Energy and Sustainable Development in Mexico [1 ed.] 9781603443241, 9781603441032

Citation preview

Ener“

and Sustainable Development in

MEXICO John R. Moroney and Flory Dieck-Assad

00-A3495 7/27/05 7:11 AM Page i

Energy and Sustainable Development in Mexico number sixteen Texas A&M University Economics Series

00-A3495 7/27/05 7:11 AM Page ii

00-A3495 7/27/05 7:11 AM Page iii

John R. Moroney and Flory Dieck-Assad

Energy and Sustainable Development in Mexico

Texas A&M university press College Station

00-A3495 7/27/05 7:11 AM Page iv

Copyright © 2005 by John R. Moroney and Flory Dieck-Assad Manufactured in the United States of America All rights reserved First edition The paper used in this book meets the minimum requirements of the American National Standard for Permanence of Paper for Printed Library Materials, Z39.48-1984. Binding materials have been chosen for durability.   Library of Congress Cataloging-in-Publication Data Moroney, John R. Energy and sustainable development in Mexico / John R. Moroney, Flory Dieck-Assad.—1st ed. p. cm. (Texas A&M University economics series ; no. 16) ISBN 1-58544-462-6 (cloth : alk. paper) 1. Energy development—Mexico. 2. Energy policy—Mexico. 3. Sustainable development—Mexico. 4. Petroleum—Prospecting— Government policy—Mexico. 5. Natural gas—Prospecting— Government policy—Mexico. I. Dieck-Assad, Flory, 1975 – II. Title. III. Series. HD9502.M62M67 2005 333.8230972 — dc22 2005004843

00-A3495 7/27/05 7:11 AM Page v

Contents

preface vii 1 Introduction 3 2 Energy, Capital, and Productivity 15 3 pemex Finances: Revenues and Taxes 25 4 Exploration and Development Drilling 39 5 Successful Exploration and Development 53 6 Additions to Oil and Gas Reserves 63 7 Production Models 77 8 Simulating the Integrated Oil and Gas Supply Model 91 9 Conclusions 103 epilogue 109 appendix 113 notes 127 references 133 index 139

00-A3495 7/27/05 7:11 AM Page vi

00-A3495 7/27/05 7:11 AM Page vii

Preface

This book is the result of collaborative research conducted during 2002 and 2003 and the subsequent analysis of data. It is an account of the requirements for achieving sustainable development in Mexico. It also provides a detailed econometric analysis of productivity and the requirements for sustainable development of energy resources. Growth in energy, chiefly oil and natural gas, is one cornerstone of development in Mexico. The goals of sustainable development are primarily economic: increases in productivity and real standards of living and long-term reductions in poverty and unemployment. These economic aspirations are inevitably linked to social and political goals. Major social goals include higher standards of public health, improved public schooling, and broader access to electricity, especially among Mexico’s lowest-income families. A key political goal of the administration of President Vicente Fox is environmental improvement: cleaner air in densely populated cities such as Mexico City, Guadalajara, and Juarez. By ratifying the Kyoto Protocol in 2002, the federal government of Mexico is committed to reducing the nation’s greenhouse-gas emissions, especially carbon dioxide. The production and consumption of commercial energy occupy center stage for achieving these goals. Mexico’s energy production consists mainly of crude oil and natural gas. Oil is the country’s major export and source of foreign exchange and will continue to be for the foreseeable future. Petroleum products and natural gas account for nearly 90 percent of the country’s total energy consumption. These fuels are essential for higher productivity and enhanced standards of living. Furthermore, because natural gas is the cleanest-burning fossil fuel, the Mexican government now encourages its substitution for coal and oil whenever economically feasible. Nearly all of the planned growth in electrical generating capacity will be fueled by natural gas. Oil and natural gas are produced exclusively by Petroleos Mexicanos (pemex), a government monopoly. Thus the recent history of oil and gas development must be a corresponding history of pemex: how its budgets are controlled by the Mexican Congress and how its revenues are taxed. Because of the dominant roles that oil and gas play, it is only a slight exaggeration to assert that pemex is the key to the future of Mexico’s energy situation. To our knowledge, we have written the first detailed account of pemex’s fiscal arrangements: most importantly, how its budgets diminished and its revenues were heavily taxed from 1979 through 2000. This tightening fiscal noose caused oil and gas

vii

00-A3495 9/9/05 11:32 AM Page viii

viii

Preface

reserves to decline for twenty years. We explain why sustainable development requires that reserves be substantially increased. We are grateful to pemex officials for providing essential information concerning the company’s fiscal and operating strategies. We are also indebted to officials of the Mexican Secretary of Energy for candid discussions concerning pemex finances and investments. Although we are listed as coauthors, this book is the result of a team effort. Clara Dieck-Assad and Maria Antonieta Dieck-Assad provided invaluable assistance in compiling volumes of data in Excel files and importing them into E-views for econometric work. David Brightwell assisted John Moroney in constructing and estimating econometric models throughout the summer of 2003. Our thanks go to Tyffanne Rowan and Teri Tenalio at Texas A&M for the skill, patience, and cheerfulness with which they typed numerous drafts of every chapter. Helen Williams improved the text at every stage with her valuable editorial work. Robert A. Wattenbarger and Richard A. Startzman, professors in the Department of Petroleum Engineering, Texas A&M University, provided exceedingly helpful reviews of chapters 5, 6, and 7. John M. Trapani, professor in the A. B. Freeman School of Business, Tulane University, gave us useful comments on chapters 1 and 2. Miguel Moreno-Tripp, professor in the Department of Accounting and Finance, Instituto Tecnológico y de Estudios Superiores de Monterrey, also furnished valuable suggestions on chapters 1 and 2. Benjamín Contreras Astiazarán, general coordinator of the Mejora Regulatoria Sectorial de la Comisión Federal de Mejora Regulatoria, improved every chapter with his careful, well-informed suggestions. We are pleased to acknowledge our gratitude to him. Without the extraordinary assistance of numerous public officials in Mexico, we could not have written this book. We greatly appreciate the expert advice we received from Germán Alarco Tosoni and Jorge Nuño Lara, Secretaría de Energía; Daniel Martínez Salinas, Mexican Institute of Petroleum; Marco Vincio Gómez, Guadalupe Gómez, and José Martínez, uanl; Angel Roberto Gonzaléz, inegi; Fernando Rodríguez Valdéz and Adrían Martínez Tamayo, conapo; Lourdes Aduna Barba, canacero; Luis Ramos, José Luis Pérez Hernández, and Jorge Jimenez Bernal, pemex Exploración y Producción; Esteban Levin-Balcells, Celina Torres Uribe, Valdéz Padilla, Adela Cárdenas, and Eloy Olivares, pemex; Erubiel González and José Antonio Puente González, Maquinaria Diesel, S.A. de C.V.; and David Shields, private energy consultant. Our research was supported by a grant from the Texas A&M Vice President for Research and Consejo Nacional de Ciencia y Tecnología (conacyt). In 2001, these institutions agreed to finance several projects of mutual interest to Mexico and the United States. Sustainable development is a profoundly important goal in both countries. We are fortunate to have been awarded one of these grants and wish to express our thanks to both organizations.

00-A3495 7/27/05 7:11 AM Page ix

Preface

ix

Apart from the text, we provide an appendix listing all of the data used in the book. This information should be helpful to others who wish to undertake research on the Mexican economy. John R. Moroney Flory Dieck-Assad

00-A3495 7/27/05 7:11 AM Page x

01-A3495 7/27/05 7:12 AM Page 1

Energy and Sustainable Development in Mexico

01-A3495 7/27/05 7:12 AM Page 2

01-A3495 7/27/05 7:12 AM Page 3

chapter one

Introduction

Vicente Fox, leader of the National Action Party (PAN), was elected president of Mexico in 2000, largely because he was seen as a “president of change.” Before his election, the country had been ruled for seventy consecutive years by the Institutional Revolutionary Party (PRI). President Fox presented visionary proposals for sustainable development, organized into four major goals: 1. The first is long-run growth in living standards spread across all segments of the population. This goal would alleviate poverty by creating new jobs, investing in human capital chiefly through improved public education, and facilitating financing for housing and small- and medium-sized firms. 2. The second objective is social and infrastructure development to provide electricity to more people, especially in rural areas. Since most new electricity will be generated by natural gas, this goal will require massive expansion in natural-gas pipelines. 3. The third goal, which is fundamental to the first two, is to increase the production and consumption of commercial energy. Growth in energy consumption goes hand in hand with economic growth, especially in developing countries.1 Much more natural gas will be required for the ambitious programs to increase electric power. Some of the additional gas can be imported, but most of it must be produced domestically. 4. The fourth objective is to improve the environment, both locally and globally. Mexico City is the world’s most populous city and suffers the world’s worst air pollution.2 This pollution is a result of its location in a huge valley that traps and holds oxides of nitrogen, sulfur, and carbon. An estimated 70 percent of Mexico City’s air pollution is attributed to transportation emissions. Even before Fox took office, the government had enacted several measures that proved to be effective in improving air quality in Mexico City. For all areas of high urbanization, industrialization, and air pollution, President Fox proposed even more stringent air-quality goals, which are stated clearly in ecological norms 085 and 086.3 The Fox administration is also committed to fighting global warming. The most effective way to do so is to reduce global emissions of nitrogen oxides and carbon dioxide. Produced by burning fossil fuels, nitrogen dioxide (NO2) is a short-lived greenhouse gas that plays a major role in the formation of smog. 3

01-A3495 7/27/05 7:12 AM Page 4

4

Chapter One

Nitrous oxide (N2O) and carbon dioxide (CO2) are long-lived greenhouse gases. The human activity responsible for almost all of the emissions of these two gases and sulfur dioxide (SO2) is the burning of fossil fuels.4 A fundamental dilemma is this: Mexico relies almost completely on fossil fuels for commercial energy. How can it increase the fossil-fuel consumption required for economic growth and at the same time reduce carbon dioxide emissions? One possibility is to substitute natural gas for oil and coal because the combustion of natural gas emits about half as much carbon dioxide per million BTUs (British thermal units) of energy as coal and about two-thirds as much as oil. Another possibility is to substitute nuclear and renewable energy for the electricity produced by fossil fuels. We stress that neither of these alternatives is feasible for at least another quarter century because Mexico will have neither the gas pipeline infrastructure nor adequate nuclear or renewable energies required to implement such large-scale substitutions. Sustainable growth in Mexico encompasses four major goals: long-run, widespread gains in standards of living; social and infrastructure development; growth in the production and consumption of commercial energy; and environmental improvement. These objectives are linked in fundamental ways. Some can be achieved harmoniously, but others pose basic conflicts.

real living standards Mexico is one of the largest of the world’s developing countries. Its population, estimated to be 99.4 million in 2001, is growing by 1.9 percent annually.5 If this growth rate persists for thirty-nine years, the population will double to 200 million by 2040. Simply to maintain a constant average standard of living would then require that real gross domestic product (GDP) grow at the same rate. Despite a rapidly increasing population, Mexico enjoyed impressive growth in real living standards from 1965 to 1979: Real gross domestic product per capita (real GDP/capita) increased by 3.7 percent annually. This brisk development was driven by high rates of capital formation, equally high growth in commercial energy consumption, and technological progress embodied in new capital stocks. If real GDP/capita grew indefinitely by an annual rate of 3.7 percent, real living standards would double every twenty years. This would have been a stunning economic achievement, but it did not occur. Instead, the period of remarkable progress, 1965 –1979, was followed by twenty years of continuous stagnation. From 1980 to 2000, real GDP/capita remained practically constant. Why were 15 years of impressive growth followed by 20 years of stagnation? That is the subject of chapter 2, where we show that the period from 1980 to 2000 was marked by a series of sharp peso devaluations. These reductions caused rapid increases in the cost of imports, notably machinery and equipment, imported chiefly from the United States. As a consequence, Mexico’s rate of real-capital formation decreased sharply. Commercial energy per worker also stagnated for 20 years.

01-A3495 7/27/05 7:12 AM Page 5

Introduction

5

In 2001, Mexico’s real GDP/capita, measured in international dollars adjusted for purchasing power parity (PPP), stood at $8,770, about one-fourth of the $34,870 in the United States.6 Real GDP/capita, adjusted for PPP, is probably the best single index for comparing international differences in real average living standards. However, it is a purely monetary measure of average living standards and is therefore incomplete. It fails to account for international differences in poverty, ignores environmental concerns, and overlooks a host of demographic variables such as life expectancy, infant mortality, illiteracy, access to public health and electricity, and unemployment.

sociodemographic concerns Demographic disparities are genuinely important in comparing real living standards. In 1997, for example, male life expectancy at birth was 69 years in Mexico but 74 in the United States; female life expectancy at birth was 75 in Mexico, compared to 79 in the United States.7 Mortality rates for children under the age of 5 were much higher in Mexico, which experienced 35 deaths per 1,000 children in 1998, compared to only 9 deaths per 1,000 children in the United States. These disparities can be narrowed by improving the overall standards of public health and extending their reach to lower-income families. Poverty and income inequality remain major problems in Mexico. According to a survey conducted in 1998, 10.1 percent of Mexico’s population lives below the national poverty line and receives only 1.3 percent of the nation’s income. By contrast, families in the top 10 percent of the income distribution receive 41.7 percent of the nation’s income.8 Household income in the United States, although highly unequal, is distributed somewhat more equally than in Mexico: In 1997, families in the top 10 percent of the distribution received 30.5 percent of income, while those in the lowest 10 percent received 1.8 percent. In 1996, per-capita consumption of electricity in Mexico was 1,381 kilowatt hours (kWh), compared to 11,796 kilowatt hours in the United States.9 An ample and widespread supply of electricity is essential for broadly based improvement in living standards. Unemployment is a serious problem in Mexico, particularly among younger people, the fastest-growing segment of the population. The Instituto Nacional de Estadística Geografía e Informática México examined employment statistics covering forty-eight urban areas in 2002. Unemployment rates averaged 6.6 percent for potential workers aged 12 –19 years; 5.3 percent for those from 20 to 24 years of age; and 1.3 percent for those 45 and older. Although unemployment rates in Mexico are measured differently than in the United States, it is clear that joblessness in Mexico is much higher among younger workers.

production, consumption, and exports of energy Mexico is rife with paradox. It is richly endowed with natural resources, particularly oil and natural gas. In 2002, its oil reserves were estimated at 25.4 billion barrels, exceeding estimated U.S. reserves of 22 billion.10 Yet Mex-

01-A3495 7/27/05 7:12 AM Page 6

6

Chapter One

ico produces only 3.8 million barrels a day, less than half the U.S. production of 7.87 million.11 Why is there so little production with such huge reserves in Mexico and so much more production from smaller reserves in the United States? The main reason is that oil in the United States is produced by thousands of unregulated producers intent on maximizing profits, so they produce reserves at a fast rate to obtain cash flow. In Mexico, however, oil is produced by pemex, a state-owned monopoly that is tightly controlled by the federal government. Chapter 3 spells out the symbiotic relation between pemex and the government. Mexico produces more energy—particularly oil—than it consumes. It also produces slightly less coal than it consumes. Furthermore, since 1990, Mexico has been a net importer of natural gas. The paradox is that Mexico has enormous untapped gas reserves. Oil has always been of the highest priority both because its reserves are quite large and because oil exports are a principal source of foreign exchange. In order for us to compare energy from different sources, it is essential to express all types of energy in a common unit. To enable such comparisons, the U.S. Energy Information Administration converts all types of energy in the United States, Mexico, and other countries into one unit of measure: quadrillion BTUs.12 This conversion to a common unit permits easy comparisons of production, consumption, and exports.

Production As table 1.1 shows, crude oil (including lease condensates) is the dominant fossil fuel. The 4.26 quadrillion BTUs of oil produced in 1980 accounted for 77 percent of Mexico’s total fossil fuels. Similarly, the 6.93 quadrillion BTUs of oil produced in 2001 amounted to 76 percent of total fossil fuels. This percentage is nearly constant from 1980 to 2001. Approximately three-quarters of Mexico’s oil production comes from the Bay of Campeche, located in the Mexican Gulf along the Yucatán Peninsula. The most productive field in the Bay of Campeche is Cantarell, which is located about sixty miles offshore. Cantarell produces about 1.9 million barrels of oil per day, half of the country’s total production. The Cantarell complex consists of four major subfields: Akal, Nohoch, Chac, and Kutz. Cantarell is one of the largest oil-bearing complexes ever discovered, with an estimated 35 billion barrels of oil originally in place.13 Production began in 1979. At the outset, production per well averaged a phenomenal thirty-five thousand barrels per day. As production continued and reservoir pressure dropped throughout the years, production fell continuously. In 1997, to increase reservoir pressure and production, pemex awarded a large contract to an international consortium of private firms to inject 1.2 billion cubic feet (BCF) of nitrogen per day. The contract paid off. By 2002, Cantarell’s production increased to 1.9 million barrels daily, double the rate in 1995. pemex continues to develop Cantarell. It now plans to drill fifty-three new wells and to build two new wellhead platforms. Despite these new investments,

01-A3495 7/27/05 7:12 AM Page 7

Introduction

7

Table 1.1. Fossil Fuel Production in Mexico, 1980 –2001 (Quadrillion BTUs)

Year

Crude oil production including lease condensate

Natural gas plant liquids

Dry natural gas

Coal

Total

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

4.26 5.07 6.03 5.90 6.12 6.02 5.34 5.59 5.53 5.53 5.60 5.88 5.87 5.86 5.89 5.74 6.28 6.63 6.74 6.37 6.63 6.93

0.26 0.32 0.34 0.35 0.34 0.36 0.47 0.45 0.49 0.51 0.57 0.60 0.60 0.61 0.61 0.59 0.56 0.51 0.56 0.58 0.58 0.57

0.95 1.02 1.10 1.08 1.06 1.05 0.93 0.94 0.95 0.91 1.00 1.00 0.98 1.05 1.08 1.06 1.21 1.25 1.34 1.36 1.39 1.38

0.08 0.08 0.09 0.12 0.13 0.13 0.14 0.17 0.15 0.16 0.17 0.15 0.14 0.15 0.16 0.16 0.18 0.20 0.21 0.19 0.21 0.22

5.54 6.49 7.56 7.45 7.65 7.56 6.88 7.15 7.12 7.11 7.34 7.64 7.59 7.68 7.74 7.56 8.23 8.59 8.85 8.51 8.81 9.09

Source: Energy Information Administration, www.eia.doe.gov.

pemex estimates that Cantarell’s production will decline to 1.5 million daily barrels by 2006 and perhaps 1.2 million per day by 2008. Dry natural gas and natural-gas plant liquids account for a much smaller share of production than oil. For example, in 1980, dry natural gas plus plant liquids accounted for 1.21 quadrillion BTUs—22 percent of the overall fossilfuel production. Their combined production in 2001 was 1.95 quadrillion BTUs, or 21 percent of all fossil fuels. Current natural-gas reserves are located primarily in the southwestern states of Tabasco and Chiapas, where large reserves were discovered in 1978. In March, 2002, pemex announced the discovery of three new gas fields in the state of Veracruz. pemex estimates that, when these fields are developed, they could account for one-quarter of Mexico’s total gas reserves. Production from these and other gas fields enabled nonassociated gas production to increase by approximately 22 percent from 2001 to 2004. Other major gas basins are located in the northern region: the Sabinas-Tamaulipas platform with 85 exploratory opportunities, the Burgos platform with 500, the Lamprea platform with 108, and the Lankahuasa platform with 72.14 Coal production is insignificant: It accounted for about 1.4 percent of total fossil fuels in 1980 and about 2.4 percent in 2001. Coal reserves, estimated to be approximately 1.3 billion tons, are located chiefly in the northeastern state of Coahuila. Almost all of Mexico’s coal is used to produce electricity.

01-A3495 7/27/05 7:12 AM Page 8

8

Chapter One

Table 1.2. Fossil Fuel and Total Primary Energy Consumption in Mexico, 1980 –2001 (Quadrillion BTUs)

Year

(1) Petroleum and petroleum products

(2) Dry natural gas

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

2.59 2.83 2.96 2.70 3.07 2.98 3.12 3.20 3.27 3.36 3.44 3.45 3.47 3.44 3.62 3.47 3.56 3.63 3.83 3.91 3.90 3.77

0.84 0.91 0.99 0.99 1.00 1.04 0.92 0.94 0.95 0.92 1.02 1.06 1.06 1.08 1.14 1.16 1.23 1.26 1.36 1.34 1.46 1.45

(3) Coal

(1)  (2)  (3) Total fossil fuels

Total primary energy consumption

0.11 0.10 0.12 0.13 0.13 0.16 0.15 0.17 0.16 0.17 0.17 0.16 0.18 0.18 0.19 0.21 0.24 0.26 0.27 0.25 0.27 0.27

3.54 3.84 4.07 3.82 4.20 4.17 4.19 4.30 4.39 4.46 4.62 4.66 4.71 4.71 4.95 4.84 5.03 5.15 5.46 5.50 5.63 5.49

3.74 4.11 4.33 4.06 4.47 4.48 4.46 4.57 4.68 4.80 4.98 5.02 5.12 5.13 5.30 5.31 5.55 5.65 5.93 6.06 6.19 6.00

Source: Energy Information Administration, www.eia.doe.gov.

Consumption Fossil fuels account for almost all of Mexico’s energy consumption. Table 1.2 shows the consumption of petroleum and petroleum products, dry natural gas, coal, and their sum (total fossil fuels). The last column lists total primary energy consumption, which is all of the energy consumed by end users, excluding electricity but including the energy consumed at electric utilities to generate electricity. Fossil fuels clearly dominate energy consumption. They account for 95 percent of the total energy consumed in 1980 and about 92 percent in 2001. Among the individual fossil fuels, petroleum and petroleum products represent 69 percent of the total primary energy in 1980 and 63 percent in 2001. Dry natural gas accounts for 22 percent of the total primary energy in 1980 and 24 percent in 2001; coal accounts for a meager 3 percent in 1980 and 4.5 percent in 2001.

Exports The difference between fossil fuels produced (table 1.1) and consumed (table 1.2) is exports. For example, in 1980, Mexico produced 4.26 quadrillion BTUs of crude oil and lease condensate, consumed 2.59 quadrillion BTUs of petroleum and petroleum products, and exported 1.67 quadrillion BTUs of crude oil. Put differently, the country exported 39 percent of the oil it produced. In 2001, the country produced 6.93 quadrillion BTUs of oil, consumed 3.77 quadrillion

01-A3495 7/27/05 7:12 AM Page 9

Introduction

9

BTUs, and exported 3.16 quadrillion BTUs, or 46 percent, of its oil.15 Oil exports are the single most important source of Mexico’s foreign exchange. By comparing tables 1.1 and 1.2, one confirms that oil has been exported every year since 1980 and natural gas imported each year since 1990. The government has given oil priority over gas for at least three reasons: (1) Oil is easily exported on tankers; (2) oil exports are the country’s principal source of foreign exchange; and (3) natural gas cannot yet be widely consumed because the country lacks the pipeline infrastructure for transmission. Mexico’s entire natural-gas pipeline network extends only about six thousand miles, so gas consumption is severely constrained.

Electricity Mexico has installed an electric-power-generating capacity of 42.3 million kilowatts (or 42,300 megawatts). Electricity production more than tripled, from 63.6 billion kWh in 1980 to 198.6 billion kWh in 2001. We are chiefly interested in two important questions: How is electricity produced, and how important are fossil fuels in its production? Table 1.3 shows that Mexico had no nuclear power until 1990. Even by 2001, the country had only two nuclear plants producing 8.3 billion kWh, or about 4 percent of its electricity.16 There are no nuclear plants under construction or being planned at present, so nuclear capacity cannot be expanded for at least ten years. These facts concerning present and near-term nuclear energy underscore the importance of fossil fuels in the foreseeable future of electricity. The rapid expansion of nuclear energy is a near-term impossibility. The burden of growth in electricity must necessarily be borne by fossil fuels, chiefly natural gas. Hydroelectricity accounts for 28.2 billion kWh, or about 14 percent of all electricity produced. There are plans to construct El Cajón, a 750-megawatt hydroelectric project on the west coast. Its estimated cost is $650 million, the largest publicly funded infrastructure to be financed by the Fox administration. Completion is slated for the end of August, 2007. When finished, this project will augment Mexico’s total installed electric-generating capacity by about 1.8 percent. No other hydroelectric projects are now being planned, so hydroelectricity offers limited opportunity for growth.17 All other renewable resources (geothermal, solar, wind, wood, and waste) account for 5.8 billion kWh, or about 3 percent of the total electricity production. Apart from hydropower, geothermal appears to be the most promising source of renewable energy. In 2002, Mexico reported 855 megawatts of installed geothermal capacity. Two new geothermal plants are under construction: the 100-megawatt Los Azufres facility in Michocán and the 10-megawatt Las Tres Vigenes plant in Baja California. Currently only two wind-power projects are in operation. Together they account for 2.1 megawatts of installed capacity. Although the government has announced goals to increase the windgenerated capacity by 2006, it is clear that nonhydroelectric renewables will continue to account for a small percentage of electricity production.

01-A3495 7/27/05 7:12 AM Page 10

10

Chapter One

Table 1.3. Sources of Electricity Production in Mexico, 1980 –2001 (Billion kWh)

Year

Nuclear

Hydroelectric

Geothermal, solar, wind, wood, and waste

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.8 4.0 3.7 4.7 4.0 8.0 7.5 9.9 8.8 9.5 7.8 8.3

16.7 24.4 22.7 20.5 23.4 26.0 19.8 18.2 21.0 24.4 23.2 21.6 25.9 26.0 19.8 27.3 31.1 26.2 24.4 32.5 32.8 28.2

0.9 0.9 1.3 1.4 1.4 1.6 3.4 4.3 4.7 4.7 4.9 5.2 5.5 5.6 5.3 5.4 5.4 5.2 5.7 5.9 6.1 5.8

Fossil fuels

Total

46.0 48.0 56.4 60.2 61.9 65.5 73.7 82.0 84.2 88.7 85.7 89.7 88.7 93.1 110.5 104.2 110.4 124.8 133.4 134.7 147.2 156.3

63.6 73.3 80.4 82.0 86.7 93.1 96.9 104.6 109.9 117.7 116.6 120.5 123.8 129.4 139.7 144.9 154.5 166.2 172.3 182.5 193.9 198.6

Source: Energy Information Administration, www.eia.doe.gov.

Fossil fuels have always been the primary source of electricity, and their importance is growing. They accounted for 46 billion of the 63.6 billion kWh of electricity produced in 1980. By 2001, however, they were responsible for 156.3 billion of the 198.6 billion kWh produced—nearly 80 percent of the total. Coal-fired plants now account for 10 percent of electricity.18 Country-wide coal consumption amounted to 15 million tons in 2001, of which 10 million were used in Rio Escondido and Carbon II, two generating plants operated by the state-owned Comisión Federál de Electricidad. Kimberly Clark, a U.S. manufacturing firm, and steelmakers Ispat and Altos Hornos de México plan to construct coal-fired generating plants near their production facilities. Nevertheless, these plants will add little to the total generating capacity.19 Oil-fired and gas-fired plants are now responsible for producing about 70 percent of electricity. The task of increasing the production of electricity will fall on these two fuels, especially natural gas. The burden will be eased somewhat by recent reforms that have reduced state control over electricity production. In 1992, Mexico adopted the Public Electricity Service Act, which allowed private companies to generate limited amounts of electricity. However, they are required to sell all of their power to the government-owned Comisión Federál de Electricidad, which resells it to consumers. The act, however, enabled the development of a very small, independent electricity-producing sector, which has attracted some foreign investment.

01-A3495 7/27/05 7:12 AM Page 11

Introduction

11

In November, 2002, the Fox administration introduced a bill to the Senate that would modify the Mexican constitution by creating separate generation, transmission, and distribution companies and allow greater participation by private firms.20 By attracting private investment, the Fox administration hopes that Mexico will be able to add 30,000 megawatts of capacity during the next ten years. If this target is achieved, then the country’s total generating capacity would nearly double from the current 42,300 megawatts. The lion’s share of this ambitious goal would be fueled by natural gas.

environmental concerns Another goal in President Fox’s vision of sustainable development is the protection of the global environment. He has pledged to reduce future emissions of carbon dioxide by switching fuel consumption from oil to natural gas. Crude oil now accounts for about 63 percent of primary energy consumption, while natural gas accounts for about 24 percent. Accomplishing such a large-scale switch in fuels will require enormous growth in both the production and the distribution of natural gas. Although vast gas resources are yet to be developed, converting latent resources into producible gas will require an ambitious, longterm drilling program. The government will simultaneously have to vastly expand the nation’s gas-pipeline system. Gas without pipelines is useless.

Environmental Paradoxes Carbon dioxide is widely viewed as the major cause of global warming in the past 140 years. Atmospheric scientists now firmly believe that global increases in human carbon dioxide emissions are responsible for 60 – 80 percent of observed global warming.21 Combustion of fossil fuels accounts for about 85 percent of the world’s synthetic carbon dioxide emissions. In 2000, world emissions of carbon dioxide were estimated at 6,417 million metric tons. Emissions originating in Mexico were 99 million metric tons, or only 1.5 percent of the total. In that same year, U.S. emissions were estimated at 1,578 million metric tons, or about 25 percent of the total.22 Yet Mexico’s Congress has ratified the Kyoto Treaty, which is designed to reduce the world’s carbon dioxide emissions. The treaty has never been considered by the U.S. Congress.23 The first paradox is clear: Acting alone, Mexico can do nothing to reduce global carbon dioxide emissions because it accounts for only 1.5 percent of the total. If, hypothetically, Mexico cut its emissions to zero, its unilateral action would have no substance. Yet the Fox administration and the Mexican Congress have pledged to reduce their nation’s meager portion. Even if they do, it will make no difference whatsoever from a global standpoint. The pledge is politically expedient but globally inconsequential. A second paradox is less transparent but equally compelling: Energy per worker must increase as a cornerstone of productivity growth. With roughly constant labor-force participation, rising energy per worker implies rising

01-A3495 7/27/05 7:12 AM Page 12

12

Chapter One

energy per capita. And rising energy per capita implies larger emissions per capita. It will be next to impossible for Mexico to achieve rising standards of living without increasing per-capita carbon dioxide emissions. Increasing percapita emissions would, of course, be compounded by population growth, currently at 1.9 percent annually. The Fox administration intends to reduce carbon emissions by switching fuel consumption from coal and oil to cleaner-burning natural gas. Combustion of coal emits about 217 pounds of CO2 per million BTUs of energy; oil emits about 173 pounds of CO2 per million BTUs; and natural gas releases approximately 117 pounds of CO2 per million BTUs.24 Thus the substitution of natural gas for oil—and especially for coal— could mitigate the growth in CO2 emissions. According to the Secretaría de Energía, “The engine of growth in natural gas demand will come from the increase in electricity generation in combined cycle plants. These plants present a new growth path in favor of natural gas because of its thermal efficiency and the reduced atmospheric contamination compared with other fossil fuels as coal and ‘combustoleo.’ We calculate that the use of natural gas in electricity generation will go from 23.3% in 2000 to 61.1% in 2010. Thus, the availability of natural gas in Mexico is a fundamental factor for the regional sustainable growth which will increase the competitiveness of the productive capacity, will be able to generate more employment in the economy, and will increase, in general, the population welfare” [italics in the original].25 It is the height of irony that Mexico has largely neglected its domestic-gas resources for so long. Indeed, the country now imports gas from the United States and is building liquefied natural gas (LNG) terminals, enabling it to import even more from other countries. Imports now account for about 10 percent of Mexico’s gas consumption. The government has outlined ambitious plans to increase consumption by approximately 8 percent annually through 2010. A second irony is this: Mexico consumes so little coal (about fifteen million tons annually, mostly to generate electricity) that, even if coal consumption were to cease, the planned growth in natural-gas usage would lead to more CO2 emissions in 2010 than in 2001. So, even if natural gas were to entirely replace coal in generating electricity (and such complete substitution is impossible), this hypothetical substitution would undoubtedly increase CO2 emissions because of greater planned electricity production.

summary and preview of remaining chapters The goals of sustainable development are clear, but they will be difficult to achieve. In the following chapters we present a realistic analysis of the objectives and major policy changes that will be required for achieving them. Long-term increases in living standards require corresponding gains in labor productivity. Productivity has stagnated for more than twenty years, and unemployment among younger members of the labor force is high.

01-A3495 7/27/05 7:12 AM Page 13

Introduction

13

Several important changes must occur to ensure productivity growth: lower unemployment, improved infrastructure for the transmission of electricity and natural gas, sustained growth in utilized capital and energy per worker, and improvements in technology. These are the topics of chapter 2, where we show that, from 1965 through 1979, productivity increased rapidly because of concomitant growth in utilized capital and energy per worker and improvements in technology. After 1979, productivity growth came to a standstill because of a slowdown in investment and stagnation in utilized capital and energy per worker. Broadly based gains in living standards will require effective employment of young, low-income workers. Greater investment in public education, including vocational training, will enable younger workers to acquire the skills necessary for effective job performance. It is essential to reduce joblessness among younger workers. Although educational investments that improve marketable skills are beyond the scope of this book, they constitute an issue of first-order importance for the country. More than half of Mexico’s new machinery and equipment is imported, mostly from the United States. Imports require foreign currency that can be earned only by exports. Since oil exports are now the single most important source of foreign currency, larger imports of machinery and equipment will require still greater exports of oil. However, to increase the volumes of oil necessary for larger exports and domestic consumption, oil production must increase substantially beyond the present 3.8 million barrels per day. Oil production and reserves are two of the major topics of chapters 6 and 7. Reserves are the essential basis of production. A long-term problem that cries out for a solution is that oil reserves decreased steadily from 1985 to 2000: Estimated reserves in 2000 were only 70 percent as high as those in 1985. Since oil is the most important export and also the most important source of domestic energy, the decline in oil reserves must be reversed. Sustainable development requires it. Natural-gas production and reserves are two of the other major topics of chapters 6 and 7. Until now, gas has been secondary to oil. As table 1.1 shows, dry gas plus gas plant liquids account for only 21 percent of fossil-fuel production. They will play a much larger role in the future, however. The first reason is quite specific: The planned expansion in electricity will be generated almost entirely by gas-fired plants. A broader reason is the Fox administration’s environmental commitment to substitute gas for oil and coal when substitution is economically feasible. The Secretaría de Energía (sener) states that gas consumption will increase not only in the production of electricity but also in the steel, chemical, glass, and paper-products industries “in search of clean combustion.” 26 Substitution of gas for fuel oils (“combustoleo”) is now targeted in areas of high urbanization, industrialization, and air contamination.27 Nevertheless, gas reserves are also in long-term decline. Apart from a shortterm aberration in 1986 and 1987, gas reserves reached a peak in 1983 and have

01-A3495 7/27/05 7:12 AM Page 14

14

Chapter One

declined continuously since then. Gas reserves in 2000 were only 69 percent as high as in 1983. Since gas consumption is projected to increase by about 8 percent annually until 2010, domestic reserves and production must increase as well. Sustainable development requires it. Oil and gas reserves are in long-term decline because of a twenty-year depression in drilling for new reserves. In 1980, for example, pemex drilled 432 wells. Then drilling went into a tailspin, reaching an all-time low of 63 wells in 1994. Drilling began a weak recovery, leading to an annual average of 184 wells from 1996 to 2000, then increased to 446 wells in 2001. Only a long-term recovery in successful drilling to the annual rates of 1980 or 2001 can reverse the serious decline in oil and gas reserves. Drilling is expensive, especially offshore, where most of the exploration and development has occurred since 1985. The drilling depression was caused by a financial squeeze that lasted for two decades. pemex’s drilling investments must be approved each year by Congress. The direct reason for the severe decline in drilling was the ever-tightening investment budgets that Congress granted to pemex. A broader reason, however, is that pemex is a creature of the government with little political control over its revenues. These revenues have been taxed at progressively higher rates, exceeding 60 percent in recent years. In chapter 3 we describe the political framework for pemex budget negotiations and the reasons for the collapse in drilling investment. Chapter 4 econometrically analyzes the drilling for oil and gas. Here we link the number of wells pemex drills each year to a cash-flow constraint: the net (after-tax) Mexican oil price. We also incorporate the short-term boost in drilling that occurred after 1998, when pemex was granted the right to negotiate private loans. Its inability to repay the principal and interest on these loans proved to be financially disastrous. Drilling is risky. And new reserves can be developed only by successful drilling. Exploration wells drilled in new areas are riskier than development wells drilled after exploration has succeeded. The fraction of exploration wells leading to the discovery of producible hydrocarbons is the exploration success ratio. Similarly, the development success ratio is the fraction of development wells proving to be commercially successful. Exploration and development success ratios exhibited impressive increases in Mexico from 1975 to 2000. In chapter 5 we analyze some of the reasons for their growth. Because successful drilling is the only way to increase producible reserves, sustainable development requires it.

02-A3495 7/27/05 7:12 AM Page 15

chapter two

Energy, Capital, and Productivity

This chapter analyzes aggregate labor productivity in Mexico for the years 1965 –2000. These thirty-six years were marked by sharp contrasts. From 1965 to 1979, real gross domestic product (GDP) per worker increased at an average annual rate of 3.7 percent. From 1979 to 2000, however, productivity stagnated, and only in 2000 did it recover to the level of 1979. What are the reasons for such exceptional growth until 1979, followed by two decades of stagnation? There are several explanations, of which the most important are clearly identifiable. Our goal in this chapter is to analyze the way in which two sets of financial disturbances—peso devaluations and sharply rising energy prices—influenced the primary determinants of macroeconomic productivity in Mexico. Identifying the major causes of growth followed by decay in productivity is vitally important because of the tight link between productivity and standards of living. Over many years, economy-wide living standards cannot rise faster than productivity. Long-term decline in productivity necessarily leads to a corresponding decay in standards of living. The first financial disturbances originated in a series of peso devaluations beginning in the 1970s. Between 1965 and 1975, the peso-dollar exchange rate remained fixed. Then the peso was devalued 25.5 percent at the end of 1976 and another 44.6 percent in 1977. The value of the peso depreciated against the dollar by more than half in 1982, then by another two-thirds in 1983. It fell by 50 percent in 1985 and yet again by more than 50 percent in 1986 and 1987. These devaluations had far-reaching consequences. Because much of Mexico’s investment in machinery and equipment is imported from the United States, these devaluations prohibitively increased the costs of imported capital. The devaluations had two further economic impacts. First, they induced a series of recessions, which caused cyclical reductions in capacity utilization, the demand for investment, and short-term productivity. Second, they inhibited longer-term growth in physical capital per worker, thereby destroying one cornerstone of long-term productivity growth. Another financial disturbance, exogenous to the Mexican economy, reinforced productivity stagnation. Following world oil prices, the weighted average price of Mexico’s crude oil rose from $11.56 in 1973 to $12.57 in 1976, then to $19.60 in 1979.1 These higher prices increased foreign exchange, which was of course a benefit to the economy. But they also increased the cost of domestic energy. Higher energy costs, in turn, caused long-term reductions in energy per worker, thereby undercutting a second vital foundation of productivity growth.2 15

02-A3495 7/27/05 7:12 AM Page 16

16

Chapter Two

After describing long-run productivity, we set out a framework for understanding its rise and subsequent decline. The analytical setup is an aggregate Cobb-Douglas production function characterized by technical change embodied in gross investment. By estimating the coefficients of the production function, one can understand the roles of utilized capital and energy per worker as sources of productivity growth. We show that growth in utilized capital and energy per worker are both essential for future productivity and economic growth. In later sections we discuss trends in productivity, utilized capital per worker, and energy per worker; specify the aggregate production framework; discuss the data; present econometric estimates; and analyze the sources of productivity growth.

outlines of productivity, utilized capital, and energy per worker Figure 2.1 tells the productivity story. Real GDP per worker, measured in constant 1993 pesos, increased from 30,420 pesos in 1965 to 50,600 in 1979, an average annual growth of 3.7 percent. At the current exchange rate of 11 pesos per U.S. dollar, GDP per worker was about $2,800 in 1965 and $4,600 in 1979. Productivity began to decline in 1980, then continued a downward path to 43,300 pesos (roughly $3,950) in 1988. For the next 12 years, productivity mounted a feeble recovery. Even so, by 2000, GDP per worker stood at 49,960 pesos (about $4,550), the same level as 21 years earlier. Why were 14 years of rapid gains followed by 21 years of stagnation?

60000

Real GOP per Worker

50000

40000

30000

20000

10000

0 1965

1970

1975

1980

1985 Year

Figure 2.1. Real GDP per worker (1993 pesos), 1965 –2000.

1990

1995

2000

02-A3495 7/27/05 7:12 AM Page 17

Energy, Capital, and Productivity

17

Utilized Real Capital Stocks per Worker

120000

100000

80000

60000

40000

20000

0 1965

1970

1975

1980

1985

1990

1995

2000

Figure 2.2. Utilized real capital stocks (1993 pesos) per worker.

There are several reasons; some are cyclical and others secular. First, physical capital was rapidly substituted for labor during the years of productivity growth. Even taking into account a devaluation-induced recession in 1977– 1978, utilized capital stocks per worker increased at an annual rate of 3 percent between 1965 and 1979. However, as figure 2.2 illustrates, between 1979 and 1997, utilized capital per worker cycled around a horizontal trend. In 1997, utilized capital per worker was 93,500 pesos, practically the same as in 1979. An interesting sidelight is that, for the entire period between 1965 and 2000, the utilized capital /GDP ratio displays cyclical variation but no trend whatsoever. The thirty-six-year average utilized capital /GDP ratio is 1.97. A regression of this ratio on a time trend yields an estimated slope coefficient of 0.00367, with an estimated standard error of 0.00251 and R 2 of 0.059. The Mexican economy exhibited a practically constant capital /output ratio from 1965 until the turn of the century. A second reason for the initial rise and decline in productivity is the corresponding trends in energy per worker. Between 1965 and 1979, energy per worker increased at an annual rate of 4.5 percent, even more rapidly than utilized capital per worker. Energy costs approximately doubled between 1979 and 1981, and energy per worker commenced a period of slow, statistically significant decline. From 1979 to 2000, a regression of energy per worker on a time trend produces an estimated slope coefficient of 0.855 with an estimated standard error of 0.198, statistically significant at P  0.001. The gradual post1979 decrease in macroeconomic energy intensity partially accounts for the stagnation in productivity. A more complete explanation requires a formal analytical framework.

02-A3495 7/27/05 7:12 AM Page 18

18

Chapter Two

specification and theory A production function is fundamentally a microeconomic concept. It describes how capital, labor, energy, and materials are combined to produce real value added. In multiproduct firms, very different technical properties characterize numerous production functions. Some processes may require that inputs be used in fixed proportions; others may allow for considerable input substitution. Some processes may be characterized by economies of scale; others by constant returns to scale. Even if one imposes the most extreme simplifying assumptions, an aggregate production function may bear no resemblance to its microeconomic constituents. For example, if one simplifies the problem of aggregation by assuming that every microproduction function requires strictly fixed input coefficients, Houthakker (1955) shows that, when aggregated, they yield an economy-wide Cobb-Douglas function.3 Yet, for understanding macroeconomic productivity, some sort of aggregate production function seems indispensable. We analyze real GDP per worker using an aggregate Cobb-Douglas production function with constant returns to scale. Some restrictions inherent in this choice of functional form are discussed later in the chapter. Before we specify the model in detail, it seems best to reiterate the facts we wish to explain. The central facts are that productivity increased rapidly from 1965 to 1979, then languished for two decades. We propose to identify the reasons for these very different trends. Assume that GDP in year t, Qt, is produced according to a production function Qt  F1UKt, Z t, Lt 2

(2.1)

with UKt, the real utilized capital stock; Zt, aggregate energy input; and Lt, aggregate labor employed. To make headway with estimation, one must choose among several possibilities that could include the Cobb-Douglas, a more general constant elasticity of substitution (CES), or more flexible, generalized Leontief or translog specifications. For analyzing economy-wide time series, we choose a simple form that sidesteps practical problems such as excessive multicollinearity (Moroney 1992). The production function should also capture the reality that most of Mexico’s investment in machinery and equipment is imported from the United States. New technologies are therefore embodied in imported capital goods. Accordingly, we choose an aggregate Cobb-Douglas function with constant returns to scale that incorporates capital-augmenting technological change: Qt  1UK*t 2 aZtbL1ab t

(2.2)

with UK*t being a measure of capital that includes technological progress embodied in new investment. The vintage capital model analyzed here originated with Solow (1959, 1962).

02-A3495 7/27/05 7:12 AM Page 19

Energy, Capital, and Productivity

19

Our main extension is to explicitly account for energy as a third input (Solow’s analysis is restricted to capital and labor). A vintage capital model seems to be natural for analyzing the consequences of higher energy prices because the optimal ex-ante energy intensity of new capital depends critically on the real cost of energy. Assume that machinery and equipment produced in any year are 100lk more productive than those produced the year before. If gross investment of Iv is made in year v, the surviving portion in some later year t is Mt –v. If we assume that labor and energy are assigned to capital of various vintages so as to yield maximum output, we can represent the stock of surviving capital goods by a productivity-augmented equivalent stock of capital services: J*t  a 11  lk 2 vMtv Iv t

(2.3)

vq

Note that J*t is sensitive to the cost of energy: If energy were to become so expensive that net quasi-rent on a particularly energy-intensive vintage vanished or became negative, it would be scrapped.4 Rates of embodied technical change (lk) and survival rates (Mt –v) are calculated for gross investment in machinery and equipment. The assumed rates of technical change range from 2 to 12 percent. The survival rate is (Mt –v)E  1  dE  0.9412. The value of dE  0.0588 is the economic depreciation rate for machinery and equipment, based on Mexican experts’ opinions for economically useful asset lives of seventeen years. The Jt series consists of the sums of vintage investments in machinery and equipment, augmented for technical change and depreciated at 0.0588 annually. We assume no technical change for investments in structures. Although this assumption is patently untrue, it seems to be innocuous because structures have economically useful lives of approximately forty years. Technical progress embodied in new structures can improve the efficiency of the aggregate stock only at a snail’s pace. Economic depreciation for structures is dS  0.025, based on U.S. Treasury guidelines, and their survival rate is (Mt –v)s  1  ds  0.975. Aggregate capital stocks are the sum of gross structures Ks(t) and gross machinery and equipment KE(t): K1t2  KS 1t2  KE 1t2

(2.4)

K*1t2  KS 1t2  J*x 1t2

(2.5)

where K(t) are “ordinary capital stocks” and K*1t2 are capital stocks augmented by technical progress, where x is the rate of capital-augmenting technical change ranging from 0, 0.02, 0.03, . . . , 0.12.5 With vintage capital, the aggregate production function is Q*t  G1UK*t , Lt , Zt 2

(2.6)

where Q*t is potential GDP; UK*t is K*t as defined in equation (2.5) but multiplied by the utilization rate U; Lt is labor input; and Zt is real energy input.

02-A3495 7/27/05 7:12 AM Page 20

20

Chapter Two

Specializing equation (2.6) as a Cobb-Douglas function with constant returns to scale, one obtains the following: 6 Q*t  g1UK*t 2 aZtb 1Lt 2 1ab

(2.7)

where a is the elasticity of GDP with respect to utilized capital, b is the elasticity with respect to energy, (1  a  b) is the elasticity with respect to labor, and g is a scaling parameter. At this level of abstraction, the model does not include two short-run variables that account for part of the gap between potential and actual GDP: underutilized capital and cyclical variation in employment. Aggregate production functions are intended to show the relationship between actual output and utilized inputs. To estimate each year’s utilized capital stocks, we multiply the productivity-augmented stocks by aggregate utilization rates. An adjustment of this sort for labor input is not necessary because it includes only paid workers: The cyclical component is already built into the employment series. Reported GDP (Qt) surely contains some measurement error. Assuming that this error is distributed log-normally so that Qt  Q*t eet lt, one obtains the expression for measured GDP: Qt  g1UK*t 2 aZtbL1ab ee lt t

(2.8)

where UKt is an estimate of utilized capital, Lt is workers employed, and e el t is the lognormal disturbance. First dividing by Lt to express the production function in labor-intensive form and then taking logarithms, we obtain the regression equation ln qt  ln g  a ln uk*t  b ln zt  e lt

(2.9)

where qt  Qt /Lt, uk*t  UK*t /Lt, and zt  Zt /Lt. The productivity-augmented capital series must be constructed. We develop four alternatives based on embodied technical change ranging from 2 to 12 percent. Solow (1964, 127) reports what he characterizes as rough estimates of capital-augmenting technical progress in U.S. manufacturing industries as follows: 9.1 percent in furniture and fixtures, 5.9 percent in chemicals, 3.25 percent in fabricated metals, 3.89 percent in electrical machinery, but practically 0 percent in nonelectrical machinery. Since these industries produce the bulk of investment goods that Mexico imports from the United States, our range of rates covers what little is known about this territory. In estimating a production function, one should choose a procedure that encompasses the entire economic model, including derived-input demands. Following the classic paper by Marschak and Andrews (1944), many authorities believe that ordinary least-squares estimates of Cobb-Douglas parameters are biased and inconsistent. This need not be. In more rigorously specified theoretical models, Mundlak and Hoch (1965) and Zellner, Kmenta, and Dreze (1966) show that ordinary least-squares estimators are unbiased and consistent.7 Accordingly, we estimate the production function using ordinary least squares.

02-A3495 7/27/05 7:12 AM Page 21

Energy, Capital, and Productivity

21

data Aggregate output Qt is constant 1993 million pesos GDP, obtained from inegi (1983, 1985b, 1986b, 1997d, and 2001f ). We express the entire series 1965 – 2000 in terms of 1993 pesos.8 Aggregate energy, Zt, is commercial energy obtained from coal, oil, natural gas, natural-gas liquids, nuclear, hydroelectric, geothermal, wind, sugar cane, firewood, imported electric power, and trivial amounts of secondary energy (e.g., coke, kerosene, diesel). These data are published by the Secretary of Energy.9 Energy input is expressed in millions of BTUs.10 Aggregate capital services are gross capital stocks adjusted for utilization rates. Banco de México published the initial series for aggregate capital stocks and depreciation from 1960 to 1975.11 These stocks include construction (i.e., physical structures), machinery and equipment, transport equipment, and office machinery and furniture. We then added real annual investment in structures, machinery, and equipment to obtain aggregate capital stocks in constant 1993 pesos. We obtained the original investment series from Nacional Financiera, S.A. (1981) for the years 1940 –1980 and from Nacional Financiera (1995) and inegi (2001) for the years 1980 –2000. Capacity-utilization rates are obtained from Nacional Financiera, S.A. (1981, 1995), and the Food and Agriculture Organization of the United Nations (1983, 1988, 1993, 1997, 2002). Labor input is the average number of employed workers, published by inegi.12 The logarithms of real GDP per worker, utilized capital stocks per worker, and energy per worker are the bases for estimating equation (2.9). We first tested each series for stationarity using augmented Dickey-Fuller tests. The critical values to reject the null hypothesis of a unit root are 2.95 at P  0.05 and 3.63 at P  0.01. The calculated test statistics are 4.27 for ln qt, 3.41 for ln ukt, and 3.78 for ln zt. Accordingly, we proceed on the assumption that the three series are stationary.

statistical estimates Because the alternative capital-input series varies little according to different assumed rates of embodied technical change, the estimated coefficients are practically identical regardless of the series employed. To avoid a large, unnecessarily repetitious table, we present estimates based on four series: the ordinary capital-labor ratio (ln ukt) and capital-labor ratios with capital augmented by investment whose efficiency improves by 4, 8, and 12 percent (ln uk*.04, ln uk*.08, and ln uk*.12 2 . Estimates shown in table 2.1 yield some obvious generalizations: (1) The estimates of a and b are unaffected by different assumed rates of technical change; (2) these coefficients are estimated with reasonable precision (estimates of b are 2.8 times their estimated standard errors; estimates of a are about 2.5 times their standard errors); (3) estimated R2 of 0.965 shows that overall

02-A3495 7/27/05 7:12 AM Page 22

22

Chapter Two

Table 2.1. Summary of Estimates of Equation (2.9), 1965 –2000* Capital / (b) Energy labor series coefficient ln ukt ln uk*.04 ln uk*.08 ln uk*.12

(a) Utilized capital coefficient

AR(1) coefficient

0.114 (0.047) 0.114 (0.046) 0.114 (0.046) 0.114 (0.045)

0.822 (0.069) 0.822 (0.069) 0.822 (0.069) 0.822 (0.069)

0.348 (0.122) 0.347 (0.122) 0.347 (0.122) 0.347 (0.122)

R

BreuschGodfrey statistic

JarqueBera statistic

0.965

0.106

1.99

0.965

0.106

2.01

0.965

0.105

2.03

0.965

0.105

2.05

2

* Estimated standard errors are listed in parentheses. The estimate of b is significant at P  0.01 and that of a is significant at P  0.02.

goodness of fit is insensitive to different assumed rates of technical change; and (4) after correcting the autocorrelated OLS residuals by an AR(1) process, regression residuals show no evidence of autocorrelation or nonnormality. The asymptotic critical values of the Breusch-Godfrey and Jarque-Bera statistics are 5.99 at P  0.05. Although the estimates of b are three times larger than those of a, we cannot reject the hypothesis that the coefficients are equal: A Wald test of the null hypothesis a  b yields a calculated x2 of 2.85, well below the critical value of 5.99 to reject the hypothesis at P  0.05. From a statistical viewpoint, growth in utilized capital and energy are equally important in spurring labor productivity. The estimated elasticity of GDP with respect to labor is 0.538, with estimated standard error of 0.121.

sources of labor productivity growth, 1965 –1979 and 1979 – 2000 Real GDP per worker increased between 1965 and 1979 at an average annual rate of 3.7 percent and then stagnated. What are the roles of energy and capital per worker and technological change in productivity growth and its subsequent erosion? A partial answer can be found by comparing the estimates of productivity growth using the estimated parameters in table 2.1 and the actual rates of productivity growth. Estimated productivity growth is found by differentiating equation (2.9) with respect to time and substituting estimates of aˆ  0.114 and bˆ  0.348: d ln qˆ 1t2 dt

 aˆ a

d ln uk1t2 dt

d ln zt b  bˆ a b dt

(2.10)

Table 2.2 shows actual and estimated rates. For the 1965 –1979 period, the estimate is about half of the actual productivity growth. We suspect that the estimate is too low because of our restrictive assumption concerning technological change.13 Note also that estimated productivity growth using the UK*0.12

02-A3495 7/27/05 7:12 AM Page 23

Energy, Capital, and Productivity

23

Table 2.2. Actual and Estimated Growth in Real GDP per Worker (1965 –1979 and 1979 –2000) Actual growth rate of GDP per worker

Estimated growth rate of GDP per worker

1965 –1979: 3.70 percent

1965 –1979 Using UK(t) /L: 1.87 percent Using UK*0.12/L: 1.89 percent

1979 –2000: 0.1 percent

1979 –2000 Using UK(t) /L: 0.002 percent Using UK*0.12/L: 0.000 percent

series is only marginally larger than the estimate based on the UK(t) series because UK*0.12 increases only slightly faster than UK(t). From 1979 to 2000, however, estimates of stagnating productivity are right on target. Neither utilized capital nor energy per worker increased during these two decades, and productivity suffered accordingly. Improvement in macroeconomic productivity will surely require restoration of growth in capital and energy per worker.

summary and conclusions Real GDP per worker increased at 3.7 percent per year between 1965 and 1979, then stopped growing for 21 years. We believe these contrasting trends are ultimately traceable to two sets of financial disturbances: a series of peso devaluations beginning in late 1976 and fluctuating world oil prices. These perturbations had profound impacts on real macroeconomic variables. Quite apart from inducing cyclical recessions, they are at least partially responsible for the long-term decay in utilized capital and energy per worker. Decreasing capital and energy intensities, in turn, are largely responsible for 21 years of decaying productivity. To restore productivity growth, sustained increases in utilized capital and energy per worker will surely be required. Future productivity should also be strengthened by the accumulation of human capital: a better-educated, bettertrained labor force with steady, long-term employment. The payoffs to Mexico from these broadly based investments in human capital remain unknown because the investments have yet to be made.14 We believe that growth in utilized capital and energy per worker are bedrocks of sustainable productivity growth. Long-run growth in capital and energy per worker almost certainly entails relatively stable peso/dollar exchange rates. Fossil fuels— chiefly oil and natural gas—are the only feasible sources of additional domestic energy in Mexico for at least twenty-five years or so. To permanently increase production and consumption of these fuels, investments in energy infrastructure and reserves on an unprecedented scale will be essential. This is a daunting but feasible task. If accomplished, it will almost certainly be possible to restore productivity growth to a range of 2.5 to 3 percent annually.

02-A3495 7/27/05 7:12 AM Page 24

03-A3495 7/27/05 7:12 AM Page 25

chapter three

pemex Finances: Revenues and Taxes To understand Mexico’s oil and gas industry, one must first understand pemex finances. As a state-owned company, pemex controls neither its revenues nor the taxes it pays. The federal government determines both of these. Moreover, the federal government also largely determines how pemex allocates revenue among its many activities— exploration, development, production, exports, refining, and sales of gasoline and other refined petroleum products. This chapter explains how pemex obtains its annual drilling budgets from the government. These budgets are submitted by pemex Finance Corporative through the Secretary of Energy, then through the Secretary of Finance, and finally to the Mexican Congress. The budget is inevitably pared by Congress, then returned through the Secretary of Finance and the Secretary of Energy to pemex. This labyrinthine process is repeated year after year. This financial process is important for this reason: To understand the twenty-five-year evolution of pemex drilling activity requires an appreciation of the fiscal constraints under which this creature of the government operates. Because pemex depended entirely on federal funding until recently, its financial constraints have been linked to the government’s fiscal balance. The government, however, relies on pemex for one-third of its tax revenue. pemex taxes depend on its total sales revenues and the rate at which they are taxed. pemex’s tax burden is the ratio of taxes paid during the year to pemex sales revenues that year. pemex’s tax rate is thus the percentage rate at which its sales are taxed. Since 1989, these tax rates have exceeded 60 percent. Such extraordinarily high rates have been required as an essential source of funds for federal programs. The following chapter shows, however, that such high rates have seriously restricted pemex drilling activity and therefore its ability to develop new reserves.

pemex revenues and taxes pemex budgets are an important part of the overall federal financial plan. pemex’s capital investments must be approved each year by the Secretary of Finance and Congress. As a state-owned company, pemex can neither issue equity capital nor borrow money by selling bonds.1 The government’s commitment to maintaining low budget deficits has seriously constricted pemex’s investment opportunities. Since pemex must receive congressional approval to issue debt and cannot sell equity (common stock), its budgets are formulated in ways that are totally different from those of private companies.

25

03-A3495 7/27/05 7:12 AM Page 26

26

Chapter Three

Table 3.1. Total pemex Taxes Paid to Federal Government and the Dominant Role of Hydrocarbon Production Taxes, 1990 –2000 (Millions of Current Pesos) Year Total Hydrocarbon production taxes Taxes on imports Value added taxes Special tax on gasoline production and sales Others

1990

Percentage

1995

Percentage

2000*

Percentage

34,698.2 25,859.5

100.0 74.5

99,500.8 64,474.5

100.0 64.8

293,768.0 180,373.6

100.0 61.4

136.1 3,477.2 5,225.4

0.4 10.0 15.1

197.2 9,605.1 17,329.2

0.2 9.7 17.4

NA NA 60,810.0

– – 20.7

0.0

0.0

7,894.8

7.9

52,584.4

17.9

Source: SHCP (1995, 1998, 2000). * Preliminary information. NA means not available.

Table 3.2. Federal Public Investment in Petroleum Industry, 1990 –2000 (Millions of Current Pesos)*

Year

(1) Total public investment by federal government

(2) Energy sector

(3) pemex

(4)  (3)/(1) pemex percentage of total federal investment

1990 1995 2000

33939.3 53251.1 142721.0

12434.8 22107.1 51707.2

5795.2 14605.3 31304.2

17.1 27.4 21.9

Source: SHCP (1995, 1998, 2000). * Energy sector includes mining, electricity, and petroleum.

Table 3.1 shows the taxes pemex paid to the government in 1990, 1995, and 2000. Taxes on oil and gas production (shown in table 3.1 as hydrocarbon production taxes) account for the lion’s share of the total taxes, ranging from 74.5 percent in 1995 to 61.4 percent in 2000. Value-added taxes amount to only about 10 percent, and taxes on gasoline production and sales account for between 15 and 21 percent of the total. Until the advent of pidiregas (an acronym for Deferred Impact Projects in the Expenditures Registry), pemex investments were determined strictly by its annual government allocations. Table 3.2 lists the total public investment by the federal government, the amount allocated to the energy sector, and the amount allotted to pemex. Table 3.2 also shows that federal allocations to pemex rose from 17.1 percent in 1990 to 27.4 percent in 1995, then decreased to 21.9 percent in 2000.

pemex Taxes and Investments pemex recovers only a small fraction of its taxes by way of public investment by the federal government. The last column of table 3.2 shows that pemex received relatively small percentages of total public investment. A stronger

03-A3495 7/27/05 7:12 AM Page 27

pemex Finances: Revenues and Taxes

27

Table 3.3. Total Federal Taxes Collected from pemex (Millions of Current Pesos)

Year 1990 1995 2000*

Total taxes collected by federal government

Taxes collected from pemex

pemex Taxes as a percentage of total federal taxes

122666.2 280144.4 868267.6

34689.2 99500.8 293768.0

28.3 35.5 33.8

Source: SHCP (1995, 1998, 2000). * Preliminary data.

contrast can be made by comparing either total pemex taxes in 1995 (99,501 million pesos) to its federal public investment (14,605 million pesos) or its taxes in 2000 (293,768 million pesos) to its federal public investment (31,304 million pesos). Put differently, pemex recovered as federal public investment only 10.7 percent of the taxes it paid in 2000 (31,304 million pesos  293,768 million pesos  0.1066, or 10.7 percent). This growing gap between its taxes and its investment budget has forced delays in drilling. In real terms (adjusted for inflation), pemex’s investment in exploratory and development drilling decreased by the year 2000 to about 39 percent of its drilling investment in 1981.

pemex Taxes and Total Federal Taxes pemex is an essential source of federal tax revenue. As table 3.3 shows, pemex taxes were 28.3 percent of federal tax revenue in 1990, 35.5 percent in 1995, and 33.8 percent in 2000. The government could scarcely afford to lose such an important source of revenue by cutting pemex tax rates. Suppose that the government were to increase pemex budgets by reducing the company’s taxes. Conventional thinking is that the government would then have to face three politically unpopular alternatives. First, it could increase taxes elsewhere in the economy through a so-called national fiscal reform to compensate for the reduction in pemex tax revenue.2 Second, it could cut spending on other projects that are vital to Mexico’s social agenda. Finally, it could maintain spending on social programs by increasing its budget deficit. An increasing deficit could be financed in either of two ways: by increasing the growth rate of the money supply (which could be inflationary) or by selling bonds on the open market. The Fox administration has pledged to minimize deficits and strive for a balanced budget. In summary, pemex is an essential source of government tax revenue. The government is understandably reluctant to reduce pemex taxes.

the pemex budgetary process It is important to understand how pemex develops its annual proposals for drilling and how it attempts to obtain the resources to finance them.3 As figure 3.1 shows, the process begins at pemex Finance Corporative. These officials

03-A3495 7/27/05 7:12 AM Page 28

28

Chapter Three

Figure 3.1. Flow diagram of the complete process of pemex funding investment projects. .

design criteria that pemex employs to evaluate project proposals and then distributes these guidelines to all pemex subsidiaries. After receiving them, pemex Exploration and Production (pep) sends these guidelines to each of its four regions: south, north, marine southwest, and marine northeast. Officials in each region evaluate their exploration and development portfolios in view of these investment criteria, then submit their project proposals to pep. pep reviews the proposals and then submits an overall investment-program proposal to pemex Finance Corporative. pemex Finance Corporative authorizes the complete investment-program proposal for pemex (all of its subsidiaries) and submits it to the Secretary of Energy, who in turn presents the proposal to the Secretary of Finance. The Secretary of Energy then begins to negotiate the pemex budgets with the Secretary of Finance. The Secretary of Finance allocates resources to specific projects and presents its federal income-expense budget proposal to Congress.4 The allocations proposed for pemex are labeled with the name and budget for each project.5 Once Congress authorizes the detailed federal budget for the year, it notifies the Secretary of Finance, who informs the Secretary of Energy of the approved projects. Finally, the Secretary of Energy sends the list of approved projects to pemex Finance Corporative. Quite a tortuous process. The projects and their budgets are quite specific. If pemex wishes to shift investment from one project to another, the entire negotiation process must be repeated, thereby causing costly delays. Once pemex receives authorization for specific projects, these investments are all but written in stone. Thus it is essential that the Secretary of Energy and pemex agree on their priorities in order to obtain approval for the most critical projects. Typically, the Secretary of

03-A3495 7/27/05 7:12 AM Page 29

pemex Finances: Revenues and Taxes

29

Finance has approved fewer than half of pemex’s recommended investments, with the result that, until 1998, pemex was in financial straits that were strictly binding. The prolonged shortage of investable funds prompted a twenty-year drilling depression.

financial crises cause a drilling depression pemex drilled 432 wells in 1980 and 405 in 1981. However, as table 3.4 shows, in 1982, the peso was devalued by 133 percent, from $24.51 to $57.18 pesos per dollar. This devaluation prompted increases in real drilling costs and pressured Congress to further reduce pemex drilling budgets. Drilling dropped to 353 wells in 1982 and 305 wells in 1983 and continued to decline through 1986. The period 1981–1986 witnessed a brutal drilling recession. The worst was yet to come, however. The peso suffered sharp devaluations in 1986 and 1987, and drilling plummeted to 103 wells in 1987. During the first four years of the administration of President Carlos Salinas de Gortari (1989 –1992), the peso remained relatively

Table 3.4. $Peso/$U.S. Exchange Rates, 1970 –2000

Year

Peso exchange rate (pesos/US$)

1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

12.50 12.50 12.50 12.50 12.50 12.50 15.69 22.69 22.76 22.82 22.95 24.51 57.18 150.29 185.19 310.28 637.87 1405.80 2289.58 2483.37 2838.35 3016.15 3094.08 3.16 3.29 6.43 7.60 7.92 9.14 9.56 9.46

Political scenario (president’s name at the beginning of his six-year term)* Luis Echeverria Alvarez

José Lopez Portillo

Miguel de la Madrid Hurtado

Carlos Salinas de Gortari

Ernesto Zedillo

Vicente Fox Quesada

* The president takes office in December of the specified year.

03-A3495 7/27/05 7:12 AM Page 30

30

Chapter Three

500

Number of Wells Drilled

400

300

200

100

0 1975

1980

1985

1990

1995

2000

Year

Figure 3.2. Total wells drilled annually.

stable, federal budgets were more nearly balanced, and drilling made a weak recovery in 1991 and 1992. Nevertheless, it was short lived. Drilling declined to 78 wells in 1993, then fell to an all-time low of 63 wells in 1994. These events coincided with a radical revaluation of the peso in 1993 and the end of Salinas de Gortari’s presidency in 1994 (see table 3.4). By 1993 and 1994, the serious drilling recession of 1981–1986 had become a depression. Figure 3.2 illustrates the astonishing depth of this event. Drilling began to recover from 1995 (101 wells) through 1997 (121 wells). The recovery strengthened between 1998 (203 wells) and 2000 (247 wells), but these drilling rates were still far below those of 1980 and 1981. The drilling from 1998 to 2000 was chiefly financed by private loans, not by a restoration of federal financing. The magnitude of the drilling depression can be illustrated in two other ways. First, according to official records, pemex drilled 450 kilometers (281 miles) in exploratory wells in 1970, but only 160 kilometers (100 miles) in 2000. A second measure of the depression is revealed by the sharp decline in real (inflation-adjusted) drilling investment. Measured in terms of 2001 U.S. dollars, pemex investment decreased from $13.3 billion in 1981 to $5.2 billion

03-A3495 7/27/05 7:12 AM Page 31

pemex Finances: Revenues and Taxes

31

in 2000. Thus, measured in real economic terms, pemex drilling in 2000 amounted to only 39 percent of that in 1981. pemex experienced a twenty-year drilling depression caused by a combination of tighter federal budgets and rising drilling costs. Even during the years of recovery, 1996 –2000, pemex drilled an average of 184 wells per year. This drilling rate is only 44 percent as high as the 419 wells per year in 1980 and 1981. Since a continuous program of successful drilling is the only way to replenish reserves, the major reason for a twenty-year decline in reserves is clear. Raúl Muñoz Leos, pemex general director, has stated that pemex would need to invest approximately $9 billion annually to stem the decline in reserves (June 28, 2001). Then, on November 27, 2001, he stressed that, given the company’s financial resources, pemex could invest $12 billion dollars per year (2002 –2006) in profitable projects that pemex prepared and evaluated. Assuming little or no inflation in real drilling costs, an annual investment of $12 billion is more than twice the rate of pemex’s actual drilling investment in 2000. Even if $12 billion were made available annually to pemex strictly for drilling, in real terms it would still fall short of the $13.3 billion pemex invested in 1981.6

pidiregas: a new source of funds and debt Before 1996, pemex investments were strictly governed by its annual allocations.7 Financial turmoil began in December, 1994, causing an abrupt reduction in public revenues and budgets. The Mexican Constitution was amended in December, 1995, when Congress reformed both Article 30 of the Federal Budget, Accounting, and Public Expenditures Law and Article 18 of the Public Debt Law. These reforms created the pidiregas program as an additional way to finance pemex projects. pidiregas investments were designed to achieve several objectives: • Permit long-term private loans to develop strategically profitable projects (this was the first time such private financing had been available) 8 • Develop more flexible governmental budgetary instruments to finance strategic projects • Approve multiyear projects deemed to be in the national interest that would generate sufficient cash flows to cover future repayment of debt 9 pidiregas allowed pemex access to private loans beginning in 1997, as figure 3.3 shows, and pemex began to borrow heavily through pidiregas in 1999 and 2000. pidiregas was designed with the laudable intent of allowing pemex greater flexibility to finance investments. However, the use of these private loans has been catastrophic. Repayment of interest and principal incurred from the additional debt has proven to be infeasible. In 2003, pemex stated that the debt burden attributable to its borrowing to finance pidiregas investments was unsustainable.10

03-A3495 7/27/05 7:12 AM Page 32

32

Chapter Three

Figure 3.3. pemex financing strategy since 1997.

By December 31, 2001, pemex had received authorization for twelve pidiregas projects.11 Total investments made under the pidiregas program were $45.7 billion pesos, of which 78.5 percent was invested in drilling. Loans initiated by pemex through pidiregas were the major source of funds for drilling from 1999 to 2001. pemex reports that 80 percent of its investment during this period was financed through pidiregas and only 20 percent through federal allocations. The renewal of drilling financed through pidiregas came at a steep price because the additional debt caused major deterioration in pemex’s balance sheet. Table 3.5 shows the consolidated balance sheet for December 31, 2000, and December 31, 2001. Long-term debt increased by 45 billion pesos, from 317 billion to 362 billion. This increase was chiefly attributable to borrowing made possible by pidiregas. The 45-billion peso growth in long-term debt was partly offset by a 24-billion peso reduction in short-term liabilities. Nonetheless, total pemex equity decreased by some 27.7 billion pesos, from 150.6 billion on December 31, 2000, to 122.9 billion pesos on December 31, 2001. The financial deterioration continued between 2001 and 2002. Table 3.6 shows the consolidated balance sheet on December 31, 2001, and December 31, 2002. Long-term debt increased by some 120.8 billion pesos, from 362.1 billion to 482.9 billion. Although pemex’s total assets increased by 140.5 billion pesos, its total liabilities rose to about 162.7 billion. Hence total equity decreased from 122.9 billion to 100.7 billion pesos. This continued decline in pemex equity is almost entirely attributable to growth in long-term loans initiated through pidiregas.

03-A3495 7/27/05 7:12 AM Page 33

Table 3.5. pemex Consolidated Balance Sheet as of December 31, 2000 and 2001 (Millions of Pesos) 2000

Assets Current assets Cash and cash equivalents Accounts, notes receivable, and other Inventories, net Fixed assets Other assets Total assets Liabilities Short-term Long-term Total liabilities Equity Total equity Total liabilities and equity

2001

Change

Amount

%

Amount

%

Amount

%

109393 27827

19.4 4.9

76534 14442

13.7 2.6

(32859) (13385)

(30.0) (48.1)

56702

10.1

44869

8.0

(11833)

(20.9)

24854 388225 65850 563468

4.4 68.9 11.7 100.0

17223 406913 73436 556883

3.1 73.1 13.2 100.0

(7641) 18688 7586 (6585)

(30.7) 4.8 11.5 (1.2)

95766 317096 412862

17.0 56.3 73.3

71948 362069 434017

12.9 65.0 77.9

(23818) 44973 21155

(24.9) 14.2 5.1

150606 563468

26.7 100.0

122866 556883

22.1 100.0

(27740) (6585)

(18.4) (1.2)

Consolidated audited financial statements prepared in accordance with the generally accepted accounting principles issued by the Instituto Mexicano de Contadores Publicos. For inflation recognition, governmental rule NIF-06-Bis “A,” Section A, was considered. The financial statements of the subsidiary companies were considered.

Table 3.6. pemex Consolidated Balance Sheet as of December 31, 2001 and 2002 (Millions of Pesos) 2001 Amount Assets Current assets Cash and cash equivalents Accounts, notes receivable, and other Inventories, net Fixed assets Other assets Total assets Liabilities Short-term Long-term Total liabilities Equity Total equity Total liabilities and equity

2002

Change

%

Amount

%

Amount

%

76534 14442

13.7 2.6

123654 43877

17.7 6.3

47120 29435

61.6 203.8

44869

8.0

55372

7.9

10503

23.4

17223 406913 73436 556883

3.1 73.1 13.2 100.0

24405 486098 87628 697380

3.5 69.7 12.6 100.0

7182 79185 14192 140497

41.7 19.5 19.3 25.2

71948 362069 434017

12.9 65.0 77.9

113771 482913 596684

16.3 69.3 85.6

41823 120844 162667

58.1 33.4 37.5

122866 556883

22.1 100.0

100696 697380

14.4 100.0

(22170) 140497

(18.0) 25.2

Audited consolidated financial statements prepared in accordance with the generally accepted Mexican accounting principles, issued by the Instituto Mexicano de Contadores Publicos. For inflation recognition, governmental rule NIF-06-BIS “A,” Section A, was considered. This information consolidates the financial statements of the principal subsidiary companies.

03-A3495 7/27/05 7:12 AM Page 34

34

Chapter Three

the unusual pemex tax burden Taxes levied on a private firm’s profits or net income are linked to the firm’s financial performance. During unprofitable years, taxes are zero. In some instances, tax credits based on negative net income can be carried forward to reduce taxes incurred in profitable years. Taxes based on a firm’s total sales are fundamentally different because they are not linked to financial performance. These taxes must be paid each year, regardless of the firm’s net income or loss. pemex taxes are based on its total sales, not its net income. And pemex taxes are an extremely high percentage of sales. The pemex tax burden in each year t is the percentage of its sales paid as taxes to the federal government. It can be calculated as pemex Tax Burden t  [Taxes and duties paidt  Total salest] 100. This tax burden can be illustrated by using pemex’s consolidated-income statements for 2000 and 2001 (table 3.7). For example, total sales in 2000 amounted to 468,268 million pesos, and taxes and duties were 293,768 million pesos. Thus pemex Tax Burden2000  [293,768 million pesos  468,268 million pesos] 100  .6273 100  62.73 percent. It is striking that, in both years, pemex paid more in taxes and duties than its income before taxes, duties, and the adoption of a new financial standard. As a consequence, each year’s net income was negative. In fact, pemex’s taxes far exceeded its total costs and operating expenses. The additional debt and interest payments pemex incurred through pidiregas became serious financial burdens. The consolidated-income statements in table 3.7 show that net interest payments nearly doubled, from 6,652 million pesos in 2000 to 13,104 million in 2001. These larger interest payments are themselves a cause for concern. Nevertheless, 13,104-million-peso interest payments pale by comparison with 263,462 million pesos in taxes. pemex’s taxes in 2001 were twenty times greater than interest payments. pemex is in financial straits for three reasons. First, its federal allocations were slashed because of fiscal austerity. Second, the loans enabled by pidiregas proved to be disastrous: Borrowing increased pemex’s long-term debt to levels that threaten the company’s financial solvency. Third, pemex taxes have approximated 60 percent of its total sales every year since 1989. It is widely recognized that such high rates are unsustainable. Although efforts to reduce them are at the forefront of public debate, the problem is not yet resolved. A direct consequence of these financial strictures is a drilling depression spanning two decades. This drilling depression is the root cause of Mexico’s declining oil and gas reserves. But reserves are the essential source of future production. The 20-year decline in oil reserves must be reversed. Otherwise, Mexico will lose its capacity to produce oil. The 20-year shrinkage in gas reserves must also be reversed. If not, Mexico will become ever more dependent on imported gas. The key—the only key—to rebuilding reserves is to revive successful drilling. This will require sustained annual investment in real terms at least three or four times larger than investments during the late 1990s.

03-A3495 7/27/05 7:12 AM Page 35

pemex Finances: Revenues and Taxes

35

Table 3.7. pemex Consolidated Income Statements, 2000 and 2001 (Millions of Pesos) Change 2000

Total Sales Exports Domestic sales** Costs and operating expenses Cost of the reserve for retirement payment, pensions, and indemnities Income before taxes, duties, interest, and other expenses Net interest Other income and expenses Income before taxes, duties, and cumulative effect of adoption of new financial instruments standard Taxes and duties** Cumulative effect of adoption of new financial instruments standard Net income (loss)

2001

Nominal

Real*

Amount

%

Amount

%

Amount

%

%

468268 175387 292881 162740

100.0 37.5 62.5 34.7

445330 141477 303853 173079

100.0 31.8 68.2 38.9

(22938) (33910) 10972 10339

(4.9) (19.3) 3.7 6.4

(10.6) (24.2) (2.5) (0.0)

29902

6.4

33849

7.6

3947

13.2

6.4

275626

58.9

238402

53.5

(37224)

(13.5)

(18.7)

6652

1.4

13104

2.9

6452

97.0

85.2

(5084) 274058

(1.0) 58.5

(5404) 230702

(1.2) 51.8

(320) (43356)

6.3 (15.8)

(0.1) (20.9)

293768

62.7

263462

59.2

(30306)

(10.3)

(15.7)

(19710)

(4.2)

1331 (34091)

0.3 (7.7)

1331 (14381)

100.0 73.0

100.0 62.6

Consolidated audited financial statements prepared in accordance with the generally accepted accounting principles issued by the Instituto Mexicano de Contadores Publicos. For inflation recognition, governmental rule NIF-06-Bis “A,” Section A, was considered. The financial statements of the subsidiary companies were consolidated. The consolidated, audited financial statements of Petroleos Mexicanos, subsidiary entities, and subsidiary companies are part of Form 20-F for registration with the Securities and Exchange Commission. * The average inflation used is 6.368 percent. ** Includes the Special Tax on Production and Services (IEPS).

the financial dilemma The federal government and pemex together face a financial dilemma. The government relies on pemex for one-third of its tax revenues, which it obtains by taxing pemex sales at rates approximating 60 percent. Taxes collected from pemex in 2001 amounted to 263.5 billion pesos, or, using an exchange rate of 11.3 pesos per U.S. dollar, about US$23.3 billion.12 In return, the government provided pemex investment allocations of 31.3 billion pesos, or roughly US$2.8 billion. pemex will require at a minimum US$13.5 billion annually just for investment in exploration and development and probably closer to US$20 billion for total investment, which includes investment in pipeline infrastructure, refineries, and construction of new gas-fired electric plants.

03-A3495 7/27/05 7:12 AM Page 36

36

Chapter Three

How can the huge gap between the US$20 billion that pemex needs for investment and the US$2.8-billion investment allocation provided in 2001 possibly be filled? pemex officials and the Secretary of Finance have stated emphatically that private-sector (pidiregas) loans are not a viable source of funds. The government might be able to close the gap, but to do so would require a radical restructuring of federal finances. A few numbers tell the realities: US$20 billion is approximately 226 billion pesos. To provide pemex 226 billion pesos annually for total investment would require 194 billion additional pesos each year and a radical cut in pemex taxes. In 2001, the government collected 263.5 billion pesos from pemex (see pemex consolidated-income statements in table 3.8). If the government had reduced these taxes by 194 billion pesos and required pemex to invest the entire tax savings, it would have collected taxes of only 69.5 billion instead of the 263.5 billion pesos actually collected in 2001.

Table 3.8. pemex Consolidated Income Statements, 2001 and 2002 (Millions of Pesos) Change 2001

Total sales Exports Domestic sales** Costs and operating expenses Cost of the reserve for retirement payments, pensions, and indemnities Income before taxes, duties, interest, and other expenses Net interest Other income and expenses Income before taxes, duties, and cumulative effect of adoption of new financial instruments standard Taxes and duties** Cumulative effect of adoption of new financial instruments standard Net income (loss)

2002

Nominal

Real*

Amount

%

Amount

%

Amount

445330 141477 303853 173079

100.0 31.8 68.2 38.9

481437 167166 314271 167773

100.0 34.7 65.3 34.8

36107 25689 10418 (5306)

8.1 18.2 3.4 (3.1)

2.9 12.5 (1.5) (7.7)

33849

7.6

37135

7.7

3286

9.7

4.5

238402

53.5

276529

57.5

38127

16.0

10.4

2.9 (0.1)

669 5063

5.1 (93.7)

0.1 (94.0)

13104 (5405)

2.9 (1.2)

13773 (342)

%

%

230703

51.8

263098

54.7

32395

14.0

8.6

263462 (1331)

59.2 (0.3)

293590 0

61.0 0.0

30128 1331

11.4 (100.0)

6.1 n /a

(34090)

(7.7)

(30492)

(6.3)

3598

(10.6)

(14.8)

Audited, consolidated financial statements prepared in accordance with the generally accepted accounting principles issued by the Instituto Mexicano de Contadores Publicos. For inflation recognition, government rule NIF-06-Bis “A,” Section A, was considered. This information consolidates the financial statements of the principal subsidiary companies. * The inflation average used for the calculation was 5.03 percent. ** Includes the Special Tax on Production and Services (IEPS). For 2001, it was 95,199 million pesos, and for 2002, it was 114,491 million pesos.

03-A3495 7/27/05 7:12 AM Page 37

pemex Finances: Revenues and Taxes

37

The actual pemex tax burden in 2001 was 263.5 billion pesos/445.3 billion pesos, or 59.2 percent. However, if the government had instead collected only 69.5 billion pesos, the pemex tax rate would have been 69.5 billion pesos/ 445.3 billion pesos, or 15.6 percent. In this hypothetical example, the additional 194 billion pesos pemex needed for investments would have been funded entirely through tax cuts. The predicament is painfully clear: With all other tax rates held constant, 194 billion more for pemex means 194 billion less for other federal programs. Since total federal taxes in 2000 amounted to 868.3 billion pesos (see table 3.3), a tax cut of 194 billion pesos is a whopping 22.3 percent reduction in total federal taxes. Assuming a balanced budget, this loss of 194 billion pesos in taxes would require almost unthinkable retrenchments in other government programs. Thus, providing pemex with investment budgets required for sustainable energy development would apparently require the following: (1) radical cuts in other government programs, (2) sharp increases in tax rates elsewhere in the economy, (3) larger federal-budget deficits, or (4) a massive increase in multiple-service contracts, which would still require larger pemex budgets.

fossil fuels for sustainable development We have stressed the importance to Mexico of rebuilding its oil and gas reserves. The reason is transparent: Larger reserves are the only means of increasing future production. Furthermore, substantial increases in oil and gas production are the very foundations of sustainable energy development for at least the next twenty-five years and probably even longer. The Fox administration announced production targets of 1,414 million barrels of oil and 2,810.5 BCF of gas in 2006.13 pemex released production targets of 1,460 million barrels of oil and 2,190 BCF of gas in 2006.14 pemex actually produced approximately 1,380 million barrels of oil in 2003. Accordingly, the production targets for oil appear to be attainable, but those for gas will be much tougher to attain. According to the natural gas bylaws (Article 109), every year the Secretary of Energy must publish Mexico’s natural-gas needs for the following ten years.15 In 2002, the Secretary of Energy suggested an annual average growth rate of 7.34 percent in domestic gas production for the period from 2003 to 2010. We believe that, without sustained increases in gas reserves, such production growth will be impossible. Chapter 1 shows that increased oil and gas production are the only realistic sources of additional energy for at least the next twenty-five years. The reasons are indisputable. Nuclear energy accounts for only 4 percent of electricity, and no additional nuclear plants are now being planned. Hydroelectricity accounts for another 14 percent, and the 750-megawatt El Cajón plant will increase the country’s total generating capacity by about 1.8 percent. New geothermal and wind-power projects may conceivably increase total generating capacity by another 5 – 6 percent. Even so, to almost double the currently installed generating capacity, the burden must be borne by oil and gas.

03-A3495 7/27/05 7:12 AM Page 38

38

Chapter Three

Fossil fuels— oil, natural gas, and coal—now account for 92 percent of Mexico’s total primary energy (table 1.2). Of this total, coal contributes a scant 4.5 percent. Like it or not, oil and gas are the only feasible fuels to support sustainable growth in the foreseeable future. Nuclear power cannot do it. Hydroelectric power cannot do it. Nonhydroelectric renewable resources cannot do it. These alternative sources of energy can at best play supporting roles of secondary importance.

04-A3495 7/27/05 7:12 AM Page 39

chapter four

Exploration and Development Drilling

Chapters 4 –7 develop an integrated model of oil and gas supply in Mexico. The model is based on earlier work by Moroney (1997), who specifies a vertically integrated system of reserves and production in Texas. A major advantage of the integrated system is its transparency. The links between exploration and development drilling, successful drilling, gross reserve additions, reserves, and production are explicit. Because the entire model is driven by a single financial variable, it is well suited for conducting tax-policy experiments, as chapter 8 explains. Our model provides explicit econometric linkage between three integrated sectors: drilling, reserve additions, and production. The entire model is driven by a single financial variable: the net (after-tax) price of Mexican oil exports. The context of our work here differs fundamentally from Moroney’s earlier work. The first difference is institutional: Moroney’s model applies to the state of Texas, where oil and gas are produced by thousands of unregulated private operators intent on maximizing profit. By contrast, only pemex produces oil and gas in Mexico. These institutional differences lead to important variations in the models for Texas and Mexico. Texas and Mexico also differ in terms of geology. Texas is a geologically mature region in which oil and gas have been produced for more than a century. Its oil and gas reserves are in a state of irreversible decline. By contrast, Mexico’s oil and gas industry is geologically youthful, with reserves that remain to be fully developed. Natural-gas resources in Mexico have scarcely been explored. Nearly 80 percent of Mexico’s oil is produced offshore, whereas only 1 percent of Texas production occurs offshore. Offshore and onshore technologies differ in important ways that account for major differences in drilling success, reserves, and production per successful well. The net price of Mexican oil exports is the financial variable that drives our model. This net price is the prevailing market price, determined by world supply and demand, then reduced by the pemex tax burden. It is therefore partly determined by the exogenous world price and partly by the endogenous tax rate. pemex produces both oil and gas. The after-tax price of oil is the relevant financial variable because natural gas has thus far been a small fraction of pemex’s production revenue. This after-tax price is a convenient key for financial policy because it can be changed by altering pemex tax rates. It is useful here to preview the research in chapters 4 –7 as links in the integrated model. The following flow chart shows the chain of causality.

39

04-A3495 7/27/05 7:12 AM Page 40

40

Chapter Four

This chapter and those that follow present a systematic econometric study of these vertically integrated activities from 1975 to 2000.

trends in drilling, 1975 – 2000 Drilling is separated into two sectors: exploration and development. Exploration occurs in unproven territory and is therefore much riskier than development. Following successful exploration, development wells are drilled in a newly proven territory in order to develop its reserves and to prepare them for production. Development wells are characterized by considerably lower geological (dry-hole) risk than exploration wells. Between 1975 and 1980, exploratory drilling was stable, ranging from 79 to 87 wells per year. As figure 4.1 illustrates, exploration declined in 1981 (63 wells), then remained reasonably stable through 1986. This decline of 1981–1986 was prompted in part by declining oil prices and in part by a rising pemex tax burden, which reached 50 percent for the first time in 1980. Following the sharp peso devaluation and the fiscal crisis of 1986, exploration plummeted to 27 wells annually from 1987 to 1989. There was a mild recovery until 1991, followed by an abrupt decrease to only 10 wells annually from 1995 to 1997. These were years of depressed oil prices and a pemex tax burden averaging 63.8 percent. The time series on exploration wells appears to be stationary. The augmented Dickey-Fuller test statistic is 3.32, compared to a critical value of 2.98 to reject the null hypothesis of a unit root at P  0.05. As figure 4.2 shows, development drilling remained stable from 1975 to 1979, then increased sharply from 1980 to 1981, following the increase in world oil prices. As oil prices weakened from 1982 to 1986, the number of devel-

04-A3495 7/27/05 7:12 AM Page 41

Exploration and Development Drilling

41

90 80

Exploratory Wells Drilled

70 60 50 40 30 20 10 0 1975

1980

1985

1990

1995

2000

Year Figure 4.1. Exploratory wells drilled annually, 1975 –2000.

opment wells declined steadily. Then, following the financial crisis of 1986, development drilling plunged to only 76 wells annually from 1987 to 1989. Development drilling remained depressed until a mild recovery in 1995, coinciding with an increase in world oil prices and weighted-average net Mexican oil prices (see chapter 7, table 7.1). The time series covering development wells may also be stationary. The augmented Dickey-Fuller test statistic is 1.70, which can be compared with critical values of 1.61 at P  0.10 and 1.95 at P  0.05. The null hypothesis of a unit root can be rejected at P  0.10 but not at P  0.05. Through the year 2000, exploration and development never recovered to the higher rates between 1975 and 1981, when exploration averaged 80 wells annually and development averaged 274 wells per year. From 1995 to 2000, exploration averaged 18 and development 152 wells annually. As chapter 3 stresses, this depression in drilling was caused primarily by ever-tightening financial constraints. It was not brought on by a lack of productive opportunities. Abundant oil and gas are undoubtedly awaiting discovery in Mexico, a fact proven by the exceedingly large reserves found by successful wells. Productive opportunities were abundant, but pemex was bound through the end of the century by a financial straitjacket.

04-A3495 7/27/05 7:13 AM Page 42

42

Chapter Four

360 320

Development Wells Drilled

280 240 200 160 120 80 40 1975

1980

1985

1990

1995

2000

Year Figure 4.2. Development wells drilled annually, 1975 –2000.

data Mexico produces three grades of crude oil: Maya-22 is a heavy, high-sulfur oil; Isthmus-34 is a lighter, low-sulfur oil; and extra-light Olmeca-39 is the topgrade oil. Maya-22 accounts for more than half of total production and approximately 87 percent of exports. The Mexican oil-export price is a volumeweighted average price of Maya, Isthmus, and Olmeca oil prices, published by Banamex (1980), Secretaría de Programación y Presupuesto (1979), pemex (1979, 1995, 1996a, 1999a, 2000a, 2001a), and Banco de México (1998). The net Mexican oil price (NMOP) is this weighted-average price reduced by the pemex tax burden. When it tallies successful wells drilled each year, pemex does not distinguish between oil and gas wells. pemex classifies each well it drills as either “exploratory” or “development” and reports successful and unsuccessful wells in each category. pemex published these data at five-year intervals from 1945 to 1974 but began publishing drilling reports annually in 1975. Accordingly, we use the sample period from 1975 to 2000. Wells are reported as having been drilled during the year in which they are completed. Thus, if a well is begun in one year but completed in the next, it is reported in the latter year.

04-A3495 7/27/05 7:13 AM Page 43

Exploration and Development Drilling

43

specification and theory We estimate separate models for exploration and development drilling. This distinction is important for two reasons. First, development wells carry far less geological risk (of being unsuccessful) than exploration wells.1 pemex may emphasize the development of existing reserves in one year but exploring for new reserves in another. It is important to keep these distinctive motives in mind when examining the data. A second reason is that far more development wells are drilled each year. From 1975 to 2000, the sample averages are 224 development wells and 48 exploration wells drilled annually. It is more informative to maintain these distinctions than to aggregate the two series. We estimate linear models that allow for incomplete adjustment to the desired level of drilling within the year.2 There are compelling reasons for treating the actual number of wells drilled as observations that represent adjustments toward long-run equilibrium rather than as equilibrium values. The first is that pemex cannot negotiate drilling contracts until it receives an approved drilling budget. A second reason, discussed in chapter 3, is that approved budgets are limited to specific projects. pemex has practically no flexibility to shift capital from one project to another within the year. The consequences of drilling constraints in one year almost certainly affect drilling in subsequent years. In short, actual drilling is constrained not only by the size of drilling budgets but also by their allocation over time by Congress. Consider the following partial adjustment model. Let Y *1t2 be the desired long-run rate of exploratory or development drilling in year t, and suppose that Y*t  a  b1NMOPt 2

(4.1)

where a 0, b 0, and NMOPt is the net Mexican oil-export price. Suppose that the actual rate adjusts to the desired rate according to the stochastic process: Yt  Yt1  11  d 2 1Y*t  Yt1 2  et

(4.2)

where 0  d  1, (1  d) is the coefficient of adjustment, and et is an error term with E(et)  0. Consider three possibilities regarding d: 1.

If d  0, Yt  Y*t  et , E(Yt)  Y*t and complete adjustment occurs within one year, subject only to random error.

2.

If d  1, Yt  Yt1  et ; thus, apart from the random error, Yt is equal to Yt1. Such a finding would mean that drilling remains almost constant year after year, fluctuating only because of the random error, et.

3.

If 0  d  1, the number of wells actually drilled adjusts partially toward a long-run equilibrium. For instance, if d  0.5, then (1  d)  (1  0.5)  0.5, or 50 percent of the adjustment occurs in one year. After

04-A3495 7/27/05 7:13 AM Page 44

44

Chapter Four

two years, (1  d)  d(1  d)  1  d2  1  0.25  0.75, or 75 percent of the adjustment is achieved; and after three years, 1  d3  1  0.125  0.875, or practically full adjustment is accomplished. Substituting equation (4.1) into equation (4.2) yields Yt  11  d2a  11  d2bNMOPt  dYt1  et

(4.3)

which can be expressed as the following regression model: Yt  b1  b2NMOPt  b3Yt1  et

(4.4)

Estimates of the coefficient of adjustment (1  d), the constant term a, and the long-run equilibrium coefficient b can be obtained from the regression coefficients as follows: bˆ 3  dˆ , bˆ 1  11  d2a;

aˆ 

bˆ 2  11  d2b;

bˆ 

bˆ 1 1  dˆ bˆ 2 1  dˆ

 

bˆ 1 1  bˆ 3 bˆ 2 1  bˆ 3

The long-run response in drilling to a change in NMOP is estimated from bˆ  bˆ 2/11  bˆ 3 2 . Econometrically, bˆ is an estimate of the response in long-run equilibrium drilling to changes in NMOP. An estimate, bˆ 3, which is not significantly different from 0, indicates that complete adjustment occurs within one year. Intuitively, the number of wells drilled last year, Yt1, has no influence on the number of wells drilled this year, Yt. However, if bˆ 3 is significantly positive, then dˆ is statistically less than 1, and adjustment of actual to desired drilling is partially complete within one year. Equation (4.4) is estimated as follows: For exploratory drilling: Y Et  b1  b2NMOPt  b3Y Et1  e1t

(4.5)

For development drilling: Y Dt  b1  b2NMOPt  b3Y Dt1  e2t

(4.6)

One would like to obtain consistent estimates of b1, b2, and b3. If the disturbance term is not autocorrelated, then the ordinary least-squares estimates are consistent. If the disturbance term is normally distributed, then ordinary least-squares estimates are also distributed normally.3 Accordingly, we use the BreuschGodfrey test for serially correlated error terms.4 We then test for normally distributed error terms, using a test statistic proposed by Jarque and Bera (1981). If the Jarque-Bera statistic is less than the critical value (5.99 with two

04-A3495 7/27/05 7:13 AM Page 45

Exploration and Development Drilling

45

degrees of freedom at the 5-percent-significance level), one can reject the hypothesis of nonnormal error terms.

statistical estimates for exploration The exploration model is TEWDt  b1  b2NMOPt  b3TEWDt1  elt

(4.7)

where TEWDt  number of exploratory wells drilled in year t NMOPt  net Mexican oil-export price in year t TEWDt1  number of exploratory wells drilled lagged one year e1t  disturbance term Estimated coefficients are shown in table 4.1. The coefficient of TEWDt1, bˆ 3  dˆ  0.612, is significant in the interval 0  d  1, showing that adjustment is incomplete within one year.5 An economic interpretation is straightforward. About 39 percent of the desired long-run adjustment occurs in one year, but complete adjustment to long-run equilibrium drilling requires roughly four years.6 The estimated coefficient of NMOP, bˆ 2  1.733, is significant at P  0.05. This coefficient has a clear short-run interpretation: For every one dollar increase in NMOP, approximately 1.73 additional exploratory wells will be drilled during that year. The long-run price response is obtained from bˆ 

bˆ 2 1  bˆ 3



1.733  4.47 0.388

The long-run response is more than twice the short-run response. For every dollar increase in NMOP, about 4.5 more exploratory wells will be drilled annually in long-run equilibrium. The short-run elasticity of exploration drilling with respect to NMOP, evaluated at sample means, is 0.31. The long-run

Table 4.1. Exploration Drilling Model Dependent Variable: TEWD Variable

Coefficient

Std. Error

t-Statistic

Probability

C NMOP TEWD(1) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood Durbin-Watson stat.

1.337312 1.732836 0.612384 0.814114 0.797950 11.91256 3263.908 99.71599 1.600362

5.980561 0.223610 0.789533 2.194760 0.092809 6.598362 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion F-statistic Probability (F-statistic)

0.8250 0.0385 0.0000 47.88462 26.50181 7.901230 8.046395 50.36587 0.000000

04-A3495 7/27/05 7:13 AM Page 46

46

Chapter Four

equilibrium elasticity is 0.80. Thus a permanent increase of 20 percent in NMOP would stimulate approximately 16 percent more exploratory wells drilled annually in long-run equilibrium. The Breusch-Godfrey statistic (1.091) allows us to reject the hypothesis of first-order autocorrelated residuals. The Jarque-Bera statistic (1.067) is well below the critical value of 5.99 required to reject normality. In summary, the exploration regression model seems to be well specified, and its economic implications are plausible. Equation (4.7) captures the dynamics of exploration drilling quite well. NMOP is the key financial variable from the viewpoint of tax policy because it includes both the exogenous Mexican oil-export price and the endogenous cash-flow constraint embodied in the pemex tax burden.

statistical estimates for development The regression for development drilling is given by TDWD t  b1  b2NMOPt  b3TDWD t1  e2t

(4.8)

where TDWDt  number of development wells drilled in year t NMOPt  net Mexican oil-export price in year t TDWDt1  number of exploratory wells drilled lagged one year e2t  disturbance term Estimated coefficients are shown in table 4.2. The estimated coefficient of TDWDt1, bˆ 3  dˆ  0.553, shows that about 1  0.553  0.447 (or 45 percent) of adjustment to long-run equilibrium occurs in one year.7 This coefficient is statistically significant in the interval 0  d  1. Adjustment to long-run equilibrium drilling requires about four years.8

Table 4.2. Development Drilling Model Dependent Variable: TDWD Variable

Coefficient

Std. Error

t-Statistic

Probability

C NMOP TDWD(1) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood Durbin-Watson stat.

3.682569 8.348142 0.552997 0.810785 0.794331 41.27727 39187.70 132.0267 1.459614

20.54645 0.179231 3.488792 2.392846 0.130490 4.237844 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion F-statistic Probability (F-statistic)

0.8593 0.0253 0.0003 176.5385 91.01790 10.38667 10.53183 49.27734 0.000000

04-A3495 7/27/05 7:13 AM Page 47

Exploration and Development Drilling

47

The coefficient of NMOP, bˆ 2  8.348, is significant at P  0.05. For every dollar change in NMOP, development drilling responds by about 8.35 wells that year. The one-year elasticity of development drilling with respect to NMOP, evaluated at sample means, is 0.40. The long-run price response is much larger: bˆ 

bˆ 2 1  bˆ 3



8.348  18.68 0.447

Thus, a permanent one-dollar change in NMOP prompts a long-run equilibrium response of about nineteen development wells drilled annually. The longrun elasticity of development drilling with respect to NMOP, evaluated at sample means, is 0.91. The Breusch-Godfrey statistic is 2.219, and the Jarque-Bera statistic is 0.943. The hypotheses of independent, normally distributed disturbances are acceptable. The regression model for development drilling appears to be adequately specified, and its economic implications make a great deal of sense.

forecasts The drilling equations are the foundations for the remainder of the vertically integrated model. Thus it is important to know how accurately these equations are able to predict drilling outside of the sample employed for estimation. The forecasts are complicated by the fact that roughly 80 percent of the development drilling in 1998, 1999, and 2000 was financed by loans made to pemex from the private sector. Recall that loans initiated through the pidiregas program were intended to be self-liquidating. Loans were restricted to projects with high probabilities of future cash flows sufficient to cover interest payments and retirement of the loan principal. Loans for high-risk investments were proscribed. Accordingly, pemex could borrow from the private sector chiefly to finance development drilling. We reestimate equations (4.7) and (4.8) using the subsample 1975 –1998, then forecast drilling in 1999 and 2000. These predictions cast some light on the reliability of these foundations for the simulations that we describe in chapter 8. Table 4.3 presents estimates of equation (4.7) based on the subsample 1975 –1998. The estimated coefficients are quite similar to the full-sample estimates in table 4.1. The forecasts of exploration wells in 1999 and 2000 are shown in table 4.4. The combined 1999 –2000 forecast of 53 wells slightly underestimates the 59 wells actually drilled in those two years. The mean absolute error (MAE) is 4 wells per year, as table 4.5 shows. Equation (4.8) is reestimated using the 1975 –1998 subsample. We estimate two specifications of equation (4.8). The first version excludes the private financing arrangements made possible by the pidiregas project. The second

04-A3495 7/27/05 7:13 AM Page 48

48

Chapter Four

Table 4.3. Exploration Drilling Model Estimated for 1975 –1998 Dependent Variable: TEWD Variable

Coefficient

Std. error

t-Statistic

Probability

C NMOP TEWD(1) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood Durbin-Watson stat.

1.199184 1.645185 0.625231 0.807433 0.789093 12.37997 3218.538 92.83807 1.587986

6.450651 0.185901 0.836471 1.966816 0.100840 6.200228 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion F-statistic Probability (F-statistic)

0.8543 0.0626 0.0000 49.41667 26.95716 7.986506 8.133763 44.02638 0.000000

Table 4.4. Exploration Wells Drilled: Forecast vs. Actual (subsample 1975 –1998) Number of exploration wells Year

Forecast

Actual

1999 2000 Total

23 30 53

22 37 59

Table 4.5. Exploration Drilling Forecast 1999 –2000 Forecast: TEWDF Actual: TEWD Forecast sample: 1999 –2000 Root mean squared error Mean absolute error Mean absolute percentage error Theil inequality coefficient Bias proportion Variance proportion Covariance proportion

4.977880 4.053307 12.02706 0.086918 0.336975 0.663025 0.000000

version includes a binary variable equal to 0 for the years 1975 –1996 and equal to 1 for the years 1997–1998. Table 4.6 presents estimates of equation (4.8) using the subsample 1975 –1998. The estimated coefficients are obviously stable by comparison with the full-sample estimates shown in table 4.2. Table 4.7 gives the forecasts of development wells in 1999 and 2000. Forecasts that exclude the possibility of private financing are obviously too low. The combined two-year forecast of 335 wells substantially understates the 422 wells actually drilled. However, when the effects of pidiregas financing are taken into account, the forecasts are more accurate: The two-year forecast differs from actual development drilling by only a single well. We tentatively suggest that, for forecasting development drilling beyond 2006, for example, the pidiregas program be ignored because it is not financially viable. How and

04-A3495 7/27/05 7:13 AM Page 49

Exploration and Development Drilling

49

Table 4.6. Development Drilling Model for 1975 –1998 Dependent Variable: TDWD Variable

Coefficient

Std. Error

t-Statistic

Probability

C NMOP TDWD(1) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood Durbin-Watson stat.

4.672928 9.971041 0.504540 0.831294 0.815227 40.53544 34505.56 121.3044 1.571876

20.77545 0.224925 3.560503 2.800458 0.131325 3.841910 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion F-statistic Probability (F-statistic)

0.8242 0.0107 0.0009 173.6667 94.30094 10.35870 10.50596 51.73863 0.000000

Table 4.7. Wells Drilled: Forecast vs. Actual (subsample 1975 –1998) Number of Development Wells Year

Forecast without pidiregas

Forecast with pidiregas

Actual

1999 2000 Total

141 194 335

184 237 421

212 210 422

when these huge debts initiated by pemex will be paid remains a major unanswered question in Mexico’s energy policy.

conclusions The exploration- and development-drilling models seem to be adequately specified. The financial constraints discussed in chapter 3 naturally lead to lags in adjustment to long-run equilibrium. The adjustment dynamics of course depend on the specified lag process. We employ a very simple process that permits a constant percentage in adjustment to occur annually. The estimated coefficients suggest that it takes about four years for exploration and development drilling to attain long-run equilibrium. The key financial variable is the net Mexican-oil export price. This price is a volume-weighted average of the prices of Maya, Isthmus, and Olmeca oil exports, adjusted downward by the pemex tax burden. It correctly incorporates exogenous changes in world oil prices and the rate at which pemex revenues are taxed each year. This tax rate is without a doubt the most powerful policy instrument at the disposal of the federal government to stimulate drilling. The models clearly show that drilling is responsive to net price. For exploration wells, the short-run elasticity (at sample means) is 0.31, and the longrun equilibrium elasticity is 0.80. For development wells, the short-run elasticity is 0.40, and the long-run elasticity 0.91. The estimated coefficients show that a one-dollar increase in the net price would induce pemex to drill

04-A3495 7/27/05 7:13 AM Page 50

50

Chapter Four

about 4 or 5 additional exploration and 19 additional development wells each year in long-run equilibrium. Out-of-sample forecasts for exploration wells are reasonably accurate. Forecasts of development wells that exclude the effects of private loans are too low, but forecasts that account for the pidiregas program are on target. We conjecture that private financing arrangements made possible by pidiregas will be discontinued after the Fox administration because these loans have proven to be financially unsound.

chapter four appendix Estimation of Drilling Equations Using Seemingly Unrelated Regressions It seems reasonable that the disturbances in the exploration- and developmentdrilling equations are correlated. A shock affecting exploratory drilling may simultaneously influence development drilling. pemex drilling-investment allocations vary considerably and should exert contemporaneous influences on exploration and development. Equations (4.9) and (4.10) define the system used in estimation by seemingly unrelated regression (SUR). Their theoretical specification is outlined earlier in this chapter. Exploration: TEWD t  b1  b2NMOPt  b3TEWD t1  et

(4.9)

Development: TDWD t  g1  g2NMOPt  g3TDWD t1  v t (4.10) If the disturbances are correlated and the explanatory variables of the equations are not identical, then SUR estimates are asymptotically more efficient than ordinary least-squares (OLS) estimates. The higher the correlation between disturbances, the larger the gain in efficiency. Ordinary least-square residuals can be used to estimate the correlation between the disturbances of equations 4.9 and 4.10. The estimated correlation between the residuals is 0.497. This correlation suggests that efficiency can be gained by estimating the system using SUR. Table 4.8 shows the SUR estimates of equation (4.9) and (4.10). The reader can compare the SUR estimates in Table 4.8 with the OLS estimates for exploration in Table 4.1 and the OLS estimates for development in Table 4.2. There is little difference between any of the corresponding parameter estimates. The SUR estimates of the coefficients of net Mexican-oil price are slightly lower than OLS estimates, while the SUR estimates of the lagged drilling terms are slightly higher. As expected, estimated SUR standard errors are smaller than the corresponding OLS standard errors.

04-A3495 7/27/05 7:13 AM Page 51

Exploration and Development Drilling

51

Table 4.8. Seemingly Unrelated Regression for Exploration and Development Wells Drilled Coefficient

Estimate

Std. Error

t-Statistic

Development Intercept 3.497186 19.34394 0.180790 Development NMOP 7.738042 3.141007 2.463555 Lagged TDWD 0.582450 0.113754 5.120249 Exploration Intercept 0.988579 5.639891 0.175283 Exploration NMOP 1.532703 0.726678 2.109193 Lagged TEWD 0.652156 0.081138 8.037612 Determinant residual covariance 143862.6 Equation: TDWD  C(1)  C(2)*NMOP  C(3)*TDWD(1) Observations: 26 R-squared 0.810366 Mean dependent var. Adjusted R-squared 0.793876 S.D. dependent var. SE of regression 41.32296 Sum of squared residuals Durbin-Watson stat. 1.470893 Equation: TEWD  C(4)  C(5)*NMOP  C(6)*TEWD(1) Observations: 26 R-squared 0.812630 Mean dependent var. Adjusted R-squared 0.796337 S.D. dependent var. SE of regression 11.96002 Sum of squared residuals Durbin-Watson stat. 1.629107

Probability 0.8573 0.0176 0.0000 0.8616 0.0404 0.0000

176.5385 91.01790 39274.51

47.88462 26.50181 3289.969

Recall that the one-year adjustments toward long-run equilibrium in exploration and development drilling are represented by 1  bˆ 3 and 1  gˆ 3 respectively. A comparison of SUR and OLS estimates shows that the adjustments toward equilibrium are slightly smaller with SUR than with OLS (0.348 vs. 0.388 for exploration and 0.418 vs. 0.447 for development). The response of long-run equilibrium drilling to a change in NMOP is determined by bˆ 2/11  bˆ 3 2 for exploration and gˆ 2/11  gˆ 3 2 for development. There is practically no difference between the SUR and OLS estimates of these long-run responses: 4.41 exploration wells using SUR estimates and 4.47 wells using OLS estimates; 18.53 development wells using SUR estimates and 18.68 wells based on OLS estimates.

04-A3495 7/27/05 7:13 AM Page 52

05-A3495 7/27/05 7:13 AM Page 53

chapter five

Successful Exploration and Development

Successful drilling is the essential source of new oil and gas reserves. It is the foundation of future production. Without successful drilling, reserves and future production must inevitably decline. This chapter analyzes successful exploration and development from 1975 to 2000. pemex reports the number of exploration and development wells drilled annually, along with the successful wells of each type. Within the categories of successful exploration and development, however, pemex does not distinguish between oil and gas wells. We necessarily employ these same pemex classifications. Exploration is motivated by the hope of discovering new reservoirs. Because exploration takes place in unproven territories, it entails high risks of failure and great geological risks. The overall success ratio in exploration (OSRE) is the number of successful exploration wells divided by the total number of exploration wells drilled. Between 1975 and 2000, the OSRE in Mexico increased around an erratic upward trend from 0.149 to 0.568. This upward drift can be explained by two phenomena: improved drilling technology and knowledge gained through experience. Development wells are drilled after hydrocarbons have been discovered by successful exploration. Development drilling thus benefits from valuable geological knowledge concerning depth zones, pressure, and other reservoir characteristics. Because development wells are drilled with the benefit of this additional information, geological risks are lower, and success ratios are considerably higher than those in exploration. In Mexico, the overall success ratio for development wells (OSRD) increased from 0.797 in 1975 to 0.910 in 2000. As one might expect, this trend in successful development is less erratic than that in exploration. A success ratio can be viewed as an index of drilling efficiency. Systematic growth in the success ratio shows progress in the battle with inevitable geological risk. By this measure, pemex has registered substantial gains in efficiency, especially since 1990. These gains are attributable both to technological improvements in drilling methods and to knowledge gained from experience.

exploration success ratios, 1975 – 2000 Successful exploration depends on several factors, including the location of wells (which, in turn, is determined by the pemex projects approved by the Mexican Congress), random luck, and drilling technology. New technology, such as three-dimensional seismic imaging, is phased in over time, not in one year. Once improved exploration technologies are adopted, they become 53

05-A3495 7/27/05 7:13 AM Page 54

54

Chapter Five

Figure 5.1. Exploration success ratios, 1975 –2000.

permanent. Because they are phased in over a span of years, their benefits might first be modeled as a simple autoregressive process. Exploration success ratios are shown in figure 5.1. They display a good deal of year-to-year variation and a large spike to 0.964 in 1978. This abnormally high success ratio is attributable to a one-time event: discovery of the southeast basin in Chiapas and Tabasco, where 80 of 83 exploration wells were successful. Setting aside this extraordinary stroke of good fortune, successful exploration exhibits a noteworthy upward trend, which is attributable to new technologies and to gains from experience.

development success ratios, 1975 – 2000 Development drilling benefits from the geological knowledge gained through successful exploration and from new technologies such as horizontal drilling. For these reasons, development success ratios are much higher and display less random variation than exploration success ratios. As figure 5.2 shows, development success ratios range from approximately 0.78 in 1976 to 0.98 in 1998. They, too, exhibit a marked upward trend, chiefly as a result of improved drilling technologies. If more effective technologies were fully adopted within one year, one would expect to see a stairstep pattern of successively higher plateaus. However, the sawtooth pattern in figure 5.2 suggests the development ratios might be modeled as an autoregressive process.

05-A3495 7/27/05 7:13 AM Page 55

Successful Exploration and Development

55

Figure 5.2. Development success ratios, 1975 –2000.

modeling success ratios We initially model exploration and development success ratios as first-order autoregressive processes.

Autoregressive Models Exploration: OSREt  b0  d1DUM 78  b1OSREt1  et

(5.1)

where DUM78 is a dummy variable  1 to adjust for the abnormal success in Chiapas and Tabasco in 1978, and b1 is the autoregressive coefficient. Development: OSRD t  a0  a1OSRD t1  v t

(5.2)

where a1 is the autoregressive coefficient. An AR(1) process has the general form Yt  d  rYt1  et

(5.3)

where et  iid1o, s2e 2 is a white-noise disturbance. In order to determine the mean, we introduce a lag operator, L. Lag operators have the following properties: 1.

LYt  Yt1

2.

L1Yt  Yt1

05-A3495 7/27/05 7:13 AM Page 56

56

Chapter Five

3.

L1L  LL1  I, where I is an identity matrix

4.

LK  K if K is constant

The lag operator allows the AR(1) process to be rewritten as Yt  d  rLYt  et

(5.4)

Solving for Yt we have Yt  d (1  rL)  et /(1  rL). Since d is constant, d (1  rL)  d/(1  r). In AR(1) processes, 0  |r|  1. In the case of |r|  1, 1/(1  rL) is the reduced form of a geometric series, giving Yt 

q 1 d  a r j etj 1r j 0

(5.5)

The expected value of equation (5.5) is E1Yt 2 

d 1r

(5.6)

In the context of equations (5.1) and (5.2) we have E1OSREt 2 

b0 1  b1

(5.7)

E1OSRD t 2 

a0 1  a1

(5.8)

and

as expected values, conditional on the sample observations.

Gains from Experience Greater drilling efficiency can also result from experience. A theory of gains from experience, or “learning by doing,” presumably originated with Arrow (1962). Improvements in efficiency attributable to experience have been documented in numerous studies of labor markets and several studies of productivity in specific industries (e.g., Sheshinski 1967, Fudenberg and Tirole 1983, Lieberman 1984, and Irwin and Klenow 1994). As contractors gain experience, they typically encounter more success. Technological improvements also increase success ratios. Experience is sometimes measured by the cumulative production of a firm or industry (Lieberman 1984, Irwin and Klenow 1994). In our context, a relevant measure of experience is the cumulative number of wells drilled. Therefore, we have constructed an index of experience in exploration by the cumulative number of exploration wells drilled and in development by the cumulative number of development wells drilled since 1975. Cumulative drilling is a commonsense indicator of

05-A3495 7/27/05 7:13 AM Page 57

Successful Exploration and Development

57

experience, and the series demonstrates the right sort of curvature to allow gains from experience to eventually exhibit diminishing returns. We measure experience in exploration as cumulative exploration wells drilled through the preceding year, CUMEW(1). Similarly, we measure experience in development as cumulative development wells drilled through the previous year, CUMDW(1). A regression equation that incorporates both new technology as an autoregressive process and experience in exploration drilling is OSREt  b0  b1OSRE112  d1DUM78  b2CUMEW112  et

(5.9)

The corresponding regression for development drilling is OSRDt  a0  a1OSRD112  a2CUMDW112  vt

(5.10)

statistical results We first estimate autoregressive equations (5.1) and (5.2), then estimate equations (5.9) and (5.10) that include gains from the experience.

First-order Autoregressive Equations Estimates of equation (5.1) are shown in table 5.1. The estimated coefficient of the binary variable DUM78 is highly significant, suggesting that the discoveries in Chiapas and Tabasco increased the 1978 success ratio by approximately 0.57. The estimated autoregressive coefficient is significant at P  0.01. An estimate of the probability limit of the exploration success ratio is bˆ0 /11  bˆ2 2  0.40 (see equation 5.7), which is close to the sample mean, excluding the abnormally high success ratio in 1978. Estimates of equation (5.2) are shown in table 5.2. The autoregressive coefficient suggests that the development success ratio can be approximated by a first-order process. An estimate of the probability limit of the development success ratio is aˆ0 /11  aˆ1 2  0.88 (see equation 5.8), which approximates the sample mean of 0.86. Table 5.1. Estimates of the Exploration Success Ratios, 1975 –2000 Dependent Variable: OSRE Variable

Coefficient

Std. Error

t-Statistic

Probability

C DUM78 OSRE(1) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood Durbin-Watson stat.

0.229549 0.574307 0.421334 0.562729 0.524706 0.122712 0.346337 19.24730 1.418137

0.061406 3.738221 0.125223 4.586281 0.136374 3.089547 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion F-statistic Probability (F-statistic)

0.0011 0.0001 0.0052 0.425007 0.177994 1.249792 1.104627 14.79950 0.000074

05-A3495 7/27/05 7:13 AM Page 58

58

Chapter Five

Table 5.2. Estimates of Development Success Ratios, 1975 –2000 Dependent Variable: OSRD Variable

Coefficient

Std. Error

t-Statistic

Probability

C OSRD(1) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood Durbin-Watson stat.

0.165844 0.811151 0.665473 0.651535 0.036834 0.032562 49.98268 2.266735

0.100612 1.648350 0.117394 6.909642 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion F-statistic Probability (F-statistic)

0.1123 0.0000 0.859246 0.062398 3.690976 3.594199 47.74315 0.000000

Table 5.3. Exploration Success Ratio with Learning by Experience, 1975 –2000 Dependent Variable: OSRE Variable

Coefficient

Std. Error

t-Statistic

Probability

C DUM78 CUMEW(1) OSRE(1) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood Durbin-Watson stat.

0.142341 0.669307 0.000186 0.281082 0.676554 0.630347 0.104798 0.230637 23.09888 1.407904

0.068414 2.080586 0.113514 5.896273 6.73E05 2.759310 0.124310 2.261134 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion F-statistic Probability (F-statistic)

0.0499 0.0000 0.0118 0.0345 0.436031 0.172368 1.527911 1.332890 14.64193 0.000023

The Breusch-Godfrey test statistic of 3.43 shows that we can reject the hypothesis of positively correlated disturbances. The Jarque-Bera statistic of 0.222 suggests that normality in disturbances is a reasonable approximation. As an experiment, we have also modeled development success ratios as a second-order process; including a second-order autoregressive term adds little information.

Gains from Experience Table 5.3 shows the results obtained for exploration equation (5.9). As one would expect, the estimated autoregressive coefficient is smaller than the estimate in table 5.1 but remains significant at P  0.05; the estimated coefficient of experience is significant at P  0.01. Since equation (5.9) exhibits a significantly lower sum of squared residuals than equation (5.1), experience complements new technology in improving exploration efficiency. Table 5.4 shows the results for development equation (5.10). The estimated autoregressive coefficient is again highly significant, and the coefficient of experience is significant at P  0.05 using a one-tail test. The gains from experience are not as pronounced in development as they are in exploration, nor should they be, because contractors already have more experience in development than

05-A3495 7/27/05 7:13 AM Page 59

Successful Exploration and Development

59

Table 5.4. Development Success Ratio with Learning by Experience, 1975 –2000 Dependent Variable: OSRD Variable

Coefficient

Std. Error

t-Statistic

Probability

C CUMDW(1) OSRD(1) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood Durbin-Watson stat.

0.333039 1.63E05 0.565055 0.697161 0.669630 0.035839 0.028257 49.34255 2.019588

0.133883 2.487531 9.10E06 1.787656 0.176515 3.201165 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion F-statistic Probability (F-statistic)

0.0209 0.0876 0.0041 0.861736 0.062352 3.707404 3.561139 25.32289 0.000002

in exploration. Nonetheless, equation (5.10) indicates that experience supplements technology to enhance the efficiency of development drilling.

summary and conclusions Exploration is always riskier than development. Exploration risk has been substantially diminished by the innovation of three-dimensional seismic imaging, which much enhances contractors’ knowledge concerning probable locations and volumes of hydrocarbons. pemex’s investment in three-dimensional seismic surveys is surely responsible for much of the growth in exploration success after the mid-1980s. Exploration and development success ratios increased between 1975 and 2000 partly because of improvements in drilling technologies and partly because of the knowledge gained from experience. We first model success ratios as first-order autoregressive processes, using the notion that new drilling technologies are adopted gradually over time. We then augment the autoregressive models to allow for gains from experience. The estimated coefficients show that experience complements the new technologies to improve exploration and development efficiency. Exploration success ratios increased from about 0.23 in 1975 and 1976 to about 0.50 in 1999 and 2000. Development success ratios increased from about 0.78 in 1975 and 1976 to roughly 0.91 in 1999 and 2000. These increases amount to impressive gains in drilling efficiency.

chapter five appendix One would suspect the disturbance terms in the exploration and development success ratio equations, (5.1) and (5.2), and also in equations (5.9) and (5.10), to be correlated. Improvements in technology, such as three-dimensional seismic imaging, may affect exploration and development contemporaneously.

05-A3495 7/27/05 7:13 AM Page 60

60

Chapter Five

Efficiency resulting from experience in exploration drilling may be partially transmitted to development. If the disturbances are correlated across equations, then estimating both equations as a seemingly unrelated regression system will increase the asymptotic efficiency of the estimates by comparison with OLS estimates. Table 5.5 shows equations (5.1) and (5.2) estimated by SUR. Although the t-statistics are unbiased estimates only in large samples, the reported values are Table 5.5. Autoregressive Models Estimated by SUR Coefficient

Std. Error

t-Statistic

OSRE constant 0.296598 0.048789 6.079242 Lagged OSRE 0.251312 0.103633 2.425021 DUM78 0.649984 0.091874 7.074710 OSRD constant 0.381332 0.083356 4.574751 Lagged OSRD 0.559071 0.097107 5.757265 Determinant residual covariance 1.17E05 Equation: OSRE  C(1)  C(2)*OSRE(1)  C(3)*DUM78 Observations: 26 R-squared 0.525157 Mean dependent var. Adjusted R-squared 0.483866 S.D. dependent var. SE of regression 0.127875 Sum of squared residuals Durbin-Watson stat. 0.906835 Equation: OSRD  C(4)  C(5)*OSRD(1) Observations: 26 R-squared 0.601204 Mean dependent var. Adjusted R-squared 0.584587 S.D. dependent var. SE of regression 0.040217 Sum of squared residuals Durbin-Watson stat. 1.470533

Probability 0.0000 0.0192 0.0000 0.0000 0.0000

0.425007 0.177994 0.376096

0.859246 0.062398 0.038818

Table 5.6. Learning by Doing Models Estimated by SUR Coefficient

Std. Error

t-Statistic

OSRE constant 0.188883 0.058816 3.211430 Lagged OSRE 0.141168 0.092446 1.527040 Lagged CUMEW 0.000200 6.12E05 3.266091 DUM78 0.685363 0.081307 8.429337 OSRD constant 0.493500 0.102802 4.800477 Lagged OSRD 0.350522 0.134705 2.602135 Lagged CUMDW 2.49E05 7.78E06 3.196057 Determinant residual covariance 6.79E06 Equation: OSRE  C(1)  C(2)*OSRE(1)  C(3)*CUMEW(1)  C(4)*DUM78 Observations: 25 R-squared 0.656637 Mean dependent var. Adjusted R-squared 0.607585 S.D. dependent var. SE of regression 0.107977 Sum of squared residuals Durbin-Watson stat. 1.081790 Equation: OSRD  C(5)  C(6)*OSRD(1)  C(7)*CUMDW(1) Observations: 25 R-squared 0.676794 Mean dependent var. Adjusted R-squared 0.647411 S.D. dependent var. SE of regression 0.037024 Sum of squared residuals Durbin-Watson stat. 1.542325

Probability 0.0025 0.1341 0.0021 0.0000 0.0000 0.0127 0.0026

0.436031 0.172368 0.244839

0.861736 0.062352 0.030158

05-A3495 7/27/05 7:13 AM Page 61

Successful Exploration and Development

61

large enough to indicate that all of the estimated coefficients are statistically significant. In the exploration success ratio equation, the estimated coefficient of autoregression (0.251) is somewhat smaller than the OLS estimate (0.421), and the estimated coefficient of the dummy variable (0.650) is slightly larger than the OLS estimate (0.574). In the development success ratio equation, the estimated coefficient of autoregression (0.559) is also smaller than the OLS estimate (0.811). These SUR estimates nonetheless suggest that the autoregressive equations (5.1) and (5.2) are reasonable first approximations. SUR estimates of the learning-by-doing equations (5.9) and (5.10) are shown in table 5.6. The lagged exploration success ratio becomes statistically insignificant, but lagged cumulative exploration wells are highly significant. In the development success ratio equation, both lagged success ratios and cumulative development wells are significant. The SUR estimate of the coefficient of lagged wells (2.49E0.05) is similar to the OLS estimate (1.63E0.05), but the estimated SUR coefficient of lagged success ratios (0.351) is somewhat below the OLS estimate (0.565). The OLS and SUR estimates of the learning-byexperience regressions yield similar interpretations.

05-A3495 7/27/05 7:13 AM Page 62

06-A3495 7/27/05 7:13 AM Page 63

chapter six

Additions to Oil and Gas Reserves

Oil and gas reserves are economically producible inventories. They are not simply oil and gas “in the ground.” Instead, they are deposits that can be drilled with a reasonable prospect of profit under current technological and economic conditions. Reserves are the basis of current and future production. They can be augmented only by successful exploration and development and are depleted by production. If current-year production exceeds current-year gross reserve additions, then end-of-year reserves are depleted. As reserves are used up, future production inevitably declines. This chapter focuses on oil and gas reserves since 1975. Reported reserves should properly be viewed as uncertain indicators rather than precise estimates. Indeed, estimated reserves differ widely, depending on the criteria used in evaluating them. We analyze reserve estimates that meet the criteria established by the World Petroleum Congress (wpc) and the Society of Petroleum Engineers (spe) and discuss these and other matters involving data reported by pemex. We also talk about trends in oil and gas production, reserves, and reserves/ production ratios, which are indices of long-run production capacity. We show that reserves/production ratios for both oil and gas have exhibited a notable decline since the late 1980s. Toward the end of the chapter we describe gross reserve additions and the sources of errors in measuring them, and we link these new reserves to their only systematic source: successful exploration and development drilling.

data All of the data concerning production, reserves, reserve additions, and successful wells are taken from pemex documents. Oil production is measured in onemillion-barrel units, which include crude oil and liquid condensates from associated gas. Production statistics are obtained from several pemex yearbooks.1 Gas production is reported in million cubic meters up to 1987 and billion cubic feet thereafter. To obtain a consistent series, we convert production prior to 1988 to billion cubic feet using conversion factors published by the Secretary of Energy.2 We believe that production of oil and gas is measured with a high degree of accuracy. Oil reserves are measured in million barrels, and gas reserves in billion cubic feet at the end of each year. Estimated reserves in the form of natural-gas liquids have been converted to billion cubic feet (BCF) equivalent to natural gas and added to the estimated reserves of dry gas. Estimated oil and gas reserves are taken directly from pemex documents. 63

06-A3495 7/27/05 7:13 AM Page 64

64

Chapter Six

Two Standards of Uncertainty: Proved and Probable Reserves We stress that reserves cannot be measured with complete accuracy but are subject to varying degrees of uncertainty. Working independently, the wpc and the spe in 1987 produced similar criteria for classifying petroleum reserves. These criteria are the preferred industry-wide standards for reserve classification. Proved reserves are quantities of oil and gas that, by analysis of geological and engineering data, can be estimated with “reasonable certainty” to be commercially producible from known reservoirs under current economic conditions, operating methods, and government regulations. Reserves are classified as proved if there is at least a 90 percent probability that the quantities actually recovered will equal or exceed the estimate. For reserves to be classified as “proved” requires that facilities to process and transport them be either operational at the time the estimate is made or highly likely to be installed. Estimates of proved reserves are required to meet the most exacting standards. Probable reserves are unproved reserves that analysis of geological and engineering data suggests are likely to be eventually recoverable. Several technical criteria permit reserves to be classified as “probable.” The overriding criterion is that there be at least a 50 percent probability that the quantities actually recovered will equal or exceed the sum of estimated proved plus probable reserves. Because probable reserves must satisfy much weaker criteria, they are subject to much greater uncertainty than proved reserves. Reported reserves may under- or overstate economically producible reserves. This problem is frustrating in econometric work for two reasons. First, the size of the measurement errors is unknown and cannot be determined. Second, the errors are almost surely not random: Errors of measurement in one year may persist for several. Estimates of reserves are necessarily based on incomplete information and rely to some degree on subjective judgment. The problem can be clarified by showing the various components of reserves. Reserves at the end of year t may be expressed as Reservest  Reservest1  Gross reserve additionst  Productiont

(6.1)

Thus year-end reserves in year t equal year-end reserves in year (t  1) plus gross reserve additions minus production during year t. Gross reserve additions consist of three parts: new pool discoveries attributable to successful exploration; extensions, which increase estimated reserves chiefly because of successful development wells; and revisions, which are made in view of changes in reservoir performance during the year. For example, if production declines unexpectedly fast, reserves are typically revised downward in the belief that earlier estimates were too high. On the other hand, if unexpected water intrusion causes the percentage of water in produced liquids to rise, reserves may also be revised downward. Year-to-year changes in reported reserves may be erratic, usually because of large revisions.

06-A3495 7/27/05 7:13 AM Page 65

Additions to Oil and Gas Reserves

65

New pool discoveries and extensions are always positive numbers, but they are uncertain estimates— or best guesses—based on incomplete information learned that year. The original reserve estimates attributable to one year’s successful exploration and development are almost always later revised. When revisions are positive, they augment new pool discoveries plus extensions, sometimes leading to very large increases in estimated reserves. By contrast, during years in which negative revisions exceed new pool discoveries and extensions, gross reserve “additions” are negative numbers, sometimes quite large. The consequence is that reserves at the end of year t are reduced both by that year’s production and by negative changes in gross reserve estimates. Large revisions produce an erratic series in end-of-year reserves. For example, huge positive revisions made by pemex in 1978 are responsible for reported year-end gas reserves that year of 66.2 trillion cubic feet (TCF), more than double the reported 1977 year-end reserves of 31.3 TCF. As another example, year-end reserves in 1986 are reported to have been 123.3 TCF, more than 55 percent higher than the 79.1 TCF reported in 1985. The large upward revision of 1986 was carried forward through 1987 (reported reserves of 122.5 TCF). Then two years of large negative revisions account for abrupt declines in reported reserves to 100.4 TCF in 1988 and 81.2 TCF in 1989. These huge revisions—positive in 1986, then negative in 1988 and 1989 —leave reported reserves in 1989 very nearly the same as in 1985. Revisions are based on information gained during the year. Because they are inherently subjective, year-to-year revisions are partially responsible for measurement errors wrapped into the estimated oil and gas reserves. Their large subjective component makes it impossible to model revisions econometrically even in a geologic area as mature as Texas (Moroney 1997). During the years when pemex revisions are known to have been exceptionally large (e.g., 1978, 1986, 1988, and 1989 for gas reserves), we make adjustments by the use of binary variables in the regressions reported in this chapter’s section on specification and estimation. It is vital to be mindful that the reserves reported by pemex necessarily contain unknown measurement errors.

Two Sets of Criteria for Reported Reserves Before 1999, pemex reported reserve estimates that satisfied the weaker wpc / spe criteria for probable reserves. Beginning in 1999, pemex has reported two sets of reserve estimates. The first set meets the strict criteria of proved reserves, as required by the U.S. Securities and Exchange Commission (sec). The sec insists that proved reserves be reported by all oil and gas firms because they are used in prudent valuation of a firm’s assets. The sec understandably requires firms to employ a conservative method of estimating reserves. pemex’s second set of reserve estimates is a continuation of its earlier series based on probable reserves. Its estimates of proved reserves are much lower than its estimates of probable ones.3 As noted earlier in the chapter, probable

06-A3495 7/27/05 7:13 AM Page 66

66

Chapter Six

Table 6.1. Proved and Probable Oil and Gas Reserves Oil Reserves (end of year million barrels) (1) Proved 1999 2000 2001

21519 20186 18767

Gas Reserves (end Ratio of of year BCF) Ratio of Column 1 to Column 3 to (2) Probable Column 2 (3) Proved (4) Probable Column 4 41495 39918 38286

0.52 0.51 0.49

18471 17365 16256

59925 59865 54676

0.31 0.29 0.30

reserves must satisfy weaker criteria than proved reserves and are subject to greater uncertainty.4 The two sets of estimates are shown in table 6.1. The more exacting standards required of proved reserves are evident: Proved oil reserves are only half as large as probable reserves; proved gas reserves are less than onethird as large as probable reserves. We employ probable reserves in this research for two reasons. First, pemex began to report proved reserves in 1999, so the series is too short for long-term analysis. Second, since we wish to analyze long-run patterns of reserves and production, probable reserves are our only choice. The comparisons in table 6.1 underscore our earlier admonition: Estimated reserves vary widely and necessarily include subjective judgments. The analysis that follows should be viewed accordingly.

oil production and reserves We first show twenty-five-year trends in production, which is measured with considerable accuracy. We then describe the trends in reported reserves, which inevitably contain measurement errors. Because of sometimes large year-toyear changes in reported reserves, we discover correspondingly large shortterm changes in reserves/production ratios.

Production In 1975, oil production amounted to a scant 262 million barrels. As figure 6.1 shows, production doubled to 537 million barrels in 1979, then nearly doubled again to 1,024 million in 1984. This spectacular growth in production followed an even more impressive increase in reserves. During the next decade, production varied little, ranging from 987 million barrels in 1985 to 955 million in 1995. Production then registered moderate growth to a peak of 1,121 million barrels in 1998. What variables can explain this remarkable growth from 1975 until 1984 and then the comparative stability until 2000? In fact, only a few important variables whose roles can be easily understood are within the context of our integrated model: reserves, successful drilling, and after-tax net income that can be invested.

06-A3495 7/27/05 7:13 AM Page 67

Additions to Oil and Gas Reserves

67

Figure 6.1. Oil production, 1975 –2000.

In the long run, production must necessarily follow reserves. However, the five- or even ten-year connection between reported reserves and production is quite elastic. The relationships among production, reserves, hydrocarbon prices, and taxes are complex and indirect. These relationships are also characterized by important lags: (1) intervals between the day the pemex budgets are approved and the time when pemex can sign drilling contracts and actually start drilling; (2) lags between the time contracts are signed and the moment drilling equipment and personnel are on site and actually begin to drill; (3) intervals between the time drilling begins and the moment a well is found to be successful or instead abandoned; and (4) lags in adjustment to long-run equilibrium production.

Oil Reserves and Investment Long-term growth and decline in Mexico’s oil reserves are traceable to a combination of geological and economic forces. Hydrocarbons in the ground are of course the essential basis for reserves. However, to convert undiscovered resources into producible reserves requires successful exploration and development, which in turn requires long-term, large-scale capital investment. pemex launched a massive drilling program that lasted from the early 1970s through 1981. Vast oil resources were discovered, and a sustained program of successful investment converted them into probable reserves. At the end of 1975, estimated reserves were 3,953 million barrels. As figure 6.2 illustrates, by 1977, estimated reserves more than doubled to 10,428 million barrels. Then, at the end of 1978, the huge discoveries in Chiapas and Ta-

06-A3495 7/27/05 7:13 AM Page 68

68

Chapter Six

Figure 6.2. End-of-year oil reserves, 1975 –2000.

basco boosted the reserves to a whopping 28,407 million barrels. Continuing successful investment enabled the reserves to double yet again, to 57,000 million barrels in 1981. Estimated reserves increased fourteenfold in six years. Reserves were stable until 1985, but the period from 1981 to 1985 was a short-lived plateau. The drilling recession began in 1982, as the number of successful wells declined to 255 from 321 in 1981. Reserves were revised downward sharply in 1986, then declined for the next fourteen years. By the end of 2000, estimated reserves stood at approximately 40,000 million barrels, only 70 percent of their level in 1981. The protracted decline in oil reserves is surely attributable to a drilling depression that spanned twenty years. This depression was caused by an evertightening shortage of investment capital ascribable chiefly to high taxes. We firmly believe that large oil resources remain to be discovered in Mexico, but to find and develop them will require a large-scale revival in successful drilling. Fortunately, drilling increased substantially from 2001 to 2003, and it is conceivable that the long-term decline has been reversed.

Reserves/Production Ratios A country’s long-run capacity to produce is limited by its reserves. Imagine for the moment that reserves could be known with certainty and that for many years there were no gross reserve additions. Equation (6.1) shows that reserves would be depleted each year by an amount equal to that year’s production. If

06-A3495 7/27/05 7:13 AM Page 69

Additions to Oil and Gas Reserves

69

70

60

Reserves/Production

50

40

30

20

10 1975

1980

1985

1990

1995

2000

Year Figure 6.3. Oil reserves/production ratio, 1975 –2000.

gross reserve additions remained at zero indefinitely, then production would eventually cease. The reserves/production (R /P) ratio is a sort of gauge that indicates how fast reserves are growing or being exhausted in relation to production. An increasing R /P ratio indicates that a country’s long-run productive capacity is growing relative to current production, and a decreasing ratio is evidence that the long-run capacity is shrinking. A declining R /P ratio is by no means cause for deep concern. However, a ratio that declines very much for many years is a clear warning that the current rate of production is not sustainable indefinitely. This long-run behavior of the R /P ratio is therefore a useful gauge of relative productive capacity. The oil R /P ratio varies enormously in Mexico.5 As figure 6.3 illustrates, the ratio was 15 in 1975. During the next six years, estimated reserves increased fourteenfold, but production increased by 3.2 times; the R /P ratio skyrocketed to 68. Then, from 1985 to 2000, reserves declined by 30 percent while production gradually increased. For these two reasons, the R /P ratio fell from 58 in 1985 to 36 in 2000.6 This decline is not cause for concern, but its continuation would be. For pemex to increase its future capacity to produce oil, the R /P ratio must eventually be stabilized.

06-A3495 7/27/05 7:13 AM Page 70

70

Chapter Six

gas production and reserves Gas production increased rapidly between 1977 and 1982 but, unlike oil production, declined between 1982 and 1988. Reported gas reserves were subject to major revisions between 1986 and 1989, which caused large short-term changes in reserves/production ratios.

Production Gas is produced from two types of wells. First, gas that is produced from oil wells is known as associated gas (sometimes called casinghead gas). Second, gas produced from gas wells is termed nonassociated gas. pemex publishes annual production reports of associated and nonassociated gas. pemex has thus far largely neglected the development of pure gas wells (see chapters 1 and 3). An important consequence of this neglect is that nonassociated gas accounted for only 17 percent of total gas reserves in the years from 1998 to 2000.7 Figure 6.4 shows gas production, which doubled from 786 BCF in 1975 to 1,550 BCF in 1982. After 1982, however, production declined nearly continuously through 1989, when production stood at 1,172 BCF. Production then rose fairly steadily until reaching a plateau at roughly 1,750 BCF between 1998 and 2000, only slightly higher than in 1982. Since 1975, gas production has obviously lagged the fourfold growth in oil.

Figure 6.4. Natural gas production, 1975 –2000.

06-A3495 7/27/05 7:13 AM Page 71

Additions to Oil and Gas Reserves

71

Figure 6.5. End-of-year natural gas reserves, 1975 –2000.

Reserves Estimated reserves registered a sixfold increase during the drilling boom of 1975 –1981, as figure 6.5 illustrates. They remained nearly stable through 1985. Then huge upward revisions lifted reported reserves abruptly from 79,110 BCF in 1985 to 123,291 BCF in 1986. These revisions proved to be overly optimistic, in part because production continued to decline after 1986 (see figure 6.4). Because higher production failed to materialize, large negative revisions were made in 1988 and 1989. These two years of negative revisions placed reported reserves in 1989 at the same level as in 1985. Since 1989, reported reserves declined steadily to some 60,000 BCF in 2000. The salient point is that estimated reserves in 2000 were substantially below those in 1981. Ignoring the erroneous peak in 1986 and recognizing that reported reserves contain measurement errors, we believe the decline between 1981 and 2000 is probably genuine, though inexact.

Reserves/Production Ratios Figure 6.6 shows that the gas R /P ratio—much like that of oil—rose sharply between 1975 and the early 1980s. Between 1980 and 1985, the ratio remained stable within the 55 – 63 range.8 The sharp rise from 1986 to 1987 should be ignored because it was caused by large upward revisions that proved to be incorrect. Between 1990 and 1995, the ratio varied in the 54 – 60 range, then decreased to about 35 between 1998 and 2000. In these three years, the R /P ratios for gas and oil are almost identical.

06-A3495 7/27/05 7:13 AM Page 72

72

Chapter Six

Figure 6.6. Gas reserves/production ratio, 1975 –2000.

We recognize that these reported ratios are suffused with measurement errors throughout the period from 1975 to 2000. Nonetheless, we believe the declining R /P ratios after 1990 depict genuine downward trends.

gross reserve additions Oil and gas reserves have decreased since the late 1980s while oil and gas production has increased. Reserves have declined because, in most years, production has exceeded gross reserve additions. Gross reserve additions consist of new pool discoveries, extensions, and revisions. pemex does not report these individual sources of reserve additions in its long-term wpc /spe reserve series, but the company began to publish them in 1999.9 Each year’s gross reserve additions can be calculated by rearranging equation (6.1) as Gross reserve additionst  Reservest  Reservest1  Productiont (6.2) Since production is estimated reliably, errors in reported reserves are transmitted as errors in gross reserve additions. Apart from the systematic components of reserves, these errors of measurement produce somewhat jagged series for estimated gross oil reserve additions (GORA t ) and estimated gross gas reserve additions (GGRA t ). We account for several of the largest changes in GORA and GGRA known to be attributable to the following specific events:10 (1) Estimated oil reserves nearly tripled between 1977 and 1978 because of the enormous discoveries in Chiapas and Tabasco; (2) estimated reserves increased by some 14 billion

06-A3495 7/27/05 7:13 AM Page 73

Additions to Oil and Gas Reserves

73

barrels at the end of 1980 because of twenty-eight large new reservoirs discovered; (3) estimated reserves decreased by nearly 9 billion barrels at the end of 1986 because of downward revisions made in response to the collapse of oil prices that year; (4) reported gas reserves and gross gas reserve additions more than doubled between 1977 and 1978 because of the Chiapas-Tabasco discoveries; and (5) reported reserves were revised upward by 44 TCF in 1986, then were completely offset by negative revisions that caused large negative spikes in GGRA in 1988 and 1989. We adjust the GORA and GGRA series by employing dummy variables in the regressions that follow. In the regression that explains GORA, the dummy variable is 1 in the two years of abnormally large reserve increases (1978 and 1980) and 0 in other years. Similarly, in 1986, when oil reserves were revised sharply downward, the dummy variable is 1, and in other years, 0. In the regression that explains GGRA, the dummy variable is 1 in 1978 (to account for the Chiapas-Tabasco discoveries) and 1986 (to control for the abnormal upward revisions) and 0 in other years. A separate dummy variable is 1 in 1988 and 1989 to control for the unusual negative revisions and 0 in other years. These dummy variables shift the estimated regression upward or downward for years in which unusual events are known to have occurred. The logic underlying the regressions is straightforward: Gross reserve additions result from successful drilling. Apart from the atypical events just discussed, the burden of explaining gross reserve additions is placed on the number of successful wells drilled. Because reserve additions are booked during the year in which wells are completed, the relationship between reported reserve additions and successful wells is contemporaneous. The series for gross oil and gas reserve additions and successful wells are tested for stationarity. The critical value to reject the hypothesis of a unit root at P  0.05 is 2.98. Using augmented Dickey-Fuller tests, the calculated test statistics are 3.03 for gross oil reserve additions, 4.29 for gross gas reserve additions, and 3.11 for total successful wells drilled. The evidence suggests that all of the series are stationary.

specification and estimation We specify gross reserve additions as linear functions of the number of successful wells drilled each year, together with dummy variables that control for specific events noted earlier. A linear model of course rules out any possibility of diminishing returns to successful drilling. We believe linearity to be a reasonable assumption for the period 1975 –2000 in the development of Mexico’s oil and gas resources.

Gross Oil Reserve Additions The equation for oil reserve additions is GORAt  b 0  b1TSWD t  b 2ODUM2H  b3ODUM1  u1t

(6.3)

06-A3495 7/27/05 7:13 AM Page 74

74

Chapter Six

where GORAt  gross oil reserve additions in year t TSWDt  total successful wells drilled in year t ODUM2H  dummy  1 for 1978 and 1980 ODUM1  dummy  1 for 1986 u1t  the disturbance term Table 6.2 shows the results. The coefficient of TSWD is astounding because it suggests that a typical successful well could add 18.85 million barrels of new reserves. This exceedingly large estimate is based on data encompassing several of the largest reserves ever discovered in Mexico. The largest ones are typically found offshore, where the number of wells drilled is severely constrained by high drilling costs per well. Reserves per successful well are almost always larger offshore.11 Taken at face value, new reserves of 18.85 million barrels per successful well seem implausibly large. Nonetheless, such a large number is entirely consistent with the following facts. Average daily production in Mexico is about 1,170 barrels per well, which translates to 427,000 barrels per year. If we view 18.85 million barrels as a reasonable estimate of average gross reserves added per successful well, then the average R /P ratio of all wells producing in Mexico is (18.85 million barrels of reserves/0.427 million barrels annual production)  44.1. This R /P ratio for the average producing well is consistent with the range of R /P ratios shown in figure 6.3. The sample mean is 46.9, and the median is 46.7. The estimated coefficient of ODUM2H shows that estimated reserves increased by nearly 12.5 billion barrels in 1978 because of the discoveries in Chiapas-Tabasco and in 1980 because of the twenty-eight newly discovered reservoirs. The estimated coefficient of ODUM1 shows the downward revision attributable to the oil-price collapse in 1986. The Durbin-Watson statistic indicates that the hypothesis of serially independent disturbances is acceptable. However, the regression residuals are highly Table 6.2. Gross Oil Reserve Additions, 1975 –2000 Dependent Variable: GORA Variable

Coefficient

Std. Error

t-Statistic

C TSWD ODUM2H ODUM1 R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood Durbin-Watson stat.

1506.585 18.85483 12445.21 9441.968 0.859951 0.840853 2043.284 91,850,250 232.9009 2.190175

978.9519 1.538978 5.567608 3.386523 1674.091 7.434014 2087.472 4.523159 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion F-statistic Probability (F-statistic)

Probability 0.1381 0.0027 0.0000 0.0002 2266.801 5121.888 18.22314 18.41670 45.02928 0.000000

06-A3495 7/27/05 7:13 AM Page 75

Additions to Oil and Gas Reserves

75

skewed in the left tail of the distribution. The Jarque-Bera statistic for testing normality is 10.29, so a normal distribution of regression disturbances is decisively rejected. As a consequence, the estimated regression coefficients are not normally distributed, and standard t-tests are inexact at reported levels of statistical significance.

Gross Gas Reserve Additions Gross gas reserve additions are estimated by the regression GGRAt  b 0  b1TSWD t  b 2GDUM2H  b 3GDUM2N  u2t (6.4) where GGRAt  gross gas reserve additions in year t TSWDt  total successful wells drilled in year t GDUM2H  dummy  1 for 1978 and 1986 GDUM2N  dummy  1 for 1988 and 1989 u2t  the disturbance term Estimates are shown in table 6.3. The coefficient of TSWD suggests that an average of 25 BCF of gas reserves are added by each successful well. Apart from the spectacular discoveries in Chiapas-Tabasco and the errant upward revision in 1986, the largest gas reserves were added during the drilling boom that lasted through 1981. The estimated coefficient of GDUM2H reveals the huge discoveries in Chiapas-Tabasco and the large upward revision in 1986. The estimated coefficient of GDUM2N reflects the downward revisions of about 18.6 BCF in 1988 and 1989. The Durbin-Watson statistic suggests that the regression residuals are not autocorrelated. The calculated Jarque-Bera statistic is 0.181, so their discrete distribution approximates normality. The estimated regression coefficients are thus approximately normal, and customary tests of significance are applicable. Table 6.3. Gross Gas Reserve Additions, 1975 –2000 Dependent Variable: GORA Variable

Coefficient

Std. Error

t-Statistic

C TSWD GDUM2H GDUM2N R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood Durbin-Watson stat

2652.938 25.35500 37992.46 18599.47 0.906294 0.893515 4317.143 4.10E08 252.3498 2.293598

2184.568 1.214399 11.39516 2.225067 3212.757 11.82550 3397.052 5.475181 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion F-statistic Probability (F-statistic)

Probability 0.2375 0.0366 0.0000 0.0000 3114.073 13229.80 19.71921 19.91277 70.92526 0.000000

06-A3495 7/27/05 7:13 AM Page 76

76

Chapter Six

summary and conclusions In this chapter we have estimated gross reserve additions of oil and gas attributable to successful drilling. pemex has published the underlying reserves series. These series are not proved reserves but instead are probable ones that meet the criteria established by the wpc and the spe. These criteria permit a broader definition of reserves than those required for proved reserves. For this reason, the two sets of estimates shown in table 6.1 differ by factors of two or three. Gross reserve additions exhibit quite a bit of year-to-year variation. However, as one would expect, reserve additions were larger during the drilling boom of 1975 –1981 than in later years. Estimated oil reserves attained a maximum of about 57 billion barrels between 1981 and 1983, then declined to about 40 billion barrels in 2000. If we ignore the errant upward revision to gas reserves in 1986, long-run reserves reached a maximum of about 85 TCF between 1981and 1984, then followed a path of steady decline to 60 TCF in 2000. Estimated reserves contain measurement errors for various reasons. One is that revisions are inherently subjective. Another is geological uncertainty: It is impossible to correctly estimate the stocks of reserves in the ground or the magnitude of ultimate recovery (Adelman 1995a, 1995b). We adjust gross reserve additions by using dummy variables in years marked by unusual events. After these adjustments, the regressions suggest that gross oil reserves increased by nearly 19 million barrels—and gross gas reserves by about 25 BCF—per successful well. Such large additions signify Mexico’s abundant endowment of oil and gas. These resources can be tapped only by reversing the twenty-year drilling depression that lasted through the year 2000. Increased drilling from 2001 to 2003 indicates that at least a shortterm reversal has occurred. pemex was able to drill 446 wells in 2001, more than the total drilling between 1996 and 1998. The Mexican Congress also approved a 40 percent increase in pemex’s investment budget to $14.7 billion in 2002, with $10.5 billion earmarked for exploration and production (U.S. Energy Information Administration, 2003c). These experiences in 2001 and 2002 might be interpreted to mean that the prolonged drilling depression will be stemmed by an adequate future supply of federal funds. Such an interpretation would be shaky, however, because much of the renewed drilling was financed by private loans. The outcome turns on the ability of President Fox and subsequent administrations to negotiate sustained increases in pemex investment budgets.

07-A3495 7/27/05 7:14 AM Page 77

chapter seven

Production Models

Here we specify and estimate oil and gas production models. Chapter 6 reviews production and reserve trends and analyzes models of exploration and development. We now link current-year oil production to reserves lagged one year and to lagged production. The connection between production and reserves is based on a model developed by Robert Pindyck (1978). Pindyck’s model applies in the strictest sense to a price-taking firm that explores for and produces an exhaustible resource without political constraints. Consistent with Pindyck’s setup, pemex is a price-taking firm that engages in exploration and production. It also faces the same reserves and depletion constraints as any other company. But, as chapter 3 stresses, pemex is subject to constraints imposed by the Mexican Congress. Year-to-year changes in production are often determined by federal mandates unrelated to long-term trends in reserves. The stunning growth in oil production from 1973 to 1982 is consistent with extraordinary reserve growth. In 1987, 1991, 1996 –1998, and 2000, pemex faced mandates to increase oil production; thus we employ a binary variable to account for this intervention. Gas production mirrors oil production because 80 percent of Mexico’s gas has thus far been produced with oil. Gas production experienced rapid growth from 1978 to 1982, consistent with the growth in oil. Then, from 1983 to 1989, gas and oil output decreased steadily, reaching troughs simultaneously in 1989. Oil and gas production rose in nearly fixed proportions throughout the 1990s. Since the lion’s share of gas has been a coproduct of oil, one can easily understand why Mexico’s oil and gas production has coincided up to now. Because oil has played such a central role in foreign exchange and domestic energy (see chapter 1), the government has always given it priority.1 The first part of this chapter focuses on theoretical development. Then we show why pemex’s oil prices are determined exogenously in the world market, and we estimate production models for the full sample 1975 –2000. Later we report two sets of forecasts. First, the models are reestimated using the sample period 1975 –1998, then employed to forecast production in 1999 and 2000. Second, the models are reestimated using samples for the period 1975 –1999, and forecasts are made for the period from 2000 to 2002. Both sets of predictions are remarkably accurate.

77

07-A3495 7/27/05 7:14 AM Page 78

78

Chapter Seven

theoretical background Exploitation of an exhaustible resource from a fixed reserve base whose volume is known with certainty originated in the theoretical work of Hotelling (1931). Hotelling demonstrates that, with constant marginal extraction costs, the optimal rate of production (which coincides with the rate of reserve depletion) requires that price minus marginal extraction cost rise at the real discount rate in a competitive market. Solow (1974) derives these same results as a condition of general-asset market equilibrium. Recent work extends Hotelling’s model by analyzing the effects of uncertainty concerning the size of the reserve base, the appearance of substitutes for the resource, and changes in demand. As one would expect, a resource is extracted more slowly (by a monopolist or a competitive industry) when the reserve base is not known with certainty. Characteristics of extraction paths under reserve uncertainty have been examined by Gilbert (1976b), Heal (1978), and Loury (1978). Dasgupta and Stiglitz (1976), Heal (1976), and Hoel (1978) have studied extraction paths when a substitute for the resource may be introduced in the future. Gilbert (1976a) analyzes the use of exploratory activity to better estimate the size of a fixed volume of reserves. These extensions of Hotelling’s model examine production without considering how reserves were found in the first place. Producers are not endowed with reserves but must instead find them by means of successful exploration. Pindyck (1978) generalizes earlier theoretical work by developing a model of optimal exploration and production. His model permits reserves to increase through successful exploration. As reserves increase, optimal production also increases. However, as reserves are extracted over time, production will eventually exceed gross reserve additions. When this occurs, reserves begin to decline. Pindyck’s model is a convenient reference point to explain the following facts. In a specific geographic or geological region, reserves are discovered through successful drilling. Although the stock of resources cannot ever be known with certainty, probable reserves typically increase for some time, as they did in Mexico. Growth in reserves is followed by growth in production. Eventually, however, reserves must decline. As they do, production inevitably decreases as well. Even though pemex estimates of probable (and therefore highly uncertain) oil reserves have decreased since 1985 (chapter 6, figure 6.2), we seriously doubt that Mexico has reached a stage of irreversible decline. We are virtually certain that Mexico’s nonassociated gas reserves remain largely undiscovered because they have thus far been largely ignored. Successful exploration and development are the only means of replenishing reserves. The desired level of reserves, however, depends in part on production costs. If production costs were independent of the stock of reserves (and if there were no uncertainty about the size of discoveries resulting from exploration), producers would maintain lower stocks of reserves. In fact, production costs rise

07-A3495 7/27/05 7:14 AM Page 79

Production Models

79

as reserves decline, although the exact relationship between the two is complex.2 A private firm striving to maximize market value simultaneously determines optimal investment in exploration and production and thereby balances exploration costs, production costs, and the user cost of depletion. pemex is anything but private. It has no stock-market value since it is a creature of the government that cannot be sold. It faces no domestic competition whatsoever. Yet its long-run production is inevitably constrained by reserves. Pindyck’s model is a starting point that helps us to understand exploration and production even for pemex.

a theory of a price-taking firm producing an exhaustible resource Our starting point is Pindyck’s model of a price-taking firm that explores for and produces an exhaustible resource. The goal is to allocate resources over time so as to maximize the present value of the firm. Although this model has been successfully employed before (Moroney 1997), it is obviously not strictly applicable to pemex. There are two related reasons. First, because pemex is a government monopoly, its exploration and production are determined politically rather than by principles of wealth maximization. Second, as chapter 3 points out, its exploration and development budgets are determined by Byzantine negotiations with the government. Oil and gas production are set to meet political goals instead of private-wealth maximization. Yet we can exploit Pindyck’s framework to develop simple econometric models that are useful for estimating oil and gas production. The most important result is that current-year production depends on both production and yearend reserves of the preceding year. Pindyck’s model can be summarized as follows. The firm takes price p as given, then chooses production Y out of proven reserves R and exploration effort w to maximize the present value of the firm. The average cost of production C1(R) increases as the proved reserve base is depleted, and C1(R) S q as R S 0. The firm adds to its gross reserves by increasing exploration effort w. The rate at which the firm adds to gross reserves depends both on w and cumulative reserve additions x: dx  f 1w, x2 dt

(7.1)

Denoting first derivatives of f with respect to its arguments by subscripts, fw 0 (the marginal product of exploration effort is positive) fx  0 (larger cumulative reserves, and thus reserve depletion, reduce the rate of new reserve findings, if w is given). As exploration and discovery continue, making new discoveries becomes more costly. The cost of exploratory effort C2(w) increases with w. The firm

07-A3495 7/27/05 7:14 AM Page 80

80

Chapter Seven

then chooses production and exploration investment to maximize its discounted present value: q

Max PV  y,w

 3 pY  C 1R2Y  C 1w2 4e 1

2

rt

dt

(7.2)

0

subject to dx dR  Y dt dt dx  f 1w, x2 dt and nonnegativity constraints R 0, Y 0, w 0, and x 0. The solution to this optimization problem is straightforward. The Hamiltonian is H  pYert  C1 1R2 Yert  C2 1w2ert  l1 c

dx dx  Y d  l2 a b dt dt

(7.3)

The Hamiltonian is a linear function of Y. Differentiating (7.3) with respect to Y, HY  pert  C 1 1R2ert  l1

(7.4)

If ert[ p  C1(R)]  l1  0, the firm produces nothing. Economically speaking, l1 is the change in the present value of future profits resulting from finding an additional unit of reserves; hence, l1 is strictly positive. In reservoir-engineering literature, this circumstance is referred to as the “economic limit.” However, if ert [ p  C1(R)]  l1 0, the firm optimizes by producing at a maximum capacity rate, constrained by its current reserves: Y*  Y*(R), where Y* is the firm’s optimal production rate.3 The economically important point is that the firm decides to produce either nothing or something by comparing price with its average cost of extraction and the change in the present value of future profits from an additional unit of reserves. If positive production is optimal, it produces at a maximum rate, constrained by reserves and geophysical variables. This intertemporal model of a price-taking firm is perfectly consistent with the reservoir- engineering maxim: If a well or reservoir is economically feasible, produce it at the maximum rate subject to the reserves constraint and independently of price. This result contrasts with the static economic model, in which a price-taking firm maximizes profit or minimizes loss in each period by producing where marginal cost equals price (provided that revenues cover variable costs).

07-A3495 7/27/05 7:14 AM Page 81

Production Models

81

pemex as a price-taking firm Although pemex is the only producer of oil and gas in Mexico, it has never influenced the world price of oil. This is because pemex has always produced such a small fraction of the world’s oil that it cannot significantly influence world supply. From 1975 to 2000, pemex production ranged from 262 million barrels (1975) to 1.103 billion barrels (2000) but never exceeded 4 percent of world production. Mexico also harbors a meager fraction of the world’s oil reserves. Even though the country’s probable oil reserves ranked ninth among all countries in 2000, its reserves amounted to only 3 percent of the combined reserves in the top eight countries.4 Mexico’s gas production in the year 2000 was 1.713 TCF, or only 2.5 percent of the top nine countries’ annual production. Since natural-gas markets are far more segmented than those for oil, pemex surely exercises some pricing power over its domestic natural gas. Yet Mexico produces so little gas that any domestic pricing power pemex exercises is unimportant compared to its role as a price taker for oil.5 pemex takes the price of West Texas Intermediate (wti) crude oil as a reference point for negotiating the prices of its three principal exports: Maya-22, Isthmus-34, and Olmeca. Then pemex International (pmi) negotiates the prices of these three exports based on the quality and physical properties of the oil. Prices (in nominal U.S. dollars) are shown in table 7.1. These bear a con-

Table 7.1. Negotiated Export Price for Different Types of Oil in Mexico (Dollars per Barrel)

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Maya-22 (heavy oil)

Isthmus-34 (light oil)

Olmeca* (super light oil)

28.65 31.01 25.23 23.95 25.32 24.04 10.56 15.12 11.15 14.38 17.04 12.20 13.06 11.37 12.57 14.41 17.04 14.65 8.56 14.18 22.81

33.24 35.94 32.81 29.54 29.00 27.15 13.50 17.48 13.89 17.09 22.70 18.08 17.96 15.79 15.33 16.66 19.77 18.19 11.81 17.47 27.67

14.33 18.76 23.58 20.02 19.50 16.95 16.27 17.51 21.12 19.52 13.11 17.92 28.94

* Exports of Olmeca began in 1988.

07-A3495 7/27/05 7:14 AM Page 82

82

Chapter Seven

sistent relationship with one another—Maya always the lowest, and Olmeca always the highest. And their year-to-year changes correspond to fluctuations in world oil prices. pemex could do nothing to prevent the price collapses in 1986 and 1998 because its negotiated prices necessarily followed the lower world prices. Note also that, between 1982 and 2000, the prices of Maya-22 and Isthmus-34 never recovered to their 1981 levels. pemex could do nothing to control the prices of its oil exports. It could only negotiate in view of prevailing world prices.

linear production models We first estimate linear oil production models. The determinants of long-term production are lagged reserves and lagged production. Reserves are the primary geophysical constraint on production because reserves are inventory from which production occurs. Second, we estimate gas production as a linear function of oil production because more than 80 percent of Mexico’s gas is produced with oil. If the R /P ratio were constant (and of course it is not), one could predict changes in production exactly from changes in proved reserves. Stated differently, if the R /P ratio were constant, the percentage change in production each year would be equal to the corresponding percentage change in reserves. The elasticity of production with respect to reserves would be one. As chapter 6 points out, however, R /P ratios vary widely. From 1973 to 1982, Mexico’s oil production increased sixfold, and estimated reserves increased by a factor of fourteen. These were years of rapid growth in exports and foreign exchange. After 1981, export prices began to weaken slightly (see table 7.1), and the peso/dollar exchange rate was sharply devalued in 1983. Then, in 1986, two financial events caused federal tax revenues to plummet. First, oil prices plunged to less than half their 1985 values (table 7.1). Second, the peso was devalued by more than half against the U.S. dollar. These events forced the Mexican Congress to seek additional tax revenues. Two obvious possibilities were to increase oil production and production tax rates. In its quest for greater revenue, Congress raised production tax rates and in several years mandated higher production. We include a binary variable that accounts for the increase in oil production in response to these mandates. It is denoted as ODUM6 for 1987, 1991, 1996 –1998, and 2000.

Lagged Adjustment to Equilibrium Production pemex cannot adjust to its desired production level within one year. Differences between actual and desired production from a specific well or reservoir can be attributed to unexpected changes in pressure, unexpected water intrusion, and a host of other technical reasons. Aggregate production is another matter: The discrepancy between desired and actual production is sharply influenced by political decisions.6 Accordingly, a simple adjustment process is specified. Let Y*t be the desired rate of production in year t, and let Yt be the observed rate of production. A

07-A3495 7/27/05 7:14 AM Page 83

Production Models

83

simple Koyck-type model is used, in which Yt1 is adjusted to the desired Y *t as Yt  Yt1  d1Y*t  Yt1 2,

0d1

(7.5)

where d is the percentage of adjustment achieved within a year. As extreme cases, if d  0, then Yt  Yt1 and adjustment is 0. Furthermore, if d  1, then Yt  Y *t and full adjustment occurs within one year. However, if 0  d  1, then annual adjustment is incomplete. If desired production depends linearly on last year’s end-of-year reserves, then desired production is Y*t  a  bRt1,

b0

(7.6)

where Rt1 is last year’s reserves. Substituting equation (7.6) in equation (7.5), one obtains Yt  dY*t  11  d2Yt1

Yt  da  dbRt1  11  d2Yt1

(7.7)

Equation (7.7) can be written as a regression model: Yt  b1  b2Rt1  b3Yt1  e1t

(7.8)

Estimates of the coefficients d, a, and b can be obtained from the estimated regression coefficients as follows: bˆ 1 dˆ

bˆ 1  da,

so

aˆ 

bˆ 2  db,

so

bˆ 2 bˆ  dˆ

bˆ 3  11  d 2,

so

dˆ  1  bˆ 3

If the disturbance term e1t in equation (7.8) is serially independent with a mean of 0 and if Rt1 is predetermined (which it must be) and measured accurately (which it is not), then ordinary least squares applied to equation (7.8) yield unbiased, efficient, and consistent estimates of b1, b2, b3, and d. If e1t is normally distributed, then the estimates of b1, b2, b3, and d are also normally distributed. Since aˆ and bˆ are ratios of the estimated coefficients bˆ1, bˆ2, and bˆ3, then aˆ and bˆ are consistent estimators. Finally, if e1t is distributed normally, then aˆ and bˆ are asymptotically normal, maximum likelihood estimators, conditional on a properly specified model. However, if the disturbance term in equation (7.8) is serially correlated, then the estimates of b1, b2, and b3 are inconsistent. If so, the estimates of a, b, and d are also inconsistent. Tests for serially independent error terms are thus

07-A3495 7/27/05 7:14 AM Page 84

84

Chapter Seven

important because their results partially govern the statistical properties of the least-squares estimators.

statistical estimates We first specify that current oil production depends on reserves and production of the preceding year and, in certain years, on federal mandates. Second, we specify a simple equation that links gas and oil production.

Oil Production The structural equation relating oil production to reserves, lagged production, and political mandates is OPROt  b1  b2ORESt1  b3OPROt1  b4ODUM6  e1t (7.9) where OPROt  oil production in million barrels for year t ORESt1  end-of-year oil reserves in million barrels for year t1 OPROt1  oil production in million barrels lagged one year ODUM6  a binary variable equal to 1 for 1987, 1991, 1996 –1998, and 2000 and zero otherwise e1t  the disturbance term We first apply augmented Dickey-Fuller tests to the series covering production and reserves and find both series to be nonstationary. However, we are able to reject the hypotheses of unit roots at P  0.05 if the series are expressed in first differences. Accordingly, we estimate b2, b3, and b4 by estimating equation (7.9) expressed in first differences, with the results given in table 7.2. Estimates of b2 and b4 are significant at P  0.01. The estimate of b4 implies that, in response to congressional directives during the six years, pemex increased production by some 63 million barrels annually. Since production in these years averaged roughly 1 billion barrels, the 63-million-barrel increase amounts to about 6 percent of actual production.

Table 7.2. Oil Production Model, 1975 –2000 Variable

Coefficient

Std. Error

t-Statistic

Probability

D(ORES(1)) D(OPRO(1)) ODUM6 R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood

0.007357 0.203998 62.68447 0.489054 0.444624 46.58980 49924.02 135.1745

0.002377 3.094961 0.177543 1.149007 19.46665 3.220095 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion Durbin-Watson stat.

0.0051 0.2624 0.0038 34.34242 62.51690 10.62881 10.77397 2.480969

07-A3495 7/27/05 7:14 AM Page 85

Production Models

85

The Breusch-Godfrey test indicates that first-order serial correlation in disturbances can be rejected at P  0.05. Moreover, the Jarque-Bera test allows rejection of the hypothesis of nonnormal disturbances.

Gas Production Until recently, approximately 80 percent of Mexico’s gas has been produced with oil.7 Since gas and oil have thus far been largely produced as coproducts, we specify a simple structural equation that connects gas with oil production: GPROt  b1  b2OPRO t  e2t

(7.10)

where GPROt  total natural gas production in year t measured in BCF OPROt  oil production in year t measured in million barrels e2t  the disturbance term The gas production series is also nonstationary. Because of nonstationarity, least-squares estimates of b1 and b2 in equation (7.10) cannot be interpreted as valid parameter estimates. First differences in the gas production series satisfy the augmented Dickey-Fuller test for stationarity at P  0.05. Accordingly, we respecify and estimate equation (7.10) in first differences, with the results shown in table 7.3. Since the estimate of b2 is five times its standard error, gas and oil production are closely linked, as one would expect. The estimate of b2 suggests that about one BCF of gas are produced with each million barrels of oil. The regression residuals are well behaved. The Breusch-Godfrey test statistic for autocorrelation is a minuscule 0.416, and the Jarque-Bera statistic for testing normality is 0.120. Equation (7.10), restated in first differences, yields a well-specified regression linking gas and oil production. Recall that only 20 percent of gas is nonassociated gas produced independently from oil. As nonassociated gas becomes a more important part of Mexico’s overall gas production, it will be advisable to model associated and nonassociated gas separately (Moroney 1997, chapter 4). But for now it only makes sense to model gas as a coproduct of oil.

Table 7.3. Gas Production Model, 1975 –2000 Variable

Coefficient

Std. Error

t-Statistic

Probability

D(OPRO) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood

1.016325 0.435737 0.435737 70.70698 124986.9 147.1047

0.197344 5.150011 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion Durbin-Watson stat.

0.0000 37.25254 94.12855 11.39267 11.44106 1.711405

07-A3495 7/27/05 7:14 AM Page 86

86

Chapter Seven

forecasting properties of the models Equation (7.9) highlights the importance of lagged reserves and lagged production in determining oil production; equation (7.10) provides a link between oil and gas. However, an important question for energy policymakers remains: How well do these equations forecast future production? If they predict with tolerable accuracy, they are reliable instruments to project oil and gas supplies, which are cornerstones of sustainable development. The practical value of these equations rests on the accuracy of their forecasts. We obtained recently published reserves and production for 2001 and unpublished production and reserve estimates for 2002.8 First, we estimate equations (7.9) and (7.10) using the subsample 1975 – 1998, then forecast oil and gas production from 1999 to 2000. Second, we estimate these equations using the subsample 1975 –1999, then forecast production for the years 2000 to 2002. Define the forecast error in year t as the difference between actual production Yt and the forecast of production, Y Ft: et  Yt  Y Ft

(7.11)

If et is positive (negative), the forecast is too low (high). To evaluate the predictive accuracy of the equations, three measures of forecast error are used: the mean absolute error (MAE), the mean absolute percentage error (MAPE), and the root mean square error (RMSE). The accuracy of a forecast in a specific year can be gauged by the absolute value of the forecast error, et. To describe the ability of a model to predict, say, three years ahead, it is best to average the three absolute errors, giving the mean absolute error: MAE 

1e2000  e2001  e20022

(7.12)

3

The MAE of course depends on the units in which the object being forecast is measured (e.g., million barrels of oil or billion cubic feet of gas). To compare forecasts of objects measured in different units, it is preferable to choose a basis that is independent of units of measurement, the mean absolute percentage error (MAPE):

MAPE 

c

e2000 Y2000



e2001 Y2001 3



e2002 Y2002

d

(7.13)

This calculation gives the average of the absolute percentage errors of forecasts in 2000, 2001, and 2002. Another frequently used measure of predictive accuracy is the root mean square error. Because the RMSE entails squaring the actual forecast errors, it

07-A3495 7/27/05 7:14 AM Page 87

Production Models

87

gives more weight to larger errors than MAE or MAPE. The RMSE is calculated as RMSE 

e 22000  e 22001  e 22002 B 3

(7.14)

Production Forecasts for 1999 and 2000 Equations (7.9) and (7.10) are expressed in first differences and reestimated using subsamples for 1975 –1998. Estimated coefficients of the oil production model are shown in table 7.4, and those of the gas production model in table 7.5. The estimated coefficients and overall goodness of fit are quite similar to the full-sample estimates (compare table 7.4 with table 7.2; also compare table 7.5 with table 7.3). Actual production and forecasts are shown in table 7.6. The oil production forecasts are somewhat too high in both years; the gas production forecast is a bit low in 1999 and slightly high in 2000.

Table 7.4. Oil Production Model: Subsample, 1975 –1998 Variable

Coefficient

Std. Error

t-Statistic

Probability

D(ORES(1)) D(OPRO(1)) ODUM6 R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood

0.007371 0.204602 65.19878 0.479620 0.430061 46.85925 46111.58 124.7837

0.002508 2.939425 0.189718 1.078457 22.15596 2.942720 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion Durbin-Watson stat.

0.0078 0.2931 0.0078 37.94562 62.06989 10.64864 10.79590 2.615014

Table 7.5. Gas Production Model: Subsample, 1975 –1998 Variable

Coefficient

Std. Error

t-Statistic

Probability

D(OPRO) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood

1.071005 0.463132 0.463132 70.72529 115047.5 –135.7551

0.201523 5.314547 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion Durbin-Watson stat.

0.0000 41.85000 96.52524 11.39626 11.44534 1.719691

Table 7.6. Forecast vs. Actual Production Oil (million barrels)

Gas (BCF)

Year

Forecast

Actual

Forecast

Actual

1999 2000 Total

1121.642 1117.697 2239.339

1061.420 1102.758 2164.178

1685.751 1792.988 3478.739

1748.715 1713.246 3461.961

07-A3495 7/27/05 7:14 AM Page 88

88

Chapter Seven Table 7.7. Oil Production Model: Summary of Forecast Errors, 1999 –2000 Forecast: OPROF Actual: OPRO Forecast sample: 1999 –2000 Included observations: 2 Root mean squared error Mean absolute error Mean absolute percentage error Theil inequality coefficient Bias proportion Variance proportion Covariance proportion

43.87436 37.58089 3.514243 0.019925 0.733690 0.181593 0.084717

Table 7.8. Gas Production Model: Summary of Forecast Errors, 1999 –2000 Forecast: GPROF Actual: GPRO Forecast sample: 1999 –2000 Included observations: 2 Root mean squared error Mean absolute error Mean absolute percentage error Theil inequality coefficient Bias proportion Variance proportion Covariance proportion

71.84436 71.35286 4.127505 0.020697 0.013635 0.249467 0.736898

Table 7.9. Oil Production Model: Subsample, 1975 –1999 Variable

Coefficient

Std. Error

t-Statistic

Probability

D(ORES(1)) D(OPRO(1)) ODUM6 R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood

0.007586 0.184750 65.89065 0.490943 0.444665 47.53638 49713.56 130.4129

0.002538 2.988699 0.191816 0.963165 22.46944 2.932456 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion Durbin-Watson stat.

0.0068 0.3459 0.0077 34.06260 63.78942 10.67303 10.81930 2.421926

The forecast errors are summarized in tables 7.7 and 7.8. The mean absolute error for oil is 37.58 million barrels annually, which translates into a mean absolute percentage error of 3.51 percent. The mean absolute error for gas is 71.35 BCF annually, or 4.13 percent. Equations (7.9) and (7.10) produce remarkably accurate two-year forecasts.

Production Forecasts for 2000, 2001, and 2002 Next, equations (7.9) and (7.10) are expressed in first differences and estimated with the 1975 –1999 subsample. Oil production estimates are shown in table 7.9, and gas production estimates in table 7.10. The estimated coefficients

07-A3495 7/27/05 7:14 AM Page 89

Production Models

89

Table 7.10. Gas Production Model: Subsample, 1975 –1999 Variable

Coefficient

Std. Error

t-Statistic

Probability

D(OPRO) R-squared Adjusted R-squared SE of regression Sum of squared residuals Log likelihood

1.041612 0.449537 0.449537 70.38659 118902.5 –141.3133

0.197771 5.266764 Mean dependent var. S.D. dependent var. Akaike info criterion Schwarz criterion Durbin-Watson stat.

0.0000 40.16140 94.86935 11.38506 11.43382 1.656205

Table 7.11. Forecast vs. Actual Production Oil (million barrels)

Gas (BCF)

Year

Forecast

Actual

Forecast

Actual

2000 2001 2002 Total

1119.656 1118.452 1105.851 3343.959

1102.758 1141.355 1192.820 3436.933

1780.507 1753.811 1699.706 5234.024

1713.246 1646.515 1651.990 5011.751

are nearly identical to the original estimates, suggesting that the estimates are robust with respect to subsamples. Oil and gas forecasts are shown in table 7.11. The oil forecasts in 2000 and 2001 are exceedingly accurate, but the prediction in 2002 is about 87 million barrels (or about 7 percent) too low. The gas forecasts are all slightly too high. You can see that the cumulative three-year forecasts correspond closely to cumulative production. A summary of forecast errors in tables 7.12 and 7.13 shows that these equations yield reliable forecasts. The accuracy of these two- and three-year forecasts reveals the practical value of these equations. The message is clear: The only key to long-term growth in production is growth in reserves.

conclusions This chapter estimates production equations for oil and gas. These equations are based on a theoretical model proposed by Pindyck (1978). This model, modified to account for political influences, proves to be a useful starting point for analyzing Mexico’s oil and gas production. The modified models closely fit the history of oil and gas production since 1975. They also produce remarkably accurate forecasts. We believe that these simple models include the essential variables required to analyze and forecast production. Although pemex produces all of Mexico’s oil and gas, it is a price taker in the world oil market. pemex takes the price of West Texas Intermediate crude oil as the basis for negotiating export prices, which change in phase with world prices. pemex is under the authority of the Mexican Congress for meeting production targets. Reserves are the ultimate constraint on production, but politics are also important.

07-A3495 7/27/05 7:14 AM Page 90

90

Chapter Seven Table 7.12. Oil Production Model: Summary of Forecast Errors, 2000 –2002 Forecast: OPROF Actual: OPRO Forecast sample: 2000 –2002 Included observations: 3 Root mean squared error Mean absolute error Mean absolute percentage error Theil inequality coefficient Bias proportion Variance proportion Covariance proportion

42.20668 38.84430 3.349160 0.018639 0.426972 0.258789 0.314239

Table 7.13. Gas Production Model: Summary of Forecast Errors, 2000 –2002 Forecast: GPROF Actual: GPRO Forecast sample: 2000 –2002 Included observations: 3 Root mean squared error Mean absolute error Mean absolute percentage error Theil inequality coefficient Bias proportion Variance proportion Covariance proportion

81.48223 77.86427 4.663884 0.023827 0.913168 0.008101 0.078731

The prolonged decline in oil prices after 1986 prompted congressional pressure for pemex to increase production. After all, oil exports are an essential source of foreign exchange, and production taxes are a major source of government revenue. Executives in pemex and the Secretary of Energy fully understand that reserves are the lifeblood of production. New reserves, however, require several years to be developed. Successful drilling over many years is the single key to long-term reserve growth. Successful drilling, in turn, is a continuous process that requires multiyear plans based on adequate drilling budgets. The provision of adequate financial resources to pemex must be the foundation of sustainableenergy development in Mexico. As of 2002, 83 percent of Mexico’s reported gas reserves were associated gas. Nonassociated gas reserves have been largely neglected. Federal officials now clearly recognize the vital importance of nonassociated gas as a future source of energy. To develop, produce, and transport nonassociated gas will require that the peso equivalent of tens of billions of U.S. dollars be invested in developing reserves and creating a country-wide pipeline infrastructure. As nonassociated gas becomes more prominent, its production should be analyzed independently of oil. This analysis will require estimates of nonassociated gas reserves that are not currently available. These estimates will quite likely become available during the next ten years for this important line of research.

08-A3495 7/27/05 7:14 AM Page 91

chapter eight

Simulating the Integrated Oil and Gas Supply Model The Fox administration understands the urgency of developing oil and gas reserves. Chapter 1 shows that these fuels dominate Mexico’s commercial energy and are the only realistic hope for increasing it in the future. Oil and gas are two pillars on which many of the goals of sustainable development ultimately rest. For this reason, Congress approved vastly larger drilling budgets enabling pemex to drill 449 wells in 2001, 457 in 2002, and 631 in 2003. This three-year total of 1,537 wells exceeds the total number of wells drilled in the ten years from 1991 to 2000. This drilling spree is surely responsible for the sharply increasing oil and gas reserves reported in 2003. The Secretary of Energy has released ambitious production targets to be reached in 2006: 1,414 million barrels of oil and 2,810 BCF of gas per year. For 2006, pemex has proposed an even higher oil production target of 1,460 million barrels but a considerably lower goal of 2,190 BCF for gas.1 Here we simulate the integrated oil and gas supply model developed in chapters 4 –7. Recall that each sector of the model— drilling, successful drilling, gross reserve additions, and production—is estimated using data from 1975 to 2000. We employ these structural estimates to simulate future values of these variables for the years 2003 –2008. These simulations allow us to compare the impact of different assumptions concerning real oil prices and pemex taxes. They also permit comparisons of oil and gas production obtained from our integrated structural model with the targets announced by the Secretary of Energy and pemex. The simulations are driven by one financial variable: the net (after-tax) real price of exported oil (Maya-22, Isthmus-34, and Olmeca). This price is a volume-weighted average of market prices, adjusted downward according to the pemex tax rate. The simulations thus require assumptions concerning future values of oil prices and tax rates. It is impossible to forecast oil prices accurately. Instead of attempting predictions, we adopt two assumptions made by the Mexican Petroleum Institute. First, we assume a low price path starting with the actual real price of $21.49 per barrel in 2003 and gradually increasing to $21.67 in 2008.2 Second, we assume a higher real price of $25 per barrel from 2003 to 2008. The pemex tax burden, which Congress determines, is also impossible to predict. We employ two tax rates in the simulations. First, we assume that the tax rate remains at 61 percent from 2003 through 2008. Second, we assume that, under a moderate tax reform, it is reduced to 50.2 percent, which was the actual pemex tax rate in 1980. 91

08-A3495 7/27/05 7:14 AM Page 92

92

Chapter Eight

Figure 8.1. Simulation of four oil-price and tax-rate scenarios.

These combinations of real oil prices and tax rates enable us to project the impact of these financial variables on drilling, successful drilling, gross reserves added, the level of reserves, and production. The assumptions concerning prices and tax rates used in the four simulations are shown in figure 8.1. Simulation 1 incorporates the most stringent financial assumptions of low real oil prices ($21.49 to $21.67 per barrel) and the actual 61 percent pemex tax rate in 2002. It therefore provides the least after-tax cash flow to pemex. Simulations 2 and 3 involve less stringent financial assumptions. Simulation 2 combines the assumption of low real oil prices with a substantially lower 50.2 percent tax rate, while Simulation 3 is based on a $25 real oil price and the higher 61 percent tax rate. Finally, Simulation 4 is based on a $25 real oil price and the lower 50.2 percent tax rate, thus providing the largest after-tax cash flow. We project different time paths for drilling, gross reserve additions, reserves, and production in response to these real oil prices and tax rates.

structural coefficients and assumptions We simulate the integrated model using gross oil and gas reserves added per successful well slightly lower than the estimates shown in tables 6.2 and 6.3. We also employ the actual success ratios in exploration and development drilling that prevailed in 2002.

Structural Coefficients The structural coefficients estimated in chapter 4 (drilling) and chapter 7 (production) are employed in the following simulations. These coefficients are estimated using statistical data for these sectors spanning the years 1975 –2000.

08-A3495 7/27/05 7:14 AM Page 93

Simulating the Integrated Oil and Gas Supply Model

93

Thus the simulations are based on a model that is grounded in recent historical experience.

Drilling We assume that drilling begins with the 631 wells scheduled in 2003. Drilling was financed by a combination of budgets approved by Congress and debt issued by pemex through pidiregas. In chapter 3 we stress that additional debt incurred through pidiregas poses a threat to pemex’s long-term viability. International debt-rating services have reacted to this problem by downgrading their ratings of pemex debt. For example, on October 3, 2003, pemex issued $500 million of bonds due to mature in 2009. Moody’s investor service rates these bonds at Bal, while Standard and Poor’s rates them at BBB–, the minimum rating for investment-grade bonds.

Successful Drilling We fix the success ratios in exploration and development at 0.50 and 0.88, their actual values in 2002. In doing so, we assume that the efficiency in exploration and development will remain as it was in 2002 throughout the simulation period.

Gross Reserves Added per Successful Well We assume an average of 16.04 million barrels of oil and 20.41 BCF of gas added per successful well. We stress that our assumption of constant reserves added per successful well ignores the possibility of reserve depletion and lower reserves added for successful wells. Although reserves added per successful well must eventually diminish, our assumption seems quite reasonable for shortterm simulations.

Dummy Variable for pidiregas The dummy variable for pidiregas is 1 for 2001–2006 and 0 for the remaining years. Our reasoning is that private financing through pidiregas was already in place during the years 2001–2004.3 We believe that Congress will continue to authorize additional debt through pidiregas until the end of President Fox’s administration in 2006. We assume that, following a change in administration, Congress will not permit pemex to continue private borrowing.4

Dummy Variable ODUM6 A dummy variable, ODUM6, is one for the entire simulation period, on the supposition that Congress will continue to stress oil exports as a source of foreign currency.

Lags in Adjustment In two sectors, drilling (chapter 4) and production (chapter 7), we distinguish between one-year and long-run equilibrium adjustments. The estimated coefficients of adjustment are explicitly incorporated in the simulations.

08-A3495 7/27/05 7:14 AM Page 94

94

Chapter Eight

Reserves/Production Ratios We place no restrictions on either the oil or gas (reserves/production) ratios because of their historical variability. Nonetheless, both ratios exhibit only slight variation throughout the simulation period.

Abnormally Large Oil or Gas Discoveries We disallow the possibility of abnormally large oil or gas discoveries but assume that discoveries of average size will continue throughout the period 2003 –2008.

Nonassociated Gas Reserves Congress now places higher priority on the development of nonassociated gas reserves in Tabasco, Veracruz, Tampico, Misantla, and the Burgos Basins in Tamaulipas. pemex will almost certainly invest more in developing these reserves. However, these investments cannot be known at present, so these prospective projects are not considered in the simulations.

Future Values Historical and simulated future values of oil and gas reserves are based on probable reserves, as discussed in chapter 3.

simulating the integrated oil and gas model In comparing the simulations, it helps to keep in mind that if oil prices remain low but the tax rate is cut from 61 to 50.2 percent (Simulation 2), the stream of annual cash flows increases by about 18 percent in relation to Simulation 1. Likewise, the combination of the higher $25 oil price and the 61 percent tax rate (Simulation 3) increases cash flows by about 16 percent in relation to Simulation 1. Finally, the combination of the $25 oil price and the lower 50.2 percent tax rate increases cash flows by approximately 34 percent and therefore produces the strongest response in drilling and gross reserve additions.

Simulations of Drilling, Successful Drilling, and Gross Reserve Additions Inspection of table 8.1 yields some obvious generalizations: (1) Simulation 1 produces the lowest time profiles of drilling and successful wells, while Simulation 4 shows the highest values. Because of substantially larger cash flows and drilling investment, Simulation 4 yields a projection of 328 more successful wells than Simulation 1; (2) Simulations 2 and 3 yield roughly equivalent drilling profiles because they are based on nearly equivalent after-tax cash flows; (3) the larger number of successful wells in Simulation 4 yields projections of much larger cumulative gross reserve additions than Simulation 1: 5,262 million barrels more oil reserves and 6,694 BCF more gas reserves; and (4) since Simulations 2 and 3 exhibit similar drilling profiles, their projected gross reserves added are about the same.

08-A3495 7/27/05 7:14 AM Page 95

Simulating the Integrated Oil and Gas Supply Model

95

Table 8.1. Simulations of Total Wells, Total Successful Wells, Gross Oil Reserve Additions, and Gross Gas Reserve Additions, 2003 –2008 Simulation 1: Real oil prices of $21.49 –$21.67; tax rate  61%

Simulation 2: Real oil prices of $21.49 –$21.67; tax rate  50.2%

Simulation 3: Real oil price of $25.00; tax rate  61%

Simulation 4: Real oil price of $25.00; tax rate  50.2%

Year 2003 2004 2005 2006 2007 2008 Sum

Total wells 631 481 394 348 281 245 2380

Total wells 635 506 430 390 327 294 2582

Total wells 633 499 425 384 319 282 2542

Total wells 638 529 470 437 376 342 2792

Year 2003 2004 2005 2006 2007 2008 Sum

Successful wells 536 406 331 290 232 200 1995

Successful wells 538 425 359 324 269 240 2155

Successful wells 537 420 356 320 262 231 2126

Successful wells 540 444 391 362 308 278 2323

Year 2003 2004 2005 2006 2007 2008 Sum

Gross oil reserve additions (million barrels) 8597 6512 5309 4652 3721 3208 31999

Gross oil reserve additions (million barrels) 8630 6817 5758 5197 4315 3850 34567

Gross oil reserve additions (million barrels) 8613 6737 5710 5133 4202 3705 34100

Gross oil reserve additions (million barrels) 8661 7122 6272 5807 4940 4459 37261

Year 2003 2004 2005 2006 2007 2008 Sum

Gross gas reserve additions (BCF) 10940 8286 6756 5919 4735 4082 40718

Gross gas reserve additions (BCF) 10981 8675 7327 6613 5490 4898 43984

Gross gas reserve additions (BCF) 10960 8572 7266 6531 5347 4715 43391

Gross gas reserve additions (BCF) 11022 9062 7980 7388 6286 5674 47412

We stress that these simulations are not forecasts. The reason is that forecasts would require, as a starting point, statistical predictions of oil prices and tax rates. We eschew such forecasts for two reasons. In the first place, world oil prices are notoriously unstable and therefore unpredictable. They are determined by economic and political events beyond the boundaries of Mexico. We avoid any attempt to forecast tax rates for this reason: The Fox administration faces majority opposition in both houses of Congress, so it is impossible to predict tax rates with this gridlock.

08-A3495 7/27/05 7:14 AM Page 96

96

Chapter Eight

Instead, the simulations show what could occur in four hypothetical financial environments. The simulations are best thought of as “what could happen if oil prices and tax rates were to change in such and such a way.” By no means do we suggest that oil prices and tax rates will in fact change in such and such a way, but the simulations tell us what could happen if they do. There are two obvious advantages in simulating the integrated model. In the first instance, the simulations are based on a complete set of sectoral coefficients. Second, the integrated model is extremely flexible because the prices and tax rates that drive it can easily be modified.

Simulating Oil Reserves and Production Given the projections of successful drilling and gross reserves added (shown in table 8.1), we are able to project alternative time paths of production and reserves for the years 2003 –2008. Our starting point is actual production in 2002 and estimated reserves at year-end 2002. Reserves are of course augmented by gross reserve additions and depleted by production. Because of the gross reserve additions in table 8.1, one expects Simulation 1 to yield the lowest projections, Simulation 4 the highest, and Simulations 2 and 3 intermediate projections. We concentrate on Simulations 1 and 4 because they bracket the results of Simulations 2 and 3. Simulation 1: Real Oil Prices of $21.49 –$21.67 and 61 Percent Tax Rate Historical values of oil reserves and production, together with projections for 2003 –2008, are shown in table 8.2 and figure 8.2. The surge in successful drilling throughout the years 2001–2003 promotes rapid growth in oil reserves. Projected oil reserves increase from 37,150 million barrels in 2002 to 60,024 million barrels in 2008. By 2007 and 2008, reserves recover to levels slightly more than those in 1983 (57,096 million barrels). Because of rapid reserve growth, oil production increases by almost 48 percent, from 1,218 million barrels in 2003 to 1,799 million in 2008. Projected production in 2006 (1,590 million barrels) slightly exceeds that year’s production targets announced by the Energy Secretary (1,414 million barrels) and pemex (1,460 million barrels). The reserves/production ratio decreases slightly from 36.57 in 2003 to 33.36 in 2008. These unrestricted ratios are obviously consistent with actual R /P ratios from 1997 to 2002. Simulation 4: $25.00 Real Oil Price and 50.2 Percent Tax Rate The $25 oil price combined with a lower tax rate provides the largest cash flow, thereby prompting the highest time path for drilling. As one would expect, accelerated drilling produces the greater reserve growth and production. Table 8.2 and figure 8.3 show that oil reserves reach a level of 65,209 million barrels in 2008. Production rises to 1,603 million barrels by 2006, easily surpassing the production targets of both pemex and the Energy Secretary.5 This combination

08-A3495 7/27/05 7:14 AM Page 97

Table 8.2. Simulating Oil Reserves and Production, 2003 –2008 Simulation 1: Real oil prices of $21.49 to $21.67; tax rate  61 percent

Simulation 4: Real oil price of $25.00; tax rate  50.2 percent

Year

Total oil reserves (million barrels)

Oil production (million Reserves/ barrels) production

Year

Total oil reserves (million barrels)

1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

57096.0 56410.0 56810.0 48041.0 47176.0 46222.0 45250.0 44560.0 44295.0 44439.0 44043.0 43127.0 42146.0 42072.0 41392.0 41064.0 41495.0 39918.0 38286.1 37150.1 44529.9 49695.7 53531.4 56592.8 58614.8 60023.7

981.222 1024.341 986.697 912.639 954.990 940.844 926.165 930.024 976.683 976.379 975.787 980.075 955.205 1046.394 1103.395 1120.550 1061.420 1102.758 1141.355 1159.605 1217.655 1346.474 1473.442 1590.248 1699.283 1799.086

1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

57096.0 56410.0 56810.0 48041.0 47176.0 46222.0 45250.0 44560.0 44295.0 44439.0 44043.0 43127.0 42146.0 42072.0 41392.0 41064.0 41495.0 39918.0 38286.1 37150.1 44594.0 50368.9 55162.0 59365.2 62583.2 65209.3

58.19 55.07 57.58 52.64 49.40 49.13 48.86 47.91 45.35 45.51 45.14 44.00 44.12 40.21 37.51 36.65 39.09 36.20 33.54 32.04 36.57 36.91 36.33 35.59 34.49 33.36

Oil production (million Reserves/ barrels) production 981.222 1024.341 986.697 912.639 954.990 940.844 926.165 930.024 976.683 976.379 975.787 980.075 955.205 1046.394 1103.395 1120.550 1061.420 1102.758 1141.355 1159.605 1217.655 1346.947 1478.492 1603.274 1722.337 1832.985

Figure 8.2. Oil simulation 1: real oil price of $21.49 –$21.67, tax rate  61%.

58.19 55.07 57.58 52.64 49.40 49.13 48.86 47.91 45.35 45.51 45.14 44.00 44.12 40.21 37.51 36.65 39.09 36.20 33.54 32.04 36.62 37.39 37.31 37.03 36.34 35.58

08-A3495 7/27/05 7:14 AM Page 98

98

Chapter Eight

Figure 8.3. Oil simulation 4: real oil price of $25.00, tax rate  50.2%.

of high oil prices and a lower tax rate understandably promotes much larger reserve growth than the stringent financial setting of Simulation 1. Larger reserve growth leads to slightly higher R /P ratios than Simulation 1.

Simulating Gas Reserves and Production Based on projections of successful drilling and gross gas reserves added, one can project alternative time paths of gas reserves and production. Again we concentrate on simulations 1 and 4 because their results bracket those of simulations 2 and 3. Simulation 1: Real Oil Price of $21.49 –$21.67 and 61 Percent Tax Rate As table 8.3 and figure 8.4 illustrate, gas reserves increase sharply, from 52,678 BCF in 2002 to 81,506 BCF in 2008. Again because of rapid reserve growth, production accelerates from 1,673 BCF in 2003 to 2,052 BCF in 2006. Even so, simulated production in 2006 is slightly below the pemex target (2,190 BCF) but far below the Energy Secretary’s target (2,810 BCF). The simulated R /P ratio remains very stable in the range 37.92 (2004) to 36.00 (2008). Simulation 4: $25.00 Real Oil Price and 50.2 Percent Tax Rate Driven by 328 more successful wells than in Simulation 1, projected gas reserves increase to 88,124 BCF by 2008 (see table 8.3 and figure 8.5). Projected production reaches 2,299 BCF in 2008, slightly exceeding pemex’s production target for 2006, but falls short of the Energy Secretary’s target. The simulated R /P ratio remains stable from 2003 through 2008.

08-A3495 7/27/05 7:15 AM Page 99

Table 8.3. Simulating Gas Reserves and Production Simulation 1: Real oil prices of $21.49 –$21.67; tax rate  61 percent

Year

Total gas reserves (BCF)

Gas production Reserves/ (BCF) production

1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

86487.3 86128.0 79109.7 123291.0 122533.0 100375.3 81226.4 79738.6 79025.5 77621.9 77144.6 75488.3 74460.8 69778.3 69272.9 60648.9 59924.6 59864.6 54676.1 52678.4 61944.8 68426.9 73249.3 77116.1 79688.3 81506.0

1479.579 1373.493 1315.401 1252.364 1276.943 1230.615 1171.571 1332.789 1326.244 1311.639 1305.396 1323.032 1372.400 1535.370 1630.455 1749.080 1748.715 1713.246 1646.515 1614.395 1673.393 1804.315 1933.356 2052.068 2162.883 2264.316

58.45 62.71 60.14 98.45 95.96 81.57 69.33 59.83 59.59 59.18 59.10 57.06 54.26 45.45 42.49 34.67 34.27 34.94 33.21 32.63 37.02 37.92 37.89 37.58 36.84 36.00

Simulation 4: Real oil price of $25.00; tax rate  50.2 percent

Year

Total gas reserves (BCF)

Gas production Reserves/ (BCF) production

1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

86487.3 86128.0 79109.7 123291.0 122533.0 100375.3 81226.4 79738.6 79025.5 77621.9 77144.6 75488.3 74460.8 69778.3 69272.9 60648.9 59924.6 59864.6 54676.1 52678.4 62026.4 69283.7 75325.5 80648.6 84748.6 88123.8

1479.579 1373.493 1315.401 1252.364 1276.943 1230.615 1171.571 1332.789 1326.244 1311.639 1305.396 1323.032 1372.400 1535.370 1630.455 1749.080 1748.715 1713.246 1646.515 1614.395 1673.393 1804.795 1938.487 2065.307 2186.314 2298.768

Figure 8.4. Gas simulation 1: real oil price of $21.49 –$21.67, tax rate  61%.

58.45 62.71 60.14 98.45 95.96 81.57 69.33 59.83 59.59 59.18 59.10 57.06 54.26 45.45 42.49 34.67 34.27 34.94 33.21 32.63 37.07 38.39 38.86 39.05 38.76 38.34

08-A3495 7/27/05 7:15 AM Page 100

100

Chapter Eight

Figure 8.5. Gas simulation 4: real oil price $25.00, tax rate  50.2%.

In summary, even in the most favorable financial setting of higher real oil prices and a 50.2 percent tax rate, it seems unlikely that gas production will reach the Energy Secretary’s ambitious target set for 2006.6 A reassuring outcome is that, by 2008, gas reserves could recover to the historical reserve levels of 1983 and 1984.

Fiscal Consequences of Permanent Tax-rate Cuts In chapter 3 we stress that taxes collected from pemex account for one-third of total federal taxes. Oil and gas production taxes constitute 60 percent of all pemex taxes. If all other taxes remained the same, a permanent cut in production tax rates would prompt an immediate reduction in federal tax collections. If oil and gas prices remained constant, and so did oil and gas production, the lower tax rates would cause a permanent decline in federal taxes. However, following a tax cut from 61 to 50.2 percent, our simulations show that, with constant oil prices, larger cash flows realized by pemex would stimulate more successful drilling, more rapid reserve growth, and greater oil and gas production. In short, the tax-rate reduction would produce a substantially larger production tax base. From the viewpoint of fiscal balance, we raise two questions of fundamental importance: How large would the short-term loss in tax revenues be? And how many years would it take before tax revenues recovered to their levels before the tax-rate cut? If the loss in pemex tax revenue were to last several years, a large cut in pemex tax rates would be politically unwise and economically unsound. If the loss in tax revenue were brief and subsequently offset by even

08-A3495 7/27/05 7:15 AM Page 101

Simulating the Integrated Oil and Gas Supply Model

101

Table 8.4. Projected Oil and Gas Production Taxes, Real Oil Price of $25/Barrel, Real Gas Price of $4/ Thousand Cubic Feet, and pemex Tax Rate of 50.2 Percent, 2003 –2008

Year 2002 2003 2004 2005 2006 2007 2008

Oil Oil sales production Gas revenue (million production (million barrels/ (BCF/ dollars/ year) year) year) 1159.6 1217.7 1346.9 1478.5 1603.3 1722.3 1833.0

1614.4 1673.4 1804.8 1938.5 2065.3 2186.3 2298.8

28990.1 30441.4 33673.7 36962.3 40081.9 43058.4 45824.6

Gas sales revenue (million dollars/ year) 6457.6 6894.4 7435.8 7986.6 8509.1 9007.6 9470.9

pemex tax rate

Oil tax revenue (million dollars/ year)

Gas tax revenue (million dollars/ year)

Total tax revenue (million dollars/ year)

0.610 0.502 0.502 0.502 0.502 0.502 0.502

17684.0 15281.6 16904.2 18555.1 20121.1 21615.3 23004.0

3939.1 3461.0 3732.7 4009.3 4271.6 4521.8 4754.4

21623.1 18742.5 20636.9 22564.3 24392.6 26137.2 27758.4

larger pemex tax collections, however, the tax cut could represent fiscally sound economics. The integrated oil and gas model is well suited for analyzing these two questions. In doing so, it is essential to hold real oil and gas prices constant in order to isolate the consequences of the tax cut alone. We first assume a hypothetical reduction in pemex’s tax rate from 61 percent, as it was in 2002, to 50.2 percent from 2003 to 2008. Second, we assume constant real prices of oil and gas of $25 per barrel and $4 per thousand cubic feet. Results are shown in table 8.4. We start by showing actual oil and gas production in 2002, the pemex sales revenue at prices of $25 per barrel and $4 per thousand cubic feet, and the federal tax revenue at a tax rate of 61 percent. In 2003, the production tax rate is lowered permanently to 50.2 percent, oil and gas prices remain constant at $25 per barrel and $4 per thousand cubic feet, and production begins to increase. As the last column shows, total production taxes decrease from $21,623.1 million in 2002 to $18,742.5 million in 2003, a oneyear loss of some $2,881 million. By 2004, increases in oil and gas output lead to higher total production taxes of $20,636.9 million. In 2005, total production taxes ($22,564 million) are higher than those in 2002. The economic reason is straightforward: Higher production more than offsets the temporary loss in production taxes attributable to the lower tax rate. After 2005, steady growth in oil and gas production stimulates continuing growth in production taxes. Are these simulated increases in production feasible? We believe they are. Oil and natural-gas liquids production was 3.80 million barrels per day (1,387 million per year) in April, 2004 (World Oil 2004, 77). If output were to increase by 7 percent annually, then production of oil and natural-gas liquids would be 5.0 million daily barrels (or 1,818 million barrels per year) in 2008. For comparison, production of 1,818 million barrels in 2008 differs from our simulated production of 1,833 million barrels by only 1 percent. Dry gas and processing plant liquids amounted to 1,614 BCF per year in 2002. If production increased by 7 percent each year, it would reach 2,422

08-A3495 7/27/05 7:15 AM Page 102

102

Chapter Eight

BCF in 2008. Our simulated production of 2,299 BCF seems to be attainable. Transporting and burning such a large volume will require a massive investment in pipelines. Whether the necessary pipeline investment will be achieved remains a crucial question. The major result is clear: A large cut in the production tax rate from 61 to 50.2 percent causes a slight reduction in tax revenue for two years, but larger tax revenue thereafter. Such a tax-rate cut could be viewed as a prospective investment in medium-term sustainable development.7 This simulated outcome of a hypothetical tax cut cannot be viewed as a fiscal panacea. After all, the results depend on estimated coefficients (with sometimes large standard errors) of the vertically integrated model. A next step, requiring too much time for us to discuss here, is to produce a series of simulations using a range of plausible values for each parameter in the model. Doing so seems to us to be the only rational way of beginning to understand the possible outcomes of fiscal reforms.

conclusions Here we employ the estimated structural coefficients to simulate drilling, reserves, and production. Combined with a cut in tax rates, higher oil prices prompt the largest response in drilling, reserve growth, and production. Although no restrictions are placed on R /P ratios for either oil or gas, the simulated ratios remain quite stable. Either higher oil and gas prices or lower tax rates produce larger cash flows for pemex. Larger cash flows stimulate more successful drilling, which in turn promotes growth in reserves and production. Our most important conjecture is that a permanent reduction in tax rates, from 61 to 50.2 percent, could cause production tax revenues to decrease for only two years, then to increase for an indefinite future. Fiscal reform along these lines could be viewed as a mediumterm investment in sustainable energy.

09-A3495 7/27/05 7:15 AM Page 103

chapter nine

Conclusions

The United Nations World Summit on Sustainable Development in Johannesburg, South Africa (August 26 –September 6, 2002), reaffirmed two principles of sustainability: to eradicate poverty and to protect the environment.1 This book is chiefly devoted to Mexico’s oil and natural gas as two foundations for sustainable development. Although we focus on oil and gas, sustainable development rests on much broader foundations: 1. The first is long-run growth in living standards spread across all segments of the population. Building this foundation will require broader access to high-quality public education that vanquishes illiteracy and enables young citizens to read, write, and compute—the essential requirements of an effective labor force.2 Improvement in living standards will also require balanced long-term increases in physical capital and energy per worker. 2. A second foundation is to provide electricity to more people, particularly in rural areas. To do so will require unprecedented investment in infrastructure: new electricity-generating plants fired by natural gas and a wide network of pipelines to deliver it. pemex’s pipeline network now extends only about six thousand miles, which is woefully inadequate to meet the demand for new gas-fired electricity plants. Broadening the availability of electricity and increasing the range of pipelines to deliver the fuel go hand in hand. 3. A third foundation, obviously complementary to the first two, is to increase the production of energy. As chapter 1 points out, energy in Mexico means oil and natural gas. It is certain that for at least the next twenty years nuclear electricity and renewable resources will be able to provide little, if any, additional energy. Mexico has only two operating nuclear-power plants and currently has no plans to add others. Renewable resources now account for 7 percent of total energy and can add only a pittance to Mexico’s overall energy requirements for at least three or four decades. It is unrealistic to imagine that anything other than oil and natural gas will be able to supply Mexico’s energy for the foreseeable future. 4. A fourth foundation is to reduce air pollution locally and globally. Mexico City suffers the worst air pollution in the world because of its intense concentrations of nitrogen oxides, sulfur oxides, carbon monoxide, and carbon dioxide. The Fox administration has proposed stringent air-quality goals in urban areas. In 2002, Congress also ratified the Kyoto Protocol, 103

09-A3495 7/27/05 7:15 AM Page 104

104

Chapter Nine

thereby committing Mexico to reducing future emissions of carbon dioxide. However, since 93 percent of its fossil-fuel carbon dioxide emissions are attributable to the combustion of oil and natural gas, this commitment will be difficult—and probably impossible—to fulfill as oil and gas consumption increase. Oil now accounts for 71 percent, natural gas 22 percent, and coal 7 percent of Mexico’s carbon dioxide emissions from fossil fuels.3 Oil and gas are constitutionally synonymous with pemex. This monolith is both servant and benefactor of the government. As chapter 3 explains, pemex depends on Congress for approval and allocation of its annual budgets but is also responsible for one-third of all federal tax revenues. Unlike private corporations, whose taxes are based on profits, pemex’s taxes are based on sales revenue. Private borrowing made possible by the constitutional amendment creating pidiregas has proven to be financially infeasible for pemex. For this reason, we believe that reduction in pemex’s production tax rates to perhaps 50 percent is a viable measure that would provide sufficient funds for future investments. Exploration and development were hampered for almost twenty years by tightly constrained drilling budgets. From 1982 to 2000, drilling collapsed (chapter 3), as did oil and gas reserves (chapter 4). Only in 2001, 2002, and 2003 did drilling recover to the rates of the late 1970s. This short-term revival stemmed a twenty-year decline in reserves. The turnaround, however, was purchased at a steep price: pemex financed much of its recent drilling by negotiating private loans that now threaten its financial viability.

the current oil and gas situation The evidence is irrefutable that Mexico possesses abundant oil and gas reserves. Average production from Mexico’s 3,065 oil wells producing in 2002 was 1,170 barrels per day, as compared to average production of 11.22 barrels per day from 518,805 oil wells in the United States.4 Mexico’s oil reserves are not only large but also flowing vigorously: Of its 3,065 producing wells, 2,800 (91 percent) flow naturally. In stark contrast, only 16,640 of the 518,805 wells in the United States (3 percent) flow naturally; 97 percent require some type of artificial lift.5 Eighty percent of Mexico’s gas is produced from oil wells. Because oil has reigned supreme, nonassociated gas resources have scarcely been tapped. Even so, pemex estimates that gas reserves on December 31, 2002, were 52,678 BCF. As nonassociated gas is developed in the years to come, reserves will increase substantially. Sustainable macroeconomic growth will be possible only if oil and gas resources are developed further. Such development requires drilling budgets in the range of $14 –$15 billion, as they were in 2002 and 2003. Such large budgets are fiscally feasible only if they are financed by Congress instead of by private debt. Prospects for developing oil and gas are strong because recent success

09-A3495 7/27/05 7:15 AM Page 105

Conclusions

105

rates are so high—about 50 percent in exploration and nearly 90 percent in development—and because reserves per successful well are so large. Our research covers the years from 1975 through 2000, when oil received top priority. As nonassociated gas receives more attention, the payoff will be substantial. In the first place, Mexico will be able to produce all of the gas it consumes instead of importing it (which drains currency). However, if the country’s gas consumption is to grow by as much as 7 or 8 percent annually, unprecedented pipeline investment must also occur. Second, the substitution of gas for oil with higher sulfur and carbon content will create environmental benefits. A few numbers illustrate this point. Combustion of oil products generates 173 pounds of carbon dioxide per million BTUs of energy, but an equivalent amount of energy from natural gas produces only 117 pounds of CO2. The substitution of natural gas for oil of equivalent energy would reduce carbon dioxide emissions by about 32 percent. Since Mexico accounts for only 1.5 percent of the world’s carbon emissions, the global reduction achieved by this substitution would be marginal. In fact, Mexico’s ratification of the Kyoto Protocol is a political act with no global significance. Mexico’s carbon emissions have not been and will not be a source of global warming.

Successful Drilling In chapter 5 we analyze success ratios. Ratios in exploration and development display notable increases from 1975 to 2000. Recent exploration success is in the range of 45 –50 percent, while development success is roughly 90 percent. Success ratios are a measure of drilling efficiency. Gains achieved in exploration efficiency are noteworthy, and development efficiency is approaching its upper limit. We find that performance has improved because drilling contractors have adopted newer technologies and have benefited from cumulative experience.

New Reserves Successful drilling is the key to new reserves. In chapter 6 we analyze oil and gas reserves per successful well. They are huge: Oil reserves appear to be roughly 18 million barrels, and gas reserves about 25 BCF per successful well. Such large reserves are essential to support such high production per well—an average of 1,170 barrels per day in 2002.

Production Chapter 7 centers on oil and gas production. Reserves are of course the lifeblood of production anywhere. But in Mexico, production is also influenced by government mandates designed to achieve economic and political goals. One of the most important objectives is to produce enough oil both to meet domestic needs and to enable larger exports. Oil exports are vital because they are Mexico’s largest “cash crop.” Accordingly, we specify an oil production equation in which output depends on reserves and annual increases that we know to have been federally mandated.

09-A3495 7/27/05 7:15 AM Page 106

106

Chapter Nine

We initially find the oil production and reserves series to be nonstationary; thus we proceed to estimate the equation using first differences in the series. Changes in production are determined by changes in reserves and by government edicts. Thus far, about 80 percent of all gas has been produced from oil wells. Therefore, it only makes sense to estimate gas as a coproduct of oil. Not surprisingly, gas and oil are closely linked: About one BCF of gas is produced with each million barrels of oil. As nonassociated gas reserves are developed, it will be advisable to model nonassociated gas production independently of associated gas and oil. Regrettably, there is now insufficient information available to permit such an independent analysis. Nevertheless, this sort of statistical work will be of primary importance as the required data become available. Our simple oil and gas equations produce accurate forecasts. We first estimate the equations using subsamples for the 1975 –1998 time period, then forecast production in 1999 and 2000. The root mean squared errors and percentage errors of the forecasts are quite small. We then estimate the equations using subsamples for 1975 –1999 and forecast production in 2000, 2001, and 2002. Again the forecast errors are reassuringly small.

A Note on Multiple-service Contracts A new device known as multiple-service contracts (MSCs) was recently introduced to enable private companies to develop nonassociated gas fields and thereby increase production. Contracts between pemex and the companies can cover services such as seismic imaging, geological reservoir modeling, field and production engineering, drilling, and gas-well maintenance and transportation. The private companies are fully responsible for all of the contracted services, but by law the natural gas belongs to the citizens of Mexico. In July, 2003, pemex held open bidding for MSC services in seven blocks in the Burgos Basin. By February, 2005, it had awarded five contracts. Companies that received contracts include Reposol-YPF, an integrated energy company in Spain; a consortium of Petrolras, Teikoku Oil in Japan, and Grupo Diavaz in Mexico; a partnership of Argentina-based Tecpetrol and Mexico’s Industrial Perforadora de Campeche; and U.S.-based Lewis Energy. pemex announced that these MSC investments would total roughly $4.5 billion. The new opportunities afforded by these private service contracts should surely enable pemex to increase production. How the payments for private services affect pemex’s already fragile balance sheet remains to be seen.

Tax-policy Simulations In chapter 8, using four sets of assumptions concerning oil prices and pemex tax rates, we simulate the integrated model. Simulation 1 relies on a relatively low oil price ($21.49 –$21.67 per barrel) and a tax rate of 61 percent from 2003 through 2008. Simulation 2 is based on the same oil price but a permanently lower tax rate of 50.2 percent, which provides an 18 percent greater annual cash

09-A3495 7/27/05 7:15 AM Page 107

Conclusions

107

flow. Simulation 3 incorporates a higher oil price of $25 per barrel with the 61 percent tax rate. Finally, Simulation 4 combines the $25 oil price with the 50.2 percent tax rate from 2003 through 2008, providing an approximately 34 percent greater annual cash flow than Simulation 1. All of the simulations show that reserves and production increase steadily beginning in 2003. They indicate that oil production targets established by pemex (1,460 barrels in 2006) and the Secretary of Energy (1,414 barrels in 2006) will be fulfilled. However, all four simulations show that gas production will fall short of those announced by pemex (2,190 BCF in 2006) and the Secretary of Energy (2,810 BCF in 2006). It will be possible to reach these lofty gas production targets only by a major effort to develop nonassociated gas reserves. In order to deliver the substantially greater gas volumes to intended end users, pemex must simultaneously invest in pipeline infrastructure.

Aggregate Productivity From 1979 to 2000, Mexico experienced no growth in real GDP per worker and average living standards. This stagnation was prompted by financial events that included a series of major devaluations in 1977, 1982, 1983, 1985 –1987, and 1995. These devaluations led to tighter federal budgets, which inevitably caused shrinkage in pemex’s investments in exploration. These financial disturbances caused disruptions throughout the economy. The devaluations increased the cost of imports, notably new machinery and equipment. Higher costs of machinery and equipment combined with periodic recessions to depress the rate of capital formation. Stringent pemex budgets were responsible for a drilling recession that lasted for twenty years. This recession, in turn, caused the country’s oil and gas reserves to decline after 1985. Mexico’s economy is not inherently stagnant. To the contrary, it has enormous potential for growth. We are certain of this possibility based on hard evidence: Aggregate labor productivity grew by 3.7 percent annually between 1965 and 1979. Such impressive productivity growth is attributable to balanced growth in capital and energy per worker coupled with technological progress. To regain productivity growth requires a successful commitment to long-term expansion in capital and energy per worker.

Policies for Sustainable Development Achieving long-term productivity growth and other goals of sustainable development will require a combination of economic, energy, and social policies. First, it is vital that governments concentrate resources to improve the reach and the quality of public education. To be effective, the labor force must be literate and healthy. Without a literate, well-nourished labor force, sustainable development cannot possibly be attained. The second is to maintain a relatively stable exchange rate, thus preserving the purchasing power of the peso, for only by maintaining purchasing power

09-A3495 7/27/05 7:15 AM Page 108

108

Chapter Nine

can Mexico continue to import technologically advanced machinery and equipment. In time, the country will develop its own production capabilities, and such capital will be produced at home. Third, Mexico must maintain low rates of inflation consistent with those of its major trading partners. If Mexico’s inflation rates are persistently higher than those of the countries with which it trades, peso devaluation will be inevitable. Measured by the percentage of change in the consumer-price index, inflation averaged roughly 4.5 percent in 2003. The Bank of Mexico has established an inflation target of 3 percent for 2004. These rates are only slightly higher than those in the United States. Fourth, Mexico must continue to develop its abundant oil reserves. Larger reserves are the lifeblood of greater oil production. Greater production is essential for increasing both domestic energy consumption and exports. After all, petroleum products now account for 63 percent of commercial energy consumption, and the country exports nearly half of the oil it produces. Finally, it is essential to concentrate on developing nonassociated gas. Gas reserves are the wellspring of production; growth in gas production and distribution is essential for the expansion of electricity. An increase in gas production and consumption is fundamental to enabling Mexico to improve urban air quality and to achieve the other goals of sustainable growth. Mexico and the United States have long been joined culturally and economically. Hispanics, mostly with family roots in Mexico, will soon outnumber Anglos as the largest ethnic group in Texas. Local cultures in south and west Texas more closely resemble Mexico than they do cultures in northern and central Texas; Spanish is the primary language spoken throughout the state’s region stretching south and west from San Antonio. Mexico is an open economy in the important sense that, in 2003, it exported $165 billion of its $629 billion GDP. Put differently, exports accounted for 26.2 percent of its GDP. Nearly 80 percent of its exports were sold to the United States. The two countries have always been integrated economically. Even in the late 1980s, before the Mexico-U.S. ratification of the North American Free Trade Agreement (NAFTA), almost three-fourths of Mexico’s merchandise exports were sold to the United States.6 Similarly, about three-fourths of Mexico’s pre-NAFTA merchandise imports originated in the United States. The Mexico-U.S. free-trade agreement only strengthened the preexisting economic ties. U.S. firms now import a vast range of consumer durables, automobile components and after-market parts, textiles, and foodstuffs from Mexico. Oil, however, is the commodity with the strongest link. In 2003, Mexico exported 1.75 million barrels of oil daily, of which 1.6 million were shipped to the United States. Oil imported from Mexico accounted for about 16 percent of total U.S. oil imports. The fates of these two major North American economies are inseparable. The destinies of both hinge in large measure on the economically and environmentally effective use of fossil fuels.

10-A3495-EPI 7/27/05 7:15 AM Page 109

Epilogue

Several changes affecting the energy sector of Mexico and the country’s prospects for sustainable development have occurred since we wrote the foregoing chapters analyzing the period 1975 –2000. The following is a brief discussion of some of the most important developments, together with their implications for the future.

crude oil prices and exchange rates The average price of crude oil exports increased to $30.22 per barrel during the first nine months of 2004, substantially higher than the range of prices reported in the preceding chapters. This price increase, by itself, boosted pemex’s export revenue by about 18 percent and reduced the rate of depreciation of the peso. For example, the Mexican currency depreciated against the U.S. dollar from December 31, 2002, to September 30, 2003, by 6.8 percent. But during the period of higher oil export prices in 2004, the peso depreciated by only 1.4 percent. A prolonged period of oil prices substantially above $30 per barrel would continue to increase pemex’s export earnings and thereby help strengthen the peso. Because world oil prices are highly unstable and beyond the control of pemex, it is impossible to know how long the currently favorable prices will continue.

drilling Immediately after the drilling depression analyzed in chapter 4, pemex sharply accelerated its drilling investments. pemex drilled only 234 wells in 1999 and 247 in 2000. Then it abruptly increased drilling to 449 wells in 2001, 457 in 2002, and 631 in 2003. pemex reports that it drilled 527 wells during the first nine months of 2004. The Secretary of Energy stated in private correspondence dated January 28, 2005, that pemex planned to drill 712 wells during the full year 2004. The central point is that the average number of wells pemex drilled during the years 2001 to 2004 is more than twice the average number drilled during 1999 and 2000. This astonishing increase in drilling has surely enabled the development of valuable new oil and gas reserves. Our main concern is that the four-year drilling spree has been purchased at too steep a price; much of this drilling was financed by additional debt incurred through pidiregas. We believe that pemex’s growing debt burden is not sustainable and may soon cause major financial problems.

10-A3495-EPI 7/27/05 7:15 AM Page 110

110

Epilogue

investment financing pemex relied on additional loans through pidiregas to finance 84 percent of its long-term investments in 2003, and 92 percent in 2004. In 2005, pemex projects total capital investments of $11.2 billion, of which 85 percent ($9.5 billion) is allocated to exploration and production. pemex reports that 88 percent of the $11.2 billion ($9.9 billion) will be financed by pidiregas. Although these investments can be underwritten by additional loans through pidiregas, we are deeply skeptical that such loans will be financially feasible. Indeed, they may soon provoke a financial crisis. Chapter 3 showed that loans through pidiregas prompted noteworthy deterioration in pemex’s balance sheets at year-end 2001 and 2002. Since 2002, the company’s financial problems have deepened. During the year September 30, 2003, to September 30, 2004, pemex’s total assets grew by 12 percent from 856.7 billion to 960.3 billion pesos; however, total liabilities increased by 23 percent from 761.5 to 936.6 billion. Accordingly, pemex’s total equity plummeted by 75 percent from 95.2 to 23.7 billion pesos. By comparing the company’s total equity of 122.9 billion pesos as of December 31, 2001, with its total equity of only 23.7 billion as of September 30, 2004, the degree of pemex’s financial impairment becomes clear. If the growth in total liabilities continues to outstrip the growth in total assets at this pace, pemex could report negative equity in the near future. Because pemex is a nationalized company, there is no market for its equity. And if its balance sheet were to report negative equity, it would not suffer the financial catastrophe that would befall a private firm. At a minimum, however, we believe that the increasing debt burden poses a serious threat to the company’s ability to meet future interest payments and repayments of loan principal. If the growing debt attributable to pidiregas becomes unserviceable, it will prompt a notable decrease in pemex’s international credit ratings, and a corresponding increase in investment costs.

oil production and exports During the last five years of our historical analysis, 1996 to 2000, crude oil production averaged 2.98 million barrels per day and exhibited no growth (see chapter 5 data appendix, table 5). But by the first nine months of 2004, production had increased to 3.395 million daily barrels. This recent growth signifies that pemex may be able to achieve its production target of 4.0 million barrels per day by 2006. Attaining this target would be a noteworthy achievement permitting both growth in exports and domestic consumption. pemex reports that during the period January 1, 2004, to September 30, 2004, crude oil exports averaged 1.838 million barrels daily. This level differs marginally from average daily exports of 1.843 million barrels in 2003, but compares favorably with the averages of 1.705 in 2002 and 1.757 in 2001. Oil production and exports may be slowly approaching sustainable growth paths.

10-A3495-EPI 7/27/05 7:15 AM Page 111

Epilogue

111

However, the rapid growth in pemex’s long-term debt casts serious doubt on the company’s capacity to increase long-run production with existing financial arrangements.

natural gas production and imports The near-term outlook for natural gas production is bleak. During the years 1996 to 2000, total (associated plus nonassociated) gas production averaged 4.59 BCF per day. In the first nine months of 2004, pemex reports that total gas production remained only 4.57 BCF per day. This actual production rate in 2004 is far short of the Energy Secretary’s ambitious target of 2810 BCF annually (equivalent to 7.70 BCF per day) to be reached in 2006. We now believe it will be impossible for pemex to increase gas production by 2006 to meets its own target of 2,190 BCF annually (6.0 BCF daily), much less the Energy Secretary’s even loftier production target. Natural gas imports are expanding rapidly. Net imports, equal to gross exports minus gross imports, averaged 208 million cubic feet per day in 2000. They increased to 267 million in 2001, 588 million in 2002, 757 million in 2003, and 758 million in the first 10 months of 2004. Recognizing that imports will continue to increase for at least several years, the government is constructing liquefied natural gas (LNG) import terminals to augment the import capacity of existing and planned pipelines. pemex is now increasing priority given to the development and production of nonassociated gas. During most years before 2000, nonassociated gas accounted for scarcely 20 percent of total gas production, but by the first nine months of 2004, it had increased to 34 percent of overall gas production. The growing emphasis pemex now places on nonassociated gas reserves and production is a valuable step toward sustainable development.

pemex taxes, debt, and future finances In the first nine months of 2004, pemex paid taxes and duties equal to 61.25 percent of total sales. An economically plausible case can be made for a fiscal change that reduces the tax rate to approximately 50 percent. Because the Fox Administration faces majority opposition in both the Chamber of Deputies and the Chamber of Senators, we seriously doubt that such a tax reduction is politically feasible during the tenure of the current administration. Substantive tax reduction will be a thorny task for future administrations as well. Sustainable development requires, among other things, substantial growth in hydrocarbon reserves and production. The foundation of such growth must be continuing high levels of annual investment in successful drilling. Noteworthy increases in drilling throughout the years 2001 to 2004 are a platform for hope, but we believe such hope rests on financial foundations that may soon buckle. There are two related reasons. The first is pemex’s continuing reliance on rapidly-growing long-term debt to finance capital expenditures. Second, this accumulating debt now threatens to negate its total equity. If its balance sheet

10-A3495-EPI 7/27/05 7:15 AM Page 112

112

Epilogue

deteriorates to a point of negative equity, pemex will encounter lower credit ratings and higher capital costs. Although these changes need not be fatal, they would be serious burdens. Far-reaching changes in pemex’s finances are urgent and two-fold. The first is to cut the growth of long-term debt. The second is to substantially increase after-tax cash flows that will be required for growing investments in reserves, production, refining, and infrastructure for many years to come.

11-A3495-APP 7/27/05 7:15 AM Page 113

Appendix

chapter 2. data appendix Table 1. Average Products of Capital and Labor

Year

Average Product of Capital; GDP(t) /K (t) (1993 pesos)

Average Product of Labor; GDP(t) /L(t) (1993 pesos)

1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982

0.365116 0.377929 0.387043 0.401517 0.407701 0.414432 0.409602 0.420783 0.431160 0.431671 0.429440 0.423760 0.419149 0.430573 0.442436 0.446386 0.445213 0.419813

30421.7547 32097.9846 33657.0419 35911.4171 37674.8011 39746.9166 39978.3480 42168.8171 43375.4276 45378.6737 45891.4825 47055.3347 46612.9789 48643.1086 50597.0570 47767.5460 48531.0736 48416.8834

113

Year

Average Product of Capital; GDP(t) /K (t) (1993 pesos)

Average Product of Labor; GDP(t) /L(t) (1993 pesos)

1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

0.387647 0.390816 0.389376 0.366696 0.365892 0.362130 0.368146 0.375125 0.377350 0.375698 0.369798 0.371652 0.343500 0.353167 0.364886 0.369250 0.367763 0.375100

46929.5838 47549.6060 47818.4008 46695.5318 47076.0461 43293.1511 43846.4937 43950.3209 44492.6175 45366.6817 45733.9393 46569.3113 45004.9131 45779.3940 47086.2871 47375.0872 47952.1236 49958.6614

11-A3495-APP 7/27/05 7:15 AM Page 114

114

Appendix

Table 2. Investment in Machinery and Equipment in Mexico

Year

Gross Investment Millions of 1993 Pesos

Domestic Investment Millions of 1993 Pesos

Imported Investment Millions of 1993 Pesos

1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

19767.1 19984.5 21883.1 24837.3 25731.4 29127.9 27953.9 32146.9 37869.1 42451.3 47672.8 45971.9 41072.1 48452.6 62978.7 74176.3 88522.1 68399.1 49694.2 55394.1 64140.8 55818.3 53816.3 64019.9 71192.2 85365.2 101634.5 117953.6 107960.9 118395.7 75347.4 92714.2 124290.6 145152.3 160138.9 185641.4

13428.8 13831.6 13873.1 15593.4 16388.6 18388.1 19197.1 21596.8 25548.9 28742.6 30294.7 30518.9 29888.2 35465.4 40962.5 45161.2 50810.8 44037.8 34340.2 38107.3 43914.8 37049.3 37064.6 41115.8 45635.1 50314.1 57939.7 60329.9 53984.1 52994.5 33858.2 40782.9 53332.9 60985.9 61382.6 68745.8

6338.3 6152.9 8010.0 9243.9 9342.8 10739.8 8756.8 10550.1 12320.2 13708.7 17378.1 15453.0 11183.9 12987.2 22016.2 29015.1 37711.3 24361.3 15354.0 17286.8 20226.0 18409.0 16751.7 22904.1 25557.1 35051.1 43694.8 57623.7 53976.8 65401.2 41489.2 51931.3 70967.7 84166.4 98756.3 116895.6

Public Private Domestic Imported Sector Sector Investment Investment Investment Investment % % % % 0.68 0.69 0.63 0.63 0.64 0.63 0.69 0.67 0.67 0.68 0.64 0.66 0.73 0.73 0.65 0.61 0.57 0.64 0.69 0.69 0.68 0.67 0.69 0.64 0.64 0.59 0.57 0.51 0.50 0.45 0.45 0.44 0.43 0.42 0.38 0.37

0.32 0.31 0.37 0.37 0.36 0.37 0.31 0.33 0.33 0.32 0.36 0.34 0.27 0.27 0.35 0.39 0.43 0.36 0.31 0.31 0.32 0.33 0.31 0.36 0.36 0.41 0.43 0.49 0.50 0.55 0.55 0.56 0.57 0.58 0.62 0.63

25.7 20.4 22.3 29.2 29.0 36.6 35.7 36.5 35.7 31.9 29.2 31.4 28.1 23.3 23.3 21.8 21.8 16.5 11.1 13.9 11.6 10.2 7.5 8.0 7.1 8.4 7.1 5.5 4.8 4.4 5.1

74.3 79.6 77.7 70.8 71.0 63.4 64.3 63.5 64.3 68.1 70.8 68.6 71.9 76.7 76.7 78.2 78.2 83.5 88.9 86.1 88.4 89.8 92.5 92.0 92.9 91.6 92.9 94.5 95.2 95.6 94.9

GDP (millions of 1993 Pesos)

365,986.82 391,355.25 415892.93 449728.44 478168.58 511264.59 532591.55 577797.13 626384.55 664661.43 701956.12 731710.45 756901.55 819344.52 894535.58 968800.59 1045800.52 1040131.48 985278.92 1021498.30 1049907.93

Year

1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985

1386.557 1496.617 1534.660 1658.342 1853.361 1875.957 1943.607 2162.122 2391.188 2524.659 2646.656 2819.295 3020.649 3363.406 3722.180 4169.065 4466.908 4812.635 4555.715 4619.645 4730.193

Total Energy Consumed (Petajoules) 1002385.831 1035524.912 1074540.488 1120073.335 1172841.205 1233651.162 1300266.868 1373147.351 1452790.562 1539739.596 1634583.713 1726709.674 1805804.377 1902915.239 2021430.789 2170319.963 2348987.990 2477607.772 2541689.122 2613757.872 2696384.413

Total Gross Capital Stocks (Millions of 1993 Pesos) 819212.962 835443.790 855961.902 881260.497 911896.521 948499.596 988444.113 1031953.163 1079262.987 1130629.668 1186327.461 1243948.377 1297825.545 1361564.107 1436276.292 1522743.402 1623928.519 1713616.374 1771755.347 1832392.241 1895300.663

Gross Capital Stocks for Structures (Millions of 1993 Pesos) 183172.868 200081.122 218578.587 238812.838 260944.685 285151.566 311822.756 341194.189 373527.575 409109.928 448256.252 482761.298 507978.832 541351.131 585154.498 647576.561 725059.471 763991.398 769933.775 781365.631 801083.750

Gross Capital Stock for Machinery and Equipment (Millions of 1993 Pesos) 0.769 0.815 0.805 0.784 0.817 0.792 0.777 0.784 0.822 0.822 0.813 0.638 0.638 0.828 0.839 0.886 0.888 0.772 0.638 0.811 0.825

12030431 12192518 12356788 12523272 12692000 12863000 13322000 13702000 14441000 14647000 15296000 15550000 16238000 16844000 17676000 20281565 21549091 21482826 20994836 21482792 21956149

L: Annual Capacity Average Utilization Number of Rate Remunerated % Employees

Table 3. Aggregate Real GDP, Energy Consumption, Capital Stocks, Capacity Utilizations Rates, and Labor Input

2.74 2.65 2.58 2.49 2.45 2.41 2.44 2.38 2.32 2.32 2.33 2.36 2.39 2.32 2.26 2.24 2.25 2.38 2.58 2.56 2.57

Capital  Output Ratio  K /GDP

2.24 2.13 2.06 1.96 1.91 1.86 1.86 1.79 1.72 1.70 1.69 1.70 1.71 1.66 1.61 1.57 1.55 1.65 1.80 1.79 1.81

(continued)

0.50 0.51 0.53 0.53 0.55 0.56 0.59 0.59 0.60 0.62 0.64 0.66 0.67 0.66 0.65 0.67 0.69 0.73 0.78 0.76 0.76

Capital Capital Machinery Structure and  Output Equipment Ratio   Output K Ratio  K structures/ machineGDP equi /GDP

11-A3495-APP 7/27/05 7:15 AM Page 115

GDP (millions of 1993 Pesos)

1010495.23 1029247.46 1042066.10 1085815.10 1140847.53 1189016.97 1232162.34 1256195.97 1311661.12 1230771.05 1294196.56 1381839.20 1451350.91 1503930.03 1603750.82

Year

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Table 3. (Continued )

4606.426 4824.919 4898.641 5175.098 5161.286 5343.640 5418.635 5407.048 5642.490 5487.190 5778.805 5979.923 6116.716 6236.137 6368.410

Total Energy Consumed (Petajoules) 2755678.420 2812981.556 2877604.360 2949418.188 3041243.736 3150963.247 3279665.690 3396981.040 3529270.268 3583028.437 3664542.516 3787045.035 3930533.233 4089394.806 4275529.173

Total Gross Capital Stocks (Millions of 1993 Pesos) 1945824.966 1996953.188 2045557.272 2095122.371 2151834.942 2212237.642 2278205.188 2346468.509 2422156.379 2465674.503 2520201.248 2585725.432 2654734.253 2728533.491 2809073.919

Gross Capital Stocks for Structures (Millions of 1993 Pesos) 809853.454 816028.368 832047.088 854295.816 889408.794 938725.606 1001460.502 1050512.531 1107113.889 1117353.934 1144341.268 1201319.604 1275798.980 1360861.315 1466455.254

Gross Capital Stock for Machinery and Equipment (Millions of 1993 Pesos) 0.773 0.773 0.773 0.779 0.779 0.780 0.763 0.747 0.740 0.638 0.676 0.725 0.749 0.765 0.828

21640084 21863507 24069999 24764012 25957661 26723916 27160072 27467478 28165783 27347482 28270286 29346956 30635319 31363158 32101557

L: Annual Capacity Average Utilization Number of Rate Remunerated % Employees

2.73 2.73 2.76 2.72 2.67 2.65 2.66 2.70 2.69 2.91 2.83 2.74 2.71 2.72 2.67

Capital  Output Ratio  K /GDP

1.93 1.94 1.96 1.93 1.89 1.86 1.85 1.87 1.85 2.00 1.95 1.87 1.83 1.81 1.75

0.80 0.79 0.80 0.79 0.78 0.79 0.81 0.84 0.84 0.91 0.88 0.87 0.88 0.90 0.91

Capital Capital Machinery Structure and  Output Equipment Ratio   Output K Ratio  K structures/ machineGDP equi /GDP

11-A3495-APP 7/27/05 7:15 AM Page 116

11-A3495-APP 7/27/05 7:15 AM Page 117

Appendix

117

Table 4. GDP and Total Gross Investment Flow

Year

GDP (Millions of 1993 Pesos)

Gross Investment Flow in Machinery and Equipment (Millions of 1993 Pesos)

1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

365986.82 391355.25 415892.93 449728.44 478168.58 511264.59 532591.55 577797.13 626384.55 664661.43 701956.12 731710.45 756901.55 819344.52 894353.58 968800.59 1045800.52 1040131.48 985278.92 1021498.30 1049907.93 1010495.23 1029247.46 1042066.10 1085815.10 1140847.53 1189016.97 1232162.34 1256195.97 1311661.12 1230771.05 1294196.56 1381839.20 1451350.91 1503930.03 1603750.82

19761.1 19984.5 21883.1 24837.3 25731.4 29127.9 27953.9 32146.9 37869.1 42451.3 47672.8 45971.9 41072.1 48452.6 62978.7 74176.3 88522.1 68399.1 49694.2 55394.1 64140.8 55818.3 53816.3 64019.9 71192.2 85365.2 101634.5 117953.6 107960.9 118395.7 75347.4 92714.2 124290.6 145152.3 160138.9 185641.4

Gross Investment Flow in Structures (Millions of 1993 Pesos)

Total Gross Investment Flow (Millions of 1993 Pesos)

37208.9 42380.9 47720.9 51204.6 55915.4 58543.5 57508.3 64409.3 72911.3 77140.2 82925.7 86192.9 83976.1 90081.1 101845.7 114576.4 130376.5 122488.2 99990.4 103900.1 107640.4 97899.4 99759.5 98527.9 100704.1 109090.6 114198.6 121273.5 125218.5 134349.6 104072.1 116168.6 128529.2 133651.9 140167.6 148753.8

56976.0 62365.4 69604.0 76041.9 81646.8 87671.4 85462.2 96556.2 110780.4 119591.5 130598.5 132164.8 125048.2 138533.7 164824.4 188752.7 218898.6 190887.3 149684.6 159294.2 171781.2 153717.7 153575.8 162547.8 171896.3 194455.8 215833.1 239227.1 233179.4 252745.3 179419.5 208882.8 252819.8 278804.2 300306.5 334395.2

11-A3495-APP 7/27/05 7:15 AM Page 118

118

Appendix

Table 5. Real GDP per Worker, Energy per Worker, and Utilized Capital per Worker

Year

GDP/L (1993 Pesos per Worker)

E /L Energy Consumption /L (Gigajoules Consumed per Worker)

1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

30421.7547 32097.9846 33657.0419 35911.4171 37674.8011 39746.9166 39978.3480 42168.8171 43375.4276 45378.6737 45891.4825 47055.3347 46612.9789 48643.1086 50597.0570 47767.5460 48531.0736 48416.8834 46929.5838 47549.6060 47818.4008 46695.5318 47076.0461 43293.1511 43846.4937 43950.3209 44492.6175 45366.6817 45733.9393 46569.3113 45004.9131 45779.3940 47086.2871 47375.0872 47952.1236 49958.6614

115.2541 122.7488 124.1957 132.4208 146.0259 145.8413 145.8945 157.7961 165.5833 172.3670 173.0293 181.3051 186.0235 199.6798 210.5782 205.5593 207.2899 224.0224 216.9922 215.0393 215.4382 212.8654 220.6837 203.5165 208.9766 198.8348 199.9572 199.5074 196.8527 200.3314 200.6440 204.4127 203.7664 199.6622 198.8364 198.3832

UK /L: Utilized Total Stocks of Capital /L (1993 Pesos per Worker) 64073.74 69218.91 70002.42 10700.30 75497.26 75958.31 75837.51 78568.64 82694.68 86411.28 86886.54 70845.07 70951.05 93541.55 95948.20 94787.51 96778.71 89040.03 77237.93 98672.35 101316.36 98434.90 99454.98 92413.31 92779.67 91268.97 91968.23 92100.80 92344.88 92683.48 83589.86 87607.80 93493.55 96108.42 99784.98 110270.58

11-A3495-APP 7/27/05 7:15 AM Page 119

Appendix

119

chapter 4. data appendix Table 1. Exploration and Development Wells Drilled

Year 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Total Exploration Successful Wells Exploration Drilled Wells 87 79 79 83 83 83 63 65 61 58 69 68 27 27 27 43 51 41 25 16 10 10 10 21 22 37

13 25 30 80 30 35 24 18 17 14 19 22 8 8 8 14 25 24 13 6 6 6 7 13 9 21

Total Development Wells Drilled

Successful Development Wells

Total Wells Drilled

Total Successful Wells Drilled

266 257 228 223 250 349 342 288 244 224 219 178 76 76 76 63 133 88 53 47 91 104 111 182 212 210

212 200 176 173 203 291 297 237 214 190 171 142 63 63 63 53 116 83 47 42 88 97 106 178 193 191

353 336 307 306 333 432 405 353 305 282 288 246 103 103 103 106 184 129 78 63 101 114 121 203 234 247

225 225 206 253 233 326 321 255 231 204 190 164 71 71 71 67 141 107 60 48 94 103 113 191 202 212

11-A3495-APP 7/27/05 7:15 AM Page 120

120

Appendix Table 2. Economic and Tax Rate Variables

Year

Mexican Oil Price ($/Barrel)

pemex Tax Rate (Percent of Total Sales)

Net Mexican Oil Price  NMOP (Net of Taxes) ($/Barrel)

1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

1.59 2.08 2.48 11.56 11.16 11.16 12.57 13.39 13.21 19.60 31.19 33.19 28.69 26.42 26.82 25.35 11.87 16.04 12.30 15.61 19.09 14.58 14.88 13.20 13.88 15.70 18.94 16.46 10.17 15.62 24.62

11.8 12.1 12.0 12.3 11.7 20.0 21.3 25.9 30.0 37.5 50.2 51.5 52.7 53.9 45.5 51.5 43.4 54.6 57.7 60.7 63.8 73.1 66.3 64.0 62.1 61.6 62.6 67.3 64.3 65.6 62.7

1.40 1.83 2.18 10.14 9.85 8.93 9.89 9.92 9.25 12.25 15.53 16.10 13.57 12.18 14.62 12.29 6.72 7.28 5.20 6.13 6.91 3.92 5.01 4.75 5.26 6.03 7.08 5.38 3.63 5.37 9.18

11-A3495-APP 7/27/05 7:15 AM Page 121

Appendix

121

chapter 5. data appendix Table 1. Exploration Success Ratios in Mexico: OSRE

Year

Total Exploration Wells

Successful Wells

Overall Exploration Success Ratio (Successful / Total)

1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

87 79 79 83 83 83 63 65 61 58 69 68 27 27 27 43 51 41 25 16 10 10 10 21 22 37

13 25 30 80 30 35 24 18 17 14 19 22 8 8 8 14 25 24 13 6 6 6 7 13 9 21

0.149 0.316 0.380 0.964 0.361 0.422 0.381 0.277 0.279 0.241 0.275 0.324 0.296 0.296 0.296 0.326 0.490 0.585 0.520 0.375 0.600 0.600 0.700 0.619 0.409 0.568

11-A3495-APP 7/27/05 7:15 AM Page 122

122

Appendix Table 2. Development Success Ratios in Mexico: OSRD

Year

Total Development Wells

Successful Development Wells

Overall Development Success Ratio (Successful / Total)

1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

266 257 228 223 250 349 342 288 244 224 219 178 76 76 76 63 133 88 53 47 91 104 111 182 212 210

212 200 176 173 203 291 297 237 214 190 171 142 63 63 63 53 116 83 47 42 88 97 106 178 193 191

0.797 0.778 0.772 0.776 0.812 0.834 0.868 0.823 0.877 0.848 0.781 0.798 0.829 0.829 0.829 0.841 0.872 0.943 0.887 0.894 0.967 0.933 0.955 0.978 0.910 0.910

Table 3. Total Wells Drilled Onshore and Offshore, 1990 –2000 Onshore Exploration Year # Wells 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

35 42 29 13 6 4 5 7 14 20 36

Offshore

Development

Total

%

# Wells

%

# Wells

46.7 29.4 33.0 59.1 25.0 6.8 5.7 7.6 8.9 9.9 15.4

40 101 59 9 1855 83 85 144 182 198 191

53.3 70.6 67.0 40.9 75.0 93.2 94.3 92.4 91.1 90.1 84.6

75 143 88 22 24 59 88 92 158 202 234

Exploration Development % # Wells % 100 100 100 100 100 100 100 100 100 100 100

8 9 12 12 10 6 5 3 7 2 1

25.8 22.0 29.3 21.4 25.6 14.3 19.2 10.3 15.6 6.3 7.7

Total

# Wells

%

# Wells

%

23 32 29 44 29 36 21 26 38 30 12

74.2 78.0 70.7 78.6 74.4 85.7 80.8 89.7 84.4 93.8 92.3

31 41 41 56 39 42 26 29 45 32 13

100 100 100 100 100 100 100 100 100 100 100

Million Barrels

275.323 281.613 275.745 264.110 249.454 248.054 266.045 264.223 260.208 244.586 229.885

Year

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Onshore

29.6 28.8 28.2 27.1 25.5 26.0 25.4 23.9 23.2 23.0 20.8

% 654.701 695.070 700.634 711.677 730.621 707.151 780.349 839.172 860.342 816.834 872.873

Million Barrels

Offshore

OIL

70.4 71.2 71.8 72.9 74.5 74.0 74.6 76.1 76.8 77.0 79.2

% 930.024 976.683 976.379 975.787 980.075 955.205 1046.394 1103.395 1120.550 1061.420 1102.758

Million Barrels

TOTAL

Table 4. Oil and Gas Production Onshore and Offshore, 1990 –2000

100 100 100 100 100 100 100 100 100 100 100

% 924.719 902.844 881.955 850.971 834.297 869.211 963.458 1028.789 1133.726 1175.774 1143.384

BCF

Onshore

69.4 68.1 67.2 65.2 63.1 63.3 62.8 63.1 64.8 67.2 66.7

%

408.070 423.400 429.684 454.425 488.735 503.189 571.912 601.666 615.354 572.941 569.862

BCF

Offshore

GAS

30.6 31.9 32.8 34.8 36.9 36.7 37.2 36.9 35.2 32.8 33.3

%

1332.789 1326.244 1311.639 1305.396 1323.032 1372.400 1535.370 1630.455 1749.080 1748.715 1713.246

BCF

TOTAL

100 100 100 100 100 100 100 100 100 100 100

%

11-A3495-APP 7/27/05 7:15 AM Page 123

11-A3495-APP 7/27/05 7:15 AM Page 124

124

Appendix

chapter 6. data appendix Table 1. Gross Oil and Gas Reserve Additions

Year 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Gross Oil Oil Reserves Reserve Additions (Million (Million Barrels) Barrels) 3953 7278 10428 28407 33560 47224 56998 56998 57096 56410 56810 48041 47176 46222 45250 44560 44295 44439 44043 43127 42146 42072 41392 41064 41495 39918

829.589 3618.117 3508.090 18421.607 5689.926 14372.593 10618.241 1003.084 1079.222 338.341 1386.697 7856.361 89.990 13.156 45.835 240.024 711.683 1120.379 579.787 64.075 25.795 972.394 423.395 792.550 1492.420 474.242

Gas Production (BCF)

Gas Reserves (BCF)

Gross Gas Reserve Additions (BCF)

786.493 771.802 746.870 934.921 1064.596 1298.591 1482.227 1549.961 1479.579 1373.493 1315.401 1252.364 1276.943 1230.615 1171.571 1332.789 1326.244 1311.639 1305.396 1323.032 1372.400 1535.370 1630.455 1749.080 1748.715 1713.246

13390.8 21795.9 31295.8 66179.3 68739.6 72440.0 84275.2 84275.2 86487.3 86128.0 79109.7 123291.0 122533.0 100375.3 81226.4 79738.6 79025.5 77621.9 77144.6 75488.3 74460.8 69778.3 69272.9 60648.9 59924.6 59864.6

1194.793 9176.902 10246.770 35818.421 3624.896 4998.991 13317.427 1549.961 3691.979 1014.193 5702.899 45433.664 518.943 20927.085 17977.329 155.011 613.144 91.961 828.096 333.268 344.900 3147.130 1125.055 6874.920 1024.415 1653.246

11-A3495-APP 7/27/05 7:15 AM Page 125

Appendix

125

Table 2. Explanatory Variables for Gross Oil and Gas Reserve Additions

Year

Gross Oil Reserve Additions (Million Barrels)

Gross Gas Reserve Additions (BCF)

1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

829.589 3618.117 3508.090 18421.607 5689.926 14372.593 10618.241 1003.084 1079.222 338.341 1386.697 7856.361 89.990 13.156 45.835 240.024 711.683 1120.379 579.787 64.075 25.795 972.394 423.395 792.550 1492.420 474.242

1194.793 9176.902 10246.770 35818.421 3624.896 4998.991 13317.427 1549.961 3691.679 1014.193 5702.899 45433.664 518.943 20927.085 17977.329 155.011 613.144 91.961 828.096 333.268 344.900 3147.130 1125.055 6874.920 1024.415 1653.246

Total Oil Oil Gas Gas Successful Dummy Dummy Dummy Dummy Wells ODUM2H ODUM1 GDUM2H GDUM2N Drilled 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0

225 225 206 253 233 326 321 255 231 204 190 164 71 71 71 67 141 107 60 48 94 103 113 191 202 212

Oil Production (Million Barrels)

261.589 293.117 358.090 442.607 536.926 708.593 844.241 1003.084 981.222 1024.341 986.697 912.639 954.990 940.844 926.165 930.024 976.683 976.379 975.787 980.075 955.205 1046.394 1103.395 1120.550 1061.420 1102.758

Year

1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

3953 7278 10428 28407 33560 47224 56998 56998 57096 56410 56810 48041 47176 46222 45250 44560 44295 44439 44043 43127 42146 42072 41392 41064 41495 39918

Oil Reserves End of Year (Million Barrels) 15.11 24.83 29.12 64.18 62.50 66.64 67.51 56.82 58.19 55.07 57.58 52.64 49.40 49.13 48.86 47.91 45.35 45.51 45.14 44.00 44.12 40.21 37.51 36.65 39.09 36.20

Oil Reserves/ Oil Production

Table 1. Oil and Gas Production and Reserves

786.493 771.802 746.870 934.921 1064.596 1298.591 1482.227 1549.579 1479.579 1373.493 1315.401 1252.364 1276.943 1230.615 1171.571 1332.789 1326.244 1311.639 1305.396 1323.032 1372.400 1535.370 1630.455 1749.080 1748.715 1713.246

Natural Gas Production (BCF) 13390.8 21795.9 31295.8 66179.3 68739.6 72440.0 84275.2 84275.2 86487.3 86128.0 79109.7 123291.0 122533.0 100375.3 81226.4 79738.6 79025.5 77621.9 77144.6 75488.3 74460.8 69778.3 69272.9 60648.9 59924.6 59864.6

Natural Gas Reserves End of Year (BCF) 17.03 28.24 41.90 70.79 64.57 55.78 56.86 54.37 85.45 62.71 60.14 98.45 95.96 81.57 69.33 59.83 59.59 59.18 59.10 57.06 54.26 45.45 42.49 34.67 34.27 34.94

Natural Gas Reserves/ Production

chapter 7. data appendix

24.65 12.05 22.17 23.60 21.31 31.97 19.14 18.81 2.18 4.39 3.67 7.51 4.64 1.48 1.56 0.42 5.02 0.03 0.06 0.44 2.54 9.55 5.45 1.55 5.28 3.89

Oil Production 16.78 84.11 43.28 172.41 18.14 40.72 20.70 0.00 0.17 1.20 0.71 15.44 1.80 2.02 2.10 1.52 0.59 0.33 0.89 2.08 2.27 0.18 1.62 0.79 1.05 3.80

Oil Reserves

5.61 1.87 3.23 25.18 13.87 21.98 14.14 4.57 4.54 7.17 4.23 4.79 1.96 3.63 4.80 13.76 0.49 1.10 0.48 1.35 3.73 11.87 6.19 7.28 0.02 2.03

Natural Gas Production

Annual Growth Rate (%)

3.15 62.77 43.59 111.46 3.87 5.38 16.34 0.00 2.62 0.42 8.15 55.85 0.61 18.08 19.08 1.83 0.89 1.78 0.61 2.15 1.36 6.29 0.72 12.45 1.19 0.10

Natural Gas Reserves

11-A3495-APP 7/27/05 7:15 AM Page 126

12-A3495-END 7/27/05 7:15 AM Page 127

Notes

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

16. 17. 18. 19. 20. 21.

22. 23.

24.

Chapter 1 Resources for the Future (1980). U.S. Energy Information Administration (2003a), pp.173 –74. Secretaría de Energía (2002), esp. p. 78ff. Nitrous oxide is also produced by nitrogen fertilization of soil and the decomposition of solid waste from animals. World Bank (2003), p. 235. Ibid. Adjusting real GDP per capita for international differences in purchasing-power parity gives the most reliable estimates of real per-capita incomes across countries. World Bank (2000), p. 233. Ibid. (2003), p. 237. Ibid. (2000), p. 265. Ibid., p. 35. In 2000, pemex reported Mexican reserves to be 40 billion barrels but has recently revised that number downward to 15.7 billion barrels. World Oil (June, 2004), p. 77. One BTU is the quantity of heat required to raise the temperature of one pound of water one degree Fahrenheit at a beginning temperature of 39.2 degrees Fahrenheit. This estimate of 35 billion barrels originally in place, although admittedly inexact, is astonishingly large when compared to the total U.S. reserves of 22 billion barrels in 2002. pemex Exploración y Producción (June 22, 2002). These numbers are consistent with estimated production, consumption, and exports reported by pemex for 2002: Production was 3.6 million barrels per day, consumption was 1.93 million barrels per day, and exports were 1.68 million barrels per day, or 46.7 percent of production. See U.S. Energy Information Administration (2003b). For further details, see ibid. (2003a), pp. 101–104. Ibid., p. 118. Coal is responsible for approximately 34 percent of electricity generated globally. See ibid., p. 137. Ibid., p. 87. The details of this bill are previewed by E. Malkin, p. C4. Nordhaus (1994), Wigley (1995), Intergovernmental Panel on Climate Change (2001). This most recent report by ipcc claims a “high level of scientific understanding” concerning the contributions of carbon dioxide, methane, and nitrous oxide to global warming observed since 1860. See Intergovernmental Panel on Climate Change (2001), p. 37. U.S. Energy Information Administration (2003a), p. 227. In February, 2002, President Bush introduced the Climate Change Initiative to address global warming. He set a target of reducing the greenhouse-gas intensity of the U.S. economy by 18 percent over the next ten years. Greenhouse-gas intensity is the ratio of greenhouse-gas emissions (in carbon dioxide equivalent) per dollar of real GDP. See ibid., p. 163. Jorgenson and Wilcoxen (1992), p. 136.

127

12-A3495-END 7/27/05 7:15 AM Page 128

128

Notes to pages 12 –25

25. Secretaría de Energía (2001a), p. 36. 26. Ibid. (2002), pp. 75 –78. 27. Ibid., esp. p. 78ff., discussing the application of ecological norms 085 and 086.

1.

2.

3. 4.

5. 6. 7. 8. 9.

10.

11.

12. 13.

14.

1.

Chapter 2 The weighted average price is a volume-weighted average of the prices of Maya-22 (heavy, high-sulfur oil), Isthmus-34 (light oil), and Olmeca (the lightest, highest-quality crude oil produced in Mexico). Of these crude oils, Maya accounts for about 87 percent of exports, and Olmeca about 12 percent. Energy price increases may also cause the composition of the GDP to shift from more energy-intensive to less energy-intensive sectors (Marlay 1984). Nonetheless, such shifts cannot possibly explain the post-1979 stagnation in Mexico’s labor productivity because stagnation pervaded virtually all industries (Bichara 1990). Sato (1970) demonstrates other striking anomalies that are obtained by aggregating microproduction functions with heterogeneous capital stocks. Technical progress can occur even if the net investment is zero. If older machines are replaced by newer ones having identical real costs of production but embodying newer technology, the constant real-cost capital stock is unchanged, but the productivity-augmented capital stock is larger. When x  0, J *x 1t2  KE 1t2. Bichara (1990) estimates Cobb-Douglas functions for Mexican industries and finds constant returns to scale to be an acceptable approximation. Moroney (1973) reviews in detail the various models and properties of the alternative estimators. The GDP for 1965 –1978 had 1960 prices as the base year and for 1979 –1985 had 1970 prices as the base year. We rebased the entire series by changing the base year to 1993. We are grateful to Germán Alarco Tosoni, director of Energy Yearbooks and Balances, and to Jorge Nuño Lara, subdirector of Energy Consumption, both at the Secretary of Energy, for their help with the energy-input series. Energy from all sources is converted to the common BTU measure of heat content. During our sample period, fossil fuels (petroleum, natural gas, natural gas liquids, and coal) account for 92 –95 percent of primary energy consumption. Having only two nuclear plants, Mexico generated small amounts of its electricity from hydropower and smaller amounts from wind, geothermal, and other renewables. Banco de México (1978). Since this series includes only private industry and commerce, we had to add the government and agricultural sector. The adjustment was made considering other surveys and publications from inegi (1997b, 2001d). inegi (1990a, 1997d, 2001f ). We also estimated several modifications of equation (2.9) by adding a time-trend regressor that allows for disembodied technical change. The time trend was never statistically significant, and its estimated coefficient was often negative, which is inconsistent with any reasonable economic theory. An oecd review of educational attainment and literacy scales in Mexico shows that more than 75 percent of the population between the ages 25 and 64 had acquired less than a highschool education in 2001, and the literacy scores for youths 15 years of age in Mexico trail those of young people of the same age in all other oecd countries. The review concludes that “Mexico’s human capital lags vis-à-vis the rest of the oecd.” See oecd (2003), pp. 56 –57. Chapter 3 pemex does not have independent authority to issue bonds or to borrow money from Mexican or international banks. pemex can initiate a debt proposal to Congress, but to actually issue debt requires approval by Congress and the Secretary of Finance.

12-A3495-END 7/27/05 7:15 AM Page 129

Notes to pages 27– 45 2.

3. 4. 5. 6.

7. 8.

9. 10.

11.

12. 13. 14. 15.

129

Francisco Gil Díaz, director of the Mexican Secretary of Finance, strongly supports national fiscal reform. His proposal has been debated in Congress since President Vicente Fox took office in 2000, but thus far without resolution. We are indebted to Germán Alarco Tosoni from the Secretary of Energy for his clear explanation of the political channels in this process. This document is called Presupuesto de ingresos y egresos de la federación, and it is the principal source of debate in Congress every year. The Secretary of Finance keeps a complete specification of each investment project submitted. In 2002, pemex succeeded in obtaining a total investment budget of $14.7 billion, with $10.5 billion earmarked for exploration and production. If $10.5 billion were invested annually for several years, it would amount to about three-fourths of real investment in 1981. pemex had no autonomy in its financing decisions. The company could present proposals, but its financing was always subject to the approval of Congress. This interpretation was confirmed by Raúl Muñoz Leos, pemex general director, during his participation at the Summit of APEC (Asia-Pacific Economic Cooperation) in Los Cabos, Baja California, México, Oct. 27, 2002. He stated, “We are investing twice the amount invested during the last 12 years in exploratory activities and production. We have delays in gas, petrochemicals, refineries, infrastructure and technology.” IPD Latin America, Inc. (2002). El Norte (2003c). This financing scheme cannot be used indefinitely because of the risk that debt might grow to levels that could threaten pemex with bankruptcy. The federal government has recognized that the pidiregas scheme is no longer viable for investments in the energy sector (Alexander’s Gas and Oil Connections 2001). Of the 12 pidiregas projects, 11 were direct investment projects and 1 was a conditional investment project. Among these projects are Burgos, Cantarell, and the Strategic Gas Program. The Cantarell investment project centers on injecting nitrogen into wells in order to increase both reservoir pressure and the average production per well, which had decreased from about 35,000 barrels per day in 1979 to about 1,400 barrels per day in 2001. Nitrogen injection substantially increased overall Cantarell production, as chapter 1 points out. See table 3.7 (pemex Consolidated Income Statements, 2000 and 2001). Secretaría de Energía (2001a). Raúl Muñoz Leos, pemex general director, released new 2006 oil and gas production targets in 2003. Secretaría de Energía (2002).

Chapter 4 pemex (2001c). See also chapter 5, where exploration and development success ratios are analyzed from 1975 to 2000. 2. Linear models are excellent approximations when the variables change by small magnitudes. However, linearity is a good approximation even when the variables change substantially, as they do in the Mexican case, if the true model structure is roughly linear. 3. The initial assumptions are E(et )  0 for all t, E(et et1)  0 (zero autocorrelation), and E1e2t 2  s2 (homoskedasticity). Further, if et is normally distributed, then the least squares estimators of b1 . . . b3 are normal. 4. Nerlove and Wallis (1966) show that the Durbin-Watson test statistic is biased toward a value of two (indicating zero autocorrelation) when a lagged value of the dependent variable is included as a regressor. In this instance, the Breusch-Godfrey test is an appropriate (large-sample) test. 5. Moroney (1997) obtained an estimate dˆ  0.508 for exploration drilling in Texas. 1.

12-A3495-END 7/27/05 7:15 AM Page 130

130

Notes to pages 45 – 80

If dˆ  0.612, then (1  dˆ )  0.388, or 38.8 percent of the adjustment occurs in one year. By repeated substitution, one finds 1  dˆ 4  0.977, or 98 percent of complete adjustment, is achieved after four years. 7. Moroney (1997) obtained an estimate dˆ  0.560 for development drilling in Texas. 8. If dˆ  0.553, then (1  dˆ )  0.447, or 45 percent of the adjustment, occurs in one year. By repeated substitution, one finds 1  dˆ 4  0.906; thus, 90.6 percent of total adjustment is

6.

achieved after four years. Chapter 5 None

1.

Chapter 6 pemex (1987, 1990, 1996a, 1999a, 2000a, 2001a).

2.

Secretaría de Energía (2001b).

3.

The reserve estimates that satisfy the sec criteria were released by pemex on Sept. 9 in Bulletin no. 259 (pemex 2002b).

4.

We are indebted to Celina Torres, advisor for the pemex Direction of Finance (asesora de la Dirección Corporativa de Finanzas de pemex), and her assistant, Lourdes Valdéz, for helpful discussions concerning the two sets of reserve estimates. We are also indebted to Robert A. Wattenbarger and Richard A. Startzman, Department of Petroleum Engineering, Texas A&M University, for exceedingly valuable discussions.

5.

The long-run R/P ratio in Texas shows less variation: It decreased from 16.5 in 1960 to 7.4 in 1978, then increased to 10.9 in 1994. See Moroney (1997, 182).

6.

Viewing the R/P ratio as an index of long-run capacity to produce in relation to current production, long-run capacity is much larger in Mexico than in Texas.

7.

Secretaría de Energía (2001b).

8.

The R/P ratio for associated gas in Texas was much lower, decreasing from 13.0 in 1975 to 6.5 in 1994. The nonassociated gas R/P was also lower, decreasing from 7.8 in 1975 to 7.0 in 1994. See Moroney (1997, 184 – 86).

9.

pemex (2002b).

10.

We are indebted to José Luis Pérez from pemex Exploración y Producción (pep), who provided careful explanations of these events.

11.

This result can be compared with the onshore results obtained by Moroney (1997). The largest oil-reserve additions in Texas occurred in 1970 (1,335 million barrels), while Mexico added approximately 3,500 million barrels each year from 1976 to 1977; 5,690 million barrels in 1979; 10,618 million barrels in 1981; and 1,492 million barrels in 1999.

1.

Chapter 7 Considering pemex’s limited resources for hydrocarbon reserve development, pemex has incentives to concentrate investments in its highest-return investment, which is crude oil. Natural gas has never been a priority for pemex; that is why Mexico has limited pipeline infrastructure for natural gas. Drilling that is targeted to develop nonassociated gas reserves has thus far been woefully inadequate to meet Mexico’s future gas requirements.

2.

Pindyck (1978) states correctly that declining reserves usually cause higher extraction costs as physical output per unit of capital equipment declines and that secondary and tertiary recovery techniques are eventually needed. At an aggregate level, reserve depletion will be accompanied by higher average extraction costs since lower-cost deposits are usually produced first.

3.

Pindyck (1978) states that the firm “should produce either nothing or at some maximum capacity level,” without stipulating the constraints on its production capacity. In producing

12-A3495-END 7/27/05 7:15 AM Page 131

Notes to pages 81–103

4. 5.

6.

7.

8.

1.

2.

3.

4.

5.

6.

7.

131

oil and gas, the major constraints are reserves and the physical conditions under which they are produced (e.g., porosity, permeability, downhole pressure, water intrusion). According to pemex (2001), the other eight countries in order of proved reserves were Saudi Arabia, Iraq, United Arab Emirates, Kuwait, Iran, Venezuela, Russia, and Libya. In 2000, pemex was the third largest oil company in the world, producing 9.6 percent of the total oil from the world’s fifteen largest companies. pemex ranked seventh among the world’s largest natural-gas firms but produced only 3.9 percent of the total produced by the fifteen biggest corporations. As chapter 3 explains, pemex is subject to congressional control over its investments. Until 2001, the pemex investment budget was tightly constrained. This is the major reason that pemex found it impossible to implement its proposed programs for drilling and production. The prominence of associated gas production in Mexico contrasts sharply with its much smaller role in Texas, where associated gas accounts for 20 –23 percent of the total production. See Moroney (1997, 185). pemex (2001a). We are grateful to Celina Torres, advisor of the Corporate Direction of Finance in pemex, for providing these data for 2002. Chapter 8 The 2006 Secretary of Energy targets appear in the Sectorial Energy Program 2001–2006, published by the Secretary of Energy (2001a). pemex released its production targets for 2006 on Nov. 27, 2001. These lower export prices were calculated by the Mexican Petroleum Institute based on predicted prices of West Texas Intermediate (wti) crude oil made by the U.S. Department of Energy, the Chemical Market Association, Inc. (cmai), and the Petroleum Industry Research Association (pira). pemex allocated 66.2 percent of its overall investment budget in 2001 to exploration and development, of which one-fourth was financed through pidiregas. It allocated 66.7 percent of its overall investment budget in 2002 to exploration and development, of which two-fifths was financed by pidiregas. The simulations were also performed without the pidiregas dummy  1 for 2001–2006. In these simulations, projected oil and gas production were lower in 2006 (1.2 percent lower for oil and 0.9 percent lower for gas). Mexico’s average daily production of oil and natural gas liquids increased from 3.6 million barrels in July, 2002, to 3.80 million in July, 2003 (World Oil 2002b, 15; 2003b, 81). This annualized increase of 5.5 percent, if sustained, would yield an annual production of 1,628 million barrels in 2006, as compared to our simulated production of 1,590 million barrels. The government is aware of these production limitations and the lags in gas-field development. Accordingly, it is preparing for an increase in gas imports in the period 2004 –2006, when real GDP growth is expected to accelerate and require more natural gas, specifically for electrification. Investment is increasing in both pipelines and liquefied natural gas (LNG) terminals that will enable LNG imports. If we had assumed a permanent tax-rate cut in 2003 but instead used lower oil and gas prices, the fiscal outcome would be qualitatively similar to that shown in table 8.4 but with lower tax collections.

Chapter 9 The oecd’s economic survey of Mexico states that, in 2002, one of every five Mexicans was living in acute poverty, defined as having income insufficient to cover basic food needs (Organization for Economic Cooperation and Development 2003, 32). 2. Ibid., p. 11. The oecd’s economic survey of November, 2003, concludes that “Mexico’s human capital lags far behind most other oecd countries. The quality of education services

1.

12-A3495-END 7/27/05 7:15 AM Page 132

132

Notes to pages 104 –108

and training is well below oecd best practice, many school-leavers have poor literacy and numeric skills, and the cost effectiveness of education programmes needs to be improved.” Mexico spends only about 3.2 percent of GDP on education, as compared with more than 6 percent in the United States, Canada, France, Germany, and Italy. 3. Relatively minor amounts of carbon dioxide are also created by the production of cement. 4. World Oil (2003a), p. 25. 5. Ibid. 6. See Hufbauer and Schott (1992), p. 48.

13-A3495-REF 7/27/05 7:15 AM Page 133

References

Adelman, M. A. 1995a. The genie out of the bottle: World oil since 1970. Cambridge: MIT Press. ———. 1995b. Sustainable growth and valuation of mineral reserves. In Sustainable economic growth, ed. J. R. Moroney, 45 – 68. Greenwich, Conn.: JAI Press. Alexander’s Gas and Oil Connections. 2001. pemex invests in pidiregas scheme. Alexander’s Gas and Oil Connections 6(19) (October), http://www.gasandoil.com. Arrow, K. J. 1962. The economic implications of learning by doing. Review of Economic Studies 29 :155 –73. Banamex. 1980. Examen de la situación económica de México: México en cifras, 1970 –1979. México, D.F.: Banco Nacional de México. Banco de México. 1978. Serie encuestas del Banco de México, S.A.: Acervos y formación de capital, cuaderno anual 1960 –1975, tomo 1. México, D.F.: Subdirección de Investigación Económica y Bancaria, Información Estadística de la Oficina de Proyectos Especiales, Proyecto de Acervos y Formación de Capital. ———. 1998. Indicadores económicos, 1942 –1997. México, D.F.: Dirección General de Investigación Económica. ———. 2000. Indicadores del sector externo, 2000. México, D.F.: Dirección General de Investigación Económica. ———. 2002. Indicadores económicos, 1996 –2001. México, D.F.: Dirección General de Investigación Económica. bbva-Bancomer. 2001. Economic report, 1980 –2001. México, D.F.: bbvaBancomer. ———. 2002. Economic report, October, 2002. México, D.F.: bbva-Bancomer. Bichara, E. 1990. Consideraciones sobre la función de producción Cobb-Douglas. Monterrey, N.L.: Universidad Autónoma de Nuevo León, Facultad de Economía. Breusch, T. S. 1978. Testing for autocorrelation in dynamic linear models. Australian Economic Papers 17 :334 –35. Comisión Federal de Electricidad (cfe). 1985. Informe de operación. México, D.F.: Comisión Federal de Electricidad, Dirección General de Energía, Subdirección de Electricidad. ———. 1995. Informe de operación. México, D.F.: Comisión Federal de Electricidad, Dirección General de Energía, Subdirección de Electricidad. ———. 2001. Informe de operación. México, D.F.: Comisión Federal de Electricidad, Dirección General de Energía, Subdirección de Electricidad. Consejo Nacional de Población (conapo). 2001. Proyecciones de población, 2001– 2010. México: conapo. Consultores Económicos Especializados, S.A. de C.V. 2002. Guía económica (October). San Pedro, Garza García, N.L.: México, D.F. Dasgupta, P., and J. E. Stiglitz. 1976. Uncertainty and the rate of extraction under alternative institutional arrangements. Technical report no. 179, Stanford University, Inst. Math. Studies Soc. Sci.

133

13-A3495-REF 7/27/05 7:15 AM Page 134

134

References

El Norte. 2002a. Elogia Fox a pemex : En el mundo reprueba (September 28). México, D.F. ———. 2002b. ¿Es pemex exitosa? (September 29). México, D.F. ———. 2003a. Detalla pemex plan alterno a Cantarell (January 23). Sección de negocios. México, D.F. ———. 2003b. Presenta pemex contratos de servicios múltiples en Houston (February 5). Sección de negocios. México, D.F. ———. 2003c. Paran paraestatales proyectos energéticos vía pidiregas (February 7). Sección de negocios. México, D.F. Food and Agriculture Organization of the United Nations (fao). 1983. Pulp and paper capacities survey 1978 –1983. Rome: United Nations. ———. 1988. Pulp and paper capacities survey 1983 –1988. Rome: United Nations. ———. 1993. Pulp and paper capacities survey 1988 –1993. Rome: United Nations. ———. 1997. Pulp and paper capacities survey 1992 –1997. Rome: United Nations. ———. 2002. Pulp and paper capacities survey 1997–2002. Rome: United Nations. Fudenberg, D., and J. Tirole. 1983. Learning by doing and market performance. Bell Journal of Economics 14 :522 –30. Gilbert, R. 1976a. Search strategies for nonrenewable resource deposits. Technical report no. 196. Stanford University, Inst. Math. Studies Soc. Sci. ———. 1976b. Optimal depletion of an uncertain stock. Technical report no. 207. Stanford University, Inst. Math. Studies Soc. Sci. Heal, G. 1976. The relationship between price and extraction cost for a resource with a backstop technology. Bell Journal of Economics 2:371–78. ———. 1979. Uncertainty and the optimal supply policy for an exhaustible resource. In Advances in the economics of energy and resources, vol. 2, ed. R. S. Pindyck, 119 – 47. Greenwich, Conn.: JAI Press. Hoel, M. 1978. Resource extraction under some alternative market structures. Mathematical Systems in Economics 39. Hotelling, H. 1931. The economics of exhaustible resources. Journal of Political Economy 39(2):137–75. Houthakker, H. S. 1955. The Pareto distribution and the Cobb-Douglas production function in activity analysis. Review of Economic Studies 23:27–31. Hufbauer, G. C., and J. J. Schott. 1992. North American free trade: Issues and recommendations. Washington, D.C: Institute for International Economics. Instituto Nacional de Estadística Geografía e Informática (inegi). 1983. Sistema de cuentas nacionales de México: Principales variables macroeconómicas, periodo 1970 –1983. México, D.F. ———. 1985a. Encuesta anual de la industria de la construcción. México, D.F. ———. 1985b. Sistema de cuentas nacionales de México: Estimación preliminar, 1985. México, D.F. ———. 1986a. Encuesta anual de la industria de la construcción. México, D.F. ———. 1986b. Sistema de cuentas nacionales de México: Estimación preliminar. México, D.F. ———. 1987. Encuesta anual de la industria de la construcción. México, D.F. ———. 1988. Encuesta anual de la industria de la construcción. México, D.F. ———. 1989. Encuesta anual de la industria de la construcción. México, D.F. ———. 1990a. Censos generales de población y vivienda: 1921–1990. México, D.F.

13-A3495-REF 7/27/05 7:15 AM Page 135

References

135

———. 1990b. Encuesta anual de la industria de la construcción. México, D.F. ———. 1992. Encuesta anual de la industria de la construcción. México, D.F. ———. 1993. Encuesta anual de la industria de la construcción. México, D.F. ———. 1994. Encuesta anual de la industria de la construcción. México, D.F. ———. 1995a. El sector energético en México. México, D.F. ———. 1995b. Encuesta anual de la industria de la construcción. México, D.F. ———. 1996a. El sector energético en México. México, D.F. ———. 1996b. Encuesta anual de la industria de la construcción. México, D.F. ———. 1997a. El sector energético en México. México, D.F. ———. 1997b. Encuesta anual de la industria de la construcción. México, D.F. ———. 1997c. Estadísticas de comercio exterior. México, D.F. ———. 1997d. Sistema de cuentas nacionales de México: Cuentas de bienes y servicios, 1988 –1996, tomo 2. México, D.F. ———. 1998a. Anuario estadístico de comercio exterior de los Estados Unidos Mexicanos. México, D.F. ———. 1998b. El sector energético en México. México, D.F. ———. 1998c. Encuesta anual de la industria de la construcción. México, D.F. ———. 1999a. Anuario estadístico de comercio exterior de los Estados Unidos Mexicanos. México, D.F. ———. 1999b. El sector energético en México. México, D.F. ———. 1999c. Encuesta anual de la industria de la construcción. México, D.F. ———. 2000a. El sector energético en México. México, D.F. ———. 2000b. Encuesta anual de la industria de la construcción. México, D.F. ———. 2001a. Anuario estadístico de los Estados Unidos Mexicanos. México, D.F. ———. 2001b. Cuaderno de información oportuna. México, D.F. ———. 2001c. El sector energético en México. México, D.F. ———. 2001d. Encuesta anual de la industria de la construcción. México, D.F. ———. 2001e. Indicadores regionales de coyuntura. México, D.F. ———. 2001f. Sistema de cuentas nacionales de México: Cuentas de bienes y servicios, 1995 –2000, tomo 2. México, D.F. ———. 2002. El sector energético en México, 1996 –2001. México, D.F. Intergovernmental Panel on Climate Change (ipcc). 2001. Climate change 2001: The scientific basis. Contribution of Working Group 1 to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge University Press. International Energy Agency. 2001. Toward a sustainable energy future. Paris: oecd /iea. ipd Latin America, Inc. 2002. pidiregas: Viability or crisis? Third Annual Energy Law Seminar. México, D.F.: amde-itam (October 10). Irwin, D. A., and P. J. Klenow. 1994. Learning by doing: Spillovers in the semiconductor industry. Journal of Political Economy 102:1200 –27. Jarque, C. M., and A. K. Bera. 1981. An efficient large-sample test for normality of observations and regression residuals. Unpublished manuscript, Australian National University, Canberra. Jorgenson, D. W., and P. Wilcoxen. 1992. Reducing U.S. carbon dioxide emissions: The cost of different goals. In Energy, growth, and the environment, ed. John R. Moroney, 125 –58. Greenwich, Conn.: JAI Press.

13-A3495-REF 7/27/05 7:15 AM Page 136

136

References

Lieberman, M. 1984. The learning curve and pricing in the chemical processing industries. Rand Journal of Economics 15:213 –28. Loury, G. C. 1978. The optimal exploitation of an unknown reserve. Review of Economic Studies 45(141):621–36. Malkin, E. 2002. Mexico’s Fox proposes opening power sectors. New York Times (August 12). Marlay, R. C. 1984. Trends in industrial use of energy. Science 226:1277– 83. Marschak, J., and W. H. Andrews. 1944. Random simultaneous equations and the theory of production. Econometrica 12:143 –205. Mexican Chamber of Pulp and Paper. 1970. Memoria estadística, 1960 –1969. México, D.F.: Cámara Nacional de las Industrias de la Celulosa y el Papel. ———. 1997. Memoria estadística, 1970 –1996. México, D.F.: Cámara Nacional de las Industrias de la Celulosa y el Papel. ———. 2001. Memoria estadística, 1997–2000. México, D.F.: Cámara Nacional de las Industrias de la Celulosa y el Papel. Moroney, J. R. 1973. The structure of production in American manufacturing. London: Oxford University Press. ———. 1992. Energy, capital, and technological change in the United States. Resources and Energy 14 :363 – 80. ———. 1997. Exploration, development, and production: Texas oil and gas, 1970 – 1995, vol. 82 of Contemporary studies in economic and financial analysis. Greenwich, Conn.: JAI Press. Mundlak, Y., and I. Hoch. 1965. Consequences of alternative specifications in the estimation of Cobb-Douglas production functions. Econometrica 33:814 –28. Nacional Financiera, S.A. (nafinsa) 1981. La economía mexicana en cifras. México, D.F.: nafinsa. ———. 1995. La economía mexicana en cifras. México, D.F.: nafinsa. Nerlove, M., and K. F. Wallis. 1966. Use of the Durbin-Watson statistic in inappropriate situations. Econometrica 34 :235 –38. Nordhaus, W. D. 1994. Managing the global commons. Cambridge: MIT Press. Organization for Economic Cooperation and Development. 2003. oecd economic surveys: Mexico (November). pemex Exploración y Producción (pep). 2002. Mexico’s gas sector: Gas potential and operating conditions in the north region (June 22). Petróleos Mexicanos (pemex). 1979. Anuario estadístico: Memoria de labores. México, D.F.: pemex. ———. 1980. Anuario estadístico: Memoria de labores. México, D.F.: pemex. ———. 1986. Anuario estadístico: Memoria de labores. México, D.F.: pemex. ———. 1987. Anuario estadístico: Memoria de labores. México, D.F.: pemex. ———. 1990. Anuario estadístico, Memoria de labores. México, D.F.: pemex. ———. 1995. Indicadores petroleros (enero). México, D.F.: pemex. ———. 1996a. Anuario estadístico: Memoria de labores. México, D.F.: pemex. ———. 1996b. Indicadores petroleros (noviembre). México, D.F.: pemex. ———. 1999a. Anuario estadístico: Memoria de labores. México, D.F.: pemex. ———. 1999b. Indicadores petroleros (enero). México, D.F.: pemex. ———. 2000a. Anuario estadístico: Memoria de labores. México, D.F.: pemex. ———. 2000b. Indicadores petroleros (mayo). México, D.F.: pemex.

13-A3495-REF 7/27/05 7:15 AM Page 137

References

137

———. 2001a. Anuario estadístico: Memoria de labores. México, D.F.: pemex. ———. 2001b. Indicadores petroleros (marzo). México, D.F.: pemex. ———. 2001c. Las reservas de hidrocarburos de México: Evaluación al primero de enero del 2002. México, D.F.: pemex Exploración y Producción pep. ———. 2002a. Anuario estadístico: Memoria de labores. México, D.F.: pemex. ———. 2002b. Boletín num. 259/2002. México, D.F.: pemex Gerencia Corporativa de Comunicación Social (septiembre 9). Pindyck, R. S. 1978. The optimal exploration and production of nonrenewable resources. Journal of Political Economy 86(5):841– 61. Resources for the Future. 1980. Trends in energy use in industrial societies. Washington, D.C.: Resources for the Future. Sato, K. 1970. Micro and macro constant-elasticity-of-substitution production functions in a multifirm industry. Journal of Economic Theory 1:438 –53. Secretaría de Energía. 2001a. Programa sectorial de energía 2001–2006. México, D.F. ———. 2001b. Prospectiva del mercado de gas natural, 2001–2010. México, D.F. ———. 2002. Prospectíva del mercado de gas natural, 2002 –2011. México, D.F. Secretaría de Hacienda y Crédito Público (shcp). 1995. Cuenta de la hacienda pública federal, 1995. México, D.F. ———. 1998. Cuenta de la hacienda pública federal, 1998. México, D.F. ———. 2000. Cuenta de la hacienda pública federal, 2000. México, D.F. Secretaría de Programación y Presupuesto. 1979. La industria petrolera en México. México, D.F. Sheshinski, E. 1967. Tests of the learning by doing hypothesis. Review of Economics and Statistics 49 :568 –78. Solow, R. M. 1959. Investment and technical progress. In Mathematical methods in the social sciences, ed. P. Suppes, 89 –105. Stanford: Stanford University Press. ———. 1962. Technical progress, capital formation, and economic growth. American Economic Review Papers and Proceedings 52:76 – 86. ———. 1964. Capital, labor, and income in manufacturing. In The behavior of income shares, vol. 27, Studies in income and wealth, 101–28. Princeton: Princeton University Press for the National Bureau of Economic Research. ———. 1974. The economics of resources or the resources of economics. American Economic Review 64(2):1–14. Uhler, R. S. 1976. Costs and supply in petroleum exploration: The case of Alberta. Canadian Journal of Economics 9:72 –90. ———. 1979. The rate of petroleum exploration and extraction. In Advances in the economics of energy and resources, vol. 12, ed. R. S. Pindyck, 93 –118. Greenwich, Conn.: JAI Press. U.S. Energy Information Administration. 2003a. International energy outlook 2003. Washington, D.C.: U.S. Department of Energy. ———. 2003b. Country analysis: Mexico, February, 2003. Washington, D.C.: U.S. Department of Energy. ———. 2003c. An energy overview of Mexico. Washington, D.C.: U.S. Department of Energy. Wigley, T. M. L. 1995. Past and projected climate change: What have we done and how rapidly can we influence the future? In Final report: An institute on climate change. Boulder: National Center for Atmospheric Research (June).

13-A3495-REF 7/27/05 7:15 AM Page 138

138

References

World Bank. 2000. World development report 1999/2000: Entering the twenty-first century. Washington, D.C.: Oxford University Press. ———. 2002. World development report 2002: Building institutions for markets. Washington, D.C.: Oxford University Press. ———. 2003. World development report 2003: Sustainable development in a dynamic world. Washington, D.C.: World Bank and Oxford University Press. World development report 1999/2000: Entering the twenty-first century. New York: Oxford University Press. World Oil. 2000. World oil. Houston: Gulf Publishing (August). ———. 2002a. World oil. Houston: Gulf Publishing (August). ———. 2002b. World oil. Houston: Gulf Publishing (September). ———. 2003a. World oil. Houston: Gulf Publishing (August). ———. 2003b. World oil. Houston: Gulf Publishing (September). ———. 2004. World oil. Houston: Gulf Publishing (June). Zellner, A. M., J. Kmenta, and J. Dreze. 1966. Specification and estimation of Cobb-Douglas production function models. Econometrica 34:784 –95.

14-A3495-IX 7/27/05 7:15 AM Page 139

Index

Page numbers in italics refer to tables.

development wells, 14, 40 – 41, 119 Dreze, J., 20 drilling budgets, 76, 91, 104 drilling depression, 29 –31 drilling recovery, 14, 41 dry natural gas, 7, 8

Adelman, M. A., 76 aggregate energy input, 21 aggregate labor input, 23 aggregate labor productivity, 15, 16, 23. See also real GDP per worker Alarco Tosoni, Germán, 128n9, 129n3 Andrews, W. H., 20 Arrow, K. J., 56 Asia-Pacific Corporation (APEC), 129n8 associated gas, as coproduct of oil, 106 Banamex, 42 Banco de México, 21, 42, 128n11 Bank of Mexico, 108 Bera, A. K., 44 – 45 Bichara, E., 128n2, 128n6 British thermal unit (BTU), as a measure of energy, 6, 127n12 Campeche, Bay of, 6 Cantarell, 6 –7, 129n11 capital /output ratio, 16 capital stocks, 18, 19, 20 carbon dioxide emissions, 11, 105 Chemical Market Association, Inc. (CMAI), 131n2 Chiapas, natural gas discoveries, 7, 54 coal, 7, 8, 12 coal and electricity production, 10 Comisión Federál de Electricidad, 10 Dasgupta, P., 78 development drilling, 40 – 41; estimation using seemingly unrelated regressions, 50 – 51; forecasts, 48 – 49; lags in adjustment, 43; number of wells, 119; short-run and long-run elasticities, 49; statistical estimates, 46 development success ratio, 14, 54; autoregressive model, 55 –56; distinguished from exploration success ratio, 53; and drilling efficiency, 57; estimation using seemingly unrelated regression, 59 – 61, 121; econometric analysis of, 57– 61; gains from experience, 56; and reasons for lower geological risk, 53; upward trend, 54; used in simulations, 92 –93

ecological norms 085 and 086, 3 education and sustainable development, 103, 107 elasticity, 20 electricity and sustainable development, 3, 103, 107 electricity consumption, 5 electricity production, 9; by coal-fired plants, 10; by fossil fuels, 10; importance of greater accessibility, 103; and investment in natural gas pipelines, 103 –105; by nuclear power, 9 –10; by oil and gas-fired plants, 10; planned expansion of, 11; by renewable resources, 9 –10 El Norte, 129n10 exchange rates, importance of stability, 15, 107–108. See also peso devaluations exploration success ratio, 14, 53; abnormality because of discoveries in Chiapas and Tabasco, 54; autoregressive model, 55 –56; econometric analysis of, 57– 61; estimation using seemingly unrelated regression, 59 – 61, 121; and gains from experience (learning by doing), 58 –59; as a measure of drilling efficiency, 53, 56; as a measure of exploration efficiency, 53; upward trend, 53; used in simulations, 92 –93 exploration wells, 14, 40 – 41, 119 exploratory drilling, 40; estimation using seemingly unrelated regressions, 50 –51; forecasts, 47– 48; lags in adjustment, 43; number of wells, 119; reasons for decline, 40 – 41; short-run and long-run elasticities, 45; statistical estimates, 45 extensions, 64 financial crises and drilling depression, 29 –31 financial dilemma shared by pemex and federal government, 35 –37 financial disturbances and peso devaluations, 15, 29

139

14-A3495-IX 7/27/05 7:15 AM Page 140

140

Index

financial investment in reserves, 31, 35 –36, 104 forecast errors: mean absolute error, 86; mean absolute percentage error, 86; root mean squared error, 87; for oil and gas production, 87– 89 fossil fuels, 6 –9, 10 foundations of sustainable development, 3, 103 Fox, Vicente, 3 Fox administration, 3 –5, 9, 11–12; plans for fuel switching; 11–13; and global warming, 3, 11; and increases in electric generating capacity, 11; and majority opposition in Congress, 95, 111 Fudenberg, D., 56 fuels for sustainable development, 37–38 gains from experience, 56 –57 gas production: nonstationary in levels, 84; stationary in first-differences, 84 gas production model, statistical estimates, 85 gas reserves, 7, 13 –14, 63, 70 –73, 75 –76 Gilbert, R., 78 Gil-Diaz, Francisco, 129n2 gross gas reserve additions (GGRA), estimation, 73, 75 gross oil reserve additions (GORA), estimates of, 72 –74 Heal, G., 78 Hoch, I., 20 Hoel, M., 78 Hotelling, H., 78 Hotelling model, of optimal exploitation of an exhaustible resource, 78 Houthakker, H. S., 18 Hufbauer, G. C., 132n6 human capital, 3, 23, 128n14 hydrocarbon production taxes, 26; as percentages of total taxes, 26; special tax on gasoline production and sales, 26; as percentages of total federal taxes, 27 hydroelectricity, 9 –10, 37

Klenow, P. J., 56 Kmenta, J., 20 Kyoto Protocol, 11; carbon dioxide emissions, 104 –105; impossibility of Mexico fulfilling commitment to, 12, 104; Mexico’s minor role, 11, 105; Mexico’s ratification of, 103 – 104 labor productivity, 16; measured by real GDP per worker, 16; reasons for growth, 16 –17, 22 –23; reasons for stagnation, 15 –17, 22 – 23 learning by doing. See gains from experience Lieberman, M., 56 life expectancy at birth, 5 living standards, 3 – 4; Mexico vs. United States, 5; remained constant, 1980 –2000, 4 Loury, G. C., 78 Malkin, E., 127n20 Marlay, R. C., 128n2 Marschak, J., 20 Mexican Petroleum Institute, 91, 131n2 Mexico: cultural links with the United States, 108; as a developing country, 4; economic integration with the United States, 108; oil exports to the United States, 108; as an open economy, 108; population growth, 4 Mexico City, 3; air pollution, 3 – 4, 103 –104; transportation emissions, 3 Moroney, J. R., 18, 39, 65, 79, 85, 97, 128n7, 129n5, 129n7, 130n5, 130n8, 130n11, 131n7 Moroney model, 39, 40 multiple service contracts, 106 Mundlak, Y., 20 Muñoz-Leos, Raúl, 31, 129n8, 129n14

imported machinery and equipment, 4, 15, 20 income inequality, 5 inegi, 128n11, 128n12 inflation, and sustainable development, 107– 108; and peso devaluation, 108 installed generating capacity, 9 Instituto Mexicano de Contadores Publicos, 33 –36 Intergovernmental Panel on Climate Change (IPCC), 127n21 Irwin, D. A., 56

natural gas bylaws (Article #109), 37 natural gas plant liquids, 6 Nerlove, M., 129n4 net price of Mexican oil exports, 39; calculation of, 42; as a financial determinant of drilling, 39; exogeneous and endogenous components, 39 new pool discoveries, 64 – 65, 72; incomplete information about, 65; as uncertain estimates, 65 nonassociated gas, 106 Nordhaus, W. D., 127n21 North American Free Trade Agreement (NAFTA), 108 nuclear energy: as percentage of electricity production, 9, 37; lack of plans for expansion, 9, 37 Nuño-Lara, Jorge, 128n9

Jarque, C. M., 44 – 45 Jorgenson, D. W., 127n24

oil exports, 3, 5, 8; and cost of domestic energy, 15; exogenous to Mexican economy,

14-A3495-IX 7/27/05 7:15 AM Page 141

Index 15, 81– 82; and foreign exchange, 15; importance of Maya-22, 42; oil prices, 15; recent increases in, 110; as a source of foreign currency, 6, 9, 105; to the United States, 108; weighted average, 81– 82 oil production, 5 – 6, 66 – 67; flowing naturally, 104; as a nonstationary series, 84; offshore and onshore, 39, 123; per well, 104; stationary in first-differences, 84 oil production model, statistical estimates, 84 – 85 oil reserves, 5, 63 – 66, 104 –105 oil-fired and gas-fired electricity production, 10 pemex: budgetary process, 27–29; budgets and federal budget, 25; consolidated balance sheets 2000 –2002, 33; consolidated income statements, 35 –37; as exporter of three types of oil, 81– 82; as a price-taking oil producer, 81– 82; requires congressional approval to issue debt, 25; as a state-owned monopoly, 6; tax burden illustrated numerically, 34 –36; tax rates, 104 Perez, Jose Luis, 130n10 peso devaluations, 15; economic consequences of, 107; inhibited growth in physical capital per worker, 15 –16, 29 –30; and inflation, 108. See also exchange rates petroleum and petroleum products, 8 –9 Petroleum Industry Research Association (PIRA), 131n2 pidiregas, 26; financially infeasible, 104; loans cause sharp increases in pemex debt, 31; as the major source of funds for drilling, 32, 109 –10; as a new source of private loans, 31–32; in simulations, 93, 131; stated by government officials to be unsustainable source of loans, 129n10 Pindyck model, as generalization of earlier models, 78; links production to reserves, 80; modified for pemex production mandates, 77, 84; summarized, 79 – 80 Pindyck, R. S., 78, 89, 130n2, 130n3 political mandates for oil production, 84 poverty in Mexico, 5, 131n1 President Bush and greenhouse gas intensity, 127n22 Presupuesto de ingresos y egresos de la federacion, 129n4 probable reserves, defined, 64; compared with proved reserves, 65 – 66; measurement errors, 64; subject to greater uncertainty than proved reserves, 64, 66 production function, 18; aggregate production function, 18; Cobb-Douglas, 18; CES, 18; estimated coefficients, 21–22; estimation procedure, 18 –20; generalized Leontief, 18; microeconomic, 18; translog, 18; and vintage capital, 18 –19

141

production targets for oil and gas in 2006, 91 productivity growth, 107 proved reserves, defined, 64; compared with probable reserves, 65 – 66; U.S. Security and Exchange Commission criteria, 65 Public Electricity Service Act, 8 purchasing power parity (PPP), 5, 127n6 real GDP per worker, 15 –16; and energy per worker, 16, 118; stagnating 1979 –2000, 15 –16; and utilized capital per worker, 17. See also aggregate labor productivity real gross domestic product (GDP) per capita, 4 –5 real price of exported oil, 91 renewable resources, and electricity production, 9 reported reserves: unknown measurement errors, 64, 66, 76; published by pemex, 65, 66 reserves/production (R /P) ratios: for oil 68 – 69, 71–72; for gas, 70 –72; in simulations, 94 revisions, 64 – 65; as a cause of reserve measurement errors, 65; are inherently subjective, 65; produce erratic year-to-year reserve changes, 65; reasons for 64 – 65 Salinas de Gortari, Carlos, 29 –30 Sato, K., 128n3 Schott, J. J., 132n6 Secretaría de Energía, 127n3, 127n25, 127n26, 127n27, 128n25, 128n26, 128n27, 129n13, 129n15, 130n1, 130n2, 130n7 Secretaría de Programacion y Presupuesto, 42 Sheshinski, E., 56 simulated cash flows, 94 –95 simulated drilling 2003 –2008, 94 simulated pemex tax rates of 61 percent and 50.2 percent, 92, 106 –107 simulated successful wells 2003 –2008, 94 – 95, 95 simulating gas reserves and production, 98 – 100; falls short of production targets, 98 – 100, 131 simulating oil reserves and production, 96 – 98, 97 simulating production taxes, 101 Society of Petroleum Engineers (SPE), 63 Solow, R. M., 18 –20, 78 Solow model, 78 Startzman, Richard A., viii, 130n4 Stiglitz, J. E., 78 structural coefficients in simulations, 92, 102 sustainable development, 3 –5, 23, 37–38, 89 –91, 103 –104, 107–12 Tabasco, large natural gas discoveries in, 7; and exploration success ratios, 54

14-A3495-IX 7/27/05 7:15 AM Page 142

142

Index

Tirole, J., 56 Torres, Celina, 130n4, 131n8

Valdez, Lourdes, 130n4 Veracruz, and new natural gas discoveries, 7

U.S. Energy Information Administration, 6, 27, 76, 127n2, 127n15, 127n16, 127n17, 127n18, 127n19, 127n22, 127n23 U.S. Securities and Exchange Commission (SEC), 65 unemployment, 5, 12; cyclical variation in, 20 United Nations World Summit on sustainable development, 103 unit root tests, 21, 40 – 41

Wallis, K. F., 129n4 Wattenbarger, Robert A., viii, 130n4 wells drilled 2001–2003, 91, 109 West Test Intermediate (WTI), 131n2 Wigley, T. M. L., 127n21 Wilcoxen, P., 127n24 World Petroleum Congress (WPC), 63 Zellner, A. M., 20