Insecurity 1 and Income 1 and Level of Insecurity 2 and Income 2; by testing how much income the average individual would be willing to forego in order to be indifferent between the status quo and a new world with no insecurity threat, a monetary value can be attributed to perceptions of insecurity. It is our belief that willingness to pay relates significantly to an individual’s perception of threat and provides a ready monetary conversion of the notion. Frey, Luechinger, and Stutzer (2009) show that the average citizen in Northern Ireland, earning an average wage, is willing to forego in excess of 25% of his/her wage not to face the threat of terrorism; the average Parisian is willing to forego only 4%. To further elucidate this point, Northern Ireland suffered (and to a small degree, still suffers from) a violent terrorist conflict that has claimed the lives of over 3,500 individuals and has been present for in excess of 30 years. The Global Terrorism Database, START (2011), suggests only one fatal terrorist incident has occurred in Paris in the last decade, although terrorism has been more prevalent elsewhere in France. It doesn’t take a huge leap of faith to believe that those in Northern Ireland have significantly higher perceptions of the risk of terrorism than those in Paris. That those in Northern Ireland are also willing to forego a significantly larger proportion of their income than Parisian counterparts .

Measuring Security

91

to prevent the threat of terrorism suggests validation for this measurement methodology and a potentially rich data source for our measure. Collecting survey data is not possible retroactively, though, and it may not be feasible to collect survey data to be able to construct a very complete picture (Bru¨ck, Justino, Verwimp, & Avdeenko, 2010). However, it may be possible to measure behaviour as a proxy for the underlying ideas about perceived security. If one were to find that news reports, for example, increase their coverage of certain security incidents despite a decrease in the occurrence thereof, this may be a signal. Of course, any of such data would need to be calibrated with survey data and a large amount of data would be required to uncover what the underlying patterns are. As a final point, we must also note that certain aspects of this measure face significant overlap with the other sections; however, an individual who feels insecure, is likely to engage in certain behaviour that also has a corresponding impact on protection.

REQUIREMENTS OF A UNIVERSAL SECURITY INDICATOR In this section, we discuss the desirable properties of an indicator of security. From the preceding sections, it has become clear how these separate variables are constructed: Threat (T) is the total per capita value of threats realised (R) and threats deterred through Protection (P), so: T = R + P. In addition to the value of Protection, there is also the Efficiency of Protection, which we define in Eq. (15), which is the marginal benefit of Protection divided by the marginal cost of Protection. Finally, Perceived Threat (X) is the citizen’s estimate of the per capita value of total Threat. To start, we set out what the conditions are to which the Security Indicator (S) should adhere to: Proposition 1. The measure of security should lie between 0 and 100 where 0 is a situation of optimal security and 100 is a situation of optimal insecurity. This proposition is introduced only to guarantee a uniform and convenient measure that we can then use in further propositions. Proposition 2. For T > X and R > R0 : S(P;R;E0 ;X) < S(P;R;E;X). This proposition suggests that, in a situation where actual threat, T, is greater than perceived threat, X, the net level of security is, ceteris paribus,

TILMAN BRU¨CK ET AL.

92

lower when the level of realised threat, R, is higher. In other words, should an exogenous increase in threat not be matched with a corresponding increase in either protection or the efficiency of protection, the net security position is lowered. Proposition 3. For E > E0 =1 > E″: S(P;R;E;X) < S(P;R;E0 ;X) and S(P;R; E″;X) < S(P;R;E;X). When the marginal benefits of Protection are either greater or smaller than the marginal costs of Protection, the result is a suboptimal level of security. When there is an undersupply of protection, security would improve with an increase in protection. Alternatively, when there is oversupply of protection, society would be better off should expenditure on protection be scaled back. In other words, ceteris paribus, the highest potential level of security is reached when the cost of providing that security is the same as the benefit that protection provides to society. Proposition 4. For P = P0 , R = R0 , E = E0 , X = X0 and T > T0 : S(P;R;E;X|T) < S(P0 ;R0 ;E0 ;X0 |T0 ). This proposition states that, even when the same proportion of threat is deterred by protection measures, a society with a greater level of underlying threat remains more insecure than one with a lower level of threat. After all, in this situation, even though the same proportion of threat is deterred, the level of realised threat is higher, ceteris paribus, in the society facing the lower underlying threat. This higher level of realised threat suggests an increased probability that an individual will be the victim of a terrorist attack. Proposition 5. For X > T = X0 : S(P;R;E;X0 ) < S(P;R;E;X). This proposition implies that when the perceived threat exceeds the actual threat level, a higher level of protection increases security, even if the spending is, technically, inefficient. In other words, the optimal level of protection provided should match with the level of threat, except in those cases where perceived threat exceeds actual threat. In those cases, it is warranted to increase the level of protection in order to alleviate the worries of the population, even if these worries are irrational.

CONCLUSION In this contribution, we discuss the possibilities of developing a measure of security. We believe that it is possible, despite severe data restrictions, to

93

Measuring Security

develop such an indicator for security in a coherent and comparable way. It is relatively easy to describe the different requirements such a quantitative measure should adhere to and we propose a specific measure for doing so. However, our proposed measure hinges on the fact that it requires significant data inputs. Without going into the country-by-country specifics, we can merely allude to these requirements. We believe, however, that it is possible to meet the necessary requirements and to develop the necessary inputs for the proposed measure of security. While this is a merely theoretical contribution that discusses the potential challenge and solutions, it should be a goal to execute the proposed system as planned. This requires the development of a complete threat map and corresponding protection and perceived threat descriptions. Doing so would yield a measure that can accurately describe developments across regions, countries and time. It may be possible to show that certain government actions do not, in fact, increase the level of security. Indeed, one outcome could be the suggestion of specific policies to combat particularly poignant security threats. Generation of ‘counterfactuals’, similarly, could help in the development of optimal security policies. Another possible suggestion could be to focus on the provision of information to society. After all, if individuals are consistently more worried than they should be, then terrorist goals are facilitated, even if they do not attack. Such a situation would stimulate the demand for an overprovision of protection, despite the fact that equilibrium utility would be higher, and the costs of provision of security lower, if individuals’ perceptions would be more in line with reality. Finally, whilst we discuss this measurement device in the specific context of terrorism, we also propose that the methodology could easily be extended to account for other forms of intentional, human-induced, insecurity such as crime, cybercrime or organised crime. After all, many of the same interactions described in this contribution are present in these other forms of insecurity, whilst the use of financial measures as a means of comparison could also be easily extended to these other threats to individual’s security and perceptions of security.

NOTE 1. Adi Shamir was referring to cryptography, but the first law still applies. The second law, To halve your vulnerability, you have to double your expenditure may also be worth exploring, although it is disconnected from ordinary economics,

TILMAN BRU¨CK ET AL.

94

which would argue for diminishing returns. The third law is interesting to keep in mind for certain forms of security: Cryptography is typically bypassed, not penetrated (Shamir, 2002).

REFERENCES Al-Rodhan, N. R. F., Stoudmann, G., & Finaud, M. (2006). Proposal for a security matrix. Geneva, Switzerland: Geneva Centre for Security Policy. Becker, G. (1968). Crime and punishment: An economic approach. Journal of Political Economy, 76(2), 169
217. Berinato, S. (2005). A few good information security metrics. CSO Magazine. Retrieved from http://www.csoonline.com/article/220462/a-few-good-information-security-metrics Bru¨ck, T. (2005). An economic analysis of security policies. Defence and Peace Economics, 16(5), 375
389. Bru¨ck, T., De groot, O. J., & Schneider, F. (2011). The economic costs of the German participation in the Afghanistan war. Journal of Peace Research, 48(6), 793
805. Bru¨ck, T., Justino, P., Verwimp, P., & Avdeenko, A. (2010). Identifying conflict and violence in micro-level surveys. HiCN Working Paper No. 79. Households in Conflict Network, Brighton, UK. Bru¨ck, T., & Mu¨ller, C. (2010). Comparing the determinants of concern about terrorism and crime. Global Crime, 11(1), 1
15. Buzan, B., & Hansen, L. (2009). The evolution of international security studies (pp. 10
19). Cambridge, UK: Cambridge University Press. Chandler, M. E. G. (2011). Old threats, new threats: New solutions? In K. Lange (Ed.), The future of security. Munich, Germany: Hanns-Seidel-Stiftung. Department of Homeland Security. (2009). Homeland security advisory council
HSAS task force
Stakeholder feedback. Retrieved from www.dhs.gov/xlibrary/assets/hsas_task_ force_stakeholder_feedback.pdf Department of Homeland Security. (2011). NTAS guide
National terrorism advisory system public guide. Retrieved from http://www.dhs.gov/xlibrary/assets/ntas/ntas-public-guide.pdf Enders, W., & Sandler, T. (2005). The political economy of terrorism. Cambridge, UK: Cambridge University Press. Enders, W., & Sandler, T. (2006). Distribution of transnational terrorism among countries by income class and geography after 9/11. International Studies Quarterly, 50, 367
393. Farrow, S. (2007). The economics of homeland security expenditures: Foundational expected cost-effectiveness approaches. Contemporary Economic Policy, 25(1), 14
26. Frey, B. S., Luechinger, S., & Stutzer, A. (2009). The life satisfaction approach to valuing public goods: The case of terrorism. Public Choice, 138, 317
345. Grigg, I. (2005). Pareto-secure: A definition of security using the theory of pareto efficiency. Retrieved from http://iang.org/papers/pareto-secure.html Haigner, S. D., Schneider, F., & Wakolbinger, F. (2012). Combating money laundering and the financing of terrorism: A survey. Economics of Security Working Paper No. 65. DIW Berlin, Berlin, Germany. Hess, G. D. (2003). The economic welfare cost of conflict: An empirical assessment. CESifo Working Paper No. 852. CESifo, Munich, Germany.

Measuring Security

95

Jansen, W. (2009). Directions in security metrics research. National Institute of Standards and Technology Interagency Report no. 7564. National Institute of Standards and Technology, Gaithersburg, MD. Landes, W. M. (1978). An economic study of US aircraft hijacking, 1961
1976. Journal of Law and Economics, 21(1), 1
31. Murray, C. J. (1994). Quantifying the burden of disease: The technical basis for disabilityadjusted life years. Bulletin of the World Health Organization, 72(3), 429
445. Schroeder, U. (2010). Measuring security sector governance
A guide to relevant indicators. Geneva Centre for the Democratic Control of Armed Forces Occasional Paper No. 20. DCAF, Geneva, Switzerland. Shamir, A. (2002). Cryptography: State of science. Turing Award Lecture. Retrieved from http://amturing.acm.org/vp/shamir_0028491.cfm Shapiro, J. N., & Cohen, D. K. (2007). Color blind: Lessons from the failed Homeland Security Advisory System. International Security, 32(2), 121
154. START. (2011). Global terrorism database. College Park, MD: National Consortium for the Study of Terrorism. Retrieved from http://www.start.umd.edu/gtd Torgerson, M. (2007). Security metrics for communications systems. 12th international command and control research and technology symposium, Newport, RI.

CHAPTER 5 MEASURING TERRORISM WITH THE GLOBAL TERRORISM INDEX Daniel Hyslop and Thomas Morgan ABSTRACT Purpose
This aim of this chapter is to outline the rationale and methodology behind the development of the Global Terrorism Index (GTI), a composite index which attempts to measure the unobserved latent impact of terrorism across 158 countries covering over 99% of the world’s population. Design methodology/approach
The GTI attempts to capture the multidimensional direct impact of terrorist related violence, in terms of its physical effect, as well as emotional wounds and fear, by attributing a single-weighted average national level score. It is based on the Global Terrorism Database (GTD). This chapter also details selection of the indicators, weighting methodology and method of normalisation. Findings
The GTI aims to be a reference point from which one can further discuss how researchers and the general public collectively define, interpret and contextualise the social, political and economic impact of terrorism.

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 97
114 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022010

97

98

DANIEL HYSLOP AND THOMAS MORGAN

Research limitations/implication
Certain issues related to the data remain intractable (e.g., missing injury data in the GTD and property damage). The chapter also suggests a number of areas in which the basic index design could be expanded upon. Originality/value of the chapter
By presenting the GTI the chapter provides a relative measure of terrorism. Its value rests in the possibility to compare the impact of terrorism between countries and scores, analyse trends over the time and assess the aggregate global impact of terrorism. Keywords: Terrorism; Global Terrorism Index; Global Terrorism Database; measuring terrorism; quantitative research; terrorism trends

INTRODUCTION The Global Terrorism Index (GTI) is a composite index which attempts to measure the unobserved latent impact of terrorism across 158 countries covering over 99% of the world’s population.1 Using the comprehensive event database, the Global Terrorism Database (GTD),2 the GTI combines four variables to develop a composite score: the number of terrorist incidents in a given year, total number of fatalities in a given year, total number of injuries caused in a given year and the approximate level of property damage in a given year. The composite score therefore captures the direct effects of terrorist-related violence, in terms of its physical effect, but also attempts to reflect the residual effects of terrorism in terms of emotional wounds and fear by attributing a weighted average to the damage inflicted in previous years. From this data, countries are attributed a score which reflects a national average. By producing a relative measure of terrorism it is possible to compare the impact of terrorism between countries, analyse trends over the time and assess the aggregate global impact of terrorism. This enables one to contextualise and place terrorism in broader global context and can provide the basis for further understanding one particular dimension of violence. Quantitative efforts, while constrained by issues of definition, which this chapter will come to, can in fact help expand rather than collapse the definition and meaning of terrorism and, over time, can provide the basis for more comprehensive measures. Providing a relative measure of the impact of terrorism and its aggregate level has policy relevance because of the significant economic and political

Measuring Terrorism with the Global Terrorism Index

99

impact it can have, and also due to the scale of the economic and political energy committed to counterterrorism. Terrorism has evident large impacts on economic growth and stability. For instance, Frey and Luechinger (2003, p. 7) noted terrorism in Spain reduced annual direct foreign investment inflow by 13.5% on average over the period 1968
1991, which translates to real direct foreign investment of almost $500 million, or 7.6% of annual gross fixed capital formation. Blomberg and Rose (2009) cite various studies which have found 9/11’s direct economic cost to be in the range of $27 billion (Looney, 2002) with total loss of output estimated by the IMF as much a ¾% of GDP, amounting to $75 billion in 2001 US dollars. While the direct cost of loss of assets and the resulting loss of output are enormous for such a large and devastating event, the after-the-fact costs of future prevention are in fact likely to be much higher. Mueller and Stewart (2011) estimate that in the decade since 9/11, the United States alone has spent over $1 trillion on domestic homeland security and counterterrorism efforts and spent approximately 30% of the 2010 Federal Budget on Homeland Security, Defense, Veterans Affairs and the Federal National Intelligence Programme when aggregated (IEP, 2012). Similarly, the international community has committed significant resources to counterterrorism efforts domestically and internationally in the form of the military interventions in Afghanistan and Iraq. Further in the European context, Kollias, Messus, Mylonidis, and Paleologou (2009) investigated the effectiveness of counterterrorism policy in Greece finding that investment in short-term counterterrorism strategies has a weak negative impact on internal terrorist actions, whereas longer-term investment in infrastructure and capital equipment can indirectly improve the capacity of security services to mitigate terrorism. Considering the sheer magnitude of state-based responses to terrorism before and since 9/11, the provision of hard data on terrorist incidents is an important starting point for assessing both the efficacy of state responses to terrorism and to counter the myth-making nature of terrorist groups themselves. In assessing state responses to terrorism, the loss of human lives in Afghanistan and Iraq demonstrates responses to terrorism and counterterrorism measures can be more destructive and costly than the original acts of terror these interventions are in response to. Conversely, ‘evidence-based’ or hard data on the actual impact of specific groups go to the very heart of challenging perceptions of the efficacy of terrorist groups themselves and the direct impact they have on specific countries. For instance, the GTI rankings reveal that based on 2011 data the United States is now ranked 41 in the world on its terrorist impact with the majority of its terrorist incidents

100

DANIEL HYSLOP AND THOMAS MORGAN

in the past 10 years (30%) being committed by domestic terrorists such as the Animal Liberation Front. Conversely, Al-Qa’ida in the Arabian Peninsula (AQAP) and Tehrik-i-Taliban Pakistan (TTP) made up less than 3% of the attacks. Similarly, commonly held perceptions about the relative impact of terrorism in Israel would likely not match with the GTI which showed it is ranked 20 in 2011, with 19 other countries with a greater impact of terrorism for 2011. In order to achieve political objectives terrorist organisations need to create a myth and reputation which makes them appear bigger and more impactful than they are in reality. This is confirmed in the analysis of the GTD by LaFree (2012, p. 40), which shows since 1970 fewer than half of terrorist attacks have resulted in fatalities, and that of the 52 foreign terrorist groups identified by the US State Department as posing the greatest threat to the United States, 90% of the 17,000 attacks carried out by these groups were actually domestic attacks. Findings such as this show how empirically assessing the direct impact of terrorist groups ought to place terrorist threats and risks into perspective.

PRINCIPLES FOR COMPOSITE INDICES AND WHY A COMPOSITE MEASURE OF GLOBAL TERRORISM IS NECESSARY Composite indices are increasingly being used by governments, think tanks, policymakers and the general public to monitor and evaluate progress of a phenomenon otherwise not easily understood via a single measure. Composite indices serve a useful purpose as they can be used as research tools and for issue-based advocacy. In this, they can drive attention to an otherwise poorly understood issue, and via statistical technique they can generate new information on an unobserved, multidimensional phenomenon for which there is minimal theoretical or practical measurement guidance. In this process a composite index can render a difficult to understand or interpret issue simple and easy for people to understand and by virtue of this help deepen understanding and promote accountability and enable the tracking of trends. More indirectly, they can highlight data gaps, help promote data advocacy and start important discussions on how we define, interpret and conceptualise social, political and economic issues. The first iteration of the GTI has subsequently been widely distributed through both research and academic networks, and also through media networks

Measuring Terrorism with the Global Terrorism Index

101

and has received broad exposure with the general public with the intent of achieving these goals. The 2012 launch of the GTI saw the index featured in over 2,000 media articles and blog posts, with coverage on CNN, the Washington Post, the BBC and the Economist magazine discussing the key trends and patterns of terrorism from 2002 to 2011.3 Conversely, composite indicators may have negatives in that they may invite overly simplistic policy conclusions, be misleading if they are poorly constructed or simply reflect incorrect subjective or normative decisions. Hence, in constructing the GTI, it has been important to follow key principles of composite index design in terms of ensuring (1) it is based on conceptually sound foundations and clear definitions, (2) it is policy relevant, (3) it is based on reliable, consistent and comprehensive data, (4) it is a clear and transparent methodology and a process of construction and (5) changes in the index are guided by principle and not by conclusions. With the key objective of the GTI aimed at capturing the impact of the latent variable of ‘terror’ on a nation’s population, the index has to capture the full spectrum of terrorism effects, including the psychological and economic impact, rather than just looking at a list of total deaths. While the number of deaths inflicted by a terrorist incident is a very key measure of the impact of an event, some incidents may be designed to inflict injuries, or more directly targeted at key physical infrastructure or economic assets. For instance, fundamentalist, religious-based terrorism has tended to focus on mass casualties to make their cause known, whereas leftist-based or nationalist terrorists have been focused on winning public favour so tend to avoid maiming or killing of people not directly connected to the political regimes they oppose (Frey & Luechinger, 2003). Hence, one would not be able to quickly deduce the relative impact of the entire spectrum of terrorism from looking at one indicator alone. In this fundamental sense, political violence as characterised by the definition of terrorism in the GTI is multidimensional in nature; its outcomes and impact cannot be reduced to one variable. The impact of terrorism can be conceptualised as having at least three distinct elements: (1) the direct impact in terms of lives lost and injuries inflicted, (2) the direct and indirect economic damage or destruction of key assets and instigation of protective measures which divert productive resources into less productive areas and (3) what can be broadly termed the social and political fear which has direct impacts on the political economy of states and behaviour of individuals. The indicators in the GTI directly measure these factors and indirectly measure the difficult-toobserve aspects of social and political fear.

102

DANIEL HYSLOP AND THOMAS MORGAN

METHODOLOGY OVERVIEW There are three important considerations in constructing an index measuring the relative impact of terrorism between countries: Firstly, because the impact of terrorism is spread and amplified across a nation by intense and constant media reporting, the impact of terrorism can be considered to be equally distributed across a population; the level or impact of terrorism is not ‘diluted’ across a population. Thus, instead of using some kind of relative measure or ‘terrorism rate’ (e.g. ‘impact of terrorism per 100,000 people’) an absolute measure was used for each country. Of course, in reality the impact of terrorism does not stop at a nation’s borders (neighbouring countries are likely to be affected by terrorist incidents, terrorist incidents targeting citizens of a certain country outside that country’s borders will also have an impact and terrorist incidents in geopolitically prominent countries are more likely to attract a disproportionate amount of media coverage). Future iterations of the GTI may attempt to take these factors into consideration. Secondly, it seems clear that there are diminishing returns to terrorism, that is, the amount of terror generated by subsequent terrorist attacks is diminished, particularly when large amounts of terrorism take place in the context of an existing war or conflict. Thus, even though certain countries receive a disproportionate amount of the global total of terrorist attacks, the impact of terrorism on these countries is not orders of magnitude larger than on other countries. In order to recognize this fact, the bands used to score the impact of terrorism in the GTI increase exponentially. This system was devised in conjunction with IEP’s expert panel, in particular Dr Ekaterina Stepanova. Thirdly, the index needed to capture how terrorism affects nations over time. An index which only included data from the immediately preceding year would fail to reflect the lingering psychological impact of terrorism. In order to address this issue, a five-year weighted average was applied to the data, such that incidents from the most immediate year counted for 52% of a nation’s total score, 26% from the previous year, 13% from two years previous and so on. Such a system is not a perfect model of the lingering impact of terrorism, as the larger a terrorism incident is, the more profound and long lasting its impact would tend to be (e.g. the psychological and cultural impact of the September 11 terrorist attacks arguably lingered for much more than five years). However, when multiple simulations were run on the index using weighted averages ranging from 1 to 15 years, the fiveyear average was unanimously accepted as the most appropriate range.

Measuring Terrorism with the Global Terrorism Index

103

There can be no perfect index of terrorism given the amount of confounding factors that make comparisons between countries difficult or near impossible. Chief among these are the enormous spread in country sizes and population and the unequal geographic spread of terrorist incidents. However, the GTI is an attempt to make a best possible compromise between these problems, and allows for the construction of comparisons between countries that are meaningful when it comes to assessing the impact of terrorism.

DATA SOURCES AND DEFINITION The GTI relies solely on the GTD as its source for terrorist incidents. Several other terrorist databases were considered, including the National Memorial Institute for the Prevention of Terrorism (MIPT) Terrorism Knowledge Base, the Worldwide Incidents Tracking System (WITS), the Research ANd Development (RAND) Database of Worldwide Terrorism Incidents and the other major event database, International Terrorism: Attributes of Terrorist Events (ITERATE). However, of all, the GTD is by far the most comprehensive database of terrorism incidents. For further background of the GTD, see LaFree and Dugan (2007). Several of the other databases have also been discontinued. Researchers working on the GTI investigated whether it might be possible to merge the databases into one comprehensive set of terrorism incidents; however, any incidents added to the GTD in this manner would lack the metadata collected by the GTD compilers, START. It was also not possible to account for the competing definitions of terrorism proffered by the various databases. Defining terrorism is not a straightforward matter. There is no single internationally accepted definition of what constitutes terrorism, and the terrorism literature abounds with competing definitions and typologies. Based on the GTD, put together by the START, the GTI accepts the terminology and the definitions agreed to by the START researchers and its advisory panel. The GTI defines terrorism as ‘the threatened or actual use of illegal force and violence by a non-state actor to attain a political, economic, religious, or social goal through fear, coercion, or intimidation’. This definition acknowledges that terrorism has not only a physical direct impact, but also a psychological impact which can affect a society for some time many years after.

104

DANIEL HYSLOP AND THOMAS MORGAN

In order to be included as an incident in the GTD the act has to be: ‘an intentional act of violence or threat of violence by a non-state actor’. This means an incident has to meet three criteria in order for it to be counted as a terrorist act: A. The incident must be intentional
the result of a conscious calculation on the part of a perpetrator. B. The incident must entail some level of violence or threat of violence
including property violence as well as violence against people. C. The perpetrators of the incidents must be sub-national actors. This database does not include acts of state terrorism. In addition to this baseline definition, two of the following three criteria also have to be met in order to be included in the START database from 1997: • The violent act was aimed at attaining a political, economic, religious or social goal. • The violent act included evidence of an intention to coerce, intimidate or convey some other message to a larger audience (or audiences) other than the immediate victims. • The violent act was outside the precepts of international humanitarian law. This importantly allows researchers flexibility to apply different definitions depending on interpretation and use of the data. In cases where there is insufficient information to make a definitive distinction about whether it is a terrorist incident within the confines of the definition, the database also codes these as ‘doubt terrorism proper’ incidents. Depending on where the definitional overlap is, between, say, a terrorist incident and potentially other forms of crime and political violence, such as insurgency, hate crime and organized crime, the GTD coders identify where a doubt may lie.4 In order to only count unambiguous incidents of terrorism the GTI does not include doubted incidents. It is important to understand how incidents are counted. According to the GTD codebook: ‘incidents occurring in both the same geographic and temporal point will be regarded as a single incident, but if either the time of occurrence of incidents or their locations are discontinuous, the events will be regarded as separate incidents’. Illustrative examples from the GTD

Measuring Terrorism with the Global Terrorism Index

105

codebook are as follows (National Consortium for the Study of Terrorism and Responses to Terrorism (START), 2012): • Four truck bombs explode nearly simultaneously in different parts of a major city. This represents four incidents. • A bomb goes off, and while police are working on the scene the next day, they are attacked by terrorists with automatic weapons. These are two separate incidents, as they were not continuous, given the time lag between the two events. • A group of militants shoot and kill five guards at a perimeter checkpoint of a petroleum refinery and then proceeds to set explosives and destroy the refinery. This is one incident because it occurred in a single location (the petroleum refinery) and was one continuous event. • A group of hijackers diverts a plane to Senegal and, while at an airport in Senegal, shoots two Senegalese policemen. This is one incident, since the hijacking was still in progress at the time of the shooting and hence the two events occurred at the same time and in the same place.

INDICATORS The four indicators in the GTI were chosen because they each capture part of the impact of terrorism, without being sufficient in of themselves to account for the total impact of terrorism. The four indicators are: • • • •

Total number of terrorist incidents in a given year. Total number of fatalities caused by terrorists in a given year. Total number of injuries caused by terrorists in a given year. A measure of the total property damage from terrorist incidents in a given year.

The total number of fatalities is the most obvious measure of terrorist impact; however, a terrorist attack with no fatalities but multiple injuries would still have a pronounced impact. Likewise, terrorist attacks targeting infrastructure with no fatalities or injuries would still lead to significant media coverage and some semblance of ‘terror’. In order to reflect the different impacts that each of these indicators would have in generating terror, each of the factors was weighted differently by the expert panel5 (where fatalities were given the highest weighting, such that one fatality is

106

DANIEL HYSLOP AND THOMAS MORGAN

the equivalent of three injuries). The accuracy of data collection was a primary concern for the expert panel in assessing appropriate weights for the indicators, as injury data was likely to be under-reported. There was also no mechanism for differentiating between serious and minor injuries. Furthermore, calculating an accurate measure of property damage is exceedingly difficult. As such, the level of property damage as recorded in the GTD is divided into four broad tranches (unknown, less than a million, more than a million but less than a billion and over a billion, with all measures being in US dollars). Thus, the property damage indicator is not supposed to be an accurate measure of the dollar value of the impact of terrorism, but rather a rough proxy for the level of terror generated by attacks on property. In order to properly assess economic damage one evidently needs to look to more detailed impact studies which qualitatively analyse direct and indirect economic costs.

WEIGHTINGS OF THE INDICATORS The greatest weighting is attributed to a fatality (Table 1). It should be noted that the property damage measure is further disaggregated into four bands depending on the measured scope of the property damage inflicted by one incident. These bandings are given in Table 2, whereby incidents Table 1. Indicator Weights Used in the GTI. Dimension

Weight

Total number of incidents Total number of fatalities Total number of injuries Sum of property damages measure

Table 2.

Property Damage Levels as Defined in the GTD and Weights Used in the GTI.

Code Damage Level 0 1 2 3

1 3 0.5 2 (contingent on sub-score)

Range Unknown Minor (likely < $1 million) Major (likely between $1 million and $1 billion) Catastrophic (likely > $1 billion)

Measuring Terrorism with the Global Terrorism Index

107

causing less than US$1 million are accorded a weighting of 1, between $1 million and $1 billion a 2 and more than $1 billion a 3. It should be noted that a great majority of incidents are coded in the GTD as an ‘unknown’ level of property damage, thus scoring nil, with ‘catastrophic’ events being extremely rare. Using the above indicators and weights, any single terrorist incident can be given an impact score. For example, a terrorist incident that resulted in three deaths, nine injuries and over a million dollars in property damage would receive a score of (3 × 3[fatalities] + 9 × 1[injuries] + 2 × 2[property] + 1[incident]) = 23. The total score for a country would then be summed for the last five years, and then the five-year weighted average would be applied to arrive at a country’s final terrorism impact raw score. This score would then be converted to a final score on a range of 1
10, where 0 is equivalent to no impact of terrorism, and 10 is the global maximum impact of terrorism. The scores were converted to the 0
10 range using the following method: 1. Define the Minimum GTI Score across all countries as having a banded score of 0. 2. Define the Maximum GTI Score across all countries as having a banded score of 10. 3. Subtract the Minimum from the Maximum GTI scores and calculate r by a. Root = 2 × (Highest GTI Banded Score − Lowest GTI Banded Score) = 20 × (10 − 0) = 20 b. Range = 2 × (Highest Recorded GTI Raw Score − Lowest Recorded GTI p Raw Score) ffiffiffiffiffiffiffiffiffiffiffiffi ffi c. r = root Range 4. The mapped band cut-off value for bin n is calculated by rn. As mentioned previously, such a method produces scores with a roughly logarithmic distribution, such that a final score of 10 requires a terrorism impact score of over 17,000, whilst a final score of 5 requires a score of approximately 130, and a final score of 2 requires an approximate raw score of 7.

POTENTIAL IMPROVEMENTS TO THE INDEX In compiling the GTI, it became apparent that certain issues related to the data are intractable. As it stands, there is no way to account for the missing injury data in the GTD. Similarly, there is no mechanism to distinguish

108

DANIEL HYSLOP AND THOMAS MORGAN

between the severities of injuries. Property damage is also patchy in this respect, as not only it is difficult to estimate but it is also unrecorded for the majority of terrorism incidents. Nevertheless, such data issues are hopefully mitigated across the entire data set over time, such that the index remains an accurate relative measure of the impact of terrorism across the globe. There are, however, a number of areas in which the basic index design could be expanded upon and improved, in order for the index to more accurately capture the impact of terrorism. For example, the current index has no indicator which captures the impact of terrorist hijackings. Although the number of hijackings has been in steep decline in the last 30 years, they still have an impact which is not captured in the current index design. The coding detail of the GTD does enable researchers to focus on different categories of terrorist and political violence. Some attempts have been made in the literature to decompose the GTD data into separate taxonomies in order to suit alternative analytical priorities. For instance, Enders, Sandler, and Gaibulloev (2011) have disaggregated the GTD into transatlantic versus domestic incidents in order to assess the empirical relationship between the two. Such an approach could be taken with the GTI to order countries by their domestic versus international impact from terrorism. Another way in which the index could be improved would be to weight specific instances by the amount of media coverage they receive, both in and out of country, and importantly where they receive it. Although it is near impossible to aggregate all media mentions of terrorism across all countries, it is currently possible to assess which terrorist incidents receive the most coverage by prominent, internationally focused media organizations (e.g. the BBC, Al-Jazeera, CNN). Rohner and Frey (2007, p. 138) find that there is a close statistical connection between the media and the actions of terrorists. They discuss how such an approach could apply by taking opinion poll data targeted towards leaders in business, politics, science and culture on the quality and influence of key media publications. They find that high-quality media publications such as the New York Times tend to focus more on Western countries and under-report terrorist acts in developing contexts, thus meaning a terrorist act in a Western country may have lower direct impact in terms of lives lost, property damage and injuries, but receive significantly more coverage than one in a developing context. Such heuristics when better understood could potentially provide the theoretical justification for an additional indicator and layer of information in the GTI. Furthermore, large and rare terrorist incidents could be weighted more heavily in order to reflect the disproportionate amount of media coverage they receive, as well as the fact that their psychological,

Measuring Terrorism with the Global Terrorism Index

109

social and cultural impact is likely to linger far longer than small-scale terrorism events. Whilst implementing such changes to the index would be difficult and time consuming, they would provide a more accurate measure of the impact of terrorism. Finally, there is the issue of what is not included in the index, namely, terrorism committed or tacitly supported by the state. Devising an index of state terrorism would be far more problematic, as the definition of state terrorism is even more contested than the definition of conventional terrorism. Coverage of such incidents is also far less extensive, and it is likely that any attempt to measure state terror would inevitably fail to capture many incidents that certain states would not want to be made public. However, it would be a useful exercise to try and construct a definition that mirrors the one used by the GTD, and to at least attempt to tabulate all the incidents that would fit within such a definition. Thus, whilst it may not be possible to construct a country-specific state terrorism index, it may be possible to try and measure the total volume of state terrorism against the total volume of conventional terrorism.

NOTES 1. The GTI has been developed by researchers at the Institute for Economics and Peace. The research team working on the GTI included Daniel Hyslop, Thomas Morgan, Dev Bal and David Hammond. 2. The GTD is put together by the National Consortium for the Study of Terrorism and Responses to Terrorism (START) headquartered at the University of Maryland. 3. Web URLs can be found in the references section. 4. Where there is doubt, GTD analysts would code an incident as conforming to one of four possible alternative designations: (1) insurgency/guerrilla action, (2) other crime type, (3) intra-/inter-group conflict or (4) lack of intentionality. 5. The Global Peace Index expert panel is composed of seven members: Professor Kevin P. Clements, Chairperson, Dr Ian Anthony, Professor Sultan Barakat, Mr Nick Grono, Dr Toshiya Hoshino, Dr Manuela Mesa and Dr Ekaterina Stepanova. The GTI also serves as a sub-composite measure in the Global Peace Index, a composite indicator that measures negative peace in over 158 countries.

REFERENCES Blomberg, S. B., & Rose, A. (2009). Editor’s introduction to the economic impacts of the September 11, 2001 terrorist attacks. Peace Economics, Peace Science and Public Policy, 15(2).

110

DANIEL HYSLOP AND THOMAS MORGAN

British Broadcasting Corporation (BBC). (n.d.). Terror attacks levelled out, says 10-year study. Retrieved from http://www.bbc.co.uk/news/world-20588238. Accessed on 4 December 2012. Economist.com. (n.d.). Fear and loathing. The Economist. Retrieved from http://www.economist. com/blogs/graphicdetail/2012/12/daily-chart-0. Accessed on 4 December 2012. Enders, W., Sandler, T., & Gaibulloev, K. (2011). Domestic versus transnational terrorism: Data, decomposition, and dynamics. Journal of Peace Research, 48, 319
337. Fisher, M. (2012). Heat map: Terrorism around the world (the U.S. is among the least affected). Washington Post, 6 December. Retrieved from http://www.washingtonpost.com/ blogs/worldviews/wp/2012/12/06/heat-map-terrorism-around-the-world-the-u-s-is-among-th e-least-affected/. Accessed on 6 December 2012. Frey, B. S., & Luechinger, S. (2003, October). Measuring terrorism. Zurich IEER Working Paper No. 171. Institute for Economics and Peace. (2012). Research brief: Violence containment industries in the United States. IEP Paper No. 18, IEP. Kollias, C., Messus, P., Mylonidis, M., & Paleologou, S. M. (2009). Terrorism and the effectiveness of security spending in Greece: Policy implications of some empirical findings. Journal of Policy Modeling, 31, 788
802. LaFree, G. (2012). Generating terrorism event databases: Results from the global terrorism database, 1970 to 2008. In C. Lum & L. W. Kennedy (Eds.), Evidence-based counterterrorism policy (pp. 41–64). New York, NY: Springer. LaFree, G., & Dugan, L. (2007). Introducing the global terrorism database. Terrorism and Political Violence, 19, 181
204. Looney, R. (2002). Economic costs to the United States stemming from the 9/11 attacks. Strategic Insights, 6(1). Mueller, J., & Stewart, M. G. (2011). Terror, security, and money: Balancing the risks, benefits, and costs of homeland security. Oxford: Oxford University Press. National Consortium for the Study of Terrorism and Responses to Terrorism (START). (2012). Global terrorism database: Codebook: Inclusion criteria and variables (p. 7). Washington, DC. Retrieved from http://www.start.umd.edu/gtd/downloads/Codebook.pdf Rohner, D., & Frey, B. S. (2007). Blood and ink! The common-interest-game between terrorist and the media. Public Choice, 133, 129
145. Zakaria, F. (Presenter). (2012). Why it’s time to end the ‘war on terror’. Fareed Zakaria GPS Show, Cable News Network (CNN), 10 December. Retrieved from http://globalpublic square.blogs.cnn.com/2012/12/10/zakaria-why-its-time-to-end-the-war-on-terror/

111

Measuring Terrorism with the Global Terrorism Index

APPENDIX: GLOBAL TERRORISM INDEX RANKS AND SCORES
2011 Country Iraq Pakistan Afghanistan India Yemen Somalia Nigeria Thailand Russian Federation Philippines Sudan Congo, The Democratic Republic of Colombia Syrian Arab Republic Algeria Sri Lanka Iran, Islamic Republic of Kenya Turkey Israel Norway Nepal China Burundi Central African Republic Greece Egypt United Kingdom Indonesia Uganda Lebanon Belarus Myanmar Coˆte d’Ivoire Eritrea Senegal Ethiopia Rwanda Bangladesh Morocco United States

Rank 2011

GTI Score for 2011

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

9.56 9.05 8.67 8.15 7.30 7.24 7.24 7.09 7.07 6.80 6.30 6.18 6.06 5.86 5.83 5.68 5.63 5.27 5.24 5.15 5.03 5.02 4.99 4.90 4.84 4.60 4.58 4.51 4.51 4.49 4.48 4.26 4.09 3.99 3.92 3.86 3.73 3.73 3.67 3.60 3.57

112

DANIEL HYSLOP AND THOMAS MORGAN

Appendix. (Continued ) Country Georgia Mali Niger Spain Chad Kazakhstan Saudi Arabia Qatar Chile Cameroon Peru Tunisia Mauritania Paraguay Ukraine Italy Netherlands Tajikistan Mexico Sweden Germany France Australia Angola Venezuela, Bolivarian Republic of Argentina Ireland Austria Bosnia and Herzegovina Zimbabwe Bhutan Honduras Canada Switzerland Serbia Denmark Portugal Moldova, Republic of Guyana Azerbaijan Bulgaria Ecuador Macedonia, The Former Yugoslav Republic of Bahrain

Rank 2011

GTI Score for 2011

42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 77 79 80 81 82 83 84 85

3.43 3.39 3.27 3.09 3.01 2.80 2.71 2.68 2.64 2.63 2.49 2.36 2.18 2.17 2.16 2.07 2.04 2.02 2.01 1.76 1.74 1.73 1.72 1.70 1.62 1.47 1.46 1.41 1.37 1.36 1.23 1.23 1.18 1.17 1.17 1.13 1.13 0.99 0.98 0.87 0.74 0.68 0.62 0.62

113

Measuring Terrorism with the Global Terrorism Index

Appendix. (Continued ) Country

Rank 2011

GTI Score for 2011

Jordan Uzbekistan Czech Republic Madagascar Bolivia, Plurinational State of Malaysia Kyrgyzstan Libyan Arab Jamahiriya Hungary Equatorial Guinea Mozambique Armenia Estonia Guinea-Bissau Kuwait Albania Cambodia Tanzania, United Republic of New Zealand United Arab Emirates Haiti Belgium Finland Japan Swaziland South Africa Cyprus Lesotho Benin Papua New Guinea Brazil Congo, Republic of Croatia Djibouti Gambia Guatemala Guinea Korea, Republic of Lao People’s Democratic Republic Latvia Liberia Montenegro Namibia Nicaragua

86 86 88 89 90 90 92 93 94 95 96 97 98 98 98 101 101 101 104 104 106 107 107 109 109 111 112 113 114 115 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116

0.58 0.58 0.52 0.43 0.42 0.42 0.40 0.35 0.28 0.22 0.20 0.19 0.16 0.16 0.16 0.12 0.12 0.12 0.08 0.08 0.07 0.07 0.07 0.06 0.06 0.05 0.05 0.04 0.03 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

114

DANIEL HYSLOP AND THOMAS MORGAN

Appendix. (Continued ) Country

Rank 2011

GTI Score for 2011

Panama Poland Sierra Leone Slovakia Trinidad and Tobago Turkmenistan Uruguay Vietnam Zambia Burkina Faso Botswana Costa Rica Cuba Dominican Republic Gabon Ghana Iceland Jamaica Lithuania Mongolia Mauritius Malawi Oman North Korea Romania Singapore El Salvador Slovenia Taiwan

Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116 Equal 116

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

CHAPTER 6 TERRORISM AND ECONOMIC SENTIMENT IN EUROPEAN COUNTRIES Christos Kollias and Stephanos Papadamou ABSTRACT Purpose
Terrorist events are unforeseen and have the potential to shake and rattle markets and investors. The purpose of this study is to examine whether major terrorist incidents have affected the Economic Sentiment Indicator (ESI) in four European countries. Methodology/approach
An index is constructed that weights the severity of each event and then used to evaluate through the use of vector autoregressive and impulse response analysis estimation techniques whether or not and to what extent the ESI has been affected. Findings
Effects were more pronounced and evident in the case of France and Germany while the ESI in Spain and Great Britain did not appear to be particularly affected by terrorist incidents. Research limitations/implications
The effects of terrorism on economic sentiment in other countries will provide additional evidence that will allow more robust and conclusive statistical inferences.

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 115
133 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022011

115

116

CHRISTOS KOLLIAS AND STEPHANOS PAPADAMOU

Originality/value of the chapter
The impact of terrorist activity on the ESI for the four European countries studied here has not been examined before using VAR and impulse response analysis. Keywords: Economic sentiment; terrorism; impulse response JEL classifications: H56; C58

INTRODUCTION Economic agents and their behaviour are subject to sentiment. Ample evidence indicates that markets and economic activity in general can be driven and influenced by the prevailing sentiment, i.e. the general attitude and feeling towards a particular security, a market or the wider prospects of an economy (inter alia: Hirshleifer, 2001; Kamstra, Kramer, & Levi, 2003; Loewenstein, 2000; Lucey & Dowling, 2005). Various indicators, which seek to quantify and hence capture and encapsulate economic and market sentiment, have been developed and used in empirical studies attempting to either explain or forecast cyclical economic fluctuations and markets’ performance and trends (inter alia: Aarle & Kappler, 2012; Easaw & Ghoshray, 2008; Ghonghadze & Lux, 2012; Lopes, 2011; Nilsson, 2000; Vuchelen, 2004). Such indicators, products of surveys of businesspersons, investors, consumers and households, do not provide factual information on the economy or the market itself, but rather strive to report the general attitudes on the current state of the economy and the markets and/or expectations towards macroeconomic prospects and future market trends. Of course, this is by no means a unidirectional association, since fluctuations in the economy and changes in the direction of markets do influence agents’ sentiment and outlook as this is reported in the surveys and conveyed by the corresponding value of the indicators that then can be utilised in economic decision making by agents, practitioners and politicians. Agents’ sentiment, however, can be influenced by factors exogenous to the economy or the market. Information and events that are not strictly economic in nature have the potential to exert significant influence on the economy and markets’ performance via the sentiment channel, since they can affect confidence, outlooks, risk perceptions, investment behaviour, economic agents’ decisions, market positioning, portfolio selection and allocation. Such non-economic events can be natural or anthropogenic in

Terrorism and Economic Sentiment in European Countries

117

nature. In the first group one may include major catastrophes such as an earthquake or tsunamis, volcano activity and others that have important and lasting ramifications (Noy, 2009). The anthropogenic category can greatly vary and include events and incidents such as for instance political change, conflict and war, optimism inducing events that can affect the general feeling of happiness and well-being and vice versa, industrial accidents, social grievances and unrest, political instability, and mega sports events. Evidence generated by a growing body of empirical literature strongly indicates that markets and economic activity in general can be highly susceptible to exogenous developments, incidents and events. For instance, major industrial and technological accidents with a heavy toll on life, grave environmental consequences and massive social and private costs become the focus of intense and negative media coverage with the concomitant widespread public attention. Hence, they generate a lot of negative sentiment and can thus exert substantial pressure on the market value of the sectors and companies involved (inter alia: Capelle-Blancard & Laguna, 2009; Kaplanski & Levy, 2010; Kollias, Papadamou, & Stagiannis, 2012). Similarly, political events and developments such as policy shifts, elections, political strife, policy decisions by governments or state agents can improve or threaten economic prospects and opportunities and hence affect economic agents’ evaluation and outlook for a particular market or economy (inter alia: Asteriou & Siriopoulos, 2003; Bru¨ck, Llussa´, & Tavares, 2011; Siokis & Kapopoulos, 2007; Wisniewski, 2009). In particular, political violence in the various forms that it is manifested has been shown to significantly affect investors’ sentiment, market agents’ asset valuations, market and country risk perceptions and portfolio allocation decisions. Inter- or intrastate conflict and war, coups, terrorist attacks live an imprint on markets and economic activity given their unsettling nature (inter alia: Blomberg, Hess, & Tan, 2011; Guidolin & La Ferrara, 2010; Schneider & Troeger, 2006). Although terrorism has been omnipresent throughout history in various forms, shapes and intensity of activity, it has emerged in the last decade or so as the prime security challenge internationally with the mega-events of recent years such as 9/11 in New York or the 2004 and 2005 Madrid and London attacks shaking economies and societies worldwide. This has spurred increased attention and research into the plethora of complex issues associated with terrorist activity and terrorist organisations, ranging from its roots and causes to its political and socioeconomic consequences (inter alia: Bru¨ck, 2007; Caruso & Gavrilova, 2012; Caruso & Schneider, 2011, 2013; Crain & Crain, 2006; Enders & Sandler, 2012;

118

CHRISTOS KOLLIAS AND STEPHANOS PAPADAMOU

Frey, Luechinger, & Stutzer, 2007, 2009; Freytag, Kruger, Meierriek, & Schneider, 2011; Krieger & Meierrieks, 2011; Shughart, 2006). A strand of this growing body of literature has concentrated its focus on the economic consequences of terrorism (inter alia: Abadie & Gardeazabal, 2008; Gaibulloev & Sandler, 2008; Rose & Blomberg, 2010). Although the probability of the occurrence of a terrorist attack is omnipresent, when it actually occurs it is by and large unanticipated. Economic agents can hedge against expected events but not so when it comes to unanticipated ones. Consequently, such incidents have the potential to generate uncertainty and exert a negative effect on economic and market sentiment. Although markets’ reaction to terrorism can significantly vary, the broad conclusion of most studies is that terrorist attacks have a negative impact on markets that nevertheless is in comparative terms short-lived and often not particularly pronounced vis-a`-vis other exogenous shocks that can rattle markets (inter alia: Amelie & Darne, 2006; Arin, Ciferri, & Spagnolo, 2008; Brounrn & Derwall, 2010; Chesney, Reshetar, & Karaman, 2011; Kollias, Manou, Papadamou, & Stagiannis, 2011a; Kollias, Papadamou, & Stagiannis, 2011b; Nikkinen, Omran, Sahlstrom, & Aijo, 2008). Recent studies by Nikkinen and Vahamaa (2010) and Drakos (2010a, 2010b, 2011) have reported evidence indicating that an important channel through which capital markets are negatively affected by terrorist acts is that of sentiment and uncertainty. In particular, as Drakos (2010a) points out, the lower returns experienced on the day of an attack are amplified as the level of the psychological effects increases. In the broader spirit of such studies, this chapter examines how major terrorist acts have affected the Economic Sentiment Indicator (henceforth ESI) in European countries that have in the past fallen victims to terrorist campaigns by domestic or transnational terrorist organisations. The data and methodology employed are presented in the section that follows. The findings are then shown and discussed in the third section, while fourth section concludes the chapter.

DATA AND METHODOLOGY For the task at hand, the chapter utilises monthly data from January 1988 to February 2008 of the ESI1 in selected European countries that have experienced major mass-casualty terrorist attacks: Spain, Germany, France and Great Britain. They are four of the largest and most important economies in the EU and as pointed out, all have been the victims of systematic

Terrorism and Economic Sentiment in European Countries

119

terrorist campaigns from both domestic as well as transnational terrorist groups (Enders & Sandler, 2012). The ESI used here is a composite indicator that is constructed by the Directorate General for Economic and Financial Affairs (DG ECFIN) and is the product of regular harmonised surveys in different economic sectors of the EU countries.2 The surveys are addressed to representatives of the manufacturing sector, the services, retail trade, construction as well as consumers. The ESI is composed of the five different confidence indicators of aforementioned sectors
Industrial Confidence Indicator; Services Confidence Indicator; Consumer Confidence Indicator; Construction Confidence Indicator; and Retail Trade Confidence Indicator
but with different weights attached to each one. It allows comparisons between the EU member states when it comes to the business cycle and economic prospects as encapsulated by the answers provided by the respondents in the different surveys and has been used in economic studies to assess for instance the impact of monetary policy shocks (inter alia: Brissimis & Magginas, 2006; Papadamou & Siriopoulos, 2012). Hence, to the extent that terrorist incidents have the potential to rattle economic agents and exert a negative effect on economic and market sentiment as mentioned above, the question examined here is whether such effects are traceable in the ESI of the EU countries that make up our sample. In a recent paper, Enders, Sandler, and Gaibulloev (2011) provide a useful decomposition of terrorist incidents into domestic and transnational. This decomposition offers a suitable framework that allows the empirical investigation of possible differences in economic agents’ reaction depending on the perpetrators of attack. This chapter study utilises this classification3 in order to examine the effects terrorist incidents had on the ESI allowing for the aforementioned grouping into domestic and transnational events. For this purpose, we estimate a series of vector autoregressive (VAR) models that include both the ESI indices as well as indices of the terrorism activity in the countries examined herein. The general form of the VAR model can be expressed as follows: ZðtÞ = C þ

m X

AðsÞZðt − mÞ þ vðtÞ

t=1

where Z(t) is a column vector of the variables under consideration, C is the deterministic component composed of a constant, A(s) is a matrix of coefficients, m is the lag length and v(t) is a vector of random error terms. The optimal lag length is defined based on Hannan
Quinn information criterion. By construction, m(t) is uncorrelated with past Z(t)s. We adopt

120

CHRISTOS KOLLIAS AND STEPHANOS PAPADAMOU

a two-step methodology. First, a trivariate VAR model is estimated for each of the four countries in question
i.e. Spain, Germany, France and Great Britain
by including in the Z(t) column vector the following variables [log(cc_ESI), cc_TTI, cc_DTI]. Where cc is the country code (DE for Germany, FR for France, ES for Spain and UK for Great Britain), ESI is the Economic Sentiment Indicator as mentioned above, TTI is a Transnational Terrorism Index and DTI is a Domestic Terrorism Index to allow for the Enders et al. (2011) classification of the events. The last two indices are calculated following the methodology of Eckstein and Tsiddon (2004). More specifically, the daily index is defined as the natural logarithm of (e + number of fatalities + number of injuries) that occurred each day for each type of event (transnational or domestic). The terrorist events that took place during the weekend are summed up to the Monday’s figure. Then daily frequency of data is converted to monthly frequency by taking the average number over one month in order to be used with the indices that make up the ESI and are available on monthly basis. These indices are then introduced in the VAR models and impulse response analysis (IR analysis) is followed. Furthermore, given that aggregate data can potentially hide significant dynamics of the times series involved, IR analysis is applied on the VAR models that includes the main sub-components of the ESI instead of the composite index itself. More specifically, the VAR includes the following variables [cc_BUIL, cc_CONS, cc_RETA, cc_INDU, cc_TTI, cc_DTI]. These variables refer to reported economic sentiment on building activity, consumption, industrial activity and retail sales respectively.4 Hence, the main goal of the analysis that follows is to shed light on the direction as well as the duration of the effects induced by terrorism on economic sentiment.

THE FINDINGS Figs. 1
4 present the IR analysis from the VAR models that include the general ESI as well as the transnational and domestic terrorism indices. As an initial, broad and general observation, the findings seem to reveal a mild division in the sample of countries used. On the one hand, the ESI in France and Germany appears to be more sensitive and susceptible to terrorist activity vis-a`-vis the cases of Great Britain and Spain. In the latter two, the ESI does not appear to be rattled in any significant manner by terrorism. A plausible explanation for this division in the findings that can be

121

Terrorism and Economic Sentiment in European Countries Response of LOG (DE_ESI) to LOG (DE_ESI)

Response of LOG (DE_ESI) to DE_TTI

Response of LOG (DE_ESI) to DE_DTI

.04

.04

.04

.02

.02

.02

.00

.00

.00

–.02

–.02

–.02

–.04

–.04 5

10

15

20

25

30

–.04 5

Response of DE_TTI to LOG (DE_ESI)

10

15

20

25

30

5

Response of DE_TTI to DE_TTI

.010

.010

.010

.008

.008

.008

.006

.006

.006

.004

.004

.004

.002

.002

.002

.000

.000

.000

–.002

–.002

–.002

5

10

15

20

25

30

5

Response of DE_DTI to LOG (DE_ESI)

10

15

20

25

30

5

Response of DE_DTI to DE_TTI

.016

.016

.012

.012

.012

.008

.008

.008

.004

.004

.004

.000

.000

.000

–.004

–.004

–.004

Fig. 1.

10

15

20

25

30

5

10

15

20

25

15

20

25

30

10

15

20

25

30

Response of DE_DTI to DE_DTI

.016

5

10

Response of DE_TTI to DE_DTI

30

5

10

15

20

25

30

Response to Cholesky One S.D. Innovations ±2 S.E. The Case of Germany.

tentatively advanced is that unlike Germany and France, Spain and Great Britain have both experienced prolonged, harsh and bloody terrorist campaigns by domestic terrorist organisations. It is therefore possible that economic agents have grown accustomed to the comparatively regular occurrence of terrorist incidents. As a result, they are not particularly unsettled and rattled by such activity since, given its regularity, it is in a sense discounted by economic agents. Indeed this finding is akin to the results reported by Kollias, Papadamou, and Arvanitis (2013) where, when it comes to the impact terrorist incidents exert on the covariance between the stock and bond markets, the same mild division is unearthed for the same group of countries. A further and somewhat alternative explanation for this mild division between the countries in our sample can also be proposed.5 Germany and France have in the past experienced domestic extreme left-wing terrorism6 that was perceived to be a potential threat to the existing polities. Groups such as RAF or Action Directe in Germany and France, respectively, were left-wing revolutionary groups aspiring to overthrow the prevailing socioeconomic and political system. Since they presented a potential threat to

122

CHRISTOS KOLLIAS AND STEPHANOS PAPADAMOU Response of LOG (FR_ESI) to LOG (FR_ESI)

Response of LOG (FR_ESI)to FR_TTI

Response of LOG (FR_ESI) to FR_DTI

.04

.04

.04

.03

.03

.03

.02

.02

.02

.01

.01

.01

.00

.00

.00

–.01

–.01

–.01

–.02

–.02 5

10

15

20

25

30

–.02 5

Response of FR_TTI to LOG (FR_ESI)

10

15

20

25

30

5

Response of FR_TTI to FR_TTI

.0025

.0025

.0025

.0020

.0020

.0020

.0015

.0015

.0015

.0010

.0010

.0010

.0005

.0005

.0005

.0000

.0000

.0000

–.0005

–.0005

–.0005

5

10

15

20

25

30

5

Response of FR_DTI to LOG (FR_ESI)

10

15

20

25

30

5

Response of FR_DTI to FR_TTI

.012

.012

.008

.008

.008

.004

.004

.004

.000

.000

.000

–.004

–.004

–.004

10

15

20

25

30

5

10

15

20

25

20

25

30

10

15

20

25

30

30

5

10

15

20

25

30

Response to Cholesky One S.D. Innovations ±2 S.E. The Case of France.

Fig. 2.

Response of LOG (ES_ESI) to LOG (ES_ESI)

Response of LOG (ES_ESI) to ES_TTI

Response of LOG (ES_ESI) to ES_DTI

.04

.04

.04

.03

.03

.03

.02

.02

.02

.01

.01

.01

.00

.00

.00

–.01

–.01

–.01

–.02

–.02 5

10

15

20

25

30

–.02 5

Response of ES_TTI to LOG (ES_ESI)

10

15

20

25

30

5

Response of ES_TTI to ES_TTI

.012

.012

.008

.008

.008

.004

.004

.004

.000

.000

.000

–.004

–.004

–.004

5

10

15

20

25

30

5

Response of ES_DTI to LOG (ES_ESI)

10

15

20

25

30

5

Response of ES_DTI to ES_TTI

.020

.020

.015

.015

.015

.010

.010

.010

.005

.005

.005

.000

.000

.000

–.005

–.005

–.005

10

15

20

25

30

5

10

15

20

25

15

20

25

30

10

15

20

25

30

Response of ES_DTI to ES_DTI

.020

5

10

Response of ES_TTI to ES_DTI

.012

Fig. 3.

15

Response of FR_DTI to FR_DTI

.012

5

10

Response of FR_TTI to FR_DTI

30

5

10

15

20

25

30

Response to Cholesky One S.D. Innovations ±2 S.E. The Case of Spain.

123

Terrorism and Economic Sentiment in European Countries Response of LOG (UK_ESI) to LOG (UK_ESI)

Response of LOG (UK_ESI) to UK_TTI

Response of LOG (UK_ESI) to UK_DTI

.04

.04

.04

.03

.03

.03

.02

.02

.02

.01

.01

.01

.00

.00

.00

-.01

-.01

-.01

5

10

15

20

25

30

5

Response of UK_TTI to LOG (UK_ESI)

10

15

20

25

30

5

Response of UK_TTI to UK_TTI

.008

.008

.008

.006

.006

.006

.004

.004

.004

.002

.002

.002

.000

.000

.000

-.002

-.002

-.002

5

10

15

20

25

30

5

Response of UK_DTI to LOG (UK_ESI)

10

15

20

25

30

5

Response of UK_DTI to UK_TTI

.020

.020

.015

.015

.015

.010

.010

.010

.005

.005

.005

.000

.000

.000

-.005

-.005

-.005

Fig. 4.

10

15

20

25

30

5

10

15

20

25

15

20

25

30

10

15

20

25

30

Response of UK_DTI to UK_DTI

.020

5

10

Response of UK_TTI to UK_DTI

30

5

10

15

20

25

30

Response to Cholesky One S.D. Innovations ±2 S.E. The Case of the United Kingdom.

the existing polity, this may still influence how individuals perceive current terrorism in Germany and France, not having updated and adapted their beliefs and reaction to the attributes of present day terrorism. To reinforce this point on whether and how people update and adjust their beliefs on momentous events such as terrorism in this case, as Caruso (2010) suggests, individuals may be viewed as being cognitive-dissonant with respect to tragic experiences such as a terrorist attack or war and conflict in general. On the other hand, in Spain and Great Britain, Basque and Irish terrorism may not be perceived to present a potential threat to the existing polity although terrorist attacks do still represent threats to personal safety and disrupt daily social and economic life. In other words, terrorist activities that present a challenge or potential threat to the existing polity would tend to generate uncertainty vis-a´-vis the rest that
ceteris paribus
would tend to affect risk perceptions. Following Knight’s (1921) famous distinction, risk is present in situations where outcomes were unknown but governed by probability distributions known at the outset, whereas uncertainty also implies random outcomes but governed by an unknown probability model. Hence, the generation of uncertainty has a more pronounced

124

CHRISTOS KOLLIAS AND STEPHANOS PAPADAMOU

and hence statistically traceable effect on market agents’ sentiment. This may explain the difference unearthed not only between the countries but also between the types of events, i.e. domestic vs. transnational terrorist attacks. In particular, as can be easily deduced from the visual inspection of Figs. 1
4, in the cases of France and Germany a shock on domestic terrorism activity has a statistically significant negative effect on the ESI. In particular, in the case of the German ESI, it bottoms after almost 12 months while in the case of France in less than 10 months. In contrast, transnational attacks do not seem to have an effect on the ESI in Germany as well as in France which, although negative, is not statistically significant. Common finding shared by France, Germany and Spain show the higher persistence that a domestic terrorist shock has on itself compared to a transnational one. Moreover, only in the case of Germany, a shock on domestic terrorist activity seems to have a significant positive effect on transnational activity over the next five months. Looking to the responses to terrorist activity in the cases of Spain and Great Britain the reactions are not statistically significant. As noted above, this can partially be explained by the fact that economic agents in these countries may be more accustomed to terrorism activity given that both countries have suffered by prolonged terrorist campaigns by domestic terrorist groups such as the Provisional IRA (Irish Republican Army) and ETA (the armed Basque nationalist and separatist organization Euskadi Ta Askatasuna). Our analysis proceeds by decomposing the ESI into its main subcomponents in order to uncover any underlying trends and mechanisms. Figs. 5
8 present the results of this analysis. Again, a mild division seems to emerge between France and Germany on the one hand and Spain and Great Britain on the other. More specifically, in the case of Germany, the negative effect of domestic terrorism on the ESI can mainly be attributed to the adverse (i.e. pessimistic) impact on consumption, industrial production and retail sales. Comparing time reactions, it appears that the latter two respond significantly from the first months while consumption follows. In the case of France, responsibility for the negative reaction of the ESI to the domestic terrorist shock seems to be the result of the aggregation of the negative reactions, the building activity, retail sales and consumption subcomponents of the ESI. The negative effect on economic sentiment in retail sales holds for almost 15 months while for the economic sentiment in the other sectors the effects wither away a bit faster, in less than 10 months. Comparing France and Germany at this stage the main difference is traceable between building and industrial activity. Disaggregation analysis for the other two countries
i.e. Spain and Great Britain
does not reveal

125

Terrorism and Economic Sentiment in European Countries Response of DE_BUIL to DE_TTI

Response of DE_BUIL to DE_DTI

4

4

3

3

2

2

1

1

0

0

–1

–1

–2

–2 5

10

15

20

25

30

5

Response of DE_CONS to DE_TTI 3

3

2

2

1

1

0

0

–1

–1

–2

15

20

25

30

–2 5

10

15

20

25

30

5

Response of DE_INDU to DE_TTI

10

15

20

25

30

Response of DE_INDU to DE_DTI

4

4

2

2

0

0

–2

–2

–4

–4 5

10

15

20

25

30

5

Response of DE_RETA to DE_TTI

10

15

20

25

30

Response of DE_RETA to DE_DTI

6

6

4

4

2

2

0

0

–2

–2 5

10

15

20

25

30

5

Response of DE_TTI to DE_TTI .012

.012

.008

.008

.004

.004

.000

.000

–.004

–.004 5

10

15

20

25

10

15

20

25

30

Response of DE_TTI to DE_DTI

30

5

Response of DE_DTI to DE_TTI

10

15

20

25

30

Response of DE_DTI to DE_DTI

.012

.012

.008

.008

.004

.004

.000

.000

–.004

–.004 5

Fig. 5.

10

Response of DE_CONS to DE_DTI

10

15

20

25

30

5

10

15

20

25

30

Responses Concerning Economic Sentiment from Main Sectors in Case of Germany.

126

CHRISTOS KOLLIAS AND STEPHANOS PAPADAMOU Response of FR_BUIL to FR_TTI

Response of FR_BUIL to FR_DTI

8

8

6

6

4

4

2

2

0

0

–2

–2

–4

–4 5

10

15

20

25

30

5

Response of FR_CONS to FR_TTI 4

4

3

3

2

2

1

1

0

0

–1

–1

–2

15

20

25

30

–2 5

10

15

20

25

30

5

Response of FR_INDU to FR_TTI

10

15

20

25

30

Response of FR_INDU to FR_DTI

4

4

3

3

2

2

1

1

0

0

–1

–1

–2

–2 5

10

15

20

25

30

5

Response of FR_RETA to FR_TTI

10

15

20

25

30

Response of FR_RETA to FR_DTI

6

6

4

4

2

2

0

0

–2

–2

–4

–4 5

10

15

20

25

30

5

Response of FR_TTI to FR_TTI .003

.003

.002

.002

.001

.001

.000

.000

–.001

–.001 5

10

15

20

25

30

5

.012

.008

.008

.004

.004

.000

.000

–.004

–.004 10

15

20

25

15

20

25

30

10

15

20

25

30

Response of FR_DTI to FR_DTI

.012

5

10

Response of FR_TTI to FR_DTI

Response of FR_DTI toFR_TTI

Fig. 6.

10

Response of FR_CONS to FR_DTI

30

5

10

15

20

25

30

Responses Concerning Economic Sentiment from Main Sectors in Case of France.

127

Terrorism and Economic Sentiment in European Countries Response of ES_BUIL to ES_TTI

Response of ES_BUIL to ES_DTI

12

12

8

8

4

4

0

0

–4

–4 5

10

15

20

25

30

5

Response of ES_CONS to ES_TTI 4

4

3

3

2

2

1

1

0

0

–1

–1

–2

15

20

25

30

–2 5

10

15

20

25

30

5

Response of ES_INDU to ES_TTI

10

15

20

25

30

Response of ES_INDU to ES_DTI

4

4

3

3

2

2

1

1

0

0

–1

–1

–2

–2 5

10

15

20

25

30

5

Response of ES_RETA to ES_TTI 6

6

4

4

2

2

0

0

–2

–2 5

10

15

20

25

30

5 .012

.008

.008

.004

.004

.000

.000

–.004

–.004 10

15

20

25

30

5

Response of ES_DTI to ES_TTI .020

.015

.015

.010

.010

.005

.005

.000

.000

–.005

–.005 10

15

20

25

20

25

30

10

15

20

25

30

10

15

20

25

30

Response of ES_DTI to ES_DTI

.020

5

15

Response of ES_TTI to ES_DTI

.012

5

10

Response of ES_RETA to ES_DTI

Response of ES_TTI to ES_TTI

Fig. 7.

10

Response of ES_CONS to ES_DTI

30

5

10

15

20

25

30

Responses Concerning Economic Sentiment from Main Sectors in Case of Spain.

128

CHRISTOS KOLLIAS AND STEPHANOS PAPADAMOU Response of UK_BUIL to UK_DTI

Response Responseof ofUK_BUIL UK_BUI tto oUK_TTI UK_TTI 6

6

4

4

2

2

0

0

–2

–2 –4

–4 5

10

15

20

25

5

30

Response of UK_CONS to UK_TTI 4

4

3

3

2

2

1

1

0

0

–1

–1 5

10

15

20

25

30

5

Response of UK_INDU to UK_TTI 6

4

4

2

2

0

0

–2

–2 10

15

20

25

30

5

Response of UK_RETA to UK_TTI 6

4

4

2

2

0

0

–2

–2 10

15

20

25

30

5

Response of UK_TTI to UK_TTI

25

30

10

15

20

25

30

10

15

20

25

30

10

15

20

25

30

Response of UK_TTI to UK_DTI

.008

.008

.006

.006

.004

.004

.002

.002

.000

.000

–.002

–.002

–.004

–.004 5

10

15

20

25

30

5

Response of UK_DTI to UK_TTI

10

15

20

25

30

Response of UK_DTI to UK_DTI

.020

.020

.015

.015

.010

.010

.005

.005

.000

.000

–.005

–.005 –.010

–.010 5

Fig. 8.

20

Response of UK_RETA to UK_DTI

6

5

15

Response of UK_INDU to UK_DTI

6

5

10

Response of UK_CONS to UK_DTI

10

15

20

25

30

5

10

15

20

25

30

Responses Concerning Economic Sentiment from Main Sectors in Case of the United Kingdom.

Terrorism and Economic Sentiment in European Countries

129

a different picture from the one presented in case of aggregate ESI data. Perhaps the only finding worth mentioning is that there is a positive shortlived response of economic sentiment in the case of the building sector for Great Britain. This result may cautiously be interpreted as suggesting that economic agents are mainly influenced by the fact that terrorist-inflicted damages in infrastructure and buildings will spur building activity. However, this is the case only for transnational terrorism. Finally, in Spain economic sentiment concerning consumption seems to be affected negatively but the magnitude of this effect is quite weak and not particularly pronounced.

CONCLUSIONS Terrorist attacks are momentous unforeseen events that can and do severely disrupt in an adverse and bloody manner the normal, daily routine of economic life. They invariably cause widespread havoc, destroy capital and infrastructure and inflict extensive damages as well as losses of life. Terrorism is widely regarded as a public bad vis-a`-vis security
a public good
affecting the subjective well-being of citizens. As studies have shown, citizens’ risk-perceptions and risk-assessment are affected by largescale terrorist acts (Bozzoli & Muller, 2011; Drakos & Muller, 2011; Echebarria-Echabe & Fernandez-Guede, 2006; Turvey, Onyango, Cuite, & Hallman, 2010). This chapter, using impulse response analysis (IR analysis), examined the impact terrorism has on the ESI and its components in the case of four major EU states: Germany, France, Spain and Great Britain. Using the Enders et al. (2011) decomposition of terrorist events into domestic and transnational, it appeared that domestic incidents seemed to exert a greater adverse impact vis-a`-vis the latter. In broad terms, the findings reported herein revealed a division in the group of countries used as the vehicles of our empirical investigation. The effects were more pronounced and evident in the case of the former two
France and Germany
while the ESI in the latter two
Spain and Great Britain
did not appear to be particularly affected by terrorist acts. Possible tentative explanations that were proposed can be summarised as follows. On the one hand, both Great Britain and Spain have experienced long-term systematic terrorist campaigns by domestic organisations such as the Provisional IRA and ETA, respectively. Given that for all intends and purposes, terrorism was for many years an embedded feature of the political scene in these two countries it is possible to cautiously argue that economic agents have grown

130

CHRISTOS KOLLIAS AND STEPHANOS PAPADAMOU

accustomed to the comparatively regular occurrence of such events and therefore they are discounted by them. Additionally, the findings may also be explained in terms of the fact that the terrorism experienced in the past by Germany and France that presented a potential threat to the polity itself, whereas this is not the case with Basque and Irish terrorism in Spain and Great Britain. Given that the former type is more prone to induce uncertainty rather than risk, it is possible that this has left indelible imprints, creating some form of path-dependency and memory among individuals in France and Germany vis-a`-vis Great Britain and Spain. This argumentation satisfactorily explains why the ESI in the former two was more susceptible to domestic rather than transnational terrorist events.

NOTES 1. Data for the Economic Sentiment Indicator and its sub-components are provided on a monthly frequency by Eurostat. 2. The surveys are also conducted in the applicant countries for EU membership. 3. We sincerely thank Todd Sandler for readily providing the Enders et al. (2011) data set used here. 4. Data for sentiment on the services sector are provided from 1997 onwards for all of the countries studied (with the exception being France). Due to this reason this indicator is not included in our analysis. 5. We thank Andrea Locatelli and Raul Caruso for this astute suggestion. 6. Terrorist organisations such as the Red Army Faction (RAF) commonly known as the Baader
Meinhof group in Germany, Action Directe in France or the Brigate Rosse in Italy operated during the 1970s and early 1980s and were left-wing revolutionary terrorist groups.

ACKNOWLEDGEMENTS The chapter has greatly benefited from insightful comments and constructive suggestions by the editors of this volume, Raul Caruso and Andrea Locatelli. The usual disclaimer applies.

REFERENCES Aarle, B., & Kappler, M. (2012). Economic sentiment, shocks and fluctuations in economic activity in the Euro area and the USA. Intereconomics Review of European Economic Policy, 47(1), 44
51.

Terrorism and Economic Sentiment in European Countries

131

Abadie, A., & Gardeazabal, J. (2008). Terrorism and the world economy. European Economic Review, 52(1), 1
27. Amelie, C., & Darne, O. (2006). Large shocks and the September 11th terrorist attacks on international stock markets. Economic Modelling, 23(4), 683
698. Arin, K., Ciferri, D., & Spagnolo, N. (2008). The price of terror: The effects of terrorism on stock market returns and volatility. Economics Letters, 101(3), 164
167. Asteriou, D., & Siriopoulos, C. (2003). The role of political instability in stock market development and economic growth: The case of Greece. Economic Notes, 29(3), 355
374. Blomberg, B., Hess, G., & Tan, D. (2011). Terrorism and the economics of trust. Journal of Peace Research, 48(3), 383
398. Bozzoli, C., & Muller, C. (2011). Perceptions and attitudes following a terrorist shock: Evidence from the UK. European Journal of Political Economy, 27(S1), S89
S106. Brissimis, S., & Magginas, N. (2006). Forward-looking information in VAR models and the price puzzle. Journal of Monetary Economics, 53, 1225
1234. Brounrn, D., & Derwall, J. (2010). The impact of terrorist attacks on international stock markets. European Financial Management, 16(4), 585
598. Bru¨ck, T. (Ed). (2007). The economic analysis of terrorism. London: Routledge. Bru¨ck, T., Llussa´, F., & Tavares, J. (2011). Entrepreneurship: The role of extreme events. European Journal of Political Economy, 27(S1), S78
S88. Capelle-Blancard, G., & Laguna, M. (2009). How does the stock market respond to chemical disasters? Journal of Environmental Economics and Management, 59(2), 192
205. Caruso, R. (2010). On the nature of peace economics. Peace Economics, Peace Science and Public Policy, 16(2). Caruso, R., & Gavrilova, E. (2012). Youth unemployment, terrorism and political violence: Evidence from the Israeli/Palestinian conflict. Peace Economics, Peace Science and Public Policy, 18(2), 1554
1597. Caruso, R., & Schneider, F. (2011). The socio-economic determinants of terrorism and political violence in Western Europe (1994
2007). European Journal of Political Economy, 27(S1), S37
S49. Caruso R., & Schneider F. (2013). Brutality of Jihadist Terrorism. A contest theory perspective and empirical evidence in the period 2002
2010. Journal of Policy Modeling, 35(5), 685
696. Retrieved from http://dx.doi.org/10.1016/j.jpolmod.2012.12.005 Chesney, M., Reshetar, G., & Karaman, M. (2011). The impact of terrorism on financial markets: An empirical study. Journal of Banking and Finance, 5(2), 253
267. Crain, N., & Crain, W. M. (2006). Terrorised economies. Public Choice, 128(1
2), 317
349. Drakos, K. (2010a). Terrorism activity, investor sentiment, and stock returns. Review of Financial Economics, 19(3), 128
135. Drakos, K. (2010b). The determinants of terrorist shock’s cross-market transmission. The Journal of Risk Finance, 11(2), 147
163. Drakos, K. (2011). Behavioral channels in the cross-market diffusion of major terrorist shocks. Risk Analysis, 31(1), 143
159. Drakos, K., & Muller, C. (2011). Terrorism risk concern in Europe. Economics Letters, 112(2), 195
197. Easaw, J., & Ghoshray, A. (2008). The cyclical nature of consumer sentiments indices in the US and UK. Journal of Socio-Economics, 37(5), 1994
1998. Echebarria-Echabe, A., & Fernandez-Guede, E. (2006). Effects of terrorism on attitudes and ideological orientation. European Journal of Social Psychology, 36(2), 259
265.

132

CHRISTOS KOLLIAS AND STEPHANOS PAPADAMOU

Eckstein, Z., & Tsiddon, D. (2004). Macroeconomic consequences of terror: Theory and the case of Israel. Journal of Monetary Economics, 51(5), 971
1002. Enders, W., & Sandler, T. (2012). The political economy of terrorism (2nd ed.). Cambridge: Cambridge University Press. Enders, W., Sandler, T., & Gaibulloev, K. (2011). Domestic versus transnational terrorism: Data, decomposition, and dynamics. Journal of Peace Research, 48(3), 319
337. Frey, B., Luechinger, S., & Stutzer, A. (2007). Calculating tragedy: Assessing the costs of terrorism. Journal of Economic Surveys, 21(1), 1
24. Frey, B., Luechinger, S., & Stutzer, A. (2009). The life satisfaction approach to valuing public goods: The case of terrorism. Public Choice, 138(3
4), 317
345. Freytag, A., Kruger, J., Meierriek, D., & Schneider, F. (2011). The origins of terrorism: Cross-country estimates of socio-economic determinants of terrorism. European Journal of Political Economy, 27(S1), S5
S16. Gaibulloev, K., & Sandler, T. (2008). Growth consequences of terrorism in Western Europe. KYKLOS, 61(3), 411
424. Ghonghadze, J., & Lux, Th. (2012). Modeling the dynamics of EU Economic Sentiment Indicators: An interaction-based approach. Applied Economics, 44(24), 3065
3088. Guidolin, M., & La Ferrara, E. (2010). The economic effects of violent conflict: Evidence from asset market reactions. Journal of Peace Research, 47(6), 671
684. Hirshleifer, D. (2001). Investor psychology and asset pricing. Journal of Finance, 56(4), 1533
1597. Kamstra, M. J., Kramer, L. A., & Levi, M. D. (2003). Winter blues: A sad stock market cycle. American Economic Review, 93(1), 324
343. Kaplanski, G., & Levy, H. (2010). Sentiment and stock prices: The case of aviation disasters. Journal of Financial Economics, 95(2), 174
201. Knight, F. (1921). Risk, uncertainty and profit. Cambridge: The Riverside Press. Kollias, C., Manou, E., Papadamou, S., & Stagiannis, A. (2011a). Stock markets and terrorist attacks: Comparative evidence from a large and a small capitalization market. European Journal of Political Economy, 27(S1), S64
S77. Kollias, C., Papadamou, S., & Arvanitis, V. (2013). Does terrorism affect the stock-bond covariance? Evidence from European countries. Southern Economic Journal, 79(4), 832–848. Kollias, C., Papadamou, S., & Stagiannis, A. (2011b). Terrorism and capital markets: The effects of the Madrid and London bomb attacks. International Review of Economics and Finance, 20(4), 532
541. Kollias, C., Papadamou, S., & Stagiannis, A. (2012). The financial spillovers of the Gulf of Mexico oil accident. Empirical Economics Letters, 11(6), 633
643. Krieger, T., & Meierrieks, D. (2011). What causes terrorism? Public Choice, 147(1–2), 3–27. Loewenstein, G. (2000). Emotions in economic theory and economic behavior. American Economic Review, 90(2), 426
432. Lopes, M. P. (2011). A psychological explanation of economic cycles. Journal of SocioEconomics, 40(5), 652
659. Lucey, B., & Dowling, M. (2005). The role of feelings in investor decision-making. Journal of Economic Surveys, 19(2), 211
237. Nikkinen, J., Omran, M., Sahlstrom, P., & Aijo, J. (2008). Stock returns and volatility following the September-11 attacks: Evidence from 53 equity markets. International Review of Financial Analysis, 17(1), 27
46. Nikkinen, J., & Vahamaa, S. (2010). Terrorism and stock market sentiment. The Financial Review, 45(2), 263
275.

Terrorism and Economic Sentiment in European Countries

133

Nilsson, R. (2000). Confidence indicators and composite indicators. Paper presented at the CIRET Conference, Paris, 10
14 October. Noy, I. (2009). The macroeconomic consequences of disasters. Journal of Development Economics, 88(2), 221
231. Papadamou, S., & Siriopoulos, C. (2012). Banks’ lending behavior and monetary policy: Evidence from Sweden. Review of Quantitative Finance and Accounting, 38(2), 131
148. Rose, A., & Blomberg, B. (2010). Total economic consequences of terrorist attacks: Insights from 9/11. Peace Economics, Peace Science and Public Policy, 16(1). Schneider, G., & Troeger, V. (2006). War and the world economy: Stock market reactions to international conflicts. Journal of Conflict Resolution, 50(5), 623
645. Shughart II, W. F. (2006). An analytical history of terrorism, 1945–2000. Public Choice, 128(1–2), 7–39. Siokis, F., & Kapopoulos, P. (2007). Parties, elections and stock market volatility: Evidence from a small open economy. Economics and Politics, 19(1), 123
134. Turvey, C., Onyango, B., Cuite, C., & Hallman, W. (2010). Risk, fear, bird flu and terrorists: A study of risk perceptions and economics. Journal of Socio-Economics, 39(1), 1
10. Vuchelen, J. (2004). Consumer sentiment and macroeconomic forecasts. Journal of Economic Psychology, 25(4), 493
506. Wisniewski, T. (2009). Can political factors explain the behaviour of stock prices beyond the standard present value models? Applied Financial Economics, 19(23), 1873
1884.

CHAPTER 7 TERRORISM AS “A POLITICAL WORLD”: IDENTITY, STRATEGY, VALUES Damiano Palano ABSTRACT Purpose
The aim of the chapter is to outline a framework for a cultural analysis of terrorism. It is based on an analogy between the logic that characterizes terrorism and the logic that characterizes the “worlds” of artistic and cultural production. Design methodology/approach
This chapter critically examines some explanations of the logic of terrorism which respectively assign a central role to individual psychology, the rational choice of the group and values. It is therefore based on secondary sources. Findings
The chapter argues that the logic of terrorism can be fully understood only as a result of the “cultural” construction of collective and individual identities. Research limitations/implication
The chapter provides a framework for a cultural analysis of terrorism.

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 135
157 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022012

135

136

DAMIANO PALANO

Originality/value of the chapter
The chapter borrows the concept of a “political world” and makes of it a central tool to explain the set of symbolic products that characterizes terrorism. Keywords: Terrorism; psychology of terrorism; instrumental rationality; axiological rationality; radicalization; art worlds

INTRODUCTION After the wave of political violence in the last decade of the 19th century, a debate on the “psychology of the terrorist” emerged among psychiatrists, criminologists, and sociologists, who tried to discover the deeper motivations that lead individuals to perpetrate terrorist attacks. The answers differed significantly. For example, in the aftermath of the assassination of French President Sadi Carnot, the criminologist Alexandre Lacassagne (1894, p. 36) wrote that the perpetrator
an Italian anarchist called Sante Caserio
should not be seen as a madman but as a “fanatic murderer,” a beˆte humaine. Instead, Cesare Lombroso (1894), the widely acknowledged father of the Italian school of criminal anthropology, argued that Caserio was a normal individual, but that social and political circumstances had temporarily undermined his psychic equilibrium. In this vein, Scipio Sighele developed a more radical thesis. While analyzing political crimes perpetrated by organized groups, Sighele (1897) argued that crimes of anarchic sects were always the consequence of altruistic behavior, namely the result of individual dedication to the interests of the group. Sighele did not consider terrorists either irrational or driven by a rationality similar to that of homo economicus. On the contrary, he argued that terrorists were driven by a sort of collective rationality, rooted in the individual psyche, which pushed the individual to commit “altruistic” behavior. Such an approach offers some insights into the rationale that triggers political violence. In this chapter, I argue that this old theoretical approach still sheds some light on the logic of terrorism, and in particular on the theoretical connection between the apparent “irrationality” of the individual and the “rationality” of a group. In particular, the idea of “fragmentation” of the psyche holds central place in our discussion, although the basis of Sighele’s argument is inadequate. Sighele explained political violence referring to a primordial and collective “unconscious,” but here we will consider terrorism as the result of a social process
whose end result is the cultural construction of individual and group identity.

Terrorism as “A Political World”: Identity, Strategy, Values

137

This chapter is constructed as follows: in the first three paragraphs, I examine some explanations of the logic of terrorism, which respectively assign a central role to individual psychology, group rationality, and values. Although each of these analytical perspectives provides important elements, I argue that they underestimate the role of “culture” or consider it as only an “external” factor. More specifically, these perspectives consider identity as an invariant element and not as a cultural construct, which is a product of social interaction and therefore subject to constant change. In the fourth section, I propose an analogy between the logic that characterizes terrorism and the logic that characterizes the “worlds” of artistic and cultural production, and I suggest the use of the notion of a “political world.” In the fifth paragraph, I propose a theoretical framework for the relationship between community, organization, and militants.

IN THE MIND OF THE TERRORIST In recent decades, research on individual motivations for terrorism has followed very different directions. In general, many attempts have been made to identify the traits of a sort of “psychology of the terrorist.” Until now, however, a well-defined profile of the individuals prone to terrorist action has not yet emerged because of two rather significant obstacles. The first problem arises from the difficulty of circumscribing criminal behavior. Defining “terrorism” is difficult because the shapes taken by political violence change significantly over time, and because the concept of “terrorism” is an “essentially contested concept” (Gallie, 1956), namely a concept with a strong judgmental component. All proposed definitions thus tend to reflect more or less explicit political objectives of legitimization or delegitimization (Evangelista, 2008; Ignatieff, 2004; Laquer, 1977). The second problem is represented by the difficulty of carrying out empirical research: data for psychological research on members of terrorist organizations are in fact quite limited and often indirect. In addition, “when researchers do manage to interview terrorists, the latter may provide a highly distorted description of their activities or even deliberately mislead the interviewer” (Moghadam, 2006, p. 16). Because of all these difficulties, contemporary research tends to provide only inconclusive answers to the most radical questions about the psychology of the terrorist. The idea of a connection between terrorist behavior and some psychopathologies has not been abandoned, but the results in this field are quite

138

DAMIANO PALANO

controversial. In general terms, contemporary psychiatry identifies adult behavioral disorders according to a multiaxial classification scheme: (1) Axis I refers to major clinical disorders, such as schizophrenia, psychosis, and major depression; (2) Axis II refers to personality disorders, such as sociopathy or antisocial personality disorders. These categories do not set clear boundaries. Usually, however, researchers tend to believe that individuals who suffer from Axis II disorders are able to distinguish between right and wrong (Victoroff, 2005, p. 12). Although data are scarce, research conducted on individuals belonging to a terrorist organization has not detected the presence of either Axis I or Axis II disorders: in this respect, the terrorist does not display strong psychopathologies. If, for this reason, scholars tend to exclude the possibility that terrorist behavior can be reliably explained by mental illness (Sageman, 2004, p. 84), some researchers have attempted to identify the traits of a “terrorist personality”: a personality that, without displaying marked disorders, is predisposed to enter terrorist organizations. In this respect, Pearlstein (1991, p. IX) points out that “the individual who becomes and remains a political terrorist generally appears to be psychologically molded by certain narcissistic personality disturbances.” Several studies have also traced the presence of the psychological mechanisms of externalizations and splitting, which are also found in individuals with borderline and narcissistic disturbances (Ferracuti, 1982; Ferracuti & Bruno, 1981; Laquer, 1987; Post, 1986a, 1986b). Therefore, although most terrorists do not show significant psychopathologies, and although researchers cannot find a real “terrorist personality,” according to Post (1998, p. 31) “it appears that people who are aggressive and action-oriented, and who place greater than normal reliance on the psychological mechanism of externalization and splitting, are disproportionately represented among terrorists.” The recent proliferation of suicide bombings has brought the issue of the psychology of the terrorist back into the debate. The old question about a correlation between terrorist behavior, which in this case coincides with the extreme choice of suicide, and the presence of personality disorders has resurfaced accordingly. In these cases, according to Ariel Merari (2010), social and cultural factors cannot be considered as the only explanatory variables, because, for example, they do not explain why only some militants become suicide bombers. Moreover, the absence of serious personality disorder does not exclude the presence of personality traits that are predispositions to political radicalization and political violence. For example, conducting a survey of a small group of Palestinian terrorists, Merari reports the presence of some common psychological traits, and in particular the presence

Terrorism as “A Political World”: Identity, Strategy, Values

139

of two models of personality that predispose individuals to suicide missions: a dependent
avoidant personality, which makes individuals vulnerable to the influence of other people and therefore easily swayed by individuals with some authority; an impulsive
unstable personality, which makes subjects significantly impulsive, fickle, self-destructive. As a result, the explanation of the behavior of suicide bombers, according to Merari, must be sought mainly in the personality characteristics of the subjects, which are particularly sensitive to outside influence (Merari, 2010, p. 222). Research on personality factors, however, has proved to be generally inconclusive (Braungart & Braungart, 1992; Ross, 1994). Attempts to unveil a connection between terrorist behavior and personality traits have been criticized, both because of the absence (or scarcity) of a robust empirical basis and because of the way in which this claim was made. Martha Crenshaw, for example, argues that it is very difficult to identify a “terrorist personality”
echoing the “authoritarian personality” identified by Adorno (Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950)
because what drives someone into an armed organization is “a combination of ego-defensive needs, cognitive process, and socialization, in interaction with a specific situation” (Crenshaw, 1981, p. 390). According to Borum (2004), personality traits are not good predictors of terrorist behavior because a more important role is played by other factors, such as perceived injustice, identity, and the need for belonging. In addition, most scholars believe that the characteristics of “new terrorism” (Laquer, 1999) are not substantially different from those of “old terrorism.” “The premise that there is a ‘new’ terrorism depends on the assumption that motivations have changed,” says Crenshaw, but “psychological studies do not support this assumption” (Crenshaw, 2006, p. 55). Thus, even in this case, the main motivations of suicide terrorism are not to be sought in psychological factors but in political and ideological causes (Horgan, 2005). In a comprehensive review of the literature, Victoroff also noted that most reflections on the psychological determinants of terrorism are based on a very small number of empirical observations. The available research, however, tends to converge on some general conclusions. In the first place, “terrorist behavior is probably determined by a combination of innate factors, biological factors, early developmental factors, cognitive factors, temperament, environmental influences, and group dynamics.” Nevertheless, “theories that claim the predominance of one of these influences over the others are premature, since no studies have systematically examined more than one or two of these factors, let alone empirically examined one while controlling for the others.” Second, “terrorists are psychologically

140

DAMIANO PALANO

extremely heterogeneous” (all quotes are from Victoroff, 2005, pp. 34
35). In spite of all these difficulties, Victoroff stresses that it is possible to identify four elements of the “typical” terrorist: a. High affective valence regarding an ideological issue. b. A personal stake
such as strongly perceived oppression, humiliation, or persecution; an extraordinary need for identity, glory, or vengeance, or to drive for expression of intrinsic aggressivity
that distinguishes him or her from the vast majority of those who fulfill characteristic a. c. Low cognitive flexibility, low tolerance for ambiguity, and elevated tendency toward attribution error. d. A capacity to suppress both instinctive and learned moral constraints against harming innocents, whether due to intrinsic or acquired factors, individual or group forces
probably influenced by a, b, and c. (Victoroff, 2005, p. 35) Therefore, in general, perpetrators of terrorist attacks are not affected by significant personality disorders. Moreover, if one restricts the analysis to the action of organized political groups that use violence as a political tool, it appears that the presence of individuals with personality disorders is at least very infrequent. In any case, the available data suggest that “the outstanding common characteristic of terrorists is their normality” (Crenshaw, 1981, p. 390). The answer to the question “What leads an individual to join an armed organization?” should therefore be sought not in individual psychological traits but elsewhere. Terrorism is indeed a group activity. Therefore it is “typically not the result of psychopathology or a single personality type,” because “shared ideological commitment and group solidarity are much more important determinants of terrorist behavior than individual characteristics” (Crenshaw, 2000, p. 409). This is one of the reasons why research on the psychology of the terrorist tends to exclude explanations based on a single factor (Bloom, 2005; Canter, 2009; Crenshaw, 1986; Pedahzur, 2005; Stern, 2003). In any case, any explanation of the phenomenon cannot neglect the collective dimension of terrorism. In Ricolfi’s words, for example, “the motivational drive to engage in suicide missions is likely to be found in a cocktail of feelings, which includes desire for revenge, resentment, and a sense of obligation towards the victims” (Ricolfi, 2005, p. 106). It is essential, however, to avoid limiting such a conclusion to merely describing a plurality of possible reasons (Kruglanski, Chen, Dechesne, Fishman, & Orehek, 2009), in order to understand the way in which the “collective” dimension of terrorism shapes and impels individual behavior.

Terrorism as “A Political World”: Identity, Strategy, Values

141

THE STRATEGIC LOGIC OF TERRORISM The collective dimension, as underlined by psychological research, gains central stage for those analyses that focus on the logic of terrorist organizations. In particular, many researchers in this field believe that terrorism is not the result of a purely irrational behavior. In contrast, the use of terrorism, even in its extreme forms, such as suicide attacks, can be interpreted as a rational behavior: it is a costs and benefits calculation that leads an organization to use certain tools to achieve specific goals (Caplan, 2006; Crenshaw, 1998; Gambetta, 2005; Pape, 2003, 2005; Sandler & Lapan, 1988; Sandler, Tschirhart, & Cauley, 1983; Sprinzak, 2000; Wilson, 2000). In other words, terrorist organizations, in their activities and in their planning of attacks, follow a strategic logic. Their action, therefore, is “rational,” because it is aimed at pursuing political objectives (e.g., the fight against a political regime, the competition with other terrorist groups, the internal competition for the conquest of leadership) (Caruso & Locatelli, 2008; Della Porta, 1990, 1995; Della Porta & Diani, 2006; Tarrow, 1989, 1994; Zald & McCarthy, 1980). The literature in this field is huge, and the relevance of these studies is widely recognized. The idea that terrorist organizations follow a strategic logic is often useful for understanding the motivations driving groups in their choice of tools and objectives. In addition, this research is an effective alternative to theories that explain terrorism exclusively as a product of “religion” or “environment.” In spite of these many merits, theories that assume that terrorist groups are rational encounter a number of relevant objections (Abrahms, 2006). First, it is hard not to recognize that sometimes the objectives pursued by the terrorists are clearly unrealistic (Crenshaw, 2000). Second, often the internal dynamics of terrorist groups are shaped by emotional experiences, which have a significant influence on the perception of reality and therefore on the behavior of individuals (Victoroff, 2005). In addition, the idea that organizations are driven by a strategic rationality leads to underestimation of the part that the organizational tradition plays in the definition of objectives, repertoires, and instruments: in other words, like any organization, even armed organizations have an organizational culture, which cannot be changed frequently (even in the presence of changes in the external environment) (Horowitz, 2010). Finally, if terrorist organizations are guided by an exclusively rational logic, then they should disband after having achieved their objectives: but (often), the groups do not dissolve; they survive, redefining their goals (Post, 1998).

142

DAMIANO PALANO

In addition to these problems, there is a relevant question concerning the specific theoretical link between the two different analytical levels
that is, the relationship between the strategic logic of the organization and the psychology of the terrorist. In contemporary studies on terrorism, three different analytical levels are usually considered: (1) the individual level, (2) the organizational level, and (3) the systemic level (Crenshaw, 1981; Hafez, 2007; Moghadam, 2006; Tosini, 2009, 2010, 2011). At the individual level, we consider the psychological, economic, social, and cultural profile of terrorists. At the organizational level, researchers study the strategies adopted by armed organizations to achieve their goals (e.g., long-term policy goals, recruitment of new members, and internal competition for leadership). With regard to the systemic level, the focus turns on the social, economic, political, and cultural factors that may favor the spread of terrorism: in this sense, we seek to understand, for example, if the presence of environmental factors determines the emergence and proliferation of terrorist activities (Caruso & Gavrilova, 2012; Caruso & Schneider, 2011, 2013). The analyses that claim that terrorism is the result of a strategic logic are focused on the organizational level: for this reason, they cannot consider the relationship between the organizational and individual levels. In fact, if a researcher considers the strategic logic of organizations as fully “rational,” the problem of explaining why an individual decides to join a clandestine organization, why s/he decides to carry out violent acts, and finally why can s/he is willing to commit an extreme form of “political sacrifice” still remains. The behavior of a suicide bomber cannot be easily accounted for in a perspective of pure instrumental rationality. In other words, although it is recognized that an organization acts in a “rational” way, it is difficult to admit that these rational criteria can motivate the militants/terrorists, who may lose their lives. A solution to this theoretical problem is found by drawing a distinction between individual and organizational logic. In general terms, it can be expected that strategy is decided by the leadership: militants are only tools in the hands of the leadership, and so they will accept orders from the organization, even in the case where such provisions entail the material risk or certainty of death. In this sense, it can also be assumed that terrorist behavior is “rational,” because it is based on the calculation of an economic cost benefit, albeit expressed in ideological terms (Elster, 2005). For obvious reasons, the interest pursued by the individual cannot be a direct material interest, and it also seems that only very rarely can their interest be identified with material interests in “after-effects” such

Terrorism as “A Political World”: Identity, Strategy, Values

143

as economic protection afforded to family members (Kihmi & Even, 2004, p. 826; Pedahzur, 2005; Tosini, 2009, pp. 113
114). The difficulty of connecting the two levels of analysis is particularly evident in the case of suicide terrorism. To solve this problem, Robert Pape (2005) in his famous study on the strategic logic of suicide terrorism applies the typology proposed by Emile Durkheim. According to Durkheim, we can distinguish three forms of suicide: (1) the egoistic suicide, which comes from a strong personal trauma and a low level of integration with the community; (2) the altruistic suicide, which comes from an excessive integration of the individual with the community: as an individual attaches great value to the community in which s/he lives, s/he is willing to kill her/himself for the public good; and (3) the fatalistic suicide, which occurs when the individual is subjected to an oppressive control that pushes her/him toward choosing suicide. According to Pape (2005), suicide terrorism can, in most cases, be attributed to altruistic suicide. In fact, suicide attacks are often organized by a group: in other words, they assume a wide social interaction and a unity of purpose within a large group of people. These elements demonstrate the presence of altruistic motives, namely the fact that bombers work in a team sharing collective goals: this is largely incompatible with the dynamics of egoistic suicide. Pape’s hypothesis offers an interesting perspective that effectively establishes a link between the individual and organizational level. Nevertheless, his hypothesis raises a problem. Although it is clear that the individual has an altruistic purpose, it remains entirely unclear who those “others” are that should benefit from the attack. In the event that the “others” are a group of family or friends, the “altruistic” connotation is clear. If the “others” are a much wider (and indeterminate) community, such as, for example, the “nation,” the “working class,” or the whole “human race,” the question becomes more complex. In the latter case, these are in fact “imagined communities” (Anderson, 1983), groups that do not really exist, but which are only symbolic, cultural, and ideological entities. In other words, Pape explores the difficulty of explaining in terms of rational behavior, a behavior that is not explainable in terms of material interests. Pape recognizes the weight of “culture” but, in this case, he seems to consider culture only as an environmental factor, a set of norms, value, and ideologies inherited from the past. In this sense, Pape’s hypothesis shows the limitations of theories that explain terrorism as a strategic logic. In other words, for rationalistic approaches, the interests that drive an individual’s actions are only material interests and must be defined in objective terms. These interests are therefore relatively constant and depend on the prospects

144

DAMIANO PALANO

of obtaining material benefits from participation in a particular collective action. In addition, we can extend the concept of personal interest to understand the choice of performing a suicide attack as the result of a “contract” or “exchange,” whereby an individual adopts the religious beliefs of a group in order to obtain solidarity and a sense of belonging (Wintrobe, 2006a, 2006b). As noted by Jon Elster (2009), however, it is difficult to define the behavior of a suicide bomber as “rational” on the basis of objective criteria (such as health, life expectancy, income). In other words, the decision of a bomber can be explained on the basis of “subjective” considerations: that is, we can try to understand the reasons that lead an individual terrorist to become a suicide bomber. It is very difficult to characterize her/his actions as “useful” or “advantageous” in terms of “objective” interest, that is, from the point of view of an external observer (Elster, 2005). From the perspective of instrumental rationality, the behavior of suicide bombers can only appear as “irrational” behavior, because it deviates from narrow self-interest, and because it fails to conform to rational expectations (Caplan, 2006). This set of difficulties, however, refers to a more general problem, which concerns the theoretical approaches that represent terrorism as a phenomenon driven by a strategic logic. At the core of these visions of collective action is the idea that each individual has a given set of preferences, and her/ his own actions descend from those interests and preferences. Consequently, the question of the origin and diversity of individual preferences can be completely neglected. In other words, it is important that the interests or preferences of an individual arise from a particular ideology, or even from psychic anomaly, but it is important that those preferences lead individual behavior toward a specific goal. The most problematic aspect is that, from this perspective, interests and preferences are always conceived as constant. Indeed, as argued by Pizzorno (1986, 1990, 1993, pp. 129
144), preferences and interests are not constant in time, because they are the product of individual identity: an identity which is formed by social interaction, through the recognition of others. In other words, the subject is constituted as a self only by participating in a collective action, and the subject makes her/his own choices to defend this identity: the choice between different preferences is in fact made to ensure continuity and consistency of the self. Although interests and preferences are based on a self, the self is not constant, because it is the result of a process of cultural construction: the subject constructs her/his own identity through recognizing others and defends this identity over time, making choices that confirm it. To this end, the rationality of an action is meaningful only within a context of social relations.

Terrorism as “A Political World”: Identity, Strategy, Values

145

TERRORISM AND AXIOLOGICAL RATIONALITY In his attempt to overcome the limits of rationalist explanations, Tosini (2009, 2010, 2011) tried to explain altruistic suicide by means of a noninstrumental rationality, an “axiological rationality” (Boudon, 2003, 2009; Cherkaoui, 2007), which corresponds to Weber’s concept of Wertrationalita¨t (rationality in accordance with moral demands) (1966). In this case, the subject thinks in rational terms, and “rationally” defines instruments to be adopted, including suicide. Rationality, however, is not aimed at achieving some sort of material interest, but only at complying with a value (and therefore complying with social, religious, or ideological norms). “Here, the decisions are made regardless of the consequences of the actions, […] the criterion to be taken into account is the commitment to certain values,” and value-commitment depends “on firm adhesion to specific beliefs which are assumed to have a binding and absolute validity” (Tosini, 2009, p. 79). The idea of axiological rationality marks an important point, because it considers a type of interest which stands in sharp contrast to material interest. Of course, there is no adequate evidence to demonstrate that the behavior of terrorists is actually driven by this kind of rationality. Nevertheless, much evidence suggests that, at least in some cases of suicide missions, conformity with political and religious values plays a significant role. From a theoretical point of view, this hypothesis is challenged by the problem of defining values. In other words, the idea of an axiological rationality assumes that societies define common beliefs and values that influence individual behavior. These values (and norms) are therefore conceived of as a reflection of social structure, which determines (or influences) actors’ behavior. Then the danger constitutes consideration of values, social norms, ideologies, and symbols only as “crystallized” elements
that is, as a legacy of the past. In essence, to get around the difficulties of an individualistic approach (which views the subject as an actor who acts like the classical homo economicus), this perspective assumes that the subject is guided by social structure (Gambetta, 1987). It shares the assumption that the action of the terrorist follows a script defined on a social level, and is therefore the result of values, norms, and ideologies which the individual has no chance to influence. From this perspective, the explanation of terrorism can actually call culture into question: for example, left-wing terrorism as perpetrated by the Red Brigade and RAF can be explained in terms of ideological

146

DAMIANO PALANO

fanaticism; the terrorism of ETA can be considered as a consequence of strong nationalist sentiment, and Islamic terrorism can be explained as a reflection of a specific religious and cultural tradition that celebrates holy war and martyrdom. The problem that this kind of explanation presents is not so different from Pape’s. Pape identifies “altruistic” motives, but he does not specify “who” are the “others” that the suicide bomber wants to help. More precisely, Pape presupposes that the “others” actually exist, whereas in fact (in cases where they do not coincide with the narrow family circle of the “martyr”) they are an “imagined community,” a symbolic construction. The explanation based on axiological rationality recognizes in a more conscious and sophisticated way the importance of culture, but tends nevertheless to transform “culture” into an “essence”: in other words, it tends to consider values, norms, symbols, and beliefs as “objective” elements, which cannot be changed by the subjects themselves. In this way, the hypothesis of axiological rationality tends to consider values, norms, ideologies, and symbols as a legacy of cultural tradition, but it seems to underestimate the fact that the “traditions”
and in particular political traditions
are “inventions,” a cultural product (Hobsbawm & Ranger, 1987). The vision of culture as a set of values that push individual behavior in a particular direction (and therefore interprets the behavior of the terrorist in terms of Wertrationalita¨t) has been radically questioned in recent decades. Arguing against functionalism, for instance, Swidler (1986) suggests the adoption of a tool-kit paradigm: according to this paradigm, culture is a tool-kit, a repertoire of symbols, styles, and meanings, which is used by active and competent social actors. As a result, culture is not to be understood as a set of elements inherited from the past and constant over time, but as a complex collection of elements constantly redefined by actors faced with specific problems and stresses. Therefore, from this perspective values, norms, beliefs, and symbols are central, but they are not “objective” elements: they are rather symbolic formations, patterns, and cultural idioms, which guide the behavior of the actors and, at the same time, empower them to take action, through the construction of identity, purpose, aspirations, preferences, and strategies for action (Laitin, 1986; Santoro, 1998, 2007; Sewell, 1992; Swidler, 1986). This view of culture, however, does not make axiological rationality pointless for our purposes: it just means that values cannot be conceived simply as a set of elements rooted in social structure. One way to overcome the difficulties of the idea of axiological rationality is perhaps to place the behavior of terrorists in a relational and symbolic context, in which the

Terrorism as “A Political World”: Identity, Strategy, Values

147

values, norms, beliefs, and symbols are not immutable elements inherited from tradition but cultural constructions in constant flux.

TERRORISM AS A POLITICAL WORLD Recognizing the role of culture is important for a broader understanding of political phenomena, but it is particularly relevant to the analysis of terrorism. Terrorism is indeed mainly a symbolic phenomenon, because it corresponds to a structured process by which actions are assigned political significance. As a matter of fact, the same action can be defined as an act of terrorism only as a result of a cultural process, and that is because an action is assigned a specific meaning. Therefore, the symbolic construction of terrorism follows the same mechanisms which characterize each process of cultural production
namely the production, distribution, reception, and contextualization of cultural products (Griswold, 1994). For this reason, it is possible to study the mechanisms of terrorist violence with the tools developed by sociology of culture to single out the roles, customs, and conventions shared by the groups involved in the creation of cultural products (Di Maggio & Hirsch, 1976; Hirsch, 1972; Powell & Di Maggio, 1991). From this perspective, since the actions of terrorist organizations can be considered as resulting from a mainly symbolic and cultural heritage, it is also possible to place them in a world similar to the symbolic art worlds studied by Howard S. Becker (1974, 1976, 1982). The notion of the art world is deeply indebted to Husserl’s phenomenological tradition and to symbolic interactionism (Danto, 1973, 2003; Dickie, 1964, 1974, 1984; Maanen, 2009). This notion focuses on shared conventions and thus escapes the drawbacks inherent in many representations of the relationship between art and society. Whereas traditional perspectives focus on the figure of the artist, or on the influences of the environment on the artist, the notion of the art world emphasizes the network of collective cooperation that makes art possible as a social phenomenon. Starting from this assumption, Becker (1976) focuses in particular on models of cooperation and on conventions that make it possible to consider a cultural product as “artistic.” In his definition, a world consists “of all those people and organizations whose activity is necessary to produce the kind of events and objects which that world characteristically produces.” Thus, an art world “consists of the people and organizations who produce those events and objects that the world defines as art” (Becker, 1976, p. 703). With this

148

DAMIANO PALANO

definition, clearly tautological, Becker cannot solve the problem of the meaning of art: in other words, Becker does not ask “why” humans produce objects that are defined as “art,” and does not clarify what “objective” elements transform an object into an artistic object. To put it simply, Becker merely recognizes that, in our society, individuals assign an “artistic” value to certain products, and he tries to reconstruct the cultural and esthetic conventions that regulate the professional art world (Becker, 1982; Maanen, 2009). There are at least three implications of the notion of the art world. First, art is not the product of a single artist, but is the result of the cooperation of all the people who collaborate to create an artistic product. This means therefore that the investigation of art worlds must consider all members of that world: people who conceive the idea of the work, those that perform it, those who prepare the tools, and, finally, the public. Second, it is perfectly possible that there are many worlds coexisting at one time. Finally, “an organized world is the source of whatever social value is ascribed to a work,” and “the interaction of all the involved parties produces a shared sense of the worth of what they collectively produce” (Becker, 1976, p. 705). The idea of art worlds cannot be mechanically applied to the study of terrorism. Obviously, armed clandestine organizations do not produce objects that are defined as art. What is common to these worlds, however, is that, in both cases, the meaning of a product (or an action) depends on the existence of a symbolic world. An object is considered as “art” only if there is a symbolic world (consisting of skilled professionals, public, academy, etc.), which, on the basis of esthetic conventions, recognizes that this product is an “artistic” object and its creator is an “artist.” In the same way, a violent action is recognized as having “political” value (or as being a terrorist action) only within a specific symbolic world, which can be defined
in analogy with Becker’ s art worlds
as a political world. In other words, only within a political world characterized by beliefs, ideologies, and conventions is it possible to recognize a specific violent action as a political action, and assign to the author the status of armed political activist or the pejorative name of “terrorist.” The same distinction between armed political struggle and terrorist violence is the result of discursive practices, which are the product of a relationship between instances of legitimization and delegitimization. From this perspective, the art worlds in which artists act and political worlds and armed organizations in which terrorists act have evident similarities. There is, however, a crucial difference concerning the activity to which meaning is assigned: in the case of artistic worlds, conventions define

Terrorism as “A Political World”: Identity, Strategy, Values

149

“what is art,” and, on the basis of this definition, establish both the boundaries between “art” and what does not have artistic value and the boundaries between “artists” and “craftsmen.” In the “political worlds” discursive practices define the identity of a political community: similarly to art worlds, the definition of a “political identity” does not take place through a clear definition of the “essence” of a community but by established boundaries. In other words, the construction of a political world is predominantly made by the classic distinction between friend (amicus) and enemy (hostis), and the end result is a process of legitimization of violence. In essence, the boundary between who belongs to the community and who is a stranger (and a threat) emerges in relation to the legitimation of violence: whereas violence against the friend is not legitimate (and therefore criminal), violence against the enemy is legitimate. This image of the political world is clearly indebted to Carl Schmitt’s conception of the political phenomenon, both for its emphasis on the contrast between friend and enemy and for the juxtaposition of political experience and aesthetic experience. In his famous essay on the concept of the “political,” Schmitt (1996) developed an analogy between what happens in the political field and what happens in the esthetic field. In essence, according to Schmitt, it is possible to distinguish between the idea of “beauty” and the criterion of “beautiful”: the idea of beauty changes over time; nevertheless, in any era it is possible to recognize the existence of a criterion of “beautiful” according to which it is possible to assess whether an object is more “beautiful” (or “uglier”) than another object. Similarly, Schmitt believes that it is not possible to say what the “essence” of “politics” is, because its forms and its contents are subject to continuous historical change. Conversely, according to the German scholar, it is possible to grasp the “criterion” of the “political,” namely the criterion for assessing whether a particular association, at a specific moment, is “more political” or “less political” than another. This criterion is, for Schmitt, the distinction between friend and enemy. More specifically, the distinction between friend and enemy is used to indicate the “utmost degree of intensity of a union or separation, of an association or dissociation” (Schmitt, 1996, p. 26). With this definition, Schmitt makes at least three points: first, the “political” is associated with the ability to “bind” humans (so that the extreme degree of the “political” coincides with the maximum degree of intensity of a union); second, this association does not rest on a stable, “natural,” durable identity, because this union is the product of a “political” division between insiders and outsiders, between friends and enemies; finally, the highest level of

150

DAMIANO PALANO

“politicization” corresponds to the maximum degree of separation and dissociation, because a “real” political union excludes the enemy from the boundaries of the community, and, overall, legitimizes physical suppression of the enemy (in other words, it legitimizes war). Relying on Schmitt’s hypothesis evidently raises a significant problem, namely the fact that the “political” is not fully comparable to the “beautiful,” the “good,” or the “profitable.” Nevertheless, the analogy between political and esthetic phenomena has an advantage at least: Schmitt does not ask “why” the concept of the “political” exists, and equally does not ask “why” the concept of “beautiful” exists. Implicitly, he supposes that it is rooted in “human nature,” but he actually focuses on another aspect, the form of political association. Despite this, Schmitt probably does not abandon the idea that “real” political association coincides with the Volk of the Romantic tradition. It is possible to give up this assumption and consider Schmitt’s hypothesis in “relativistic” (and non “essentialistic”) terms: in other words, we can assume that a political world has a purely contingent basis which establishes the boundary between insiders and outsiders, just as art worlds are rooted in historically changing conventions, which establish the boundary between art and craft and between artists and craftsmen. Similar boundaries are probably crucial for both the strategy pursued by armed organizations and the actions of terrorists themselves.

THE MAKING OF IDENTITIES IN A POLITICAL WORLD Drawing from a pattern used to consider suicide attacks (Gill, 2007; Moghadam, 2006; Pape, 2005; Tosini, 2009, 2010, 2011), we can consider terrorism as the result of the interaction between: (1) a community of support; (2) armed organizations, which are more or less structured and long-lasting, and claim to plan actions to achieve certain political goals, and (3) activists, who are willing to join armed organizations. Albeit each of these levels usually tends to be considered in isolation from the others, the perspective of the political world allows us to develop a more coherent vision. First, the community as a whole can be considered as a set of poorly structured entities, whose provisional and contingent units are given by the presence of discursive practices which define the collective identity and establish the (immaterial) boundaries between friends and enemies (Laclau,

Terrorism as “A Political World”: Identity, Strategy, Values

151

2005; Mouffe, 2005; Palano, 2012). This community, like any “imagined community,” does not really exist but is only the product of a “cultural” construction. For this reason, the boundaries of the community (immaterial and symbolic boundaries) are unstable over time: they may change and even completely dissolve. Reference to the legitimacy of violence is especially important in the definition of identity. In other words, the definition of the boundaries is achieved mainly through the legitimization of violence against the “enemies”: individuals who are regarded as threats to the “imagined community.” In this process, a decisive role is played by rituals, symbols, and ideologies, which tend to “dehumanize” the enemy in order to legitimize the use of violence and therefore to radicalize the clash. Of course, the armed organizations contribute to radicalization, but, to a large degree, they use an ideological and symbolic repertoire which is built up and consolidated within the political world. Second, the action of an armed organization and its strategic logic can be designed as components of a political world, but we must consider the community and the organization as two autonomous entities: in other words, the armed organization acts like an artistic movement within the art world, as it attempts to change the discursive practices in at least three directions. First, similarly to those movements that shift the boundaries of artistic expression, terrorist organizations seek to modify (through action and ideological production) the boundaries that define the identity of the “imagined community”: in other words, they move the dividing line between friend and enemy (the subject against whom it is legitimate to use the most extreme form of violence). In addition, like the artistic avantgarde, they change the repertoires of legitimate action and, at the same time, try to get new forms of violent action recognized as legitimate forms of political struggle. Finally, they try to gain political recognition and leadership within the political world, often with the aim of ensuring the stable recruitment of new members. Just as the relationship between avantgarde and art worlds is often problematic, so the relationship between armed organizations and community support is always precariously balanced. The use of violence (and the forms in which it is exercised) is increasingly at the center of a complex process of legitimization and delegitimization
usually about organizations’ control. Third, the action of individual armed activists can also be positioned differently, and the problem of an “altruistic” action can be considered in a new light. From this perspective, terrorist actions cannot be explained as actions dictated by values (political and religious), or as conduct directed at material gain, but as the result of the search for stable recognition by

152

DAMIANO PALANO

the community of support. In this way, the behavior of a member in a terrorist organization can be considered as the result of a specific identification, which is possible only within a group which allows stable recognition. As argued by Pizzorno (1993), it is indeed very difficult to support the idea that participation in a political movement is only a behavior by which an individual seeks to obtain material advantages. Incentives related to the purpose of the movement (purposive goals) are often ineffective because of the option of free-riding. The incentive of solidarity, which relates to social recognition, is much more relevant to participation. This incentive is closely linked to membership in a group: first, it acts on the need for an individual to enter into solidarity relations to facilitate the recognition of their identity; in addition, it affects “the need to meet the expectations that an individual perceives as related to its action” (Pizzorno, 1993, p. 140). In this case, individual behavior cannot be entirely watered down to a rationalistic logic, but it is not contradictory. From this perspective, individuals change the elements of their same self through participation in a collective movement, and this experience changes the criteria of selection and calculations. Through participation in organizations and movements, individuals change their own interests, and, in particular, come to a different assessment of the costs and benefits of certain actions. In other words, what is perceived by others as a cost can be defined by the activist of an underground organization as a prize, because it reinforces their social identity. For example, spending time and energy on the development of the organization can be perceived by activists not as a cost but as a benefit, because these activities increase solidarity with the group. Obviously, participation affects identity, as do the expectations the individual has of social recognition and the rules that the individual feels bind her/him. In particular, the interests pursued by a subject cannot be distinguished from participation in a specific group in the sense that the expectations of social recognition have meaning only within that specific social system in which the individual participates, and cannot be satisfied outside the group. In this respect, therefore, the problem of which rationality terrorist behavior reflects finds an answer in belonging to an “identifying community,” that is, a political world. In general, the incentives that explain participation in an armed organization are no different from those that explain participation in other social movement organizations. More specifically, however, the former has characteristics that are partly owed to the clandestine conditions of the organization (the dislocation from everyday life and the rupture of relations with the outside world). The most significant point, though, is that in this case the idea of the political world helps

Terrorism as “A Political World”: Identity, Strategy, Values

153

to critically review a crystallized image of “values” and invites us to consider values, norms, beliefs, and symbols within a context of social relations, in which given actions warrant a specific meaning and recognition. From this perspective, just as an art world can exist if there are relatively institutionalized mechanisms for the recognition of reputation, so the political world within which the armed organizations act can exist only if there are mechanisms that ensure recognition of those who carry out an activity that does not produce material benefits.

CONCLUDING REMARKS In this chapter I outlined the framework for a cultural analysis of terrorism. In particular, I argued that the logic of terrorism can be fully understood only as a reflection of the “cultural” construction of collective and individual identities. For this reason, I proposed the concept of the political world to represent the set of symbolic production that characterizes terrorism. In particular, the interactions which take place in this world can be seen as symbolic interactions in which, at different levels, identities are at stake: first, at a level of community support, the dispute is in fact over the identity of the community (and thus on the legitimacy of violence); second, at the level of the armed organization, interaction concerns the identity of the organization, its role within the community, and the meaning of its actions; finally, at the individual level, interaction concerns the identity of individual activists, who get their recognition of self and carry out actions that confirm their identity over time. At each level the definition of identities is therefore the result of a complex interaction process: a process that involves, on the one hand, a dimension of conflict, but that, on the other hand, always requires a component of recognition.

REFERENCES Abrahms, M. (2006). What terrorists really want. International Security, 32(4), 78
105. Adorno, T. W., Frenkel-Brunswik, E., Levinson, D., & Sanford, N. (1950). The authoritarian personality. New York, NY: Harper & Row. Anderson, B. (1983). Imagined communities: The origin and spread of nationalism. London: Verso. Becker, H. S. (1974). Art as a collective action. American Sociological Review, 39(6), 767
776. Becker, H. S. (1976). Art worlds and social types. American Behavioral Scientist, 19(6), 703
718.

154

DAMIANO PALANO

Becker, H. S. (1982). Art worlds. Berkeley, CA: University of California Press. Bloom, M. (2005). Dying to kill: The allure of suicide terrorism. New York, NY: Columbia University Press. Borum, R. (2004). Psychology of terrorism. Tampa, FL: University of South Florida. Boudon, R. (2003). Beyond rational choice theory. Annual Review of Sociology, 29, 1
21. Boudon, R. (2009). Rational choice theory. In B. S. Turner (Ed.), The new Blackwell companion to social theory (pp. 179
195). Oxford: Blackwell. Braungart, R. G., & Braungart, M. M. (1992). From protest to terrorism. The case of SDS and the Weathermen. In D. Della Porta (Ed.), Social movements and violence: Participation in underground organizations (pp. 45
78). Greenwich, CT: Jai Press. Canter, D. (2009). The multi-faceted nature of terrorism: An introduction. In D. Canter (Ed.), The faces of terrorism: Multidisciplinary perspectives (pp. 1
17). London: Wiley-Blackwell. Caplan, B. (2006). Terrorism: The relevance of the rational choice model. Public Choice, 128(1), 91
107. Caruso, R., & Locatelli, A. (2008). Deadly contests: An economic note on al-Qaeda’s reward system. The Economics of Peace and Security Journal, 3(2), 62
68. Caruso, R., & Gavrilova, E. (2012). Youth unemployment and political violence: Evidence from Israeli/Palestinian conflict (1997
2007). Peace Economics, Peace Science and Public Policy, 18(2), Article 2. Caruso, R., & Schneider, F. (2011). The socio-economic determinants of terrorism and political violence in Western Europe (1994
2007). European Journal of Political Economy, 27(S1), s37
s49. Caruso, R., & Schneider, F. (2013). Brutality of jihadist terrorism. A contest theory perspective and empirical evidence in the period 2002
2010. Journal of Policy Modeling, 35(5), 685
696. Retrieved from http://www.sciencedirect.com/science/article/pii/S0161893813000124 Cherkaoui, M. (2007). Good intentions: Max Weber and the paradox of unintended consequences. Oxford: Bardwell Press. Crenshaw, M. (1981). The causes of terrorism. Comparative Politics, 13(4), 379
299. Crenshaw, M. (1986). The psychology of political terrorism. In M. G. Hermann (Ed.), Political psychology (pp. 379
413). San Francisco, CA: Jossey-Bass. Crenshaw, M. (1998). The logic of terrorism: Terrorist behavior as a product of strategic choice. In W. Reich (Ed.), Origins of terrorism: Psychologies, ideologies, theologies, states of mind (pp. 7
24). Cambridge: Cambridge University Press. Crenshaw, M. (2000). The psychology of terrorism: An agenda for the 21st century. Political Psychology, 21(2), 405
420. Crenshaw, M. (2006). Have motivations for terrorism changed? In J. Victoroff (Ed.), Tangled roots: Social and psychological factors in the genesis of terrorism (pp. 51
57). Amsterdam: Ios Press. Danto, A. (1973). Artworks and real things. Theoria, 39, 1
17. Danto, A. (2003 [1964]). The artworld. In C. Korsmeijer (Ed.), Aesthetics: The big questions (pp. 33
43). Oxford: Blackwell. Della Porta, D. (1990). Il terrorismo di sinistra. Bologna: Il Mulino. Della Porta, D. (1995). Social movements, political violence and the state: A comparative analysis of Italy and Germany. Cambridge: Cambridge University Press. Della Porta, D., & Diani, M. (2006). Social movements. Oxford: Blackwell. Dickie, G. (1964). The myth of the aesthetic attitude. American Philosophical Quarterly, 1(1), 56
65.

Terrorism as “A Political World”: Identity, Strategy, Values

155

Dickie, G. (1974). Art and aesthetics: An institutional analysis. Ithaca, NY: Cornell University Press. Dickie, G. (1984). The art circle: A theory of art. New York, NY: Haven. Di Maggio, P. J., & Hirsch, P. M. (1976). Production organizations in the arts. American Behavioral Scientist, 19(6), 735
752. Elster, J. (2005). Motivations and beliefs in suicide missions. In D. Gambetta (Ed.), Making sense of suicide missions (pp. 233
258). Oxford: Oxford University Press. Elster, J. (2009). Le de´sinte´ressemnt. Paris: Seuil. Evangelista, M. (2008). Law, ethics, and the war on terror. Cambridge: Polity Press. Ferracuti, F. (1982). A sociopsychiatric interpretation of terrorism. Annals of the American Academy, 463, 129
140. Ferracuti, F., & Bruno, F. (1981). Psychiatric aspects of terrorism in Italy. In I. L. BarakGlantz & C. R. Huff (Eds.), The mad, the bad and the different: Essays in honor of Simon Dinitz (pp. 199
213). Lexington, MA: Heath. Gallie, W. B. (1956). Essentially contested concepts. Proceedings of the Aristotelian Society, 56(1), 167
198. Gambetta, D. (1987). Were they pushed or did they jump? Cambridge: Cambridge University Press. Gambetta, D. (2005). Can we make sense of suicide missions. In D. Gambetta (Ed.), Making sense of suicide missions (pp. 259
299). Oxford: Oxford University Press. Gill, P. (2007). A multi-dimensional approach to suicide bombing. International Journal of Conflict and Violence, 1(2), 142
159. Griswold, W. (1994). Culture and society in a changing world. Thousand Oaks, CA: Pine Forge Press. Hafez, M. (2007). Suicide bombers in Iraq. Washington, DC: United States Institute of Peace Press. Hirsch, P. M. (1972). Processing fads and fashions: An organization-set analysis of cultural industry systems. American Journal of Sociology, 77(4), 639
655. Hobsbawm, E., & Ranger, T. (1987). The invention of tradition. Cambridge: Cambridge University Press. Horgan, J. (2005). Psychology of terrorism. London: Routledge. Horowitz, M. (2010). Nonstate actors and the diffusion of innovations: The case of suicide terrorism. International Organization, 64(1), 33
64. Ignatieff, M. (2004). The lesser evil: Political ethics in an age of terror. Princeton, NJ: Princeton University Press. Kihmi, S., & Even, S. (2004). Who are Palestinian suicide bombers? Terrorism and Political Violence, 16(4), 815
840. Kruglanski, A. W., Chen, X., Dechesne, M., Fishman, S., & Orehek, E. (2009). Fully committed: Suicide bombers’ motivations and the quest of personal significance. Political Psychology, 30(3), 331
357. Lacassagne, A. (1894). L’assassinat du Pre´sident Carnot. Lyon-Paris: Storck-Masson. Laclau, E. (2005). On populist reason. London: Verso. Laitin, D. D. (1986). Hegemony and culture: Politics as religious change among the Yoruba. Chicago, IL: University of Chicago Press. Laquer, W. (1977). Interpretation of terrorism: Fact, fiction and political science. Journal of Contemporary History, 12(1), 1
42. Laquer, W. (1987). The age of terrorism. Boston, MA: Little, Brown.

156

DAMIANO PALANO

Laquer, W. (1999). The new terrorism: Fanaticism and the arms of destruction. New York, NY: Oxford University Press. Lombroso, C. (1894). Gli anarchici. Torino: Bocca. Maanen, H. V. (2009). How to study art worlds: On the societal functioning of aesthetic values. Amsterdam: Amsterdam University Press. Merari, A. (2010). Driven to death: Psychological and social aspects of suicide terrorism. Oxford: Oxford University. Moghadam, A. (2006). The roots of terrorism. London: Chelsea House. Mouffe, C. (2005). On the political. London: Routledge. Palano, D. (2012). La democrazia e il nemico. Saggi per una teoria realistica. Milano: Mimesis. Pape, R. (2003). The strategic logic of suicide terrorism. American Political Science Review, 97(3), 343
361. Pape, R. (2005). Dying to win: The strategic logic of suicide terrorism. New York, NY: Random House. Pearlstein, R. M. (1991). The mind of the political terrorist. Wilmington, DE: SR Books. Pedahzur, A. (2005). Suicide terrorism. London: Polity Press. Pizzorno, A. (1986). Some other kinds of otherness: A critique of “rational choice” theories. In A. Foxley, M. S. McPherson, & G. O’Donnel (Eds.), Development, democracy and art of trespassing: Essays in honor of Albert O. Hirschman (pp. 355
373). Notre Dame, IL: University of Notre Dame. Pizzorno, A. (1990). On rationality and democratic choice. In P. Birnbaum & J. Leca (Eds.), Individualism: Theory and methods (pp. 295
331). Oxford: Clarendon Press. Pizzorno, A. (1993). Le radici della politica assoluta e altri saggi. Milano: Feltrinelli. Post, J. M. (1986a). Narcissism and the charismatic leader-follower relationship. Political Psychology, 7(5), 675
688. Post, J. M. (1986b). ‘Hostilite´’, ‘Conformite´’, ‘Fraternite´’: The group dynamics of terrorist behavior. International Journal of Group Psychotherapy, 36(2), 211
224. Post, J. M. (1998). Terrorist Psycho-logic: Terrorist behavior as a product of psychological forces. In W. Reich (Ed.), Origins of terrorism. Psychologies, ideologies, theologies, states of mind (pp. 25
40). Cambridge: Cambridge University Press. Powell, W. W., & Di Maggio, P. J. (Eds.). (1991). The new institutionalism in organizational analysis. Chicago, IL: Chicago University Press. Ricolfi, L. (2005). Palestinians, 1981
2003. In D. Gambetta (Ed.), Making sense of suicide missions (pp. 77
129). Oxford: Oxford University Press. Ross, J. J. (1994). The psychological causes of oppositional political terrorism: Toward an integration of findings. International Journal of Group Tensions, 24, 157
185. Sageman, M. (2004). Understanding terror networks. Philadelphia, PA: University of Pennsylvania Press. Sandler, T., & Lapan, H. E. (1988). The calculus of dissent: An analysis of terrorists’ choice of targets. Synthe´se, 76(2), 245
261. Sandler, T., Tschirhart, J. T., & Cauley, J. (1983). A theoretical analysis of transnational terrorism. American Political Science Review, 77(4), 36
54. Santoro, M. (1998). Mafia, cultura e politica. Rassegna Italiana di Sociologia, 39(4), 441
476. Santoro, M. (2007). La voce del padrino. Mafia, cultura, politica. Verona: Ombre corte. Schmitt, C. (1996 [1927]). The concept of the political. Chicago, IL: University of Chicago Press. Sewell, W., Jr. (1992). A theory of structure. Duality, agency and transformation. American Journal of Sociology, 98(1), 1
29.

Terrorism as “A Political World”: Identity, Strategy, Values

157

Sighele, S. (1897). La Delinquenza Settaria: Appunti di sociologia. Milano: Treves. Sprinzak, E. (2000). Rational fanatics. Foreign Policy, 120, 66
73. Stern, J. (2003). Terror in the name of God: Why religious militants kill. New York, NY: Harper Collins. Swidler, A. (1986). Culture in action: Symbols and strategies. American Sociological Review, 51(2), 273
286. Tarrow, S. (1989). Democracy and disorder. Oxford: Oxford University Press. Tarrow, S. (1994). Power in movement: Collective action, social movements and politics. Cambridge: Cambridge University Press. Tosini, D. (2009). A sociological understanding of suicide attacks. Theory, Culture & Society, 26(4), 67
96. Tosini, D. (2010). Calculated, passionate, pious extremism: Beyond a rational choice theory of suicide terrorism. Asian Journal of Social Science, 38(3), 392
413. Tosini, D. (2011). Agents and mechanisms of terrorist campaigns: A contribution to a general theory of rationality. Revue europe´enne des sciences sociales, 49(2), 43
70. Victoroff, J. (2005). The mind of the terrorist: A review and critique of psychological approaches. Journal of Conflict Resolution, 49(3), 3
42. Weber, M. (1966). The theory of social and economic organization. New York, NY: The Free Press. Wilson, M. A. (2000). Toward a model of terrorist behavior in hostage-taking incidents. Journal of Conflict Resolution, 44(4), 403
424. Wintrobe, R. (2006a). Rational extremism. Cambridge: Cambridge University Press. Wintrobe, R. (2006b). Extremism, suicide terror, and authoritarianism. Public Choice, 128(1), 169
195. Zald, M. N., & McCarthy, J. D. (1980). Social movement industries: Competition and cooperation among movement organizations. In L. Kriesberg (Ed.), Research in social movements, conflict and change (Vol. 3, pp. 1
20). Greenwich, CT: JAI Press.

CHAPTER 8 THE ECONOMIC ANALYSIS OF LONE WOLF TERRORISM Peter Phillips ABSTRACT Purpose
This chapter explains how economic analysis can contribute to the delineation of the lone wolf’s opportunities and choices in a manner that allows operationally relevant advice to be contributed to the investigative process. Approach
Using a risk-reward analytical framework we examine the lone wolf’s attack method opportunities and choices and identify those attack methods that would be chosen by lone wolves with different levels of risk aversion. We also use prospect theory as an alternative methodology for the determination of the lone wolf’s preference orderings over the available attack methods in a context where he references his actions against those of a predecessor whom he wishes to emulate. Findings
We find that lone wolf terrorists with different levels of risk aversion can be expected to choose different attack methods or combinations of attack methods. More risk averse lone wolf terrorists will choose attack methods such as assassination. Less risk averse lone wolf terrorists will choose attack methods such as bombing, hostage-taking and unconventional attacks. Also, we find that lone wolf terrorists who

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 159
191 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022013

159

160

PETER PHILLIPS

reference their actions against ‘predecessor’ lone wolf terrorists will choose differently from among the available attack methods depending on which predecessor lone wolf is being referenced. Limitations
The analysis provides two different perspectives on terrorist choice but by no means exhausts the analytical alternatives. The analysis focuses on the fatalities and injuries inflicted whereas other perspectives might include different ‘payoffs’ series, including news or media coverage. Originality
The chapter contributes an analysis of the order in which lone wolf terrorists with particular characteristics will choose from a set of available attack methods. During the course of our discussion we point out the consistency between the ‘rise’ of the lone wolf terrorist and the diseconomies to scale that are evident within the terrorism context. This presents the opportunity for new debates. Keywords: Lone wolf terrorism; Opportunities; Choices; Attack methods; Economic Analysis; Prospect theory JEL classifications: H56; D74; D81

INTRODUCTION The purpose of this chapter is to explore some of the ways in which economics may be helpful in the pre-emption, pursuit and apprehension of the lone wolf terrorist. The ‘true’ or ‘pure’ lone wolf terrorist operates alone, outside of any command structure or formal organisation. He presents a challenge to law enforcement because his independence makes it difficult to detect evidence of his existence before he strikes. Lone wolf terrorists do not have a network that can be infiltrated by government security agencies. Even the ‘quasi’ lone wolf terrorist who engages in some level of interaction with known potential terrorists or maintains some links with extremist organisations will be more difficult to identify, track and apprehend prior to an attack than those who choose to operate within a formal command structure and information and supply network. We must not fall into the trap of thinking that a single individual cannot obtain the information and materials required to inflict human tragedy upon civilians. Most lone wolf attacks have been relatively crude in terms of the materials required but

The Economic Analysis of Lone Wolf Terrorism

161

have sometimes demonstrated sophistication in planning that rivals or exceeds that of known terrorist groups. Some lone terrorists have inflicted more injuries and fatalities than many terrorist groups. For these reasons, the lone wolf terrorist has figured prominently in the Federal Bureau of Investigation’s (FBI) strategic planning for almost a decade. The then FBI’s (acting) general counsel, Patrick Rowan, highlighted the threat of lone wolf terrorism in testimony before the House Intelligence Committee in 2003 (Webster et al., 2012, p. 9). In 2011, both the president of the United States, Barack Obama, and the U.S. Homeland security director, Janet Napolitano, named lone wolf terrorism as the greatest threat to be guarded against. According to the Associated Press, Janet Napolitano said, ‘The risk of “lone wolf” attackers, with no ties to known extremist networks or grand conspiracies, is on the rise as the global terrorist threat has shifted’. She was quoted as saying, ‘There’s been a lot of evolution over the past three years. The thing that’s most noticeable to me is the growth of the lone wolf’ (Associated Press, 2011). The recent attention that lone wolf terrorism has received from governments and their security agencies should not lead us to overlook the long history of lone wolf terrorism in both the United States and Europe. Indeed, if lone wolf terrorism is a ‘rising threat’, then there is certainly an argument to be made that the threat has been rising for a long time. The narrative of lone wolf terrorism stretches back at least several decades. In the United States between 1978 and 1999, 26 per cent of the victims of terrorism were victims of lone wolf terrorism and the number of recorded incidences of lone wolf terrorism rose from just 2 during the 1960s to 13 in the 1990s (Spaaij, 2010, pp. 859
860). Thirty cases of lone wolf terrorism were reported in the United States between 1968 and 2007. Over the same period nine cases were reported in Germany, seven in France, six in Spain and five in Italy.1 The actions of Anders Behring Breivik in July 2011, which resulted in the deaths of 77 people, became the most deadly incidence of lone wolf terrorism, overshadowing the 18 fatalities attributed to the actions of Joseph Paul Franklin in the United States in the 1970s. What makes the challenge facing law enforcement more acute is the absence of any particular ‘profile’ for lone wolf terrorists. Lone wolves often hold ‘extreme’ views but different lone wolves hold these extreme views about different things and many people with extreme views do not end up expressing them violently in an act of terrorism. A number of cases of lone wolf terrorism in the United States have apparently been motivated by extreme views on race and abortion. Racially motivated acts of terrorism have been among the most deadly instances of lone wolf terrorism.

162

PETER PHILLIPS

For example, Joseph Paul Franklin, a white supremacist who engaged in a series of racially motivated shootings between the late 1970s and early 1980s, killed 18 and injured 5 people and Mark Essex, a black militant, accumulated an identical number of victims
though with 10 fatalities and 13 injuries
in a one week spree of violence in 1973. Ten of Essex’s victims were police officers, whom he deliberately targeted in an attack on the New Orleans P.D. on New Year’s Eve 1972. One officer was killed and one injured in the initial attack. Others were killed or wounded in the ensuing pursuit along with several civilians. Neither the backgrounds nor motivations of lone wolf terrorists reveal any strongly discernible patterns. Indeed, it is hard to construct a representative ‘profile’ for lone wolf terrorists in general. Just about the only thing that lone wolves appear to have in common is that nearly all lone wolf terrorists have been male. Rachelle Shannon, an anti-abortion activist who shot and wounded a doctor in Wichita, Kansas, is the only female to engage in an act of lone wolf terrorism in the United States. Beyond this, the characteristics and motivations of lone wolf terrorists are divergent enough to prevent a profile for an archetype lone wolf from being constructed. This is essentially the same situation that has been found to characterise terrorists in general and far-reaching and comprehensive analysis of offenders from psychological and socio-economic-environmental perspectives have failed to identify any finite list of things that terrorists have in common (Rasch, 1979; Victoroff, 2005). This being said, work is ongoing in defining and refining the categories of lone wolf terrorism and it might still be possible that within particular sub-sets of motivations patterns will emerge that will help to develop a typology of the characteristic lone wolf offender. One of the most recent and most deadly acts of lone wolf terrorism in the United States occurred on 5 November 2009 when Nidal Malik Hasan, a U.S. Army Major, used two pistols to kill 13 people
12 U.S. soldiers and 1 Department of Defence employee
and injure 43 others in a shooting spree at the U.S. Army base in Fort Hood, Texas. The William H. Webster Commission, which was established to investigate the FBI’s responses to pieces of intelligence obtained prior to the shooting, has further highlighted the challenges that lone wolf offenders present to law enforcement agencies. On the face of it, events such as 9/11 and the Fort Hood shootings may lead us to the conclusion that ‘religious extremism’, particularly of the Islamic type, is a prominent category of modern lone wolf terrorism within which most potential lone wolf terrorists can be grouped and on which law enforcement efforts can be concentrated. However, the Commission’s report highlights the fact that Islamic

The Economic Analysis of Lone Wolf Terrorism

163

religious extremism has accounted for only 7 per cent of terrorist incidences in the United States. The two other overarching categories of extremism, ‘political’ and ‘social’, are much more significant. This, of course, highlights immediately the problem of classifying lone wolf terrorists. The task would be easy if most of them could be placed within the category of ‘religious extremists’ and further classified into the subset Islamic extremism. When political and social extremism are dominant overarching categories, each with diverse subsets of motivations, the task of deciding which types of behaviour will attract law enforcement attention is more complex. The purpose of the FBI’s model of the radicalisation process, which is consistent with similar models developed in Coolsaet (2011) and Ranstorp (2010), is to determine a set of ‘indicators’ that identify a potential lone wolf terrorist before he strikes. The model is relatively straightforward (Webster et al., 2012, p. 8): Pre-radicalisation → Identification → Indoctrination → Action This involves a dual challenge. First, the radicalisation of a particular ‘lone’ individual may not be observable prior to action. Second, even if there are some observable indicators of radicalisation, these may or may not be ‘actionable’ within the legal and constitutional frameworks that exist. As the Webster Commission’s report clearly states, radicalisation is not a crime. Evidence that the potential offender is planning a violent action is required before an individual can be taken into custody. One important reason to justify a heightened level of concern about the potential for lone wolf terrorism perpetrated by radicalised or extremist individuals is that the process of radicalisation may be becoming less visible, more multifaceted in its origins and faster in its evolution towards violent action. The last two stages of the radicalisation process are known to proceed with greater pace than the first two but growing prominence of the Internet as a means to obtain and distribute information at little or no cost and hidden amongst large volumes of other electronic information may contribute to both a general acceleration in and ‘cloaking’ of all four stages of radicalisation making it even more difficult to detect a lone wolf terrorist before he engages in a violent action (Webster et al., 2012, p. 10). Economic analysis can contribute several things to investigative processes designed to pre-empt and pursue the lone wolf terrorist. Fundamentally, economics is about ‘opportunities’ and ‘choices’. It is a set

164

PETER PHILLIPS

of analytical tools and theoretical frameworks that are used to identify the opportunities that individuals face
called the ‘opportunity set’
and the choices that they make from the available opportunities. Applied to the analysis of terrorism, economic analysis will focus on determining the opportunities and choices that terrorists face. Because violent actions are the most pernicious actions of ‘radicalised’ or ‘extremist’ individuals and because these actions are those which law enforcement efforts are most concerned with pre-empting, it is useful to consider those opportunities and choices that are relevant to violent action and its consequences. In doing so, it becomes possible to align ‘rational’ choice with ‘most dangerous’ choice. If the terrorist makes a mistake, his actual choice will be less dangerous than that which we have prepared to face. Traditionally, economic analysis has tended to avoid investigating the underlying motivations of terroristic individuals, leaving that side of things to the psychologists (see Chapter 7 within this volume). Given what we have said about the difficulties experienced in the attempts to identify a set of discernible characteristics for ‘the terrorist’ this might have been quite wise. However, it is important to note that the boundary between economics and psychology is becoming less and less demarcated as time goes by. Over the past two decades, work in both disciplines has reflected work in the other. Akerlof and Kranton (2000) have shown that significant psychological concepts such as identity can be encompassed within traditional economic analysis. Our economic approach is flexible enough to be adjusted for important results generated within psychology and should it one day be determined that the lone wolf terrorist is characterised by a particular set of psychological-economic-sociological features, this would be a positive development that would expand the horizons of even the most orthodox economic analysis of terrorism. This being said, we shall explore some nonorthodox models of choice emanating from behavioural economics, especially prospect theory, and the ways in which it might be relevant to the investigative process that aims to pre-empt and pursue the lone wolf terrorist.

THE DISECONOMIES OF TERRORISM Let us begin with a discussion of a more overarching subject. That is, the place for lone wolf terrorism within the terrorism context. In fact, two

165

The Economic Analysis of Lone Wolf Terrorism

prominent features of the terrorism context are consistent with each other. These two features are, on the one hand, the existence of lone wolf terrorism or terrorism at the smallest possible scale of operation and, on the other hand, the absence of evidence for economies of scale as it pertains to terrorist groups. Just as we find strong evidence for economies of scale in the tendency for business organisation to become larger in scale with independent operators and small-sized firms superseded and replaced by larger corporations (see Lucas, 1978; Viner, 1932), we observe evidence for diseconomies of scale in the absence of this tendency within the terrorism context (Phillips, 2011b). The Global Terrorism Database (GTD) contains many thousands of data entries referring to terrorist groups that, with very few exceptions, never reached beyond a small number of members and supporters. During the period 2000
2008, there were 400 different active terrorist groups involved in some form of terrorism that inflicted injuries and fatalities. Over time many terrorist groups fade away and are replaced by others (Table 1). Within the terrorism context, there is a positive relationship between average fatalities inflicted by terrorist attacks and the variability of those fatalities over time. Variability reflects the risk to the terrorist that the actual outcome of a terrorist attack will be different, higher or lower, than that which was expected. Attacks that are expected to inflict more fatalities on average are also more risky or variable in their outcomes. In 2008, for example, there were 106 terrorist groups which inflicted one or more fatalities. The five groups with the highest average number of fatalities across Table 1.

Number of Terrorist Groups Involved in Acts of Violence Each Year.

Year

Number of Groups Involved in Acts of Violence

2000 2001 2002 2003 2004 2005 2006 2007 2008

103 26 95 26 67 111 99 99 165

Data Source: Global Terrorism Database (GTD).

166

Table 2.

PETER PHILLIPS

Average Monthly Fatalities and Risk: Terrorist Groups 2008.

Terrorist Group Justice and Equality Movement (JEM) Al-Qaeda Lord’s Resistance Army (LRA) Islamic State of Iraq (ISI) Deccan Mujahideen Caucasus Emirate Terai Army Madhesi People Rights Forum (MPRF) Ogaden National Liberation Front (ONLF) Basque Fatherland and Freedom (ETA)

Average Fatalities Per Month

Risk (Standard Deviation)

12 5.875 5.739 7.451 2.178 0.333 0.306 0.250 0.250 0.308

38.527 14.058 10.811 10.340 7.546 0.888 0.870 0.866 0.866 0.861

Data Source: Global Terrorism Database (GTD).

the 12 months of 2008 and five groups with a lower average number of fatalities across the 12 months of 2008 are listed in Table 2. The groups with the highest average also experienced the highest amount of variability in the outcomes of their terrorist activities. The monthly correlation between the monthly average fatalities inflicted by each group and the level of monthly standard deviation that characterised the outcomes of the group’s terrorist activities was 0.90 in 2008. This relationship is reflected in the aggregate data year after year (see appendix). The important thing about this relationship between variability or risk and the outcomes of terrorists’ violent actions is that it varies considerably from terrorist group to terrorist group. Some groups ‘dominate’ others by inflicting a higher average number of fatalities upon civilians than other groups while experiencing a lower amount of variability (Sharpe, 1966, p. 123). The outcomes of these groups’ attacks are more deadly and more certain. Comparing simple averages unadjusted for risk or variability overlooks this fact. A statistic that accounts for the amount of variability experienced by terrorist groups relative to the average fatalities inflicted by their attacks is the fatalities-to-variability ratio. This is the ratio of the average fatalities inflicted by a terrorist group to the variability of those fatalities across attacks over time. What we find when we analyse the fatalities-tovariability ratios for individual terrorist groups is that it is very difficult for terrorist groups to generate and maintain a relatively high fatalities-tovariability ratio over time and, in general, the fatalities-to-variability ratios of terrorist groups are quite low. For the period 2000
2008, the average

167

The Economic Analysis of Lone Wolf Terrorism

fatalities-to-variability ratio or risk adjusted capability to inflict fatalities through an act of terrorism of the ten groups with the highest ratio in each year was 0.7847 fatalities per unit of variability. The majority of the 400 or more terrorist groups that were active during this time recorded much lower fatalities-to-variability ratios. The ten groups with the highest ratios in 2008 are presented in Table 3. The lone wolf terrorist also must confront the trade-off between the average number of fatalities he can expect to inflict and the risk that the actual outcomes of his terrorist actions may be quite different from that which he expected. When terrorist operations are examined at the ‘group’ scale, it is found that terrorist groups find it difficult to generate and maintain a fatalities-to-variability ratio of greater than 1.00. Most groups are characterised by a ratio that is significantly lower than this and those groups that do manage to generate a higher ratio cannot do so persistently over time. At the smallest scale of terrorist operation at the level of the individual terrorist, the story is different. If the lone wolf terrorists that were active in the United States during the period 1960
2009 were treated as an ‘unconnected group’ the fatalities-to-variability ratio recorded by the group for the entire period is 0.7510, indicating that the aggregate of terrorist activity perpetrated by America’s lone wolf terrorists inflicted a higher average number of fatalities per unit of variability than most of the terrorist groups contained within the GTD. This is important because terrorist groups, perhaps having recognised the diseconomies that we have been talking about, encourage individuals who share a similar ideology to participate in lone wolf terrorism (MEMRI, 2012). Table 3.

Highest Fatalities-to-Variability Ratios, 2008, Monthly Data.

Terrorist Group Communist Party of India-Maoist (CPI-M) Al-Qaeda in Iraq New People’s Army (NPA) Liberation Tigers of Tamil Eelam (LTTE) Taliban Revolutionary Armed Forces of Colombia (FARC) Baloch Liberation Army (BLA) Islamic State of Iraq (ISI) Al-Shabaab

Fatalities-to-Variability Ratio 1.245 1.179 0.990 0.928 0.907 0.863 0.726 0.721 0.651

Data Source: Calculations made using the data contained in the Global Terrorism Database (GTD).

168

PETER PHILLIPS

The possibility that the fatalities-to-variability (or risk-reward trade-off) is more effectively managed at a smaller scale of terrorist operation is also reflected in the summary statistics for the attack method category, ‘armed attacks’ within the RAND-MIPT transnational terrorism database. Across the period 1968
2008, the average number of fatalities per ‘armed attack’ was 5.32 and variability (standard deviation) 15.00 fatalities per attack per year. The fatalities-to-variability ratio of ‘armed attacks’ was 0.3546. Many of the lone wolf terrorists who were active in the United States over the same period engaged acts of violence that would be categorised as ‘armed attacks’, especially shootings or shooting sprees. In ‘armed attacks’
with assassinations and bombings removed from the dataset
America’s lone wolf terrorists inflicted an average of 2.63 fatalities per armed attack with variability 2.87 and a fatalities-to-variability ratio of 0.9157, almost twoand-a-half times the ratio that characterises the transnational ‘armed attacks’ perpetrated predominantly by terrorist groups. Although the armed attacks of the lone wolf terrorists could be expected to inflict about half as many fatalities as an act of transnational (group) terrorism of the same category, the outcomes were characterised by about one-fifth the amount of variability across the different lone wolf terrorists’ actions. The armed attacks perpetrated by lone wolf terrorists were 50 per cent as deadly but 80 per cent more certain (less variable). On average, if a lone wolf terrorist is successful in engaging in an act of violence, the outcome is both more deadly and more certain than would be the case for many of the terrorist groups contained within the GTD. Both the size distribution of terrorist groups and the relative capability of the aggregate of lone wolf terrorists in the United States to inflict a level of fatality with less uncertainty than terrorist groups point towards the conclusion that the terrorism context is characterised by diseconomies of scale. If this is so and if the terrorism context reflects over time some tendency towards an optimal size distribution of terrorist operation, lone wolf terrorism may be expected to become a more prominent feature of the terrorism context. Because of the lone wolf’s common modus operandi of engaging in a ‘single spree of violence’ governments and security agencies may be advised to view the individual lone wolf as the subject of investigation whilst viewing the aggregate of single-attack (or spree) and serial-attack lone wolves not as a disconnected set of individuals and outcomes but as an unconnected group emerging from the nature of the economies that characterise the scale at which terrorism is most optimally undertaken.

The Economic Analysis of Lone Wolf Terrorism

169

THE OPPORTUNITIES AND CHOICES OF THE LONE WOLF TERRORIST Economics is especially well-suited to the analysis of autonomous individual choice. In fact, even when defence economists analyse terrorist groups they often make use of the ‘representative agent’ theoretical device which effectively treats the group as a single independent decision-maker. Even so, the lone wolf terrorist presents significant analytical challenges if we wish to push beyond a statement of the obvious
for example, that the lone wolf terrorist can be analysed within a rational choice framework where he is viewed as making his choices in a manner that maximises his expected utility
and obtain results that are relevant to policymakers and, more importantly, law enforcement and security agencies engaged in investigative processes that aim to pre-empt or pursue particular lone wolf terrorists. Because the sociological-economic-psychological characteristics of lone wolf terrorists (and terrorists in general) have yet to yield discernible and reliable patterns, abstracting from the standard constituents of a potential lone wolf ‘profile’, as economic analysis is predisposed to do, will not abstract from as much analytical detail as might be expected and, what is more, may open new perspectives onto lone wolf terrorist behaviour that allows us to extract more from our existing catalogue of sociological-economic-psychological characteristics than has been the case up till now. At the most fundamental level, the economic analysis of the lone wolf terrorist would proceed by writing down an expected utility function that encompasses the particular opportunities and choices that have been deemed pertinent to the analysis. This type of approach follows Becker (1968) and Ehrlich (1973) who were among the first economists to use expected utility theory to investigate crime and criminal behaviour. Within defence economics, Landes (1978) is one of the earliest examples of an application of this traditional theoretical framework to a type of terrorist behaviour. Landes (1978) analysed hijacking. Similar approaches have been taken to the analysis of other types of terrorist activity, especially those involving negotiations or bargaining (Gaibulloev & Sandler, 2009; Sandler & Scott, 1987; Sandler, Tschirhart, & Cauley, 1983). Landes describes the opportunities and choices of the hijacker with the expected utility function: EU = ð1 − Pa ÞUðWj Þ þ Pa Pc UðWi − SÞ þ Pa ð1 − Pc ÞUðWj − CÞ

ð1Þ

170

PETER PHILLIPS

Here the hijacker’s expected utility depends upon the probability of apprehension in country i, Pa , the probability (given apprehension) of being convicted and sentenced to prison, Pc , the hijacker’s wealth, Wi and Wj , in each of two countries, i and j, the monetary equivalent of the sentence, S, in country i and the monetary costs, C, associated with apprehension but no sentence (example, costs of defending oneself at trial). In Landes’ (1978) model, the potential hijacker’s opportunities are twofold: (1) hijack; (2) do not hijack. The potential hijacker’s choice will be the one that maximises his expected utility. If the government takes steps to increase the probability of apprehension and conviction it will decrease the expected utility of hijacking and we should see a decline in the number of hijackings undertaken. Landes’ (1978) empirical analysis revealed that such security initiatives as placing air marshals on flights, installing metal detectors at airports, arranging for hijackers to be extradited back to the United States and substantially increasing the prison sentences imposed on hijackers were significant in bringing about a rapid decline in the number of hijackings undertaken in the United States. On the face of it, the expected utility analysis appeared to be quite good at explaining the behaviour of potential hijackers. Given their particular objective, which was usually to obtain money or get to Cuba, hijackers had previously found that hijacking ranked higher (had a higher expected utility) than doing something else (encompassed by a broad ‘do not hijack’ in Landes’ model). Following several steps by the U.S. government, this ranking was reversed. All of this sort of analytical work must be careful not to fall prey to the same criticisms that have been directed at Becker’s work (Blaug, 1992, especially p. 223). If we can use expected utility theory to obtain the prediction that hijacking will decrease when air marshals are put on planes, metal detectors are put in airports and harsher prison sentences are put in place and then we test that theoretical prediction empirically and find that it was indeed the case that a decrease in hijacking followed these particular security measures, have we proven anything at all? Aren’t such a theory and its empirical verification somewhat banal? The line is a very fine one and Blaug (1992) points out that Becker might have crossed it in his later work, though not necessarily his work on crime and criminal behaviour. What prevents analysis like Landes’ (1978) from slipping into banality is the fact that the analysis explains why we see a decline in attempted hijackings and not simply an increase in the number of hijackers being apprehended. That is, the analysis explains the ‘weighing up’ that potential hijackers do and predicts changes to the results of that weighing up when they are

The Economic Analysis of Lone Wolf Terrorism

171

confronted with changed circumstances. When there are changes to the payoffs, potential hijackers reorder their preferences away from hijacking and do not even attempt it. It is not the case that they simply proceed as usual and get caught in greater numbers. It is the case that the opportunities and choices of the lone wolf terrorist could be cast in terms of the ‘full’ expected utility analysis of the type used by Landes (1978). The analysis of the lone wolf within such a theoretical framework presents two main obstacles. First, a full expected utility analysis requires payoff and probability estimates for each outcome under consideration. Even if we concentrated on a particular attack method or target type, this would be a difficult task. Second, the individuality of the lone wolf and the absence of a clearly discernible ‘archetypal’ profile mean that any attempts to develop a full expected utility analysis around particular motivations, attack methods and targets (to name but a few aspects of the terrorism context) must always verge on the boundary between pure formalism and useful analysis. Rather than a full expected utility analysis of a particular scenario, we may find it useful to concentrate our efforts on a broader set of opportunities and choices and the alternative to ‘full’ expected utility analysis called mean-variance utility analysis. One of the most important choices that the lone wolf makes is his choice of attack method. The attack method that he chooses plays a part in determining the outcomes of his terrorist actions. The most immediate outcome of terrorism is the fatalities and injuries that it inflicts on civilians. It is possible to define the opportunities and choices of the lone wolf terrorist with respect to attack methods and the fatalities and injuries expected to result from the deployment of those attack methods. This approach depicts the lone wolf as choosing attack methods on the basis of the fatalities and injuries that he expects to inflict and the possibility that the actual outcome diverges (higher or lower) from that which he expected. Terrorists, through word and deed, have indicated their intention to inflict fatalities and injuries through violent acts. For example, the man arrested for plotting a metro bombing in Washington had told contacts that he planned to kill as many people as possible (AFP, 2010). Although the infliction of fatalities and injuries might be a subsidiary goal that is correlated with some more final goal, such as press coverage, it is clear that the actions of terrorists are designed to inflict a level of human tragedy. Furthermore, this cannot be undertaken without bearing the risk that the actual amount of human tragedy that is inflicted is less than or greater than that which was expected. If we look at attack method choice through this analytical lens, we see a much more delineated set of opportunities from

172

PETER PHILLIPS

which the lone wolf can choose. There are not more than ten attack method categories.2 Each represents an opportunity to inflict human tragedy and each is attended by the risk that the actual outcomes will be different from the expected outcomes. The lone wolf weighs the different opportunities in terms of the expected outcomes and the risk that the actual outcomes will be different. The ‘full’ expected utility analysis is collapsed into an analytical framework of risk versus reward. Our rewards could be anything that conveys positive utility to a terrorist. This could be press coverage. We have chosen the immediate outcome of terrorism as the unit of analysis: fatalities and injuries. The expected utility function that describes the lone wolf’s opportunities and choices across all attack methods is one that contains just the two moments of the distribution (mean and standard deviation) of the fatalities expected to result from a single attack method or a combination of attack methods: U = f ðEF ; σ F Þ

ð2Þ

The lone wolf chooses his attack method (or combination of attack methods) on the basis of the fatalities that he expects to inflict and the variability of those outcomes. We must not think that the lone wolf will find it difficult to determine approximately what the possible outcomes of his actions will be. The press coverage of terrorist incidences is comprehensive and there is a significant amount of information about both successful and failed attacks embedded within such coverage that may calibrate expectations with regards to the possible outcomes of a particular type of attack. The utility function, Eq. (2), depicts the lone wolf as trading off expected fatalities against variability. If we applied this to Landes’ (1978) hijacking analysis, our expected utility analysis would be much simpler and we would not need probability estimates for each outcome. Instead of ordering his different opportunities on the basis of an expected utility computed over the probabilities of all possible outcomes and their associated utility, Eq. (2) says that the lone wolf will order his different opportunities (attack methods) on the basis of the expected fatalities and the variability of the possible outcomes. Both of these things are inferred from the ‘averages’ that the lone wolf might reasonably be expected to see reflected in the publicly available information regarding the outcomes of actual and attempted acts of terrorism. In turn, this allows our analysis to be based on historical data without the need for further inferences regarding the possible outcomes and their probabilities.

The Economic Analysis of Lone Wolf Terrorism

173

THE LONE WOLF TERRORIST’S ATTACK METHODS: OPPORTUNITIES AND CHOICES The RAND-MIPT transnational terrorism database lists several attack method categories that encompass acts of transnational terrorism perpetrated since the late 1960s. When trying to obtain data that may be used to analyse the opportunities and choices of the lone wolf, the transnational terrorism database emerges as the most appropriate starting point. The number of lone wolf terrorists is relatively small and their attacks relatively idiosyncratic (with the notable exception of the shootings and shooting sprees). The complete historical record of terrorist incidences such as that contained within the GTD includes many attacks that are not relevant to the analysis of the lone wolf’s opportunities and choices. In particular, attacks on civilians in a war-torn region such as Iraq or Afghanistan are not reflective of the outcomes that a lone wolf terrorist is likely to achieve in attacks undertaken in Western countries. So the RAND-MIPT database on transnational terrorism is the most extensive record of terrorist incidences that most accurately reflects the fatalities and injuries that a lone wolf terrorist could expect to inflict if he deploys an attack method that is encompassed within one of the attack method categories covered by the database. Each attack method category has an expected number of fatalities and injuries and each expected amount of human tragedy is accompanied by a variability that reflects the risk that the actual outcomes of such a type of attack may diverge from those which were expected. When the level of analysis is ‘attack method’ the same statistical structure is evident as that which manifests itself in the aggregate fatalities-tovariability ratios
a positive relationship between inflicted fatalities and variability. Just like the terrorist groups whose actions populate the historical record of terrorism, the lone wolf terrorist faces a trade-off between the number of fatalities he can expect to inflict with a particular attack method and the possibility that the actual outcome will be different from his expectation. The opportunities that the lone wolf has to inflict human tragedy are contained within Table 4. In essence, these opportunities are ‘pairs’ of fatalities and variability. If he chooses an attack method that is expected to inflict a higher amount of human tragedy, he must bear the greater risk that his attack method will inflict more or less human tragedy than he expected. If he wants a more certain outcome, he must choose an attack method with a lower expected amount of human tragedy. Something that is particularly important to consider, though, is the case where the lone

174

PETER PHILLIPS

Table 4. Attack Type Armed attacks Arson Assassination Hostage Bombing Hijacking Kidnapping Other Unconventional Unknown

Attack Type, Fatalities and Variability of Outcomes. Average Fatalities and Injuries Per Attack Per Year

Standard Deviation

5.32 0.72 1.54 11.46 5.88 3.91 0.46 1.14 7.48 1.11

15.00 1.85 0.71 38.18 6.09 10.82 0.35 3.02 42.94 4.04

Data Source: Calculations made using the data contained in the RAND-MIPT transnational terrorism database.

wolf terrorist can combine attack methods. This is difficult for the lone wolf terrorist to do, particularly at short horizons or, at the limit, simultaneously. A terrorist group can allocate its resources across different attack methods and deploy terrorist operatives to undertake different types of attack simultaneously. A lone wolf as a single individual cannot easily do this. However, Anders Breivik did manage to combine ‘bombing’ and ‘armed attack’, making use of a timed bombing device to enable him diversify his strategy by combining two different attack methods. When attack methods are able to be combined, a feature of the same statistical structure that is manifested in a positive fatalities-to-variability relationship emerges with a great deal of significance for the risk-reward trade-off. This feature of the statistical structure is the correlation between the fatalities generated by the different attack methods. Because the correlation between the outcomes of the attack methods is imperfect, the terrorist may inflict more fatalities per unit of risk than law enforcement would expect if they did not take into consideration the imperfect correlation and the lone wolf’s potential, though limited, ability to combine imperfectly correlated attack methods together in an overall terrorist strategy. Rather than a linear ‘list’ of attack methods, the lone wolf terrorist may choose from a non-linear or concave set of attack methods and combinations of attack methods. The fact that the opportunity set confronting terrorists is concave rather than linear was identified by Phillips (2009). The critically important feature of a concave set of opportunities is that the expected payoffs at each level of variability are higher than they would be if all the payoffs to the different attack methods were perfectly

The Economic Analysis of Lone Wolf Terrorism

175

positively correlated. There are gains, measured in terms of higher expected payoffs per unit of variability, that are obtainable through combining two or more attack methods together. When you compute the average number of fatalities and variability for every possible combination of the attack methods listed in Table 4 and plot those attack methods or combinations that have the highest expected payoff at each level of variability, the result is a concave set. An important question can now be answered with the assistance of economic analysis. When making his decision on the basis of expected fatalities and variability, in what order of preference will the lone wolf place the attack methods listed in Table 4? Which attack method (or combination of attack methods) is the lone wolf terrorist’s best-ranked choice from the available opportunities? The standard approach in economics is to determine the ‘optimal’ or ‘rational’ choice or the one that maximises expected utility. This has been criticised both within and without of the economics discipline but there is reason to expect this approach to be somewhat more useful to us within the context that we have been considering. In this context, a choice from the set of attack method opportunities whose risks and payoffs are presented in Fig. 1 will be the most dangerous choice. By making the fatalities that are expected to be inflicted by an act of terrorism the key variable in the terrorist’s objective function, we equate the terrorist’s most dangerous choice with the optimal one of economic analysis. If the terrorist errs, as well he might, he will inflict less fatalities per unit of risk than we were prepared for. This is much better than the inverse scenario where we expect the terrorist to choose sub-optimally and essentially rule out the worst possible outcomes. Any choice from the concave set of opportunities presented in Fig. 1 will be optimal in the sense that such a choice will have the highest expected number of fatalities at its particular level of risk or variability. Although the number of ‘points on the line’ in Fig. 1 is infinite, certain attack methods and combinations of attack methods emerge as dominant optimal choices over particular ranges of the opportunity set. The range which is inhabited by the lone wolf terrorist will depend on his preference for risk. If he is more averse to risk, he will inhabit a range in the south-western region of the opportunity set, whereas if he is less averse to risk or more risk seeking he will inhabit a range in the north-eastern region of the set. In each range, particular attack methods dominate the others (Fig. 2). The range that the lone wolf terrorist inhabits will depend on his risk preference. It is almost certainly the case that terrorists, including lone wolf terrorists, are risk averse rather than risk seeking. A terrorist who is averse

176

PETER PHILLIPS

Expected or Average Fatalities

9.00 8.00 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00 0.00

2.00

4.00

6.00

8.00 10.00 12.00 14.00 16.00

Standard Deviation or Risk of Attack Method Combinations

The Lone Wolf’s Opportunity Set (Based on RAND-MIPT Data).

Fig. 1.

Expected or Average Fatalities

9.00 8.00

Assassination and Bombing, and Armed Attacks

7.00 6.00

Bombing, Hostage-Taking and Uncoventional Attracks

5.00

Bombing and Hostage-Taking

4.00 3.00 2.00 Assassination and Bombing

1.00 0.00 0.00

2.00

4.00

6.00

8.00

10.00 12.00 14.00

16.00

Standard Deviation or Risk of Attack Method Combinations

Fig. 2.

Dominant Attack Methods at Different Risk Ranges.

to risk will allocate his resources to more risky terrorist attack methods only if he receives a commensurate increase in the payoffs that he can expect to yield on average. A risk seeking terrorist would allocate all of his resources to the single riskiest attack method. This type of behaviour is seldom, if ever, observed. If we take the historical record of lone wolf terrorism as our guide, we know that most lone wolf terrorists have chosen either

The Economic Analysis of Lone Wolf Terrorism

177

assassination, bombing and armed attacks as their attack methods. This indicates a medium-high level of risk aversion. Although this does not rule out the possibility of the emergence of a lone wolf characterised by a much lower level of risk aversion, it does give us an indication of the risk ranges that have usually been inhabited by lone wolves in the past. The lone wolf terrorist is most likely to be a relatively risk-averse individual. When he engages in violence, his optimal attack methods choices are assassination at the highest level of risk aversion and armed attacks and bombing at medium levels of risk aversion. His expected number of fatalities per attack is no more than 7.00 but the outcomes of his attacks exhibit a relatively small amount of variability. This highlights the threat of the lone wolf. The challenge facing law enforcement is to prevent the lone wolf from striking. Once his attack commences, it is unlikely that it will end without a level of harm being inflicted upon his victims.

PURSUING THE LONE WOLF TERRORIST The investigative process that is applied to lone wolf terrorism must address two challenges: (1) the pre-emption of the lone wolf terrorist, and (2) the pursuit of the lone wolf terrorist if he strikes and remains at large. Aspects of the economic analysis of the lone wolf terrorist may contribute to both stages of the investigative process. The pre-emption of the lone wolf is extremely challenging but it is in just such a context that even small contributions may be important. The ways in which economic analysis may be used to delineate the lone wolf’s opportunities and choices provide a new perspective to the problem. It is unlikely, for example, that without economic analysis to provide such an insight that law enforcement would be aware that the terrorist’s opportunity set is characterised by higher expected payoffs per unit of variability when attack methods can be combined, though they might have had an intuition to that effect based on cumulative experience in law enforcement and the observation of the outcomes of terrorist attacks. Similarly, economic theoretical frameworks may prove to be useful in that part of the investigative process that is designed to identify, pursue and apprehend a lone wolf terrorist who has struck and who remains at large. In this case, the type of attack method that the lone wolf has chosen will provide insights into the amount of variability or risk he is willing to bear and, in turn, the type of risk aversion that most likely characterises him.

178

PETER PHILLIPS

Risk preference is a fundamental and critically important part of the analysis of behaviour under conditions of risk and uncertainty. Analysis of terrorist behaviour that does not take risk preference into account will overlook many of the most important aspects of terrorist decision-making. For example, it is the conventional wisdom that concessions by the government to a terrorist group will make that terrorist group more risk seeking and more likely to engage in additional and more intense acts of violence. However, the matter is not so clear cut. Sandler et al. (1983), Atkinson, Sandler, and Tschirhart (1987) and Lapan and Sandler (1988) have shown that the ‘never negotiate’ position is not always optimal. Work on incentives structures in other parts of economics, especially Ross (2004) and attempts to bring this work within the purview of the defence economics literature (see Phillips & Pohl, 2013), have shown that an increase in government concessions does not unambiguously decrease terrorist risk aversion and, in fact, might even increase it under certain conditions. The outcome depends on both the type and level of risk aversion that characterises the terrorist or terrorist group. When the lone wolf’s choices are the subject of analysis, his level of risk aversion and the type of risk aversion that he exhibits are important considerations. He may be more or less risk averse. His level of risk aversion may be high or low. He may exhibit increasing, decreasing or constant relative risk aversion as the outcomes of his terrorist actions unfold and accumulate. His type of risk aversion may be increasing, decreasing and constant. We have spoken of the lone wolf ordering his preferences on the basis of expected utility. The traditional approach in orthodox economic science is to treat this preference ordering process formally by assuming that the individual behaves as if he ‘computes’ the expected utility of each of his opportunities. The individual is assumed to apply his expected utility function to this process. Economics makes use of a small number of expected utility functions: quadratic, logarithmic, exponential and power. Each has different mathematical properties that reflect different ways in which risky opportunities are weighed up and choices made. Within economic science, the von Neumann
Morgenstern (NM) (1944) axioms for rational choice are accorded a good deal of significance. Preference orderings across opportunities are viewed as ‘rational’ if they are consistent with the NM axioms. The mean-variance expected utility analysis introduced in the previous section yields a preference order that is approximate to a full expected utility analysis but will only be guaranteed to be consistent with the NM axioms if the payoffs are distributed normally or if expected utility is quadratic. The quadratic expected utility function has a place of prominence within

The Economic Analysis of Lone Wolf Terrorism

179

the development of mean-variance expected utility analysis. It should be noted, however, that a mean-variance preference ordering will approximate an ordering determined by a full expected utility analysis even when payoffs are not normally distributed and when utility is not quadratic. Much has been made of the properties of quadratic utility in some parts of the financial economics literature but it does not alter the fact that the method provides a computationally tractable preference ordering that will approximate a full expected utility analysis which will likely be much more burdensome to compute. In any case, the quadratic utility function approximates a broader class of functions (Elton, Gruber, Brown, & Goetzmann, 2003, p. 220). If the lone wolf terrorist’s preference ordering process can be assumed to be exactly or approximately described by a quadratic expected utility function, it follows that the lone wolf may be expected to be characterised by a particular type of risk aversion. The quadratic expected utility function may be written as: UðRÞ = ð1 þ b0 ÞR þ b1 R2

ð3Þ

The utility function simply relates utility or satisfaction with whatever contributes to it, either positively or negatively. In Eq. (3), the terrorist’s expected utility depends upon the payoffs or rewards to terrorism, R. Since R is a ‘good’, increases in R increase expected utility. However, variability or risk is ‘bad’ and increases in risk decrease expected utility (i.e. b1 < 0). The quadratic utility function has several interesting properties. First, the function exhibits increasing risk aversion. As the terrorist’s actions accumulate fatalities over time, he allocates less resources (including time) to terrorism. Second, and more interesting, the terrorist can become satiated within a relatively narrow range of payoffs. Satiation means that a point is reached where more of something is not preferred. In this context, past some point more fatalities are no longer preferred to less and the terrorist may drift away from terrorist activity for some period of time. For most other utility functions, more is always preferred to less and the fact that the quadratic function allows for the opposite case has been the main point of criticism directed towards it and mean-variance analysis. Of course, in the context of terrorism it would be unwise to accept uncritically a conclusion that emerged from shortcomings attending the application of the quadratic utility function to a context that was solely concerned with monetary gain (investor behaviour) and where satiation was less easy to understand. Phillips (2011a) and Phillips and Pohl (2012) discuss the case of Theodore Kaczynski (the Unabomber) with reference to mean-variance

180

PETER PHILLIPS

utility, the quadratic expected utility function and the tendency of the terrorist to come and go from the terrorism context as fatalities accumulate over time. Kaczynski is a prominent example of what may be called ‘serial’ lone wolf terrorism, which may be contrasted with ‘spree’ lone wolf terrorism. Unlike the ‘spree’ lone wolf who engages in a time-concentrated act of violence such as a shooting spree, the serial lone wolf is more akin to the serial killer in the sense that he engages and withdraws from terrorism and may continue his attacks for extended periods if he is not apprehended. Kaczynski engaged in a series of bombings over two decades. Between 1978 and 1995, Kaczynski mailed sixteen parcel and letter-bombs to various targets around the United States. In May 1978, Kaczynski apparently left a parcel in the parking lot of the University of Illinois at Chicago. On the parcel, Kaczynski had written as the return address the address of a materials engineering professor who worked at Northwestern University at Evanston, Illinois. The parcel was found and sent by mail to the return address. When the parcel was received campus police were alerted. When a campus police officer attempted to open the parcel it exploded causing minor injuries. In each of the years 1978
1982, Kaczynski engaged in terrorism at least once. Following two attacks in 1982 that caused severe injuries to a secretary and a professor at Vanderbilt University (in Tennessee) and University of California (Berkeley), Kaczynski faded away and did not engage in violent terrorism for almost three years. His most deadly attacks occurred after his re-emergence in 1985. In May 1985, a graduate student at Berkeley was severely injured when he opened a parcel containing a bomb. In November, Kaczynski mailed another device to the home of biologist James McConnell in Michigan. Both McConnell and his research assistant were injured when the package was opened. The very next month, Kaczynski left a bomb in the parking lot of a computer store in Sacramento, California. The store’s owner was killed. Two months later another computer store owner was severely injured by a device left in the parking lot of his store in Salt Lake City, Utah. This was Kaczynski’s last attack for six years. The Unabomber re-emerged again in 1993. Over the next three years, Kaczynski would attack four more times. On 22 June and 24 June 1993, a geneticist at the University of California and a professor of computer science at Yale in New Haven, Connecticut were injured by parcels sent by Kaczynski. Almost 18 months later, in December 1994, an advertising executive became Kaczynski’s second fatality when he opened a parcel bomb that was delivered to his home. In April 1995, a timber industry lobbyist was killed when he opened a parcel addressed to the president of the California Forestry Association. In the same month, Kaczynski contacted

The Economic Analysis of Lone Wolf Terrorism

181

The New York Times. In the letter that he sent to the newspaper, Kaczynski stated that he would cease his violent terrorist activities if the paper published his manifesto. The Unabomber Manifesto, Industrial Society and Its Future, was published by The New York Times and The Washington Post in 1995. Kaczynski’s brother recognised the ideas contained in the manifesto as belonging to or being similar to those held by his brother. The Unabomber was arrested in April 1996 at his remote cabin in Montana. If we were asked, before the Unabomber was identified, to list some points relevant to the unknown suspect’s profile that emerge from the economic analysis of lone wolf terrorism (with risk, reward and quadratic utility) we would be able to say that the Unabomber’s choice of attack method
parcel bombings
is more akin to assassination than bombing in terms of its ‘targetedness’. This would place the suspect in the higher range of risk aversion and lead us to believe that he would continue to involve himself in attacks where the outcomes did not exhibit a great deal of variability. Because of the need to find low-reward-low-risk targets at which to deploy his attack methods, the suspect will need to drift from place to place and he is unlikely to be found in the vicinity of his attacks. Of course, parcel bombing is suited to such a modus operandi. As fatalities accumulate, he is likely, because of his increasing risk aversion, to allocate fewer resources to terrorism and more resources to other activities or forms of political expression. Because of his quadratic utility, he is likely to be easily satiated at intervals, especially after a series of successful attacks (or a single very successful attack). At such times, he may discontinue his involvement in terrorism, allowing ‘time diversification’ between terrorism and other activities to regain a semblance of balance. The ‘engage and withdraw’ type of serial lone wolf terrorism exemplified by Kaczynski is reflected in the types of choice behaviour embedded within our economic analysis. This, however, is never going to be a case of perfect match and economic analysis, like investigative psychology and offender profiling, will require a mixture of pragmatism and law enforcement experience in order to be effective at the operational level.

THE ‘COPYCAT’ LONE WOLF TERRORIST ACT The mean-variance or risk-reward approach to analysing the opportunities and choices of the lone wolf terrorist is, because of its computable results regarding opportunities and preference orderings, more capable of yielding

182

PETER PHILLIPS

operationally relevant advice for law enforcement and security agencies than the more computationally burdensome full expected utility analysis. However, the mean-variance approach is still more or less orthodox from the viewpoint of those critics of economic theory and rational choice. Over the past 40 years there have been significant advances in the analysis of choice from a behavioural or psychological perspective (DellaVigna, 2009). Of special note is the work of Kahneman and Tversky (1979, 1992). Kaheman and Tversky’s ‘prospect theory’ is an alternative to expected utility analysis in which preference orderings over ‘lotteries’ are determined by the application of a more complicated probability weighting scheme that accounts for many of the violations from the axioms of expected utility theory that have been observed in experimental economics laboratories. Kahneman and Tversky (1979, p. 265) discuss a series of examples of violations of the axioms of expected utility theory in hypothetical choice scenarios. All of Kahneman and Tversky’s (1979, pp. 265
271) results indicate that people do not weight the utilities of outcomes by their probabilities and therefore violate one of the principles of expected utility theory. Kahneman and Tversky’s (1979) prospect theory is perhaps the most prominent of several models of choice that propose an alternative to the probability weighting scheme of orthodox expected utility theory. Other weighting schemes are discussed by Machina (1987). In attempting to develop an alternative model, Kahneman and Tversky assume that there are two phases in the choice process. First, people are assumed to edit or frame prospects so as to simplify evaluation and choice between prospects. It is during the editing or framing phase, as people grapple with the choice problem and try to simplify it, that some of the anomalies such as the ‘isolation effect’ are presumed to enter into the choice process. After editing, people are assumed to evaluate (rank) the prospects and choose the prospect with the highest value (utility). As such, the functional form for the preference function is non-linear: V=

X

vðΔxi Þπðpi Þ

ð4Þ

In Eq. (4), the value V of a prospect is expressed in terms of v and π. Both v and π are scales which respectively (1) measure the value of deviations (gains and losses) from a reference point, and (2) associate with each probability p a decision weight π(p) that reflects the impact of p on the overall value of the prospect (Kahneman & Tversky, 1979, p. 275). This setup enables the authors to assume, in light of their experimental results, that values are attached to deviations or changes from some reference point

The Economic Analysis of Lone Wolf Terrorism

183

rather than to final states and that decision weights do not coincide with stated probabilities (Kahneman & Tversky, 1979, p. 277). As an alternative to expected utility theory, prospect theory has enjoyed significant attention in the literature (Kim, Morse, & Zingales, 2006). From the point of view of terrorism research, the ‘reference point’ and the relevance of ‘losses’ against the reference point are likely to be the most significant features of prospect theory. This has been recognised in the context of bargaining problems by Butler (2007) who explores the implications for the outcomes of strategic bargaining interactions when players’ behaviour is described by prospect theory and different reference points such as the ‘status quo’ are important to the players. Another way to apply prospect theory in terrorism research is by making use of the idea that the outcomes of previous acts of terrorism may be used as reference points by terrorists contemplating a future action. This approach would be relevant to existing work undertaken by terrorism researchers that involves the analysis of situations where terrorists may be thought to reference their actions or demands against those of other terrorists, past outcomes or recent events and may complement research that has used contest theory to examine terrorist behaviour (Caruso & Locatelli, 2008; Caruso & Schneider, 2013). We might guess, for example, that a terrorist who seizes hostages may not be satisfied with a ransom of $12 million if he knows that another terrorist group was able to extract a ransom of $25 million during a recent comparable hostage-taking incident. Prospect theory provides a theoretical framework for the analysis of the effect of this type of consideration on the decisions of the terrorist. It may be particularly useful, as we discuss, to situations where a terrorist is driven to emulate or copy the actions of a predecessor. ‘Copycat’ acts of violence are something that must be taken into consideration when assessing the threat of lone wolf terrorism. Prospect theory allows us to obtain a preference ordering over the available opportunities
attack methods
that takes into account the possibility that the potential lone wolf terrorist assesses each opportunity not on the basis of some final outcome (e.g. total number of fatalities) but with reference to some specific number of fatalities recorded by a predecessor lone wolf whom he seeks to emulate. With regard to ‘copycat’ acts of violence, the behaviour that is most directly relevant to acts of lone wolf terrorism, particularly when those acts take the form of sprees of violence, is the behaviour of ‘school shooters’. Robertz (2007) reports that perpetrators of ‘school shootings’ frequently ‘…state a desire to do it [the spree] ‘better’ than their predecessors
which generally means killing even more people’.

184

PETER PHILLIPS

Let us consider the cases of Kaczynski, whose actions have already been detailed, and Anders Behring Breivik. In Norway in July 2011 Breivik engaged in a bombing and shooting spree that resulted in the deaths of 77 people. The attacks were undertaken in two parts. First, a car bomb was left outside the government building that housed the offices of the prime minister. The bomb was detonated and killed eight people. This is exactly what would be expected from an ‘average’ bombing attack. Breivik then travelled approximately 40 kilometres to a youth camp where he murdered another 69 people in an hour-long shooting spree. For a terrorist attack that might be categorised as an ‘armed assault’, 69 fatalities is many standard deviations from the mean number of fatalities historically generated in such attacks.3 If Kaczynski is a prominent example of ‘serial’ lone wolf terrorism then Breivik is certainly the archetypal example of ‘spree’ lone wolf terrorism. The question that we wish to answer concerns the attack methods that would be chosen by an individual seeking to emulate the actions of either Kaczynski or Breivik. Prospect theory can answer this question if we determine the prospect value (utility) of the alternative attack methods with reference to the number of fatalities inflicted by Kaczynski and Breivik. Although Kaczynski inflicted an accumulated total of 23 injuries and fatalities, when considered individually his attacks inflicted a single injury or fatality. A potential lone wolf who would seek to emulate Kaczynski’s parcel bombing campaign may operate against a reference point of 1 or 2 for each single attack. The total injuries and fatalities inflicted by Breivik during a time-concentrated spree were 228. We seek to determine how a potential lone wolf seeking to emulate the actions of either Kaczynski or Breivik would order his preferences over the attack methods encompassed by the RAND-MIPT attack method categories. This analysis involves computing the prospect values for each attack method using 1, 2, 23 and 228 as reference points. Using the mean and standard deviation (Table 4) and assuming a normal distribution, the probabilities and the cumulative probabilities for each number of injuries and fatalities per attack per year are computed for each attack method.4 Each of the probabilities are then weighted to find πðpi Þ. Using a particular reference point, the change in x is computed for each level of injury and fatality. If, for example, the reference point is 3, then the change in x for 2 injuries and fatalities is
1. The result is a set of values for the various changes in x (above and below the reference point). The final step is multiply each of these values by the weighted probabilities as in Eq. (5) and sum to find the value or utility of the prospect. The preference orderings, based on prospect value, that emerge from the analysis are presented in Table 5. The prospect values are ‘reference point

The Economic Analysis of Lone Wolf Terrorism

185

Table 5. Preference Orderings by Prospect Value: Highest to Lowest. Reference Point = 1 Fatality Assassination Bombing Kidnapping Arson Hijacking Armed attacks Hostage ‘Miscellaneous’ Unconventional

Reference Point = 2 Fatalities

Reference Point = 23 Injuries and Fatalities

Reference Point = 228 Injuries and Fatalities

Bombing Assassination Hijacking Armed attacks Hostage Arson ‘Miscellaneous’ Unconventional Kidnapping

Bombing Armed attacks Hijacking Hostage Unconventional ‘Miscellaneous’ Arson Assassination Kidnapping

Bombing Armed Attacks Hijacking Hostage Unconventional ‘Miscellaneous’ Arson Assassination Kidnapping

dependent’ and the preference ordering changes depending on how the potential lone wolf references his attacks. If he were to reference his attacks to Kaczynski, we should expect him to choose either assassination or bombing (if those opportunities are available to him). On the other hand, if he references his attacks to Breivik, we should expect him to choose either bombing or armed attacks. This type of analysis may prove useful to the investigative process in cases where law enforcement believes there is a chance that a predecessor lone wolf’s actions may be emulated by an unknown suspect. For example, a suspect, who may be identified or unidentified, is known to be obsessed with the actions of a predecessor lone wolf. Under these conditions, if the suspect’s decision making is described by prospect theory5 and the predecessor’s outcomes are a salient reference point, prospect theory may provide some indication to law enforcement about the value or utility that the suspect may attach to particular attack methods. This may help to guide an investigation or the security initiatives undertaken by local law enforcement at potential targets.

CONCLUDING REMARKS Economics is about opportunities and choices. Economic analysis first of all delineates some set of opportunities from which the terrorist or terrorist group chooses and then proceeds to the analysis of the choice that is made from the opportunity set. Usually, this choice is the ‘rational’ or expected utility maximising choice but sub-optimal choices are implicit

186

PETER PHILLIPS

within the delineated set of opportunities and these possible choices are not ignored. Indeed, there is nothing within economic analysis to say that the decision-maker always makes a perfect decision based on a complete calculation and weighing up of alternatives. However, it is easy to see why ‘rational choice’ is the target of criticism. It is not always clear how it is relevant to actual decisions even if it is accepted that the ‘rational choice’ is a benchmark against which actual choices might be compared. This problem is mitigated considerably if we equate the rational choice with something meaningful and significant. Although we may have chosen any relevant payoff, we chose to analyse terrorist opportunities and choices from within a context where the fatalities inflicted by different attack methods are the main consideration of terrorist decision-maker. In such a context the set of rational or expected utility maximising opportunities is equivalent to the most damaging or dangerous opportunities. If we prepare, in this context, for a lone wolf who chooses optimally or rationally from his set of opportunities we are preparing for a lone wolf to choose the most deadly attack method combinations. It would seem unwise to forgo an analysis of these most dangerous opportunities on the basis of an ideological conclusion that ‘terrorists do not choose rationally’ and would therefore not choose them. Economic analysis in general may provide some insights into the terrorist’s logic of choice but it is economic analysis that provides or is based upon quantitative computable results that will be most effectively used in the investigative process. It is no good knowing that the terrorist chooses optimally (or not) unless we can at least narrow down his opportunities and choices. To do this necessitates a certain willingness to abstract from certain details and focus on a few key features of terrorism context. Our analysis focuses on a small list of key features: attack methods, expected fatalities and the risk that the actual outcome may diverge from that which was expected. Rather than try to figure out how the terrorist would order his attack method preferences if he were to apply a full expected utility analysis and weigh up all the payoffs and all the probabilities, we try to determine how the terrorist would order his attack method preferences if he makes an assessment of risk and reward. Terrorists could make such an assessment simply on the basis of what they see reported about different attacks. Such an analysis opens the way for a delineation of the terrorist’s opportunities as a set of risk-reward ‘pairs’ rather than a more complex amalgam of different characteristics. Furthermore, we can define the set of opportunities that dominate all others in terms of reward per unit of risk. A utility maximising terrorist would choose from this sub-set. His choice

The Economic Analysis of Lone Wolf Terrorism

187

depends on his preference for risk. We can identify the attack methods that would be chosen by terrorists who are more risk averse as well as those that would be chosen by more risk seeking terrorists. Some attack methods, such as arson, are all but ruled out and are dominated by other attack methods at each level of risk. If we are willing to attach a particular expected utility function (and risk aversion parameter) to a lone wolf terrorist we can determine the terrorist’s utility maximising attack method choice with precision. However, attack methods are dominant over large enough ranges of the opportunity set to enable us to draw conclusions about the choices that would be made by particular types of terrorists without the need to make additional assumptions. At the lowest levels of risk and the highest levels of risk aversion, the lone wolf terrorist would choose assassination or a combination of assassination and bombing. At the highest levels of risk and the lowest levels of risk aversion, the lone wolf terrorist would choose ‘unconventional attacks’ or a combination of hostage-taking, unconventional attacks and bombing. What is most interesting about the opportunities available to the lone wolf is the concavity of the trade-off between risk and reward. This property of the opportunity set emerges from the imperfect correlations that characterise the payoffs to the different attack types and must not be overlooked by law enforcement for the simple reason that concavity implies higher average fatalities at each level of variability when compared to a linear nonconcave opportunity set. When we turn our attention to the operational details of the investigative process we find that an economic analysis of opportunities and choices can be used to generate inferences about a lone wolf terrorist who has struck and who remains at large. The attack method that the lone wolf chose reveals something about him. Among other things, it reveals his preference for risk. If he has chosen an attack method that is characterised by a relatively lower variability of payoffs we may infer that he is more averse to risk. Further inferences flow, including the inference that he will not be found in the vicinity of his targets and that he will drift from location to location in order to continue his habitation of the low-risk-low-payoff sector of the opportunity set.6 Similar pieces of investigative advice emerge from the application of behavioural economics, especially prospect theory, to the analysis of terrorism and terrorist behaviour. It cannot be ruled out that a potential lone wolf will be driven to copy the actions of a predecessor. If the predecessor’s actions become a reference point for the potential lone wolf, we may use prospect theory to determine how the potential lone wolf will order his preferences over the available attack methods in his act

188

PETER PHILLIPS

of emulation. In this way, economic analysis may contribute to the investigative process. When stripped to its barest fundamental features, economic analysis is about opportunities and choices. Along with other analytical approaches, economic science can contribute to the investigative process designed to pre-empt or pursue the lone wolf terrorist.

NOTES 1. Though when Europe is considered as a whole, there were 38 lone wolves active during the period throughout Europe (Instituut voor Veiligheids en Crisismanagement, 2007). 2. The RAND-MIPT database lists these as: assassination, arson, armed attacks, bombing, kidnapping, hijacking, hostage-taking, unconventional attacks, ‘other’ and ‘unknown’. 3. What might be required is some estimate of an expected number of injuries and fatalities from shooting sprees in confined locations, such as schools, campuses or youth camps. The injuries and fatalities inflicted by lone shooters (though not necessarily lone wolf terrorists) have sometimes been very substantial. The mean and standard deviation of the injuries and fatalities inflicted by such specialised armed attacks may represent the statistical foundation for a new attack method category. 4. The normality assumption is not as unjustifiable as it might seem. The data series we are considering is injuries and fatalities per attack per year not the ‘raw’ series of injuries and fatalities per year. 5. It may not be. That is why contributions from orthodox expected utility theory and mean-variance expected utility should be used in conjunction with each other to sketch several possible profiles for each offender. 6. This is distinct from actions taken to avoid being apprehended. If law enforcement efforts have been enhanced, the average payoff may be expected to have fallen and the risk-reward trade-off will no longer be appropriate. The lone wolf might need to move to another type of attack or another type of location in order to obtain his desired risk-reward trade-off.

REFERENCES AFP. (2010). US man arrested in Washington metro bombing plot. Retrieved form http://www. news.com.au/breaking-news/world/us-man-arrested-in-washington-metro-bombing-plot/storye6frfkui-1225944669226. Accessed on 1 December 2012. Akerlof, G. A., & Kranton, R. A. (2000). Economics and identity. Quarterly Journal of Economics, 115(3), 715
753. Associated Press. (2011). Napolitano: Lone wolf terror threat growing. Retrieved from http:// www.cbsnews.com/8301-201_162-57336080/napolitano-lone-wolf-terror-threat-growing/. Accessed on 30 November 2012.

The Economic Analysis of Lone Wolf Terrorism

189

Atkinson, S. E., Sandler, T., & Tschirhart, J. (1987). Terrorism in a bargaining framework. Journal of Law and Economics, 30, 1
21. Becker, G. S. (1968). Crime and punishment: an economic alternative. Journal of Political Economy, 78, 169
217. Blaug, M. (1992). The methodology of economics or how economists explain. Cambridge, UK: Cambridge University Press. Butler, C. K. (2007). Prospect theory and coercive bargaining. Journal of Conflict Resolution, 51, 227
250. Caruso, R., & Locatelli, A. (2008). Deadly contests: an economic note on Al-Qaeda’s reward system. Economics of Peace and Security Journal, 3, 62
67. Caruso, R., & Schneider, F. (2013). Brutality of jihadist terrorism: A contest theory perspective and empirical evidence in the period 2002
2010. Journal of Policy Modelling, 35(5), 685
696. Coolsaet, R. (Ed). (2011). Jihadi terrorism and the radicalisation challenge in Europe. London: Ashgate. DellaVigna, S. (2009). Psychology and economics: Evidence from the field. Journal of Economic Literature, 47, 315
372. Ehrlich, I. (1973). Participation in illegitimate activities: A theoretical and empirical investigation. Journal of Political Economy, 81, 521
565. Elton, E. J., Gruber, M. J., Brown, S. J., & Goetzmann, W. N. (2003). Modern portfolio theory and investment analysis. New York, NY: Wiley. Gaibulloev, K., & Sandler, T. (2009). Hostage taking: Determinants of terrorist logistical and negotiation success. Journal of Peace Research, 46, 739
756. Instituut voor Veiligheids en Crisismanagement. (2007). Lone wolf terrorism. Instituut voor Veiligheids en Crisismanagement, Netherlands. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263
292. Kim, E. H., Morse, A., & Zingales, L. (2006). What has mattered to economics since 1970? Journal of Economic Perspectives, 20, 189
202. Landes, W. M. (1978). An economic study of U.S. aircraft hijackings, 1961
1976. Journal of Law and Economics, 21, 1
31. Lapan, H. E. & Sandler, T. (1988). To bargain or not bargain: That is the question. American Economic Review Papers and Proceedings of the 100th Annual Meeting of the American Economic Association, 78, 16
21. Lucas, R. E. (1978). On the size distribution of business firms. The Bell Journal of Economics, 9, 508
523. Machina, M. J. (1987). Choice under uncertainty: Problems solved and unsolved. Journal of Economic Perspectives, 1, 121
154. Middle East Media Research Institute. (2012). AQAP releases its English language magazine ‘Inspire’ calling for lone-wolf jihad attacks. Memri Report 6342. Retrieved from www. memri.org Neumann, J. Von, & Morgenstern, O. (1944). The theory of games and economic behaviour. Princeton, NJ: Princeton University Press. Phillips, P. J. (2009). Applying portfolio theory to the analysis of terrorism: Computing the set of attack method combinations from which the rational terrorist group will choose in order to maximise injuries and fatalities. Defence and Peace Economics, 20, 193
213. Phillips, P. J. (2011a). Lone wolf terrorism. Peace Economics, Peace Science and Public Policy, 17(1), 1
29. Article 1.

190

PETER PHILLIPS

Phillips, P. J. (2011b). The diseconomies of terrorism. World Economics, 12, 171
192. Phillips, P. J., & Pohl, G. (2012). Economic profiling of the lone wolf terrorist: Can economics provide behavioural investigative advice? Journal of Applied Security Research, 7, 151
177. Phillips, P. J., & Pohl, G. (2013). Does negotiating with terrorists make them more risk seeking? Journal of Politics and Law, 6, 1–13. Ranstorp, M. (Ed.). (2010). Understanding violent radicalisation: Terrorist and jihadist movements in Europe. London: Routledge. Rasch, W. (1979). Psychological dimensions of political terrorism in the Federal Republic of Germany. International Journal of Law and Psychiatry, 2, 79
85. Robertz, F. J. (2007). Deadly dreams. Scientific American Mind, 18, 52
59. Ross, S. A. (2004). Compensation, incentives and the duality of risk aversion and riskiness. Journal of Finance, 59, 207
225. Sandler, T., & Scott, J. L. (1987). Terrorist success in hostage-taking incidents: An empirical study. Journal of Conflict Resolution, 31, 35
53. Sandler, T., Tschirhart, J. T., & Cauley, J. (1983). A theoretical analysis of transnational terrorism. American Political Science Review, 77, 36
54. Sharpe, W. F. (1966). Mutual fund performance. Journal of Business, 39, 119
138. Spaaij, R. (2010). The enigma of lone wolf terrorism: An assessment. Studies in Conflict and Terrorism, 33, 854
870. Tversky, A., & Kahneman, D. (1992). Advances in prospect theory: Cumulative representation of uncertainty. Journal of Risk and Uncertainty, 5, 297
323. Victoroff, J. (2005). The mind of the terrorist: A review and critique of the psychological approaches. Journal of Conflict Resolution, 49, 3
42. Viner, J. (1932). Cost curves and supply curves. The Journal of Economics, 3, 23
46. Webster, W. H., Winter, D. E., Steel, A. L., Baker, W. M., Bruemmer, R. J., & Wainstein, K. L. (2012). Final report of the William H. Webster Commission on the Federal Bureau of Investigation, counter-terrorism intelligence and the events at Fort Hood, Texas, on November 5, 2009. Federal Bureau of Investigation, Washington DC.

The Economic Analysis of Lone Wolf Terrorism

191

APPENDIX

Fig. A.1.

Average Fatalities and Variability: All Active Terrorist Groups, 2000
2007.

CHAPTER 9 A COMPARATIVE ANALYSIS OF HOMEGROWN TERRORISM Kaisa Hinkkainen ABSTRACT Purpose
An investigation of parallels between homegrown, international, and domestic terrorism. Methodology/approach
A comparative method is used to analyze data from two main sources, ITERATE data on international and the TWEED data on domestic terrorism. The similarities are tested in various dimensions
target types, severity, and the method of the attacks. Findings
Homegrown terrorism is inherently motivated by domestic issues. Moreover, variables of ethnic heterogeneity, political inclusiveness of fringe groups, and problems in the democratization process are good predictors of the occurrence of other forms of domestic and homegrown terrorism alike. Research limitations/implications
Number of observable cases of homegrown terrorism are low. The two main datasets have potentially overlapping incidents.

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 193
216 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022014

193

194

KAISA HINKKAINEN

Originality/value
Provides and operational definition of homegrown terrorism and test empirically the similarity between homegrown and other types of terrorisms. Keywords: Domestic terrorism; homegrown terrorism; international terrorism; rational choice; terrorist targeting

INTRODUCTION The issue of homegrown terrorism and the suspected rise of religious terrorism was just a part of the debates taking place in the aftermath of 9/11, London and Madrid bombings. The scholarly community started unpacking the events from financing the attacks to the recruitment and training the operatives. One of the important questions to answer was, naturally, what caused the phenomenon. The general perception was that al-Qaeda had successfully spread the ideology of political Islam even in Western Europe and recruited foot soldiers to carry out attacks such as in Madrid and London. This, moreover, indicated that homegrown terrorism has a strong international element and that the counterterrorism campaigns in Afghanistan would help to eradicate the instances. The international component, however, missed crucial explanation to a question: why would seemingly “normal” and integrated Muslim men turn to terrorism against their own country? This suggests that perhaps the international element is misleading and to this end it is important to ask the question: is homegrown terrorism similar to international or domestic terrorism, or both? Finding causal relationships in investigating homegrown terrorism is challenging. In Western Europe the total amount of attacks is nine1 which render most statistical analysis meaningless. Nevertheless, it is important to seek explanations to homegrown terrorism in order to counter it effectively. This chapter deals with this problem in particular. By using the comparative method, the chapter examines the similarities and differences between homegrown and other types of terrorism in Western Europe. The empirical results confirm many parallels between homegrown terrorism and other forms of domestic terrorism. First, the targets of their attacks are within the boundaries of the country that they were either born or raised in. The logic of targeting for homegrown terrorists suggests that they have groups of allegiance and loyalty domestically and the attacks are intended to influence these bonds. The aim of homegrown terrorism is domestic political change, not to change the actions of a foreign government. Finally, this

A Comparative Analysis of Homegrown Terrorism

195

chapter illustrates that country propensities for both domestic and homegrown terrorism in Western Europe are very similar. These similarities bear important opportunities for analyzing homegrown terrorism. Since the causes for homegrown terrorism are still largely understudied, finding parallels between domestic terrorisms and homegrown terrorism can help using information of the former to study the latter. Also, knowing how domestic terrorism ends can provide counterterrorism implications to tackle homegrown terrorism. One of the problems in terrorism research are definitional issues that subsequently lack empirical testing (Young & Findley, 2011). This holds true particularly for homegrown terrorism. Some definitions do exist, but they are not tested empirically. This chapter will however address the issue by providing an operational definition as well as using the small number of observations through comparison. The study focuses on 18 Western European countries. The reason for this narrow focus is that even though homegrown terrorism is perceived as a problem in the Western states as a whole, the issue mainly stems from Muslim immigrant populations, which are the largest in Western Europe (Peach, 2007). Some existing literature suggests that the likelihood of a country being targeted by international terrorist groups does depend not only on regime type, welfare policies, trade, population size, inequalities, etc., but also on their international alliances and the strategic value of the foreign targets (Neumayer & Plu¨mper, 2009; Plu¨mper & Neumayer, 2010). These studies provide some intuition as to why Western European governments invest money and military forces in fighting against terrorism abroad although the occurrence of domestic terrorist attacks in Western Europe from 1968 to 2004 is twice as frequent as international terrorist attacks (see Fig. 1). More profoundly perhaps this suggests that the confusion over what is international and what is domestic terrorism influences the decisions to allocate resources in countering terrorism. Obviously frequency of attacks is not the only factor guiding counterterrorism decisions, but also the severity of attacks and the target types of both international and domestic terrorisms. Attacks with large civilian casualties are more likely to exert more pressure for the governments to act proactively in response. Method of the attack, severity, and target types of international, domestic, and homegrown terrorisms are analyzed as indicators of differences or similarities in this chapter. This chapter consists of the following sections. The first section provides a short literature review of homegrown terrorism. The second section outlines a definition of homegrown terrorism, followed by a theoretical frame

KAISA HINKKAINEN

800 600 400 0

200

Number of attacks

1000

1200

196

1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 Year Domestic terrorism

Fig. 1.

International terrorism

Number of Terrorist Attacks in Western Europe.

on similarities in targeting. The research design and empirics are presented in the third section and the final part provides some discussion on counterterrorism and suggestions for future research.

Homegrown Terrorism in the Literature Literature on Islamist or jihadist terrorism has expanded in recent years. The primary focus used to be al-Qaeda and the threat of international jihadist groups to the West (Bergen, 2001; Burke, 2004; Caruso & Schneider, 2013; Roy, 2004; Sageman, 2004, 2008). After the Madrid and London bombings, however, the literature shifted slightly toward Europe investigating recruitment and mobilization, structure of jihadist cells (Neumann, Rogers, Alonso, & Martinez, 2007), radicalization processes (King & Taylor, 2011; Precht, 2007; Ranstorp, 2010), and the characteristics and circumstances of joining jihad in Europe (Bakker & Donker, 2006). Bakker (2008) studied individual level attributes of European jihadists in comparison to the global Salafi network studied by Sageman (2004). The findings suggest that European jihadists are in fact fairly different to the global Salafi studied in Sageman’s sample. The differences can be identified as follows: the European sample had higher average age of joining
they were single men, self-recruited in their home countries and part-time jihadists. The subgroup the European jihadists mainly resemble are the North

A Comparative Analysis of Homegrown Terrorism

197

African Arabs in Sageman’s sample. This study, moreover, demonstrates the gap between the international dimension of Islamist terrorism and the local (European) version, which makes the use of the term homegrown suitable. A common conclusion in all of the studies in Islamist terrorism is that the individuals are heterogeneous; hence no common terrorist profile exists. Many articles conclude that more research is needed into recruitment, mechanisms of radicalization, and active participation in violent acts of terrorism. Homegrown terrorism has not generated much comparative literature and it seems that the debate of old and new terrorism has fused the concepts of international Islamist terrorism and homegrown terrorism into one. I, however, argue that homegrown terrorism is more similar to the other types of domestic terrorisms in Western Europe rather than its international Salafi counterpart. One main reason for this is that it is rooted in the internal scope of the country, whose government is in most cases the target of its action.

DEFINITION OF HOMEGROWN TERRORISM Terrorism is a disputed concept. While no mutually agreed definition of terrorism exists, a commonly used definition in the literature views terrorism as use or threatened use of violence to achieve political gains through intimidation (Sandler, Tschirhart, & Cauley, 1983). While studies in homegrown terrorism have increased, they often lack a clear definition of the term. The definition has to be specific enough to be able to operationalize the concept and collect any, even if very limited, data on the phenomenon. Mullins (2010) identified some features, which are often related to the concept of homegrown terrorists; they were born and/or spent most of their lives in the West, lack direct foreign (non-Western) international support or control, were radicalized within their Western home countries, trained and achieved attack-capability in their Western home countries and planned/carried out attacks in their Western home countries. However, if all these elements are put together, it would constitute such a strict criterion that hardly any organization would fit the category. Crone and Harrow (2011) embark on the same journey to reduce the ambiguity in the use of the term homegrown terrorism by breaking the concept down into two dimensions: belonging and autonomy of the West. Their article provides a useful fourfold categorization of the phenomena at hand and tests the typologies with data on 65 Islamist terrorist attacks and 228 individuals

198

KAISA HINKKAINEN

from 1989 to 2008. Only four events in their data took place in the West. Although the authors introduce an important discussion on the scope and meaning of the term homegrown terrorism, the article fails to address a vital issue of motivation. Regardless of the level of belonging and autonomy of the group, what determines the modus operandi of the group or an individual is their aim. This study mirrors the argument of Engene (2004) suggesting that what constitutes domestic terrorism is dependent on whether the perpetrator is trying to influence the legitimacy of the government in the target country. I also suggest that a definition needs to help in distinguishing one form of terrorism from another, namely homegrown from domestic or international. My definition of homegrown terrorism therefore is individuals, sub-state groups, or organizations who hold a citizenship or residency in the country within which they perpetrate or intend to perpetrate acts of terrorism guided by the ideology of political Islam. Since by definition homegrown refers to internal individuals and groups, the ideological separation differentiates homegrown terrorism from other domestic typologies.

THEORY OF SIMILARITY IN TARGETING In this chapter a communication perspective of terrorism is adopted. This definition focuses on the element of instrumentality in defining terrorism. Terrorist groups are separated from other forms of political violence because the intended audience of violence is different to the immediate targets (Arce & Sandler, 2007; Engene, 2004; Lapan & Sandler, 1993; Schmid & De Graaf, 1982). Engene (2004) specifies that violence by terrorists contains a political message, which is aimed for various different groups in the society. The intention of the message is to influence bonds of loyalty and allegiance between the diverse groups in the society, either strengthening or weakening them. One of these groups can be the public, the state, or other groups within the society, i.e., ethnic groups. The goal of the organization naturally determines who they perceive as the enemy and the group they want to maintain loyalty to. Terrorist organizations strategically choose their targets in order to signal the message to different groups in the society. If an organization wants freedom to its ethnic homeland, the likely targets to signal animosity are the occupying forces, i.e., the police and the military. By doing this the group can at the same time signal allegiance to their ethnic kin group. Target selection, however, is not just

A Comparative Analysis of Homegrown Terrorism

199

a function of goals translated to suitable targets. Terrorist organizations are constrained in their actions not only by their internal capabilities and resources, but also by external factors such as counterterrorism measures hardening their preferred targets. The chapter is based on an overarching framework of rational actor theory. This, moreover, means that terrorist groups are assumed to be rational actors. The contemporary rational choice models follow utilitarian philosophy where individuals choose from a set of alternatives each with given benefits and costs the ones that would maximize their satisfaction (Dugan, LaFree, & Piquero, 2005). Even though rational choice theory applied to terrorism does not take into account individual preferences, it suggests similar behavior at a group level. Furthermore, terrorist groups often have some form of structure, be it hierarchical or network-like structure. The structure does not necessarily suppress individual preferences, but the preferences are most often the same from top to the bottom of the chain. If this is not the case, terrorist groups disintegrate and splinter groups emerge to follow a more unitary approach. At a group level a shared utility can derive from common political goals within a given organization (Enders & Su, 2007). Terrorist organizations are often elites working on behalf of the deprived masses. This is an indication that individualistic preferences are unlikely to play a role in such organizations in any case. The shared political goal can be motivated by any perceived grievance by the group such as lack of political participation, unemployment, and social exclusion (Ross, 1993). Terrorism by definition is obtaining goals through intimidation of a large audience; hence the media attention is essential in achieving these goals. On top of media attention, political instability, fear, intimidation and popular support aid in achieving the utility. These means to achieve the goals are known as the basic commodities. Every terrorist group has certain personnel, financial benefits, buildings and weapons, and entrepreneurial abilities which all constitute a finite set of resources. Given the resources, the terrorist organizations will choose activities that the likelihood of achieving the shared goal is maximized. The relative prices of the basic commodities and the value of the resources constitute the overall price of an attack. The value of resources can be measured in probability of apprehension, severity of the punishments, loss of group members via deaths, likelihood of infiltration and the resources used to plan and execute the attack. In order to make rational choices all the costs must be calculated prior to an attack (Elu & Price, 2007; Enders & Su, 2007). Terrorism is a phenomenon of the margins and terrorists do not often identify with large segments of society. Those large segments of society are

200

KAISA HINKKAINEN

therefore considered to be the legitimate targets of their attacks. The legitimate targets are measured through the frequency of victim typologies. Of course terrorism like any other form of violence generates some unintended consequences, but regardless frequency of target types should imply terrorist groups’ constituencies and their legitimate targets, i.e. who they do not identify with. Based on the theoretical frame, the general hypothesis of terrorist group targeting is as follows: terrorist groups have specific constituencies and they do not attack their own constituents. The legitimate targets of terrorist groups are outside their constituencies and formed by the groups of the society that they do not identify with. From this I derive terrorist type specific hypothesis. Domestic terrorism is disaggregated along ideological typologies, whereas international terrorism, due to data limitations, can only be observed in the data as one aggregate category.

Homegrown Terrorism The actual causes for homegrown terrorism unfortunately are still largely understudied and very few practitioners or scholars have devised empirical tests for these. Understandably for quantitative studies, the lack of testable event data on the dependent variable makes it hard to generalize to a larger population or to achieve significance in statistical tests. Given that the young Muslims or converts are recruited from Western Europe (democratic welfare societies), explanations of abject poverty and injustices seem less relevant. Franz (2007) suggests that Europe has failed to integrate its Muslim immigrant population leading to self-seclusion by the communities. This trend according to Franz is due to the socioeconomic and political conditions in terms of relative deprivation, identity formation and discrimination. Franz provides a good theoretical and comparative approach into the Muslim disaffection in different European countries, but does not provide systematic empirical evidence. The ideological characteristic of homegrown terrorism stems from political Islam. Political Islam as a concept can be defined in various ways and has generated a lot of discussion in recent years.2 In general political Islam refers to “… a form of instrumentalization of Islam by individuals, groups and organizations that pursue political objectives. It provides political responses to today’s societal challenges by imagining a future, the foundations for which rest on reappropriated, reinvented concepts borrowed from the Islamic tradition” (Denoeux, 2002). It is beyond the scope of this

A Comparative Analysis of Homegrown Terrorism

201

chapter to engage in a debate whether political Islam is inherently a violent ideology, but the interpretation of the ideology justifying violence is that of the homegrown terrorists. For the homegrown terrorists killing Kuffar3 is their obligation and fulfilling jihad4 is Allah’s wish (Cole & Cole, 2009). The ultimate goal of political Islam in the minds of homegrown terrorists is to establish the rule of Islam, hence the violent terrorist campaigns are intended to weaken the legitimacy of the government. Moreover by attacking civilian population the government seems unable to protect its citizens and can be consequently perceived less legitimate. The justified targets according to the homegrown terrorists are all non-Muslims, hence civilian target types should be the preferred ones.

International Terrorism International terrorism in contrast to domestic terrorism is intended to influence the policy of an external actor in relation to the country that the terrorist group originates from. As much as international terrorism may stem from domestic grievances, the domestic political actors are not the target of the pressure for political change. An example of the first incident of international terrorism is the case of hijacking of an Israeli airplane in 1968 by the Palestinian Liberation Organization (PLO) (Hoffman, 2006).5 Alexander and Pluchinsky (1992a) make a specific reference to international terrorism experienced in Europe as the Middle-Eastern spill over terrorism. Of course terrorism could spill over from other regions of the world to Western Europe, as it has done for example in the case of the Lockerbie bombing in 1989 where Libyan terrorists sought revenge for the U.S. foreign policy at the Gulf of Sidra by bombing an aircraft over Scottish airspace (Eland, 1998). Many international terrorist attacks in Western Europe have not been claimed by the terrorist organizations in the same manner as domestic terrorist attacks. Partially due to this lack of information on the perpetrators, it is also harder to exactly identify what the reasons for the attacks were. Some researchers however suggest that Western Europe has become the stage for the disputes between Middle-Eastern states and groups, their political and socio-economic grievances, as well as their historical hostilities (Alexander & Pluchinsky, 1992a; Anderson, 1997; Esposito & Voll, 1996; Huband, 1998; Sayre, 2009). Four broader categories for goals in targeting Europe have been suggested by Alexander and Pluchinsky (1992a) as follows: (i) intention of silencing or intimidating political dissidents in exile, (ii) pressuring

202

KAISA HINKKAINEN

the target state to alter its actions and policies, (iii) revenge against persons or states, and (iv) demanding the release of imprisoned comrades. These suggested goals combine both internal and international elements for what the authors call spillover terrorism. Of course having such a broad category for analysis as international terrorism is, it is harder to find a general ideology or motive in explaining the occurrence. It can be however argued that the motivations to target foreign countries have less to do with the internal situation of the target country than the country of origins of the perpetrators. Some foreign policy decisions from the target countries and political alliances that have worsened the situation in the country of origins are more likely to have an impact on the target of the international terrorist groups. As also exemplified in the case of the hijacking of the Israeli airplane, the reason to operate away from the terrorists’ home region was to attract international attention. The mechanisms of influencing the legitimacy of a government can be observed for international terrorism as well. By targeting the civilian population in, for example, Western European states, the target country’s government can lose legitimacy by not being able to protect its citizens or by implementing policies that are viewed as causal factors for the attacks. This can moreover apply pressure for the target government to alter the policies directed at the perpetrators’ country of origin. They try to achieve this pressure by influencing the legitimacy, and this most likely will be best achieved by causing large civilian casualties. Since the group that the terrorist organization wants to show their loyalty to has most likely nothing to do with the target population, they can attack indiscriminately. Indiscriminate attacking is also presumably logistically easier than targeting protected public persons.

Right Wing Terrorism Right wing terrorism appeared in the European landscape initially as a response to the threat of communism although another simultaneous process of post-war immigration occurred in Europe in the 1960s and 1970s, increasing the multicultural environment. Right wing terrorist groups were in favor of maintaining the existing governments and sometimes the strategy involved carrying out attacks and blaming them on the leftist counterparts. This would, moreover, turn people against the leftist groups and boost the legitimacy of the government.6 Some research on right wing terrorist groups does however suggest that the ultimate aim was to overthrow existing governments in order to replace them with fascists ones rather than

A Comparative Analysis of Homegrown Terrorism

203

supporting the status quo.7 One example of right wing terrorist groups is Combat 18, a violent neo-Nazi group originally from the United Kingdom. The organization was formed in 1989 and since then has conducted several attacks on members of ethnic minorities, immigrants and leftists (Lowles, 2001). Once the Cold War was over and communism did not threaten the existing governments, the right wing terrorist groups shifted their focus solely on anti-immigrant nationalist agenda. Again, the anti-immigration attitudes hardly rose from the dust, but most likely the economic crisis fueled the perception of competition between the native population and the immigrants on the job market. The recent homegrown terrorist attacks have increased anti-immigrant sentiment, more specifically against Muslim immigrants. Based on these three categories of terrorism and their groups of animosity and loyalty, I hypothesize: H1. Homegrown, international, and right wing terrorist groups attack the civilian population most frequently rather than state apparatuses. Even though homegrown, international, and right wing terrorism are all hypothesized most likely to attack civilian population, one could expect that international terrorist groups are most likely to cause mass casualties than homegrown or right-wing terrorist groups. This is because for homegrown and right wing groups some of the individuals in their potential supporter pool are located in the same country where the attack takes place; they in theory should be more discriminate in order to avoid casualties of that constituency.

Separatist Terrorism The most dominant form of ideologically motivated domestic terrorism in Western Europe from 1968 to 2004 has been separatist terrorism. The main separatist terrorist organizations in Western Europe from the late 1960s were the IRA in the United Kingdom and ETA in Spain. Both of these organizations wanted independence for their ethnic homelands. Therefore for these groups the preferred targets are the forces of occupation, i.e., military and police personnel. Ethnic struggles would presumably take place within their ethnic region, avoiding civilian casualties since the organizations need support from their ethnic kin group members. In other words, as (Engene, 2004) suggested, the separatist organizations were trying to weaken the legitimacy of the occupying state, while tightening the links between the terrorist organization and their ethnic population. Based on

204

KAISA HINKKAINEN

the rational choices the separatist organizations make in communicating their aims, I hypothesize: H2. Separatist terrorist organizations attack military and police targets most frequently, avoiding civilian casualties in their own ethnic homelands.

Leftist Terrorism The second most frequent type of domestic terrorism strain in Western Europe has been leftist terrorism. Alexander and Pluchinsky (1992b) call the leftist terrorist organizations in Europe fighting communist organizations (FCOs), whose ideological source for the struggle is based on Marxism
Leninism with the aim to get the working class to rise up against the capitalist exploiters. The most prominent group in Western Europe was the RAF, formed in 1978 as a continuation of the Baader-Meinhof Group formed in 1968. The preconditions for the leftist terrorism were founded in the new generation of youth, who in several industrialized countries protested against their parent’s generation and the fascist past (especially Nazism in Germany), the war in Vietnam (and anti-imperialism), changes in the educational system, poor living conditions of the students, and the lack of democracy8 (Crenshaw, 1995). Granted that the aim for the FCOs was to replace the status quo government with a communist one, they wanted to gain a large number of supporters, mainly from the working class masses. This would mean that the terrorist attacks were designed to weaken both the legitimacy of the government as well as the status of the capitalist entrepreneurs. From this I hypothesize: H3. Leftist terrorist organizations attack corporations and government officials most frequently, avoiding civilian casualties.

DATA Data on homegrown terrorism are collected from academic sources (Bakker & Donker, 2006; Cole & Cole, 2009; Harrow, 2010), news reports, as well as from the Global Terrorism Database (GTD).9 The overall amount of incidents is nine incidents dispersed through years 2001
2011. This of course suggests that this type of terrorism only came to existence in

A Comparative Analysis of Homegrown Terrorism

205

Western Europe after the 9/11 attacks in the United States. Out of the nine homegrown terrorist attacks in Western Europe, three (37.50%) took place in the United Kingdom, two in Germany, and the one attack in each of the following countries: Spain, Sweden, the Netherlands, and Italy. Unlike some other academic sources, I exclude foiled attacks and plots, since the information does not allow me to compare homegrown terrorism in relation to other types of domestic terrorisms in Western Europe. The data for the latter is not available. Including arrest data of individuals and foiled plots naturally increases the amount of observations. Harrow (2010) justifies this inclusion as it compensates for the increased counterterrorism measures. However, this is not very useful for comparisons, since counterterrorism measures are often heightened by the target government when the threat of terrorism is perceived high, whether this takes place in the form of legislative change or not. This is not an exception with regards to homegrown terrorism. Counterterrorism has also been argued to be counterproductive,10 which on the other hand means that more attacks could take place regardless of increased counterterrorism efforts. The collected cases for the actual attacks are very few (N = 9) without the foiled incidents, and two of these cases were failed attacks. Regardless of the fact that number of incidents is very low, researchers and policymakers should put emphasis in learning about homegrown terrorism merely due to the high number of injuries and killing generated just by a few attacks in Western Europe. The data on domestic terrorism are compiled from the TWEED dataset and on international terrorism from the ITERATE dataset. For the purpose of comparison, the TWEED dataset years are limited to 1968
2004 (instead of 1950
2004) to match those years covered in the ITERATE dataset. The ITERATE dataset is also limited in its coverage to those 18 Western European countries covered in the TWEED dataset.11 The TWEED dataset codes attacks also by the state agents, which for the purpose of this analysis are deleted. Similarly the ITERATE dataset codes killings of the actual perpetrators into the total amount of individuals killed, which were also deleted. The international terrorism dataset (ITERATE) defines transnational (which I refer to as international) terrorism as “intended to influence the attitudes and behavior of a target group wider than the immediate victims and when, through the nationality or foreign ties of its perpetrators, its location, the nature of its institutional or human victims, or the mechanics of its resolution, its ramifications transcend national boundaries.”12 This loose definition of international terrorism can code virtually all incidents of terrorism as international and therefore biases the distribution of domestic

206

KAISA HINKKAINEN

vs. international attacks. A good example of this is that the attacks by the IRA in mainland Britain are considered as international regardless of the phenomenon being a separatist struggle. The same attacks are of course also included in the domestic terrorism dataset, TWEED. TWEED dataset also uses a criterion in which a terrorist attack is based on the communicative nature of the violent act meaning that there are immediate victims through which terrorists send a message to a larger audience. The coding rules state that any act where the actor originates from outside Western Europe or the act is imported does not constitute domestic terrorism.13 It is however not clear what is meant by actors originating from outside Western Europe as, for example, in the case of homegrown terrorism, the actors often have either residency or citizenship in the countries they target regardless of being born or having family background, for example, in the Middle-Eastern countries. Also the notion of imported terrorism seems rather ambiguous.

EMPIRICS The comparative method examines patterns of similarities and differences across a relatively small number of cases. While it has been often assumed that homegrown terrorism has a lot of distinct features from domestic terrorism, this comparative study aims to identify the patterns of similarities between these typologies. The similarities are tested through target types, country propensities, method of the attack, distributions, and severity. The types of terrorisms are arguably comparable since they already share many similarities. Most importantly, they all exercise terrorist attacks in the Western European countries within the years selected for the sample. According to the theoretical framework, terrorist groups act rationally and choose their targets based on their capabilities and constraints with the aim of maximizing their utility in weakening or strengthening their bonds with a given group in the society (including the state). From this standpoint I derived three testable hypothesis about the likely target types for four ideological types of domestic terrorisms and international terrorism. Homegrown, international and right wing terrorism was predicted to have the highest frequency of civilian targeting. The collected nine cases of homegrown terrorism confirm the hypothesis of civilian targeting as 77.78% of the target types are civilians. With the aim of comparing the types of terrorisms, all types of domestic terrorism

207

A Comparative Analysis of Homegrown Terrorism

follow the logic of legitimate targets outside their own constituency, which translates into most frequent target types with the aim of influencing the bond of allegiance and animosity between different segments of the society. Table 1 summarizes the attacks. Table 2 shows the results for the target types of the four ideological types of domestic terrorisms and the frequencies in percentages. Right wing terrorists target civilians most frequently in 74.02% of the time, thus also confirming the first hypothesis. In addition, the hypothesis that separatist terrorist groups attack military and police targets most frequently is verified. The table, however, also shows that the frequency of civilian targets is very high. The potential explanation for the high number of civilian casualties could be, first, that once the separatist group moves their campaigns away from their ethnic homelands, civilians are not part of their constituencies in the mainland of the occupying forces and therefore they become legitimate targets.14 Second, a substitution effect might take place once Table 1.

Homegrown Terrorist Attacks in Western Europe.

Year

Country of Attack

2004 2004 2005 2005 2005 2006 2007 2009 2011

Spain Netherlands UK UK Sweden Germany UK Italy Germany

Table 2. Target Type Public Military and police Public services/ administration Political institution Business Trade unions Other militants Other

Killings

Injuries

Type of Victim

191 1 52 0 0 0 1 0 4

2050 2 700 1 0 0 5 1 4

Civilian Civilian Civilian Civilian Civilian Civilian Civilian Military Military and civilian

Success Yes Yes Yes Yes Yes No Yes Yes Yes

Target Types of Domestic Terrorism by Ideology. Ethnic/ Separatist

Left Wing

Right Wing

Other

Missing

2,057 27.54% 132 16.82% 282 74.02% 8 15.09% 99 34.26% 2,123 28.42% 130 16.56% 19 4.99% 7 13.21% 28 9.69% 205 2.74% 78 9.94% 6 1.57% 15 28.30% 12 4.15% 150 348 1 102 78

2.01% 74 9.43% 4.66% 149 18.98% 0.01% 2 0.25% 1.37% 3 0.38% 1.04% 62 7.90%

12 11 3 3 6

3.15% 2.89% 0.79% 0.79% 1.57%

3 5.66% 27 8 15.09% 16 0 0.00% 0 2 3.77% 1 7 13.21% 7

9.34% 5.54% 0.00% 0.35% 2.42%

208

KAISA HINKKAINEN

carrying out attacks against the preferred target types becomes too difficult or costly. Counterterrorism efforts can create income effect or substitution effect on terrorism. The former refers to reducing the levels of resources terrorists have at their disposal, whereas the latter refers to counterterrorism designed to reduce certain types of attacks (Enders & Sandler, 1993, 2004). While the substitition effect can indeed alter the results of targets types, it is likely that all terrorist groups are at different stages of their life cycle exposed to counterterrorism efforts. Therefore, the assumption here is that most frquent target types still reflect the underlying preference of targeting even when taking into account the potential substitution effect. The hypothesis for left wing terrorist group target types is also confirmed. The organizations attack business representatives or institutions most commonly.15 The results must be interpreted with some caution due to the possibility of overlapping incidents in the datasets. Table 3 demonstrates the target types for international terrorism and the hypothesis of civilian casualties is not confirmed by the percentages of the target types. The most frequent target type for international terrorism in Western European countries is actually the target country’s government officials and military. The second highest frequency is however civilians with 30.04% of the target types, which might be due to the unintended consequences or subsitution effect. The unconfirmed hypothesis can be an indication that international terrorism does not, as a generic typology, follow the same logic of targeting as domestic terrorisms. Perhaps, as Alexander and Pluchinsky (1992a) note, the motives for attacks could be revenge, demanding a release of imprisoned comrades, intention of silencing or intimidating political dissidents in exile, and pressuring the target state to alter its actions and policies. It might be that the mechanisms to achieve these aims is not the same for that of influencing the bonds between different groups in the target society. It is Table 3. Type of Immediate Victims. Target Types Host government officials and military Foreign government officials and military Corporation officials Prominent opinion leaders Private parties, including tourists, missionaries, and students Indeterminate Total

Number

Percentage

186 1,143 805 124 1,095 292 3,645

5 31 22 3 30 8 100

A Comparative Analysis of Homegrown Terrorism

209

also possible that international terrorist groups have an advantage from an element of surprise since domestic terrorist groups are under more scrutiny in the given country of attack. On the other hand the logistical implementation of an attack should be harder for international terrorist organizations. One can also speculate that the international terrorist groups have better resources and capabilities to attack government and military officials than the domestic terrorist groups. If however the strategic logic of targeting applies similarly to domestic terrorisms, the preferred target type should be civilian population, which should in principle be easier to conduct attacks against. Testing similarities in methods of attack, distribution, and severity yielded somewhat mixed results. The most frequent method of attack for all the three typologies of terrorisms is bombings. Due to data limitations, it is not possible to disaggregate bombings into further categories, but approximately 78% of homegrown terrorist attacks has involved some form of bombing, 72% of domestic terrorism and 62% of international terrorism. This would lend some support to more similarities between homegrown and domestic terrorism. Domestic and homegrown terrorism have a second preference over armed attacks, whereas in international terrorism this type of attack is fourth most frequent. Using the dissaggregated GTD dataset (Enders, Sandler, & Gaibulloev, 2011), I compared domestic and international terrorism using the Kolmogorov
Smirnov equality-ofdistributions test. Due to the small number of observations, it is not meaningful to use an equality-of-distributions test for homegrown terrorism in relation to international or domestic terrorism. However, given that the disaggregated GTD data codes homegrown terrorism as domestic terrorism, it is possible to establish that at minimum, when homegrown terrorism is observed as part of domestic terrorism, it differs from international terrorism. The null hypothesis that the distributions are equal is rejected (p-value 0.005). This, moreover, lends support to the hypothesis that homegrown terrorism as part of domestic terrorism differs from other forms, namely, international terrorism. Severity comparisons produce slightly different results. Due to the low number of observations, the values of injuries and killings of homegrown terrorist attacks are not useful for interpretation as such. Moreover, one can infer that the range of killings from homegrown terrorism is 0
191 and the range of injuries 0
2050. The range of killings in domestic terrorism is 0
85, injuries 0
220. The corresponding figures for international terrorism are 0
190 and 0
231. The range of the figures in themselves show that the range of injuries of homegrown terrorism is much wider than for other types of domestic terrorisms as well as

210

KAISA HINKKAINEN

international terrorism, but that the range of killings is more similar to international terrorism. The chapter sets out to analyze homegrown terrorism not as a distinct independent phenomenon, but either as part of broader typology of international or domestic terrorism. Previous literature has been inconclusive in explaining which category homegrown terrorism belongs to. Even though the term has clear implications over the fact that homegrown is somewhat a domestic phenomenon, recent counterterrorism efforts by the Western European governments in the Middle-East suggest otherwise. Domestic terrorism is a function of decreased legitimacy in a country and domestic terrorist groups send messages of allegiance and animosity by using terrorist attacks as an instrument of pressure. According to Engene (2004) legitimacy can be reduced down to three variables, which helps us to categorize Western European countries into groupings of more and less likely to experience domestic terrorism. These variables measure the ethnic heterogeneity of the country, continuity (operationalized as problems in the democratization process), and integration of the fringe groups in the society to the political arena. The most likely countries to experience terrorism are the ones that are ethnically heterogeneous, having problems of continuity and inclusiveness in the political system. By classifying the countries according to the values on these variables, France, Italy, and Spain are most likely to experience terrorism domestically scoring positively in all the legitimacy problem categories. Table 4 illustrates the 18 Western European countries in relation to the 3 legitimacy measures as well as levels of domestic terrorism, homegrown terrorism, and percentage of Muslim populations. Domestic and homegrown terrorism provide many parallels in this analysis. First, homegrown terrorism has taken place in six separate Western European countries (the United Kingdom, Italy, Spain, Germany, the Netherlands, and Sweden). Of these countries, the United Kingdom, Spain, Italy and Germany are in the top five countries for most frequent domestic terrorism. According to Engene (2004) data, three out of the six countries with a homegrown terrorist event are ethnically heterogeneous. The measure used to capture this is very arbitrary and coded such that when a country’s dominant group constitutes 96% of the population, a country is ethnically homogeneous. Ethnic heterogeneity in turn influences the likelihood of terrorism through potential cultural and social discrimination. While it is perfectly sensible to assume that social and political discrimination can lead to higher risk of terrorism, the size of the ethnic minority might not be in any ways correlated to the occurence of terrorism. Even small minorities can use terrorism as a weapon, moreover

211

A Comparative Analysis of Homegrown Terrorism

Table 4. Country

Austria Belgium Denmark Finland France Germany Greece Iceland Ireland Italy Luxembourg Netherlands Norway Portugal Spain Sweden Switzerland United Kingdom

Domestic and Homegrown Terrorism in Western Europe.

Group Continuity Integration Ethnicity Domestic Percentage of Homegrown Terrorism Muslim Terrorism Immigrant 3 5 1 7 8 4 4 1 3 8 2 1 1 4 8 1 5 5

Yes No No Yes Yes Yes Yes No Yes Yes No No No Yes Yes No No No

No No No Yes Yes Yes Yes No No Yes Yes No No Yes Yes No No No

No Yes No No Yes No No No No Yes No No No No Yes No Yes Yes

16 52 5 0 2801 296 248 0 26 351 1 31 2 204 855 6 33 4050

4.1 3.9 5.0 0.3 6.7 4.2 1.3 N/A N/A 1.4 N/A 5.8 2.0 N/A 1.2 2.3 4.2 2.7

0 0 0 0 0 2 0 0 0 1 0 1 0 0 1 1 0 3

The data for domestic terrorism and country categorization are from Engene (2004) and the percentages of Muslim immigrants per country from early to mid-2000s are from Peach (2007).

the conventional wisdom suggests exactly terrorism to be the weapon of the weak. The more apprpriate measure would be to look which countries, for example, have minorities at risk and use that as a predictor of more likely occurences of domestic terrorism. Another parallel observed in Table 4 is that France, the Netherlands, and Germany are in the top five countries with most Muslim immigrants in Western Europe and have also experienced homegrown terrorism. The data on Muslim immigrants further casts a shadow over the reliability of the measure of ethnic homegeneity. Just by having over 4% of their population Muslim immigrants should give countries a score of yes in the problems of ethnic diversity categorization. This is not the case for Austria, Denmark, Germany, and the Netherlands. Inherently, domestic terrorism is set out to change the conduct of domestic players, be it the government or other groups in the society. As an example the Red Army Faction (RAF) in Germany wanted to overthrow the fascist government and spread leftist ideology in Germany. Regardless of having widespread international attention and alleged

212

KAISA HINKKAINEN

alliances, the RAF was seeking political change in Germany. By attacking business targets in Germany, a message of animosity was sent to the capitalist exploiters, pressure was exerted to the government who could not protect its citizens, and a message of loyalty was projected to the masses, most importantly working class masses. This similar logic of terrorism as an instrument of communication can be observed in the nice cases of homegrown terrorism. The groups are seeking for a pan-Islamist caliphate in their given countries and are often unhappy about the foreign policy choices of their governments. These groups or individuals may feel a lack of integration, which has driven them to feel animosity toward the large segments of society. The target types of the nine attacks reflect this. By attacking the civilian population, pressure to the government is applied in order to change the political, economic, or societal landscape of the country. Given that homegrown terrorism has the same logic of targeting as other domestic terrorisms, revealing the motivations and aims of the groups, we can apply lessons learned from domestic terrorism to homegrown terrorism. The next section provides explicit counterterrorism lessons to curb homegrown terrorism.

DISCUSSION Since homegrown terrorism has only a few observations, this chapter introduced a comparison of the logic of targeting, method of attack, distribution, country propensities, and severity in relation to other types of domestic terrorism and international terrorism. This moreover enables us to identify whether homegrown terrorism indeed resembles international or domestic terrorism and draw lessons from one or the other. In order to do this the study provided an operational definition of homegrown terrorism and used empirical evidence about the similarities and differences. The empirical evidence shows that homegrown terrorism indeed resembles domestic terrorism in its logic of targeting, country propensities, and methods. In order to study homegrown terrorism further, we should learn from other types of events that are similar, i.e., domestic terrorism. The counterterrorism lessons should also be drawn from other domestic terrorist groups and how their campaigns came to an end. Majority of terrorism actually ends in joining the political process (43%). The second most common way of terrorism to cease is local police and intelligence work, but military force has rarely ended a terrorist campaign (Jones & Libicki, 2008).

A Comparative Analysis of Homegrown Terrorism

213

In order to counter the threat of terrorism, the disaggregation into different types of terrorisms has to be taken into account. Since this study established that homegrown terrorism indeed has many similarities with domestic terrorism, the phenomena should be countered using the methods that have worked for other types of domestic terrorisms. The feelings of exclusion in the societies should be addressed, whereas military interventions in Afghanistan and Iraq are unlikely to have an impact on homegrown terrorism in the West. Even though it is beyond the scope of this chapter to study the motivations of homegrown terrorism, it should be noted that there may be some universality in the motivations of domestic terrorisms. Most scholars of terrorism would duly reject the possibility of generalizing such a complex phenomena, but perhaps there is room for generalizability within the two main typologies: domestic and international. Again, we may know a lot about the roots of separatist, leftist and right wing terrorism, but we do not know enough about the roots of homegrown terrorism. And from a conflict resolution perspective, this would be the only way to solve the problem. Moreover, empirical research on the motivations for homegrown terrorism should be the next step in terrorism research.

NOTES 1. Two attacks in Germany (2006 and 2011), three attacks in the United Kingdom (two in 2005 and one in 2007), and one attack in the Netherlands (2004), Italy (2009), Spain (2004), and Sweden (2005). 2. For further discussion see, for example, Fuller (2002) and Ayoob (2008). 3. Kafir is a non-believer, plural Kuffar. The homegrown terrorists use this as a derogatory term, dehumanizing the non-Muslims to theologically justify killing them. 4. Based on the interpretation of Qur’an, jihad can be either violent or peaceful. Homegrown terrorists perceive themselves as soldiers fighting a war against the enemies of Islam, hence violent jihad is acceptable. 5. The terrorists boarded the plane in Rome and diverted the plane into Algiers where they kept some passengers as a hostage while negotiating with the Israeli government. This event is a watershed episode marking the beginning of international terrorism, since it was the first time that the PLO took their struggle outside their region in order to bring international attention to their cause. One of the claims made by the PLO was to exchange imprisoned Palestinian comrades for the Israeli hostages. Time, 21.9. 1970, World: Drama of the Desert: The Week of the Hostages, available at: http://www.time.com/time/magazine/article/0,9171,942267-1, 00.html, last accessed April 21, 2011.

214

KAISA HINKKAINEN

6. Transnational Terrorism, Security and the Rule of Law, 2008, 20th Century Right Wing Groups in Europe Prone to Extremism or Terrorism? Available at: http://www.transnationalterrorism.eu/tekst/publications/Rightwingterrorism.pdf, last accessed April 21, 2011. 7. The National Consortium for the Study of Terrorism and Responses to Terrorism (START) Terrorist Organization Profiles for Right Wing Reactionary Groups, available at: http://www.start.umd.edu/gtd/features/TOPs.aspx, last accessed April 21, 2011. 8. The Observer, 20 January 2008, available at: http://www.guardian.co.uk/ world/2008/jan/20/1968theyearofrevolt.features, last accessed April 21, 2011. 9. Global Terrorism Database (GTD), available at: http://www.start.umd.edu/ gtd/, last accessed April 20, 2011. 10. Several scholars have studied the potential counter-productiveness of proactive measures by governments in tackling terrorism as well as the positive influence of a spectacular attack on terrorist recruitment. See, for example, Bueno de Mesquita and Dickson (2007). 11. The countries covered are Austria, Belgium, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and the United Kingdom of Great Britain and Northern Ireland. 12. ITERATE codebook, 2003, p. 2. 13. TWEED codebook, 2006. 14. As evidence of this please see Eggen and Jo (2010) on the logic of Palestinian terrorist groups targeting differently in the occupied territories vs. in the state of Israel. 15. The DTV dataset gives a slightly different result; nationalist (pro-secessionist) groups attack civilians most frequently, left wing terrorists attack police most frequently, and right wing groups attack civilians most frequently.

REFERENCES Alexander, Y., & Pluchinsky, D. A. (1992a). European terrorism: Today and tomorrow. Brassey’s Washington: Brassey’s. Alexander, Y., & Pluchinsky, D. A. (1992b). Europe’s red terrorists: The fighting communist organizations. London: Frank Cass. Anderson, L. (1997). Fulfilling prophecies: State policy and Islamist radicalism. In John L. Esposito (Ed.), Political Islam: Revolution, radicalism, or reform (pp. 17
31). Boulder, Colorado: Lynne Reinner Publishers. Arce, D. G., & Sandler, T. (2007). Terrorist signalling and the value of intelligence. British Journal of Political Science, 37(4), 573
586. Ayoob, M. (2008). The many faces of political Islam: Religion and politics in the Muslim world. Ann Arbour: University of Michigan Press. Bakker, E. (2008). Jihadi terrorists in Europe and global Salafi jihadis. In Jihadi terrorism and the radicalisation challenge in Europe. The Hague: Ashgate.

A Comparative Analysis of Homegrown Terrorism

215

Bakker, E., & Donker, T. H. (2006). Jihadi terrorists in Europe: Their characteristics and the circumstances in which they joined the jihad: An exploratory study. Clingendael, The Hague: Netherlands Institute of International Relations. Bergen, P. L. (2001). Holy war, Inc. New York: Simon & Schuster. Bueno de Mesquita, E., & Dickson, E. S. (2007). The propaganda of the deed: Terrorism, counterterrorism, and mobilization. American Journal of Political Science, 51(2), 364
381. Burke, J. (2004). Al-Qaeda: The true story of radical Islam. London: IB Tauris. Caruso, R., & Schneider, F. (2013). Brutality of jihadist terrorism: A contest theory perspective and empirical evidence in the period 2002
2010. Journal of Policy Modeling, 35(5), 685
696. Cole, J., & Cole, B. (2009). Martyrdom: radicalisation and terrorist violence among British Muslims. London: Pennant Books. Crenshaw, M. (1995). Terrorism in context. Philadelphia: Penn State University Press. Crone, M., & Harrow, M. (2011). Homegrown terrorism in the west. Terrorism and Political Violence, 23(4), 521
536. Denoeux, G. (2002). The forgotten swamp: Navigating political Islam. Middle East Policy, 9(2), 56
81. Dugan, L., LaFree, G., & Piquero, A. (2005). Testing a rational choice model of airline hijackings. In P. Kantor, G. Muresan, F. Roberts, D. Zeng, F.-Y. Wang, H. Chen, & R. Merkle, (Eds.), Intelligence and security informatics (Vol. 3495, pp. 513
529). Lecture Notes in Computer Science. Berlin: Springer. doi:10.1007/11427995_28 Eggen, H., & Jo, R. (2010). The logic of Palestinian terrorist target choice? Examining the Israel defense forces’ official statistics on Palestinian terrorist attacks 2000
2004. Studies in Conflict & Terrorism, 33(2), 134
148. Eland, I. (1998). Does US intervention overseas breed terrorism? Cato Institute Foreign Policy Briefing, 50, 1
24. Elu, J. U., & Price, G. N. (2007). Remittances and the financing of terrorism in sub-Saharan Africa: 1974
2006. Peace Economics, Peace Science and Public Policy, 18(1), 1
42. Enders, W., & Sandler, T. (1993). The effectiveness of antiterrorism policies: A vectorautoregression-intervention analysis. American Political Science Review, 829
844. Enders, W., & Sandler, T. (2004). What do we know about the substitution effect in transnational terrorism. Researching Terrorism: Trends, Achievements, Failures, 119, 137. Enders, W., Sandler, T., & Gaibulloev, K. (2011). Domestic versus transnational terrorism: Data, decomposition, and dynamics. Journal of Peace Research, 48(3), 319
337. Enders, W., & Su, X. (2007). Rational terrorists and optimal network structure. Journal of Conflict Resolution, 51(1), 33
57. Engene, J. O. (2004). Terrorism in Western Europe: Explaining the trends since 1950. Cheltenham: Edward Elgar Publishing. Esposito, J. L., & Voll, J. O. (1996). Islam and democracy. New York: Oxford University Press. Franz, B. (2007). Europe’s Muslim youth: An inquiry into the politics of discrimination, relative deprivation, and identity formation. Mediterranean Quarterly, 18(1), 89
112. Fuller, G. E. (2002). The future of political Islam. Foreign Affairs, 81(2), 48
60. Harrow, M. (2010). The effect of the Iraq war on Islamist terrorism in the West. Cooperation and Conflict, 45(3), 274
293. Hoffman, B. (2006). Inside terrorism. New York, NY: Columbia University Press. Huband, M. (1998). Warriors of the prophet. Boulder: Westview Press.

216

KAISA HINKKAINEN

Jones, S. G., & Libicki, M. C. (2008). How terrorist groups end: Lessons for countering Al Qa’ida. Santa Monica: Rand Corporation. King, M., & Taylor, D. M. (2011). The radicalization of homegrown jihadists: A review of theoretical models and social psychological evidence. Terrorism and Political Violence, 23(4), 602
622. Lapan, H. E., & Sandler, T. (1993). Terrorism and signalling. European Journal of Political Economy, 9(3), 383
397. Lowles, N. (2001). White riot: The violent story of Combat 18. Bury: Milo Books. Mullins, S. (2010). Home-grown terrorism: Issues and implications. Perspectives on Terrorism, 1, 3. Neumann, P. R., Rogers, B., Alonso, R., & Martinez, L. (2007). Recruitment and mobilisation for the Islamist militant movement in Europe. ICSR. Neumayer, E., & Plu¨mper, T. (2009). International terrorism and the clash of civilizations. British Journal of Political Science, 39(4), 711
734. Peach, C. (2007). Muslim population of Europe: A brief overview of demographic trends and socioeconomic integration, with particular reference to Britain. In Muslim integration: Challenging conventional wisdom in Europe and the United States. A Report of the Center for Strategic and International Studies Transatlantic Dialogue on Terrorism. Plu¨mper, T., & Neumayer, E. (2010). The friend of my enemy is my enemy: International alliances and international terrorism. European Journal of Political Research, 49(1), 75
96. Precht, T. (2007). Home grown terrorism and Islamist radicalisation in Europe. From coversion to terrorism. An Assessment of the factors influencing violent Islamist extremism and suggestions for counter radicalisation measures. Copenhagen: The Danish Ministry of Justice. Ranstorp, M. (2010). Understanding violent radicalisation: Terrorist and jihadist movements in Europe. Routledge, Abingdon: Taylor & Francis. Ross, J. I. (1993). Structural causes of oppositional political terrorism: Towards a causal model. Journal of Peace Research, 30(3), 317
329. Roy, O. (2004). Globalised Islam: The search for a new Ummah. London: C. Hurst and Co. Publishers. Sageman, M. (2004). Understanding terror networks. Philadelphia: University of Pennsylvania Press. Sageman, M. (2008). Leaderless jihad: Terror networks in the twenty-first century. Philadelphia: University of Pennsylvania Press. Sandler, T., Tschirhart, J. T., & Cauley, J. (1983). A theoretical analysis of transnational terrorism. The American Political Science Review, 77(1), 36
54. Sayre, E. A. (2009). Labor market conditions, political events, and palestinian suicide bombings. Peace Economics, Peace Science and Public Policy, 15(1), 1. Schmid, A. P., & De Graaf, J. (1982). Violence as communication: Insurgent terrorism and the Western news media. Beverly Hills, CA: Sage. Young, J. K., & Findley, M. G. (2011). Promise and pitfalls of terrorism research. International Studies Review, 13(3), 411
431.

CHAPTER 10 CLOSE TO THE EDGE: CYBERTERRORISM TODAY Giampiero Giacomello ABSTRACT Purpose
This chapter examines the phenomenon of cyberterrorism, which is a form of terrorism that could be potentially devastating as terrorists could, through computer networks, wreak havoc in the critical infrastructure of an advanced country. Methodology/approach
Review of relevant literature. Findings
The chapter argues that there are actually two types of cyberterrorism: “Hard-Rock” and “Lite.” Both are potential “force multiplier” for terrorists. The former relies on attacking a country’s infrastructures via computer networks and blocking the working of critical functions like water and energy distribution and the like. Because of its complexity and costs, this type of cyberterrorism is far from frequent and actually almost non-existent. The latter type (Lite) is much more commonplace and is the capability by terrorist organizations to use the Web (and the Internet) to communicate with their affiliates, recruit new elements and, most importantly, publicize their deeds.

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 217
236 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022015

217

218

GIAMPIERO GIACOMELLO

Research limitations/implications
Because the use of the Web and the Internet by terrorists give intelligence agencies more opportunities to actually monitor them, this form of terrorism is much more a “doubleedged sword” than regular terrorism. Originality/value of the chapter
The chapter tries to better classify and compare the two forms of cyberterrorism; it also shows how remote the prospect of Hard-Rock Cyberterrorism still is. Keywords: Cyberterrorism; information operations; psychological operations; computer networks; terrorists’ communications

INTRODUCTION If a terrorist organization uses the Internet to spread its messages and show gory videos of beheaded hostages or of civilians killed by NATO airstrikes, would that be an act of “cyberterror” comparable to a group of highly skilled hackers taking over the air control system of the, say, New York JFK airport and ground all the planes?1 As many governments and observers (see Colarick, 2006 for example) tend to see both instances belonging to the class of cyberterrorism, the answer should be affirmative. The author would define cyberterrorism only those attacks that, via computer networks, can disable or disrupt computers and their networks causing serious material and/or economic damage and, possibly, loss of life. This, however, is not the prevailing view. Inevitably, this state of affairs is a source of confusion (Embar-Seddon, 2002; Stohl, 2006). To (partially) correct this situation, it could be argued that there are two distinctive types of “cyberterrorism”: the “Hard-Rock” version includes attacks by terrorists of a country’s infrastructures via computer networks and to block critical functions such as water and energy distribution and cause disruption and possibly casualties. The “Lite” version is the competence by terrorist organizations to use the Web (and the Internet) to communicate with their affiliates, recruit new elements and, most importantly, publicize their deeds. Disrupting water distribution or air traffic is quite a different deed than posting fundamentalist or political extremist material on the Web and perhaps only the former should be termed and technically considered “cyberterrorism.” However, there are several reasons why the same term is used with reference to both forms and thus both instances will be examined in this chapter.

Close to the Edge: Cyberterrorism Today

219

The first of such reasons is the inner structure of computer networks: although the Internet and the Web are not the same thing (as the information transmitted and the infrastructure are not two different realms), they are two parts of the same architecture. Libicki (2007, pp. 236
240), for example, suggests a three-layer structure, comprising the “physical” (the hard physical infrastructure of computers, routers, and cables), the “syntactic” (the software and protocols which control network flows), and the “semantic” (the actual information and ideas exchanged in human language through the network). The former two are subject to “Hard-Rock” cyberattacks, whereas the latter is the object of “Lite cyberterrorism.” The second reason is that because there is no universally accepted definition of terrorism, hence, there is no agreed-upon definition of cyberterrorism, so that almost any “terrorist” action related to or involving computers can be “cyberterrorism.” A more viable alternative would be to define cyberterrorism those terrorist actions that target computers and networks, while recruiting, intelligence-gathering and propaganda via computers should be defined as “computer-assisted” terrorism, for example. For example, for Colarick (2006, p. 47), it was “attacks against information and computer systems that result in physical violence” to spread fears in non-combatants. Hughes (2004) defined cyberterrorism as “a diverse set of technologies that ranges from viruses and denial-of-service attacks to posting messages, pictures and videos on Web sites whose purpose is to scare people.” For Denning (2001), it was the convergence of cyberspace and terrorism, with its unlawful attacks and threats of attacks against computers, networks and the information stored therein when done to intimidate or coerce a government or its people in furtherance of political or social objectives. In 2002, for the “old” experts of the Computer Emergency Response Teams (CERT) Centers at the Carnegie Mellon University, it was the deliberate destruction, disruption or distortion of digital data or information flows with widespread effect for political, religious or ideological reasons.2 Even the military cannot help as they bundle together many diverse activities because the U.S. Army Field Manual (Department of the Army, 2003, pp. 1
5) defines the activities of terrorists engaged in cyberspace “range from gaining unauthorized access to C2 [Command and Control] systems to physical attacks against commanders and decision-makers.” According to the manual, information networks are used by terrorists to pass intelligence and technical data across international borders. Using the Internet and the Web to spread terrorist messages thus should not qualify as “cyberterrorism.”

220

GIAMPIERO GIACOMELLO

More “conservative” views of cyberterrorism suggest to take into account the “efficiency” of cyberterror, measured in number of people killed (Giacomello, 2003, 2004). As such, cyberterrorism would not and cannot be cost-efficient. As the scope of terrorists is to spread fear. Today car bombs are deadly, but lunatics myths and false news on the Web may also be “terror multipliers,” hence quite useful for the terrorists’ goals. Introducing terms such as cracktivism, cyber jihad, and cyber dissent in connection to “cyberterrorism” has, unfortunately, only reinforce such effect. And what is briefly summarized here is just a tiny sample of the vast literature retrieved today on the topic. Today, many observers are alarmed that both Hard-Rock and Lite cyberterrorism belong to the terrorists’ arsenals. But, which is the most common online? And which one is the most likely to produce effects that justify the investment in resources and personnel? At the end of the 1990s, a white paper by the Center for the Study of Terrorism and Irregular Warfare noted that “the majority of literature dealing with cyberterrorism has focused principally on the vulnerabilities of critical infrastructures” (Nelson, Rodney, Iacobucci, Mitchell, & Gagnon, 1999, p. 3). The scenarios for exercises like “Black Ice” or “Blue Cascade”
held in the United States in 2002 and described by Verton, 2003
all focus on penetration of critical infrastructure and disruption of SCADA systems.3 Quite soon, however, scholars found that that was not the case with cyberterrorism. Conway (2002) identified a pattern of Internet use by terrorists that did not imply destruction (or even disruption) of “the medium” (the Web). Moreover, Weimann (2004) observed that no terrorist organization (including al-Qaeda) appeared to have tried to stage a serious cyberattack, with the most damaging attacks and intrusions typically carried out either by disgruntled corporate insiders intent on embezzlement or sabotage or by individual hackers (young and male) seeking thrills and notoriety. Along the same lines were the conclusions of Lewis (2002), Schneier (2003), Giacomello (2004), Singer (2011) and Helms, Costanza, and Johnson (2012). It is, therefore, not Hard-Rock cyberterrorism that seems to be prevailing online. Paradoxically, it is actions in “support of cyberterrorism” (Conway, 2002; Nelson et al., 1999; Thomas, 2003; Weimann, 2004), that is, the Lite type, that are now common. It is those actions commonly defined as “cyber-planning” (Thomas, 2003) that are the most “popular” among terrorist organizations when it comes to the Internet and computer networks. For almost a decade, it seemed that the United States and other advanced democracies
all of them in the OECD club that invented, built

Close to the Edge: Cyberterrorism Today

221

and, for a time, “owned” the Internet
could only concentrate on nongovernmental terrorist groups (and likewise cyberterrorists) as adversaries. In the post-9/11 trauma period, it was all understandable. Furthermore, for a while, Islamic fundamentalists and ideological terrorists showed a certain aptness at seizing that icon of modernism and Western culture that was the Internet. Those groups turned the Internet (and the Web more so) into an effective means to spread their message and enhance their global presence, while the very countries that invented and developed the network have almost been taken by surprise. Today many a government considers that other state actors will be their most likely adversaries. Thus, if statesponsored cyberterrorism, as part of information operations (IO) or part of computer network operations (CNO), may be a force multiplier in a war between states, what should we have to expect for tomorrow?4

“HARD-ROCK” CYBERTERRORISM Hard-Rock cyberterrorism would be quite close to cyberwarfare, as the tools are basically the same. As it is intended today by all major military powers, cyberwarfare (or more precisely, computer network operations (CNO), which for NATO is part of information operations) is an “enlarged” version of (old) electronic warfare (EW). Much like EW would have introduced a major conflict between the Soviets and the West, CNO would certainly characterize the opening stages of any major armed conflict. Thus cyberwarfare is a most serious phenomenon and, for such reason, it is today part of any military planning undertaken by national armies of those countries that are dependent on information infrastructures. Much as cyberwarfare, Hard-Rock cyberterrorism would use multiple techniques to penetrate the defenses and take control, via remote computers and their networks, of critical infrastructures, with the intent of causing disruption and collapse of daily activities and possibly casualties. In the late 1990s, a serious concern was that cyberterrorists would take over the remote control of power distribution systems or air traffic control systems, and, by compromising the software managing those systems, cause major physical destruction and fatalities (Zannini & Edwards, 2001). “Scaremonger” exercises, like Eligible Receiver,5 Solar Sunrise, or Black Ice (Vernon, 2003), organized by the U.S. federal and state government agencies, implied disruption of SCADA systems to cause physical and economic damage. Some press reports even seemed to substantiate that

222

GIAMPIERO GIACOMELLO

forecast (Anderson, 2002; Arena & Ensor, 2002; Embar-Seddon, 2002), a trend that continues unabated (Molfino, 2012). Yet, nothing of that disastrous magnitude has materialized. Indeed, severe power outages on the Eastern Coast of the United States and Canada in August 2003 and in Italy in September of the same year demonstrated that modern societies are more resilient in recovering than many would expect.6 Undoubtedly, Hard-Rock cyberterrorism may serve as multiplier to increase the number of people injured or killed by making rescue and recovery more difficult. This increased efficiency would, of course, come with a cost: to train or hire teams of hackers that remotely could disrupt communications would be expensive compared to more “traditional” terrorist tactics (Denning, 2001; Giacomello, 2004; Lederer, 2004; Shane, 2010; Vittori, 2011). In the end, this type of cyberterrorism is the most difficult to accomplish, as it requires the deep pockets and extensive expertise that are basically monopoly of sovereign states, and not all of them (Giacomello, 2004). When Stuxnet hit the Iranian nuclear program, experts identified only a handful of countries capable of achieving such results (Broad & Sanger, 2010; Sanger, 2010). Today such “cyber-elite” is made of 12 of the most advanced militaries of the world (Shane, 2012). It is unsurprising thus that several authors (Conway, 2002; Denning, 2001; Giacomello, 2004; Green, 2002; Lewis, 2002; Ranum, 2004; Schneier, 2003) agree that not only this type of cyberterrorism is not a realistic threat and that resources should be focused elsewhere, but also that there has never been a recorded act of Hard-Rock cyberterrorism pre- or post-September 11. Terrorist organizations probably regard strikes at computer networks as a tempting, new tactic, but such attacks should be sufficiently destructive or disruptive to generate fear comparable to that from physical acts of terrorism. Successful Hard-Rock cyberattacks would have to lead to death or bodily injury as result of extended power outages, plane crashes, or water contamination. None of this has yet happened as consequence of terrorist actions. For terrorist groups, more traditional forms of attacks like car bombings are still far more dangerous and attention-grabbing (Denning, 2001). The most dangerous evolution for Hard-Rock cyberterrorism would be if it were to become state-sponsored, so that terrorist groups could tap into the wealth and technology prowess of sovereign governments. Some authors (Colarick, 2006; Lewis, 2002) already demonstrated that the most devastating impact of cyberterrorism would be on the American economy. Such undertaking, coupled with the long tradition of deniability of state-sponsored terrorism and the near-impossibility of identifying an

Close to the Edge: Cyberterrorism Today

223

unambiguous culprit for skilled cyberattacks, would have lasting and extremely serious consequences for the United States and other advanced economies. It may as well be that the attack on Saudi ARAMCO Oil Company of Fall 2012 to send a warning to the United States was an instance of statesponsored cyberterrorism (Perlroth, 2012). Nonetheless, it is hard to see how Iran could have pressed on to the extreme consequences (stop of the Saudi oil production, skyrocketing of price and turmoil in the global oil market) without suffering serious setbacks that, ultimately, may harm Iran more than the United States.7 As any computer security expert would say, “people are the weak link” of cyber-defense because even advanced users tend to bypass security policy recommendations for greater speed and ease in their work. If terrorist cyberattacks might yet not to be that good for causing direct physical damage, they are excellent to strike the will and “hearts and minds” of the enemy, or, in other words, for “Lite cyberterrorism.”

CYBERTERRORISM “LITE” In the introduction to the 2001 edition of A History of Terrorism, Walter Laqueur (2001, p. xi) wrote that “terrorism was always, to a large extent, about public relations and propaganda.” Modern technologies have certainly put “unprecedented destructive power” (Laqueur, 2001, p. xi) in the hands of terrorist groups. More than the actual number of victims, however, what makes a terrorist attack successful seems to be the media coverage (Huddy, Feldman, Lahav, & Taber, 2003). The spectacular acts of violence, Ranum (2004, p. 115) noted, are “to show the people that the government is helpless to protect them.” September 11 became 9/11 because it was broadcast live. Hence, more than the material damage, what counts for terrorists is the “damage to the minds.” A 2004 BBC report on terrorism on the Web noted that “one of the most important weapons in al-Qaeda’s armory is its ability to generate fear. Jihadist groups in general use videos and still images of attacks
and especially gruesome attacks like beheadings
as part of psychological warfare” (Corera, 2004; see also Olimpio, 2006; Ranstorp, 2007; Santucci, 2008). Hence, for terrorists, whether cyber or real, the first “battlefield” is the mind of people, victims, suicide bombers, and political leaders. As Schweitzer and Shay (2003, p. 221) note, “terror is psychological warfare.”8 Propaganda and perception management, now widely used by cyberterrorists,

224

GIAMPIERO GIACOMELLO

are ideal tools to increase that sense of threat and stress, particularly if the message to be conveyed is that it is the cultural identity of the attacked that they want to annihilate. Under these circumstances, the role of media as amplifier of terror is crucial (Cassidy, 2009
2010; Helms et al., 2012; Huddy et al., 2003; Merkin, 2004; Norris, Kern, & Just, 2003; Santucci, 2008).9 “Islamic fundamentalists,” mainly but not exclusively, have mastered the techniques to take advantage of the Web and the Internet so as to spread their violent message. The increasing relevance of social networks has only strengthened this attitude. Democratic governments not only have been incapable of stopping them (though ability to contrast them vary between countries), but also have appeared to be on the defensive psychologically, politically and, not least, technically (Weimann, 2004). Psychological operations (including perception management or old “propaganda”) are a critical component of IO (Brown, 2003; Cassidy, 2009
2010; Fulbright, 1971; Lennon, 2003). Despite the success of IO (especially PSYOPS) in Vietnam and, later on, in Panama in 1989 and in the 1991 Gulf War,10 American public has never felt really comfortable with the notion of using “propaganda” (Waller, 1994; Wiest, 2002). For the same reasons, Americans have long disliked psychological warfare. In fact, Harris (1967) called it the “Un-American Weapon.”11 Unsurprisingly, the U.S. federal government and its professional soldiers tend to think along the same line. If it had not been proved that IO/PSYOPS indeed have an impact on military operations, the U.S. military would have long done without IO/PSYOPS. Nevertheless, if in wartime, American (and Western) public opinion tolerates IO/PSYOPS, the latter strongly dislikes the use of such tool in peacetime. Cyber-terrorists Lite see such moral aversion as a weakness to be exploited. By 2001, despite the already long experience with terrorism (Laqueur, 2001), another limit of the American/Western approach to counterterrorism emerged, that IO on computer networks (especially perception management and PSYOPS) would be undertaken overwhelmingly by foreign governments, and only as an afterthought by terrorists groups. It was a little puzzling that the societies that invented and made marketing, exit polls and “spin doctors” ubiquitous turned out to be rather slow in accepting that terrorists might soon be waging a “war of ideas” on the Net (Cassidy, 2009
2010; Colarik & Janczewski, 2008; Neumann, 2004; Norris et al., 2003; Steinberg, 2004). A clear consequence of such state of affairs was that, for some time, al-Qaeda’s and Islamic fundamentalism’s war of ideas against the West appeared to be quite successful (Brown, 2003; Brooks, 2004).12 On the

Close to the Edge: Cyberterrorism Today

225

New York Times Magazine, Daphne Merkin (2004, p. 13) wrote that “we have been glued to the Internet throughout this long summer of beheading and terrorist scares.” Americans were puzzled and confused by the way the federal government issues terrorist warnings and do not know how to respond. The message was somehow confusing: (“be alert, but go on with your lives”) and did not contribute to a more sober assessment of what terrorism (and cyberterrorism as well) really is. The 2003 Iraq war (and the subsequent occupation) and the 2001 Afghan insurgency seemed to become a catalyst for the war of ideas on the Internet between Western democracy and Islamic fundamentalism. It was a type of warfare close to what Arquilla and Ronfeldt (1997, 2001) called “netwars.” Arquilla and Ronfeldt (1997, p. 275) defined netwars “increasingly at the societal end,” that is, where operations other than war (OOTW) are located, pointing out that while “cyberwar” would be waged by “formal military forces,” netwar is more likely to involve “nonstate, paramilitary, and other irregular forces.” According to Arquilla and Ronfeldt, netwars are about “who knows what, when, where, and why, and about how secure a society” (Arquilla & Ronfeldt, 1997, p. 275). Their definition, however, should have also included perception management, as “to appear to know” is as important as “to know” (and often, more important). In fact, their work addresses extensively many possible facets of low-intensity operations conducted by network-centric groups, but does not examine cyberwarfare and cyberterrorism specifically. Based on a broad definition of cyberterrorism as the “unlawful destruction or disruption of digital property” (Nelson et al., 1999, pp. 9, 10), Table 1 outlays five typologies of terrorist groups that may have interests in pursuing cyberterrorism. Accordingly, Arquilla and Ronfeldt’s netwar campaigns per se would hardly cause any victim or physical damage or even economic and financial disruption but they would well fit the type (a) of cyberterrorism indicated in Table 1. The war on terror and the subsequent wars in Afghanistan and Iraq have provided a formidable basis to “globalize” the sort of clash of civilization that fundamentalists hope for. Religion could be a potent motivation for cyberterrorism. Nelson et al. (1999, p. 58) estimated that “cyberterror use by the religious terrorist would be ideologically consistent with regard to complex-coordinated cyberterror attacks against targets which may cause mass disruptions.” In fact, five years after the study, religious terrorists were hardly engaged in those complex, coordinated attacks causing mass disruption. On the contrary, they had invested their “meager” (compared to those of the U.S. federal

226

GIAMPIERO GIACOMELLO

Table 1.

Typology of Targets and Damage by Cyberterrorism.

Type

Target

(a) Perception Image and selfmanagement/ confidence of the propaganda enemy (b) Web defacement/taking Private companies, over credit card and banks, stock banking accounts exchange, etc. (c) “Breaking things and Air control/power killing people” and water distribution systems etc.

Damage

Fatalities Personnel Requireda

Psychological

None

One to three

Economic/ psychological

None

Two to several

Human/physical/ economic and psychological

Yes

Several

a

Estimated. Source: Giacomello (2003, 2004).

government or of most OECD countries) resources in activities that the study termed “support to cyberterrorism.” Perception management, propaganda and PSYOPS might “only” be supporting actions to destruction or disruption, but, as of 2012, terrorists have amply exploited the Web to promote their ideas and views. Another possible classification is summarized in Table 1, which outlines the characteristics of three main types of cyberterrorism, based on the nature of targets that terrorists would like to strike and the kind of damage they hope to provoke. As it is often the case with the Internet, the same tools may have different applications.13 For example, Web sites that demonize the adversary try to undermine the latter’s will to fight, but at the same time they convey a powerful message that helps to attract new recruits. Seizing the control of bank accounts may have the ultimate goal of wrecking the reputation of the bank itself, but it could also be done to deposit money that needs to be laundered somewhere else. Other authors (e.g., Schneier, 2000, 2003) have used the propensity to risk of the attacker for classification, with terrorists as bold as the “infowarriors” (uniformed, government employees, trained in IO). Despite the scary prose displayed by authors like Vernon (2003), realworld examples of cyberterrorism remain elusive (Helms et al., 2012). For instance, Iraqi insurgents and al-Qaeda terrorist never attempted at remotely compromising SCADA systems. If, as defined by the U.S. Department of the Army (2003), cyberterrorism implies unauthorized penetration of

Close to the Edge: Cyberterrorism Today

227

government or private sector computer networks, then alleged cyberterrorist “actions” by al-Qaeda have been, thus far, more about mobilization, spreading propaganda, and communication rather than cyberattacks proper (Conway, 2002; Egan, 2004; McGuire, 2004; Ranstorp, 2007; Weimann, 2004). On the contrary, while the terrorist organization has changed and adapted to become less spectacular but more elusive, it has also become more unpredictable (Shane, 2010; The Economist, 2004). What it lost in the post-9/11 phase, it now is trying to make it up with ensuring that the message of terror still goes out and reach as far as possible. The ideology of al-Qaeda has thrived and spread. Iraqi insurgents have revised their structure accordingly: many have developed a “media” section in addition to their “military” branch (Olimpio, 2006; Santucci, 2008; Sarzanini, 2004). The former prepares texts and videos and make certain that they reach the media and are posted on the Web; there are now even specialized “production companies” that guaranteed a high-level outcome.14 The IO strategy of al-Qaeda and of Iraqi and Afghan insurgents has thus remained consistent in targeting audiences. It was only a matter of time that the Web and social media, because of their availability, user-friendliness and cost, became central in that strategy. The psychological blow to the morale and cohesion of Western countries of well-planned kidnappings15 was amplified via Web sites (Lipton & Lichtblau, 2004; Santucci, 2008; Steinberg, 2004). Some of the Web sites terrorists use function as “disposable cyberspace” (Cha, 2004), which are viewable only for limited time and hosting organizations are unaware of the content and, often, cannot held liable for it (virtually all advanced democracies have laws that “protect” Internet providers in this respect). While it is not always possible to exactly pinpoint who is behind these Web sites,16 their message is echoed in other media (Egan, 2004; Olimpio, 2006; Santucci, 2008). A few of the “permanent” Web sites17 host chat and discussion groups where it is possible to notice that these actions do have an impact on some sectors of the Arab public (Steinberg, 2004; Zecchinelli, 2004). At times, advanced democracies appear that they ceded the initiative of deciding what is news and what to their terrorist foes (Harmon, 2004; Steinberg, 2004), but such impression is incorrect. In the post-9/11 decade, it seemed that terrorism had become a “state of mind” (Dowd, 2004) or “brand name,” as al-Qaeda (Johnston & Sanger, 2004). Moreover, some terrorists have adopted “network-centric” organization: they have no center of gravity and are less likely to be sponsored by a sovereign government (unlike what happened in the 1970s and 1980s).

228

GIAMPIERO GIACOMELLO

As Arquilla and Ronfeldt (1997 and 2001) noted, state actors, with their centralized bureaucracies and institutionalized militaries, are woefully inadequate to tackle network-centric groups. Cyberterrorism have become a sort of network-centric “dark force” because average computer users tend to attribute “arcane” and “mysterious” power to computers (and their networks).18 Terrorists can be chased and killed (as it happened with the most famous of them) and states are still the most powerful adversaries with their ability to gather resources and concentrate them at focal points. Likewise, only states, with their impressive bureaucracy and military machineries, could have designed and developed Stuxnet and Flame. The last point worth stressing here is that, along with IO, terrorists rely on computer networks for teaching tactics and disseminating operational knowledge (such as making bombs and so on). Inevitably, these activities tend to be rather theoretical, much as learning any other skill or technique exclusively online, without instructors and teachers ready to correct mistakes or misunderstandings. Logically, a scant or defective familiarity with the chemicals necessary to build a bomb or how to shoot properly can have serious consequences for the amateurs and could never make an individual a trained terrorist (Stewart, 2011). More importantly, the existence of dedicated Web sites and “open sources” allow several counter-terrorist agencies to monitor terrorists’ activities (online and offline), gather intelligence and facilitate infiltration, at least to some extents. Thus, once again, computer networks are a double-edged sword for the terrorists and their adversaries.

CONCLUSIONS: WINNING THE HEARTS AND MINDS In a reveling article (Singer, 2011) for the Council of Foreign Relations, Peter Singer observed that cyberterrorism discourse is like the Discovery Channel’s “Shark Week,” where those obsessed with sharks learned that it is much more likely (about 15,000 times) to be hurt or killed by a toilette accident than by a real shark. Likewise, compared to the huge number (about 31,300) of articles on cyberterrorism of various types, in 2011, the number of actual cyberterrorist attacks was the equivalent of zero (Singer, 2011). The situation, as this chapter argues, has not changed since and it is not likely to change in the near future, despite claims to the contrary (Molfino, 2012).

Close to the Edge: Cyberterrorism Today

229

Today all advanced and developing economies pour money in strengthening cyber-defenses. At the same time, modern terrorists not only have not disrupted computer networks, but also have learned how to use them (and other digital technologies) to make sure that their messages and declaration reached a wider audience and were reported, on a cascade effect, on Westerns media. Under these circumstances, it is likely that terrorist organizations will continue to refine their skills to spread their messages on the Web in their attempt to win the hearts and minds. Although nobody can say the final word about how cyberterrorism will look like in the future, one of the main conclusions of this chapter is that, when it comes to computer networks and the war on terror, advanced democracies have concentrated on the wrong menace. The United States and also Europe, Japan and other advanced democracies have focused on the material capabilities of other state actors, Russia and China, in primis. Even when considering terrorism the law enforcement and intelligence agencies of those countries emphasized material damage and searched for those terrorist groups that might seize the capability to strike. Once those capabilities are acquired, fundamentalist terror groups would use computer networks to penetrate the national information infrastructures and cause planes to crash into each other or leave entire cities in pitch darkness. Not only these nightmare scenarios have not occurred (and will not occur for still some time), but, because democracies have put their resources and attention to the wrong direction, in the meantime, Islamic fundamentalists have also exploited the Internet at the best, spreading their message of hatred. An alternative explanation is that the political structure of actors involved is only marginally important. More important is the nature of the actors. Perhaps, state-actors, with their organized bureaucracies, simply do not cope well with non-state, network-based actors. This situation should not be surprising. Neumann (2004, p. A15) argued that nobody saw 9/11 coming, because “10 years after, a cold-war mentality was still in place” in the intelligence community. Then National Security Advisor Condoleezza Rice stated that, in coming into office in January 2001, she “emerged from a generation of scholars trained in great power politics, with terrorism seen as a troubling but subordinate element” (Jehl & Sanger, 2004, p. A1). When searching for an enemy, the “natural” reaction for a state is to look for another state, whose armed forces and government buildings can be bombed and destroyed. These are actions that the public can grasp and that an audience can watch on news channels. The overall government

230

GIAMPIERO GIACOMELLO

approach (“only a nation-state can threaten another state: terrorists are just means in the hands of nation-states”) thus persuades the public itself that the only way to measure success and victory in war is the control of the enemy’s territory, or the number of prisoners or of captured colors. Indeed, Hard-Rock cyberterrorism will be a serious occurrence only if it will be state-sponsored, because only governments have the necessary knowledge and funding to undertake that (Stuxnet docet). In other cases, for still quite some time, the enterprise would not be cost-efficient and convenient to terrorists. Lite cyberterrorism will probably remains as it is now, with a growing number of supporters increasingly apt at computer skills to wage their IO, but without the sufficient technical skills and “deep pockets” to switch to the Hard-Rock version. There is certainly the possibility that powerful cyber-weapons like Stuxnet and Flame could be developed by states and then turned to “independent” terrorist organizations so that they could be employed against an adversary without compromising the original developers. But if a government is funded and technically advanced enough to create those weapons, it is also quite capable of “covering its tracks,” if it decided to employ them in the first place. Moreover, much like with nuclear or biological weapons, passing them to autonomous terrorists always implies the risk that the latter may use them without consent or knowledge of the sponsoring government, with embarrassing consequences. Incidentally, despite disturbing forecasts that the majority of online threat would be crime-related rather than terrorism-related (Ghosh & Turrini, 2010; Menn, 2010; Schneier, 2003, 2004), the worst scenarios of such predictions also failed to materialize (Dinei & Herley, 2012; Singer, 2011). But cybercrime assessments are definitively more likely approximate reality than forecasts of Hard-Rock cyberterrorism will ever be. Cyberterror as the main cause for physical destruction and fatalities will then remain highly unlikely, at least for the foreseeable future. Finally, it could be asked whether it would make sense for states to engage in cyberterrorism: yes, but to what purpose? If they have cyberwarfare capabilities it would be a waste of time; if they don’t, why bother? It would be counterproductive if the targeted adversary is a capable cyber-power, because it probably had strong defenses and highly effective cyber and conventions arsenals for retaliation. On the other hand, if the adversary is technologically backward, not heavily dependent on information infrastructures and computer networks, cyberterrorism (much as cyberwarfare) would be ineffective and thus irrelevant. This is one of

Close to the Edge: Cyberterrorism Today

231

the cardinal issues of security in cyberspace about which future research should focus.

NOTES 1. The definition of critical infrastructures
that is those essential services for the well-being and functioning of a modern society
varies to some degree, depending on the country considered. But a valid list would have to include water and power distribution, banking and finance systems, air traffic control, transportation and all the forms of communication. Moreover, all these services rely on computers and computer networks for their daily activities. For a remarkable survey of national infrastructures see Brunner and Sutter (2008). 2. See the PowerPoint presentation (2002) by Tim Shimeall of the Software Engineering Institute at the Carnegie Mellon, which has long operated as the coordination center for various American CERT teams, downloadable from www.cert. org/archive/ppt/cyberterror.ppt. Last accessed December 15, 2012. 3. “SCADA (supervisory control and data acquisition) is a category of software application program for process control, the gathering of data in real time from remote locations in order to control equipment and conditions. SCADA is used in power plants as well as in oil and gas refining, telecommunications, transportation, and water and waste control. SCADA systems include hardware and software components. The hardware gathers and feeds data into a computer that has SCADA software installed. The computer then processes this data and presents it in a timely manner. SCADA also records and logs all events into a file stored on a hard disk or sends them to a printer. SCADA warns when conditions become hazardous by sounding alarms.” This definition by Whatis. com can be retrieved from http://searchcio-midmarket.techtarget.com/definition/ SCADA. 4. As it happened in the case of the Russian-Georgian war of 2008 (Markoff, 2008). When information warfare appeared with the 1991 Gulf War, military professionals quickly included “perception management” (i.e. “what others think”) in what they saw would be a new way of war-fighting. Later on, as the term information warfare entered the media jargon, the U.S. military switched to information operations (IO). IO involves “actions taken to affect adversary information and information systems while defending one’s own information and information systems” (Joint Command, Control and Information Warfare School, Joint Forces Staff College, 2003, p. I-1). 5. For two divergent interpretations of “Eligible Receiver” see George Smith in The Crypt Newsletter, 1999 available from www.soci.niu.edu/∼crypt/other/eligib. htm and Globalsecurity.org (no date given) “Eligible Receiver,” available from www.globalsecurity.org/military/ops/eligible-receiver.htm 6. In the case of the United States and Canada, the outage affected 50 million people. In Italy, virtually the whole country was without electricity for 24 hours. Economic damages were serious (in Italy the event took place on a Sunday

232

GIAMPIERO GIACOMELLO

morning, however, which reduced its economic impact). In both instances very few people were hurt, which, considering the dimension of the two events, is unquestionably surprising. 7. It is the same logic that rules out any serious thought of cyber-guerrilla against the U.S. economy by the Chinese, as their economy would terribly suffer from a collapse of the American economy. It is clear that Chinese leaders would accept to crash the U.S. economy only if they were on the verge of initiating a larger armed conflict with the United States (and perhaps not even in that case, given the overlapping of the two economies). 8. It is somehow ironic that the United States wants to fight such war primarily with such a Soviet-style institution like the Pentagon, not known for its independent and flexible thinking. 9. During World War II, both Hitler and Stalin understood and appreciated the psychological implications of propaganda. Communists in many countries were apt at applying the Soviet lesson, as the French in Indo-China, for example, had to learn. As Nye (2003, p. 232) noted, “Milosevic’s control of television was crucial to his power in Serbia in the 1990s. In Moscow in 1993, a battle for power was fought at a TV station. In Rwanda in 1994, the Hutu people controlled radio station that incited genocide against the Tutsi minority.” The importance of broadcasting is widely recognized, but, as Nye concludes, “it will be increasingly supplemented by the Internet, with its multiple channel of communication, controlled by multiple actors who cannot use force to control each other.” After the experience of World War II and Korea, the United States began to develop a structured approach to psychological operations that became the politically correct term for “propaganda” in Vietnam (Waller, 1994; Wiest, 2002). On this point, it is also useful to consult Herman (1997). 10. Under certain conditions, information operation and PSYOPS may even spare lives. 11. Attempts at influencing the U.S. public opinion by the Pentagon, however, continued in the 1960s and 1970s as U.S. Senator J. William Fulbright (1971) noted. It is a fair guess that such efforts have never actually stopped. 12. Admittedly, such psychological attacks found receptive and predisposed ears in the United States and Europe; see Herman (1997). 13. Multiple other categorizations are possible. Here, we have not included the cases in which the Internet is used for recruiting, data mining, and actual communications (Weimann, 2004). 14. The best known is “Al Sahab” (Olimpio, 2006). 15. The description of the Web site of the “Islamic Army” (one of the many terrorist groups operating in Iraq seizing and killing hostages) by Italian journalists (Rampoldi, 2004; Zecchinelli, 2004) conveys the astonishment and psychological shock that average tolerant individuals may experience before such displays of raw horror. Rampoldi portrays the “mix of sinister technological modernity and primordial religiosity for which the Internet becomes the altar for sacrifices.” Such reaction is the ultimate success of “terrorism.” It is the perfect fulfillment of the term “terror”. 16. Intelligence agencies, however, have learned to become much more proficient in this business, by learning to exploit the Web and social networks for their datagathering.

Close to the Edge: Cyberterrorism Today

233

17. See, for example, www.ansarnet.ws/ 18. The whole Terminator saga is a clear illustration of this point, where computers with advanced artificial intelligence would take over.

REFERENCES Anderson, K. (2002). US “fears al-Qaeda hack attack”. BBC News, June 27. Retrieved from http://news.bbc.co.uk/1/hi/sci/tech/2070706.stm Arena, K., & Ensor, D. (2002). U.S. infrastructure information found on al-Qaeda computers. CNN.com, June 27. Retrieved from http://edition.cnn.com/2002/US/06/27/alqaeda.cyber. threat/index.html Arquilla, J., & Ronfeldt, D. (1997). The advent of netwar. In J. Arquilla, & D. Ronfeldt (Eds.), In Athena’s camp: Preparing for conflict in the information age. MR-880-OSD. Santa Monica, CA: Rand. Retrieved from www.rand.org/publications/MR/MR880/MR880.ch12.pdf Arquilla, J., & Ronfeldt, D. (2001). The advent of netwar (revisited). In J. Arquilla & D. Ronfeldt (Eds.), Networks and netwars. MR-1382-OSD. Santa Monica, CA: Rand. Retrieved from www.rand.org/publications/MR/MR1382/MR1382.ch1.pdf Broad, W. J., & Sanger, D. E. (2010). Worm was perfect for sabotaging centrifuges. The New York Times, November 19, p. A1. Brooks, D. (2004). War of ideology. The New York Times, July 24, p. A13. Brown, R. (2003). Spinning the war: Political communications, information operations and public diplomacy in the war on terror. In D. K. Thussu & D. Freedman (Eds.), War and the media: Reporting conflict 24/7 (pp. 87
100). Thousand Oaks, CA: Sage. Brunner, E. M., & Sutter, M. (2008). International CIIP handbook 2008/2009. Zurich: Center for Security Studies, ETH Zurich. Cassidy, R. M. (2009
2010). War in the information age (review essay). Parameters, (Winter), 117
122. Cha, A. E. (2004). From a virtual shadow, message of terror. The Washington Post, October 2, p. A01. Colarik, A. M., & Janczewski, L. J. (2008). Introduction to cyber warfare and cyber terrorism. In L. J. Janczewski & A. M. Colarik (Eds.), Cyber warfare and cyber terrorism (pp. xiii
xxx). Hershey, PA: Information Science Reference. Colarick, A. M. (2006). Cyberterrorism: Political and economic implications. Hersey, PA: Idea. Conway, M. (2002). Reality bytes: Cyberterrorism and terrorist ‘use’ of the internet. First Monday, 7(11). Retrieved from http://doras.dcu.ie/498/1/first_mon_7_11_2002.pdf Corera, G. (2004). A web wise terror network. BBC News. October 6. Retrieved from http:// news.bbc.co.uk/1/hi/world/3716908.stm Denning, D. (2001). Is cyber terror next? “After September 11”. Social Science Research Council, New York, NY. 1 November. Retrieved from www.ssrc.org/sept11/essays/ denning.htm Department of the Army. (2003). Information operations: Doctrine, tactics, techniques, and procedures. Field Manual 3
13. Washington, DC. November. Retrieved from www.iwar. org.uk/iwar/resources/doctrine/fm-3-13.pdf Dinei, F., & Herley, C. (2012). The cybercrime wave that wasn’t. The New York Times, April 15, p. SR5.

234

GIAMPIERO GIACOMELLO

Dowd, M. (2004). Spinning our safety. The New York Times, July 25, p. 11. Egan, T. (2004). Computer student on trial for aid to Muslim web sites. The New York Times, April 27, p. A16. Embar-Seddon, A. (2002). Cyberterrorism: Are we under siege? American Behavioral Scientist, 45(6), 1033
1043. Fulbright, J. W. (1971). The Pentagon propaganda machine. New York, NY: Vintage Books. Giacomello, G. (2003, October). Measuring digital wars: Learning from the experience of peace research and arms control. Infocon Magazine. Issue One. Retrieved from http://7.iwar.org.uk/infocon/measuring-io.pdf Giacomello, G. (2004). Bangs for the buck: A cost-benefit analysis of cyberterrorism. Studies in Conflict and Terrorism, 27(5), 387
408. Green, J. (2002). The myth of cyberterrorism. Washington Monthly. November. Retrieved from www.washingtonmonthly.com/features/2001/0211.green.html Ghosh, S., & Turrini, E. (Eds.). (2010). Cybercrimes: A multidisciplinary analysis. Heidelberg: Springer. Harmon, A. (2004). New technology loosens controls over images of war. The New York Times, May 14, p. A12. Harris, E. (1967). The “un-American” weapon: Psychological warfare. New York, NY: M. W. Lads. Helms, R., Costanza, S., & Johnson, N. (2012). Crouching tiger or phantom dragon? Examining the discourse on global cyber-terror. Security Journal, 25, 57
75. Herman, A. (1997). The idea of decline in western history. New York, NY: The Free Press. Huddy, L., Feldman, S., Lahav, G., & Taber, C. (2003). Fear and terrorism: Psychological reactions. In P. Norris, M. Kern, & M. Just (Eds.), Framing terrorism: The news media, the government, the public (pp. 256
278). New York: Routledge. Hughes, G. (2004). Tech threats: The new front in the war on terror. CBC News Viewpoint, September 1. Retrieved from www.cbc.ca/news/viewpoint/vp_hughes/20040901.html Jehl, D., & Sanger, D. E. (2004). New to the job, Rice focused on more traditional threats. The New York Times, April 5, pp. A1, A16. Joint Command, Control and Information Warfare School, Joint Forces Staff College. (2003, July). Joint information operations planning handbook. Norfolk, VA: National Defense University. Retrieved from http://mail.iwar.org.uk/iwar/resources/jiopc/io-handbook-2003. pdf Johnston, D., & Sanger, D. E. (2004). New generation of leaders is emerging for al Qaeda. The New York Times, August 10, p. A1. Laqueur, W. (2001). A history of terrorism: With a new introduction by the author. New Brunswick, NJ: Transaction (originally published in 1977). Lederer, E. (2004). UN calculates the cost of terrorism. The Age, August 27. Retrieved from www.theage.com.au/articles/2004/08/27/1093518081060.html?from = storylhs&oneclick = true Lennon, A. (Ed.). (2003). The battle for the hearts and minds. Cambridge, MA: MIT Press. Lewis, J. (2002, December). Assessing the risk of cyber terrorism, cyber war and other cyber threats. Washington, DC: Center for Strategic and International Studies. Retrieved from www.csis.org/tech/0211_lewis.pdf Libicki, M. C. (2007). Conquest in cyberspace: National security and information warfare. New York, NY: Cambridge University Press. Lipton, E., & Lichtblau, E. (2004). Online and even near home, a new front is opening in the global terror. The New York Times, September 23, p. A12.

Close to the Edge: Cyberterrorism Today

235

Markoff, J. (2008). Before the gunfire, cyberattacks. The New York Times, August 13. Retrieved from www.nytimes.com/2008/08/13/technology/13cyber.html?hp = &p McGuire, D. (2004). Al Qaeda messages post on U.S. server. Washington Post. June 13. Retrieved from www.washingtonpost.com/ac2/wp-dyn/A47681-2004Jul13?language = printer Menn, J. (2010). Fatal system error: The hunt for the new crime lords who are bringing down the internet. New York, NY: Perseus Books. Merkin, D. (2004). Terror-filled. The New York Times Magazine, August 15, pp. 13, 14. Molfino, E. (2012). Viewpoint: Cyberterrorism: Cyber pearl harbor is imminent. In S. Costigan & J. Perry (Eds.), Cyberspaces and global affairs (pp. 75
92). Farnham: Ashgate. Nelson, B., Rodney C., Iacobucci, M., Mitchell, M., & Gagnon, G. (1999, October). Cyberterror: Prospects and implications. White Paper. Center for the Study of Terrorism and Irregular Warfare, Monterey, CA. Retrieved from www.nps.navy.mil/ctiw/files/Cyberterror %20Prospects%20and%20Implications.pdf Neumann, P. R. (2004). Why nobody saw 9/11 coming. The New York Times, March 27, p. A15. Norris, P., Kern, M., & Just, M. (Eds.). (2003). Framing terrorism: The news media, the government, the public. New York: Routledge. Nye, J. (2003). Understanding international conflict: An introduction to theory and history (4th ed.), New York, NY: Longman. Olimpio, G. (2006). Cameramen, Corrieri, Studi: Ecco la Propaganda del Califfo. Il Corriere della Sera, July 1, p. 13. Perlroth, N. (2012). Cyberattack on Saudi firm disquiets U.S. The New York Times, October 24, p. A1. Rampoldi, G. (2004). L’Assassinio di Baldoni: L’ultimo orrore in rete. La Repubblica. September 9. Retrieved from www.repubblica.it/2004/i/sezioni/esteri/itarap/rampbald/ rampbald.html Ranstorp, M. (2007). The virtual sanctuary of al-Qaeda and terrorism in an age of globalization. In J. Eriksson & G. Giacomello (Eds.), International relations and security in the digital age (pp. 31
56). London: Routledge. Ranum, M. (2004). The myth of homeland security. Indianapolis, IN: Wiley. Sanger, D. E. (2010). Iran fights malware attacking computers. The New York Times, September 26, p. A4. Santucci, G. (2008). A Scuola di Terrorismo. Online. Il Corriere della Sera, June 10, pp. 8, 9. Sarzanini, F. (2004). Individuati Dieci Membri della Banda di Rapitori. Il Corriere della Sera, August 29, p. 3. Shane, S. (2010). Qaeda branch aimed for broad damage at low cost. The New York Times, November 21, p. A16. Shane, S. (2012). Cyberwarfare emerges from shadows for public discussion by U.S. officials. The New York Times, September 27, p. A10. Schneier, B. (2000). Secrets and lies: Digital security in a networked world. New York, NY: Wiley. Schneier, B. (2003). Beyond fear: Thinking sensibly about security in an uncertain world. New York, NY: Copernicus Book. Schneier, B. (2004). Secrets and lies: Digital security in a networked world. New York, NY: Wiley. Schweitzer, Y., & Shay, S. (2003). The globalization of terror: The challenge of Al Qaeda and the response of the international community. New Brunswick, NJ: Transaction Publishers.

236

GIAMPIERO GIACOMELLO

Singer, P. W. (2011, November). The cyber terror bogeyman. Research Article. The Brookings Institution. Retrieved from www.brookings.edu/research/articles/2012/11/cyber-terror-singer Steinberg, J. (2004). Kidnappings, beheadings and deciding what’s news. The New York Times (Week in Review), August 1, pp. 1
3. Stewart, S. (2011). How to tell if your neighbor is a bomb maker. Stratfor Security Weekly, April 7. Retrieved from www.stratfor.com/weekly/20110406-how-tell-if-your-neighborbombmaker Stohl, M. (2006). Cyber terrorism: A clear and present danger, the sum of all fears, breaking point or patriot games? Crime, Law and Social Change, 46(4
5), 223
238. The Economist. (2004). Plots, alarms and arrests, August 14, pp. 22
24. Thomas, T. (2003). Al Qaeda and the internet: The danger of “cyberplanning”. Parameters, (Spring), 112
123. Vernon, D. (2003). Black ice: The invisible threat of cyber-terrorism. New York, NY: McGrawHill. Vittori, J. (2011). Terrorist financing and resourcing. New York, NY: Palgrave Macmillan. Waller, D. (1994). The commandos. New York, NY: Simon & Schuster. Weimann, G. (2004, March). Www.terror.net: How modern terrorism uses the internet. Special Report. United States Institute for Peace, Washington, DC. Wiest, A. (2002). The Vietnam war 1965
1975. London: Osprey. Zannini, M., & Edwards, S. J. A. (2001). The networking of terror in the information age. In J. Arquilla & D. Ronfeldt (Eds.), Networks and netwars. MR-1382-OSD. Santa Monica, CA: Rand. Retrieved from www.rand.org/publications/MR/MR1382/MR1382.ch2.pdf Zecchinelli, C. (2004). Gli Arabi sul Web “Sgozzateli!”. “No, Loro sono Diversi”. Il Corriere della Sera, August 31, p. 3.

CHAPTER 11 SOME INSIGHTS ON THE LINK BETWEEN TERRORISM, ORGANISED CRIME AND ‘NEW WARS’ Carla Monteleone, Raul Caruso and Andrea Locatelli ABSTRACT Purpose
The purpose of this chapter is to investigate on the relationship between terrorism, criminal organisations, and the so-called ‘new wars’. Methodology/approach
Review of the existing literature on the ‘new wars’ and on the terrorism
crime nexus; estimate of the financial revenues gained by terrorist groups engaged in illicit activities. Findings
Terrorist and criminal groups can develop several forms of collaboration. Whether terrorists convert to criminal activities or not depends on a variety of factors, both internal and external. In some cases these links are occasional and opportunistic, associated with the possibility of exploiting the availability of specialised competences in a stateless

Understanding Terrorism: A Socio-Economic Perspective Contributions to Conflict Management, Peace Economics and Development, Volume 22, 237
254 Copyright r 2014 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 1572-8323/doi:10.1108/S1572-8323(2014)0000022016

237

238

CARLA MONTELEONE ET AL.

environment, in other cases the warring parties are directly involved and create stable relationships as a way to finance their war activities. Research limitations/implications
Lack of reliable data limits a comprehensive analysis. Nevertheless, the study has important consequences for designing and adopting more effective policies regarding terrorism, organised crime, as well as conflict resolution. Originality/value of the chapter
The study presents an overview of different strands of the literature on terrorism; a discussion is articulated on the interplay between ‘new wars’ and the terror
crime connection. Potentially fruitful avenues of investigation are suggested for future research. Keywords: Terrorism; organised crime; armed conflicts; non-state actors.

INTRODUCTION The link between terrorism and organised crime has long caught the attention of governments and international institutions as an international security issue. This is witnessed by a series of initiatives, like the Political Declaration of the 1994 Naples World Ministerial Conference on Transnational Organized Crime, the concern ‘at the growing and dangerous links between terrorist groups and drug traffickers and their paramilitary gangs’ expressed by the UN General Assembly in the Declaration of Measures to Eliminate International Terrorism annexed to resolution 49/60 in 1994, and again the 1995 resolution 50/186. All these indicate a slow but progressive securitisation of the issue (see also Mylonaki, 2002). After 11 September 2001, the UN Security Council with resolution 1373 considered the phenomenon a ‘serious challenge and threat to international security’. The link between terrorism and organised crime is so strong that, in 2008, 19 out of the 44 foreign terrorist organisations (FTOs) designated by the US State Department had been involved in the business of illegal drugs (Luna, 2008). In particular, drug trafficking seems to be the criminal activity of choice for most terrorist groups: according to the US Drug Enforcement Administration (DEA), in 2011, 19 out of 49 FTOs ‘confirmed links to the drug trade’ (quoted in Rollins & Wyler, 2013, p. 3). Many of those are rebel or fighting parties that have entered the list of

Some Insights on Link Between Terrorism, Organised Crime and ‘New Wars’

239

terrorist organisations of the United States, the EU and Canada, and have become involved with drug trafficking over time in order to fund their activities. While the Revolutionary Armed Forces of Colombia (FARC) involved in the Colombian armed conflict is probably the most renowned case, the PKK in Turkey, Hizbollah in Lebanon, Shining Path in Peru, the AUC in Colombia, Abu Sayyaf in the Philippines, or the IMU in Uzbekistan are other well-known instances. Whether and how much their activities are typical of a rebel organisation, a terrorist organisation or a criminal organisation is a matter of debate, but such a remarkable number of examples seems to point towards a facilitating environment played by armed conflict situations in creating a link between organisations of a different type (Monteleone, 2010). Despite the growing attention to the linkage between terrorism, organised crime and armed conflict, so far only partial explanations are available. Literature on armed conflicts has highlighted important changes in dynamics, actors and features that have increasingly pointed in the direction of a blurring of the dividing lines between rebel groups and terrorist and criminal organisations, especially since the end of the Cold War (Holsti, 1996; Kaldor, 1999; Keen, 1998; Naylor, 1993). Nevertheless, causal relations are still rather unclear. Literature on the relationship between terrorism and organised crime has pointed out a number of links between terrorist organisations and criminal organisations and their potential transformations, but it has mostly relied on anecdotes and case studies, and has strongly divided scholars and analysts as to whether the nexus between the two types of organisations is a growing or impossible one (among others, Dishman, 2005; Hutchinson & O’Malley, 2007; Makarenko, 2004; McCarthy, 2003; Napoleoni, 2003). What the above mentioned literature agrees on, however, is the possibility that contemporary armed conflict situations taking place in failed states create a favourable environment for the wielding of terrorism and organised crime. This chapter will try to get a better understanding of the terrorism-organised crime connection in armed conflict situations. In this view, the chapter is organised as follows: the next section highlights the main features of the so-called new wars
in particular, who fights, why, and how
to understand why changes in contemporary warfare have allowed criminal and terrorist organisations to be part of the game, often involved in joint practices or businesses. In order to build a comprehensive framework of the crime
terrorism nexus, the second section investigates on the factors that ease or deter cooperation, and the types of interaction that eventually blur the difference between criminal and terrorist

240

CARLA MONTELEONE ET AL.

organisations. The third section discusses the available empirical evidence on the criminal activities perpetrated by terrorists. The fourth section summarises and concludes.

CHANGES IN CONTEMPORARY ARMED CONFLICTS1 The end of the Cold War gave spur to new research paths regarding contemporary armed conflicts. Two phenomena, in particular, caught scholarly attention. The first one was the decline of interstate conflicts and the contemporary rise of intrastate conflicts, including a change in actors and dynamics. The second one was the very long duration of existing conflicts. The resilience of conflicts that were born during the Cold War and protracted after its end, in particular, gave impetus to further inquiring into the economy of armed conflicts and sources of financing (among others, Keen, 1998; Naylor, 1993). Literature concentrating on the former phenomenon focuses on the ‘new wars’, as they have become known after Kaldor (1999). These are mostly intra-state, and fought not by armies, but by actors (regular and irregular troops, cells of various kind, including terrorist groups) that do not respond to a central authority and have to fund themselves through illegal activities. Frontal clashes between the fighting groups are limited, while rebels and governmental troops alike use violence against civilians, transforming the armed conflict into a war of terror. According to Holsti (1996), this is a radical change compared to past armed conflicts that relied on the presence of centrally controlled professional armed forces capable to win over internal enemies and avoid any disruption in tax collection. Contemporary armed conflicts tend to take place in weak or failed states, in particular in those post-colonial states commonly characterised by (a) the presence of multiple centres of power that do not recognise a legitimate authority; (b) the personalisation of the state; (c) significant internal fragmentation into a number of communities (being them ethnic, religious, linguistic, based on class or other identities), with no chance of creating forms of horizontal legitimacy or social contracts (Holsti, 1996; see also Gurr, 2000; Jackson, 2007, p. 149; Migdal, 1988). Under these circumstances, black markets gain central stage, creating an alternative economy that further weakens the state, incapable of collecting taxes and therefore to offer services that could legitimate it.

Some Insights on Link Between Terrorism, Organised Crime and ‘New Wars’

241

This environment encourages opportunistic behaviours of terrorist and criminal organisations. Another strand of literature concentrates on exogenous factors related to globalisation that have an impact on contemporary armed conflicts and can favour the creation of a link between terrorist and criminal organisations. The effects of globalisation in the information, transport and finance sectors have transformed contemporary armed conflicts into netwars: more than hierarchical groups, they see the involvement of military, criminal and terrorist networks that are distinct from an organisational point of view, but also capable of communicating, coordinating and converging without a central authority (Arquilla & Ronfeld, 2001). Because of globalisation, states are weakened by their incapacity to answer efficiently to contemporary problems by acting alone, as well as by the increasing intrusiveness of international institutions (Adler, 2006; Keohane, 2006). Moreover, globalisation has increased the role of external actors. Among them, diasporas have gained an important role (Shain & Barth, 2003). Thanks to globalisation, they can easily keep links with their motherland and lobby foreign governments and international institutions. But they can also help organising illegal traffics to fund fighting groups, provide material and logistic support, and facilitate the illegal transfer of resources (Leander, 2001; Sheffer, 1994). Expatriate Lebanese communities, for instance, have long been known to be at the basis of a global commercial network of licit and illicit businesses that support Hizbollah (Rollins & Wyler, 2012, p. 19). Likewise, Irish expatriates were known for sending money for humanitarian purposes that were diverted to buy weapons for the IRA (Naylor, 1993). According to Kaldor (1999), because of all of these changes, contemporary armed conflicts are different from conflicts that have taken place in the past. In ‘new wars’ economic aspects play an important part, since ‘a new type of globalized informal economy is established in which external flows […] are integrated into a local and regional economy based on asset transfer and extra-legal trading’ (Kaldor, 1999, p. 104). This encourages both rebels and governments to use illegal traffics
normally drugs, weapons, cigarettes, human smuggling and any other good that can be traded to fund conflicts. For this reason, rebels have an incentive to create links with criminal groups (Naylor, 1993). In a conflict economy, in the absence of a proper operating productive cycle, what takes place is the allocation of any resource into military resource through black market and illegal trafficking. However, while recognising that the described dynamics of the new wars sheds light on

242

CARLA MONTELEONE ET AL.

armed conflicts, their ‘new-ness’ has been debated (Newman, 2004), leaving scholars wondering when they really started and what causal factors are at work. One of the issues that deserve further scrutiny is why the terror
crime connection takes place. In the next session we will discuss how this usually happens, applying it to cases taking place in contemporary armed conflicts.

LINKING TERRORISM TO ORGANISED CRIME One of the main hurdles in tracing the connections between terrorist and criminal organisations is the lack of reliable information. Empirical evidence is limited, fragmented, and based on anecdotal evidence rather than on systematic research (Rollins & Wyler, 2012, p. 1). This makes generalisations very difficult. Given the complexity of the issue at stake, we will start here with a deductive investigation of the factors that might spur collaboration, and what type of relationship they produce (for an early attempt, see Gupta, Horgan, & Schmid, 2009). We will then back up these considerations with empirical evidence based on terrorists’ illicit activities. Our first question is then what makes a connection or even convergence appealing for terrorist and criminal organisations. The main answer is basically the same for both actors: by exploiting their own differences and complementarities, terrorist and criminal organisations may better serve their final goals. In other words, the main rationale to establish a connection is strategic calculation. So, as we will see in the next paragraph, terrorists may engage in illicit traffic to sponsor their cause, as well as criminal organisations may perpetrate terrorist actions to increase their financial stature (Rollins & Wyler, 2012, p. 5). Alex Schmid (1996, 2004) has identified further factors that encourage the establishment of a link. Among them, of particular relevance for terrorist organisations are (a) access to specialist skills and expertise, (b) support in smuggling people and goods, and (c) the possibility to increase their recruiting pool. For criminal organisations, incentives range from the possibility to defend their illegal businesses in areas controlled by terrorist organisations and the destabilisation of adverse political and economic structures. On the other hand, we could also list a number of cons that make such a relationship less likely. This explains why terrorists and criminals seldom

Some Insights on Link Between Terrorism, Organised Crime and ‘New Wars’

243

make business with each other, regardless of the potential benefits. The first problem derives from their diverging goals: as stressed by Schmid (2011), terrorist groups and criminal groups have different motivations,2 so they cannot really go any further than occasional collaboration. Likewise, Bjo¨rnehead (2004) sees many obstacles on a converging path, from lack of trust to lack of a common cause, and sees the two kind of organisations competing as well as cooperating. Second, by cooperating, they both raise the bar of their own action, so attracting governments’ attention. In other words, by increasing their finances or range of activities, they become more visible, so rising on governments’ security agenda. Another disincentive shared by both types of organisation is related to security concerns. For terrorist groups, joining forces with criminals may compromise internal security, as well as for criminals ties with terrorists may increase the possibility of infiltration
and ultimately lead to the capture of the leadership. This is due to the secrecy surrounding both types of activities: by establishing ties with external partners, organisations weaken their defences to police and counter-terrorism. Moving to peculiar differences in disincentives, for criminals, perpetrating acts of political violence can erode the ties created with corrupt regimes. For terrorists, ideological motivations frequently clash with the prospect of engaging in criminal activities: for this reason, as witnessed among others by the Tamil Tigers, illicit activities (like drug trafficking, gunrunning) remain a marginal part of the organisation, officially justified only on the ground of selfpreservation (Rabasa et al., 2006, pp. 103
110).3 The incentives to establish a connection also shape the kind of relationship between criminal and terrorist groups. Here the literature highlights different possible configurations. Three analyses have attracted particular attention. According to Schmid (2004), interactions between terrorism and crime range among the following positions: national criminal and terrorist groups (e.g. in Colombia, the FARC and drug cartels), national and foreign criminal organisations and terrorist groups (e.g. Colombian drug dealers and Shining Path in Peru, or the triads and paramilitary protestants in Northern Ireland), ‘sleeping’ foreign terrorists as parts of local criminal organisations (e.g. GIA members in France), foreign terrorist groups training local ones (e.g. Provisional IRA-FARC in Colombia), criminal organisations evolving into terrorist ones (e.g. Arkan’s tigers in the former Yugoslavia), and declining terrorist groups transforming into criminal organisations (e.g. Abu Sayyaf in the Philippines). Similarly, Rollins, Wyler, and Rosen (2010) identify a variety of configurations ranging from full convergence/fusion (like the D-Company in

244

CARLA MONTELEONE ET AL.

Pakistan and India) to the constitution of in-house criminal structures (this is the case of FARC and the Madrid bombers), from terrorist organisations with criminal sympathizers (like Hizbollah) to terrorist organisations with peripheral connections to criminal groups (like al-Qaeda), and from terrorist cells without major criminal involvement (like the 2005 London bombers) to the ambiguous involvement of criminal groups that use violence for political purposes (like Medellin in Colombia). According to Makarenko (2004), four different types of cooperation are possible. In particular, she posits a crime/terror continuum depending on the environment in which it operates: (a) Terrorist groups and criminal organisations can form alliances, whose nature can vary from one-off to short-term and long-term relationships. This is the case of the FARC, that has entered into alliances with criminal groups outside of Colombia. But it is also the case of the strategic relationship between the Islamic Movement of Uzbekistan and the Afghan drug mafia and Central Asian criminal groups for the smuggling of heroin4 (Makarenko, 2004, p. 131
132). (b) Terrorist organisations can also engage in criminal activities (and, vice versa, criminal organisations can use terrorism) as an operational tool, so blurring the line between the two phenomena. Examples range from the involvement of illicit drug trade of the FARC, ETA, PKK, Sendero Luminoso and Hizbollah, to involvement in fraud, counterfeiting and human smuggling of al-Qaeda (Makarenko, 2004, p. 134, 135). (c) A deeper type of connection implies ‘convergence into a single entity that initially displays characteristics of both groups simultaneously’, but it can ‘transform itself into an entity situated at the opposite end of the continuum from which it began’ (Makarenko, 2004, p. 135). Abu Sayyaf, the IMU and the FARC all enter into this category. The FARC, in particular, in the mid-1990s ‘shifted their involvement in the regional drugs trade from that of protector of crops and laboratories, to “middlemen” between farmers and illicit drug cartels. This shift directly resulted in the groups acquiring more profits from the drug trade’ (Makarenko, 2004, p. 137). (d) The final point of the continuum is the so-called ‘black hole’, which ‘refers to the situations in which weak or failed states foster the convergence between transnational organised crime and terrorism, and ultimately create a safe haven for the continued operations of convergent groups’ (Makarenko, 2004, p. 138). This means that the primary motivations of groups engaged in a civil war can evolve from a focus on

Some Insights on Link Between Terrorism, Organised Crime and ‘New Wars’

245

political aims to a focus on criminal aims, but it can also mean that a hybrid group takes over the state. Under these circumstances, civil war is promoted to secure economic and political power. Conflicts in Afghanistan, Angola, Myanmar, Sierra Leone and Tajikistan are all examples of this situation (see also Monteleone, 2010). The main difference between these modes of collaboration can be summarised as the one between ‘in-house’ development of tactics and outsourcing. Put it differently, once the organisation has decided to engage in activities beyond its original scope, what is the best strategy? One option is to develop autonomously the capabilities and skills required; the second option is to outsource the services required to other, specialised, organisations. Three basic factors intervene to shape this choice. The main explanation probably depends on the availability or lack of inhouse capabilities and expertise. Wherever these are easy to procure or develop (as, e.g., when members of terrorist groups also have ties or experiences with criminal organisations and vice versa) it is likely that the nexus will take the first form. As argued among others by Gunaratna (2001), the need to grant financial resources forces terrorist groups to adapt their organisations accordingly, and eventually recruit members among criminals. It is therefore plausible that as terrorist groups become more articulated in terms of internal division of labour, they will tend to develop in-house criminal capabilities. By reverse, less articulated groups will tend to rely on external collaboration with criminals. A second internal element is to be found into the ideology underpinning the group. As we have seen, this is particularly the case with terrorist groups that are sometimes motivated by ideals that reject criminal actions and organisations. As ideologies and ideological fervour are hardly constant, we can infer a second observable implication: whenever the group’s ideology ‘becomes less sharp, the group will tend to veer toward criminal activities’ (Gupta et al., 2009, p. 131). Moreover, as the ideology tends to fade, the group will be more prone to engage in alliances with criminal organisations. Third, the possibility to resort to partnerships of some sort is made more likely by the growing decentralisation of terrorist groups. Dishman (2005) in particular points out that, because of the breakdown of hierarchical structures in illicit organisations, terrorist and criminal cells have been forced to generate funds independently from leadership, and this has led to new opportunities to collaborate. This leaderless nexus can push terrorist and criminal organisations to external convergence and long-term alliances,

246

CARLA MONTELEONE ET AL.

as in the case of one Hizbollah cell’s alliance with Mexican drug cartels, or to hybrid organisations.

ILLICIT ACTIVITIES OF TERRORIST ORGANISATIONS5 The above discussion highlights the rationale and forms of cooperation/ merging between criminal and terrorist organisations. As mentioned above, the data available is just indicative of the phenomenon. Needless to say, there are only estimates on the overall financial capacity of terrorist organisations. In most cases, it is possible to have only guesstimates. With regard to Islamist organisations, Freeman (2011) reported that it is around US$30 million per year. Feiler (2007) estimated that it was around US$1 billion in the period 1995
2005. A more accurate empirical study reports the financial capacity of al-Qaeda over 1999
2004 varied between US$20 and 50 million per year (Schneider, 2010). The mere fact that terrorist and criminal organisations overlap or interact in some illicit businesses should not come as a surprise. Indeed, many terrorist groups of the Marxist-Leninist type of the 1960s and 1970s relied on various criminal activities
particularly kidnapping, bank robberies and extortion
to supplement their funding. Other organisations including the FARC in Colombia and Shining Path in Peru financed their activities through drug trafficking (Yepes, 2008). What is worth considering is rather the magnitude of wealth generated by these activities. In what follows, we mention some complementary illicit sources of funding for terrorist groups. First, as noted above, smuggling of illegal drugs has been a significant source of funding for terrorist organisations for years. In particular, in latest years several Islamist groups in central Asia have developed strong ties and involvement with the narcotic industry. Drug trafficking has provided 90% of funds to rebels in Tajikistan (see also Paoli, Rabkov, Greenfield, & Reuter, 2007) and also for the Islamic Movement of Uzbekistan (Hardouin & Weichhardt, 2006), while in Colombia it provided rebels with US$800 million per year (more than half the money spent in defence by the Colombian government) (Kaldor, 1999, p. 102). Afghanistan produces more than 90 per cent of world heroin and this has long been the main source of funding for the Taliban (Van Ham & Kamminga, 2006
2007); recently Peters (2009) and Cornell (2005, 2006, 2007) reported information on the ties between drug-trafficking and al-Qaeda in Afghanistan and

Some Insights on Link Between Terrorism, Organised Crime and ‘New Wars’

247

Pakistan. According to Schneider (2010) drug trafficking was worth 30
35% of total al-Qaeda budget in the period 1999
2006. Drug trafficking is one of the funding sources for Hizbollah in Lebanon (Arena, 2006; Naylor, 1993, p. 35) and it is rapidly growing in importance in West Africa (Cockayne & Williams, 2009; Mazzitelli, 2007), an area rich in armed conflicts and difficult post-conflict stabilisations. A report by UNODC, entitled Addiction, Crime and Insurgency: The Transnational Threat of Afghan Opium (2009), also highlights the complex web of ties between Afghan opiate economy, transnational terrorism and local insurgency. In particular, the report remarks that ‘[…] Al-Qaeda does not appear to have a direct role in the Afghan opiates trade. Beyond ideological considerations, involvement in drug trafficking would increase the visibility of its secretive core group […]’.6 However, extremist groups linked or ideologically aligned with al-Qaeda have been involved in drugtrafficking. For example, the report also presents the case of sea-borne trafficking through Pakistan to Europe via the Gulf States, which appears to have been significant for al Qaeda’s funding. In a two-week period in December
January 2004, the US Navy intercepted four boats carrying tons of hashish, heroin and methamphetamines. One of these was carrying two tons of hashish with a street value of US$8
10 million and was led by suspected al-Qaeda members. According to Williams (1994, p. 100), drug today can be considered almost as a currency, and it is much appreciated by parties in armed conflicts because it can easily be exchanged with money and arms. Overall, this means that drug trafficking has become the largest source of profits for both international organised groups and terrorists (Shelley & Picarelli, 2002). Estimates range from US$200 billion (Baker, 2005) to US$300 billion every year (Costa, 2010). It comes as no surprise, then, that the United Nations takes into account the presence of drug trafficking in its conflict prevention, conflict analysis, and peace-building operations (Cockayne, 2010). In fact, Piazza (2011) highlights the specific relationship between narcotics and recrudescence of violence. In particular, it is shown illicit drug production is a significant positive predictor of transnational and domestic terrorist attacks. Another kind of illegal smuggling is related to gems and diamonds. Diamond trafficking has been a major determinant of civil conflicts in SubSaharan Africa, at least until the institution of the certification scheme known as the Kimberley Process (Hartung, 2002), but nowadays it constitutes a relevant source of funding for terrorist groups (Levi & Gilmore, 2002; Schneider, 2004, 2008, 2009; Yepes, 2008). In particular, Raphaeli (2003) reports the role of al-Qaeda affiliates in Liberia in the gem business

248

CARLA MONTELEONE ET AL.

in Africa. Other evidence on al-Qaeda’s involvement is provided by Basile (2004) in Liberia and Burkina Faso since 1998 and by Dishman (2005) and Passas and Jones (2006). In recent years other journalistic accounts uncovered the linkages between al-Qaeda and the illicit trade in so-called ‘blood diamonds’ bought from rebel groups in Africa.7 Hu¨bschle (2007) also reports about al-Qaeda’s interest in Tanzanite and Arena (2006) reports the involvement of Hizballah in diamonds trade. Eventually, another relevant illicit source of funding is kidnapping. Kidnapping as a terrorist-related activity has earned widespread recognition mostly in the aftermath of the Iraq war. Many journalists, civil servants and NGO operators in 2004
2005 became profitable sources of revenues for petty criminals and insurgents in Iraq (Looney, 2005). It is not possible to assess the real amount of money generated by this illicit business, mostly because there are not reliable statistics (the main issue is underreporting, as well documented in a number of case studies). However, some estimates of the revenues generated by this practice are indicative. For instance, figures about the FARC suggest that in the mid-2000s kidnapping earned about US$220 million per year (Riascos & Vargas, 2011, p. 5). Likewise, in Chechnya, according to Mukhina (2005) more than a thousand people were kidnapped between 1997 and 1999, their ransom ranging from US$5000 to 145,000. Overall, the annual global income is estimated to exceed US$500 million (Briggs, 2001, 2002). What seems to be particularly relevant for our purposes is that while the cases mentioned above are certainly outliers, the practice of kidnapping for ransom has been documented in a huge number of states. Latin American states have a sad record in this respect, but even Europe is not immune. For example, traditional terrorist groups like the Red Brigades in Italy or ETA in Spain resorted to this practice in their formative stages. This shows that while this practice currently seems to be spread mostly in failed states, it is also typical of young clandestine terrorist organisations. The main point that needs to be articulated here is that, as claimed by Phillips (2009), we can draw a distinction between kidnapping as a sporadic experience and kidnapping as a business. The former is perpetrated by small groups as a handy option to raise money. It does not require contacts with more structured networks to seize and keep hostages, or launder the ransom. It is basically a bilateral relationship between the victim and the perpetrator. On the other hand, when kidnapping is seen as a business, planning and investments get much more complicated, since kidnappers need to make their endeavour a stable activity. As such:

Some Insights on Link Between Terrorism, Organised Crime and ‘New Wars’

249

The analytical process that a radical group goes through is comparable to a corporate business plan. Kidnappers have to identify the risks to their business, variables that include: detection from law enforcement, the individual skills set of the group and even the current political and social climate. This must be examined relative to the market value of the hostage, the turnaround of the crime and the supply of hostages. The ability to store the hostage and shield them from detection is also an additional consideration. (Phillips, 2009, p. 203)

Evidently, the choice to enter the business depends on a cost-benefit analysis. In conclusion, the final decision will depend as much on the external context (i.e. the costs that are imposed by legal authorities and the opportunities to find valuable preys) as on internal factors. This explains at least partially why failed states are more likely to experience kidnap for ransom,8 but it also elucidates why they host established organisations devoted to making this activity a long-term and lucrative business.

CONCLUSIONS The end of the Cold War and the decline or reshaping of state sponsorship for terrorist groups contributed to consolidate a crime
terror nexus. According to Sanderson (2004), the war on terror has just made things worse, as it prompted terrorist organisations to establish links with criminal organisations in order to fund their activities. Admittedly, the dividing line between criminals and terrorists has traditionally been blurred. What has changed in the past 20 years, according to many authors (Ehrenfeld, 2011, p. 8; Hutchinson & O’Malley, 2007; Rabasa et al., 2006, p. 101; Rollins & Wyler, 2012, p. 4), is their growing reliance on these forms of funding. Both superpowers provided financial and logistical support to fighting groups in the then Third World. In particular, according to Ehrenfeld (2011, p. 8): Most modern international terrorist organizations were trained, sponsored, and supported by the Soviet Union and its surrogates to help expand Marxism and Leninism. […] The Soviet Bloc continued to train PLO terrorists in countries as diverse as Cuba, Vietnam, South Africa, Bulgaria, and Hungary, until the Soviet Union’s demise in 1991.

With the abrupt collapse of the bipolar confrontation, the number of states sponsoring terrorism (and their funds) was significantly reduced. Beyond that, what marks a difference today is most importantly the transformation occurred in the security scenario. As discussed in the previous

250

CARLA MONTELEONE ET AL.

pages, ‘new wars’ imply a peculiar kind of globalised informal economy, where illicit activities play a critical role. Borrowing Angel Rabasa’s words: The activities of these groups tend to weaken and corrupt political and social institutions, particularly when they involve trafficking in a lucrative and social destructive commodity such as cocaine. To the extent that these groups are successful, they also displace state and government institutions, usually weak to begin with, in the areas where they establish a foothold. Unchecked, the groups will expand their resource base, increase their recruiting pool, and generate greater capacity at the expense of the state. Therefore, there is a high correlation between the development of these groups and failed or failing states. (Rabasa et al., 2006, p. 102)

While most of the current literature has concentrated either on changes in the characteristics of contemporary armed conflicts or on the crime
terror nexus, we believe that a fruitful avenue for future research is the way these phenomena interact with each other. As shown in the previous pages, the literature on ‘new wars’ correctly emphasises the growing relevance of private actors as fighting parties and, equally important, the need to extract resources from black markets. In a similar vein, scholars debating on the crime
terrorism connection have analysed the structure of incentives that might lead criminal and terrorist organisations to converge. Interestingly, this literature has also pointed out how convergence is made easier by a number of external factors that are usually present in conflict situations, although more research is needed to test this hypothesis. For one thing, the empirical analysis on the illicit activities perpetrated by terrorist groups clearly shows that at least some terrorist organisations have become relevant players in a variety of criminal
predominantly drugrelated
businesses. In conclusion, the growing collaboration of criminal and terrorist organisations in conflict situations calls for further investigation, both in terms of theoretical speculation and empirical analysis. What are the mechanisms at play that make terrorists turn into criminals and vice versa? What kind of illicit activities will terrorist engage in? What calculations are made by terrorists when deciding whether to ally with criminal organisations or develop in-house capabilities? As stated elsewhere in this volume, the complexity of these phenomena requires an interdisciplinary effort to tackle the mechanisms at play in this articulated relationship.

NOTES 1. This section elaborates on Monteleone (2010).

Some Insights on Link Between Terrorism, Organised Crime and ‘New Wars’

251

2. As recognised among others by Hoffman (1998, p. 43), ‘the terrorist is fundamentally an altruist’, while criminals are motivated by personal gain. Moreover, the intent of terrorist actions is revolutionary, no matter if it is aimed to change given policies, regimes or societies. By reverse, criminal organisations are inherently conservative (Jamieson, 2005, p. 165), since they prefer to uphold the status quo
especially when they can take advantage of weak state institutions. As a result, in Gupta et al. (2009), terrorists are public good seekers; criminals are quasi-public good seekers. 3. Interestingly, both the FARC and the Taliban were initially strongly opposed to drug trafficking, to the point of destroying drug cultivations in their controlled areas. However, the FARC “were rapidly compelled to acknowledge that coca cultivation was vital to the very survival of the peasants who comprised its social base” (Labrousse, 2005, p. 172), and ‘the Taliban realized that in a country devastated by war, with its infrastructure destroyed, the peasants had access to no other resources than poppy cultivation’ (Labrousse, 2005, p. 173). 4. Other examples include the connection between al-Qaeda and Bosnian criminal organisations to smuggle Afghan heroin into Europe via the Balkans; criminal networks in southern Thailand aimed at smuggling arms into the conflict zones of Sri Lanka, Aceh, Sulawesi and Maluku with the intent of arming terrorist groups; the links between the Albanian mafia and the Kosovo Liberation Army (KLA) during the Kosovo conflict, which allowed the KLA to buy weapons in ‘drugs-forarms’ arrangements. 5. This section elaborates on Schneider & Caruso (2011). 6. UNODOC (2009, p. 102). 7. See the account by Lucy Jones on BBC news website http://news.bbc.co.uk/2/ hi/2775763.stm 8. The flipside of the coin is that democracies experience fewer events of kidnapping and hostage-taking, as empirically shown in Santifort and Sandler (2013).

REFERENCES Adler, E. (2006). Communitarian multilateralism. In E. Newman, R. Thakur, & J. Tirman (Eds.), Multilateralism under challenge? Power, international order and structural change. Tokyo: United Nations University Press. Arena, M. P. (2006). Hizballah’s global criminal operations. Global Crime, 7(3), 454
470. Arquilla, J., & Ronfeld, D. (Eds.). (2001). Networks and netwars. The future of terror, crime, and militancy. Santa Monica, CA: RAND. Baker, R. (2005). Capitalism’s Athilles Heel
Dirty money and how to renew the free-market system. Hoboken, NJ: Wiley. Basile, M. (2004). Going to the source: Why Al Qaeda’s financial network is likely to withstand the current war on terrorist financing. Studies in Conflict & Terrorism, 27(3), 169
185. Bjo¨rnehead, E. (2004). Narco-terrorism: The merger of the war on drugs and the war on terror. Global Crime, 6(3&4), 305
324. Briggs, R. (2001). The kidnapping business. London: The Foreign Policy Centre.

252

CARLA MONTELEONE ET AL.

Briggs, R. (2002). Hostage Inc. Foreign Policy, 131, 28, 29. Cockayne, J. (2010). Enter the Security Council: A milestone in drug control policy. Retrieved from http://www.ipacademy.org/news/comment-a-analysis/139-enter-the-security-council-amilestone-in-global-drug-control-policy.html. Accessed on 3 January 2013. Cockayne, J., & Williams, P. (2009). The invisible tide: Towards an international strategy to deal with drug trafficking through West Africa. New York: International Peace Institute. Cornell, S. (2005). The interaction of narcotics and conflict. Journal of Peace Research, 42(6), 751
760. Cornell, S. (2006). The narcotics threat in Greater Central Asia: From crime-terror nexus to state infiltration. China and Eurasia Forum Quarterly, 4(1), 37
67. Cornell, S. (2007). Narcotics and armed conflict: Interactions and implications. Studies in Conflict and Terrorism, 30(3), 207
227. Costa, A. (2010). The economics of crime: A discipline to be invented and a nobel prize to be awarded. Journal of Policy Modeling, 32(5), 648
661. Dishman, C. (2005). The leaderless nexus: When crime and terror converge. Studies in Conflict and Terrorism, 28(3), 237
252. Ehrenfeld, R. (2011). Funding evil. How terrorism is financed
and how to stop it (2nd ed.). New Rochelle, NY: Multieducator. Feiler, G. (2007). The globalization of terror funding. Mideast Security and Policy Studies Series No. 74. The Begin-Sadat Center for Strategic Studies, Bar-Ilan University, Tel-Aviv. Freeman, M. (2011). The sources of terrorist financing: Theory and typology. Studies in Conflict and Terrorism, 34(1), 461
475. Gunaratna, R. (2001). The lifeblood of terrorist organizations: Evolving terrorist financial strategies. In A. Schmid (Ed.), Countering terrorism through international cooperation (pp. 180
205). Milan: ISPA. Gupta, D., Horgan, J., & Schmid, A. (2009). Terrorism and organized crime: A theoretical perspective. In D. Canter (Ed.), The faces of terrorism: Multidisciplinary perspectives (pp. 123
136). Hoboken, NJ: Wiley. Gurr, T. (2000). Peoples versus States: Minorities at risk in the new century. Washington, DC: United States Institute of Peace. Hardouin, P., & Weichhardt, R. (2006). Terrorist fund raising through criminal activities. Journal of Money Laundering Control, 9(3), 303
308. Hartung, W. D. (2002). The new business of war: Small arms and the proliferation of conflict. Ethics and International Affairs, 15(1), 79
96. Hoffman, B. (1998). Inside terrorism. New York, NY: Columbia University Press. Holsti, K. (1996). The state, war and the state of war. Cambridge: Cambridge University Press. Hutchinson, S., & O’Malley, P. (2007). A crime-terror nexus? Thinking on some of the links between terrorism and criminality. Studies in Conflict & Terrorism, 30(12), 1095
1107. Hu¨bschle, A. (2007). Terrorist financing in Southern Africa: Are we making a mountain out of a Molehill. ISS Paper No. 132. Jackson, R. (2007). Regime security. In A. Collins (Ed.), Contemporary security studies (pp. 146
163). Oxford: Oxford University Press. Jamieson, A. (2005). The use of terrorism by organized crime: An Italian case study. In T. Bjorgo (Ed.), Root causes of terrorism: Myths, reality and ways forward (pp. 164
177). London: Routledge. Kaldor, M. (1999). New wars and old wars. Organized violence in a global era. Oxford: Polity Press.

Some Insights on Link Between Terrorism, Organised Crime and ‘New Wars’

253

Keen, D. (1998). The economic functions of violence in civil war. Adelphi Paper No. 320. Keohane, R. O. (2006). The contingent legitimacy of multilateralism. In E. Newman, R. Thakur, & J. Tirman (Eds.), Multilateralism under challenge? Power, international order and structural change. Tokyo: United Nations University Press. Labrousse, A. (2005). The FARC and the Taliban’s connection to drugs. Journal of Drug Issues, 35(1), 169–184. Leander, A. (2001). Globalisation and the eroding state monopoly of legitimate violence. COPRI Working Paper No. 24, Copenhagen. Levi, M., & Gilmore, W. (2002). Terrorist finance, money laundering and the rise of mutual evaluation: A new paradigm for crime control. In M. Pieth (Ed.), Financing terrorism (pp. 87
115). Netherlands: Kluwer Academic. Looney, R. (2005). The business of insurgency. The expansion of Iraq’s shadow economy. The National Interest, 81, 67
72. Luna, D. M. (2008). Narco-trafficking: What is the nexus with the war on terror? Retrieved from http://2001-2009.state.gov/p/inl/rls/rm/110828.htm. Accessed on 5 January 2013. Makarenko, T. (2004). The crime-terror continuum: Tracing the interplay between transnational organised crime and terrorism. Global Crime, 6(1), 129
145. Mazzitelli, A. L. (2007). Transnational organized crime in West Africa: The additional challenge. International Affairs, 8(3), 1071
1090. McCarthy, D. (2003). Narco-terrorism: International drug trafficking and terrorism: A dangerous mix. Testimony of the Deputy Assistant Secretary for International Narcotics and Law Enforcement Affairs, U.S. Department of State, before the U.S. Senate Judiciary Committee, 20 May 2003. Migdal, J. (1988). Strong societies and weak states: State-society relations and state capabilities in the Third World. Princeton, NJ: Princeton University Press. Monteleone, C. (2010). La criminalita` organizzata come problema di sicurezza internazionale nella nuova agenda transatlantica. In G. Fiandaca & C. Visconti (Eds.), Scenari di mafia. Turin: Giappichelli. Mylonaki, E. (2002). The manipulation of organized crime by terrorists: Legal and factual perspectives. International Criminal Law Review, 2(3), 213
235. Napoleoni, L. (2003). Modern jihad
Tracing the dollars behind the terror network. London: Pluto. Naylor, R. T. (1993). The insurgent economy: Black market operations of guerrilla organizations. Crime, Law and Social Change, 20(1), 16
23. Newman, E. (2004). The ‘New Wars’ debate: A historical perspective is needed. Security Dialogue, 35(2), 173
189. Paoli, L., Rabkov, I., Greenfield, V. A., & Reuter, P. (2007). Tajikistan: The rise of a narcostate. Journal of Drug Issues, 37(4), 951
980. Passas, N., & Jones, K. (2006). Commodities and terrorist financing: Focus on diamonds. European Journal on Criminal Policy and Research, 12(1), 1
33. Peters, G. (2009). How opium profits the Taliban. Peaceworks n.626, United States Institute of Peace. Phillips, E. (2009). The business of kidnap for ransom. In D. Canter (Ed.), The faces of terrorism. Multidisciplinary perspectives (pp. 189
208). Hoboken, NJ: Wiley. Piazza, J. A. (2011). The illicit drug trade, counternarcotics strategies and terrorism. Public Choice, 149, 297
314. Rabasa, A., Chalk, P., Cragin, K., Daly, S., Gregg, H., Karasik, T., … Rosenau, W. (2006). Beyond al-Qaeda, Part 2: The outer rings of the terrorist universe. Santa Monica, CA: RAND.

254

CARLA MONTELEONE ET AL.

Raphaeli, N. (2003). Financing of terrorism: Sources, methods and channels. Terrorism and Political Violence, 15(4), 59
82. Riascos, A., & Vargas, J. (2011). Violence and growth in Colombia: A review of the quantitative literature. Economics of Peace and Security Journal, 6(2), 15
20. Rollins, J., & Wyler, L. S. (2012). Terrorism and transnational crime: Foreign policy issues for Congress. CRS Report for Congress. Rollins, J., & Wyler, L. S. (2013). Terrorism and transnational crime: Foreign policy issues for Congress. CRS Report for Congress. Rollins, J., Wyler, L. S., & Rosen, S. (2010). International terrorism and transnational crime: Security threats, U.S. policy, and considerations for Congress. CRS Report for Congress. Sanderson, T. M. (2004). Transnational terror and organized crime: Blurring the lines. SAIS Review, 24(1), 49
61. Santifort, C., & Sandler, T. (2013). Terrorist success in hostage-taking missions: 1978
2010. Public Choice, 156(1
2), 125
137. Schmid, A. P. (1996). The links between transnational organized crime and terrorist crimes. Transnational Organized Crime, 2(4), 40
82. Schmid, A. P. (2004). Links between terrorist and organized crime networks: Emerging patterns and trends. Milan: ISPA. Schmid, A. P. (Ed.). (2011). The Routledge handbook of terrorism research. London: Routledge. Schneider, F. (2004). The financial flows of islamic terrorism. In D. Masciandaro (Ed.), Global financial crime: Terrorism, money laundering and offshore centres (pp. 97
126). Aldershot: Ashgate. Schneider, F. (2008). Turnover of organized crime and money laundering: Some preliminary empirical findings. Discussion Paper, Department of Economics, University of Linz. Schneider, F. (2009). Die Finanzstro¨me von Organisierter Kriminalita¨t und Terrorismus: Was wissen wir (nicht)? Vierteljahreshefte zur Wirtschaftsforschung, 78(4), 73
87. Schneider, F. (2010). Turnover of organized crime and money laundering: Some preliminary empirical findings. Public Choice, 144(3
4), 473
486. Schneider, F., & Caruso, R. (2011). The (hidden) financial flows of terrorist and transnational crime organizations: A literature review and some preliminary empirical results. Economics of Security Working Paper No. 52, DIW Berlin. Shain, Y., & Barth, A. (2003). Diasporas and international relations theory. International Organization, 57(3), 449–479. Sheffer, V. (1994). Ethno-national diasporas and security. Survival, 36(1), 60–79. Shelley, L. I., & Picarelli, J. T. (2002). Methods not motives: Implications of the convergence of international organized crime and terrorism. Police Practice and Research: An International Journal, 3(4), 305
318. UNODOC. (2009). Addiction, crime and insurgency: The transnational threat of Afghan opium. UNODOC, Vienna. Van Ham, P., & Kamminga, J. (2006
2007). Poppies for peace: Reforming Afghanistan’s opium industry. The Washington Quarterly, 30(1), 69
81. Williams, P. (1994). Transnational criminal organisations and international security. Survival, 36(1), 96
113. Yepes, V. (2008). International cooperation in the fight against terrorist financing. Dissertation. Universitat de Barcelona, Barcelona, Spain.