Global Poverty: The Role of Economic Freedom, Democracy, and Foreign Aid

Florida State University Libraries Electronic Theses, Treatises and Dissertations

The Graduate School

2011

Global Poverty: The Role of Economic Freedom, Democracy, and Foreign Aid Joseph S. Connors

Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected]

THE FLORIDA STATE UNIVERSITY COLLEGE OF SOCIAL SCIENCES AND PUBLIC POLICY

GLOBAL POVERTY: THE ROLE OF ECONOMIC FREEDOM, DEMOCRACY, AND FOREIGN AID

By JOSEPH S. CONNORS

A Dissertation submitted to the Department of Economics in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Degree Awarded: Summer Semester, 2011

The members of the committee approve the dissertation of Joseph S. Connors defended on June 15, 2011.

_______________________________________ Thomas W. Zuehlke Professor Co-Directing Dissertation _______________________________________ James D. Gwartney Professor Co-Directing Dissertation _______________________________________ Charles J. Barrilleaux University Representative _______________________________________ Bruce L. Benson Committee Member _______________________________________ R. Mark Isaac Committee Member

The Graduate School has verified and approved the above-named committee members.

ii

I dedicate this to Mom and Dad.

iii

ACKNOWLEDGEMENTS To begin, I thank the members of my dissertation committee. Your advice, feedback, and willingness to read 175 pages of my prose are greatly appreciated. I am blessed to have been a part of the Department of Economics at Florida State University. I couldn’t have asked for a better environment to study economics. Many professors have devoted much of their time and effort to me, but before thanking them I wish to single out one member in particular. In my view, Dr. Carol Bullock is the best hire this department ever made. She will be missed. While she only gets to take one retirement she deserves about three or four. I also thank Drs. Jim Gwartney, Bruce Benson, Mark Isaac, Thomas Zuehlke, Randy Holcombe, Milton Marquis, Tim Salmon, Don Schlagenhauf, Anastasia Semykina, Joe Calhoun and Joab Corey. I have also been lucky to be part of the Stavros Center while in graduate school. Harriet Crawford is the best secretary that center has ever had. I appreciate her help and for always laughing whenever I tried to be funny. Shana Case, whose office in the Stavros Center was always next to mine, has become a close friend and an excellent economist. Having started our tenure at FSU together it is appropriate that we finish together. I thank the Economics Department for funding while I was a teaching assistant, Gus A. Stavros for funding as a research assistant, and BB&T for funding while a dissertation fellow. I will never be able to repay the kindness, help, company, meals, and root beer floats given to me by Jim and Amy Gwartney. I thank Jim especially for the many hours spent advising and mentoring me. Any measure of success I achieve is because of him. I also thank my family for their support and my nieces and nephews for their artwork and skype conversations. Mom, Dad, Paul, Amy, Brendan, Dennis, Emily, Sarah, Ciara, Charles, Collin, Cristie, Reagan, Chris, Elizabeth, Henry, and Lindsay have all helped to make my life a whole lot of fun. I thank my dear friends Hilary and Pascal. Your friendship, support, and patience have sustained me through graduate school. I always look forward to spending time with you when I head west. I am lucky to have you both as dear friends. I befriended many wonderful classmates during graduate school. Notable among them is Tonya Elliot. I highly doubt there is a more giving person in this world. She has become a good friend and a fantastic economist. Mr. John Janeski is the best teacher I have ever had. He stands head and shoulders above his peers. He deserves the credit for my success at the undergraduate and graduate level. I thank him for resisting considerable pressure to make his classes and grading easier. I also thank him for expecting and demanding the very best out of me. I will consider myself a successful professor when I become at least half as good as Mr. Janeski. Lastly, I thank the Christian Men’s Nicotine Research Collective whose conversations and camaraderie were nourishment for a weary man’s soul. Our gatherings for beer and fine cigars were the perfect endings for many a rough week.

iv

TABLE OF CONTENTS List of Tables ................................................................................................................................ vii List of Figures ............................................................................................................................... xii Abstract ........................................................................................................................................ xiii 1.

INTRODUCTION ....................................................................................................................1

2.

GLOBAL POVERTY: MEASUREMENT AND REGIONAL ANALYSIS ..........................6 2.1 The World Bank, World Development Indicators Extreme and Moderate Poverty Rates ...........................................................................................................................................7 2.2 Preparation of the World Bank Extreme and Moderate Poverty Rates for Use in Statistical Analysis ..........................................................................................................10 2.2.1 Part 1: Adjusting the World Bank Poverty Rates At 5-Year Intervals and Filling In Several Missing Values.....................................................................................10 2.2.2 Part 2: Regression Analysis to Fill In Missing Poverty Values ............................12 2.3 Sala-i-Martin and Pinkovskiy Poverty Dataset: An Alternative Measure of Poverty .....14 2.4 Analysis of Global Poverty Using the World Bank and Pinkovskiy and Sala-i-Martin Poverty Measures ............................................................................................................15 2.5 Conclusion .......................................................................................................................22

3.

THE RELATIONSHIP BETWEEN ECONOMIC FREEDOM AND GLOBAL POVERTY ................................................................................................................................................24 3.1 Economic Freedom, Growth, and Global Poverty...........................................................24 3.1.1 Economic Freedom and Growth............................................................................25 3.1.2 Economic Freedom and Poverty ...........................................................................27 3.2 Cross-Country Empirical Framework..............................................................................28 3.3 Results..............................................................................................................................32 3.4 Alternate Specifications and Multicollinearity................................................................49 3.5 Conclusion .......................................................................................................................53

4.

DEMOCRATIC POLITICAL INSTITUTIONS AND GLOBAL POVERTY......................55 4.1 Why Do Political Institutions Matter...............................................................................55 4.2 Empirical Framework ......................................................................................................58 4.3 Results..............................................................................................................................61 4.4 Conclusion .......................................................................................................................75

5.

ECONOMIC FREEDOM, FOREIGN AID, AND GLOBAL POVERTY ............................77 5.1 Foreign Aid and Growth..................................................................................................78 5.2 Foreign Aid and Poverty..................................................................................................83 5.3 Empirical Framework ......................................................................................................84 5.4 Results..............................................................................................................................92 5.5 Conclusion .....................................................................................................................113

v

6.

CONCLUSION.....................................................................................................................115 6.1 Summary of Findings ....................................................................................................115 6.2 Implications ...................................................................................................................120 6.3 Future Research .............................................................................................................121

APPENDICES .............................................................................................................................123 A. EXTREME AND MODERATE POVERTY RATES, 1980-2005 ......................................123 B. DESCRIPTION OF VARIABLES .......................................................................................128 B.1 Economic Freedom of the World Index ........................................................................129 B.2 Polity IV ........................................................................................................................133 B.3 Freedom in the World....................................................................................................135 B.4 World Development Indicators .....................................................................................137 B.5 Geographic Variables ....................................................................................................137 C. SUMMARY STATISTICS ...................................................................................................139 D. CHAPTER 3 ALTERNATIVE SPECIFICATIONS TABLES............................................144 E. ANCILLARY TABLES FOR CHAPTER 5.........................................................................163 REFERENCES ............................................................................................................................169 BIOGRAPHICAL SKETCH .......................................................................................................176

vi

LIST OF TABLES 2.1

World Bank mean extreme and moderate poverty rate for Africa, Latin America, Asia, India, and China, 1980-2005................................................................................................20

2.2

Pinkovskiy and Sala-i-Martin extreme and moderate poverty rate for Africa, Latin America, Asia, India, and China, 1970-2005 ......................................................................21

3.1

Correlation between economic and political institution variables, 1980-2005....................30

3.2

Description and source of regression variables ...................................................................31

3.3

Determinants of the 2005 extreme poverty rate...................................................................34

3.4

Determinants of 2005 moderate poverty rate.......................................................................35

3.5

Impact of the average economic, political, and geographic factors on reductions of the extreme poverty rate, 1980-2005 .........................................................................................38

3.6

Impact of the average economic, political, and geographic factors on reductions of the moderate poverty rate, 1980-2005 .......................................................................................39

3.7

The impact of changes in economic, political, and geographic factors on reductions of the extreme poverty rate, 1980-2005 .........................................................................................41

3.8

The impact of changes in economic, political, and geographic factors on reductions of the moderate poverty rate,1980-2005 ........................................................................................42

3.9

The impact of changes in economic, political, and geographic factors on reductions of the extreme poverty rate, 1990-2005 .........................................................................................44

3.10 The impact of changes in economic, political, and geographic factors on reductions of the moderate poverty rate, 1990-2005 .......................................................................................45 3.11 The impact of changes in economic, political, and geographic factors on reductions of the extreme poverty rate, 1990-2005 .........................................................................................47 3.12 The impact of changes in economic, political, and geographic factors on reductions of the moderate poverty rate, 1990-2005 .......................................................................................48 4.1

The impact of political institutions on subsequent changes in economic freedom (pooled OLS for ten-year periods, 1985-1995 and 1995-2005)........................................................62

4.2

The impact of political institutions (1975-1985) on subsequent changes in economic freedom (1985-1995) ...........................................................................................................63

vii

4.3

The impact of political institutions (1985-1995) on subsequent changes in economic freedom (1995-2005) ...........................................................................................................64

4.4

The impact of prior and concurrent changes in political institutions on changes in economic freedom (pooled OLS for ten-year periods, 1985-1995 and 1995-2005) ...........66

4.5

The impact of prior and concurrent changes in political institutions on changes in economic freedom, after controlling for geographic factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005)....................................................................................67

4.6

The impact of political institutions on subsequent changes in economic freedom for countries with poverty data (pooled OLS for ten-year periods, 1985-1995 and 19952005) ....................................................................................................................................68

4.7

The direct impact of political institutions on reductions of the extreme poverty rate, after controlling for economic freedom and geographic factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005)....................................................................................70

4.8

The direct and indirect impact of political institutions on reductions of the extreme poverty rate, after controlling for economic freedom and geographic factors (pooled OLS for tenyear periods, 1985-1995 and 1995-2005) ............................................................................71

4.9

The direct impact of political institutions on reductions of the moderate poverty rate, after controlling for economic freedom and geographic factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005)....................................................................................73

4.10 The direct and indirect impact of political institutions on reductions of the moderate poverty rate, after controlling for economic freedom and geographic factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005) ................................................................74 5.1

Empirical papers on foreign aid and growth, 1996-2011 ....................................................80

5.2

Countries with high levels of aid during each decade (ten percent or more), 1980-2005 ...89

5.3

Countries with increasing amounts of aid (eight percentage points increase or more), 1970-2005 ............................................................................................................................90

5.4

Countries with decreasing amounts of aid (a decline of eight percentage points or more as a share of GNI), 1970-2005.............................................................................................91

5.5

Determinants of foreign aid: Pooled OLS regressions with extreme poverty, economic freedom, and political institutions, 1981-2005 (five-year average).....................................94

5.6

Determinants of foreign aid: Pooled OLS regressions with moderate poverty, economic freedom, and political institutions, 1981-2005 (five-year average).....................................95

viii

5.7

Determinants of foreign aid: Pooled OLS regressions with extreme poverty, economic freedom, and political institutions, 1986-2005 (ten-year average)......................................96

5.8

Determinants of foreign aid: Pooled OLS regressions with moderate poverty, economic freedom, and political institutions, 1986-2005 (ten-year average)......................................97

5.9

The impact of foreign aid on changes in economic freedom after controlling for political institutions and geographic/locational factors, 1985-1995 ..................................................99

5.10 The impact of foreign aid on changes in economic freedom after controlling for political institutions and geographic/locational factors, 1995-2005 ................................................101 5.11 The impact of foreign aid on changes in economic freedom after controlling for political institutions and geographic/locational factors (pooled OLS for ten-year periods, 19851995 and 1995-2005) .........................................................................................................102 5.12 The impact of Foreign aid on reductions in the extreme poverty rate after controlling for political institutions and geographic/locational factors (pooled OLS for five-year periods, 1980-2005).........................................................................................................................105 5.13 The impact of Foreign aid on reductions in the moderate poverty rate after controlling for political institutions and geographic/locational factors (pooled OLS for five-year periods, 1980-2005).........................................................................................................................107 5.14 The impact of Foreign aid on reductions in the extreme poverty rate after controlling for political institutions and geographic/locational factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005)................................................................................................109 5.15 The impact of Foreign aid on reductions in the moderate poverty rate after controlling for political institutions and geographic/locational factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005)................................................................................................111 A.1

Extreme ($1.25 per day) and moderate ($2 per day) poverty rate by country...................124

B.1

Scoring of political rights and civil liberties......................................................................136

C.1

Economic Freedom of the World (EFW) summary statistics ............................................140

C.2

Polity IV summary statistics ..............................................................................................140

C.3

Executive constraints summary statistics...........................................................................141

C.4

Freedom House Political Rights Index summary statistics................................................141

C.5

Freedom House Civil Liberties Index summary statistics .................................................142

ix

C.6

Sachs geographic variables summary statistics .................................................................142

C.7

Foreign aid (ODA) summary statistics ..............................................................................143

D.1

Determinants of the 2005 SP extreme poverty rate (similar to table 3.3) .........................145

D.2

Determinants of the 2005 SP moderate poverty rate (similar to table 3.4) .......................145

D.3

The impact of changes in economic, political, and geographic factors on reductions of the SP extreme poverty rate, 1980-2005 (similar to table 3.7)................................................146

D.4

The impact of changes in economic, political, and geographic factors on reductions of the SP moderate poverty rate, 1980-2005 (similar to table 3.8)..............................................147

D.5

Determinants of the 2005 extreme poverty rate, bifurcated EFW (similar to table 3.3) ...148

D.6

Determinants of the 2005 moderate poverty rate, bifurcated EFW (similar to table 3.4) .149

D.7

The impact of changes in economic, political, and geographic factors on reductions of the extreme poverty rate with bifurcated EFW, 1980-2005 (similar to table 3.7) ..................150

D.8

The impact of changes in economic, political, and geographic factors on reductions of the moderate poverty rate with bifurcated EFW, 1980-2005 (similar to table 3.8) ................152

D.9

Correlation among economic freedom, Polity IV and geography (from regressions of table 3.7).....................................................................................................................................154

D.10 Correlation among economic freedom, executive constraints, and geography (from regressions of table 3.7).....................................................................................................155 D.11 Correlation among economic freedom, political rights, and geography (from regressions of table 3.7) ............................................................................................................................156 D.12 Multicollinearity regressions for extreme poverty, economic freedom, and Polity IV (similar to table 3.7)...........................................................................................................157 D.13 Multicollinearity regressions for extreme poverty, economic freedom, and executive constraints (similar to table 3.7) ........................................................................................158 D.14 Multicollinearity regressions for extreme poverty, economic freedom, and political rights (similar to table 3.7)...........................................................................................................159 D.15 Multicollinearity regressions for moderate poverty, economic freedom, and Polity IV (similar to table 3.8)...........................................................................................................160

x

D.16 Multicollinearity regressions for moderate poverty, economic freedom, and executive constraints (similar to table 3.8) ........................................................................................161 D.17 Multicollinearity regressions for moderate poverty, economic freedom, and political rights (similar to table 3.8)...........................................................................................................162 E.1

Countries with poverty rate data in 1980 and a population larger than 1 million in 2005 included in chapter five regressions (86 countries) ...........................................................164

E.2

The impact of Foreign aid on reductions in the extreme poverty rate after controlling for changes in economic freedom, initial political institutions, and geographic/locational factors, 1995-2005 .............................................................................................................167

E.3

The impact of Foreign aid on reductions in the moderate poverty rate after controlling for changes in economic freedom, initial political institutions, and geographic/locational factors, 1995-2005 .............................................................................................................168

xi

LIST OF FIGURES 2.1

World Bank extreme poverty rate ($1.25 per day, 2005 international $) of the world, 1980-2005 ............................................................................................................................16

2.2

World Bank moderate poverty rate ($2 per day, 2005 international $) of the world, 1980-2005 ............................................................................................................................17

2.3

Pinkovskiy and Sala-i-Martin extreme poverty rate ($1 per day, 2000 international $) of the world, 1970-2005...........................................................................................................18

2.4

Pinkovskiy and Sala-i-Martin moderate poverty rate ($2 per day, 2000 international $) of the world, 1970-2005...........................................................................................................19

xii

ABSTRACT This research is an empirical study of how economic freedom, democratic political institutions, and foreign aid influence global poverty rates. The updated extreme and moderate poverty rates recently released by the World Bank indicate that significant reductions in global poverty occurred during 1980-2005. Chapter two regularizes these data for statistical analysis in later chapters. This chapter also looks at global poverty trends over time, both in aggregate and for various regions. The data indicate that Asia had remarkable reductions in poverty during 1980-2005. The world’s two most populous countries, China and India, were the primary drivers of these reductions. Latin America also experienced reductions in poverty rates over the period while sub-Saharan Africa achieved very little reductions in poverty. Using the World Bank poverty rates, chapter three examines the impact of economic freedom on global poverty during 1980-2005. Prior research indicates that economic freedom is a significant determinant of economic growth. Some argue that only the wealthy are able to take advantage of movements toward increased economic freedom.

The results of this chapter

indicate that the poor benefit from economic freedom as well. Countries with more economic freedom over the period 1980-2005 had lower poverty rates in 2005. Moreover, movements toward institutions consistent with economic freedom corresponded to larger reductions in poverty over the period. This was especially true for changes in economic freedom that occurred earlier in the period. Prior research indicates that democratic political institutions exert very little influence on economic outcomes after accounting for the impact of economic freedom.

Chapter four

examines whether democratic political institutions exert an indirect impact on reductions in poverty through economic freedom. Milton Friedman and others have argued that political institutions exert an impact on economic freedom. This research provides some evidence in support of this view.

Movements toward democratic political institutions in one decade

corresponded to increases in economic freedom in the subsequent decade. After accounting for the indirect impact of political institutions, the results indicate that movements toward democracy exert an influence on subsequent reductions in extreme poverty through economic freedom. A similar result was not found for reductions in moderate poverty.

xiii

Chapter five examines the impact of foreign aid on global poverty during 1980-2005. Over the last half century, a significant amount of foreign aid was given to developing countries to reduce the incidence of poverty. The results of this chapter indicate that a countries poverty rate is an important determinate of the amount of aid received. Economic and political institutions of a recipient country had little to no impact on the allocation of foreign aid. The results also indicate that foreign aid exerted little influence on movements toward economic freedom or democracy. Lastly, the results of this chapter indicate that foreign aid exerted little to no impact on reductions in either extreme or moderate poverty during 1980-2005. These results indicate that poor countries receive foreign aid, but do not become economically free or democratic. Moreover, it appears that aid, as historically practiced, exerted little impact on global poverty. Chapter six is the concluding chapter. The empirical results are summarized and placed in context. This research indicates that economic freedom is a significant determinant of reductions in poverty. Democratic political institutions facilitate reductions in the extreme poverty rate through changes in economic freedom. Lastly, no evidence was found that foreign aid exerted a significant impact on either institutions or poverty.

xiv

CHAPTER 1 INTRODUCTION Modern growth theory can be disaggregated into three alternative theories. First, there is the production function theory associated with Solow (1956), Lucas (1988), and Romer (1990). This theory holds that inputs, namely savings, investment, and human capital are the source of economic growth. The second is the geography and location theory associated with the work of Sachs (2003) and Diamond (1999). The third is the new institutional theory of North (1990) and Acemoglu, Johnson, and Robinson (2001). The three theories are not necessarily mutually exclusive. This research is within the framework of the institutional theory. Considerable research examining the relationship between institutions and economic growth has already been undertaken. But, there is an absence of research investigating the impact of institutions on poverty in the developing world. This project will focus on the latter topic. The literature indicates that institutions supportive of economic freedom are particularly important. The central elements of economic freedom are: “personal choice, voluntary exchange coordinated by markets, freedom to enter and compete in markets, and protection of persons and their property from aggression by others (Gwartney and Lawson 2009).” Research has shown that countries with more or increasing economic freedom have higher growth rates and income levels than countries that are less free (Berggren 2003). This relationship between economic freedom and growth suggests that a negative association between economic freedom and the rate of poverty is a likely occurrence. However, many argue that when economic freedom is present, the rich will get richer and the poor will become poorer. Is this really true or does economic freedom and growth lead to higher incomes for the poor as well as others? The institutional literature focuses on political, as well as economic institutions. Milton Friedman argued, “History speaks with a single voice on the relation between political freedom and a free market. I know of no example in time or place of a society that has been marked by a large measure of political freedom, and that has not also used something comparable to a free market to organize the bulk of economic activity (Friedman 1962, 9).”

Friedman’s view

suggests that political and economic institutions are linked. Do political institutions influence the likelihood of economic reforms? Might political institutions facilitate reductions in poverty either directly, or indirectly through the promotion of liberal economic reforms? 1

Proponents of foreign aid argue that it is an effective tool with which to reduce poverty. For example, the Millennium Development Corporation seeks to reduce the world poverty rate by 2015 to one-half of the 1990 level. An expansion of foreign aid is an integral part of the strategy to achieve this goal. During the past half century a considerable amount of foreign aid has been allocated to less developed countries.

Has foreign aid helped to reduce poverty in these

countries? An answer to this question is important given the magnitude of the aid. This research is an empirical study of the impact of economic freedom, political institutions, and foreign aid on global poverty. Each of these topics is developed in subsequent chapters. However, before an empirical investigation of poverty is undertaken, a homogeneous measure across countries is needed. The second chapter focuses on this issue. The starting point is the extreme and moderate poverty rates, $1.25 and $2 per day in 2005 international dollars respectively, from the World Bank, World Development Indicators. The data used to generate these poverty rates come from household surveys conducted in various countries during 19802005. The household surveys used by the World Bank are conducted irregularly. In order to make the poverty rate data more comparable with other variables, these data were regularized to fiveyear intervals. Chapter two explains the statistical techniques utilized to make these adjustments. The result is a comprehensive poverty database that covers up to 128 countries at five-year intervals during 1980-2005. Several stylized facts are directly observable from the cross-country poverty rate data. First, global poverty has fallen considerably since 1980. Second, the largest reductions took place in Asia with China being the primary driver of those reductions. India also achieved reductions in poverty, but the declines were modest prior to 1990.

Third, Latin America had modest

reductions in their poverty rates, however, the average level of poverty during the period was much lower than the rest of the developing world. Finally, sub-Saharan Africa had persistently high poverty throughout the period. The fact that Asia had high economic growth during the period, Latin America moderate growth, and Africa practically none lends merit to the view that growth and poverty reduction are linked. The extreme and moderate poverty rates both declined in most countries during 1980-2005. Many countries implemented reforms increasing economic freedom during this same time frame.

2

The third chapter explores whether this increasing level of economic freedom played a role in the reductions in poverty. As far back as Smith and Ricardo economists have recognized that the gains from trade, specialization, and entrepreneurship are central ingredients of the growth process. Economic freedom is likely to enlarge the gains from these sources. Thus, economic freedom may help to reduce poverty. In addition, as Jeffrey Sachs has argued, geographic and location factors may affect the growth process. Countries in the tropical zones contend with diseases such as malaria and yellow fever.

These debilitating diseases lead to higher mortality rates and lower

productivity. Furthermore, countries a great distance from international markets face sizable transaction costs. A large proportion of the world’s poor live in areas with these geographic and location disadvantages. Chapter three examines the impact of economic freedom on poverty, controlling for the affects of geographic, locational, and political factors. Countries that on average had a higher level of economic freedom during 1980-2005 had lower extreme and moderate poverty rates in 2005. Moreover, countries with increases in economic freedom, especially during the earlier periods, experienced larger reductions in extreme and moderate poverty. These results were robust for alternative quantifications of economic freedom regardless of the measure of poverty utilized. The findings of this chapter also indicate that political institutions do not exert a statistically significant impact on poverty.

This result is peculiar in light of Friedman’s

conjecture that economic freedom and political freedom are closely related. The fourth chapter investigates this issue in more detail. After controlling for economic freedom, the results of prior research generally indicate that political institutions exert an insignificant impact on economic growth. However, there is some evidence that political institutions influence the adoption of reforms consistent with economic freedom. This raises the possibility that political institutions may indirectly affect poverty rates. Weingast (1995) argues that political institutions exert an impact on economic freedom through market-preserving federalism.

Federalism facilitates competition among different

sources of power within a country. In theory, this competition may curtail the tendency of a centralized authority to reduce the economic freedom of its citizens. The fourth chapter provides insight on this issue. The results indicate that movements toward democracy and constraints on the power of the executive are associated with increases in economic freedom in subsequent

3

periods. A two-step approach was used to determine if these movements indirectly facilitated reductions in poverty. The results are supportive of the view that movements toward democracy and limits on executive power facilitate subsequent reductions in the extreme poverty rate. These results, however, did not hold for the moderate poverty rate. Chapter five explores the relationship between foreign aid and reductions in poverty. The amount of foreign aid provided by wealthy countries increased steadily until the mid 1990s at which time it leveled off. The aid levels once again began to rise during the first decade of this century. Several countries were highly dependent on the receipt of foreign aid. Even though most sub-Saharan African countries received large amounts of aid, their high poverty rates persisted.

Botswana was the only sub-Saharan African country to significantly reduce its

dependence upon aid, increase per-capita income, and reduce poverty. This chapter focuses on three questions. First, what are the factors that determine the amount of aid received by various countries? International aid agencies for a time stressed the direction of aid “selectively” toward countries with better policies where it was thought aid would be used more effectively. Was this really the case? Second, did foreign aid exert an impact on economic and political institutions? Spurned on by the Washington Consensus, the international aid agencies often claimed that their aid was contingent upon democratic and market based reforms. Was aid really allocated in this manner, and if so, did it influence economic and political institutions? Third, did foreign aid exert an impact, either direct or indirect, on the reduction of global poverty during 1980-2005? The effectiveness of foreign aid is dependent upon the answer to this latter question. What does this research indicate with regard to the answers to these three questions? First, the primary determinant of the level of aid received was a country’s poverty rate. Neither economic nor political institutions exerted an impact on the level of aid received. Second, there was no evidence that aid exerted an impact on either economic freedom or political institutions. This indicates that the Washington Consensus either exerted little impact on the allocation of aid or, if it did, the aid was ineffective. Third, the statistical findings indicate that foreign aid failed to exert an impact, either positive or negative, on the poverty rate of countries during 1980-2005. This is consistent with the view that foreign aid has been largely ineffective as a tool for the achievement of its stated goal. These results are consistent with the influential study of Rajan and Subramanian (2008) regarding the linkage between foreign aid and growth.

4

The final chapter summarizes the findings of this research. The results are supportive of the following conclusions: 1) movements toward institutions consistent with economic freedom have been a primary driver of reductions in global poverty; 2) democracy and limitations on executive power facilitate movements toward economic freedom and thereby indirectly promote reductions in the extreme poverty rate; 3) foreign aid is largely ineffective as a means for the promotion of economic freedom, democratic reforms, and reductions in global poverty.

5

CHAPTER 2 GLOBAL POVERTY: MEASUREMENT AND REGIONAL ANALYSIS Many developing countries experienced rapid economic growth over the period 1980-2005. China’s real per capita income increased by almost a factor of eight while that of Botswana, Chile, and India more than doubled. These statistics indicate remarkable expansions of wealth in developing countries since the 1980s, but reveal very little about the economic circumstances of the poor. Increasing average income levels could potentially mask large increases in income inequality, i.e. the rich become richer and the poor poorer. Thus it is important to examine how the poor fared during this period. Was the economic expansion of 1980-2005 accompanied by declines in global poverty? If so, were the reductions in poverty widespread or localized within specific countries or regions? The aim of this chapter is to explore these questions. The most widely used measures of world poverty come from the World Bank, World Development Indicators. These measures are based upon household income and consumption surveys conducted irregularly in various countries.

This irregularity complicates statistical

analysis. Therefore, the World Bank poverty measures used here, and in subsequent chapters, were regularized to five-year intervals.

An alternative poverty measure constructed by

Pinkovskiy and Sala-i-Martin is included in order to account for a possible discrepancy that exists with income data.

Income levels and growth rates computed from survey data are

consistently lower than those from national income accounts. Thus, there is a possibility that the World Bank poverty measures overstate the level of poverty. The alternative measure utilizes both types of data and unsurprisingly indicates lower levels of poverty. Both the World Bank and Pinkovskiy and Sala-i-Martin poverty rates indicate significant reductions in world poverty occurred during 1980-2005. In fact, both measures show that the first Millennium Development Goal has almost been achieved. That is, halving the number of people who live on $1.25 a day or less, in 2005 international dollars, between 1990 and 2015. These poverty reductions, however, occurred primarily in Asia and Latin America. China had by far the largest reductions in poverty as it began the 1980s with the third highest poverty rates but by 2005 had poverty rates approaching those of Latin America. India also experienced poverty reductions, most of which took place after 1990. Poverty reductions in Latin America were

6

modest, but the overall level of poverty was much lower than other world regions throughout the period. Very little, if any, reductions occurred in sub-Saharan Africa. This is in stark contrast to the reductions that took place in Asia. In 1980, Asia’s poverty rates were higher than those of sub-Saharan Africa, but by 2005 Asia’s rates were much lower. While sub-Saharan Africa’s poverty situation stagnated during the period, Asia achieved significant reductions. The material in this chapter proceeds as follows. Section 2.1 discusses the latest World Bank poverty rates and how they differ from previous versions. Section 2.2 discusses in detail the statistical techniques used to regularize the World Bank data for statistical analysis. Section 2.3 introduces the Pinkovskiy and Sala-i-Martin poverty rates while section 2.4 uses both measures to examine world poverty over the period 1980-2005. Section 2.5 sets up questions, explored in subsequent chapters, concerning the regional differences in poverty reduction over the period and finally concludes.

2.1 The World Bank, World Development Indicators Extreme and Moderate Poverty Rates The most widely used measures of poverty are the extreme and moderate poverty rates from the World Bank, World Development Indicators. They are the percentage of a country’s population living on $1.25 and $2 per day or less, respectively, in 2005 international dollars. These rates are reported for 115 developing countries over the period 1978-2007.1 The initial measurements of global poverty began with the work of Ahluwalia, Carter, and Chenery in 1979.

These

measurements were refined and improved over time with the latest poverty rates generated by two World Bank researchers, Shaohua Chen and Martin Ravallion. The extreme and moderate poverty rates for each country are derived from household survey data. These household surveys are used to create an income distribution, which is then adjusted to 2005 international dollars using PPP conversion ratios. The $1.25 and $2 per day poverty lines are then applied to the income distribution of a country to determine the number of people living in either extreme or moderate poverty. The respective poverty rates are computed by dividing the number of people living at or below these poverty lines by the country’s population. 1

In 1978, 1979, and 1980 there are poverty rates for only one country each year. These countries are: India (1978), Panama (1979), and Madagascar (1980). It is not until the early 1980s that significantly more countries have reported poverty rates.

7

The latest World Bank poverty rates differ from previous measures in four major respects. The first, and most significant is the use of the PPP conversion ratios from the 2005 round of the International Comparison Program (ICP).

The 2005 ICP was a worldwide pricing survey

conducted to generate PPP conversion ratios. The previous poverty rates used PPP conversion ratios from the 1993 ICP, which did not include China, the world’s most populous country. In addition, the older ICP was less accurate due to insufficient allowances for differences in the quality of goods and services. PPP ratios are calculated by dividing the average price (in the local currency) of a basket of goods and services in a particular country by the price of the same basket in U.S./international dollars. Failing to account for differences in the quality of the goods and services in the basket will result in underestimation of the PPP ratios for developing countries. For example, if a market basket in a developing country contains goods that are of a lower quality than can be found in the U.S., this would imply that the prices in the numerator of the PPP ratio are lower than they should be, resulting in a lower PPP ratio. This leads to underestimation of the cost of living in developing countries. Hence, the older PPP ratios overestimated income levels and underestimated the level of poverty. The latest PPP ratios also adjust for differences in staple goods between regions. The type of food consumed in Asia is very different from food consumed in Africa. Failing to account for these differences complicates cross-country comparisons. This takes on additional importance when the subject is food, as a major consequence of poverty is hunger and inadequate nutrition. Hence, these adjustments allow for a more accurate comparison of all goods and services across regions. The 2005 ICP is a significant improvement over the previous ICP.2 China was included and the price surveys for each country included more goods and services with much more detail in order to compensate for differences in quality and variety throughout the world. In addition, the number of countries included in the price surveys increased from 117 to 146 (Chen and Ravallion 2008). On average, the 1993 PPP conversion rates resulted in an understatement of poverty. Therefore the use of the 2005 PPP ratios corresponds with a slightly higher level of poverty. The second way that the latest World Bank poverty rates differ from previous measures is due to the inclusion of statistical techniques to control for an urban bias inherent in the PPP 2

For a thorough discussion of suggestions to improve the 1993 ICP see the Ryten Report (UN, 1998).

8

ratios for some countries. An urban bias often exists in poor countries, as it is easier to conduct surveys in urban areas than poor rural areas. The ICP gathers prices on a number of goods and services in many countries in order to determine the PPP conversion ratios. Gathering prices in rural areas is challenging, as travel can be difficult. Moreover, inhabitants of rural areas are often responsible for the production of their own goods and services. In addition, markets that do exist in rural areas are often informal which further complicates price surveys. The new methodology compensates for this urban bias. The older poverty rates made no adjustments for this bias, which suggests that, on average, they contributed to an understatement of poverty. Accordingly, adjustment for this bias in the latest poverty rates resulted in, on average, slightly higher levels of poverty. The third difference is an increase in the number of household surveys used to compute the poverty rates. The latest poverty rates used 675 surveys from 115 countries, up from 454 surveys and 97 countries in the previous estimates (Chen and Ravallion 2004 2008). This increase in the number of surveys suggests that the latest poverty estimates cover more of the world’s poor than the previous measures. The fourth change is an adjustment of the international poverty lines for both the extreme and moderate poverty rates. The new income thresholds for the extreme and moderate poverty rates are $1.25 and $2.00 per day, respectively, measured in 2005 international dollars. The previous poverty thresholds were $1.08 per day for the extreme poverty rate and $2.15 per day for the moderate poverty rate, measured in 1993 international dollars. The previous poverty measures used 1993 international dollars because this was the year in which the previous ICP was conducted. The new poverty rates rely upon the PPP conversion ratios of the 2005 ICP and therefore are denominated in 2005 international dollars.

While the extreme and moderate

poverty thresholds have changed, the methodology for determining the threshold has not. The income threshold for extreme poverty is computed as the mean of the poverty lines of the 15 poorest countries with household survey data. The moderate poverty threshold of $2.00 per day is the mean of the poverty lines of the developing countries excluding the poorest 15. It should be noted that the previous extreme poverty threshold of $1.08 per day in 1993 international dollars is $1.45 in 2005 dollars, after adjusting for U.S. inflation. The previous moderate poverty threshold of $2.15 per day in 1993 international dollars is $2.88 in 2005 after adjusting for U.S. inflation. This implies that, in real terms, the poverty thresholds are lower for the latest poverty

9

rates, especially for moderate poverty. This change will result in, on average, lower poverty rates than previous World Bank estimates. The lowering of the poverty lines reduced measured poverty while the use of the new PPP conversion ratios, as well as statistical adjustment for an urban bias, increased measured poverty. Overall, these changes resulted in a higher estimated incidence of poverty for both the extreme and moderate rates. While the newer rates indicate a higher level of poverty they are more comprehensive and there is good reason to expect that they are a more accurate measurement of world poverty. With a change in methodology there is a concern about whether continuity with previous rates is maintained. It is expected that these changes had an impact on the measurement of poverty levels while preserving time trends between the older and newer measures. This was indeed the case. The correlation between the old and new measures was computed with 322 country-year observations.

The correlation coefficient between the old and new poverty

measures was 0.88 for extreme poverty and 0.92 for moderate poverty. These high correlation coefficients suggest that, while the latest poverty measures are, on average, higher than the previous measures, they are capturing similar time trends.

2.2 Preparation of the World Bank Extreme and Moderate Poverty Rates for Use in Statistical Analysis The World Bank extreme and moderate poverty rates were created from surveys conducted irregularly and for different years in different countries. This complicates the use of the poverty rates in statistical analysis as international data is reported regularly at least every five years. Therefore, statistical procedures were used to adjust the World Bank poverty rates and to derive estimates at five-year intervals during 1980-2005. The adjustment of the data was split into two parts. The first part was the adjustment of the data at five-year intervals while the second part was a regression analysis that estimated missing values. Appendix A provides these estimates for the 128 countries for which these poverty rates could be derived.

2.2.1 Part 1: Adjusting the World Bank Poverty Rates At Five-Year Intervals and Filling In Several Missing Values Adjustment of the World Bank poverty rates involved five steps.

10

1. When the World Bank data were available for a year ending in zero or five for a country, these poverty rate figures were used. There were 101 observations for each poverty rate that fit this category during 1980-2005, which is 15 percent of the total observations. 2. If the World Bank did not provide the poverty rate figures for a country in a zero or five year, but there was a value in each of the adjacent years, the average of the adjacent years was used as the value for the missing year ending in zero or five. For example, the 2000 data for Brazil were unavailable, but the data were available in both 1999 and 2001. Brazil’s extreme poverty rate was 11.2 in 1999 and 11.0 in 2001. Thus, the 2000 figure for Brazil’s extreme poverty rate was merely the average of these two figures, 11.1. There were ten observations for each poverty rate that fit this category during 1980-2005, which is roughly 1 percent of the total observations. 3. If a country had a value for only one adjacent year, that value was used for the missing year ending in zero or five. Poverty rates seldom change much from one year to the next. Therefore the figure for an adjacent year will nearly always be a good estimate for the missing value. For example, Botswana had an extreme poverty rate of 31.2 in 1994 but no data were available for 1995, 1996, or 1997. Thus, the 1994 figure of 31.2 was used as the value for Botswana's extreme poverty rate in 1995. There were three observations for each poverty rate that fit this category during 1980-2005, which is less than 1 percent of the total observations. 4. If none of the above scenarios fit a country during a specific five-year period, but it still had poverty data in a five-year window centered on the year ending in zero or five, the value for that period became the average of all the values in the five-year window. For example, Ghana had an extreme poverty rate of 50.6 in 1988, 49.4 in 1989, no values for 1990 or 1991, and 51.1 in 1992. Therefore, Ghana's extreme poverty rate for 1990 became the average of the values over the fiveyear window centered on 1990 (i.e. 1988-1992), which was 50.3. There were 245 observations for each poverty rate that fit this category during 1980-2005, which is 36 percent of the total observations. 5. In cases where a gap of a decade existed after the above procedures were used, the real per capita GDP data were used to fill in the middle year in cases where the per capita income and poverty rates moved in opposite directions. This was the case for 15 countries. Again, Ghana can be used to illustrate the procedure. Ghana had a poverty rate in 1990 and 2000, but was missing a value for 1995. The real per capita GDP data for Ghana were available for 1990, 1995, and 2000 and the pattern of these figures was used to adjust and predict the poverty rate

11

value for the missing middle year (1995). Ghana had an extreme poverty rate of 50.3 in 1990 and 39.1 in 2000. Ghana’s per capita GDP (measured in 2005 constant international dollars) increased from $861 in 1990 to $925 in 1995 and $1,015 in 2000. Thus, there was an upward trend in per capita GDP and a downward trend in poverty over this period so the per capita GDP figures were used to estimate the missing poverty rate value. Equation 1 was used to generate the missing poverty values under these conditions.

(1) povertyt =povertyt-5 + (povertyt+5 - povertyt-5)*(GDPt - GDPt-5)/(GDPt+5 - GDPt-5) Where povertyt is the missing poverty rate, povertyt-5 is the poverty rate five years prior to the missing rate, povertyt+5 is the poverty rate five years after the missing rate, GDPt is the per capita real GDP for the same year of the missing poverty rate, GDPt-5 is the per capita real GDP five years prior to the missing rate, and GDPt+5 is the per capita real GDP five years after the missing rate. Equation 1 uses real per capita GDP as a scale factor to predict the missing poverty value. The GDP ratio in the last term of the equation represents the proportion of the increase (decrease) of GDP that took place in the first five years of the decade. This is then multiplied by the decline (rise) in poverty over the decade. Lastly, this change in poverty is then added to the poverty rate at the start of the decade. There were 14 observations for each poverty rate that fit this category during 1980-2005, which is 2 percent of the total observations. The methodology described in this first part, steps 1-5, was used to derive the extreme and moderate poverty rates for 373 country-year observations, which comprised 55 percent of the total. At least one observation was present for 115 countries and at least three observations for 79 countries over the period 1980-2005.

2.2.2 Part 2: Regression Analysis to Fill In Missing Poverty Values The log of per capita real GDP and the under-five mortality rate are major determinants of both the extreme and moderate poverty rates. Using the 373 observations derived in the previous section, the log of per capita real GDP and the under-five mortality rate was regressed against the extreme poverty rate. The r-square for this equation was 0.76. When the same equation was run

12

with the moderate poverty rate as the dependent variable, the r-square was 0.80. These high rsquare values indicate that taken together, the per capita real GDP and under-five mortality rate are excellent predictors of both the extreme and moderate poverty rates. The per capita real GDP and under-five mortality data are available for countries and years beyond which the procedures of the previous section could be used to derive the poverty rates for years ending in zero or five during 1980-2005. Because the log of per capita real GDP and the under-five mortality rate are major determinants of the extreme and moderate poverty rates, these data can be used to estimate the poverty rates for a broader range of countries and years. The 373 observations were used to estimate the following regression:

(2) povertyit = log(GDPit) + u5Mortit + dSSaharanAfrica + dOutlierCountries Where povertyit is the poverty rate of country i in time period t, log(GDPit) is the log of per capita real GDP of country i in time period t, u5Mortit is the under-five mortality rate of country i in time period t, dSSaharanAfrica is a dummy for sub-Saharan Africa, and dOutlierCountries is a dummy for several outlier countries. A country was considered an outlier if its dummy variable was significant at the 90 percent level or more in the above regression. The r-square for this equation was 0.90 when the extreme poverty rate was the dependent variable and 0.91 when the dependent variable was the moderate poverty rate. The country data for per capita real GDP and under-five mortality rate along with the dummies (if they applied) were then inserted into the regression equations for the missing years and used to estimate the country’s extreme and moderate poverty rates for those years. This methodology was used to estimate a country’s poverty rate when it could not be derived for a specific year by steps 1-5 in section 2.2.1 above. In addition to the 115 countries for which data for at least one year were available from the World Bank, this regression procedure was also used to estimate the extreme and moderate poverty rates for another 13 countries with a population of more than 1 million. The World Bank limits the poverty rate at 2 percent for countries whose poverty rate falls below this level. This convention was used for countries whose predicted poverty rates were below 2 percent. Similarly, countries whose predicted poverty rates were greater than 99 percent were limited to that figure.

13

In total, it was possible to estimate the extreme and moderate poverty rate figures for 683 country-year observations among the 128 countries. Appendix A provides these estimates for the period 1980-2005 in five-year intervals. The bold values in the table are the actual poverty rates as reported by the World Bank after adjustment by the procedures explained in section 2.2.1 steps 1-5. The 310 non-bold values for each poverty rate were calculated by the regression methodology explained here. This methodology accounted for 45 percent of the poverty values. With these adjustments, the World Bank extreme and moderate poverty rates are now more comprehensive and easier to use for statistical analysis than the original World Bank figures. The coverage of the original poverty rates was 11 percent in 2000 and 59 percent of the world’s population in 2005. The adjustments described here increased the coverage to 84 percent in 2000 and 85 percent in 2005. When the high-income countries are included, the coverage increases to 99 percent for both 2000 and 2005, which suggests that the adjusted poverty rates encompass nearly the entire developing world. All references to the World Bank extreme and moderate poverty rates, both in the rest of this chapter and in the chapters that follow, will be to the adjusted World Bank poverty dataset described here.

2.3 Pinkovskiy and Sala-i-Martin Poverty Dataset: An Alternative Measure of Poverty The income distributions used to derive the World Bank poverty rates are derived from household surveys.

Income levels can also be calculated from national income accounts.

Interestingly, income levels as measured by surveys and national income accounting differ. Income levels and growth rates from survey data are much lower than those computed from national income accounts; this gap has been growing larger over time (Deaton 2005). This divergence is not restricted to poor countries as high-income countries also exhibit this pattern. It has been observed in the United States, the United Kingdom, and India (Deaton 2005). This increasing difference over time implies that poverty rates generated from survey data will on average be higher and show less reduction over time than those using national income accounting data. A hybrid approach, which uses both household surveys and national income accounts, was implemented by Pinkovskiy and Sala-i-Martin in a 2009 NBER working paper. This dataset

14

includes 191 countries, which cover 98 percent of the world’s population in 2006, annually over the period 1970-2006. The extreme poverty rate in this dataset represents the percentage of a country’s population living on $1 per day or less in 2000 international dollars. This is slightly below the $1.25 per day in 2005 international dollars threshold used in the World Bank poverty measures. The moderate poverty rate created by Pinkovskiy and Sala-i-Martin represents the percentage of a countries population living on $2 per day or less in 2000 international dollars. To avoid confusion it should be noted that the base year for the World Bank poverty rates is 2005 while the base year for the Pinkovskiy and Sala-i-Martin poverty rates is 2000. Pinkovskiy and Sala-i-Martin constructed their poverty measures using lognormal income distributions for each country with a mean fitted to match that from the national income accounts. The shape, or variance of the income distribution, is estimated using the quintile shares as reported by household surveys. After the income distribution for each country and year is estimated, the extreme and moderate poverty income thresholds ($1 and $2 per day in 2000 international dollars, respectively) are applied to the income distributions to determine the poverty rates.

2.4 Analysis of Global Poverty Using the World Bank and Pinkovskiy and Sala-i-Martin Poverty Measures Figures 2.1 and 2.2 present the World Bank extreme and moderate poverty rates for all countries and for developing countries during 1980-2005. The figures for all countries include 27 highincome countries (and one territory) from Western Europe and North America along with Japan, Australia, and New Zealand as well as the 128 developing countries listed in appendix A. These 155 countries and 1 territory cover 99 percent of the world’s population in 2000 and 2005. As was mentioned previously, the World Bank imposes a lower limit of 2 percent for any country whose poverty estimate falls below that level. Therefore the 27 high-income countries and 1 territory were assigned a poverty rate of 2 percent to compute the world average for the extreme and moderate poverty rates.

15

70

Extreme Poverty Rate (%)

60 50

58.4 47.5

47.7 39.4

40

42.0 35.6

35.9 30.8

31.7 27.4

30

25.1 21.8

20 10 0 1980

1985

1990 Developing Countries

1995

2000

2005

All Countries

Figure 2.1: World Bank extreme poverty rate ($1.25 per day, 2005 international $) of the world, 1980-2005

As figure 2.1 illustrates, the extreme poverty rate for the world’s developing countries fell from 58.4 percent in 1980 to 42.0 percent in 1990 and 25.1 percent in 2005. Thus, over the 25year time frame, the extreme poverty rate fell by more than 30 percentage points.3 When the high-income countries are included, the world’s poverty rate fell from 47.5 percent in 1980 to 35.6 percent in 1990 and 21.8 percent in 2005. The gap between the extreme poverty rate for developing countries and the parallel rate for all countries (including those with high-incomes) fell from 11 percent in 1980 to approximately 3 percent in 2005. The narrowing of this gap reflects both the substantial reduction of the extreme poverty rate in developing countries, as well as the increasing share of the world’s population residing in the less developed world. While a little more than one in five persons in the world still lives in extreme poverty, this is less 3

The number of developing countries for which data were available ranged from 92 in 1980 to 128 in 2005. The poverty rate data were available for 89 countries continuously (for years ending in five or zero) during 1980-2005. These 89 countries comprised 91 percent of the developing world population in 2005 and an even higher percentage in the earlier years. The extreme poverty rate of these 89 countries fell from 59 percent in 1980 to 46 percent in 1990 and 26.8 percent in 2005. Thus, the extreme poverty rates (and changes in those rates) for the set of countries with data throughout the period were quite similar to those presented in figure 2.1 for all developing countries with data during the specified year.

16

than half the rate of 1980. In all cases, the aggregate poverty rate calculations were derived by weighting the poverty rate for each country by its population during the year. As a result, the rates presented in the figures here are also equal to the mean poverty rate for the grouping.

90 Moderate Poverty Rate (%)

80 70

75.7 70.4 61.5

60

58.0

62.2

58.0 52.7

50

49.5

53.2 45.7

45.6 39.4

40 30 20 10 0 1980

1985

1990

Developing Countries

1995

2000

2005

All Countries

Figure 2.2: World Bank moderate poverty rate ($2 per day, 2005 international $) of the world, 1980-2005

Figure 2.2 presents similar data for the moderate ($2 per day) poverty rate. Because this standard implies a higher level of income, the poverty rate for this measure will always be higher than the parallel extreme poverty rate. The moderate poverty rate of developing countries declined from 75.7 percent in 1980 to 62.2 percent in 1990 and 45.6 percent in 2005.4 Thus, there was a 29.7 percentage point reduction in moderate poverty in the world’s developing countries during the quarter of a century following 1980. The moderate poverty rate in the developing countries in 2005 was approximately three-fifths the level of 1980. When the high-income countries are included, the moderate poverty rate for the world was 39.4 percent in 2005, down from 61.5 percent in 1980. These figures indicate that in 2005 4

When only the 89 countries with data available for all years are considered, the moderate poverty rate fell from 76.5 percent in 1980 to 68.6 percent in 1990 and 48.8 percent in 2005. This is a reduction in the moderate poverty rate of 27.7 percentage points, which is slightly less than the reduction based on all countries with data during a specific year.

17

approximately two out of every five people in the world lived on an income of less than $2 per day, compared with three out of every five in 1980.

40

Extreme Poverty Rate (%)

30

36.9 29.6

31.6 25.7 23.3 19.2

20

13.0 10.8 10

9.7

8.3

8.7

7.5

7.0 6.0

5.8 5.1

2000

2005

0 1970

1975

1980

1985

Developing Countries

1990

1995

All Countries

Figure 2.3: Pinkovskiy and Sala-i-Martin extreme poverty rate ($1 per day, 2000 international $) of the world, 1970-2005

Figures 2.3 and 2.4 present the extreme and moderate poverty rates generated by Pinkovskiy and Sala-i-Martin for all countries and for developing countries for 1970-2005. While the Pinkovskiy and Sala-i-Martin poverty rates cover more countries than the World Bank poverty rates, figures 2.3 and 2.4 only include countries with corresponding World Bank poverty rates.5 The similarity between the World Bank poverty rates and the figures presented here is apparent. The extreme poverty rate for the developing world, as measured by Pinkovskiy and Sala-i-Martin and shown in figure 2.3, fell from 36.9 percent in 1970 to 23.3 percent in 1980 to 5.8 percent in 5

When all countries in the Pinkovskiy and Sala-i-Martin poverty dataset are included in the world average the extreme poverty rate for the developing world falls from 29.6 percent in 1970 to 19.1 percent in 1980 to 5.6 percent in 2005. The moderate poverty rate for the developing world falls from 49.8 percent in 1970 to 42.2 percent in 1980 to 13.9 percent in 2005. These values are almost identical to the values when just the countries with World Bank data are considered. This insignificant difference is due to the fact that the countries excluded from the Pinkovskiy and Sala-i-Martin world averages shown in figures 2.3 and 2.4 have very small populations and do not exert a significant influence on the world poverty average. The same pattern emerges when the high-income countries are included in the averages for the entire world.

18

2005. A decline of 31.1 percentage points over the entire 35-year period and 17.5 percentage points over the later 25-year period. While the level of extreme poverty is lower than the corresponding World Bank measure during the period, the reductions over time are similar. The moderate poverty rate for the developing world, shown in figure 2.4, exhibits a fall from 62.0 percent in 1970 to 51.4 percent in 1980 to 14.9 percent in 2005. This represents a reduction of 47.1 percentage points from 1970 and 36.5 percentage points from 1980. The fall in the moderate poverty rate is substantial, and similar to the 29.7 percentage point reduction in the World Bank moderate poverty rate during 1980-2005 shown in figure 2.2.

70 62.0 58.2

Moderate Poverty Rate (%)

60 49.9 50

51.4 47.4 42.4 40.7

40

34.0 29.5 25.3 24.7 21.3

30 20

18.3 15.9 14.913.0

10 0 1970

1975

1980

1985

Developing Countries

1990

1995

2000

2005

All Countries

Figure 2.4: Pinkovskiy and Sala-i-Martin moderate poverty rate ($2 per day, 2000 international $) of the world, 1970-2005

As figures 2.1-2.4 illustrate, the poverty rates generated by Pinkovskiy and Sala-i-Martin are lower and exhibit larger reductions over time compared to the World Bank rates. This is to be expected, as the income levels from national income accounts are higher than those reported by household surveys. The Pinkovskiy and Sala-i-Martin dataset uses both sources of data to generate the poverty rates while the World Bank uses only the household surveys. However, both poverty measures are capturing similar trends over time.

19

Table 2.1: World Bank mean extreme and moderate poverty rates for sub-Saharan Africa, Latin America, Asia, India, and China, 1980-2005 Extreme Poverty Rate ($1.25 per day) No. of Countries Country/Region Sub-Saharan Africa 39 Latin America 24 Asia 15 China India Asia, omitting China 13 and India Moderate Poverty Rate ($2 per day) No. of Countries Counrty/Region Sub-Saharan Africa 39 Latin America 24 Asia 15 China India Asia, omitting China 13 and India

1980 60.8 15.6 69.1 84.0 65.9 47.1

1985 58.1 14.7 55.7 61.7 55.5 46.0

1990 60.3 11.2 53.5 60.2 53.6 42.6

1995 57.7 9.7 43.7 45.0 49.4 34.8

2000 57.1 10.9 36.4 32.0 46.5 30.9

2005 51.3 8.1 26.9 15.9 41.6 24.7

1980 77.3 25.5 88.3 97.8 89.0 71.1

1985 77.3 26.8 82.2 88.3 84.8 69.2

1990 78.1 21.3 79.2 84.6 83.8 65.0

1995 77.4 20.6 71.3 71.8 81.7 57.8

2000 76.1 21.4 63.4 56.3 79.4 54.6

2005 72.3 17.0 52.5 36.3 75.6 48.0

Table 2.1 lists the average World Bank extreme and moderate poverty rates for various regions and countries during 1980-2005. Countries with data available continuously for years ending in either zero or five are included in this table. By far, Asia had the largest reductions over the 25-year period as the extreme poverty rate fell from 69.1 percent to 26.9 percent and the moderate poverty rate fell from 88.3 percent to 52.5 percent. These reductions were largely driven by China and India as these two countries account for half of the extreme poverty rate reduction over the period and two-fifths of the moderate poverty rate reductions. Latin America had much lower poverty rates than other regions and achieves consistent reductions over the 25year period. The extreme poverty rate fell from 15.6 percent to 8.1 percent and the moderate poverty rate fell from 25.5 percent to 17.0 percent. In contrast to the reductions in Asia and Latin America, table 2.1 indicates that sub-Saharan Africa had very little poverty reduction over the 25-year period, with reductions occurring only in the last five years. The extreme poverty

20

rate for sub-Saharan Africa was 60.8 percent in 1980 falling slightly in 1985 and increasing back to 60.3 percent in 1990 before falling to 51.3 percent by 2005. The moderate poverty rate in subSaharan Africa exhibits a similar trend as it began the 1980s at 77.3 percent, increased to 78.1 percent in 1990 before dropping to 72.3 percent by 2005.

Table 2.2: Pinkovskiy and Sala-i-Martin extreme and moderate poverty rates for sub-Saharan Africa, Latin America, Asia, India, and China, 1970-2005 Extreme Poverty Rate ($1 per day) No. of Countries 1970 Country/Region Sub-Saharan Africa 39 40.9 Latin America 24 11.8 Asia 15 44.4 China 73.7 India 21.4 Asia, omitting China 13 15.0 and India Moderate Poverty Rate ($2 per day) No. of Countries 1970 Country/Region Sub-Saharan Africa 39 66.4 Latin America 24 25.9 Asia 15 72.8 China 92.4 India 61.1 Asia, omitting China 13 47.8 and India

1975 39.3 6.1 37.7 62.2 20.1 11.1

1980 41.1 4.0 25.9 44.2 11.6 8.9

1985 44.4 4.7 9.8 13.4 6.7 7.3

1990 43.4 5.0 5.8 7.9 3.3 5.5

1995 43.4 4.7 3.8 3.0 2.5 6.7

2000 38.5 4.3 1.7 0.6 0.6 4.8

2005 32.8 3.1 1.1 0.3 0.3 3.4

1975 65.2 17.0 69.1 89.3 59.6 39.6

1980 66.4 12.5 60.4 81.7 49.5 32.8

1985 68.6 14.1 44.3 55.8 39.1 30.9

1990 67.4 14.8 30.7 37.2 27.4 24.0

1995 68.9 14.2 21.6 19.6 23.0 23.2

2000 66.2 13.1 11.5 7.6 11.0 18.2

2005 61.3 10.5 7.4 4.1 5.9 13.9

Table 2.2 is similar to table 2.1, but lists the Pinkovskiy and Sala-i-Martin extreme and moderate poverty rates for various regions and countries. The same trends that appeared in the World Bank poverty rates of table 2.1 can be seen in table 2.2. Asia had the largest reductions as the extreme poverty rate fell from 44.4 percent in 1970 to 1.1 percent in 2005, while the moderate poverty rate fell from 72.8 percent to 7.4 percent over the same period. Again, the reductions for Asia were largely driven by China and India as the percentage point reductions for the region, when those two countries are omitted, are significantly less. Latin America had consistent reductions of poverty during 1970-2005 as the extreme poverty rate fell from 11.8

21

percent in 1970 to 3.1 percent in 2005 while the moderate poverty rate fell from 25.9 percent to 10.5 percent over the same period. Similar to table 2.1, Latin America had much lower poverty rates than the other regions throughout the period. These poverty rates also show very little poverty reduction in Sub-Saharan Africa with any reduction occurring in the last five years. Interestingly, in both tables 2.1 and 2.2, Asia had the higher initial extreme and moderate poverty rate, but by 2005 Sub-Saharan Africa had the higher rates as reductions in Asia over the period moved it closer to the rates of Latin America. Tables 2.1 and 2.2 illustrate that while the two poverty measures differ in levels, they both measure similar aspects of global poverty as they demonstrate similar time trends. In addition to the tables, the correlation between these two measures of poverty was examined. The correlation coefficient between the extreme poverty rates was 0.76 and 0.84 for the moderate rates. Again, this illustrates that the two poverty measures are measuring similar aspects of global poverty. The World Bank poverty rates are the most commonly used measure of poverty in the literature and by international organizations. They are the primary measure of poverty here. They are based solely upon household surveys in developing countries and as such contain data that comes directly from the poor (Chen and Ravallion 2008; Deaton 2005). In addition, Deaton argues that, “…if we need to measure poverty in a way that will convince those who are skeptical of the idea that average growth reaches the poor, there is little choice but to use the surveys (Deaton 2005, 18).” The poverty measures of Pinkovskiy and Sala-i-Martin will be used throughout in order to verify the robustness of the statistical results. Factors that affect poverty should be statistically significant with either measure.

2.5 Conclusion The World Bank and Pinkovskiy and Sala-i-Martin measures of poverty indicate that the poor improved their standard of living during the rapid economic expansion of 1980-2005. According to the World Bank poverty rates, extreme poverty fell from 58.4 percent of the developing world’s population in 1980 to 25.1 percent by 2005, nearly achieving the 2015 Millennium Development Goal of halving extreme poverty. While global poverty fell over the period, the reductions occurred primarily in China, India, and Latin America. The incidence of extreme poverty in China fell from 84.0 percent in 1980 to 15.9 percent by 2005. This is an almost 70

22

percentage point reduction in the number of people living on $1.25 a day or less in 2005 international dollars. China’s reduction in moderate poverty is no less remarkable as nearly the entire country lived in moderate poverty in 1980, but by 2005 it had fallen to 36.3 percent of the population. India’s reductions were not as significant as China’s, as much of the changes took place during the latter half of the period. While the poverty reductions in Latin America were modest, the overall level of poverty was much lower than other developing regions throughout the period. Unfortunately, sub-Saharan Africa experienced little, if any, reductions in poverty. While much of the developing world achieved poverty reductions, conditions in Africa however, continue to lead to stagnating levels of poverty. The differences in poverty reductions in Asia, Latin America, and sub-Saharan Africa suggest questions for further study. Why did Asia experience dramatic reductions in poverty while Sub-Saharan Africa experienced little if any? Are there institutional, geographic, or investment factors that can explain some of the differences in levels of poverty as well as changes over time in various counties and regions? The following chapters will use the poverty rates discussed here to examine these questions.

23

CHAPTER 3 THE RELATIONSHIP BETWEEN ECONOMIC FREEDOM AND GLOBAL POVERTY In aggregate, the world became more economically free over the period 1980-2005 (Gwartney and Lawson 2009). As the previous chapter discussed, the world also became much less poor over the same period. The extreme poverty rate fell from 58.3 percent of the world’s developing population in 1980 to 25.1 by 2005. Over the same period the moderate poverty rate fell from 75.3 to 45.6 percent. Shleifer (2009) argues that these two trends are related. Based upon recent scholarship from modern growth theory he concludes that increases in economic freedom played an important role in these reductions of poverty rates. Preliminary evidence is supportive of this view. The rating of economic freedom, as measured by the Economic Freedom of the World index, in China and India increased from 4.41 and 4.42 to 6.07 and 6.5, respectively, over the period. These two countries also had significant reductions in poverty that coincided with movement toward greater levels of economic freedom. While this evidence is consistent with the hypothesis that economic freedom is associated with reductions in poverty, it is hardly convincing. Therefore, the goal of this chapter is to use empirical techniques and implications of modern economic growth theory to examine the relationship between economic freedom and poverty. More specifically, did the increase in economic freedom over the period 1980-2005 help to reduce the incidence of worldwide poverty? In presenting this research, section 3.1 discusses how the current economic growth literature is supportive of the view that economic freedom can help to reduce poverty rates. Section 3.2 introduces the empirical framework used in the analysis as well as the variables. The growth literature suggests that factors other than economic freedom have an impact on economic growth and development. Therefore variables representative of these factors are included along with the poverty and economic freedom measures. Section 3.3 contains the empirical results while section 3.4 investigates whether the results are robust to different specifications. The final section concludes.

3.1 Economic Freedom, Growth, and Global Poverty

24

Before discussing how economic freedom relates to global poverty, it is important to consider the relationship between poverty and growth. If reductions in poverty occur in a particular country they will primarily be the result of sustained economic growth (Bhagwati and Srinivasan 2002).

Sustained economic growth creates opportunities for people at all income levels.

Individuals in more advantageous positions may have larger gains, but in general the gains from trade are available to all. The poverty data from the previous chapter is supportive of this view. The regions of the world with the highest sustained growth rates, primarily parts of Asia, also had the largest reductions in poverty. Sub-Saharan Africa, however, had stagnant growth rates during 19802005 and exhibited very little, if any, reductions in poverty. This does not imply that all economic growth leads to poverty reduction, but does suggest that reductions in poverty that do occur will take place in countries with sustained economic growth.

3.1.1 Economic Freedom and Growth

If economic growth is a necessary condition for reductions in poverty then the implications of modern growth theory are relevant to global poverty. The recent scholarship on economic growth has highlighted institutions as an important, if not the most important, factor for long run growth. Specifically, the literature has highlighted the significance of economic institutions. Economic institutions are the rules of the game for economic interactions.

This literature

indicates that institutions, more compatible with economic freedom, lead to higher income levels and growth rates. Institutions more compatible with economic freedom are those that respect private property rights and the rule of law, enforce contracts in an even handed manner, promote monetary stability, have low trade barriers, and have limits on government regulation and taxation.

In short, economically free institutions help to reduce the uncertainty of market

transactions. But how does economic freedom drive economic growth? There are two primary channels through which economic freedom impacts growth: a direct channel as a result of the growth environment and an indirect channel through investment. The first channel was explored by Dawson in 1998.

He demonstrated that the total factor productivity (TFP) term of the

25

neoclassical growth model captures the direct impact of the growth environment. Parente and Prescott (2000) used a different approach by describing the TFP term as being representative of the types of barriers facing businesses. Despite the slightly differing emphasis, these authors are both capturing the impact of the growth environment. To illustrate this, equation (1) contains a simplified form of the Parente and Prescott cobbdouglas production function.6 -

(1) yt = A(πi)ktαnt1 α Equation (1) contains the usual variables: y is output, k is the capital stock, n is labor (hours worked), and α is the capital share of production. A(πi) is the TFP term and is a function of barriers, π, faced by businesses in country, i and is defined as follows. -

(2) A(πi) = W(1+πk,i) α(1+πn,i)

-(1-α)

Here πk, and πn represent barriers associated with using capital and labor in country i and W represents the level of global technical knowledge. Despite the simplicity of the expression, it illustrates the impact of economic freedom on the productive ability of a particular country. The negative exponent on the terms representing the barriers illustrates the relationship between TFP and economic freedom. Businesses in countries with more economic freedom will be able to utilize the latest productivity enhancing technology in the production process. These countries will have lower barriers and consequently a higher TFP. The opposite is true for countries with less economic freedom. They are characterized by higher barriers and lower TFP. The second channel highlights the indirect impact of economic freedom on growth through investment. Again, Dawson (1998) explored this relationship and found supportive empirical evidence. In the context of the neoclassical growth model this direct relationship between economic freedom and investment can be seen with the basic investment equation. In the neoclassical model, consumers use output to purchase consumption goods or invest. With s representing the share of output directed toward investment we have the following expression. (3) I = sy 6

As a simplification, the intangible capital term used in the Parente and Prescott production function has been dropped and the growth term for technical knowledge has been absorbed into A(πi).

26

I and y are per capita investment and output, respectively. Substituting the production function of equation (1) into this investment equation yields, -

(4) I = sA(πi)ktαnt1 α This equation illustrates the direct impact of economic freedom on investment as it operates through the TFP term, A(πi). Countries with more economic freedom are those that have clearly defined property rights, the rule of law, and even handed enforcement of contracts. These factors reduce the uncertainty and transaction costs associated with investment. Therefore, one would expect that countries with more economic freedom will, on average, have higher levels of investment and consequently higher growth rates and income levels. Indeed, the empirical evidence suggests that economic freedom affects growth and income levels through these two channels. Gwartney, Lawson, and Holcombe (1999) found empirical evidence of the direct impact of economic freedom on growth. Countries with increases in economic freedom had correspondingly higher rates of economic growth. In addition, Dawson (2003) found that the causation was from economic freedom to growth. Along with Dawson’s work discussed previously, Gwartney, Holcombe, and Lawson (2006) found that a higher level of economic freedom also corresponded to higher levels of investment and higher levels of the productivity of private investment. In addition to the impact of economic freedom on growth and investment, Cole (2003) found evidence to suggest that convergence, a prediction of the neoclassical growth model, occurs after accounting for the level of economic freedom. This result, called conditional convergence, is similar to that of Barro (1997).7 In addition to higher growth rates, countries with consistently high levels of economic freedom also have higher income levels and higher quality of life measures (Gwartney and Lawson 2009).

3.1.2 Economic Freedom and Poverty

7

Barro used a measure of democracy rather than economic freedom in his work. At the time, measures of economic freedom were unavailable.

27

Thus far, it has been argued that economic growth is a necessary condition for reductions in poverty. In addition, modern growth theory stresses the importance of economic freedom for promoting growth.

Taken together this suggests that economic freedom is important for

achieving reductions in poverty. Shleifer (2009) made a similar argument after examining various quality of life indicators, including poverty rates.

The mechanism through which

economic freedom influences growth also applies to poverty. More economic freedom implies increased opportunities for the poor to participate in this process. In addition, barriers that exist in less economically free countries stifle the entrepreneurial aspirations of the poor. Increased economic freedom reduces these barriers, thereby unleashing the entrepreneurial spirit of the poor (Vargas Llosa 2008). Norton and Gwartney (2008) found that increases in economic freedom correspond to reductions in poverty, but that analysis relied on measures of poverty prior to major revisions of the PPP conversion ratios and changes to its methodology. Others are less sanguine about how economic freedom may impact the poor. Increased economic growth may only benefit those in the highest income strata (Galor and Zeira 1993) and in spite of improved institutions (i.e. more economic freedom) the poor may still be left behind (Ray 2010). Therefore, to thoroughly examine the impact of economic freedom on poverty an empirical investigation is necessary.

3.2 Cross-Country Empirical Framework To maintain consistency with the growth literature, this analysis relies upon the empirical framework used in recent work regarding economic freedom and growth. The workhorse of this framework is an estimation equation derived from a linear approximation of the equation describing how an economy transitions to the steady state level of output in the neoclassical growth model. A general form of this equation is listed below. (5) y = α + βEFWave + δZ + u Typically y is the natural logarithm of per capita income or its annual growth rate, α is the intercept term, β is the coefficient of interest, EFW is the average level of economic freedom over a given period as measured by the Economic Freedom of the World index, Z is a matrix of control variables, and u is the white noise error term. More recent work has focused on how

28

changes in economic freedom affect income and growth rates. Hence, a modified version of equation (5) is typically used. (6) Δy = α + βΔEFW + δZ + u For this analysis, per capita income is replaced by a country’s poverty rate and is expressed by the pair of equations below. (7) poverty = α + βEFWave + δZ + u (8) Δpoverty = α + βΔEFW + δZ + u The dependant variables are the extreme and moderate poverty rates produced by the World Bank and described in the previous chapter. The remaining variables are the same as those discussed in equations (5) and (6). Throughout this analysis the focus will be on the sign and significance of β.

A positive and significant β indicates that, on average, higher levels

(increases) of economic freedom correspond to lower levels (decreases) of poverty rates. In the current growth literature the proximate causes of economic growth are grouped into three general categories. The first is the view that inputs to production (i.e. physical capital and human capital) are the key to the growth process. The second view is that geographical factors are what encourage or discourage growth. The third view stresses the importance of institutions such as property rights, the rule of law, and even handed enforcement of contracts as fundamental for economic growth. The implications of these three theories are incorporated into this research. While the focus is on institutions, and hence falls into the third category, the other two theories are important and may have an impact on poverty. Therefore, it will be necessary to control for the impact of inputs and geography on poverty when possible. The extreme and moderate poverty rates from the World Bank, World Development Indicators are the measures of poverty used in this analysis. The data has been regularized to five-year intervals over the period 1980-2005 using the methods described in the previous chapter. These poverty rates represent the percentage of a country’s population that live on $1.25 and $2 per day or less, respectively, in 2005 international dollars. There are 5 broad measures of institutions used in this analysis. The first and primary measure is the Economic Freedom of the World (EFW) index published annually by the Fraser Institute. It is a measure of a country’s economic institutions and comprises five areas: the size 29

of the central government, security of property rights, stability of the currency, openness to trade, and regulation of credit, labor, and business. It covers 141 countries and is comprised of 42 different subcomponents from external sources. The version of the index used here is chainlinked and spans the period 1980-2005 at five-year intervals covering at least 102 countries. The second and third institutional variables are the polity and constraints on the executive measure from the Polity IV project. These variables are included to control for any impact political institutions may have on poverty.8 While Gwartney, Lawson, and Holcombe (1999) found that political institutions did not have a significant impact on growth after controlling for economic freedom, these two variables have been frequently used in the growth literature and are, therefore, included. The fourth and fifth institutional variables, again, are proxies for political institutions. They are measures of political rights and civil liberties published annually by Freedom House. These measures are used often in the growth literature and are, therefore, included. While both the political rights and civil liberties variables were used in the analysis, only the results for the political rights variable are included. These two measures are highly correlated (a correlation during 1980-2005 of 0.93) which makes the inclusion of both variables redundant.9

The

correlation between these variables and the other institutional variables is shown in table 3.1.10

Table 3.1: Correlation between economic and political institution variables, 1980-2005

EFW Polity Constraints on the executive Political rights Civil liberties

EFW 1.00 0.47 0.51 0.54 0.59

Polity

Constraints on the executive

Political rights

Civil liberties

1.00 0.92 0.90 0.86

1.00 0.86 0.82

1.00 0.93

1.00

8

Recent research indicates that there is a relationship between economic freedom and political institutions. This relationship is examined in greater detail in the chapter 4. The analysis here follows the current growth literature and controls for a direct impact of political institutions on poverty only. 9 A similar argument could be made for the other political variables. As they are all highly correlated one could easily use only one of the variables in the analysis. The literature has relied upon the Polity IV index, the constraints on the executive measure, and the Freedom House Political Rights measure. All three are included in this analysis to make the results more comparable. Different political measures are never used together in the same regression, however. 10 Correlations for each five-year period were run in addition to the correlations of the period as a whole. There was very little difference between the two, which is why five-year period correlations are not included. However, it is worth mentioning that the correlations between the variables increased very slightly over time.

30

Investment is a key element in the theory where inputs to the production process are the source of economic growth. The lack of investment in developing countries is a significant rational behind the provision of foreign aid. However, a measure of foreign aid is not included in this chapter. This analysis was conducted with and without a measure of foreign aid with very little difference in the results. Therefore, it was decided to exclude the level of foreign aid for two reasons. First, chapter five will focus exclusively on the impact of foreign aid on poverty. Second, when poverty is regressed on the level of foreign aid, the coefficient is positive and significant.

The interpretation of this coefficient is difficult due to endogeneity.

One

interpretation is that higher levels of foreign aid correspond to higher levels of poverty. However, a different interpretation is that poorer countries receive more foreign aid precisely because they are poor. Therefore, the task of understanding the relationship between foreign aid and poverty is left for chapter five.11 A listing of all the variables, along with a brief description is shown in table 3.2. Appendix B contains a much more thorough description of all the variables.

Table 3.2: Description and source of regression variables Variable (Category) Extreme and moderate poverty rate

Description The percentage of a country’s population that lives on $1.25 and $2 per day or less, respectively, in 2005 international dollars.

Source World Bank

Economic freedom of the world (Economic Institutions)

An index that measures the degree to which a country’s institutions adhere to the free market ideal. It is measured on a scale from 0-10 with 10 indicating “most free”. There are 5 areas of the index made up of 42 different subcomponents.

Gwartney, James D. and Robert Lawson. “Economic Freedom of the World Annual Report”, (2009).

Polity (Political Institutions)

A measure that captures the autocratic and democratic, or lack thereof, characteristics of a country’s polity. It ranges from -10 to 10 with the low value corresponding to complete autocracy and the high corresponding to representative democracy.

Polity IV Project

11

The foreign aid and growth literature has found no overall impact of foreign aid on growth. In addition, the chapter on foreign aid and poverty in this research found no statistically significant relationship between foreign aid and poverty after controlling for various factors. This suggests that its exclusion here will have little affect on the results.

31

Table 3.2 – continued Variable (Category) Constraints on the executive (Political Institutions)

Description This is a subcomponent of the Polity measure and reflects the degree to which the power of a country’s chief executive is restrained. This measure spans from 1-7 with 1 indicating unchecked executive power and 7 indicating executive parity.

Source Polity IV Project

Political rights (Political Institutions)

A measure of the degree of a country’s political freedom. It ranges from 1-7 with 7 indicating the largest degree of political freedom. (Note: This 1-7 range has been reversed in order to remain consistent with the other institutional measures.)

Freedom House

Civil liberties (Political Institutions)

A measure of the degree to which a country’s citizens civil liberties are protected. This measure ranges from 1-7 with 7 indicating the highest protection of civil liberties. (Note: This 1-7 range has been reversed in order to remain consistent with the other institutional measures.)

Freedom House

Percentage of population within 100km of coast (Geography)

This captures the percentage of a country’s population that lives within 100km of the coastline.

Sachs, Jeffrey D. “Tropical Underdevelopment”, (2001).

Tropics (Geography)

This measure is the percentage of a country’s land area that resides in the tropics (south of the Tropic of Cancer and north of the Tropic of Capricorn).

Sachs, Jeffrey D. “Tropical Underdevelopment”, (2001).

Air distance (Geography)

A country’s minimum air distance in kilometers to one of the three major world markets: New York, Amsterdam, or Tokyo.

Sachs, Jeffrey D. “Tropical Underdevelopment”, (2001).

3.3 Results If economic freedom corresponds to lower levels of poverty than one would expect to observe this relationship in a basic OLS regression. Table 3.3 contains OLS regression results of the extreme poverty rate in 2005 regressed on economic and political institutions along with geographic control variables. Indeed, column one indicates that there is a strong correlation between the average level of economic freedom during 1980-2005 and the level of extreme poverty in 2005. A one unit increase in the average level of EFW over the period corresponds to

32

an 11.41 percentage point lower extreme poverty rate in 2005. This result is significant at the one percent level. The three geography variables in the regression are significant and have the sign that theory would predict. The coastal population variable is the percentage of a country’s population that lives within 100km of the coastline and has a negative coefficient in column one, confirming Sach’s geography hypothesis. Proximity to navigable waterways lowers transaction costs and increases the ability of people to access international markets. Therefore, as this variable increases one would expect increased access to global markets and lower levels of poverty. The tropical location variable is a measure of a countries land area located in the tropical zone.

These areas have much higher incidences of malaria and other debilitating

diseases resulting in lower productivity and higher mortality rates for children under five. Accordingly, one would expect this to have a negative impact on economic growth and correspond to a higher incidence of poverty in these areas and hence have a positive coefficient in this regression. Lastly, the variable representing the shortest distance to major world markets, measured in thousands of kilometers, also has the expected sign. As the distance between a country and major areas of international trade increase one would expect this to be associated with a higher level of poverty. Again, this corresponds to a positive regression coefficient. These results are all supportive of Sach’s geography hypothesis. Unfavorable geographic and locational factors are an impediment to growth and are associated with higher poverty rates. Columns two through four of table 3.3 tell a similar story. These regressions examine the impact of political institutions on the extreme poverty rate in 2005. Each measure of political institutions is significant at the one percent level. However, the magnitudes of the coefficients are much less than the corresponding economic freedom variable in column one, even after adjusting for differences in scale. A one unit increase in the average of the polity score over the period corresponds to a 1.07 percentage point lower poverty rate in 2005. However, the scale of the polity measure is double that of the EFW index implying a 2.14 percentage point lower poverty rate in 2005. The executive constraints and political rights variables are measured on a seven point scale and have coefficients of -4.25 and -4.37, respectively corresponding to roughly -3.0 after adjusting the scale.

This is close to an order of magnitude lower than the

corresponding impact of the economic freedom measure. In addition, the explanatory power of the regressions containing the measures of political institutions (r-square between 0.49-0.52) is slightly less than the economic freedom regression that has an r-square value of 0.55. The

33

coefficients on the geography variables in columns two through four all have similar significance, sign, and magnitude as the first regression.

Table 3.3: Determinants of the 2005 extreme poverty rate Independent variable EFW, average 1980-2005a

(1) -11.41 *** (2.48)

Polity IV, average 1980-2005

(2)

Dependent variable: Extreme poverty rate, 2005 (3) (4) (5) (6) (7) -10.61 *** -10.10 *** -9.88 *** (2.46) (2.48) (2.55) -0.79 ** (0.35)

-1.07 *** (0.33)

Executive constraints, average 1980-2005

-4.25 *** (1.06)

Political rights, average 1980-2005

-2.88 *** (1.14) -4.37 *** (0.98)

-3.01 ** (1.21)

Coastal population (% within 100km)

-16.77 *** -21.88 *** -20.12 *** -18.19 *** -14.50 *** -13.88 ** (6.28) (4.36) (4.21) (5.79) (5.70) (4.20)

Tropical location (% area in tropics)

19.99 *** 23.24 *** 20.02 *** 21.15 *** 20.89 *** 18.85 *** 19.90 *** (4.33) (3.67) (3.51) (4.37) (4.13) (4.20) (3.50)

Distance to major markets b

2.23 ** (0.92)

Intercept

70.04 *** 4.18 (3.07) (12.05)

-13.88 ** (5.85)

3.00 *** 2.99 *** 3.24 *** 2.43 *** 2.52 *** 2.48 *** (0.77) (0.77) (0.84) (0.86) (0.82) (0.75) 21.93 *** 18.63 *** 63.54 *** 72.41 *** 70.80 *** (4.03) (4.92) (12.47) (11.64) (11.76)

R2 (adjusted)

0.55

0.49

0.52

0.51

0.57

0.58

0.57

Number of countries

76

123

123

127

76

76

76

Notes: Countries with at least five of the six observations over the period 1980-2005 were included. b The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors are listed in parenthesis. a

Columns five through seven of table 3.3 list the regression results for the economic freedom variable and each of the political measures. Both the economic freedom variable and the political institutions variables are significant with coefficients of a similar magnitude to those in the first four regressions. The coefficient on the economic freedom variable ranges from -9.88 to -10.61 while those of the political variables range from -0.79 to -3.01. Again, the coefficient for the average level of economic freedom is nearly an order of magnitude larger than the corresponding political variables. The goodness of fit of these last three equations indicates that economic freedom, political institutions, and geographic measures during 1980-2005 explain

34

roughly 57 to 58 percent of the cross-country variation in the extreme poverty rate in 2005. It should be noted that in this table and in subsequent tables that contain the average level of economic freedom, only countries that contained EFW ratings for 5 or more of the 6 years over the period 1980-2005 were included.

Table 3.4: Determinants of 2005 moderate poverty rate Independent variable EFW, average 1980-2005a

(1) -16.48 *** (3.03)

(2)

Dependent variable: Moderate poverty rate, 2005 (3) (4) (5) (6) (7) -15.45 *** -14.78 *** -14.46 *** (3.13) (3.15) (3.25)

-1.49 *** (0.46)

Polity IV, average 1980-2005

-1.02 ** (0.48) -5.72 *** (1.42)

Executive constraints, average 1980-2005

-3.73 ** (1.59) -3.99 ** (1.63)

-6.31 *** (1.32)

Political rights, average 1980-2005

-14.68 ** (7.41)

-14.61 * (7.50)

25.25 *** 30.77 *** 26.43 *** 27.93 *** 26.41 *** (5.78) (4.47) (4.28) (4.19) (5.87)

23.77 *** (5.57)

25.12 *** (5.59)

Distance to major markets b

2.84 ** (1.22)

3.23 *** (1.12)

3.19 *** (1.05)

Intercept

105.95 *** 12.43 *** 36.42 *** 33.19 *** 97.55 *** 109.01 *** 106.95 *** (15.41) (16.62) (4.26) (6.85) (15.54) (5.60) (15.62)

Coastal population (% within 100km)

-18.44 ** (7.95)

Tropical location (% area in tropics)

R2 (adjusted) Number of countries

-28.79 *** -26.63 *** -23.44 *** -15.51 ** (5.62) (5.37) (5.34) (7.47)

3.78 *** 3.76 *** 4.11 *** 3.11 *** (0.94) (0.92) (1.09) (0.90)

0.56

0.50

0.53

0.53

0.58

0.59

0.59

76

123

123

127

76

76

76

Notes: Countries with at least five of the six observations over the period 1980-2005 were included. b The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors are listed in parenthesis. a

Table 3.4 contains identical regressions to those of table 3.3 except that the dependent variable is the moderate poverty rate in 2005. Unsurprisingly, the results are very similar. The measures of the average level of economic freedom and political institutions during 1980-2005 are significant and have the expected sign. A one unit increase in the average level of economic freedom over the period corresponds to a 14.46 to 16.48 percentage point lower moderate

35

poverty rate in 2005, depending upon the regression.

Again, the coefficients of the three

geography variables are significant, have the expected sign, and are similar to those of table 3.3. Tables 3.3 and 3.4 taken together lend support to the ideas explored in section 3.1; that higher levels of economic freedom correspond to lower levels of poverty. These regressions explain 50 percent or more of the cross-country variation in the extreme and moderate poverty rate in 2005. These results, however, provide no information about poverty reductions nor do they indicate if increased economic freedom precedes reductions in poverty. The following tables examine these aspects more fully. Table 3.5 presents OLS regression results for reductions in the extreme poverty rate during 1980-2005 regressed on the same economic, political, and geographic variables in the two previous tables. In addition, the extreme poverty rate in 1980 is included as an independent variable in order to control for the level of extreme poverty in each country at the beginning of the period. It should be noted that the dependent variable represents the percentage point reduction in poverty (i.e. a positive value indicates there was a fall in poverty over the period), therefore positive regression coefficients correspond to percentage point reductions in poverty. The coefficient on the extreme poverty rate at the beginning of the period is significant in all of the regressions of table 3.5 and ranges from 0.31 to 0.40. While small, this result indicates a type of conditional poverty convergence that is consistent with the corresponding conditional convergence theory in the growth literature.12 In column one, a ten percentage point higher level of poverty at the beginning of the period corresponds to a 3.8 percentage point reduction in extreme poverty during 1980-2005 after controlling for institutions and geography. Turning to the institutional variables one finds that countries with higher levels of economic freedom and political measures exhibited larger reductions in poverty during 1980-2005. These results are significant at the 5 percent level or higher in the first four columns and at the ten percent level or higher in the last three columns. Column one of table 3.5 indicates that a one unit higher average economic freedom rating over the period corresponds to a 3.78 percentage point reduction in the extreme poverty rate after controlling for geographic factors. The political variables tell a similar story, but again with smaller coefficients. Similar to the previous tables, 12

Conditional convergence, as used in the growth literature, is the idea that after accounting for various factors, poor countries tend to grow faster then rich countries. As a result, per capita income levels converge over time. Here we see that after accounting for economic freedom, poorer countries had larger reductions than wealthier – but still developing – countries.

36

the last three columns (five through seven) are regressions that include the average economic freedom rating and the average political rating. The results from these three equations are indicative of a pattern that exists in the literature and becomes more pronounced in the remaining results.

That is, the decreasing significance of political institutions after controlling for

economic freedom. While all of the measures for political institutions are significant at the five percent level or higher in columns two through four, this is no longer the case in the last three columns when economic freedom is included. The variables remain statistically significant, but at a lower level. Lastly, the geographic coefficients have the expected sign, however, the coastal population variable is no longer significant in any of the regressions. This result should not be considered evidence against the geography hypothesis. Rather it is the result of the dependent variable being a change in the poverty rate as opposed to the level. The geographic variables do not change over time and therefore one would expect them to have less impact on reductions in poverty. For example, these regressions illustrate that a country whose land area lies outside of the tropical zone will on average have roughly a 10 to 11 percentage point larger reduction in extreme poverty over the period as compared to a country whose land area lies entirely in the tropics. Obviously, countries cannot change their location. Rather, these results illustrate the geographic barriers faced by poor countries in various parts of the world. Table 3.6 is identical to table 3.5 except that the dependent variable is now the percentage point reduction in moderate poverty during 1980-2005. Again, we observe similar results. The average level of economic freedom over the period is significant at the five percent level and has a coefficient that ranges from 5.28 to 5.70 in the respective regressions.

The statistical

significance of the political variables is slightly reduced when economic freedom is included (columns five through seven), with the average level of political rights becoming insignificant in column seven. The geographic variables also display similar trends as those found in table 3.5. The coastal population variable is generally less significant than the other two geography variables as it is only significant in columns two through four. The tropical location variable remains very significant and the distance to major markets variable is less so.

37

Table 3.5: Impact of the average economic, political, and geographic factors on reductions of the extreme poverty rate, 1980-2005 Independent variable Extreme poverty rate, 1980 EFW, average 1980-2005a

Dependent variable: Reduction in extreme poverty, poverty, 1980-2005 (1) (2) (3) (4) (5) (6) (7) 0.37 *** 0.31 *** 0.33 *** 0.32 *** 0.39 *** 0.40 *** 0.39 *** (0.07) (0.06) (0.06) (0.06) (0.07) (0.07) (0.07) 3.78 ** (1.92)

3.70 ** (1.82) 0.42 ** (0.20)

Polity IV, average 1980-2005

1.59 ** (0.68) 1.43 * (0.75)

1.62 ** (0.70)

Political rights, average 1980-2005 4.82 (3.52)

3.40 * (1.87)

0.41 ** (0.21) 2.05 *** (0.74)

Executive constraints, average 1980-2005

3.48 ** (1.76)

3.77 (3.45)

4.12 (3.56)

3.86 (3.81)

3.41 (3.80)

3.65 (3.86)

Coastal population (% within 100km)

5.31 (4.00)

Tropical location (% area in tropics)

-10.35 *** -10.81 *** -10.09 *** -10.09 *** -11.47 *** -10.55 *** -10.88 *** (3.23) (3.07) (3.13) (3.36) (3.22) (3.31) (3.24)

Distance to major markets b

-1.31 * (0.69)

-1.80 *** -1.97 *** -1.86 *** -1.57 ** (0.62) (0.63) (0.61) (0.71)

-1.67 ** (0.73)

-1.58 ** (0.70)

Intercept

-11.38 (9.24)

14.82 *** 7.12 ** (4.42) (3.61)

-9.11 (8.84)

-14.08 (8.80)

-12.95 (8.69)

8.63 ** (4.11)

R2 (adjusted)

0.39

0.31

0.34

0.32

0.41

0.42

0.40

Number of countries

71

89

89

92

71

71

71

Notes: Countries with at least five of the six observations over the period 1980-2005 were included. b The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors are listed in parenthesis. a

The previous tables indicate that, on average, higher levels of economic freedom are associated with both lower levels of poverty and reductions over time. While these results are suggestive that economic freedom can have an impact on poverty, the analysis up to this point has ignored the impact of a change in economic freedom on poverty. This is the focus of the remaining regression results. Countries with less economic freedom cannot instantly transform into countries with more economic freedom. Increases in economic freedom take place over time. Therefore, understanding the impact of these changes is arguably more important. This can be thought of as understanding the transition path, as opposed to the steady state.

38

Table 3.6: Impact of the average economic, political, and geographic factors on reductions of the moderate poverty rate, 1980-2005 Independent variable Moderate poverty rate, 1980 EFW, average 1980-2005a

Dependent variable: Reduction in moderate poverty, poverty, 1980-2005 (1) (2) (3) (4) (5) (6) (7) 0.26 *** 0.21 *** 0.22 *** 0.22 *** 0.28 *** 0.28 *** 0.28 *** (0.06) (0.05) (0.05) (0.05) (0.07) (0.07) (0.07) 5.70 ** (2.38)

Polity IV, average 1980-2005

5.57 ** (2.24) 0.42 * (0.25)

Executive constraints, average 1980-2005

5.33 ** (2.19)

5.28 ** (2.28)

0.46 * (0.27) 2.06 ** (0.86)

Political rights, average 1980-2005

1.71 ** (0.87) 1.67 * (0.89)

1.52 (0.95)

Coastal population (% within 100km)

5.33 (5.12)

Tropical location (% area in tropics)

-11.12 *** -12.55 *** -11.79 *** -11.83 *** -12.32 *** -11.29 *** -11.67 *** (3.32) (3.50) (3.34) (3.03) (3.39) (3.17) (3.03)

Distance to major markets b

-1.20 (0.81)

-1.31 ** (0.65)

Intercept

-22.03 * (12.50)

8.29 ** (3.98)

7.15 * (3.84)

7.91 ** (4.04)

3.61 (4.98)

3.18 (4.98)

3.52 (5.12)

-1.38 ** (0.64)

-1.50 * (0.81)

-1.59 * (0.83)

-1.50 * (0.81)

12.62 *** 4.80 (4.49) (4.53)

5.98 (4.73)

-19.26 (11.80)

-24.78 ** (12.26)

-23.73 ** (12.07)

-1.48 ** (0.66)

R2 (adjusted)

0.29

0.24

0.27

0.25

0.31

0.32

0.31

Number of countries

71

89

89

92

71

71

71

Notes: Countries with at least five of the six observations over the period 1980-2005 were included. b The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors are listed in parenthesis. a

In the regression analysis below the changes in both economic freedom and the political measures are broken up into several time periods. The rational for this is that a change to economic and political institutions affects economic decisions with a delay. This is largely due to expectations, credibility, and the transmission of information. For example, businesses in many poor countries often go through an onerous bureaucratic and regulatory process to obtain permission from the government to operate. This costly process shapes the expectations of business people and entrepreneurs, discouraging the creation of businesses and forcing some underground. A new regulatory and approval process that makes it easier for businesses affects people’s expectations over time. In addition, a new process for businesses has to be credible in order to have any affect on behavior. This new credibility is not immediate. Rather, the new

39

system becomes credible over time if the process is consistently implemented.

Lastly,

information about changes to the approval process takes time to reach all those who will be affected. The net affect of these factors suggests that a change in economic freedom or political factors will have an impact on economic decisions over an extended period of time. Table 3.7 presents OLS regression results for the percentage point reduction in extreme poverty during 1980-2005. The independent variables include the same geographic variables from the previous tables, but the economic freedom and political variables now represent changes over time. The change is split into two periods, an earlier period, 1980-1995 and a later period, 1995-2005. These variables are calculated by subtracting the later period from the earlier period, implying that an increase in the measure over the period has a positive value. In addition, the value of the economic freedom and political variables in 1980 is included to control for each country’s starting point. A change in economic freedom over the earlier period, 1980-1995, is significant at the five percent level or higher in all the regressions of table 3.7 in which it is included, but is not significant in the later period. This result is supportive of the view that a change in economic freedom impacts economic decisions over an extended period of time as changes in the later period have yet to have an impact. The coefficients on the change in the EFW variable during 1980-1995 range from 4.29 to 4.95 implying, ceteris paribus, that a one unit increase in economic freedom corresponds to a more than 4.2 percentage point reduction in extreme poverty during 1980-2005. The coefficient on the change of economic freedom variable in the latter period, 1995-2005, is small and slightly negative, however, it is not statistically significant. Interestingly, the coefficients for the change in political variables are not significant in any of the regressions, even the equations that do not include economic freedom, columns two through four. This result is fairly consistent throughout the remaining tables: changes in the various political measures appear to have very little statistical impact on changes in poverty over time. The level of the political measures in 1980 appears to have a significant affect on reductions in extreme poverty. The initial level of each political measure is significant at the ten percent level or higher in all but the last regression. Both the extreme poverty rate in 1980 and the geographic variables have similar signs, significance, and magnitudes as compared to previous tables, suggesting that the regression results remain consistent across tables.

40

Table 3.7: The impact of changes in economic, political, and geographic factors on reductions of the extreme poverty rate, 1980-2005 Independent variable Extreme poverty rate, 1980 EFW rating, 1980

Dependent variable: Reduction in extreme poverty, poverty, 1980-2005 (1) (2) (3) (4) (5) (6) (7) 0.41 *** 0.31 *** 0.31 *** 0.30 *** 0.44 *** 0.41 *** 0.43 *** (0.08) (0.06) (0.06) (0.06) (0.09) (0.08) (0.09) 3.16 (1.94)

Polity IV rating, 1980

3.29 * (1.85)

3.34 (2.09)

0.46 ** (0.22)

0.41 * (0.22)

Executive constraints rating, 1980

2.66 (1.84)

1.42 * (0.73)

1.81 ** (0.77)

Political rights rating, 1980

1.43 * (0.77)

0.88 (0.83)

Change in EFW, 1980-1995

4.95 *** (1.78)

4.29 *** 4.38 *** 4.88 ** (1.68) (1.68) (1.99)

Change in EFW, 1995-2005

-0.52 (2.58)

-1.55 (2.46)

Change in polity IV, 1980-1995

-0.03 (0.20)

0.05 (0.23)

Change in polity IV, 1995-2005

0.08 (0.36)

0.35 (0.36)

-0.60 (2.71)

Change in executive constraints, 1980-1995

0.82 (0.58)

0.93 (0.65)

Change in executive constraints, 1995-2005

-0.09 (0.86)

0.41 (1.02)

-0.53 (2.69)

Change in political rights, 1980-1995

0.37 (0.62)

-0.24 (0.85)

Change in political rights, 1995-2005

-0.26 (1.02)

-0.16 (1.04)

Coastal population (% within 100km)

3.38 (3.75)

Tropical location (% area in tropics)

-9.43 *** -12.08 *** -10.43 *** -10.06 *** -13.35 *** -10.41 *** -10.51 *** (3.02) (3.81) (3.61) (3.34) (3.39) (3.18) (3.55)

Distance to major markets a

-1.64 ** (0.68)

-1.61 *** -1.86 *** -1.68 *** -1.75 *** -1.88 *** -1.57 ** (0.66) (0.72) (0.64) (0.65) (0.63) (0.60)

Intercept

-8.57 (10.78)

16.19 *** 8.64 ** (3.63) (4.56)

5.23 (3.56)

4.54 (3.51)

4.91 (3.69)

9.15 ** (4.13)

2.35 (3.47)

2.29 (3.71)

3.04 (3.67)

-5.02 (9.57)

-9.29 (10.40)

-12.38 (11.61)

R2 (adjusted)

0.44

0.31

0.32

0.30

0.47

0.44

0.44

Number of countries

67

89

89

92

67

67

67

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors are listed in parenthesis.

a

41

Table 3.8: The impact of changes in economic, political, and geographic factors on reductions of the moderate poverty rate, 1980-2005 Independent variable Moderate poverty rate, 1980 EFW rating, 1980

Dependent variable: Reduction in moderate poverty, poverty, 1980-2005 (1) (2) (3) (4) (5) (6) (7) 0.30 *** 0.21 *** 0.21 *** 0.21 *** 0.33 *** 0.31 *** 0.32 *** (0.07) (0.05) (0.05) (0.05) (0.08) (0.08) (0.08) 4.77 ** (2.42)

Polity IV rating, 1980

4.77 ** (2.25)

4.84 * (2.51)

0.57 * (0.29)

0.42 (0.29)

Executive constraints rating, 1980

4.26 * (2.18)

1.70 * (0.91)

1.83 * (0.94)

Political rights rating, 1980

1.53 (0.96)

1.26 (0.99)

Change in EFW, 1980-1995

6.44 *** (2.09)

5.45 *** 5.45 *** 5.95 *** (1.82) (1.83) (2.17)

Change in EFW, 1995-2005

-0.86 (2.88)

-2.42 (2.49)

Change in polity IV, 1980-1995

-0.01 (0.21)

0.14 (0.24)

Change in polity IV, 1995-2005

0.21 (0.37)

0.61 * (0.33)

-1.63 (2.80)

Change in executive constraints, 1980-1995

1.03 (0.68)

1.29 * (0.69)

Change in executive constraints, 1995-2005

0.55 (0.84)

1.47 * (0.83)

-1.31 (2.75)

Change in political rights, 1980-1995

0.51 (0.63)

0.07 (0.82)

Change in political rights, 1995-2005

0.41 (0.91)

0.78 (0.80)

Coastal population (% within 100km)

3.76 (4.70)

Tropical location (% area in tropics)

-10.42 *** -14.05 *** -12.41 *** -12.01 *** -15.42 *** -12.54 *** -12.23 *** (3.19) (4.39) (3.95) (3.63) (3.48) (3.11) (3.68)

Distance to major markets a

-1.55 * (0.80)

-1.13 * (0.63)

Intercept

-18.07 (14.36)

8.55 ** (3.93)

7.53 * (3.99)

8.37 ** (4.09)

2.27 (4.18)

2.50 (4.55)

2.82 (4.40)

-1.26 ** (0.63)

-1.70 ** (0.78)

-1.83 ** (0.82)

-1.56 ** (0.72)

14.09 *** 6.58 (4.47) (4.75)

6.60 (4.66)

-12.92 (12.06)

-19.10 (13.60)

-21.67 (15.25)

-1.42 ** (0.68)

R2 (adjusted)

0.36

0.23

0.23

0.23

0.41

0.37

0.36

Number of countries

67

89

89

92

67

67

67

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors are listed in parenthesis.

a

42

Table 3.8 lists the results of OLS regressions where the dependent variable is the reduction in the moderate poverty rate during 1980-2005. The independent variables are the same as table 3.7 with the exception of the control variable for the level of moderate poverty in 1980. The pattern of the results is the same. The change in economic freedom over the early period, 19801995, is highly significant in all the regressions where it is included and a one unit change in economic freedom implies a 5.45 or more percentage point reduction in moderate poverty over the twenty-five year period. The change in the later period is not significant and the political measures are occasionally significant at only the ten percent level. The pattern emerging from tables 3.7 and 3.8 suggest that changes in economic freedom played a far more significant role in reductions in extreme and moderate poverty over the period 1980-2005 than measures of political institutions. The statistical significance of increases in economic freedom during the earlier time period is consistent with the view that economic freedom affects poverty over time. However, one cannot conclude that increases in economic freedom in an earlier period influence poverty in a later period from these regressions, as the changes in economic freedom were contemporaneous with the reduction in poverty rates. Also, it is possible that the results of the previous tables were driven by the particular time periods examined. The following regressions examine these issues. Tables 3.9 and 3.10 list the regression results for reductions in extreme and moderate poverty during 1990-2005, respectively, regressed on changes in economic and political factors as well as geographic variables. The time periods of the analysis are slightly changed. The dependent variable no longer covers the entire twenty-five year period, instead covering the fifteen years spanning 1990-2005. The economic and political variables span the entire period, 1980-2005, but are split into ten-year increments, with the exception of the final five-year period. The results are largely consistent with the previous two tables. The change in the economic freedom variable in the earlier time period is significant at the 5 percent level or higher throughout the regressions in which it is included. The coefficients in table 3.9, which examine reductions in extreme poverty, range from a low of 4.78 to 5.54, while those of table 3.10, on reductions in moderate poverty, range from 5.92 to 7.40. A change in economic freedom over the middle period, 1990-2000 was only marginally significant in a few of the regressions. Again, the change in political variables for all of the various time periods is not significant in any of the regressions.

43

Table 3.9: The impact of changes in economic, political, and geographic factors on reductions of the extreme poverty rate, 1990-2005 Independent variable Extreme poverty rate, 1990 EFW rating, 1980

Dependent variable: Reduction in extreme poverty, poverty, 1990-2005 (1) (2) (3) (4) (5) (6) (7) 0.38 *** 0.33 *** 0.35 *** 0.34 *** 0.42 *** 0.40 *** 0.43 *** (0.06) (0.06) (0.06) (0.06) (0.07) (0.07) (0.08) 3.46 ** (1.59)

Polity IV rating, 1980

3.41 ** (1.64) 0.27 (0.19)

Executive constraints rating, 1980

3.17 * (1.64)

3.95 ** (1.89)

0.30 (0.24) 1.13 ** (0.56)

Political rights rating, 1980

0.67 (0.64) 1.10 (0.71)

0.83 (0.80)

Change in EFW, 1980-1990

4.96 *** (1.78)

5.00 *** 4.78 ** (1.85) (2.00)

5.54 *** (2.13)

Change in EFW, 1990-2000

2.84 * (1.70)

2.31 (1.72)

2.51 (1.71)

2.98 (1.87)

Change in EFW, 2000-2005

0.69 (2.40)

1.80 (2.95)

1.26 (2.85)

1.48 (3.38)

Change in polity IV, 1980-1990

0.08 (0.20)

0.10 (0.22)

Change in polity IV, 1990-2000

-0.02 (0.26)

-0.28 (0.36)

Change in polity IV, 2000-2005

-0.32 (0.40)

0.05 (0.49)

Change in executive constraints, 1980-1990

0.94 (0.66)

0.57 (0.70)

Change in executive constraints, 1990-2000

0.28 (0.84)

0.03 (1.11)

Change in executive constraints, 2000-2005

-0.73 (0.70)

-0.07 (0.84)

Change in political rights, 1980-1990

0.68 (0.56)

0.26 (0.80)

Change in political rights, 1990-2000

-0.08 (0.63)

-0.99 (0.90)

Change in political rights, 2000-2005

-0.29 (1.07)

0.24 (1.68)

Coastal population (% within 100km)

5.02 (3.12)

6.38 ** (2.93)

6.23 ** (2.97)

5.72 ** (2.91)

3.52 (3.38)

4.28 (3.62)

4.57 (3.03)

Tropical location (% area in tropics)

-6.51 * (3.40)

-3.37 (3.07)

-2.40 (2.88)

-3.07 (2.88)

-7.12 * (3.67)

-6.19 * (3.39)

-7.09 ** (3.46)

Distance to major markets a

-1.08 ** (0.54)

-1.24 *** -1.22 *** -1.30 *** -1.13 ** (0.40) (0.42) (0.56) (0.39)

-1.16 ** (0.59)

-1.03 * (0.54)

44

Table 3.9 – continued Independent variable Intercept

(1) -18.76 ** (8.94)

(2) 1.96 (2.95)

(3) -4.24 (3.97)

(4) -2.47 (3.53)

(5) -16.56 * (8.79)

(6) -19.53 ** (8.93)

(7) -25.20 ** (10.46)

R2 (adjusted)

0.44

0.33

0.34

0.33

0.44

0.41

0.43

Number of countries

67

111

111

115

67

67

67

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors are listed in parenthesis.

a

Table 3.10: The impact of changes in economic, political, and geographic factors on reductions of the moderate poverty rate, 1990-2005 Independent variable Moderate poverty rate, 1990 EFW rating, 1980

Dependent variable: Reduction in moderate poverty, poverty, 1990-2005 (1) (2) (3) (4) (5) (6) (7) 0.30 *** 0.29 *** 0.30 *** 0.30 *** 0.33 *** 0.31 *** 0.36 *** (0.06) (0.06) (0.06) (0.06) (0.07) (0.07) (0.07) 4.80 ** (2.31)

Polity IV rating, 1980

4.70 ** (2.26)

5.54 ** (2.45)

0.41 (0.30)

0.42 (0.28)

Executive constraints rating, 1980

4.14 * (2.22)

1.18 (0.81)

1.61 * (0.83)

Political rights rating, 1980

1.70 * (1.03)

1.29 (1.07)

Change in EFW, 1980-1990

6.71 *** (2.62)

6.47 *** 5.92 ** (2.62) (2.52)

Change in EFW, 1990-2000

3.81 * (2.23)

3.07 (2.07)

2.98 (2.03)

3.95 * (2.31)

Change in EFW, 2000-2005

2.02 (2.74)

2.85 (3.01)

2.25 (2.89)

2.71 (3.66)

Change in polity IV, 1980-1990

0.18 (0.26)

0.12 (0.26)

Change in polity IV, 1990-2000

0.13 (0.37)

-0.13 (0.32)

Change in polity IV, 2000-2005

-0.41 (0.53)

0.13 (0.44)

Change in executive constraints, 1980-1990

1.44 (0.96)

0.96 (0.76)

Change in executive constraints, 1990-2000

1.15 (1.18)

0.89 (0.94)

Change in executive constraints, 2000-2005

-0.36 (0.81)

0.30 (0.82)

45

7.40 *** (2.77)

Table 3.10 – continued Independent variable Change in political rights, 1980-1990

(1)

(2)

(3)

(4) 1.15 (0.81)

(5)

(6)

(7) 0.45 (0.91)

Change in political rights, 1990-2000

0.31 (0.88)

-1.17 (0.79)

Change in political rights, 2000-2005

0.04 (1.21)

0.91 (1.42)

Coastal population (% within 100km)

6.13 (4.22)

10.42 *** 10.30 *** 9.43 *** 4.52 (3.68) (3.58) (3.48) (4.23)

5.75 (4.56)

5.33 (4.20)

Tropical location (% area in tropics)

-6.92 ** (3.28)

-4.68 (3.33)

-7.53 * (3.88)

-8.02 ** (3.67)

Distance to major markets a

-1.22 * (0.68)

-1.47 *** -1.41 *** -1.58 *** -1.30 * (0.52) (0.51) (0.54) (0.68)

-1.34 * (0.74)

-1.20 * (0.63)

Intercept

-27.97 ** (13.13)

-0.85 (4.34)

-9.89 (6.73)

-7.38 (5.95)

-28.42 ** (13.10)

-37.76 *** (14.85)

-3.54 (3.14)

-4.28 (2.97)

-8.52 ** (4.25)

-24.71 ** (12.24)

R2 (adjusted)

0.33

0.26

0.27

0.26

0.32

0.30

0.34

Number of countries

67

111

111

115

67

67

67

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors are listed in parenthesis.

a

The consistency of these results with those of tables 3.7 and 3.8 suggest that the results are not driven by the length of the time period in which the difference in economic freedom is calculated. Changing how the twenty-five year period was divided had no affect on the pattern of the results. But more importantly, these results illustrate that a change in economic freedom that occurs before the time period in which the reduction in poverty is examined has a positive and statistically significant impact on poverty. These results are supportive of the view that an increase in economic freedom leads to reductions in poverty over an extended period of time. The final regression results, tables 3.11 and 3.12 contain regressions similar to those of tables 3.9 and 3.10. The dependent variable remains the reduction in extreme and moderate poverty, respectively, during 1990-2005, but the earlier time period for the economic and political variables has been dropped. These variables now cover changes during 1990-2000 and 2000-2005. These results are again supportive of the view that a change in economic freedom leads to reductions in poverty with a lag. The change in economic freedom during the earlier period, 1990-2000, is significant at the five percent level in all of the regressions in which it

46

Table 3.11: The impact of changes in economic, political, and geographic factors on reductions of the extreme poverty rate, 1990-2005 Independent variable Extreme poverty rate, 1990 EFW rating, 1990

Dependent variable: Reduction in extreme poverty, poverty, 1990-2005 (1) (2) (3) (4) (5) (6) (7) 0.38 *** 0.33 *** 0.35 *** 0.34 *** 0.41 *** 0.39 *** 0.41 *** (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) (0.07) 4.61 *** (1.36)

Polity IV rating, 1990

4.49 *** 4.52 *** 4.85 *** (1.35) (1.42) (1.55) 0.14 (0.17)

0.18 (0.17)

Executive constraints rating, 1990

0.53 (0.51)

1.08 ** (0.54)

Political rights rating, 1990

0.90 * (0.53)

0.50 (0.63)

Change in EFW, 1990-2000

3.50 ** (1.51)

3.08 ** (1.57)

3.28 ** (1.53)

3.49 ** (1.67)

Change in EFW, 2000-2005

1.63 (1.88)

2.19 (2.20)

1.82 (2.19)

2.21 (2.67)

Change in polity IV, 1990-2000

-0.01 (0.26)

-0.27 (0.32)

Change in polity IV, 2000-2005

-0.32 (0.42)

0.11 (0.46)

Change in executive constraints, 1990-2000

0.35 (0.82)

-0.06 (0.96)

Change in executive constraints, 2000-2005

-0.69 (0.70)

0.11 (0.80)

Change in political rights, 1990-2000

-0.12 (0.61)

-0.75 (0.79)

Change in political rights, 2000-2005

-0.21 (1.09)

0.56 (1.47)

Coastal population (% within 100km)

5.00 * (2.82)

6.63 ** (2.93)

6.40 ** (2.94)

6.33 ** (2.78)

4.01 (2.99)

4.15 (3.18)

4.56 (2.81)

Tropical location (% area in tropics)

-7.32 ** (3.32)

-2.69 (2.77)

-2.49 (2.78)

-3.07 (2.78)

-6.77 ** (2.85)

-6.79 ** (3.00)

-7.46 ** (3.27)

Distance to major markets a

-0.99 ** (0.51)

-1.22 *** -1.22 *** -1.28 *** -1.01 ** (0.51) (0.38) (0.40) (0.39)

-1.07 ** (0.52)

-1.04 ** (0.52)

Intercept

-25.28 *** 0.46 (2.38) (7.57)

-4.51 (3.90)

-2.54 (3.47)

-24.25 *** -26.74 *** -28.67 *** (7.28) (7.57) (8.22)

R2 (adjusted)

0.44

0.35

0.36

0.35

0.45

0.43

0.44

Number of countries

77

112

112

116

77

77

77

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors are listed in parenthesis.

a

47

Table 3.12: The impact of changes in economic, political, and geographic factors on reductions of the moderate poverty rate, 1990-2005 Independent variable Moderate poverty rate, 1990 EFW rating, 1990

Dependent variable: Reduction in moderate poverty, poverty, 1990-2005 (1) (2) (3) (4) (5) (6) (7) 0.30 *** 0.31 *** 0.32 *** 0.32 *** 0.32 *** 0.31 *** 0.34 *** (0.05) (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) 6.38 *** (2.04)

Polity IV rating, 1990

6.23 *** 6.13 *** 6.72 *** (1.96) (2.06) (2.08) 0.18 (0.20)

0.34 (0.25)

Executive constraints rating, 1990

0.81 (0.61)

1.71 ** (0.83)

Political rights rating, 1990

1.58 * (0.83)

0.77 (0.75)

Change in EFW, 1990-2000

4.57 ** (1.93)

4.11 ** (1.88)

4.09 ** (1.84)

4.52 ** (2.01)

Change in EFW, 2000-2005

2.85 (2.21)

3.34 (2.26)

2.71 (2.23)

3.53 (2.88)

Change in polity IV, 1990-2000

0.16 (0.36)

-0.20 (0.29)

Change in polity IV, 2000-2005

-0.38 (0.55)

0.13 (0.44)

Change in executive constraints, 1990-2000

1.36 (1.13)

0.60 (0.83)

Change in executive constraints, 2000-2005

-0.24 (0.80)

0.36 (0.80)

Change in political rights, 1990-2000

0.28 (0.86)

-0.93 (0.77)

Change in political rights, 2000-2005

0.22 (1.23)

1.02 (1.25)

Coastal population (% within 100km)

4.79 (3.91)

11.02 *** 10.75 *** 10.47 *** 3.70 (3.97) (3.52) (3.33) (3.68)

Tropical location (% area in tropics)

-7.68 ** (3.15)

-4.50 (3.07)

Distance to major markets a

-1.15 * (0.64)

-1.48 *** -1.43 *** -1.60 *** -1.18 * (0.64) (0.51) (0.53) (0.52)

Intercept

-35.95 *** -3.03 (3.81) (11.15)

-4.26 (3.03)

-11.06 * (6.66)

-4.91 * (2.95)

-8.23 (5.90)

-7.24 ** (3.12)

4.30 (4.19)

3.95 (4.06)

-7.63 ** (3.26)

-7.89 ** (3.22)

-1.23 * (0.68)

-1.25 ** (0.62)

-34.95 *** -37.97 *** -41.15 *** (10.59) (11.52) (11.74)

R2 (adjusted)

0.35

0.27

0.29

0.28

0.34

0.33

0.36

Number of countries

77

112

112

116

77

77

77

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors are listed in parenthesis.

a

48

is included. The results of the other variables are similar to previous tables. Changes in the political variables are not significant in either time period in all of the regressions and the geographic variables, when significant, have the expected signs.

3.4 Alternate Specifications and Multicollinearity The results thus far are based upon the World Bank extreme and moderate poverty rates. Investigating whether these results hold for an alternative measure of poverty tests the validity and robustness of the results. In addition, various subcomponents of the EFW index contain ordinal data. Therefore, these results may be questioned as the EFW index was used in a cardinal fashion in the previous regressions. These two potential criticisms of the results are examined in turn. The previous chapter discussed the poverty rates generated by Pinkovskiy and Sala-i-Martin (hereafter referred to as PS). The PS poverty rates will be used here as an alternative measure. Before proceeding, it is worth mentioning that there are two general differences between the World Bank and PS poverty measures. The first is that the World Bank poverty rates are based solely on data that comes from household surveys while the PS poverty rates use both surveys and national income accounts. The second difference is that the World Bank defines the extreme and moderate poverty rate as the percentage of a country’s population living on $1.25 and $2 per day or less in 2005 international dollars, respectively. This is in contrast to the PS poverty rates which are $1 and $2 per day or less in 2000 international dollars. It should be noted that while the PS poverty rates are more comprehensive, covering 191 countries as opposed to 128 for the World Bank rates, only the countries with corresponding World Bank poverty rates are included in the regressions so that the results are comparable to tables 3.4-3.4 and 3.7-3.8. Tables D.1-D.4 of appendix D contain regressions identical to those contained in tables 3.3, 3.4, 3.7, and 3.8, respectively, except that the PS poverty rates are used instead of the World Bank rates. Comparing the results of the tables that use the World Bank rates to those that use the PS poverty rates one can see that they are very similar. The results of tables D.1 and D.2 indicate that countries with, on average, higher levels of economic freedom and democracy during 1980-2005 had lower levels of both extreme and moderate poverty in 2005 as measured by the PS rates.

49

Tables D.3 and D.4 follow a similar pattern as the previous two tables. They are identical to tables 3.7 and 3.8 of the previous section except that the dependent variable is now the reduction in the PS extreme and moderate poverty rates during 1980-2005. Again, the results of the regressions using the PS poverty rates are similar to those with the World Bank rates. An increase in economic freedom during 1980-1995 is significant at the five percent level or higher and had a positive impact on reductions in both extreme and moderate poverty during 19802005. The change over the period 1995-2005 is insignificant in all the regressions similar to what was found in the previous section. Tables D.1-D.4 indicate there is little difference in the overall results when either poverty measure is used. This suggests that the implications of the previous section are largely robust to the measure of poverty utilized. The second alternative specification examines whether the ordinal representation of subcomponents of the EFW index have an impact on the regression results. Grether (1974 and 1976) demonstrated that the magnitude of the scale with which ordinal data is ranked could have a significant impact on regression results. While a majority of the subcomponents contained within the EFW index are cardinal values, there are a few that are based upon ordinal rankings. To test whether the ordinal properties of the EFW index are driving the results one simply needs to bifurcate the EFW index and run the regressions as before. If the bifurcated EFW index remains significant with the identical sign as in the previous regressions this indicates that the arbitrary nature of the ordinal data is not driving the results. Implementing this, the EFW index becomes a binary with 1 indicating levels of economic freedom above a particular threshold and 0 below. Bifurcating the EFW index in this fashion removes the possibility that the Euclidian distance between the ordinal data values influences the results. As there is no single correct threshold value, the analysis here uses several thresholds based upon the following criteria. The mean EFW index of the countries included in the regressions of tables 3.3-3.12 was used as the middle threshold. The lower threshold was chosen so that one-third of the countries in the regressions had an EFW index below that value and the upper threshold was chosen so that onethird of the countries had an EFW index above that value. Regressions were run using these three thresholds with the World Bank extreme and moderate poverty rates. Tables D.5 and D.6 of appendix D contain regressions similar to those of table 3.3 and 3.4. The economic freedom variable is now a binary defined according to one of three threshold

50

values. Rather than present results for each regression of table 3.3 with the different EFW index thresholds, table D.5 only contains regressions five through seven of table 3.3 with the three different thresholds. The same holds for table D.5 of appendix D. The first row of table D.5 lists the coefficient and robust standard error for the bifurcated EFW index. The low EFW threshold, a value of 5.2, is used in the first three regressions. In columns four through six the threshold for the EFW index is 5.6, the mean value for the countries contained in the regressions. The last three columns, seven through nine, are those where the highest EFW threshold, 6.0, is used. The coefficients and significance levels of the EFW index in tables D.5 and D.6 illustrate that bifurcating the index at various values has little impact on the results. In every regression the EFW index is significant at the five percent level or higher and negative suggesting that, on average, countries with more economic freedom during 1980-2005 had lower extreme poverty rates in 2005, regardless of the particular threshold used. Table D.6 of appendix D is identical to table D.5 except that the dependent variable is now the moderate poverty rate in 2005. Again the results of table D.6 are similar to table 3.4. Tables D.7 and D.8 of appendix D mimic tables 3.7 and 3.8 of the previous section much like tables D.5 and D.6 do. However, these regressions now deal with the change in economic freedom over a given period and what impact that has on reductions in poverty. In these regressions the change in the EFW index during 1980-1995 is bifurcated according to the threshold criteria explained above. The lower threshold is set at a change of 0.37 during 19801995, the middle threshold is the mean change for the countries contained in the regressions, 0.74, and the upper threshold is a change of 1.0. Again, the results of table D.7 and D.8 are similar to those of table 3.7 and 3.8 in the previous section. The variable representing the change in economic freedom during 1980-1995 indicates that countries with a larger increase in economic freedom exhibited larger reductions in poverty during 1980-2005. This result was significant in all but three of the regressions of table D.7 and D.8. In addition, the coefficient on the change in economic freedom term during 1980-1995 is positive in every regression, indicating that there is little evidence that the ordinal data contained in the EFW index would lead to a change in sign of the regression coefficient.

These results suggest that while

subcomponents of the EFW index are indeed ordinal, the scale of the rankings has little impact on the overall results.

51

Correlation among the independent variables in this analysis varies from high to low depending upon the variables considered. When highly correlated variables are included as independent variables in the same regression there is the possibility of the problem of multicollinearity.

Multicollinearity is an artifact of the matrix operations that are used to

compute the vector of coefficient estimates.

The least squares estimator, "ˆ = ( X #X)$1 X #Y ,

requires computing the inverse of the product of the matrix of independent variables. When two or more independent variables are highly correlated, the matrix of independent variables can get close to becoming singular. The inversion of a matrix involves dividing by the determinant of that matrix, which is zero if the matrix is singular. Obviously, this cannot be done. However, when a matrix is close to being singular and an attempt is made to compute the inverse, the result is that division by small numbers that are close to zero can occur. This presents both a problem and a solution. The problem is that estimates obtained with a matrix that is nearly singular change significantly when certain independent variables are either removed or added to the regression. But, the solution is that this presents a relatively straightforward way to determine if the problem of multicollinearity is affecting the results. Perform a series of regressions with different combinations of the independent variables to see if the coefficient estimates of the variables of interest change significantly. As table 3.1 demonstrated the measures of economic freedom and political institutions are highly correlated. The correlation between the EFW index and the political measures used in the regressions ranges from 0.47 to 0.54. The correlation among the political measures is much higher, typically 0.9 or more.

As multicollinearity could be a problem from these high

correlations, the regression tables 3.3-3.12 never included more than one of the three political measures in each regression. In addition, regressions with just the economic freedom measure, just one of the political measures, and then both are shown in the tables. Throughout all of these various combinations the coefficient and significance for economic freedom and the Sachs geography and location measures remained largely unchanged. The coefficients of the political measures did not change by much, but did become less significant when economic freedom was included in the regressions. This finding is consistent with the literature, however. The correlations listed in table 3.1 are only the correlations between the levels of the economic and political institutions. The regressions in table 3.7-3.12 involve changes in these institutional measures. Multicollinearity could be an issue for these regressions if changes in the

52

various institutional measures are highly correlated. Tables D.9-D.11 of appendix D list the correlation coefficients for all the independent variables used in tables 3.7 and 3.8. While many of the variables are largely uncorrelated there are some variable combinations that are reasonably correlated. Several combinations have a correlation coefficient as high as 0.5. To check for multicollinearity, tables D.12-D.17 contain regressions examining the impact of the independent variables on changes in poverty. These regressions are different combinations of the regressions found in tables 3.7 and 3.8. The dependent variable in tables D.12-D.14 is the reduction in extreme poverty during 1980-2005 while it is the reduction in moderate poverty for tables D.15D.17. The first column of each table begins with very few variables that are largely uncorrelated with one another. Moving from left to right across the columns, variables are slowly added to the regressions. In all of these tables the results are very stable. The change in coefficients from column to column is small and the variable of primary interest in this chapter, the change in economic freedom, remains highly significant throughout. The only change of significance that occurs as more variables are added is the reduced significance of the coastal population variable. However, this pattern is repeated throughout the tables and is more a factor of its weak explanatory power when other variables are included. These tables suggest that while some of the independent variables are correlated to varying degrees, multicollinearity does not appear to be a problem in these regressions.

3.5 Conclusion The results of the previous section are supportive of the following conclusions. First, a higher average level of economic freedom during 1980-2005 was correlated with a lower level of both extreme and moderate poverty in 2005 after controlling for political and geographic factors. Second, an increase in economic freedom in early time periods (1980-1995 and 1980-1990) had a statistically significant impact on reductions in poverty over the entire period, 1980-2005, and the latter fifteen years, 1990-2005. This was a consistent result throughout regressions covering changes in economic freedom and political factors. This result is supportive of the view that increases in economic freedom can help to bring about reductions in poverty over time. It is also supportive of the view that a change in economic freedom operates with a lag.

53

Third, while higher average ratings of the political variables during 1980-2005 were correlated with lower levels of poverty in 2005, there appeared to be no relation between changes in the political variables and reductions in poverty in any of the time periods examined. While these results suggest that reforms of political institutions may have little impact on poverty, they may hide the fact that political institutions play an indirect role by facilitating increases in economic freedom. A more thorough discussion is left for the next chapter. These results, in combination with the previous literature on economic growth, suggest that economic freedom is conducive to both economic growth and reductions in poverty. After controlling for various factors, there appears to be little evidence to suggest that economic freedom is only beneficial to the wealthy or is confounded by poverty traps. The results are consistent with the theory that geographic factors such as, barriers to trade routes and malaria prevalence have a negative impact on poor countries. But, these results suggest that economic freedom can help poor countries overcome these and other barriers in order to achieve reductions in poverty.

54

CHAPTER 4 DEMOCRATIC POLITICAL INSTITUTIONS AND GLOBAL POVERTY The previous chapter found that both the level and change of economic freedom had a significant impact on reductions in the extreme and moderate poverty rates during 1980-2005. However, there was no evidence that changes in political institutions directly reduced either the extreme or moderate poverty rate. This chapter investigates whether political institutions indirectly facilitate reductions in poverty through economic freedom.

Prior research indicates that economic

freedom and political institutions, specifically democracy, are interrelated. If so, then changes in political institutions may impact poverty indirectly through changes in economic freedom. There are two parts to the analysis of this chapter. The first examines whether changes in political institutions, movements toward democracy, have an impact on subsequent changes in economic freedom. The second uses a two-step approach to account for the indirect impact of changes in political institutions on reductions in poverty. With regard to the first topic, the analysis indicates that movements toward democracy over a ten-year period correspond to increases in economic freedom during the subsequent ten-year period. These findings are consistent with Friedman’s view that political freedom is highly supportive of economic freedom (Friedman 1962). They also indicate that political institutions may indirectly facilitate reductions in poverty rates through subsequent changes in economic freedom. The second part of the analysis provides some evidence for this. After accounting for the their impact through economic freedom, movements toward democracy were associated with larger subsequent reductions in the extreme poverty rate. However, these results did not hold for the moderate poverty rate.

Taken together, the findings of this chapter (a) reaffirm the

significance of changes in economic freedom on reductions in poverty and (b) indicate that movements toward democracy also facilitate reductions in the extreme poverty rate, after accounting for the indirect impact.

4.1 Why Do Political Institutions Matter?

55

Previous empirical research indicates that the relationship between political institutions and economic growth is weak (Tavares and Wacziarg 2001).

However, there are compelling

theoretical arguments that suggest a positive link between growth and democracy. In general, democracies have constitutional constraints on the exercise of government power, which limits the expropriation of private property.

Such constraints strengthen property rights, reduce

uncertainty, lead to increased rates of investment, and provide an environment more hospitable to entrepreneurship. This facilitates economic growth and prosperity. Weingast suggests that federalism, “market preserving federalism” specifically, is the important aspect of democracy that leads to economic growth.

More autocratic regimes do not have these constitutional

protections and are more likely to have less secure property rights and hence lower growth rates and income levels. Moreover, democracies are more stable over long time horizons because the transfer of power between competing groups and leaders is handled through an orderly and predefined process.

This reduces the uncertainty accompanying long-term investments.

While non-

democratic regimes can be stable during a leader’s tenure, the transfer of power after their death or coup is often unstable. Protests, violence, and even civil wars are often a result. The uncertainty of future violence as well as the potential that future rulers will confiscate the property of those who opposed them can lead to decreased levels of investment, entrepreneurship, and lower growth rates. However, there are also some adverse elements of democracy. Taxes that transfer income from taxpayers to non-taxpayers can reduce the incentive to work and invest. The democratic political process is susceptible to interest groups and rent seeking. Lastly, political decisions in democratic countries tend to be shortsighted. Policy choices that lead to immediate benefits with costs that materialize later are generally preferred. Promising goodies in the short-run is a much easier way for politicians to win elections than pursuing long-term policy goals. These factors – transfers, interest group lobbying, and shortsightedness of the political process – lead to higher taxes or higher debt or sometimes both. This reduces the incentive to invest and discourages entrepreneurship. Autocratic regimes can, in theory, limit the effects of these factors. In practice, however, autocratic regimes may not be less susceptible to these factors. These conflicting attributes of democracy may explain its weak statistical relationship with growth present in the existing literature. Results from earlier empirical research, however, found

56

a more robust relationship. For example Scully (1988) and Barro (1991) found that democracy had a significantly positive impact on growth. Using the Gastil index of political rights and civil liberties, which is now the Freedom House index, they found that more democratic countries had higher rates of economic growth. While this index is considered a measure of democracy, the authors used it primarily as a proxy for economic institutions as no such measure existed at the time.

In later work, Barro (1997) found a non-linear relationship.

Movements toward

democracy were growth enhancing to a point, but growth reducing thereafter providing possible evidence for democracy’s shortcomings. More recent literature indicates that a weak relationship between democratic political institutions and growth is a result of accounting for the impact of economic freedom (Knack and Keefer 1995; Dawson 1998; Gwartney, Lawson, and Holcombe 1999; Wu and Davis 1999). The findings of the previous chapter regarding economic freedom and poverty are similar. Both levels and changes of democratic political institutions were generally insignificant in poverty regressions, after controlling for economic freedom and other factors. Combining these results suggests that economic freedom is a contributing factor to both growth and reductions in poverty, but that political institutions are, for the most part, unimportant. However, this ignores the possibility that political institutions may influence economic institutions and hence indirectly impact growth and poverty. Recent empirical studies have found a statistically significant relationship between political and economic institutions.

Lawson and Clark (2010) found

preliminary evidence that movements toward economic freedom were related to a country’s level of political freedom. Others found a statistically significant relationship between political and economic institutions with Granger causality tests (de Haan and Sturm 2003; Dawson 2003; Pitlik and Wirth 2003; Vega-Godillo and Alvarez-Arce 2003; Aixala and Fabro 2009). Farr, Lord, and Wolfenbarger (1998) did not find a direct link between political and economic institutions. However, they indicated that economic freedom leads to higher income levels, which corresponds to increased political freedom. This result is consistent with Lipset (1959) who suggested that higher income levels would lead to increases in political freedom. While Rigobon and Rodrik (2005) used a different measure – a rule of law measure was used as a proxy for economic freedom – and a different estimation technique, they found that democratic political institutions had a positive impact on the rule of law. Taken as a whole, this literature suggests that political and economic institutions are interrelated.

57

This may explain the insignificant impact of political institutions on growth and poverty found in the literature and the previous chapter. The overall impact may be understated if the impact of political institutions on economic freedom is not taken into account. The results of the previous chapter may not accurately reflect the true impact of democratic political institutions on poverty. The empirical analysis that follows accounts for this indirect impact in order to gain a more accurate understanding of the relationship between political institutions and poverty.

4.2 Empirical Framework The empirical analysis uses regression equations similar to the previous chapter. Both levels and changes of variables were considered in that analysis. The focus here, however, will be on changes only. Changes in institutions have been shown to be much more robust in regression equations regarding growth (Gwartney, Lawson, and Holcombe 1999). The purpose of this analysis is to ascertain the overall impact of political institutions on poverty both directly and indirectly. Therefore, the analysis first considers the impact of political institutions on economic freedom. Second, in order to obtain a “truer” measure, the analysis accounts for both the direct and indirect impact of political institutions on poverty. To examine the impact of political institutions on economic freedom the following regression equation is used. (1) ΔEFWit = α + βΔPolit-10 + δXit + γdt + uit The dependent variable, ΔEFWit, is the change in the level of economic freedom over a ten-year period for country i. The polity term, ΔPolit-10, is the change in the level of political institutions over the previous ten-year period for country i. Xit contains various control variables including the level of economic freedom at the beginning of the ten-year period as well as the level of political institutions at the beginning of the previous ten-year period. The last two terms are the period dummy and the white noise error term, respectively. This analysis uses the data largely in panel form. The period of interest spans 1985-2005 and is broken into two, ten-year periods: 1985-1995 and 1995-2005. As the political institutions measures are lagged ten-years they correspond to earlier ten-year periods: 1975-1985 and 1985-1995. Because periods of ten years are the unit of analysis, the empirical work could focus on the period 1980-2000 or 1985-2005.

58

The pattern of the results was identical. The findings for the more recent period are presented here. Regression equation 1 investigates whether movements toward more democratic political institutions during a decade correspond to increases in economic freedom in the subsequent decade. This equation does not examine changes in economic freedom and political institutions during the same period because it will take time for institutional changes to exert their impact. However, to verify the robustness of the results, the contemporaneous impact of political institutions is also considered later in the chapter. The second stage of the analysis examines the overall impact of political institutions on poverty by considering both the direct and indirect affects. The following equation captures the direct channel. (2) Δpovertyit = α + βΔEFWit + θΔPolit-10 + δXit + γdt + u Here the dependent variable is the change in either the extreme or moderate poverty rate for country i during two, ten-year periods: 1985-1995 and 1995-2005. The other variables in the equation are the same as equation 1. Equation 2 captures the direct impact of changes in political institutions in one decade on reductions in poverty during the subsequent decade, after controlling for economic freedom and other factors. However, this equation does not account for the possible impact of political institutions through economic freedom. Hence, it may understate the impact of political institutions on poverty. Gwartney, Holcombe, and Lawson (2006) used a statistical technique to isolate both the direct impact of economic freedom on growth and the indirect impact through investment. Parallel analysis is used here to measure both the direct impact of political institutions on poverty and the indirect impact through economic freedom. This involves taking the residuals from equation 1, ΔEFWResit, and using them in place of the economic freedom variable in equation 3 below. (3) Δpovertyit = α + βΔEFWResit + θΔPolit-10 + δXit + γdt + u The residuals of equation 1 represent the change in economic freedom that is unexplained by prior changes in political institutions. These residuals, when used in place of the change in economic freedom variable in equation 3, represent the impact of economic freedom on reductions in poverty, excluding that which is attributable to changes in political institutions.

59

Therefore, the coefficient on the political institutions term of equation 3, θ, will represent both the direct and indirect impact of political institutions on reductions in poverty. If political institutions exert an impact on poverty through economic freedom, the magnitude and significance of θ will be larger in equation 3 than equation 2. As in the previous chapter, the measure of economic institutions is the Economic Freedom of the World Index. The political institutions measures are the Polity IV index, the constraints on the executive measure, which is part of the Polity IV index, and the Freedom House Political Rights index. Except for the Freedom House variable, larger values of these measures represent more freedom. The scale of the Freedom House measure was reversed so that its directional change would parallel that of the other measures. Each of these political variables is a measure of the degree to which a country’s polity conforms to the democratic ideal.

The Sachs

geographic and locational variables are also included to control for the influence of various geographic and climactic factors. See table 3.2 of the previous chapter for a list of the variables and their sources, appendix B for a more thorough description of the variables, and appendix C for their summary statistics. This analysis does not contain an explicit measure of corruption even though the impact of corruption on democratic political institutions and poverty has been considered in the literature (Rose-Ackerman 1996; Gupta, Davoodi, and Alonso-Terme 2002).

The reasons for not

including a measure of corruption are twofold. First, measures of corruption are fairly new and only cover the last few years of the time period used in the analysis, 1998-2005. The most widely used measure is the Corruption Perceptions Index compiled yearly by Transparency International. This index was created largely for media and governmental use and not for academic research. The methodology used to compile the index changes frequently from year to year limiting the ability to make comparisons across time. Second, the EFW index contains subcomponents that may reflect the influence of corruption on the legal and regulatory environment.13 Thus, inclusion of a corruption measure could introduce multicollinearity into the regression analysis. 13

These are the sub-components of the EFW index that include measures of corruption. Sub-component A of area 2, legal structure and security of property rights, is a question from the Global Competitiveness Report (GCR), “Is the judiciary in your country independent from political influence of members of government, citizens, or firms?” Sub-component B of area 2 is also from the GCR, “The legal framework in your country for private businesses to settle disputes and challenge the legality of government actions and/or regulations is inefficient and subject to manipulation, or is efficient and follows a clear, neutral process?” Sub-component v of part C of area 5 is also from

60

4.3 Results Examining the impact of political institutions on economic freedom is a first step toward determining whether political institutions affect poverty indirectly through economic freedom. Table 4.1 contains the results of regressing changes in economic freedom on prior changes in political measures. There are two ten-year time periods that span 1985-2005 in the pooled OLS regressions listed in the table. The first independent variable is the EFW index at the start of the ten-year period. This corresponds to 1985 for the first ten-year period and 1995 for the second. The change for each of the three political measures during the prior period is shown as well their initial value at the beginning of the earlier period. For example, the change in the Polity IV index from 1975-1985 and the value in 1975 correspond to a change in the dependent variable during 1985-1995. A period dummy is included in the regressions to account for any timevarying effects. This dummy is generally insignificant indicating that the statistical relationships were time invariant. Lastly, the standard errors in this and all subsequent tables are robust to heteroskedasticity and are clustered by country. The first three columns of table 4.1 provide evidence that changes in political institutions affect subsequent changes in economic freedom. Both the initial level and the change of the Polity IV index, the constraints on the executive measure, and the political rights measure are positive and significant at the one percent level. In addition, the coefficients on the change in the political institutions measures are very similar. A one unit increase in the constraints on the executive and the political rights measure corresponds to a 0.16 and 0.18 increase in economic freedom over the subsequent period, respectively. The marginal impact of the Polity IV measure is similar after adjusting for its scale of -10 to 10. (Note: the scale of the other two measures ranges from 1 to 7.) These small marginal values should be understood in context. Using the measure of political rights, a move from autocracy, a value of 1 as this index has been inverted, to democracy, a value of 7, corresponds to an increase in the EFW index of 1.08 over the subsequent decade. This is a large increase and one that would be associated with significant poverty reductions based upon the results of the previous chapter.

the GCR, “In your industry, how commonly would you estimate that firms make undocumented extra payments or bribes connected with the following: A-import and export permits; B-Connection to pubic utilities (e.g., telephone or electricity); C-Annual tax payments; D-Awarding of public contracts (investment projects); E-Getting favorable judicial decisions.”

61

Table 4.1: The impact of political institutions on subsequent changes in economic freedom (pooled OLS for ten-year periods, 1985-1995 and 1995-2005) Dependent variable: Change in economic freedom, 1985-1995 and 1995-2005 Independent variable EFW, beginning of period

All countries (1) (2) (3) -0.36 *** -0.40 *** -0.44 *** (0.05) (0.05) (0.06)

Low and middle income countries (4) (5) (6) -0.38 *** -0.42 *** -0.45 *** (0.06) (0.06) (0.07)

Polity IV, beginning of previous 10-year period

0.03 *** (0.01)

0.03 *** (0.01)

Change in polity IV, previous 10-year period

0.06 *** (0.01)

0.05 *** (0.01)

Executive constraints, beginning of previous 10-year period

0.11 *** (0.03)

0.09 *** (0.03)

Change in executive constraints, previous 10-year period

0.16 *** (0.03)

0.14 *** (0.04)

Political rights, beginning of previous 10-year period

0.14 *** (0.03)

0.12 *** (0.04)

Change in political rights, previous 10-year period

0.18 *** (0.03)

0.17 *** (0.03)

Period dummy, 1985-1995

0.11 (0.10)

Intercept

2.47 *** 2.34 *** 2.47 *** (0.31) (0.26) (0.26)

0.04 (0.10)

-0.06 (0.08)

0.05 (0.13)

-0.04 (0.12)

-0.14 (0.10)

2.57 *** 2.55 *** 2.61 *** (0.37) (0.33) (0.34)

R2 (adjusted)

0.35

0.33

0.37

0.33

0.30

0.34

Number of observations

216

216

227

176

176

183

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis.

The last three columns of table 4.1 examine whether this relationship is present when the countries in the sample are low and middle-income countries. The results indicate that the relationship is largely unchanged whether or not high-income countries are included.14 The coefficients and significance levels are similar and there is little difference in the explanatory power of the regressions between columns one through three and four through six. The r-square 14

Countries were considered high income for this study if their per capita GDP in U.S. dollars in 1980 was 5,670 or higher. This cut off is roughly the per capital income level of Greece in 1980. Several oil rich countries of the Middle East and small island nations, which had income levels above this threshold, were excluded from the highincome group. The 23 high income countries using this criteria are: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Hong Kong, Iceland, Ireland, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Spain, Sweden, Switzerland, United Kingdom, and the United States.

62

values indicate that the model explains approximately one-third of the variation in the dependent variable.

Table 4.2: The impact of political institutions (1975-1985) on subsequent changes in economic freedom (1985-1995) Dependent variable: Change in economic freedom, 1985-1995 All countries (1) (2) (3) -0.39 *** -0.41 *** -0.49 *** (0.07) (0.07) (0.08)

Low and middle-income countries (4) (5) (6) -0.39 *** -0.43 *** -0.51 *** (0.09) (0.10) (0.10)

Polity IV, 1975

0.05 *** (0.01)

0.04 *** (0.01)

Change in Polity IV, 1975-1985

0.08 *** (0.02)

0.08 *** (0.02)

Independent variable EFW, 1985

Executive constraints, 1975

0.14 *** (0.04)

0.12 *** (0.04)

Change in executive constraints, 1975-1985

0.22 *** (0.05)

0.21 *** (0.06)

Political rights, 1975

0.22 *** (0.04)

0.20 *** (0.05)

Change in political rights, 1975-1985

0.22 *** (0.05)

0.21 *** (0.06)

2.70 *** 2.28 *** 2.41 *** (0.37) (0.35) (0.40)

2.72 *** 2.43 *** 2.51 *** (0.52) (0.50) (0.49)

Intercept R2 (adjusted)

0.35

0.33

0.39

0.34

0.32

0.38

Number of observations

101

101

106

81

81

84

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis.

The period dummy included in the regressions of table 4.1 indicates that there was very little impact of unobserved time-varying effects. Tables 4.2 and 4.3, which display the regression results for each of the ten-year periods separately, confirm this finding. Table 4.2 examines the 1985-1995 period for all countries and low and middle-income countries. The pattern is virtually identical to that of table 4.1.

Table 4.3, which covers the period 1995-2005, has lower

significance levels, but the same overall results. The change in the Polity IV and the constraints on the executive measure in table 4.3 are significant at the five percent level or higher, while they are significant at the one percent level in table 4.2. The change in the political rights

63

measure is significant at the one percent level in both tables 4.2 and 4.3, although the magnitude of the coefficient is slightly smaller in the latter table. In addition, the initial level of the political measures is largely insignificant in table 4.3. The lack of significance of the initial levels, however, is not important for this analysis as the focus is on changes. The initial levels were included to control for institutional quality at the beginning of the period. The two tables illustrate that the relationship between changes in political institutions and changes in economic freedom in subsequent periods, is consistent across decades.15

Table 4.3: The impact of political institutions (1985-1995) on subsequent changes in economic freedom (1995-2005) Dependent variable: Change in economic freedom, 1995-2005 All countries (1) (2) (3) -0.31 *** -0.37 *** -0.33 *** (0.05) (0.05) (0.05)

Low and middle-income countries (4) (5) (6) -0.34 *** -0.39 *** -0.34 *** (0.06) (0.06) (0.05)

Polity IV, 1985

0.01 (0.01)

0.00 (0.01)

Change in Polity IV, 1985-1995

0.04 ** (0.02)

0.04 ** (0.02)

Independent variable EFW, 1995

Executive constraints, 1985

0.05 * (0.03)

0.04 (0.03)

Change in executive constraints, 1985-1995

0.10 ** (0.04)

0.10 ** (0.04)

Political rights, 1985

0.03 (0.03)

0.01 (0.04)

Change in political rights, 1985-1995

0.13 *** (0.03)

0.12 *** (0.03)

2.27 *** 2.47 *** 2.33 *** (0.30) (0.27) (0.35)

2.37 *** 2.60 *** 2.42 *** (0.37) (0.34) (0.32)

Intercept R2 (adjusted)

0.38

0.36

0.41

0.33

0.30

0.35

Number of observations

115

115

121

95

95

99

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis.

15

Because the regressions of table 4.2 and 4.3 cover different time periods and hence different data, a Hausman specification test in order to determine if the results between the two periods are statistically different cannot be performed here.

64

The previous tables focused on how changes in political institutions impact subsequent changes in economic freedom. Contemporaneous changes of political institutions were excluded from these tables. The following two tables include the contemporaneous period in order to determine if the change in the prior period remains significant even after inclusion of the change during the current period. The first of these tables, table 4.4, contains pooled OLS regressions for both all countries and low and middle-income countries during the 1985-1995 and 1995-2005 periods. It is clear from the regressions that changes in political institutions during the current period are significantly related to changes in economic freedom. The changes in constraints on the executive and political rights during the current period are positive and significant at the one percent level for all countries and for low and middle-income countries. The change in the Polity IV index during the current period is also positive and significant, but at the five percent level or higher in the regressions. Even after inclusion of changes during the current period, however, changes in the prior ten-year period are still significant and are largely unchanged from the results of table 4.1. These results indicate that both prior and current changes in political institutions positively impact changes in economic freedom.

But, the magnitude of the

coefficients suggests that the prior period has a larger impact. Table 4.5 is similar to table 4.4 except two of the Sachs geography variables are included. The distance to major markets variable was excluded from this analysis because it was insignificant in all regressions. Sachs has argued that geographic and locational factors can hinder growth. Moreover, Acemoglu, Robinson, and Johnson (2001) argue that geography directly impacted institutions through the institutional arrangements utilized by early settlers. These institutions could persist through time resulting in a geographic influence on institutional change. The results of table 4.5 indicate that geographic factors are associated with changes in economic freedom. The coastal population variable, which is the percentage of a countries population that lives within 100 kilometers of a coastline, is positive and significant at the one percent level throughout the regressions. The positive coefficient indicates that a larger share of the population close to the coast was more conducive to increases in economic freedom after controlling for other factors. These areas are generally closer to trade routes and hence global markets. The tropical location variable is significant at the five percent level or higher in all the regressions indicating that tropical countries had smaller increases in economic freedom during 1985-2005.

Nonetheless, the regressions indicate that the impact of changes in political

65

institutions on subsequent changes in economic freedom is largely unchanged after the inclusion of geographic factors. In summary, tables 4.1 through 4.5 indicate that changes in political institutions exert a subsequent impact on changes in economic freedom. This suggests that political institutions may have an indirect impact on reductions in poverty. This is the topic to which I now turn.

Table 4.4: The impact of prior and concurrent changes in political institutions on changes in economic freedom (pooled OLS for ten-year periods, 1985-1995 and 1995-2005) Dependent variable: Change in economic freedom, 1985-1995 and 1995-2005 Independent variable EFW, beginning of period

All countries (1) (2) (3) -0.36 *** -0.40 *** -0.45 *** (0.05) (0.05) (0.06)

Low and middle-income countries (4) (5) (6) -0.38 *** -0.42 *** -0.46 *** (0.06) (0.06) (0.07)

Polity IV, beginning of previous 10-year period

0.04 *** (0.01)

0.03 *** (0.01)

Change in polity IV, previous 10-year period

0.07 *** (0.01)

0.07 *** (0.01)

Change in polity IV, current 10-year period

0.02 ** (0.01)

0.02 ** (0.01)

Executive constraints, beginning of previous 10-year period

0.15 *** (0.03)

0.13 *** (0.03)

Change in executive constraints, previous 10-year period

0.21 *** (0.04)

0.20 *** (0.04)

Change in executive constraints, current 10-year period

0.11 *** (0.03)

0.11 *** (0.03)

Political rights, beginning of previous 10-year period

0.18 *** (0.04)

0.17 *** (0.04)

Change in political rights, previous 10-year period

0.22 *** (0.03)

0.21 *** (0.03)

Change in political rights, current 10-year period

0.10 *** (0.04)

0.10 *** (0.04)

0.01 (0.09)

-0.07 (0.08)

Period dummy, 1985-1995

0.08 (0.10)

Intercept

2.38 *** 2.08 *** 2.38 *** (0.27) (0.27) (0.32)

0.02 (0.13)

-0.07 (0.12)

-0.14 (0.10)

2.48 *** 2.26 *** 2.47 *** (0.38) (0.35) (0.35)

R2 (adjusted)

0.37

0.37

0.40

0.34

0.34

0.37

Number of observations

216

216

227

176

176

183

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis.

66

Table 4.5: The impact of prior and concurrent changes in political institutions on changes in economic freedom, after controlling for geographic factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005) Dependent variable: Change in economic freedom, 1985-1995 and 1995-2005 Independent variable EFW, beginning of period

All countries (1) (2) (3) -0.45 *** -0.47 *** -0.51 *** (0.05) (0.05) (0.06)

Low and middle-income countries (4) (5) (6) -0.49 *** -0.52 *** -0.55 *** (0.07) (0.07) (0.07)

Polity IV, beginning of previous 10-year period

0.03 *** (0.01)

0.03 *** (0.01)

Change in polity IV, previous 10-year period

0.06 *** (0.01)

0.06 *** (0.01)

Change in polity IV, current 10-year period

0.02 ** (0.01)

0.02 ** (0.01)

Executive constraints, beginning of previous 10-year period

0.11 *** (0.03)

0.10 *** (0.03)

Change in executive constraints, previous 10-year period

0.19 *** (0.04)

0.18 *** (0.04)

Change in executive constraints, current 10-year period

0.10 *** (0.03)

0.10 *** (0.03)

Political rights, beginning of previous 10-year period

0.13 *** (0.03)

0.12 *** (0.04)

Change in political rights, previous 10-year period

0.19 *** (0.03)

0.19 *** (0.03)

Change in political rights, current 10-year period

0.09 *** (0.03)

0.09 *** (0.03)

Coastal population (% within 100km)

0.54 *** 0.51 *** 0.50 *** (0.13) (0.12) (0.13)

0.66 *** 0.65 *** 0.64 *** (0.15) (0.15) (0.16)

Tropical location (% area in tropics)

-0.34 *** -0.29 *** -0.25 *** (0.10) (0.10) (0.10)

-0.32 *** (0.11)

-0.28 *** -0.24 ** (0.11) (0.10)

Period dummy, 1985-1995

0.01 (0.09)

-0.07 (0.12)

-0.15 (0.11)

Intercept

2.88 *** 2.60 *** 2.78 *** (0.27) (0.27) (0.32)

-0.06 (0.09)

-0.12 (0.08)

-0.21 ** (0.09)

3.04 *** 2.85 *** 3.02 *** (0.35) (0.38) (0.34)

R2 (adjusted)

0.43

0.42

0.45

0.41

0.40

0.43

Number of observations

216

216

227

176

176

183

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis.

The previous tables contained regressions with all countries and low and middle-income countries. But, the poverty rate data are unavailable for all of these countries. Table 4.6 shows

67

the regressions of table 4.1 and table 4.5, but only for the countries for which the poverty rate data are available. Unsurprisingly, these results are similar to those of the previous tables. The changes in the political institutions variables are positive and significant at the one percent level in all regressions. The initial levels of the political institutions variables are less significant than before, but these were included only to control for the type of institutions present at the start of the period.

Table 4.6: The impact of political institutions on subsequent changes in economic freedom for countries with poverty data (pooled OLS for ten-year periods, 1985-1995 and 1995-2005)

Independent variable EFW, beginning of period

Dependent variable: Change in economic freedom, 1985-1995 and 1995-2005 (1) (2) (3) (4) (5) (6) -0.42 *** -0.44 *** -0.46 *** -0.52 *** -0.53 *** -0.56 *** (0.07) (0.06) (0.07) (0.07) (0.07) (0.07)

Polity IV, beginning of previous 10-year period

0.02 ** (0.01)

0.02 * (0.01)

Change in polity IV, previous 10-year period

0.04 *** (0.01)

0.03 *** (0.01)

Executive constraints, beginning of previous 10-year period

0.08 ** (0.04)

0.06 * (0.03)

Change in executive constraints, previous 10-year period

0.10 *** (0.03)

0.09 *** (0.03)

Political rights, beginning of previous 10-year period

0.12 *** (0.05)

0.10 ** (0.05)

Change in political rights, previous 10-year period

0.12 *** (0.03)

0.11 *** (0.03)

Coastal population (% within 100km)

0.63 *** 0.63 *** 0.61 *** (0.19) (0.19) (0.18)

Tropical location (% area in tropics)

-0.29 ** (0.12)

-0.24 * (0.13)

-0.25 ** (0.12)

Period dummy, 1985-1995

-0.04 (0.11)

-0.08 (0.11)

-0.16 * (0.10)

Intercept

2.83 *** 2.66 *** 2.63 *** 3.32 *** 3.18 *** 3.27 *** (0.33) (0.39) (0.37) (0.34) (0.35) (0.34)

R2 (adjusted)

0.31

0.30

0.32

0.36

0.35

0.37

Number of observations

145

145

145

145

145

145

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis.

68

The results of regressing the reduction in the extreme poverty rate during 1985-1995 and 1995-2005 on changes in economic freedom, political institutions, and the Sachs geography variables are shown in table 4.7. These are pooled OLS regressions with changes over ten-year periods. The extreme poverty rate at the beginning of the period is positive and significant at the one percent level in all regressions. This indicates that poorer countries had larger reductions in poverty during 1985-2005, after controlling for other factors. Both the level and change in economic freedom over the period are positive and highly significant in all the regressions. A one unit increase in economic freedom corresponds to a 2.87 percentage point reduction or more in the extreme poverty rate, after controlling for other factors. The three political measures are largely insignificant in all six regressions of table 4.7. The Polity IV index at the beginning of the prior period is significant at the ten percent level in column four, while the initial political rights measure is significant at the ten percent and five percent level in columns three and six, respectively. The coefficients for the change in political institutions measures are insignificant in all regressions. The last three columns include the coastal population and tropical location variables. The introduction of the geographic and locational variables into the model does not affect either the sign or significance of the economic freedom and political institutions variables. The results of table 4.7 are consistent with those of the previous chapter. Economic freedom and geographic factors appear to matter for reductions in extreme poverty, while political institutions are largely insignificant. The next table investigates whether the coefficients on the changes in political institutions in table 4.7 are understated as a result of not accounting for the impact of political institutions on economic freedom. This is accomplished using the residuals from the regressions of columns one through three of table 4.6 in place of the change in economic freedom variable. The residuals from each of the first three columns of table 4.6 represent the change in economic freedom during 1985-1995 and 1995-2005 that is unexplained by the initial level and change of the political measure in the previous ten-year period. When these residuals are used in the equation, the political measures will reflect both the direct and indirect impact of political institutions on poverty. If prior changes in political institutions influence current changes in economic freedom, as tables 4.1-4.6 indicate, then one would expect a larger and more significant coefficient on the change in political institutions measures once this indirect impact is taken into account.

69

Table 4.7: The direct impact of political institutions on reductions of the extreme poverty rate, after controlling for economic freedom and geographic factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005) Independent variable Extreme poverty rate, beginning of period

Dependent variable: Reduction in extreme poverty rate, 1985-1995 and 1995-2005 (1) (2) (3) (4) (5) (6) 0.16 *** 0.16 *** 0.16 *** 0.22 *** 0.22 *** 0.22 *** (0.03) (0.03) (0.04) (0.04) (0.04) (0.03)

EFW, beginning of period

2.71 *** 2.67 *** 2.37 *** 2.27 *** 2.30 *** 2.01 ** (0.84) (0.87) (0.84) (0.87) (0.88) (0.84)

Change in EFW, current 10-year period

3.41 *** 3.33 *** 3.03 *** 3.19 *** 3.14 *** 2.87 *** (1.06) (1.03) (1.00) (1.02) (0.99) (1.06)

Polity IV, beginning of previous 10-year period

0.10 (0.08)

0.18 * (0.10)

Change in polity IV, previous 10-year period

0.03 (0.14)

0.05 (0.13)

Executive constraints, beginning of previous 10-year period

0.38 (0.31)

0.41 (0.33)

Change in executive constraints, previous 10-year period

0.30 (0.38)

0.33 (0.39)

Political rights, beginning of previous 10-year period

0.68 * (0.38)

0.85 ** (0.38)

Change in political rights, previous 10-year period

0.63 (0.46)

0.60 (0.47)

Coastal population (% within 100km)

4.30 ** (1.89)

Tropical location (% area in tropics)

-6.08 *** -5.52 *** -5.81 *** (2.05) (1.94) (2.00) -1.01 (1.13)

-1.24 (1.22)

-1.37 (1.22)

4.33 ** (1.92)

-1.32 (1.18)

4.15 ** (1.94)

-1.51 (1.26)

Period dummy, 1985-1995

-1.17 (1.18)

Intercept

-16.81 *** -18.34 *** -17.57 *** -13.51 *** -15.91 *** -15.46 *** (6.03) (6.31) (5.39) (5.59) (5.79) (5.72)

R2 (adjusted)

0.19

0.19

0.21

0.26

0.25

0.27

Number of observations

145

145

145

145

145

145

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis.

Table 4.8 is identical to 4.7 except that the measure of economic freedom is the residuals from table 4.6. Put another way, the change in economic freedom variable in table 4.8 reflects that portion that is unrelated to prior changes in political institutions. The magnitude of the coefficient of the change in the Polity IV index in columns one and four of table 4.8 is larger

70

than the corresponding values in table 4.7, but remains insignificant. However, the coefficient of the change in executive constraints measure is significant at the ten percent level in column two

Table 4.8: The direct and indirect impact of political institutions on reductions of the extreme poverty rate, after controlling for economic freedom and geographic factors (pooled OLS for tenyear periods, 1985-1995 and 1995-2005) Independent variable Extreme poverty rate, beginning of period

Dependent variable: Reduction in extreme poverty rate, 1985-1995 and 1995-2005 (1) (2) (3) (4) (5) (6) 0.16 *** 0.16 *** 0.16 *** 0.22 *** 0.22 *** 0.22 *** (0.03) (0.03) (0.04) (0.04) (0.04) (0.03)

EFW, beginning of period

1.28 ** (0.59)

1.20 ** (0.59)

0.98 (0.63)

0.93 (0.60)

0.92 (0.61)

Change in EFW proxy, residuals from table 4.6

3.41 *** 3.33 *** 3.03 *** 3.19 *** 3.14 *** 2.87 *** (1.06) (1.03) (1.00) (1.02) (0.99) (1.06)

Polity IV, beginning of previous 10-year period

0.18 ** (0.09)

0.25 *** (0.10)

Change in polity IV, previous 10-year period

0.15 (0.13)

0.17 (0.13)

Executive constraints, beginning of previous 10-year period

0.64 ** (0.32)

0.66 * (0.34)

Change in executive constraints, previous 10-year period

0.64 * (0.38)

0.66 * (0.38)

0.68 (0.63)

Political rights, beginning of previous 10-year period

1.06 *** (0.42)

1.21 *** (0.42)

Change in political rights, previous 10-year period

0.98 ** (0.45)

0.94 ** (0.45) 4.33 ** (1.92)

4.15 ** (1.94)

Coastal population (% within 100km)

4.30 ** (1.89)

Tropical location (% area in tropics)

-6.08 *** -5.52 *** -5.81 *** (1.94) (2.00) (2.05)

Period dummy, 1985-1995

-1.17 (1.18)

-1.01 (1.13)

-1.24 (1.22)

-1.37 (1.22)

-1.32 (1.18)

-1.51 (1.26)

Intercept

-7.15 * (3.89)

-9.47 ** (4.35)

-9.59 ** (4.55)

-4.47 (3.65)

-7.56 * (3.88)

-7.90 * (4.08)

R2 (adjusted)

0.19

0.19

0.21

0.26

0.25

0.27

Number of observations

145

145

145

145

145

145

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis.

71

and the magnitude of the coefficient is more than twice as large as the corresponding value in table 4.7. The coefficient on the change in political rights in column three is also much larger than before and is now significant at the five percent level. A movement from no political rights, a score of 1, to full political rights, a score of 7, corresponds to a reduction in the extreme poverty rate of 5.9 percentage points over the subsequent decade. This is a significant reduction in extreme poverty. Column two tells a similar story. A change from having an unconstrained executive to one that is constitutionally constrained, an increase from 1 to 7, results in a 3.8 percentage point reduction in the extreme poverty rate over the subsequent decade. Columns four through six add the Sachs geography variables and the results are unchanged. While the Polity IV variable remained insignificant (column 4), increased political rights and constraints upon the chief executive (columns five and six) significantly reduced the extreme poverty rate, after controlling for geographic and other factors. In table 4.8, the levels of each political measure at the beginning of the earlier period are positive and significant at the ten percent level or higher in all regressions. This indicates that countries with more democratic political institutions at the start of the period had larger reductions in the extreme poverty rate, after controlling for other factors. The results of table 4.8 are consistent with the view that changes in political institutions facilitate reductions in the extreme poverty rate both directly and indirectly through changes in economic freedom.

Two out of the three changes in political institutions measures were

significant – with and without the geography variables – when residuals from table 4.6 were used in place of the change in economic freedom variable. Tables 4.9 and 4.10 examine whether a similar relationship exists between political institutions and the moderate poverty rate. Table 4.9 is similar to table 4.7 except that the dependent variable is now the reduction in the moderate poverty rate during 1985-1995 and 1995-2005. Both the level and change of economic freedom exert a positive and significant impact on reductions in the moderate poverty rate. This indicates that countries with higher levels of economic freedom had larger reductions in moderate poverty. Correspondingly, countries with larger increases in economic freedom achieved larger reductions in moderate poverty during 1985-2005. The structure of table 4.10 is similar to table 4.8 except that the dependent variable is now the reduction in the moderate poverty rate. Table 4.10 uses the residuals from table 4.6 as the measure of economic freedom. In this case, the coefficients on each of the change in the

72

political institutions measures are insignificant in all regressions. The results for both the economic freedom variables and the geographic and locational variables are unchanged. These results indicate that economic freedom and geographic factors play a much larger role regarding

Table 4.9: The direct impact of political institutions on reductions of the moderate poverty rate, after controlling for economic freedom and geographic factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005) Independent variable Moderate poverty rate, beginning of period

Dependent variable: Reduction in moderate poverty rate, 1985-1995 and 1995-2005 (1) (2) (3) (4) (5) (6) 0.13 *** 0.13 *** 0.14 *** 0.18 *** 0.18 *** 0.19 *** (0.03) (0.03) (0.03) (0.03) (0.03) (0.04)

EFW, beginning of period

4.12 *** 4.14 *** 4.04 *** 3.60 *** 3.69 *** 3.60 *** (1.07) (1.08) (1.10) (1.10) (1.14) (1.14)

Change in EFW, current 10-year period

4.50 *** 4.45 *** 4.37 *** 4.20 *** 4.19 *** 4.14 *** (1.34) (1.33) (1.30) (1.28) (1.31) (1.28)

Polity IV, beginning of previous 10-year period

0.06 (0.10)

0.13 (0.11)

Change in polity IV, previous 10-year period

-0.07 (0.16)

-0.04 (0.15)

Executive constraints, beginning of previous 10-year period

0.11 (0.32)

0.13 (0.36)

Change in executive constraints, previous 10-year period

-0.14 (0.43)

-0.08 (0.43)

Political rights, beginning of previous 10-year period

0.46 (0.41)

0.61 (0.42)

Change in political rights, previous 10-year period

-0.10 (0.42)

-0.07 (0.43) 4.17 * (2.28)

Coastal population (% within 100km)

4.51 ** (2.24)

Tropical location (% area in tropics)

-5.89 *** -5.45 *** -5.85 *** (2.05) (2.01) (2.03)

4.58 ** (2.32)

Period dummy, 1985-1995

-1.87 (1.61)

Intercept

-25.41 *** -26.13 *** -27.20 *** -22.15 *** -23.61 *** -24.84 *** (7.56) (7.71) (8.04) (7.21) (7.35) (7.64)

-1.77 (1.52)

-1.50 (1.43)

-2.07 (1.60)

-2.09 (1.54)

-1.79 (1.47)

R2 (adjusted)

0.20

0.20

0.20

0.25

0.24

0.26

Number of observations

145

145

145

145

145

145

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis.

73

reductions in the moderate poverty rate.

Table 4.10: The direct and indirect impact of political institutions on reductions of the moderate poverty rate, after controlling for economic freedom and geographic factors (pooled OLS for tenyear periods, 1985-1995 and 1995-2005) Independent variable Moderate poverty rate, beginning of period

Dependent variable: Reduction in moderate poverty rate, 1985-1995 and 1995-2005 (1) (2) (3) (4) (5) (6) 0.13 *** 0.13 *** 0.14 *** 0.18 *** 0.18 *** 0.19 *** (0.03) (0.03) (0.03) (0.03) (0.04) (0.03) 1.86 ** (0.81)

1.69 ** (0.81)

EFW, beginning of period

2.23 *** 2.19 *** 2.02 *** 1.84 ** (0.79) (0.81) (0.80) (0.78)

Change in EFW proxy, residuals from table 4.6

4.50 *** 4.45 *** 4.37 *** 4.20 *** 4.19 *** 4.14 *** (1.33) (1.30) (1.28) (1.31) (1.28) (1.34)

Polity IV, beginning of previous 10-year period

0.15 (0.10)

0.22 * (0.11)

Change in polity IV, previous 10-year period

0.09 (0.15)

0.11 (0.14)

Executive constraints, beginning of previous 10-year period

0.46 (0.33)

0.46 (0.36)

Change in executive constraints, previous 10-year period

0.33 (0.40)

0.35 (0.41)

Political rights, beginning of previous 10-year period

1.01 ** (0.45)

1.13 ** (0.46)

Change in political rights, previous 10-year period

0.41 (0.40)

0.41 (0.40)

Coastal population (% within 100km)

4.51 ** (2.24)

Tropical location (% area in tropics)

-5.89 *** -5.45 *** -5.85 *** (2.05) (2.01) (2.03)

Period dummy, 1985-1995

-1.87 (1.61)

-1.77 (1.52)

-1.50 (1.43)

-2.07 (1.60)

Intercept

-12.66 ** (5.43)

-14.30 *** -15.68 *** -10.26 ** (5.09) (5.93) (5.70)

4.58 ** (2.32)

4.17 * (2.28)

-2.09 (1.54)

-1.79 (1.47)

-12.48 ** (5.26)

-13.95 *** (5.48)

R2 (adjusted)

0.20

0.20

0.20

0.25

0.24

0.26

Number of observations

145

145

145

145

145

145

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis.

74

This empirical analysis indicates that changes in political institutions exert an impact on subsequent changes in economic freedom.

This in turn corresponded to reductions in the

extreme poverty rate in table 4.8. But, this was not the case for the moderate poverty rate. Even after accounting for the indirect affect through economic freedom, changes in political institutions did not reduce the moderate poverty rate. Robust standard errors were used in all the tables presented here. There are two potential reasons why this is important. First, given the heterogeneity across countries, the assumption of common variance is unlikely to be met. Second, pooled OLS regressions are susceptible to errors resulting from serial correlation in the error term. However, statistical tests indicated that serial correlation was not a problem in these regressions.16 While serial correlation appears not to be a problem, the significant heterogeneity of the countries in the sample warranted the use of standard errors robust to heteroskedasticity.

4.4 Conclusion This chapter examined whether changes in political institutions had an impact on reductions in poverty during 1985-2005. Prior research indicated that political institutions exert an impact on economic freedom.

This analysis extended this result in order to determine if political

institutions indirectly facilitate reductions in poverty. The first set of regression tables examined whether movements toward more democratic political institutions were associated with subsequent movements toward economic freedom.

The findings presented here are highly

supportive of this view. Changes in democratic political institutions during 1975-1985 and 1985-1995 were positive and significantly related to increases in economic freedom during 19851995 and 1995-2005. This relationship held after controlling for initial institutional levels and geographic and locational factors. Moreover, this was true both for all countries and for low and middle-income countries alone. The second half of the analysis measured both the direct impact of changes in political institutions on poverty and the indirect impact through changes in economic freedom. The

16

The test for the existence of serial correlation in the error term comes from Wooldridge (2002, 176). It is implemented by including the lagged residuals in the regression and then testing the null hypothesis of no serial correlation in the error term. This is done with a simple t-test on the coefficient of the lagged residuals. Under the assumption of no serial correlation, the coefficient should not be statistically different from zero. The null hypothesis could not be rejected for all the pooled OLS regressions.

75

results indicated that movements toward democracy were associated with subsequent reductions in the extreme poverty rate, after including the indirect impact of political institutions. But, this was not true for reductions in the moderate poverty rate. Thus, while the results indicate that democratic political institutions facilitate movements toward economic freedom, their impact on poverty rate reductions is more tenuous.

76

CHAPTER 5 ECONOMIC FREEDOM, FOREIGN AID, AND GLOBAL POVERTY A primary goal, if not the primary goal, of foreign aid is to alleviate poverty in developing countries. The poverty rate data presented in chapter two suggests that while global poverty rates have fallen, there remain billions of individuals who live in extreme poverty and even more who live in moderate poverty. During the past several decades, vast sums of foreign aid funds have been transferred to developing countries. This chapter seeks to empirically determine the impact of foreign aid during 1976-2005 on the extreme and moderate poverty rates during recent decades. There is a vast literature on the subject of foreign aid. Much of this work focuses on weather foreign aid is growth enhancing. The findings of this literature have varied greatly. Evidence exists that aid facilitates economic growth. Others have found inconclusive statistical results, while still others have found that aid has a negative impact on growth. Recent influential work by Rajan and Subramanian (2008) indicates that a statistically significant relationship between aid and growth is difficult to find. Public Choice theory suggests why this may be the case. Developing countries often have poor governing institutions and high levels of corruption. As a result, a sizeable portion of aid funds may be lost to rent seeking and other nonproductive activities. In addition, modern growth theory stresses the importance of institutions supportive of economic freedom.

Pressures for constructive reforms often stem from poor economic

performance. Provision of aid to countries experiencing sluggish growth or reductions in income levels may reduce the likelihood of growth enhancing institutional reforms. The link between foreign aid and economic growth is important and has been well researched. The relationship between aid and poverty is equally important, however, it is less developed in the current literature. Using data for the period 1976-2005, this chapter attempts to develop this literature by empirically testing three questions. First, what factors influence the level of foreign aid a country receives? Second, has foreign aid exerted any impact on economic freedom? Third, has foreign aid had any impact on the reductions in the extreme and moderate poverty rate during 1980-2005? The results suggest the answer to the first question is that poverty is by far the most significant determinant of foreign aid dollars. However, the answers to the second and third 77

questions are less clear. The data presented here finds very little evidence that foreign aid had a positive impact on changes in economic freedom or reductions in poverty during 1980-2005, after controlling for economic institutions, political institutions, geographic factors, and unobservable time-varying effects. This is in spite of the result that the single most important factor determining the level of foreign aid a country receives is the poverty rate. Before beginning an extensive discussion of these findings, it is necessary to place this work within the context of the existing literature. The following section focuses on this topic. Section 5.2 explores the theoretical reasons for why foreign aid may or may not impact poverty. Section 5.3 presents the empirical structure used in the analysis and some preliminary data on foreign aid. Section 5.4 discusses the empirical results and section 5.5 concludes.

5.1 Foreign Aid and Growth The literature on foreign aid and growth is massive, spanning more than fifty years and covering both theoretical and empirical analysis. In order to accomplish a manageable review of this literature the focus will be primarily on the most recent empirical papers. Several non-empirical papers will be discussed as they provide theoretical insight for aspects of the empirical results. Before the flurry of empirical work on aid and growth, the aid literature was strongly supportive of the idea that aid was necessary for growth in developing countries. The two lone dissenters of this orthodoxy were Peter T. Bauer and Milton Friedman. The orthodoxy at the time held that poor countries were poor due to an insufficient savings rate. This was known as the savings gap and is best understood in the context of the neoclassical growth model. In this model the difference in the steady-state income levels of countries is a result of differences in savings rates. Therefore, it was believed that if this savings gap was filled with outside aid, developing countries would “take off” due to increased growth rates in the short run and transition to a higher steady-state income level. Verifying whether this theory held was difficult due to a lack of empirical data on aid flows. This changed in the mid 1990s when data on foreign aid became available for a large number of developing countries. Table 5.1 lists the papers that used this aid data to examine the relationship between aid and growth between 1996 and 2011.17 In total 27 papers empirically investigated the aid-growth 17

Five of these papers looked at income level, income inequality, poverty, or health measures rather than growth.

78

relationship. Fourteen found statistical results which suggested that aid was growth enhancing, with the results often conditional on good policies or geographic factors. Eight concluded that neither a positive nor negative statistically significant relationship existed between aid and growth. And five found statistically significant evidence that aid inhibited growth. This volume of papers with conflicting results in such a short time span (16 were published between 2000 and 2004) implied that there was no general consensus regarding aid and growth. In addition, many of the papers used different specifications and examined different time periods limiting the comparability of the results. This allowed aid policy makers to choose studies that supported their particular position. Rather than review each empirical paper, this discussion will focus on the most influential. Boone (1996), the first such paper to perform an empirical analysis with aid data, did not find a statistically significant relationship between aid and growth but did find that aid was fungible. Poor countries demonstrated statistically significant increases in government consumption as a result of increased aid flows. In addition, the results suggested that governments with better policies – those where aid flows were less fungible – might be better candidates for the receipt of aid. Burnside and Dollar (2000) showed that there was a statistically positive relationship between aid and growth in countries with good policies. They performed a panel analysis over the period 1970-1993 with data averaged over four-year periods. They found that a standalone variable for aid was insignificant in the regressions, but that aid interacted with a quality of government policy measure was positive and significant.18 At the time the current trend in the aid community was to provide aid in the form of Structural Adjustment (SA) loans. These SA loans were given with a share of the money up front and the remainder contingent upon various government reforms. Burnside and Dollar argued that their result suggested that SA loans were ineffective for promoting growth. Instead, they argued that the aid community needed to be selective and channel aid to countries that already had good policies.

See the note of table 5.1 for more details. 18 Burnside and Dollar aggregated central government budget surplus, a measure of inflation (M2/GDP), and the Sachs-Warner trade openness index into a policy index.

79

Table 5.1: Empirical papers on foreign aid and growth, 1996-201119 Papers Svensson, Jakob (1999) Burnside and Dollar (2000) Hansen and Tarp (2000) Hansen and Tarp (2001) Lensink and White (2001) Guillaumont and Chauvet (2001) Collier and Dehn (2001) Dalgaard and Hansen (2001) Collier and Dollar (2002) Burnside and Dollar (2004) Dalgaard, Hansen, and Tarp (2004) Clemens, Radelet, and Bhavnani (2004) Heckelman and Knack (2009) Bearce and Tirone (2010)

Impact of aid on growth + + + + + + + + + + + + + +

Boone (1996) Arvin and Barillas (2002) Easterly (2003) Easterly, Levine, and Roodman (2004) Kraay and Raddatz (2007) Rajan and Subramanian (2008) Williamson (2008) Chong, Gradstein, and Calderon (2009)

~ ~ ~ ~ ~ ~ ~ ~

Brumm (2003) Ovaska (2003) Rajan and Subramanian (2005) Djankov, Montalvo, and Reynal-Querol (2006) Rajan and Subramanian (2011)

– – – – –

The Burnside and Dollar paper was extremely influential in the aid community. It provided the impetus behind the Millennium Development Corporation, was used as justification for doubling aid flows (Easterly, 2003), and was responsible for a shift in the approach of the aid community from SA to selectivity.

However, subsequent work by Easterly, Levine and

19

While a majority of these papers examined the relationship between aid and growth several focused on poverty and health indicators. Collier and Dollar (2002) focused on poverty and attempted to determine the poverty efficient allocation of aid. Chong, Gradstein, and Calderon (2009) examined poverty and inequality. Arvin and Barillas (2002) examined poverty, but used GNP per capita as the measure of poverty. Therefore this paper should be considered a study of how aid influences average income levels rather than poverty. Kraay and Raddatz (2007) focused on poverty traps and the link between aid, investment, and growth. Williamson (2008) explored the impact of aid on numerous health measures.

80

Roodman (2004) and Easterly (2003) found that the Burnside and Dollar result was not robust. They demonstrated that the statistical significance of the aid-policy interaction term disappeared when: additional countries were added to the data set, both the time period and time horizon were changed, alternative measures of aid were used, and alternative measures for good policy were used. In short their empirical results suggested that neither SA nor the selective use of aid would lead to growth in poor countries. As table 5.1 illustrates, subsequent papers failed to reach an unambiguous conclusion regarding the aid-growth question. There were also differing conclusions among papers where aid positively influenced growth. Burnside and Dollar (2000) argued that aid contributed to higher growth rates in countries with good policies. This was subject to diminishing returns, however. Dalgaard, Hansen, and Tarp (2004) found that aid was growth enhancing in countries with favorable geographic characteristics.

And Dalgaard and Hansen (2001) found that aid

contributed to growth regardless of the policy environment. Comparability across these results was difficult, however, due to differing specifications, time periods, and explanatory variables. In addition, time periods as short as four years were used restricting the relevancy of the results to the short run.20 Rajan and Subramanian (2008), hereafter referenced as RS, addressed these shortcomings with their influential paper.

They emphasized that a general empirical

specification was needed to test the differing conclusions of the literature. They also stressed that short run analysis failed to account for the influence of business cycle fluctuations. More importantly they argued that it was the long run impact of aid that mattered, not the short run. To ensure comprehensive results, RS conducted both cross-section and panel data analysis. Time periods no shorter than ten years were used so as to focus on the long run impact of aid. In addition, IV estimation was used to account for any endogeneity between aid and growth. Using this specification they were unable to find any evidence supportive of a positive or negative statistical relationship between aid and growth. conditional on policies or geographic factors.

This included whether aid was effective In concluding their analysis, RS suggest a

theoretical justification for why one would expect an insignificant impact of foreign aid on growth. Assuming that at least some portion of foreign aid is used productively, one should expect an impact on growth no greater than that of typical investment funds. Estimates from empirical growth regressions indicate that the coefficient of the investment to GDP ratio is 20

For example see Burnside and Dollar (2000) and Collier and Dollar (2002).

81

roughly 0.03. This implies that a one percentage point increase in the investment to GDP ratio corresponds to a 0.03 percentage point increase in the growth rate. This is a small impact and one that is most likely hidden in noisy data or measurement error. Therefore, RS suggest that the absence of a robust statistical relationship between aid and growth is unsurprising. This result adds weight to the view that foreign aid, in its current form, has been an ineffective tool for producing economic growth in developing countries. This leaves open, however, the possibility that aid, if implemented differently in the future, might facilitate economic growth in the developing world. However, there are several reasons why aid, even if implemented differently, may be largely ineffective in promoting growth. First, institutions consistent with economic freedom are lacking throughout much of the developing world.

The growth literature suggests that institutions more consistent with

economic freedom are a major determinant of growth and prosperity. The results of the previous chapter are also supportive of this view. Therefore, it is unlikely that aid can promote growth in an institutional environment that is largely inconsistent with both growth and poverty reduction. Second, Vasquez (2003) suggests and Pitlik and Wirth (2003) find evidence indicating that countries often undertake productive reforms in response to various crises. If this is true then aid, which is often given to recipient countries in a time of crisis, could soften the impact of a crisis and reduce the need for productive reforms. In short, aid may reduce the urgency for reforms that lead to increased economic freedom. Third, the primary goal of the international aid agencies is to move as much aid as possible from developed countries to those that are less developed. This is evidenced by the former head of the World Bank urging a doubling of aid flows (Easterly 2003) and the constant push by international aid organizations for the developed world to meet the 0.7% target.21 Aid agencies forfeit the ability to promote productive reforms in developing countries when maximizing aid flows is their primary objective. Any attempt to use aid as a carrot is correctly perceived by recipient countries as not credible. This interaction between aid agencies and recipients has been described as a ritual dance (Vasquez 1998). The aid agencies structure the aid package in such a way as to encourage reform. The recipient country promises reform, but reneges once the aid

21

The 0.7% target is a goal where 0.7% of the developed world’s GNP is directed toward aid (Official Development Assistance more specifically). See the website of the Millennium Project, “the 0.7% target: An in-depth look”: http://www.unmillenniumproject.org/press/07.htm

82

has been disbursed. The dance repeats so long as the primary aim of the aid agencies is “moving money” (Easterly 2003). While this is by no means an exhaustive list of why aid may be ineffective in promoting growth, it does highlight several flaws underlying the motivation for aid. Addressing these issues does not appear to be a high priority of the aid community. This suggests that the disbursement of aid in the future will most likely have little impact on economic growth in the developing world.

5.2 Foreign Aid and Poverty Rajan and Subramanian (2008) found no statistically significant relationship between foreign aid and long-term economic growth. This does not imply, however, that the same is true for aid and poverty. Quite often aid is channeled to poor countries for the alleviation of poverty. The Millennium Development Goal of using aid to halve the percentage of people living in extreme poverty between 1990 and 2015 is such an example. In addition, humanitarian aid is sent to countries during times of natural disasters to assist the poor. The possibility that aid can have an impact on those living in poverty is very real. On a trip to Ethiopia, William Easterly, a former economist for the World Bank, witnessed one such example. … I visited a project of a British aid organization called Water Aid, which receives funds from official aid agencies. Water Aid has put in a water pipe to carry clean water from springs on top of the mountains bordering the Great Rift Valley to villages down in the Valley. The project was run entirely by Ethiopians, with representatives from the villages on the board of the agency. At a bustling water tap in one village, the villagers watered their cattle and collected drinking water for a nominal fee paid to Water Aid, to be used for maintenance of the system. Previously, the villagers had walked every other day two miles to collect water from a polluted river that transmitted disease. Children had been kept out of school, farmers kept out of farming, all to pursue the all-consuming and backbreaking task of fetching water.

With the new water pipe, life was better

(Easterly 2003, 40).

83

While reducing poverty in the developing world is a significant impetus for aid, very few papers have examined the relationship between aid and poverty.22 Collier and Dollar (2002) were the first to explore this relationship and based much of their analysis upon the results of Burnside and Dollar (2000), which found that aid facilitated growth in countries with good policies. Collier and Dollar indicated that aid, operating through increased economic growth, was responsible for lifting approximately 10 million people out of extreme poverty each year. They argued that aid could be more effective if it was allocated according to the quality of policies in recipient countries. If the aid agencies adopted their aid allocation strategy they believed roughly 19 million people could be lifted out of poverty each year rather than 10 million. Chong, Gradstein, and Calderon (2009) used the second-generation World Bank poverty rates to examine the impact of aid on both poverty and income inequality. Utilizing GMM-IV panel estimation, which accounts for possible endogeneity through lagged values of the regressors, they failed to find a robust statistical relationship between aid and poverty or inequality. However, establishing a robust statistical relationship with the GMM-IV estimator is difficult as it is an inefficient estimator. These two papers comprise the extent of the aid-poverty literature. While Collier and Dollar found a relationship between aid and poverty through growth, this result appears uncertain in light of other findings that aid exerts very little measurable impact on growth. In addition, Chong et al. achieved inconclusive results. The remainder of this chapter seeks to further this literature in the following manner.

The latest poverty data from the World Bank, which

encompass more countries than the previous two papers discussed as well as a larger time period, is used.23 In addition, both cross-section and panel estimation methods are used. Lastly, the analysis accounts for the impact of both economic and political institutions.

5.3 Empirical Framework The previous aid-poverty literature relied upon an empirical framework from the aid-growth literature. This approach will be used here. Both panel estimation methods as well as cross22

Boone (1996) was published before the World Bank poverty rates became available. Instead he utilized public health and quality of life measures as proxies for poverty. As a result his paper will not be discussed here. However, it is worth mentioning that Boone found no impact of aid on these various measures. 23 For a thorough discussion of the latest World Bank poverty rates see chapter 2.

84

sectional regressions will be used in a cross-country analysis of foreign aid and poverty over the period 1980-2005.24 The analysis can be thought of as addressing three general questions. The first is what factors contribute to the level of aid developing countries receive. Has aid been channeled to countries with better institutions or government policies as many aid advocates claim?

In addition, investigating the determinants of aid will suggest which variables are

potentially endogenous in regressions involving aid, institutions, and poverty. Equation (1) specifies the estimation equation used to explore this question.

(1) Aidit = α + ηPovit + δEFWit + θPolit + γdt + uit

The dependent variable is the average level of aid received by country i over a five or tenyear period, t. There are five periods in the five-year pooled OLS regressions: 1981-1985, 19861990, 1991-1995, 1996-2000, and 2001-2005. The ten-year regressions contain two periods: 1986-1995 and 1996-2005. Povit is the level of poverty at the beginning of the five or ten-year period. In the first period of the five-year regressions this corresponds to the poverty rate in 1980. EFWit is the level of economic freedom and Polit is a measure of the political institutions at the beginning of the period as measured by the Economic Freedom of the World (EFW) index and the Freedom House Political Rights index, respectively. dt is a time dummy used to capture time-varying effects and uit is the white noise error term. For a thorough explanation of the poverty, economic freedom, and political institutions variables see section 3.2 of chapter three. The second question is whether foreign aid has had any impact on economic and political institutions during 1985-2005.

The previous chapter indicates that movements toward

institutions more consistent with economic freedom contribute to reductions in the poverty rate. If aid exerts an influence upon institutions this suggests both a direct and indirect channel through which aid can reduce poverty. Investigation of this secondary channel is a method of evaluating whether the SA loans, the preferred method of aid funding during the 1990s, were effective. Equations (2) and (3) contain the regression equations used in this portion of the analysis. (2) ΔEFW = α + βAid + δX + u 24

The measure of foreign aid spans 1975-2005 in some of the regressions where lagged values are used.

85

(3) ΔEFWit = α + βAidit + δXit + γdt + uit

Equation (2) is a cross-section equation used to examine the impact of foreign aid on economic freedom over the period 1985-1995 and 1995-2005, while equation (3) is used in the pooled OLS regressions. The dependant variable is the change in the EFW index during either of the ten-year periods just mentioned. This variable has been signed such that increases in economic freedom over the period correspond to a positive value while decreases are negative. Aid and Aidit are as before, the average level of aid over the ten-year period in country i, however, both the current period average as well as the previous period are used in the regressions. X and Xit are matrices of covariates containing the level of economic freedom and political institutions at the start of the time period as well as the Sachs geography variables. The remaining variables are the time dummy and the white noise error term. The third and final question is whether aid contributed to the reductions in poverty rates that occurred between 1980-2005.

Determining whether aid has been an effective method of

alleviating poverty is the primary aim of this chapter as it has been a significant reason for the provision of aid. Equation (4) contains the regression equation used in the final section of the analysis. (4) ΔPovit = α + βAidit-y + δXit + γdt + uit

Here the dependent variable is the reduction in the extreme and moderate poverty rate over five-year periods from 1980-2005 and ten-year periods from 1985-2005 for country i. The independent variable of interest in this equation is the average level of foreign aid, Aidit-y, over the previous five or ten-year period. The y in the t-y subscript of the variable is meant to denote a time lag of five years for the regressions for 1980-2005 and ten years for the period 1985-2005. For example, the average level of aid during 1976-1980 corresponds to the change in poverty during 1980-1985 in the regressions with five-year time periods. The remaining variables, Xit, dt, and uit are the same as described above. As the first set of regressions will show, the higher a country’s poverty rate the more aid a country receives. This implies that endogeneity may be an issue when estimating the final equation, equation (4). In addressing this issue the previous literature has utilized lagged values

86

of aid, proxies for influence, and colony dummies as instruments for aid. As it seems likely that any potential impact of aid would take place over time, the previous values of aid will be used here. Aid has been used to build schools, create health facilities and factories, and to increase access to clean water. These projects impact a country’s population over time and it is likely that previous levels of aid will be better predictors. Before proceeding further, it is necessary to define what is meant by foreign aid. Foreign aid, in this analysis, is taken to be net Official Development Assistance (ODA) as defined by the OECD.

This definition is as follows: “Flows of official financing administered with the

promotion of the economic development and welfare of developing countries as the main objective, and which are concessional in character with a grant element of at least 25 percent (using a fixed 10 percent rate of discount). By convention, ODA flows comprise contributions of donor government agencies, at all levels, to developing countries (“bilateral ODA”) and to multilateral institutions.

ODA receipts comprise disbursements by bilateral donors and

multilateral institutions. Lending by export credit agencies—with the pure purpose of export promotion—is excluded (IMF 2003).” Loans that are concessionary in character are those that have below market interest rates. This definition excludes all military aid. Development aid that goes directly to NGOs, bypassing the government of donor countries, is included in ODA as long as it coordinated through the OECD. However, a majority of ODA goes directly to donor governments. While most aid comes from the member countries of the OECD, development aid from non-OECD members is also included in ODA as long as it meets the definition given above and is coordinated through the OECD. While most of the empirical literature on foreign aid uses ODA as the measurement of development assistance, several studies have used an alternative definition known as Effective Development Assistance (EDA). This measure was created by Chang, Frenandez-Arias, and Serven in 1999. The primary difference between ODA and EDA is the exclusion of loans and grants tied to technical assistance. Technical assistance is aid that must be used to build or implement a certain project in the recipient country, but the materials and expertise must be purchased from the donor country. Often this is a way for politicians in donor countries to create business for companies they favor. Chang et. al convincingly argue that aid of this type is of little benefit to the recipient country. This suggests that EDA would be a truer measure of development assistance. However, most studies use ODA as it is readily available from the

87

World Bank, World Development Indicators, and it is highly correlated with EDA.25 Several studies have used both ODA and EDA and concluded that there is little empirical difference between the two (Ovaska 2003; Burnside and Dollar 2004; Chong, Gradstein, and Calderon 2009). Therefore, the analysis here will follow the literature and use ODA as a share of Gross National Income (GNI) as the measure of foreign aid. While there are a number of countries that received aid during 1970-2005, the analysis here will be restricted to countries with poverty rate data in 1980 and with a 2005 population of more than one million. The 86 countries meeting these criteria are listed in table E.1 of appendix E. As the aid and political institutions variables have broad coverage, excluding countries that lack poverty data from this analysis ensures that regressions involving political institutions and foreign aid contain the same set of countries as regressions involving aid and poverty. In addition, excluding countries with low populations ensures that the analysis continues to cover a majority of the world’s poor. The aid data used in this analysis is averaged over five and tenyear periods. A country’s average level of aid is computed if two observations are present for the five-year periods and five observations for the ten-year periods. There are various theories pertaining to the relationship between aid and development. Theories where aid is supportive of the development process differ on whether aid should be used to kick-start development or as a reward for implementing productive reforms. The view that aid is ineffective contends that aid will primarily encourage rent seeking, both in the recipient and donor countries, and deter implementation of needed reforms in developing countries. Therefore, a preliminary examination of aid trends over time may suggest which theories better describe aid as it is actually implemented. This will also provide insight for the statistical results of the next section. Table 5.2 lists the twelve countries that received foreign aid funding of 10 percent or more of GNI during the period 1980-2005 in order of highest average aid receipt to the lowest. The first column lists the average over the period 1980-2005 and columns two through five list the average in each of the last four decades. This table indicates that there are a number of countries that have consistently received a large amount of aid since the 1970s. Only four countries on the list received an average level of aid less than 20 percent of GNI during 1980-2005. The country at the top of the list, Guinea-Bissau, is notable as their level of aid averaged almost 52 percent of 25

Ovaska (2003) indicates the correlation is “very close to one” and Easterly (2003) found it to be 0.93.

88

GNI during the 1990s. In addition, their average level of aid never fell below 20 percent of GNI during any of the decades. Table 5.2: Countries with high levels of aid during each decade (ten percent or more), 1980-2005

Country Guinea-Bissau Mozambique Gambia, The Malawi Mauritania

1980-2005 47.4 29.7 23.3 21.7 21.4

Average Aid (ODA as a share of GNI) 1970s 1980s 1990s 2000-2005 20.7 49.3 51.7 37.2 15.8 44.3 28.5 10.5 32.5 19.0 15.2 9.9 16.8 27.0 21.1 21.3 25.4 18.8 18.9

Burundi Rwanda Zambia Mali Niger

21.1 20.5 20.1 17.8 14.8

10.3 13.1 3.4 11.7 8.7

14.8 10.7 13.6 20.0 13.4

19.7 29.7 26.5 17.9 16.3

34.0 21.4 20.2 14.1 14.8

Burkina Faso Chad

13.9 12.9

8.8 7.8

11.7 12.5

16.6 15.0

13.2 10.0

RGDP per Average Level of Aid capita, PPP as a Share of (constant 2005 Government international Expenditures dollars) 1990s 2000-2005 2005 497 677 197.5 1,142 648 1,684 83.7 118.2 82.1

100.5 97.7 167.6 118.0

340 793 1,127 1,004 584 1,026 1,468

Columns six and seven of table 5.2 list the average level of aid as a share of government expenditures for the 1990s and during 2000-2005. As can be seen from the table, there are many missing observations due to a lack of government expenditure data for many developing countries. Despite the lack of data, it is apparent that aid is a substantial source of funds for recipient governments. Gambia’s 197.5 percent of aid as a share of government expenditures is the highest on the list, while Niger with 167.6 percent during the latter period, is not far behind. The country with the lowest level listed still had a value larger than 80 percent.26 The last column of the table lists the real GDP per capita for each of the countries in PPP adjusted 2005 international dollars. While these countries have received a large amount of aid during 19702005, the level of per capita income in these countries is still very low. Mauritania has the highest per capita income level of $1,684 while the income level of Burundi is the lowest. Burundi’s per capita income level of $340 indicates that the average citizen lived below the extreme poverty level in 2005. 26

Aid as a share of government expenditures higher than 100 percent could be a result of aid going to NGO’s, bypassing the recipient government. However, a majority of a country’s aid is received by the central government.

89

Table 5.3: Countries with increasing amounts of aid (eight percentage points increase or more), 1970-2005

Country Sierra Leone Congo, Dem. Rep. Liberia Burundi Zambia

RGDP per capita, PPP (constant Percentage 2005 international point increase dollars) Average Aid (ODA as a share of GNI) 1970-2005 2005 1970s 1980s 1990s 2000-2005 2.8 9.2 19.7 36.2 33.4 640 2.0 5.1 4.1 31.2 29.1 273 4.2 11.8 30.3 26.0 323 10.3 14.8 19.7 34.0 23.8 340 3.4 13.6 26.5 20.2 16.8 1,127

Nicaragua Guinea-Bissau Uganda Malawi Madagascar

3.1 20.7 1.7 9.9 3.5

8.4 49.3 6.6 16.8 8.8

30.7 51.7 15.9 27.0 13.2

19.9 37.2 14.1 21.1 13.8

16.8 16.4 12.4 11.1 10.3

2,311 497 901 648 882

Ghana Rwanda

2.8 13.1

6.3 10.7

9.9 29.7

12.6 21.4

9.9 8.3

1,193 793

Table 5.3 presents data similar to that of table 5.2 except that the twelve countries listed are those that exhibited an eight percentage point or more increase in aid as a share of GNI between 1970 and 2005. The countries are sorted in order of the highest percentage point increase during the period. The first four columns list the level of aid for each decade, while the fifth column lists the corresponding percentage point increase. Five of the countries listed on the previous table – Burundi, Zambia, Guinea-Bissau, Malawi, and Rwanda – are also found on this table, indicating these countries had both high and increasing levels of aid during 1970-2005. In addition, eight of the twelve countries listed had aid levels larger than 20 percent of GNI during 2000-2005. The last column of the table lists the level of real GDP per capita in 2005. Similar to the previous table, these per capita income levels are extremely low. Nicaragua, the only nonAfrican country, has a per capita income level of $2,311, the highest on the list. Only three countries have a per capita income level greater than $1,000. One country, the Democratic Republic of Congo, has an income level of $273, implying that the average citizen of the Democratic Republic of Congo lives well below the extreme poverty line. This table indicates that in spite of increasing receipt of foreign aid during 1970-2005, these countries remained poor in 2005.

90

The final table of this section, table 5.4, lists the four countries that have decreased their receipt of aid as a share of GNI by eight percentage points or more during 1970-2005. The columns of this table are identical to that of the previous table, 5.3. This table is noticeably different from the two previous tables in the following ways. First, the number of countries with decreasing aid levels is three times less than the number that have high or increasing levels of aid. Second, only one country in table 5.4, Cambodia with an aid level of 10.5 percent, received a substantial amount of aid during 2000-2005. Third, Botswana is the only African country that appears on this list while the previous two tables are populated entirely by African countries excluding Nicaragua in table 5.3. Lastly, the per capita income levels in 2005 are significantly higher than those in the other tables. Botswana’s income is roughly four times higher than that of Nicaragua, which had the highest income level of any country in the previous two tables.

Table 5.4: Countries with decreasing amounts of aid (a decline of eight percentage points or more as a share of GNI), 1970-2005

Country Jordan Botswana Egypt, Arab Rep. Cambodia

RGDP per capita, PPP (constant Percentage 2005 international point decrease dollars) Average Aid (ODA as a share of GNI) 1970s 1980s 1990s 2000-2005 1970-2005 2005 20.7 14.4 10.2 6.5 14.2 4,342 13.3 8.5 2.4 0.5 12.7 12,088 10.4 5.4 5.9 1.4 9.1 4,319 18.6 11.3 10.5 8.1 1,443

Only the 86 countries with populations larger than 1 million in 2005 and poverty data in 1980, listed in table E.1 of the appendix, are included in these tables. While these restrictions exclude several countries, the excluded countries are by and large island countries with small populations. While tables 5.2-5.4 only provide preliminary analysis, they do suggest several things. First, countries that received large amounts of aid in previous decades continue to do so. Second, many countries increased the level of aid received, while few decreased the level of aid received. Third, countries that are highly dependent upon aid had low per capita income in 2005, especially countries in sub-Saharan African. Botswana appears to be the only exception to this rule. The high levels of aid as both a share of national income and government expenditures is consistent with the view that the institution of foreign aid retards reform in developing countries 91

and is susceptible to rent seeking. While these results are suggestive they do not constitute a robust investigation of the relationship between foreign aid and poverty. This is the focus of the next section.

5.4 Results The previous section highlighted three questions. The first is, what factors influence the level of aid a country receives? The next four tables address this. Table 5.5 contains five pooled OLS regression equations examining the influence that extreme poverty, economic freedom, and political institutions had on aid flows during 1981-2005. The dependent variable in each of the regressions is the five-year average of foreign aid (ODA) as a share of GNI. This constitutes five periods in the pooled regressions: 1981-1985, 1986-1990, 1991-1995, 1996-2000, and 20012005. The independent variables, excluding the period dummies, are the initial value at the beginning of the five-year period. The period dummies are included to capture time-varying effects throughout the twenty-five year period. The base period for the dummy variables is 1991-1995, the period with the highest average level of aid. This and subsequent tables presents the standard errors in parenthesis. These are robust to heteroskedasticity and are clustered by country in the pooled regressions. The table indicates that the extreme poverty rate is a significant determinant of foreign aid dollars as its coefficient is significant at the one percent level in all regressions. The lowest tvalue for the extreme poverty rate in these regressions is 7.04. Column one contains only the poverty level at the beginning of each five-year period and the period dummy variables as independent variables.

This specification alone explains 44 percent of the cross-country

variation in aid levels. The coefficient on the extreme poverty level indicates that, ceteris paribus, a ten percentage point higher level of extreme poverty at the beginning of the five-year period corresponds to a 1.8 percentage point increase in the level of aid as a share of GNI over the period. As the average level of aid dispersed during 1991-1995, the period with the highest level, was 7.58 percent, a 1.8 percentage point increase of aid is substantial. The subsequent columns of table 5.5 add the level of economic freedom and various measures of political institutions to the regression equation. The addition of these variables does little to increase the explanatory power of the regression as the r-squared value increases from 0.44 to 0.46, even in the regressions containing both economic freedom and measures for political institutions,

92

columns three through five.27 There are two points worth noting about regressions two through five. The first is that the coefficients for economic freedom and the political measures are marginally significant throughout the regressions. The EFW index is significant at the ten percent level in columns two and three, but insignificant in four and five. The constraints on the executive measure from the Polity IV index is just significant at the ten percent level in column four while the remaining political institutions variables are not significant.

The second

observation is that when the economic freedom measure and political institutions measures are significant, the coefficient is negative implying that, on average, more aid is directed toward countries with less economic freedom and less democratic institutions. Before moving to the next table, it is worth briefly discussing the period dummy variables in the regression equations of table 5.5. These variables are significant at the five percent level or higher throughout the regressions except for the period 2001-2005, which is only significant at the ten percent level in the first regression. As was mentioned previously, 1991-1995 is the base for the period dummy, corresponding to the period with the highest level of aid and resulting in negative coefficients for the other dummy variables. This aid trend, increasing until the mid 1990s and then decreasing later in the period, is consistent with the previous literature.28 This trend is still observable even after controlling for the level of poverty. Table 5.6 contains regressions identical to those of table 5.5, except the moderate poverty rate is used in place of the extreme poverty rate. The implications of this table are similar to the previous table. During 1981-2005 the moderate poverty rate had a significant impact on aid flows, as the moderate poverty coefficient is significant at the one percent level in all regressions.

The higher the poverty rate the higher the level of aid received.

While the

executive constraints variable is the only political measure significant in these regressions, the economic freedom level is now significant at the ten percent level or higher in columns two through five. But, again, the sign of the coefficient is negative implying that countries with less economic freedom on average receive higher levels of aid. The r-squared values from the regressions with the moderate poverty rate, those of table 5.6, are lower than the regressions with 27

While the inclusion of the economic freedom and political institutions variables adds little explanatory power, a likelihood ratio test rejects the null that these variables should be excluded. 28 See Ovaska (2003) and Dichter (2005). The five-year average of aid as a share of GNI for the 86 countries of table E.1 in the appendix was: 1976-1980 (3.74), 1981-1985 (4.20), 1986-1990 (6.29), 1991-1995 (7.58), 1996-2000 (4.82), and 2001-2005 (5.68). This slump in aid levels during the latter part of the period was partly responsible for the renewed effort of the aid agencies to significantly increase aid levels.

93

the extreme poverty rate found in table 5.5. This suggests that not only is poverty a major determinant of aid, but that the severity of poverty also influences the flow of aid.

Table 5.5: Determinants of foreign aid: Pooled OLS regressions with extreme poverty, economic freedom, and political institutions, 1981-2005 (five-year average)

Independent variable Extreme poverty rate, beginning of period

Dependent variable: Average verage foreign aid (ODA as a share of GNI), five-year average (1) (2) (3) (4) (5) 0.18 *** 0.16 *** 0.16 *** 0.16 *** 0.16 *** (0.02) (0.02) (0.02) (0.02) (0.02)

EFW, beginning of period

-1.01 * (0.54)

Polity IV, beginning of period

-0.97 * (0.56)

-0.90 (0.56)

-0.08 (0.07)

Executive constraints, beginning of period

-0.38 * (0.23)

Political rights, beginning of period Period dummies: 1981-1985

-0.92 (0.56)

-0.28 (0.21) -4.13 *** -4.33 *** -4.57 *** -4.67 *** -4.41 *** (0.91) (0.94) (0.95) (0.96) (0.94)

1986-1990

-1.56 ** (0.65)

1996-2000

-2.63 *** -2.23 *** -2.03 *** -2.04 *** -2.22 *** (0.58) (0.53) (0.50) (0.51) (0.51)

2001-2005

-1.44 * (0.77)

-0.63 (0.74)

-0.42 (0.72)

-0.47 (0.71)

-0.60 (0.73)

1.90 ** (0.78)

7.87 ** (3.71)

7.82 ** (3.77)

9.02 ** (3.73)

8.61 ** (3.71)

Intercept

-1.84 ** (0.75)

-2.01 *** -2.08 *** -1.88 *** (0.76) (0.76) (0.75)

R2 (Adjusted)

0.44

0.46

0.46

0.46

0.46

Number of observations

345

345

345

345

345

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors, clusterd by country, country, are listed in parenthesis.

94

Table 5.6: Determinants of foreign aid: Pooled OLS regressions with moderate poverty, economic freedom, and political institutions, 1981-2005 (five-year average) Dependent variable: Average verage foreign aid (ODA as a share of GNI), five-year average Independent variable (1) (2) (3) (4) (5) M
    

  
0.15 *** 0.13 *** 0.13 *** 0.12 *** 0.13 *** (0.02) (0.02) (0.02) (0.02) (0.02) E, beginning of period

-1.11 ** (0.57)

Polity IV, beginning of period

-1.06 * (0.57)

-0.99 * (0.57)

-0.10 (0.07)

E     

  


-0.42 * (0.24)

Political rights, beginning of period P

  1981-1985

-1.01 * (0.57)

-0.31 (0.23) -3.85 *** -4.10 *** -4.39 *** -4.50 *** -4.20 *** (0.90) (0.94) (0.96) (0.97) (0.94)

1986-1990

-1.53 ** (0.66)

1996-2000

-2.73 *** -2.28 *** -2.05 *** -2.07 *** -2.27 *** (0.58) (0.53) (0.51) (0.51) (0.51)

2001-2005

-1.49 ** (0.75)

-0.60 (0.74)

-0.35 (0.74)

-0.41 (0.72)

-0.55 (0.73)

0.79 (0.79)

7.51 * (3.92)

7.49 * (3.97)

8.87 ** (4.01)

8.40 ** (3.98)

Intercept

-1.85 ** (0.75)

-2.04 *** -2.11 *** -1.88 *** (0.78) (0.78) (0.76)

R2 



0.40

0.42

0.42

0.43

0.42

N    

345

345

345

345

345

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors, clusterd by country, country, are listed in parenthesis.

While the regressions in tables 5.5 and 5.6 examine the determinants of aid over five-year periods, those in the following two tables, 5.7 and 5.8, focus on ten-year periods. These tables analyze the relationship between aid and poverty, economic freedom, and political institutions with two ten-year intervals during 1986-2005. It is possible that the length of the time period used influences the results, but tables 5.7 and 5.8 suggest this is not the case. The coefficients for both the extreme and moderate poverty rates are similar to those found in tables 5.5 and 5.6

95

in almost all of the regressions and all are significant at the one percent level. The economic freedom variable is significant at the ten percent level or higher in columns two through five and always negative. However, in these two tables, the political institutions measures are always insignificant. In addition, the same pattern of the r-squared values between the two tables is also observed. Regressions utilizing the extreme poverty rate explain slightly more of the crosscountry variation observed in aid allocations during 1986-2005, than regressions utilizing the moderate poverty rate.

Table 5.7: Determinants of foreign aid: Pooled OLS regressions with extreme poverty, economic freedom, and political institutions, 1986-2005 (ten-year average)

Independent variable Extreme poverty rate, beginning of period

Dependent variable: Average verage foreign aid (ODA as a share of GNI), ten-year average (1) (2) (3) (4) (5) 0.19 *** 0.17 *** 0.16 *** 0.16 *** 0.16 *** (0.02) (0.03) (0.03) (0.03) (0.03)

EFW, beginning of period

-1.36 ** (0.69)

Polity IV, beginning of period

-1.33 * (0.69)

-1.29 * (0.70)

-1.27 * (0.69)

-0.07 (0.07)

Executive constraints, beginning of period

-0.30 (0.25)

Political rights, beginning of period

-0.24 (0.24)

Period dummy, 1996-2005

-1.09 ** (0.51)

-0.18 (0.68)

0.15 (0.73)

0.17 (0.72)

-0.15 (0.67)

Intercept

0.52 (0.70)

8.21 * (4.21)

8.04 * (4.27)

9.04 ** (4.27)

8.75 ** (4.33)

R2 (Adjusted)

0.51

0.54

0.54

0.54

0.54

Number of observations

138

138

138

138

138

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors, clusterd by country, country, are listed in parenthesis.

96

Table 5.8: Determinants of foreign aid: Pooled OLS regressions with moderate poverty, economic freedom, and political institutions, 1986-2005 (ten-year average) Dependent variable: Average verage foreign aid (ODA as a share of GNI), ten-year average Independent variable (1) (2) (3) (4) (5) ! "#!$ !% &'())()' * "( 0.16 *** 0.13 *** 0.13 *** 0.12 *** 0.13 *** (0.02) (0.02) (0.02) (0.02) (0.02) +,-, beginning of period

-1.56 ** (0.70)

Polity IV, beginning of period

-1.53 ** (0.71)

-1.49 ** (0.71)

-1.47 ** (0.70)

-0.08 (0.08)

+./0!(# /)1!()!1% &'())()' * "(

-0.34 (0.27)

Political rights, beginning of period

-0.25 (0.26)

2( 033$, 1996-2005

-1.23 *** -0.17 (0.50) (0.70)

0.20 (0.78)

0.24 (0.77)

-0.13 (0.69)

Intercept

-0.51 (0.81)

8.43 * (4.49)

9.62 ** (4.60)

9.19 ** (4.61)

8.57 * (4.44)

R2 45601!7

0.45

0.48

0.48

0.49

0.48

803& * &1#!()1

138

138

138

138

138

Notes: *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors, clusterd by country, country, are listed in parenthesis.

Three implications are suggested by the results of the four tables just discussed, tables 5.55.8. First, the level of poverty appears to be the single most important factor influencing the level of foreign aid during 1981-2005. Second, the level of economic freedom was generally significant and negative throughout the regressions. However, this negative correlation is most likely the result of the impact of economic freedom on poverty, as highlighted in the previous chapter.

Countries with lower levels of economic freedom also tend to be poor.

As the

regressions just discussed indicate, higher levels of aid were directed to the poorest countries during the period. Thus, countries with low levels of economic freedom most likely received, on average, more aid as a result of high poverty and not as a result of poor quality institutions. But, the results do appear to suggest that no effort is made to restrict aid from countries with low

97

levels of economic freedom as well as those that are autocratic.29 Lastly, the aid literature has examined the determinants of aid and found population to be a significant explanatory variable. Smaller countries on average receive more aid per capita (Burnside and Dollar 2000). While population is not included in the regressions presented here, it was included in other regressions. The results of the regressions were unchanged, however. Therefore, population was excluded from this analysis. The previous set of tables examined the determinants of aid flows. The following three tables focus on the relationship between aid and economic freedom. The results of chapter three suggest that examining this relationship is important. Those results indicated that countries with low levels of economic freedom were generally poor. However, the results also demonstrated that poor countries that increased their level of economic freedom also had significant reductions in poverty. Thus, if aid exerts any impact on economic freedom, there is a possibility that it may indirectly influence poverty through economic freedom. The next set of tables examines the impact of foreign aid on economic freedom. Because this analysis utilizes ten-year periods, the period 1980-2000 or 1985-2005 could be used. As either period provided similar results the latter was chosen. The first two tables examine each of the ten-year periods while the third table includes the entire period in a pooled OLS regression. The first table, table 5.9, contains cross-country OLS regressions for the period 1985-1995. In these regressions a positive dependent variable indicates there was an increase in economic freedom over the period. Also, the level of economic freedom in 1985 is included as an independent variable in these regressions to control for the differing levels of economic freedom of countries at the beginning of the period. The first column of table 5.9 regresses the increase in economic freedom over 1985-1995 on the average level of aid as a share of GNI during the period, while the second column shows the results for aid in the previous ten-year period.30 In both regressions the coefficient of the average level of foreign aid as a share of GNI is negative and significant at the one percent level, suggesting that aid may have a negative influence on economic freedom. While the negative sign of the coefficient of the contemporaneous average aid level could be expected due to the relationship between economic freedom and poverty, the negative sign on the average level of aid during 1976-1985 is interesting. It indicates that 29 30

Others have come to the same conclusion; see Collier and Dollar (2002). The average level of aid was computed if there were five or more observations during the period.

98

countries that received a higher level of aid during 1976-1985 had, on average, slight reductions in economic freedom during the subsequent ten-year period. However, the size of the coefficient indicates that any impact was slight as a one percentage point increase in the average level of aid as a share of GNI over the period corresponded to a 0.04 and 0.06 decrease in the EFW index for columns one and two respectively.

Table 5.9: The impact of foreign aid on changes in economic freedom after controlling for political institutions and geographic/locational factors, 1985-1995 Independent variable EFW, 1985

Dependent variable: Change in Economic Freedom, 1985-1995 (1) (2) (3) (4) (5) (6) -0.55 *** -0.46 *** -0.51 *** -0.45 *** -0.63 *** -0.58 *** (0.12) (0.11) (0.11) (0.10) (0.12) (0.12)

Polity IV, 1985

0.04 *** 0.04 *** 0.03 (0.02) (0.02) (0.02) -0.06 *** (0.02)

Average aid (ODA as a share of GNI), 1976-1985 Average aid (ODA as a share of GNI), 1986-1995

-0.04 *** (0.01)

-0.04 * (0.02) -0.02 * (0.01)

0.03 * (0.02) -0.02 (0.02)

-0.02 (0.02)

Coastal population (% within 100km)

1.07 *** 1.02 *** (0.36) (0.37)

Tropical location (% area in tropics)

-0.22 (0.27)

-0.29 (0.26)

Distance to major markets a

0.08 (0.06)

0.06 (0.06)

Intercept

3.76 *** 3.31 *** 3.56 *** 3.27 *** 3.37 *** 3.25 *** (0.61) (0.64) (0.57) (0.58) (0.55) (0.70)

R2 (Adjusted)

0.29

0.28

0.36

0.36

0.42

0.41

Number of observations

69

68

69

68

69

68

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis. a

Columns three and four of table 5.9 add the Polity IV index, a measure of political institutions, to the regressions.31 The results change slightly with the inclusion of this variable. 31

The Polity IV index was chosen from among the other political measures because it yielded the strongest regression results.

99

First, the size and significance of the coefficient of the average level of aid as a share of GNI for both periods is reduced. However, the coefficients remain negative and significant at the ten percent level. Second, the explanatory power of the regressions increases from 0.29 and 0.28 in columns one and two to 0.36 in three and four. Lastly, the regressions indicate that more democratic political institutions were associated with increases in economic freedom over the period at the one percent significance level. This finding is similar to what was found in the previous chapter. However, the coefficient of 0.04 on the political measure in both regressions indicates that any impact was quite small. The last two regressions of table 5.9 indicate that, after controlling for the influence of geographic factors, the average level of aid as a share of GNI is no longer significant either in the contemporaneous or previous period. The inclusion of the Sachs geographic variables also reduces the significance of the political institutions measure, as it remains significant at the ten percent level in only the last regression. The explanatory power of the regression is increased however, with the inclusion of these variables.32 The final two regressions explain slightly more than 40 percent of the cross-country variation in increases of economic freedom during 19851995. Of the three Sachs variables included, only the variable representing the percentage of a country’s population that lives within 100 kilometers of the coast is significant. The sign of the coefficient and its significance suggest that increased access to trade routes is conducive to increases in economic freedom. Table 5.10 contains regressions identical to those of table 5.9, except that the time period of interest is now 1995-2005. Interestingly, changing the time period yields slightly different results. While in the previous table the coefficient of the average level of aid as a share of GNI, both in the contemporaneous period and the previous period was negative and significant in the first four regressions, this is not the case here. The sign of the coefficient of the average aid variable is positive in all six regressions, but it is not significant. A possible explanation for this sign change between the two ten-year periods is the influence of the Washington Consensus. This consensus held that development and economic growth were possible through democratic and free market political reforms. This view was fairly influential during the 1990s and into the early part of this century.

Foreign aid was ostensibly used to further this objective as is

32

A likelihood ratio test for the exclusion of the Sachs geography variables is rejected for the regressions of columns five and six, suggesting that the increase in explanatory power from these variables is significant.

100

evidenced by both the SA grants and selectivity of foreign aid discussed in the previous section. While the sign change of the aid coefficients between these two tables is not strong evidence for the influence of the Washington Consensus, it is however, one possible explanation. A thorough investigation of this link is not addressed here and is left for future research.33

Table 5.10: The impact of foreign aid on changes in economic freedom after controlling for political institutions and geographic/locational factors, 1995-2005 Independent variable EFW, 1995

Dependent variable: Change in Economic Freedom, 1995-2005 (1) (2) (3) (4) (5) (6) -0.30 *** -0.29 *** -0.27 *** -0.26 *** -0.31 *** -0.31 *** (0.07) (0.06) (0.07) (0.06) (0.07) (0.07)

Polity IV, 1995

-0.02 ** (0.01) 0.01 (0.01)

Average aid (ODA as a share of GNI), 1986-1995 Average aid (ODA as a share of GNI), 1996-2005

0.01 (0.01)

-0.02 ** (0.01)

-0.02 ** (0.01)

0.01 (0.01) 0.01 (0.01)

-0.02 ** (0.01) 0.02 (0.01)

0.02 (0.01)

Coastal population (% within 100km)

0.35 * (0.18)

0.37 ** (0.19)

Tropical location (% area in tropics)

-0.16 (0.17)

-0.17 (0.18)

Distance to major markets a

0.02 (0.03)

0.01 (0.03)

Intercept

2.19 *** 2.14 *** 2.06 *** 1.99 *** 2.18 *** 2.15 *** (0.43) (0.43) (0.41) (0.43) (0.41) (0.46)

R2 (Adjusted)

0.34

0.35

0.36

0.38

0.36

0.38

Number of observations

72

71

72

71

72

71

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis. a

An additional difference between tables 5.9 and 5.10 is the coefficient on the Polity IV index. In the previous table the coefficient indicated that countries with more democratic

33

It may be interesting to observe the sign of the aid coefficient when data for 2010 and 2015 become available. If the coefficient is negative in the later periods, this may indicate that the Washington Consensus did have some influence for a short period of time.

101

institutions in 1985, on average, had slightly larger increases in economic freedom over the subsequent ten-year period as compared to countries with less democracy. Here, the results indicate the opposite is true and are significant at the five percent level or higher. Two aspects of the results of table 5.10 are similar to the previous table.

First, the

coefficient of the political institutions measure is 0.02 in columns three through six indicating that any impact during the period was small. Second, when the Sachs geography measures are included, columns five and six, the coefficient of the coastal population variable, is again the only geographic variable that is significant at the ten percent level or higher. Table 5.11: The impact of foreign aid on changes in economic freedom after controlling for political institutions and geographic/locational factors (pooled OLS for ten-year periods, 19851995 and 1995-2005) Independent variable EFW, beginning of period

Dependent variable: Change in Economic Freedom, 1985-1995 and 1995-2005 (1) (2) (3) (4) (5) (6) -0.45 *** -0.40 *** -0.45 *** -0.41 *** -0.55 *** -0.53 *** (0.07) (0.07) (0.07) (0.07) (0.08) (0.08)

Polity IV, beginning of period

0.02 * (0.01)

Average aid (ODA as a share of GNI): Previous 10-year period Contemporaneous 10-year period

-0.01 (0.01) -0.02 * (0.01)

0.02 ** (0.01)

0.01 (0.01)

-0.01 (0.01) -0.02 (0.01)

0.01 (0.01) 0.00 (0.01)

-0.01 (0.01)

Coastal population (% within 100km)

0.80 *** 0.84 *** (0.23) (0.23)

Tropical location (% area in tropics)

-0.17 (0.17)

-0.24 (0.16)

Distance to major markets a

0.05 (0.04)

0.05 (0.04)

0.11 (0.11)

0.10 (0.11)

0.11 (0.11)

Period dummy, 1995-2005

0.07 (0.10)

Intercept

3.14 *** 2.83 *** 3.17 *** 2.92 *** 3.11 *** 2.96 *** (0.42) (0.45) (0.42) (0.45) (0.45) (0.43)

-0.01 (0.11)

0.01 (0.12)

R2 (Adjusted)

0.29

0.26

0.30

0.29

0.36

0.35

Number of observations

141

139

141

139

141

139

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively Heteroskedastic robust standard errors are listed in parenthesis. a

102

Table 5.11 combines the time periods of the two previous tables, 1985-1995 and 1995-2005 into a set of pooled OLS regressions for the period 1985-2005 with the dependent variable representing the increase in economic freedom over both of the ten-year periods. While the coefficient for the average level of foreign aid during the contemporaneous and previous period is negative in all but the last regression, it is largely insignificant throughout with the first column being the only exception. The coefficient of the Polity IV index is positive in this table, but is only statistically significant in regressions without the Sachs geography variables, columns three and four. Of the geography variables, again, only the coastal population variable is significant. These tables indicate that the evidence that foreign aid exerted a positive impact on economic freedom during 1985-2005 is weak, at best. This is in spite of efforts during the 1990s to use aid as a means to encourage market based reforms (Easterly 2003). The 1985-1995 period showed the strongest results for a statistically significant relationship. Those results however, indicate that higher levels of aid were associated with smaller increases, or possibly a decrease, in economic freedom over the period. But, this result was not robust as the inclusion of geographic variables eliminated all significance of the aid variables. The only robust result from these tables is the impact of the Sachs coastal population variable. Regressions containing this variable consistently indicated that, ceteris paribus, countries with a larger percentage of the population within 100 kilometers of the coast had larger increases in economic freedom during 1985-2005. This variable is largely a proxy for the transaction costs of accessing markets. Countries with a significant percentage of their population close to the coast will generally have cheaper transportation costs and better access to international markets. In these regressions it appears that countries with this attribute also experienced larger increases in economic freedom. While investigating this relationship is not the aim of this chapter, it is interesting nonetheless. The tables presented here did not include regressions with political institution measures as the dependent variable. Such regressions were analyzed, but the results were insignificant with very little explanatory power. Therefore, they are not included here. This analysis is not the first to investigate aid and institutions.

Several papers have

examined this relationship, however, not all focused on economic freedom. Heckelman and Knack (2009) found that aid was associated with an increase of some and a decrease of other sub-components of the EFW index. They found that the net result of the changes was positive

103

overall for growth. Knack (2004) found no statistical relationship between foreign aid and political institutions when using the Freedom House Political Rights index. Three papers (Knack 2001; Brautigam and Knack 2004; Heckelman and Knack 2008) found a slightly negative relationship between aid and institutions, however, only one used the EFW index and another focused exclusively on Africa. Based on the literature and the results presented here, there is little evidence that foreign aid exerts an independent positive impact on economic freedom, either contemporaneously or in the future. If a relationship does exist the results here indicate that aid may retard improvements in economic freedom. However, this result is not robust. Therefore, it does not appear that aid influences poverty indirectly through increases in economic freedom. The last, and most important, question this chapter seeks to investigate is whether the disbursement of foreign aid during 1976-2005 contributed to the reductions in poverty that occurred during 1980-2005. Foreign aid may help alleviate poverty despite the fact that the aid literature indicates that foreign aid has had little impact on economic growth. Table 5.12 contains pooled OLS regressions of five-year reductions in extreme poverty rate on aid, economic freedom, political institutions, and geographic and locational factors during 19802005. The dependent variable is the reduction in extreme poverty during, 1980-1985, 19851990, 1990-1995, 1995-2000, and 2000-2005. A positive value indicates a reduction in poverty during the period. The extreme poverty rate at the beginning of each five-year period is included as an independent variable to control for the differing initial poverty levels of the countries in the sample. The positive sign and high significance level of the initial extreme poverty rate in all the regressions of table 5.12 indicate that, ceteris paribus, poorer countries exhibited larger reductions in poverty during the period. This result is similar to the conditional convergence result observed in the empirical literature concerning institutions and growth. Column one of table 5.12 lists the results of five-year reductions in poverty regressed on the extreme poverty rate at the beginning of the period, the average level of foreign aid as a share of GNI during the previous five-year period, and period dummy variables. The period dummies are included to control for unobservable time-varying effects. The coefficient on the aid variable in this regression of -0.15 is just significant at the ten percent level and negative suggesting that higher levels of aid during a five-year period correspond to smaller reductions in poverty during the subsequent period. This result is extremely tenuous however as the inclusion of additional

104

independent variables eliminates the significance of the aid term. Columns two through five add the economic freedom index and the political institutions measures to the regressions. While the

Table 5.12: The impact of Foreign aid on reductions in the extreme poverty rate after controlling for political institutions and geographic/locational factors (pooled OLS for five-year periods, 1980-2005) Independent variable Extreme poverty rate, beginning of period

Dependent variable: Reduction in extreme poverty rate, five-year periods (1) (2) (3) (4) (5) (6) (7) (8) 0.07 *** 0.08 *** 0.08 *** 0.08 *** 0.09 *** 0.12 *** 0.12 *** 0.12 *** (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02)

Average aid (ODA/GNI), -0.15 * previous 5-year period (0.09)

-0.15 (0.09)

-0.14 (0.09)

-0.14 (0.09)

-0.14 (0.09)

-0.09 (0.07)

-0.10 (0.07)

-0.10 (0.08)

EFW, beginning of period

0.45 (0.34)

0.38 (0.35)

0.32 (0.35)

0.26 (0.37)

0.46 (0.36)

0.37 (0.37)

0.33 (0.38)

Polity IV, beginning of period

0.12 ** (0.06)

Executive constraints, beginning of period

0.19 *** (0.06) 0.38 * (0.19)

Political rights, beginning of period

0.49 ** (0.20) 0.52 ** (0.21)

0.65 *** (0.21)

Coastal population (% within 100km)

1.61 * (0.91)

Tropical location (% area in tropics)

-3.46 *** -3.05 *** -3.05 *** (0.85) (0.78) (0.82)

Distance to major markets a

-0.41 *** -0.39 ** (0.16) (0.16)

-0.43 *** (0.17)

Period dummies: 1981-1985

1.74 * (0.93)

1.41 (0.93)

2.19 * (1.33)

2.31 * (1.33)

2.68 ** (1.31)

2.67 ** (1.34)

2.47 * (1.32)

2.72 ** (1.27)

2.62 ** (1.31)

2.37 * (1.28)

1986-1990

0.46 (1.42)

0.59 (1.41)

0.83 (1.43)

0.84 (1.42)

0.66 (1.42)

1.01 (1.40)

0.94 (1.40)

0.73 (1.40)

1996-2000

1.08 (1.18)

0.87 (1.21)

0.59 (1.15)

0.68 (1.16)

0.88 (1.18)

0.32 (1.12)

0.53 (1.14)

0.78 (1.16)

2001-2005

2.59 ** (1.12)

2.21 * (1.16)

1.93 * (1.14)

2.07 * (1.14)

2.19 * (1.14)

1.80 (1.11)

2.07 * (1.12)

2.19 ** (1.12)

Intercept

-0.31 (0.88)

-2.93 (2.12)

-2.79 (2.21)

-4.00 * (2.20)

-4.20 * (2.25)

-0.74 (2.13)

-2.53 (2.00)

-2.59 (2.10)

R2 (Adjusted)

0.06

0.06

0.07

0.07

0.08

0.12

0.11

0.12

Number of observations

337

337

337

337

337

337

337

337

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors, clusterd by country, country, are listed in parenthesis.

a

105

average level of aid in the previous five-year period remains negative it is no longer significant after controlling for the level of economic freedom and political institutions at the start of each period. At first glance, the insignificance of the economic freedom level at the beginning of the five-year period in regressions two through five is surprising given the results of the previous chapter.

But this result indicates that the level of economic freedom exerts a statistically

insignificant impact on reductions in poverty after controlling for various factors. The previous chapter found that increases in economic freedom rather than levels had a substantial impact on poverty reduction.

Therefore, this result and those of the previous chapter indicate that

reductions in poverty are most affected by changes in economic freedom and not initial levels of either institutions or wealth. While the economic freedom variable is insignificant in these equations, the three variables representing political institutions are significant at the ten percent level or higher in regressions three through eight, even after the inclusion of the geography variables. This result implies, that after controlling for various factors, more democratic countries exhibited larger reductions in poverty during 1980-2005. Columns six through eight add the Sachs geography and locational measures to the regressions. Every geography variable is significant except for the coastal population variable in the last equation. The signs of the variables indicate that geographic factors are a significant impediment to poverty reduction. The positive coefficient of the coastal population variable is consistent with the view that countries with minimal or no coastline confront higher transaction costs when transporting goods or attempting to engage with international markets. Malaria and other common diseases in the tropics increase mortality rates and lower productivity levels of individuals, which is consistent with the negative coefficient on the tropics term. And lastly, the negative coefficient of the air distance term suggests that a greater distance to global markets increases the transaction costs of international trade. All these factors make economic growth more difficult to achieve and hence are less conducive for reductions in poverty. Table 5.13 lists regressions identical to those contained in table 5.12, except that the dependent variable is now the reduction in the moderate poverty rate over five-year periods during 1980-2005. The coefficient of the average level of aid as a share of GNI in the previous period has results similar to those of table 5.12. The coefficient is negative and significant at the

106

ten percent level in the first regression, but than insignificant thereafter. Thus, while the first regression indicates that aid hinders reductions in moderate poverty, this result is not robust to

Table 5.13: The impact of Foreign aid on reductions in the moderate poverty rate after controlling for political institutions and geographic/locational factors (pooled OLS for five-year periods, 1980-2005) Independent variable Moderate poverty rate, beginning of period

Dependent variable: Reduction in moderate poverty rate, five-year periods (1) (2) (3) (4) (5) (6) (7) (8) 0.04 *** 0.05 *** 0.05 *** 0.05 *** 0.05 *** 0.08 *** 0.07 *** 0.07 *** (0.01) (0.01) (0.02) (0.02) (0.01) (0.02) (0.01) (0.01)

Average aid (ODA/GNI), -0.13 * previous 5-year period (0.08)

-0.12 (0.08)

-0.11 (0.08)

-0.11 (0.08)

-0.11 (0.08)

-0.06 (0.06)

-0.06 (0.06)

-0.07 (0.07)

EFW, beginning of period

0.82 ** (0.40)

0.77 * (0.41)

0.77 * (0.41)

0.72 * (0.41)

0.80 * (0.41)

0.75 * (0.43)

0.72 * (0.42)

Polity IV, beginning of period

0.08 (0.06)

0.13 ** (0.06)

Executive constraints, beginning of period

0.17 (0.19)

Political rights, beginning of period

0.24 (0.20) 0.28 (0.20)

0.38 * (0.19)

Coastal population (% within 100km)

1.79 * (1.03)

Tropical location (% area in tropics)

-3.32 *** -2.98 *** -3.00 *** (0.78) (0.74) (0.75)

Distance to major markets a

-0.25 (0.16)

-0.22 (0.16)

-0.25 (0.17)

Period dummies: 1981-1985

2.01 * (1.06)

1.78 * (1.07)

1.69 (1.39)

1.92 (1.40)

2.18 (1.42)

2.09 (1.45)

2.02 (1.40)

2.26 (1.39)

2.08 (1.43)

1.97 (1.37)

1986-1990

1.27 (1.38)

1.49 (1.37)

1.66 (1.40)

1.61 (1.40)

1.54 (1.38)

1.81 (1.38)

1.69 (1.39)

1.59 (1.36)

1996-2000

2.12 ** (1.02)

1.73 * (1.03)

1.54 (1.02)

1.65 (1.01)

1.74 * (1.03)

1.30 (1.00)

1.51 (1.00)

1.63 (1.02)

2001-2005

2.97 *** 2.29 * (1.24) (1.17)

2.10 * (1.25)

2.23 * (1.24)

2.28 * (1.24)

2.04 * (1.24)

2.29 * (1.24)

2.34 * (1.24)

Intercept

-0.42 (0.90)

-5.34 ** (2.52)

-5.27 ** (2.57)

-5.84 ** (2.65)

-6.10 ** (2.61)

-3.94 * (2.37)

-5.01 ** (2.45)

-5.15 ** (2.43)

R2 (Adjusted)

0.03

0.03

0.04

0.03

0.04

0.07

0.06

0.07

Number of observations

337

337

337

337

337

337

337

337

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors, clusterd by country, country, are listed in parenthesis.

a

107

the inclusion of additional explanatory variables. In this table, and the previous one, after controlling for other factors, the average level of aid in a particular five-year period appears to have little impact on reductions in poverty in the subsequent period. Interestingly, the level of economic freedom at the beginning of the period is now significant in all the regressions in which it is included. It also indicates that the level of economic freedom has a positive impact on reductions in moderate poverty, while the previous table indicated the impact on extreme poverty was statistically insignificant. The political institutions measures are less significant in table 5.13 than they were in the previous table. While these measures were significant in columns three through eight of the previous table, only two of the three measures are significant in these regressions, columns six through eight. Lastly, while only two of the three geographic factors are significant in the last three regressions of table 5.13, they indicate that unfavorable geographic and locational factors retard reductions in moderate poverty as well. The explanatory power of the regressions of tables 5.12 and 5.13 is fairly low. The adjusted r-squared values of the first table range from 0.06 to 0.12 and those of the later table, 0.03 to 0.07, are even lower. This indicates that while the regressions are informative, they explain very little of the cross-country variation in poverty reductions during 1980-2005. The focus of the previous two tables was on how lagged values of aid affect future reductions in poverty. While one could argue that using the lagged value of aid does not capture its true impact on poverty, this seems unlikely for the following reason. Whether aid is given for economic development or to reduce the affects of poverty, there will be a time lapse between when the aid is disbursed and its actual economic impact. Often, aid allocated for economic development is used to develop a country’s infrastructure.

Infrastructure projects, even in

developed countries, take several years to implement indicating that any economic impact occurs several years in the future. Aid used specifically for poverty alleviation is often put toward improving access to clean water, malaria prevention, and regional health clinics.

While

implementing these projects most likely occurs at a faster pace than infrastructure projects, they still take time. Therefore, any impact of aid on poverty reduction will likely occur a number of years after the aid money is given to the recipient government. Moreover, the initial set of regressions of this chapter indicates that the level of aid is endogenous in regressions where the dependent variable is poverty, and where the two variables are contemporaneous. Using the,

108

level of poverty in the previous period, while arguably a better way to ascertain the true impact on poverty, should reduce the endogenous variable problem. It is possible that five-year periods are too short to ascertain the impact of foreign aid on poverty reduction. The long run impact of aid is arguably more important after all. In order to investigate the possibility that aid may impact poverty over a longer time frame, the next set of tables examines reductions in poverty over ten-year periods.

Table 5.14: The impact of Foreign aid on reductions in the extreme poverty rate after controlling for political institutions and geographic/locational factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005) Independent variable Extreme poverty rate, beginning of period

Dependent variable: Reduction in extreme poverty rate, 1985-1995 and 1995-2005 (1) (2) (3) (4) (5) (6) (7) (8) 0.13 *** 0.14 *** 0.15 *** 0.16 *** 0.16 *** 0.21 *** 0.21 *** 0.21 *** (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04)

Average aid (ODA/GNI), -0.15 previous 10-year period (0.17)

-0.14 (0.17)

-0.14 (0.17)

-0.14 (0.17)

-0.16 (0.17)

-0.07 (0.14)

-0.08 (0.14)

-0.11 (0.15)

EFW, beginning of period

0.89 (0.57)

0.84 (0.58)

0.76 (0.58)

0.50 (0.61)

0.70 (0.63)

0.59 (0.64)

0.40 (0.64)

0.17 * (0.09)

Polity IV, beginning of period Executive constraints, beginning of period

0.26 *** (0.10) 0.62 ** (0.31)

0.77 ** (0.34) 1.09 *** (0.42)

Political rights, beginning of period

1.27 *** (0.43)

Coastal population (% within 100km)

3.73 * (2.13)

3.81 * (2.17)

3.00 (2.11)

Tropical location (% area in tropics)

-6.12 *** -5.51 *** -5.56 *** (1.94) (1.88) (1.86)

Distance to major markets a

-0.42 (0.35)

-0.40 (0.35)

-0.51 (0.36)

Period dummy, 1995-2005 2.91 ** (1.40)

2.30 (1.49)

1.58 (1.45)

1.59 (1.39)

2.28 (1.42)

1.21 (1.45)

1.49 (1.43)

2.29 (1.42)

Intercept

-0.26 (0.98)

-5.30 (3.41)

-4.95 (3.50)

-7.06 * (3.70)

-7.78 ** (3.70)

-1.39 (3.33)

-4.46 (3.24)

-4.70 (3.21)

R2 (Adjusted)

0.12

0.12

0.13

0.14

0.17

0.21

0.20

0.24

Number of observations

136

136

136

136

136

136

136

136

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors, clusterd by country, country, are listed in parenthesis.

a

109

Table 5.14 contains regressions of the reduction in the extreme poverty rate on foreign aid, economic freedom, political institutions, and geographic factors over ten-year periods during 1985-2005. The structure of these regressions is identical to those in tables 5.12 and 5.13. The extreme poverty rate at the beginning of the period is included as a control variable and the economic freedom and political institutions measures are measured at the start of each ten-year period. The average level of aid as a share of GNI used in these regressions is for the ten-year period just prior to the period of interest. Therefore, the average level of aid for the period 19761985 is associated with changes in poverty during 1985-1995 and the level of aid for 1986-1995 corresponds to poverty changes during 1995-2005. The relationship between aid and reductions in the extreme poverty rate is largely unchanged when the time period is ten years rather than five. The coefficient for the average level of foreign aid as a share of GNI in the previous ten-year period is negative in all regressions, but also insignificant. Again, it appears that the level of economic freedom is less important for reductions in the extreme poverty rate as the coefficient on the EFW index is insignificant in columns two through eight. The measures of a country’s political institutions however, are positive and significant at the ten percent level or higher in the last six regressions of table 5.14. This is similar to that of the table 5.12, which uses five-year periods. After controlling for other factors, more democratic countries appeared to have larger reductions in extreme poverty during 1985-2005. The last three regressions of the table again illustrate the influence of geographic factors. The coastal population variable is significant in two of the three regressions, while the coefficient of the distance to major markets variable is insignificant. The tropics variable, however, is significant at the one percent level and negative indicating that reductions in extreme poverty are difficult to achieve in areas susceptible to tropical diseases. Table 5.15 presents the results for the moderate poverty rate. The structure of this table is identical to that of the previous table, except now the dependent variable is the ten-year reduction in the moderate poverty rate during 1985-2005. In these regressions the level of aid in the previous period has a more significant negative result than in the previous tables. Ceteris paribus, a ten percentage point higher level of aid as a share of GNI during a ten-year period corresponds to a smaller decrease or possibly an increase in moderate poverty over the subsequent period of 2.9 to 3.0 percentage points, after controlling for economic and political institutions. This result becomes insignificant, however, in the last three equations of the table

110

after the geography and locational variables are included. Similar to the five-year regressions with reductions in moderate poverty, table 5.13, the level of economic freedom at the beginning

Table 5.15: The impact of Foreign aid on reductions in the moderate poverty rate after controlling for political institutions and geographic/locational factors (pooled OLS for ten-year periods, 1985-1995 and 1995-2005) Independent variable Moderate poverty rate, beginning of period

Dependent variable: Reduction in moderate poverty rate, 1985-1995 and 1995-2005 (1) (2) (3) (4) (5) (6) (7) (8) 0.10 *** 0.12 *** 0.13 *** 0.13 *** 0.13 *** 0.17 *** 0.17 *** 0.17 *** (0.03) (0.03) (0.03) (0.03) (0.03) (0.04) (0.04) (0.04)

Average aid (ODA/GNI), -0.31 * previous 10-year period (0.17)

-0.29 * (0.17)

-0.29 * (0.17)

-0.29 * (0.17)

-0.30 * (0.17)

-0.20 (0.15)

-0.21 (0.15)

-0.22 (0.16)

EFW, beginning of period

1.73 ** (0.74)

1.70 ** (0.76)

1.67 ** (0.76)

1.51 * (0.78)

1.43 * (0.83)

1.35 (0.84)

1.24 (0.84)

Polity IV, beginning of period

0.12 (0.11)

0.19 * (0.11)

Executive constraints, beginning of period

0.31 (0.34)

Political rights, beginning of period

0.39 (0.37) 0.65 * (0.37)

0.76 ** (0.38)

Coastal population (% within 100km)

4.65 * (2.83)

Tropical location (% area in tropics)

-5.69 *** -5.22 *** -5.28 *** (2.01) (1.99) (1.99)

Distance to major markets a

-0.25 (0.43)

-0.20 (0.43)

-0.28 (0.43)

2.19 (1.70)

2.60 (1.69)

2.99 * (1.68)

-9.75 ** (4.22)

-10.12 ** (4.22)

2.92 * (1.68)

4.91 * (2.90)

4.38 (2.88)

Period dummy, 1995-2005 4.17 *** 2.94 * (1.45) (1.67)

2.43 (1.71)

2.58 (1.66)

Intercept

-0.32 (1.41)

-10.23 ** (4.57)

-11.35 *** -12.08 *** -7.84 * (4.52) (4.54) (4.37)

R2 (Adjusted)

0.09

0.11

0.11

0.11

0.12

0.16

0.16

0.17

Number of observations

136

136

136

136

136

136

136

136

-10.39 ** (4.41)

Notes: The minimum air distance in thousands of kilometers from a country to any one of the following major markets: New York, ork, Tokyo, or Amsterdam. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. respectively. Heteroskedastic robust standard errors, clusterd by country, country, are listed in parenthesis.

a

of the period is more significant, while the level of political institutions are less so. Overall, whether the time period is five or ten years, the level of economic freedom appears to be more significant for reductions in moderate poverty while the level of democracy appears to matter

111

more for reductions in extreme poverty. Again, the Sachs variables continue to indicate that geographic factors matter regarding reductions in both extreme and moderate poverty, even after controlling for the level of aid and institutions. These two tables explain very little of the cross-country variation in poverty rate reductions during 1985-2005. While the adjusted r-squared values are slightly higher than those from the regressions involving five-year periods, the highest value being 0.24 in table 5.14, they are still relatively low. However, this is partially to be expected as the tables are investigating how levels of the independent variables impact changes in the dependent variable. This suggests that analysis of the impact of changes in the level of aid over time on reductions in poverty might be productive. However, this was not the case. The three tables discussed in the previous section, section 5.3, explain why. Countries that receive aid in prior decades generally receive similar or increased amounts of aid in subsequent decades. Very few countries reduced their reliance upon aid as is evidenced by the small number of countries contained in table 5.4. In addition, most countries received similar levels of aid as a share of GNI since the 1970s. Therefore, regressions containing a change in aid levels between decades as an independent variable have very little explanatory power in cross-country regressions when a reduction in poverty is the dependent variable. Tables 5.12-5.15 indicate that foreign aid failed to exert a significant impact on reductions in either the extreme or moderate poverty rate. While these tables controlled for the initial level of economic freedom, they did not include the impact of changes in economic freedom. Tables E.2 and E.3 of the appendix add the change in economic freedom to the regressions of tables 5.14 and 5.15. A difference between tables 5.14 and 5.15 and those of the appendix is that the regressions in the appendix are cross-country regressions where the dependent variable is the reduction in the poverty rate during 1995-2005. The regressions of tables 5.14 and 5.15 are pooled OLS regressions. Chapter three found that changes in economic freedom during earlier periods exerted a positive and significant impact on reductions in the extreme and moderate poverty rate. EFW measures for many countries do not exist prior to 1980. This lack of data availability makes it difficult to include prior changes in economic freedom in pooled OLS regressions and is why the period 1995-2005 is examined in tables E.2 and E.3. Adding the change in economic freedom during 1985-1995 yields results that are consistent with the tables of this chapter and those of chapter three. The coefficient of the average level of

112

foreign aid during 1985-1995 is negative but is not significant. However, changes in economic freedom during 1985-1995 exert a significant impact on reductions in poverty. This is true for both the extreme and moderate poverty rates. After controlling for the initial poverty rate, economic freedom, political institutions, geographic factors, and unobservable time-varying effects, the evidence indicates that foreign aid had little or no impact on poverty rate reductions during 1980-2005. The only statistically significant relationships found in the last four tables indicate that higher levels of aid hindered rather than facilitated reductions in poverty during the twenty-five year period. These results, however, were generally not robust once additional factors were added to the regressions. Thus, despite a considerable effort on the part of the aid community, the evidence presented here indicates that the provision of foreign aid during 1976-2005 failed to significantly reduce either the extreme or moderate poverty rate.

5.5 Conclusion Fostering economic growth, encouraging institutional reform, and reducing poverty in the developing world are among the primary goals of foreign aid.

The first goal, promoting

economic growth, has been extensively examined in the literature. While the findings have been mixed, recent studies suggest that foreign aid has been largely ineffective in promoting economic growth. This chapter empirically investigates the latter of these two goals. To examine whether foreign aid was successful in promoting these goals three questions were raised. 1) What factors contribute to the level of foreign aid countries receive? 2) What impact does foreign aid have on economic freedom? 3) Has foreign aid contributed to the reductions in poverty during 19802005? Regarding the first question, the results presented here indicate that the most significant determinant of foreign aid received is the level of poverty in the recipient country. Countries with higher extreme and moderate poverty rates received substantially larger amounts of aid as a share of GNI during 1981-2005, after controlling for various factors. Neither economic freedom nor democratic political institutions exerted a statistically significant impact on the level of aid received by countries during the period. This indicates that attempts to channel aid toward countries with institutions more supportive of economic freedom and political democracy have been largely unsuccessful.

113

Turning to the second question, the results indicate that foreign aid failed to exert a significant impact on changes in economic freedom during 1985-2005. These findings are consistent with the view that foreign aid, as historically practiced, has failed to facilitate the adoption of market-based reforms. In theory, if foreign aid were to have an impact on changes in economic freedom, the results of chapter three raised the possibility of an indirect impact on poverty. Chapter three found that countries implementing reforms, consistent with economic freedom, achieved larger reductions in both the extreme and moderate poverty rates during 19802005. However, the results of this chapter indicate that reductions in poverty due to changes in economic freedom were not facilitated by foreign aid. Lastly, the analysis indicates that foreign aid failed to exert a significant impact on either the extreme or moderate poverty rate during 1980-2005.

As the first set of regression tables

illustrates, the extreme and moderate poverty rates are important determinants of the level of aid a country receives. However, while aid is given to countries because they are poor, there is no indication that poverty declines as a result. In summary, there is little evidence that foreign aid has been an effective tool for the alleviation of poverty. Public choice analysis provides insight with regard to why this may be the case. Foreign aid is prone to rent seeking in both the donor and recipient countries, often resulting in the unproductive use of resources. In addition, reforms compatible with economic freedom are often the result of poor economic performance. To the extent that foreign aid reduces the urgency for constructive reforms, it may undermine institutional change supportive of long-term prosperity. While a thorough public choice analysis is beyond the scope of this study, this is a potentially fruitful area for future examination.

114

CHAPTER 6 CONCLUSION This research examined the impact of economic freedom, political institutions, and foreign aid on global poverty during 1980-2005. Each chapter empirically tested whether these factors contributed to the decline in both the extreme and moderate poverty rates over the twenty-five year period. This final chapter summarizes the major findings, considers their implications, and suggests areas for future research.

6.1 Summary of Findings In 2008, the World Bank provided revised and updated estimates for the extreme and moderate poverty rates for 115 countries.

These new estimates were based on household surveys

conducted irregularly during 1980-2005. In chapter two, statistical techniques were utilized to adjust the data and derive poverty rate estimates at five-year intervals during 1980-2005. These data indicate that the extreme poverty rate for the developing world fell from 58.4 percent to 25.1 percent during this quarter century. The moderate poverty rate fell from 75.7 percent to 45.6 percent during same time frame. When high-income countries are included, the extreme poverty rate for the entire world fell from 47.5 percent in 1980 to 21.8 percent in 2005 and the moderate poverty rate fell from 61.5 percent to 39.4 percent. These figures cover 99 percent of the world’s population in both 2000 and 2005. Clearly, substantial reductions in both the extreme and moderate poverty rates occurred during 1980-2005. These aggregate numbers hide large differences in poverty rate reductions across countries and regions. In Asia, the extreme poverty rate fell from 69.1 percent in 1980 to 26.9 percent in 2005. During the same time frame, Asia’s moderate poverty rate declined from 88.3 percent to 52.5 percent. China was the primary driver of these poverty rate reductions. China’s extreme poverty rate fell from 84 percent in 1980 to 15.9 percent in 2005, while its moderate poverty rate declined from 97.8 percent to 36.3 percent. These huge reductions are remarkable, particularly when one considers that approximately one-sixth of the world’s population lives in China and that 98 percent of the Chinese population had incomes below the moderate poverty rate in 1980.

115

Even more important, the numbers represent movement of hundreds of millions of people out of poverty in a single generation. While Latin America’s poverty rates declined over the same period, their poverty rates in 1980 were much lower than those of China and the rest of Asia. The extreme poverty rate of Latin America fell from 15.6 percent in 1980 to 8.1 percent in 2005. The corresponding decline of the moderate poverty rate was 25.5 percent to 17 percent. In contrast with Asia and Latin America, the poverty rate reductions in sub-Saharan Africa were small, particularly during 1980-2000. In 2000, the extreme poverty rate of sub-Saharan Africa was 57.1 percent, only slightly lower than the 60.8 percent figure in 1980. The moderate poverty rate was 76.1 percent in 2000, compared to 77.3 percent in 1980. Some progress was made during 2000-2005. The extreme poverty rate of sub-Saharan Africa fell to 51.3 percent in 2005, nearly six percentage points lower than the figure in 2000. The moderate poverty rate declined to 72.3 percent, down from the 76.1 percent figure in 2000. Chapter two also uses the Pinkovskiy and Sala-i-Martin poverty rate estimates to examine the regional patterns and changes in poverty rates during 1980-2005. The World Bank poverty rate estimates are based on household consumption and expenditure surveys, while those of Pinkovskiy and Sala-i-Martin supplement the survey data with national income figures. As a result, the poverty rate estimates of Pinkovskiy and Sala-i-Martin are lower then the World Bank poverty rates. Nonetheless, the Pinkovskiy and Sala-i-Martin poverty measures reflect the same regional pattern of poverty rates and changes in those rates during 1980-2005. Chapter three examines the impact of economic freedom on poverty, using both the World Bank and Pinkovskiy and Sala-i-Martin poverty rate measures. The results indicate that the average level of economic freedom exerts an independent impact on both the level and change in poverty rates.

This was the case for both the extreme and moderate poverty rates after

controlling for political institutions and geographic and locational factors. See tables 3.3-3.6. Further, increases in economic freedom over the period were associated with larger reductions in poverty. After controlling for the initial poverty rate and political, geographic, and locational factors, a one unit increase in the EFW index during 1980-1995 was associated with a minimum 4.29 percentage point reduction in the extreme poverty rate during 1980-2005. A unit increase in EFW reduced the moderate poverty rate by 5.45 percentage points or more during the twenty-five year period. See tables 3.7 and 3.8.

116

To test for the robustness of changes in economic freedom, changes over alternative time frames were examined. Rather than the change in EFW during the fifteen-year period 19801995, the changes during two ten-year periods (1980-1990 and 1990-2000) were considered. The dependent variable remained the reduction in the poverty rate during 1980-2005. In these regressions, increases in EFW during the initial decade always exerted a significant independent impact on poverty rate reductions for both the extreme and moderate rates. The increases in EFW during the second decade had the correct sign, but were not always significant. This is not surprising because there would be only a brief time frame for changes in economic freedom during the 1990s to exert an impact on the poverty rate during 1980-2005. See tables 3.9 and 3.10. Finally, the impact of changes in economic freedom during the 1990s on the poverty rate during 1990-2005 was analyzed.

Increases in EFW during the 1990s always exerted a

significant independent impact on poverty rate reductions for both the extreme and moderate rates. See tables 3.11 and 3.12. Additional tests for robustness were performed using the Pinkovskiy and Sala-i-Martin poverty rate data as the dependent variable. The pattern was identical to that of tables 3.7 and 3.8. Changes in economic freedom during 1980-1995 exerted a significant impact on poverty rate reductions during 1980-2005, after controlling for political, geographic, and locational factors. See appendix D tables 3 and 4. Taken together, the findings of chapter three are consistent with the view that countries that move toward economic freedom achieve more rapid reductions in poverty rates. This was true for both the extreme and moderate poverty rates and it was true for both the World Bank and Pinkovskiy and Sala-i-Martin poverty measures. Of course, it will take time for changes in economic freedom to exert their full impact. Thus, changes in economic freedom exert a larger impact when they are present over a more lengthy time period. Chapter three also investigated the impact of the geography and location variables. The percentage of a countries population within 100 kilometers of a coastline was significant in poverty regressions involving levels, but was generally insignificant when the dependent variable was a change in the poverty rate. In all regressions, the sign of the coefficient was consistent with the view that proximity to navigable waterways corresponds to a lower poverty rate and larger rate reductions. The percentage of a country’s land area in the tropics was significant throughout the analysis. The sign of the coefficient was always consistent with the view that the

117

tropical zones were plagued by both higher poverty rates and less rapid reductions in those rates. Finally, a country’s air distance to major markets was a significant determinant of both levels and changes in the poverty rates. The coefficient indicated that a greater distance to major markets corresponded to higher poverty rates and smaller rate reductions. The three measures of political institutions affected the level of poverty in chapter three. In these regressions the average level of political institutions during 1980-2005 was significant and associated with lower poverty in 2005, after controlling for other factors. However, there was little or no evidence that changes in political institutions exerted an impact on changes in poverty rates. In regressions with reductions in the poverty rate as the dependent variable, changes in political institutions were rarely significant. Chapter four explores the possibility that political institutions might enhance future economic freedom and thereby exert an indirect impact on the poverty rate. Initially, the relationship between political institutions during one decade and economic freedom during the next decade was examined. This analysis indicated that movements toward more democratic political institutions, as measured by the three political variables, were associated with subsequent increases in economic freedom. This was the case after controlling for time-varying effects in pooled OLS regressions, contemporaneous changes in political institutions, and geographic and locational factors. Moreover, this result held for all countries as well as for low and middle-income countries only. Next, the indirect impact of changes in political institutions on the poverty rate through economic freedom was examined. A two-step approach was used to estimate this indirect impact. First, a variable was constructed which represented the change in economic freedom over a decade unexplained by changes in political institutions from the previous decade. Second, the constructed variable was substituted for the change in economic freedom variable in poverty regressions. In this specification, the change in the political institutions variable will reflect both the direct and indirect impact of political institutions on the poverty rate. Using this statistical method, changes in political institutions exerted a significant impact on reductions in the extreme poverty rate in the subsequent decade, after controlling for other factors. These findings are consistent with the view that after accounting for the indirect impact, changes in political institutions exert a significant impact on the extreme poverty rate. The results were insignificant, however, for the moderate poverty rate.

118

Chapter five examined the impact of foreign aid on reductions in poverty during 1980-2005. Foreign aid as a share of GNI is both high – ten percent or more – and increasing in many countries, particularly those in sub-Saharan Africa. See tables 5.2 and 5.3. In contrast, there were only four countries with aid levels as a share of GNI during 2000-2005 that were eight percentage points or more lower than during the 1970s. These four countries are Botswana, Cambodia, Egypt, and Jordan. See table 5.4. Clearly, not many countries have been able to successfully transition from high to low aid levels. In examining the impact of foreign aid on poverty, chapter five posed three questions. The first was what factors influence the level of foreign aid a country receives? The results indicate that the single most important factor was the extreme and moderate poverty rate in the recipient country. Countries with a higher extreme or moderate poverty rate at the beginning of either a five or ten-year period subsequently received more foreign aid as a share of GNI than countries with lower poverty rates. This was after controlling for economic freedom, political institutions, and time-varying effects. See tables 5.5-5.8. The economic freedom and political institutions variables were largely insignificant in these regressions, indicating these factors had very little impact on the amount of aid received as a share of GNI. The second question was, does foreign aid exert an impact on economic freedom or political institutions? Foreign aid as a share of GNI failed to exert a significant impact on the change in economic freedom in either the contemporaneous or subsequent decade. See tables 5.9-5.11. The same findings were present for changes in political institutions. Thus, there is no evidence that foreign aid, as currently practiced, has enhanced either economic freedom or democracy. The third question was, what impact did foreign aid have on reductions in poverty during 1980-2005? The relationship between foreign aid as a share of GNI and subsequent reductions in poverty were examined for both five and ten-year periods.

There was no significant

relationship between aid levels and subsequent reductions in poverty rates. This was the case after controlling for economic freedom, political institutions, geographic and locational factors, and time-varying effects. These findings held for both the five and ten-year time periods and for both the extreme and moderate poverty rates. See tables 5.12-5.15. The findings indicate that foreign aid failed to reduce poverty during either the current or subsequent periods.

Furthermore, it also failed to promote either economic freedom or

democracy. Thus, the results are consistent with the view that aid, as currently practiced, is an

119

ineffective tool for either the reduction of poverty or the promotion of institutions conducive for poverty rate reductions.

6.2 Implications Movements toward economic freedom exerted a significant impact on reductions in poverty during 1980-2005.

Unfavorable geographic and locational factors impeded poverty rate

reductions. However, changes in economic freedom exerted a significant impact on both the extreme and moderate poverty rate, after accounting for the adverse impact of geographic and locational factors.

Thus, the adverse affects of geographic and locational factors can be

overcome through movements toward economic freedom. Many of the conditions necessary for economic freedom are maintained to a large degree by political institutions. In this research, political institutions aided movements toward economic freedom and facilitated reductions in the extreme poverty rate. Taken together, the results of chapters three and four are consistent with the view that institutions matter for reductions in poverty. Despite disadvantages, such as geographic and locational factors and initial economic and political institutions, movements toward economic freedom, democracy, and restrictions on executive power were associated with reductions in poverty. Even though foreign aid levels increased substantially during last half-century, there was no evidence that this expansion in aid exerted an impact on economic or political institutions or facilitated reductions in poverty during 1980-2005. Instead of serving as a temporary means to help start the development process, foreign aid has fostered dependency. While many countries received high levels of aid year after year, the number that have reduced or eliminated their dependency on aid can be counted on one hand. The ineffectiveness of foreign aid, as currently practiced, is interesting for two reasons. First, the objective of foreign aid is the elimination of global poverty. Second, the findings of this research indicate that aid had no impact on reductions in the extreme or moderate poverty rate. While poverty reduction is the motivation for foreign aid, it contributes little to the achievement of this outcome. This research is consistent with the view that institutions supportive of economic freedom contribute to the alleviation of poverty. The findings also indicate that democratic political 120

institutions facilitate movements toward economic freedom and thereby indirectly contribute to reductions in the extreme poverty rate. While unfavorable geographic and locational factors impeded progress against poverty, movements toward economic freedom and democracy were able to overcome these factors. Some argue that foreign aid is needed to hasten the alleviation of poverty. The results presented here indicate that foreign aid has been largely ineffective in this regard.

Both economic theory and prior research indicate that institutions supportive of

economic freedom are an important source of economic growth. This research indicates that these institutions are also an important source of reductions in global poverty.

6.3 Future Research It is clear that institutions matter, but it is less clear what brings about changes in institutions. Cultural values and norms may play a role in institutional change. Measures of values and norms have recently been created enhancing the potential productivity of future research in this area. Chapter four found that political institutions facilitated reductions in the extreme poverty rate through changes in economic freedom. Examining whether a similar relationship exists between political institutions and economic growth would be a fruitful and natural extension of this research. Legal institutions, specifically the security of property rights, are critical for the future of sub-Saharan Africa. In Africa, property rights are poorly defined and often depend upon the whim of both tribal leaders and political rulers. The Ghanaian economist George Ayittey argues that indigenous property rights, based upon the family and traditional culture, could facilitate economic growth (Ayittey 2006). Is his assertion correct? What are the different types of traditional African property rights? Are there cases where these property rights have been supportive of economic growth? Answers to these questions may shed light on the direction of sub-Saharan Africa’s development. Can changes be made to the current practice of foreign aid to make it more effective? The majority of foreign aid is channeled through the government of recipient countries. These governments are susceptible to corruption and rent seeking, undermining the effectiveness of aid. Could aid organizations change this aspect of foreign aid? Could the structure of foreign aid be

121

altered in a manner that would improve its effectiveness? All of these questions are potentially productive areas for future research. Rather than use foreign aid to promote economic growth and general development, it could be targeted toward objective goals such as increased access to clean water, malaria prevention, or reducing infant mortality rates. These more quantifiable objectives provide information about effectiveness of the aid and how it might be better allocated. Some projects of this type are already operational. Analysis of their effectiveness is a topic for future research.

122

APPENDIX A EXTREME AND MODERATE POVERTY RATES, 1980-2005

123

Table A.1: Extreme ($1.25 per day) and moderate ($2 per day) poverty rate by country

Albania Algeria Angola Argentina Armenia

Percentage of Population Living on $1.25 per Day or Less 1980 1985 1990 1995 2000 2005 2.0 2.0 2.0 25.2 18.1 6.6 6.8 10.9 8.5 64.0 63.5 61.9 61.2 54.3 44.2 5.6 2.0 2.0 2.0 6.0 4.5 17.5 16.5 10.6

Percentage of Population Living on $2 per Day or Less 1980 1985 1990 1995 2000 2005 6.5 8.7 7.8 31.7 25.1 23.8 23.6 20.4 16.8 70.4 69.5 67.9 71.3 70.2 52.0 10.9 2.0 3.2 7.0 14.3 11.3 38.9 47.7 43.4

Azerbaijan Bangladesh Belarus Benin Bhutan

15.6 6.3 2.0 39.3 77.5 72.2 66.8 59.4 57.8 49.6 99.0 99.0 92.5 87.4 2.0 2.3 2.0 2.0 2.0 11.1 65.2 62.6 61.3 57.3 53.4 47.3 81.6 79.0 79.1 75.6 56.1 48.7 39.9 33.7 28.5 26.2 76.8 68.5 56.7 49.0

Bolivia Bosnia and Herzegovina Botswana Brazil Bulgaria

37.0 34.5

18.9 23.8 19.6 46.7 46.7 17.2 29.9 34.9 30.3 14.6 2.0 2.0 25.0 2.0 2.0 42.0 35.6 31.9 31.2 32.3 23.1 65.1 54.7 50.2 49.4 46.0 36.3 17.1 17.5 15.5 10.5 11.1 7.8 31.1 31.5 27.8 21.9 22.6 18.3 2.0 2.0 2.6 2.0 2.0 2.2 7.8 2.4

Burkina Faso Burundi Cambodia Cameroon Cape Verde

74.7 71.0 68.3 71.2 70.0 72.3 70.4 84.2 85.7 86.4 48.6 45.8 52.8 45.6 46.9 51.5 32.8 43.0 38.3 36.0 33.1 20.6

Central African Republic Chad Chile China Colombia

62.9 71.3 12.3 84.0 13.7

Comoros Congo, Dem. Rep. Congo, Rep. Costa Rica Cote d'Ivoire

56.7 69.0 37.9 21.4 16.9

44.3 74.1 39.5 4.4 23.7

Croatia Czech Republic Djibouti Dominican Republic Ecuador

2.0 2.0 2.0 2.0 2.0 2.0 11.7 4.8 18.8 24.4 16.4 8.4 5.9 4.4 20.2 12.2 14.0 15.9 14.9

27.1 2.0 85.4 81.3 2.0 2.0 71.3 75.3 42.6 49.5

4.0

94.8 90.9 88.7 85.8 87.6 97.1 94.7 95.2 95.3 95.4 77.8 74.6 65.3 56.3 60.2 74.4 57.7 62.9 55.9 53.4 49.7 40.2

81.2 93.4 68.2 62.6 42.6

61.2 61.6 82.8 64.9 62.4 65.9 64.6 65.3 66.3 61.9 10.5 4.4 2.3 2.0 2.0 61.7 60.2 45.0 32.0 15.9 12.3 9.5 11.2 16.8 15.7

81.5 90.3 22.9 97.8 24.4

80.5 82.8 23.4 88.3 23.1

82.0 82.5 13.6 84.6 19.4

90.7 83.6 9.1 71.8 23.3

85.6 84.9 6.0 56.3 29.1

81.9 83.3 3.9 36.3 27.1

51.9 68.4 33.6 10.4 9.5

46.1 59.2 54.1 2.4 15.5

74.0 88.8 49.3 35.7 34.9

69.2 88.8 42.4 21.5 23.9

67.5 91.0 46.1 18.7 35.1

66.3 99.0 49.1 16.4 47.9

64.6 99.0 50.2 11.5 47.9

65.0 79.5 74.4 8.6 38.9

2.0 2.0 8.8 5.0 9.8

2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 24.4 15.1 41.2 26.1 37.9 30.4 20.8 15.7 12.4 15.1 28.6 22.3 24.0 28.2 27.7 20.4

49.1 69.1 36.1 9.2 13.8

46.7 73.2 38.3 7.5 21.1

124

56.5 81.3 40.2 49.1 27.9

Table A.1 – continued ^B@OBFU