DIFFERENTIAL ECONOMIC PERFORMANCE IN DEVELOPING COUNTRIES Except where reference is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory committee. This thesis does not include proprietary or classified information. ___________________________ Charumporn Fon Vijakkhana Certificate of Approval: ________________________ ______________________ T. Randolph Beard John D. Jackson, Chair Professor Professor Economics Economics _________________________ _______________________ Richard P. Saba George T. Flowers Associate Professor Dean Economics Graduate School DIFFERENTIAL ECONOMIC PERFORMANCE IN DEVELOPING COUNTRIES Charumporn Fon Vijakkhana A thesis Submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Master of Science Auburn, Alabama December 19, 2008 iii DIFFERENTIAL ECONOMIC PERFORMANCE IN DEVELOPING COUNTRIES Charumporn Fon Vijakkhana Permission is granted to Auburn University to make copies of this thesis at its discretion, upon request of individuals or institutions and at their expense. The author reserves all publication rights. ____________________________ Signature of Author ____________________________ Date of Graduation iv ? VITA Charumporn Vijakkhana, daughter of Sittisak Vijakkhana and Roongnapa Vijakkahana, was born August 22, 1983, in Surin, Thailand. She graduated from Loveless Academic Magnet Program High School, Montgomery, Alabama in 2002. She attended Auburn University, Auburn, Alabama, and graduated in May 2006, with a Bachelor of Science in Economics. She entered graduate school at Auburn University in August 2006. ? ? ? ? ? ? ? ? ? ? ? ? v ? THESIS ABSTRACT DIFFERENTIAL ECONOMIC PERFORMANCE IN DEVELOPING COUNTRIES Charumporn Vijakkhana Master of Science, December 19, 2008 (B.S., Auburn University, 2006) 68 Typed Pages Directed by John D. Jackson This thesis examines the economic performance in developing countries. The 2003 data is collected from 74 developing countries. The level of economic performance is measured by GDP per capita. The independent variables that are used in this study are physical capital, human capital, percentage of service in GDP, political rights, inflation, foreign direct investment, and trade openness. FDI and education also used as interaction variable to investigate their effect on GDP per capita. ? ? vi ? ACKNOWLEDGEMENTS The author would like to thank Dr. John Jackson for guidance and support and his incredibly patient throughout this process. The author also would like to thank committee members Dr Randolph Beard and Dr Richard Saba for additional inputs into this project. The author also would like to thank her parents for their love and support and her great aunt and great uncle for their advices and guidance throughout my stay at Auburn. ? ? ? ? ? ? ? ? ? ? ? vii ? Style manual or journal used: American Economic Review Computer software used: Microsoft Word 2007 Microsoft Excel LIMDEP 8.0 viii ? TABLE OF CONTENTS LIST OF TABLES ............................................................................................................. ix LIST OF FIGURES .............................................................................................................x INTRODUCTION ...............................................................................................................1 BACKGROUND .................................................................................................................4 LITERATURE REVIEW ..................................................................................................15 METHODOLOGY ............................................................................................................22 RESULTS ..........................................................................................................................40 CONCLUSION ..................................................................................................................54 REFERENCES ..................................................................................................................57 ix ? LIST OF TABLES Table 1. GDP Distribution in 1960 and 2006 .....................................................................6 Table 2. GDP per Capita in 1960 and 2006 ........................................................................7 Table 3. Countries with Population living less than $2 a day ..........................................10 Table 4. Percentage of Population living less than $1 a day by Region ..........................11 Table 5. Growth in GDP per Capita from 1960 to 2006 ...................................................13 Table 6. Country Classification .......................................................................................27 Table 7. Political Rights Rating ......................................................................................35 Table 8. Regression Results (Model I) ..........................................................................44 Table 9. Regression Results (Model II) ..........................................................................48 Table 10. Regression Results (Model III) .......................................................................50 Table 11. Regression Results (Model IV) .......................................................................52 x ? LIST OF FIGURES Figure 1. GDP per Capita..................................................................................................28 Figure 2. Capital................................................................................................................29 Figure 3. Education ..........................................................................................................31 Figure 4. GDP Structure ..................................................................................................32 Figure 5. Service ...............................................................................................................33 Figure 6. Inflation ............................................................................................................36 Figure 7. FDI ....................................................................................................................38 Figure 8. Openness ...........................................................................................................39 1 ? CHAPTER I INTRODUCTION The Industrial Revolution brought many changes to the world. Prior to this period, the world?s Gross Domestic Product (GDP) stayed relatively constant; there was not that much growth in GDP in any economy. With the burst of new technology, countries started to move from agrarian to manufacturing and began to experience growth in output. Originating in Britain and spreading throughout Europe and later to North America, this process of industrialization brought wealth and elevated people?s standard of livings. Years passed, and wealth kept on growing. Today, growth is a very important indicator of the well-being of the economy. However, growth in output is uneven across countries. Although poor countries? GDP may increase, they are increasing at a slower rate. Many countries in South America, Africa, and Asia are low in GDP per capita. This thesis is the study of economic growth in such countries. There are total of six chapters in the thesis. The contents of each chapter are explained in the following paragraphs. Chapter II presents background in economic growth. This chapter has three sections. The first section takes a look of GDP and GDP per capita of the world with 2 ? countries classified into high income, upper middle income, lower middle, and low income countries. The second section discusses current aspects of the problem of poverty in the world. The third section presents a brief summary of growth theory from the past to the present. The fourth section is a conclusion of the chapter. Chapter III is a literature review. The chapter is divided into two sections. The first section reviews the literature of growth theory, which includes the Harrod-Domar, neoclassical, and endogenous growth theories. The second section reviews empirical studies of growth using cross-country data. These previous studies report the result on what factors affect growth. Robert J. Barro (1997) examines the relationship between political freedom and inflation and growth. Paul Segerstrom (1991) studies the link between technological innovation and imitation and their affect on growth in developing countries. The study by Borensztein, Gregorio, and Lee (1998) investigates the effect of foreign direct investment on the process of growth. Halit Yanikkaya (2003) examines the effect of the openness of a country on economic growth. Finally, this section presents the study by Mukesh Eswaran and Ashok Kotwal (2002) which looks into the role of service sector play in the process of agriculture sector and how the service sector affect economic growth. Chapter IV is methodology. There are two sections in Chapter IV. The first chapter describes the conceptual model. This section also discusses the challenge faced in selecting independent variables, since there are many countries with different characteristics and many factors that contribute to growth. The first section concludes 3 ? with the conceptual model for this thesis. The second section defines and discusses each variable in detail and presents their descriptive statistics. Chapter V presents the regression results. Chapter V is divided into four sections. The first section discusses the OLS assumptions and the consequences of violation of these assumptions. The second section reports the results of the Breush-Pagan test for heteroskedasticity. The third section discusses the misspecification problem and presents the results for the Ramsey?s RESET test. The last section presents the regression results and the discussion of the interpretation of the results. Chapter VI is a conclusion. The conclusion summarizes and comments on the results of this study. It also discusses ideas for further research. 4 ? CHAPTER II BACKGROUND This chapter addresses the definition and measurement of economic growth and the development process; it has four sections. The first section examines the composition of the world?s GDP in the past and present. This section examines both GDP and GDP per capita of the world and countries which are classified into high income, upper middle income, lower middle income, and low income countries. The second section looks at the problem of poverty in the world. It defines what poverty is and presents a discussion of the current state of poverty in the world. The third section gives a brief overview of growth theory from the past to the present. This section also presents the challenges that are faced in cross-country studies of growth today. Finally, the last section presents a concluding remark. A. Overview of the World?s Wealth Compared to fifty years ago, the world is producing and consuming more. Using 2000 as the base year, the world GDP was $7,265,893,529,448?in 1960 and has grown to 5 ? $37,866,383,439,225 in 2006. For the past fifty years, the world?s output was on a steady rise. World production as a whole is growing, but the distribution of wealth is still far from even. For 2006, 78% of the world GDP goes to high income countries, 9% belongs to upper middle income, 9% goes to lower middle income, and only 3% are from low income countries (World Bank). This distribution has not changed much in the past fifty years. The high income countries produce the majority of world GDP. In 1960, the high income nations produce 84% of the world output, upper middle income countries produce 9%, middle income and lower income countries produce 4% and 2%, respectively. The largest gains are among the lower middle income countries, but that gain is only 5% (World Bank). The upper middle income nations? shares remain the same at 9%, and the lower income countries only gain one more percent of shared outputs. Table 1 below presents this information. Although the overall wealth of the world has increase, the distribution of wealth does not change much. The rich nations still get most of the wealth, and the income gap is still wide open. Those statistics were based on total output; a more meaningful measure of economic well-being is GDP per capita. The share of GDP indicates that rich nations still enjoying more wealth, but overall GDP itself does not tell much about the welfare of people. An increase in overall GDP could mean that labor force has grown and thus the country produced more. GDP per capita needs to be examined to see if the average person actually enjoys more income, or whether instead the growth in GDP suggests that everybody still enjoys the same or smaller income, but there are just proportionately more 6 ? people. We observed that not only output is bigger, but GDP per capita also indicates people are getting more income. The world GDP per capita is $2,405 in 1960 compared to $5,810 in 2006. However, the distribution of the GDP per capita is very uneven. For 2006, the high income countries? GDP per capita is $28,689. For developing countries, their GDP per capita are all under the world average. The GDP per capita for low income, low middle income, and upper lower income are $509, $1,565, and $4,422 respectively (World Bank). The Table 2 below compares the GDP per capita in 1960 and 2006. In 2006, the world GDP per capita is closer to the developing countries than to the higher income countries. This indicates that the majority of people in the world are making relatively little money. In fact, about a third of the world population is still living in poverty. This topic is discussed in detail in the next section. Table 1 ? GDP distribution in 1960 and 2006 Countries 1960 2006 High Income 84% 78% Upper Middle Income 9% 9% Lower Middle Income 4% 9% Low Income 2% 3% 7 ? Table 2 ? GDP per capita in 1960 and 2006 Countries 1960 2006 High Income $8,790 $28,689 Upper Middle Income $1,626 $4,422 Lower Middle Income $262 $1,565 Low Income $206 $509 World $2,405 $5,792 B. Poverty From the previous section, both GDP and GDP per capita reveal that the distribution of wealth does not change much. Despite the increase in overall outputs and per capita output, there is still a large disparity in distribution of wealth among countries in the world. In 2006, the world average GDP per capita is $5,810, while the high income countries? average GDP per capita is $28,689. All countries except those in the ?high income? group have GDP per capita figures that are below the world?s average. These statistics indicate that the majority of people in the world are still living in poor conditions. 8 ? In fact, about a third of the world?s population is living in moderate or extreme poverty. The World Bank defines extreme poverty as people that are living on less than $1 a day and moderate poverty as people that are living less than $2 a day. The following World Bank statistics were gathered by sampling 1.1 million people in over 100 countries (Ref). The surveys were conducted by government officials of the particular country. From these surveys, assessments can be made about the distribution of consumption in each country. From this distribution, the proportion of people who do not meet the level of poverty line is calculated. According to the World Bank?s estimate, in 2001, 1.1 billion people have a consumption level less than $1 a day, and 2.1 billion people live on less than $2 a day. That is roughly a third the world population living in moderate poverty, and about one sixth of the world population live in extreme poverty. According to the data from Human Development Report published by the United Nations, 85 countries have populations that are living under $2 a day (Ref). The Table 3 below shows percentage of population of those countries that are living on less than $2 a day. Table 3 shows that many countries in Africa and Asia have a majority of their population living in poverty. Nigeria has more than 90% of its population living in at least moderate poverty. Countries such as Zimbabwe, Zambia, Madagascar, Rwanda, Tanzania, and Niger have more than 80% living below poverty line. These countries are in Africa. However, in Asia, India and Bangladesh also have more than 80% of their population living on less than $2 a day. Poverty is still everywhere in the world. In Eastern Europe, countries such as Slovakia, Latvia, Bulgaria, In South America, Mexico, 9 ? Argentina, Colombia, Panama, Peru, Guatemala, Venezuela, and El Salvador have their population living in poverty. In Asia, China has about 30%-40% living below $2 a day. Many countries are still living in at least moderate poverty, and these countries are everywhere in the world. Table 4 further illustrates the fact that poverty is everywhere in the world; it groups the percentage of people that live on less than $1 a day by regions. Table 4 reveals that Sub-Sahara Africa has 41.09% of the population living on less than $1 a day. In South Asia, 31.11% of its population is living in extreme poverty. East Asia and Latin America have about 8% in extreme poverty. Middle East and North Africa have 1.47%. Finally, Europe and Central Asia have 0.95% living in extreme poverty. The Table 4 shows that poverty is still an important problem everywhere, and many people are still living with a consumption level less than $1 a day. But what does $1 a day mean exactly? With $1 or less a day, basic needs are not met. People that live in this condition do not have enough to eat, do not have proper shelter, and cannot afford proper health care. This means that people can die of illness that can be easily treated with proper medical care. When basic needs are not met, these people are set up for a continuation of the vicious cycle of poverty. They cannot afford proper education, so they do not have a chance to better their skills and attain better jobs. Poor health also hinders them from getting education or training they need. The opportunity to get out of this condition is slim to none. This creates a vicious cycle that people are stuck in for generations. 10 ? Table 3 ? Countries with population living on less than $2 a day < 10% 10% - 20% 20% - 30% 30% - 40% 40% - 50% Slovakia, Latvia, Ukraine, Chile, Uruguay, Bulgaria, Tunisia, Jordan, Iran, Lithuania, Malaysia, Costa Rica Mexico, Russia, Romania, Morocco, Jamaica, Algeria, Kazakhstan, Dominican Republic, Argentina, Colombia, Panama, Turkey Moldova, Brazil, Kyrgyzstan, Thailand, Georgia, Paraguay Peru, Armenia, Guatemala, Azerbaijan, South Africa, China, Trinidad and Tobago Venezuela, El Salvador, Ecuador, Sri Lanka, Bolivia, Tajikistan, Philippines, Egypt, Mongolia Yemen, Cote d?Ivoire 50% - 60% 60% - 70% 70% - 80% 80% - 90% > 90% Cameroon, Indonesia, Botswana, Namibia, Lesotho, Senegal, Kenya Malawi, Mauritania, Nepal Burkina Faso, Mali, Pakistan, Benin, Laos, Mozambique, Sierra Leone, Cambodia, Ethiopia Swaziland, Haiti, Ghana, Nicaragua India, The Gambia, Zimbabwe, Bangladesh, Central Africa Republic, Madagascar, Niger, Zambia, Burundi, Rwanda, Tanzania Nigeria 11 ? Table 4 ? Percentage of population living on less than $1 a day by region Regions 2004 East Asia and Pacific 8.88% Europe and Central Asia 0.95% Latin America and the Caribbean 8.71% Middle East and North Africa 1.47% South Asia 31.11% Sub-Sahara Africa 41.09% Poverty is a very important problem in economic development. It?s important for these countries to catch up with the develop nations in term of economic well-being. Economic growth is one of the goals that will help alleviate the problem of poverty. The question of what causes economic growth is quite relevant. The study of economic growth is important; finding the cause of growth could bring improvement to millions of people?s lives. C. Growth Theory Generally for there to be growth, an economy has to expand its production capacity. For the production capacity to increase there needs to be investment made into 12 ? physical capital. With investment in physical capital today, tomorrow?s productive capacity is expanded, and therefore the whole economy will produce more output. One well-known growth theory is the neoclassical model. In the neoclassical model, output is a function of capital and labor. Technology is given and assumed to be identically implemented everywhere. One of the well known facts that come out of the neoclassical model is that outputs will converge. This suggests that the poor countries will catch up with the rich nations in term of growth. A more complete discussion of the neoclassical growth model can be found in chapter III to follow. The neoclassical growth model says that less developed nations will grow at a faster rate. Therefore, they will eventually catch up with the developed nations? output. As the data in the previous section shows, there is a huge disparity between developed and developing nations. The distributions of GDP today and fifty years ago are about the same. There is little sign of closing the gap. Many cross-country studies also have not been able to verify a convergence in rate of growth. With this flaw in the neoclassical theory, many economists look for a more relevant theory to model growth in cross- country samples. The information from the Table 5 shows that the world GDP per capita grows by 141%. The lower middle income countries have almost a 500% increase in their GDP per capita. However, for the rest of the developing countries, their growth in output is less than that of the high income countries. 13 ? Table 5 ? Growth in GDP per Capita from 1960 to 2006 Countries GDP per capita Growth High Income 226% Upper Middle Income 172% Lower Middle Income 497% Low Income 147% World 141% These observations lead to endogenous growth theory. Now economists are studying growth in developing countries with assumption that technology is not the same in all countries and that many factors besides capital and labors affect growth in output. For one country to grow is a very complicate situation. The conditions will vary from one country to others. Each country has its own characteristics. The first and most important thing for growth is physical capital. A country needs an investment today in order to grow tomorrow. But there are also other factors that affect growth. A more complete discussion is provided in chapter III. D. Conclusion 14 ? This chapter has shown that the overall GDP and GDP per capita are growing. However, the distribution of wealth does not change much in the past fifty years. In 2006, high income countries produce almost 80% of the output. Moreover, the upper middle income, lower middle income, and low income countries? GDP per capita is below the world average, which indicates that many people in the world are making little money. Further examination reveals that 2.1 billion people live in moderate or extreme poverty. Therefore, poverty is an important and relevant problem. Investigating the factors that affect economic growth is one part of solving this problem. To say that the study of the cause of growth in these developing countries is important is a dramatic understatement. 15 ? CHAPTER III LITERATURE REVIEW This chapter is divided into two parts. The first part reviews the literature of growth theory. This section discusses the economic growth theory from Harrod-Domar model, to neoclassical growth model, and to endogenous growth model. The second part of this chapter discusses the cross countries empirical studies of economic growth. A. Growth Theory The Great Depression brought about a change in the field of economics. The hands-off policy of the classical economists was no longer appropriate. Say?s Law that supply creates its own demands had its limitations. This is where Evsey Domar (1947) and Ray Harrod (1948) pick up. They independently investigate the full employment conditions given that people hoard. Therefore, the main focus in this paper is investment?s role. When people save, this saving turns into investment, and capital will then accumulate, thus the economy?s productivity capacity increases. When capacity increases, Harrod and Domar argue three things can happen ? the new capitals remain 16 ? unused, the new capital will substitute the old, or the new capital will substitute for labor and there will be unemployment. This suggests that the way to maintain full employment is to keep income growing. Domar is interested in finding the magnitude of investment that will keep economy at full employment. The condition for full employment is that the increase in income equals the increase in productive capacity. Solow (1956) criticizes Domar-Harrod?s assumption of fixed proportions in production. Solow argues that there is a possibility of substituting labor for capital in production. He developed what has become the heart of the neoclassical growth model. Output is a function of both capital and labor, Y=F(K,L). Net investment is a rate of increase capital stock, which is equals to that part of income which is being saved. The model centers on the role of capital-to-labor ratio, r = K/L. Solow defined r * as when the rate of change in the capital-to-labor ratio is equal to zero. When this happens, capital stock expands at the same rate as the labor force. Solow called this the warranted rate of growth. If r>r*, then r will decrease and move toward r * , which means capital will grow slower than labor. The opposite will happen if r $11, 116 1. GDP per capita. GDP per capita is included as the dependent variable. There are 74 developing countries in the sample. The average GDP per capita from the sample is $2233.56. This average falls in the range of the lower middle income countries according to the World Bank?s classification. The maximum for GDP per capita is $8769.26. That country is Oman. The least GDP per capita is Ethiopia with $121.19. The scatter plot below shows the distribution of the data. There are a large number of countries that GDP per capita is below $1,000. The majority of countries in this sample are in the range of lower middle income countries which is $906-$3,595. The standard deviation for this sample is $1926.63. 28 ? ? Figure?1??GDP?per?capita 2. Capital Capital plays a vital role in the process of growth. For a country to grow in the future, saving needs to turn into investment for the future. For physical capital, this paper uses gross fixed capital. Past literature has used many variables to measure physical capitals. One study used the number of telephone lines as a proxy. For this paper, fixed capital is used. The maximum is $146,106,122,240.00, and the minimum is $109,850,952.00. The average is $12,058,250,494.92, the median is $3,249,333,248.00, and the standard deviation is$ 25,515,467,699.57.? From the scatter plot, the majority of countries have their capital below $20,000,000,000. Capital is expected to positively affect GDP per capita; the more capital, the higher economic growth. ? 29 ? ? Figure?2???Capital 3. Education Physical capital is not the only essential element for growth. There will be no use for physical capital if there is no one there to use it. Therefore, human capital is also a very important factor in the process of economic growth. Higher education means higher skill level. Education brings specialization. This means that the labor force will be more productive. Human capital is included here because many empirical researchers attest that human capital plays a vital role in growth. There are many empirical studies that look into the effect of human capital on growth. For example, human capital will contribute to the research sector. The higher human capital, the better the research sector will be. Thus, this will lead to faster technological progress, and ultimately to growth. Higher 30 ? human capital will lead to a quicker absorption of new ideas from outside the country. Thus, countries with higher human capital will grow faster. Romer (1990) and Nelson and Phelps (1966) Therefore, human capital will expected to be positively related to GDP per capita. The variable used to measure human capital in this study is the rate of primary school completion. There are many variables that can be used as proxy for education. Some proxies that are used for education are teacher-to-student ratio, spending on education, literacy rate. For this thesis, the primary completion rate is included as a proxy for education. This study use developing countries; thus, the primary education is sufficient. Primary completion rate is defined as the total number of students minus the number of repeaters in that grade divided by the total number of children of official graduation age. In this sample, the maximum is Brazil with the completion rate of 108.8%, and the minimum is Burkina Faso with the rate of 28.17%. The average is 83.15, and the standard deviation is 20.05. 31 ? ? Figure?3???Education 4. Service GDP is generally taken as originating in of the three sector ? agriculture, manufacture, and service. There is a theory that an economy will go through structural changes. It will moves from agriculture to manufacturing, and finally to service. The size of the service sector can be used to measure the maturity of the economy. The world GDP is made up of 3% from agriculture products, 18% from manufacturing goods, and 69% from services. However, the high income countries? GDP?s comprised of 72% service, 17% manufacture, and 2% agriculture. For low income countries, 50% of GDP made up of service, 15% is from manufacturing sector, and 22% is from the agriculture sector. For lower middle income countries, 45% is from 32 ? service sector, 26% is from manufacturing, and 13% is from agriculture. For upper middle income, 62% is from service, 19% is from manufacture, and 7% is from agriculture. The Figure shows the composition according to the classification of high income, upper middle income, lower middle income, and low income countries. This data shows that agriculture is the smallest component of GDP in the high income countries. But it is interesting to note that, the lower middle income countries? GDP is made up of more manufacturing goods than the upper middle income countries?. ? Figure?4???GDP?Structure? The size of the service sector is measured by service as percentage of GDP. This variable proxies the maturity of the economy. The assumption here is that an economy goes through stages of change. An economy starts off as a agarian society, then an economy will progress to manufacturing. And finally, the economy will enter the service sector. Thus, service output as a percentage of GDP is also included. The mean is 52.82. The 33 ? median is 53.87. The standard deviation is 11.809. Equatorial Guinea has the minimum percentage of service in GDP at 5.12, while Panama has the maximum percentage of GDP at 78.53%. ? ? Figure?5??Service 5. Political Rights The level of freedom in a country and the level of government corruption have an impact on growth. A political rights variable is included in the model to account for these effects. The explanation of what political rights encompass is given in the following paragraphs. The political rights variable that is included in this study is ordinal data, which has a questionable role in regression analysis. The details of how this index is derived are in the following paragraphs. 34 ? Political rights data was obtained from Freedom House. Countries are rated on the scale of 1-7 with 1 being the most free. The political rights ratings are based on questions in three categories ? electoral process, pluralism and participation, and functioning of government. The survey has 10 questions. Three questions are for electoral process, four questions are for pluralism, and three questions are for functioning of government. The electoral questions are as follows: is the head of the government elected through free and fair election, are the representative elected through free and fair election, and are the electoral laws fair? Political pluralism, which measures the degree in which the political power does not reside only with a few groups, and participation ask questions such as do people have the rights to organize political parties and is the system open to the rise and fall of these competing parties? Is there significant opposition votes to increase support and gain power through elections? Are people?s political choices free from military, foreign powers, or any other power groups? Do ethnic, cultural, religious, or minority groups have full political rights and electoral opportunities? The third group of questions deals with the functioning of government: Do the elected governments determine the policies of government? Do government free from corruption? Is the government accountable to the electorate between elections, and does it operate with openness and transparency? The score range from 0-4 in for these questions. Then they add up and convert to a 1-7 scale. The rating of 1 means election is fair, those elected rule, and the opposition play an important role and has actual power. The rating of 2 means political rights is less free. 35 ? There is corruption, discrimination against minority, and military or foreign power may have influence on the country?s politics. The ratings of 3, 4, and 5 have the same conditions as rating 2. In addition, these countries have other factors such as civil war, heavy military involvement in politics, unfair election, and one-party dominance. Countries whose rating is 6 are ruled by military juntas, one-party dictatorships, religious hierarchies, or autocrats. A rating of 7 means political rights are absent or nonexistent. Table 7 ?Political Rights Rating Total Score PR Rating 36-40 1 30-35 2 24-29 3 18-23 4 12-17 5 6-11 6 0-5 7 The mode of the sample is 6, and the average is 3.5. Countries such as Belize, Bulgaria, Chile, Hungary, and South Africa have the best political rights rating of 1. Three 36 ? countries from the sample have the political rights rating of 7; these countries are Equatorial Guinea,?Swaziland, and Vietnam. 6. Inflation Inflation is also included. How does inflation related to growth? Inflation should be negatively related to growth. If there is inflation, households and businesses are expected to be worse off. Inflation affects consumers? and producers? behavior in the economy. Moreover, when there is an expectation of inflation, people expect the economy to slow down. They will adjust their behavior, which in turn might actually slow down the economy. Past literature has found that low inflation has no significant effect on growth, only high inflation does. However, it is important to include this Chad, ? Figure?6??Inflation? 37 ? variable, since countries are affected by high inflation. The mean in the sample is 6.16. The maximum inflation is 31.11, which occurred in Venezuela. There is a deflation in which is -1.75. The median is 4.21. The standard deviation is 7.089.? 7. Foreign Direct Investment With the endogenous growth theory, technology is not longer treated as the same in every country. Each country has its own different level of technology. Technology is developed within a country through research and development. But for many developing countries, one of the easier ways to have new technology is through the adoption of new technology from the developed nations. This can happen through many channels. A country can adopt new technology from developed country or import the technology. One of the ways is through foreign direct investment. From past literature, foreign direct investment has a direct effect on growth. Therefore, this variable is also included in this study. The maximum is $15,256,200,000.00, and the minimum is -$596,923,827.79. Mexico is the country with the most foreign direct investment, while Indonesia is the least. The mean is $1,177,258,072.32, and the standard deviation is $2,277,003,579.29. 38 ? ? Figure?7???FDI? ? 8. Openness Trade is also essential to the process of growth. A country should focus on producing goods that it has lower opportunity cost and trade for other goods. The measure of trade openness is also included. There is a consensus among economists that trade liberalization is supposed to aid the process of economic growth. The study done by Yanikkaya (2003) yields ambiguous results. When using variables that measure trade intensity, Yanikkaya found that openness is directly related to growth. However, when he used variables that measure trade barrier such as tariff, he found that the higher the barrier, the higher the output growth. For this paper, openness measure is the sum of export and import divided by total GDP. This is one of the measures that Yanikkaya used for trade intensity. In the sample of 74 countries, the maximum is Malaysia with an 39 ? openness ratio of 2.109, while the minimum is Argentina with a ratio of 0.205. The mean is 0.867, and the standard deviation is 0.416. ? Figure?8???Openness? ? C. Conclusion Chapter IV has presented the conceptual model and data. The first section gives the conceptual model and discuesses the challenge in cross-country data. The the second section of this chapter gives the descriptive statistics of the data and the justification of including each variable. This chapter has laid the groundwork for the next chapter where the regression results are presented. 40 ? CHAPTER V RESULTS This chapter is divided into five parts. The conceptual model in the previous chapter will be estimated by using ordinary least square (OLS). Thus, the first section presents OLS assumptions and consequences of violation of these assumptions. The second section discusses the Breush-Pagan test for heteroskedasticity. The third section discusses the Ramsey?s RESET test of misspecification. The fourth section reports the regression results and discusses the interpretation. Finally, the last section is a concluding remark. A. OLS assumptions In classical least square regression analysis, the following assumptions are essential to the model. The regression model is linear in parameters, y i = ? 1 + ? 2 X i + ? i . The expected value of the random disturbance given the value of X is zero, E(? i | X) = 0. This assumption implies that there is no specification error. In other words, there is no factor in the residual that correlated with the explanatory variables. Another assumption states that the conditional variance of the residual given the value X is constant for or all observations, var(? i | X) = ? 2 . This is a homoskedasticiy assumption. If it is violated, 41 ? then there is heteroskedasticity. Next, there is no autocorrelation. There is no multicolinearity, which is an exact linear relationship among some or all explanatory variables in the model. Finally, the last assumption is normality. If these assumptions hold, the OLS estimates will be BLUE ? best linear unbiased estimator. A violation of OLS assumption can result in an estimator that is not BLUE. An estimator is unbiased when the expected value of the estimator is equal to the true parameter, E(b j ) = ? j . The best estimator is the one with the smallest variance. According to the Gauss Markov theorem, OLS estimators are ?best? at least among all other estimators that are also linear and unbiased if the above assumptions are met. Moreover, the estimates need to be unbiased, consistent, and efficient. An estimate is a consistent estimator of a true parameter, ?, if it approaches the true value of ? as the sample size gets larger without bound. In other words, if plim b = ?, i.e. attain its Rao- Cramer lower bound. An estimator is efficient if has the smallest variance. These conditions are important. It can be shown that the OLS estimator meets these conditions if the assumptions held. The violation of E(? i | X) = 0 implies that there is specification error. One type of specification error is omitting a relevant variable from the model. The consequence of omitting a variable is a biased, inconsistent, and inefficient estimate. A violation of homoskedasiticity results in a variance that is not the smallest and, unless correctly estimated, also inconsistent. The consequence of this is, at the minimum, inflated t or F statistics. A violation of no multicolinearity results in inflated variance and a greater 42 ? probability of making type I error, which is failing to reject the null hypothesis when it is false. Violation of an assumption can have a serious consequence. Therefore, the following sections report some tests to ensure that there is no violation of the assumptions. B. Testing for Heteroskedasticity As mentioned earlier, homoskedasticity is one of the assumptions of OLS. With heteroskedasticity, the estimates are no longer BLUE. The estimates are still linear, unbiased, and consistent. However, they are not the best estimates; they do not have the minimum variance. As a result, the t and F statistics are no longer reliable. To make sure that this assumption is not violated, the Breush-Pagan test is used to test for heteroskedasticity. To do so, first regress the model by OLS and save the residuals and square, e i 2 . Next, create g i = e i 2 /?( e i 2 /n). Regress g i on the independent variables of the model (i.e. g i = ? 1 + ? 2 Z 2i +?+ ? m Z mi + v i ) and save the explained sum of square or SSR. Next construct the LM statistics, LM = ?(SSR) ~ ? 2 m-1 . The null hypothesis is that there is homoskedasticiy, in other words, H 0 = ? 2 = ? 2 =?= ? m = 0. If the null hypothesis is rejected, that means there is heteroskedasticiy. If there is heteroskedasticity in the sample, the OLS variance estimates can be easily corrected by reporting the White?s heterskedasticiy consistent standard error in the final model. C. Testing for Specification Errors 43 ? Specification errors happen in the case of omitted variables, irrelevant variables, incorrect functional form, or errors in measurement. Omitted variable is the case of leaving out the variables that belong in the model. The consequence of omitting a variable is a biased, inconsistent, and inefficient estimate. Irrelevant variable is the case of including too many variables that do not belong in the model. In this case, the estimates are all still unbiased and consistent; however, they are not efficient. To make sure that there is no specification error, the Ramsey?s RESET can be used to test for misspecification. As mentioned in the previous chapter, many variables can be included as the independent variables. Past literature is used as a guideline in terms of what variables to include in the model to be estimated. Therefore, the Ramsey?s RESET is necessary to test for any omitted or irrelevant variables and incorrect functional form. The functional form to be used to estimate the conceptual model of chapter IV is log-log. The steps for the RESET test are following. First, regress the model and obtain the estimated ? i . Next regress the model with additional ? i 2 , ? i 3 , and ? i 4 as independent variables. Construct the F-statistics as follows: F = [(R 2 new ? R 2 old )/3]/[(1- R 2 new )/(n-k-3)] ~ F(3,n-k-3). The null hypothesis is that there is no misspecification. That?s it, H 0 : ? i 2 = ? i 3 = ? i 4 = 0. If the computed F statistics is greater than the critical value, then the hypothesis is rejected and there is specification error (Gujarati 2003). D. Regression Results 44 ? ln GDP per capita = ? 0 + ? 1 ln capital + ? 2 ln education + ? 3 ln service + ? 4 ln political rights + ? 5 ln inflation + ? 6 ln FDI + ? 7 ln openness +? Table 8 ? Regression Results Variable Coefficient Standard Error t-statistics Intercept 0.2507 2.3888 0.105 ln Capital** 0.1446 0.0707 2.045 ln Education* 1.6944 0.2956 5.732 ln Service*** -0.7361 0.5199 -1.416 ln Political Rights* -0.5513 0.1649 -3.342 ln Inflation** -0.1647 0.7540 -2.184 ln FDI 0.0053 0.2402 0.222 ln Openness 0.7294 0.1968 0.371 R 2 = 0.5336 N=74 *significant at the 1% level **significant at the 5% level ***significant at the 10% level The results shown in Table 8 are reported in White?s Standard Error, since the data shows heteroskedasticity. With heteroskedasiticity, the variance is not the minimum 45 ? variance. Thus, the t-statistics are no longer reliable. The White?s Standard Error corrects this problem. The computed F statistics for the RESET test is 0.4419. Therefore, we fail to reject the hypothesis of no misspecificaton. Thus, there are no omitted and irrelevant variables in the model. There are 74 observations in this sample. The regression results report R 2 of 0.5336. This means that the independent variables explain 53.36% of the dependent variables. The capital coefficient is 0.1446. This means that with a 1 percent increase in capital, the GDP per capita will increase by 0.1446. Its t-statistics is 2.045. Therefore, Capital is statistically significant at the 5% level and is positively related to GDP per capita. The relationship between capital and GDP per capita is what anticipated by theory ? more capital means more growth in output. However, the magnitude of the coefficient is quite small. Once again, it is shown that physical capital is essential to growth. Therefore, an economy needs to invest and accumulate physical capital in order to expand their production capability and enable the economy to produce more outputs. Human capital also plays an important part in economic growth. The variable Education is positively related to GDP per capita. Its coefficient is 1.6944, which means a 1 percent increase in education leads to a by 1.6944 percent increase in GDP per capita. It is statistically significant at a 1% level with a t-statistics of 5.732. This means that the null hypothesis that the coefficient is equal to zero is rejected with 99% confident. This result is expected. Education should affect output positively. The higher the training of the labor force, they more productive they are. Thus, the more output an economy can 46 ? produce. This result once again confirms how important education is the process of growth. Developing countries need to invest in education of their labor force for economic growth in the future. The service variable coefficient is -0.7361. It is negatively related to GDP per capita. This means that a 1 percent increase service sector as the percentage of GDP will cause GDP per capita to decrease by 0.736 percent. Service is statistically significant at 10% level with a t-statistics of -1.416. Service sector is used as the measure of maturity of the economy. Measuring output in the service sector presents some challenges because the outputs are intangible. The negative relationship could also means that the economy is not using its resources efficiently. This is a signal that these resources might be better used somewhere else. Political rights are inversely related to economic growth. Its coefficient is - 0.5513. This means a 1 percent increase in political rights, GDP per capita decrease by 0.5513 percent. This makes sense because the index of 1 means the most free. Therefore, the less free the country, the lower its growth. The variable?s t-statistics is - 3.342. It is statistically significant at a 1% level. Therefore, the result suggests that the freer the government, the higher the growth. However, it is important to note that political rights are ordinal data. The countries are ranked. A country with 1 is more free the one with 2 political index; however, how much freer is unknown. This kind of data is hard to quantify. A change in one unit could mean many things. It?s hard to pinpoint exactly what the results imply. Moreover, the political rights variable measure three 47 ? main areas ? electoral process, pluralism and participation, and functioning of government. Thus, an improvement in any of these areas means an increase in output. One of the ways to improve the regress is to include political rights as a dummy variable. Inflation is inversely related to economic growth. The t-statistics is -2.184. Therefore, the hypothesis that the coefficient is different from zero is rejected at the 95% confidence level. The coefficient is -0.1647. That means a one percent increase in inflation causes economic growth to decrease by 0.1647 percent. This result is what was expected. FDI is positively related to GDP per capita. Its coefficient is 0.0053. This means a 1 percent increase in FDI, GDP per capita will increase by 0.0053. However, FDI is not statistically significant. Its t-statistics is 0.222. This is too small thus we cannot reject the null hypothesis of zero. This is consistent with the past finding. In the study by Borensztein, Gregorio, and Lee (1998), foreign direct investment is found to be statistically insignificant in the model. However, when Borensztein, Gregorio, and Lee interact FDI with education, the new variable is statistically significant and is positively related to output. This result leads to the conclusion that the effect of foreign direct investment has on growth depends on the level of education in the country. This result only confirms further the importance of human capita in the process of economic growth. Openness is also positively related to growth. The coefficient is 0.7294. This means that a 1 percent increase in openness, the GDP per capita will increase by 0.7294 percent. However, openness is not statistically significant. Its t-statistics is 0.371. 48 ? As mentioned earlier, political rights are ordinal data. We attempt to improve the model by including political rights as dummy variable. Countries with political rights index of 6 or 7 are assigned 1, whereas countries with political right index of 1, 2, 3, 4, or 5 are assigned the value of 0. In other words, the value of 1 means a country is not free; while the value of 0 means that a country is relatively free. Table 9 ? Regression Results with Political Rights as a dummy variable Variable Coefficient Standard Error t-statistics Intercept -2.5154 2.6437 -0.9510 ln Capital ** 0.1456 0.0746 1.952 ln Education* 1.8718 0.3177 5.891 ln Service -0.3798 0.6575 -0.578 ln Political Rights -0.2884 0.2412 -1.196 ln Inflation** -0.1501 0.0759 -2.014 ln FDI 0.0094 0.2501 0.373 ln Openness 0.1738 0.2115 0.822 R 2 = 0.4614 N=74 *significant at the 1% level **significant at the 5% level 49 ? ***significant at the 10% level The results are reported in Table 9. The results here are reported with White?s Standard Error. The R 2 is 0.4614. The Ramsey?s RESET was not conducted for this model because the previous model is considered to be the best model. Therefore, the test for misspecification is not necessary here. Capital is positively related to economic and statistically significant at the 5% level. Education is also positively related to growth; it is statistically significant at the 1% level with a t-statistics of 5.891. Service is negatively related to output; however, in this model, service is not statistically significant. Political right also has a negative relationship with growth. Inflation is inversely related to growth, and it is statistically significant at the 5% level. FDI and openness are positively related to economic growth. Like the previous model, FDI and openness are not statistically significant. From both models, FDI is found to be statistically insignificant. This is the same result that Borensztein, Gregorio, and Lee found in their study. To improve the model, Borensztein, Gregorio, and Lee interact FDI with education. We follow their lead by interacting FDI with education. The results are found in Table 10. The results here are reported with White?s Standard Error. The R 2 is 0.5336. The results here show that education is not significant. Capital is still statistically significant and directly related to economic growth. The service variable is negatively related to growth, and it is statistically significant at the 1% level. Political rights and inflation are 50 ? also negatively related to growth, and they both are statistically significant at the 1% Table 10 ? Regression Results with interacting FDI and Education (I) Variable Coefficient Standard Error t-statistics Intercept 8.2209 6.4171 1.281 ln Capital ** 0.1356 0.0678 2.001 ln Education 0.0344 1.3347 0.026 ln Service*** -0.8539 0.5228 -1.633 ln Political Rights* -0.5875 0.1637 -3.589 ln Inflation* -0.1767 0.0743 -2.831 ln FDI*** -0.4129 0.3161 -1.306 ln Openness 0.0469 0.1869 0.251 ln FDI*ED*** 0.0953 0.0731 1.303 R 2 = 0.5336 N=74 *significant at the 1% level **significant at the 5% level ***significant at the 10% level 51 ? level. Unlike the previous two models, FDI is statistically significant at the 10% level. The FDI?s coefficient is negative. However, FDI is not negatively related to GDP per capita. FDI is interacted with education. Separating the effect of FDI has on GDP per capita by taking the partial derivative shows that FDI?s coefficient is 0.00839. Thus, FDI is positively related to economic growth. The openness variable is once again found be statistically insignificant, and it is positively related to growth. Finally, the new interaction term between FDI and education is positive, and it is significant at the 10% level. The previous study by Borensztein, Gregorio, and Lee has found that the coefficient of the interaction term is positive, which is the same result that is found here. They suggested that the effectiveness of the foreign direct investment on growth depends on the level of education. However, in this model the education variable is found to be insignificant. Since education is found to be insignificant when including the interaction variable, the next model drops the education and FDI variable and includes the interaction term. Therefore the model becomes: ln GDP per capita = ? 0 + ? 1 ln capital + ? 2 ln service + ? 3 ln political rights + ? 4 ln inflation + ? 5 ln openness + ? 6 ln FDI*education + ? 52 ? Table 11 ? Regression Results with interaction FDI and Education (II) Variable Coefficient Standard Error t-statistics Intercept 2.9919 2.5143 1.190 ln Capital* 0.2559 0.0615 4.158 ln Service -0.2392 0.5247 -0.456 ln Political Rights* -0.6288 0.1831 -3.435 ln Inflation -0.1096 0.8676 -1.264 ln Openness** 0.4207 0.2132 1.937 ln FDI*ED 0.0061 0.0051 1.199 R 2 = 0.4112 N=74 *significant at the 1% level **significant at the 5% level ***significant at the 10% level The results of the regression are presented in Table 11. The estimates are reported with White?s Standard Error. The R 2 is 0.4112. When dropping both education and FDI, the interaction term between FDI and education turns out to be statistically insignificant. Capital is statistically significant at the 1% level. Its coefficient is 0.2559 Political rights are statistically significant at the 1% level, and it is negatively related to growth with a coefficient of -0.6288. The service variable is insignificant. Inflation is 53 ? also statistically insignificant in this model. However, the openness variable for the first time is found to be statistically significant at the 5% level, and it is directly related to GDP per capita with a coefficient of 0.4207. E. Conclusion In this chapter, the first section has discussed the assumptions of OLS and the consequence of any violations. The next two sections present the of assumptions violations. The second section talks about the Breusch-Pagan test of heteroskedasticity, while the third section discusses the Ramsey?s RESET test of misspecification. Finally, the fourth section presents the regression results. Four models were estimated including the original model, the model with political rights as a dummy variable, model with interaction term of FDI and education, and the model with the interaction term excluding the education and FDI variables. These four models give the same results with a few exceptions. Capital and education are found to be positively related to economic growth. Service, political rights, and inflation are found to be negatively related to growth. FDI and openness are found to be directly related to economic growth. However, in many models they are found to be statistically insignificant. The interaction term of FDI and education is also found to be positively related to economic growth. 54 ? CHAPTER VI CONCLUSION From the results in Chapter V, the following can be concluded. Physical and human capitals positively affect economic growth. More physical capitals mean larger production capacity. Similarly, more human capitals mean better trained labor, which also leads to growth. In reality, the situation is more complicated. We know that more physical and human capital will mean more growth. However, the challenge here is how can a country gain more physical capital or human capital? For physical capital to accumulate, first there needs to be saving. But how can a country save if the majority of its population do not have their basic needs met. There will be nothing left to save. The similar situation happens with human capital. There are many people that cannot afford education. Their health prevents them from pursuing education or training to attain better skills. These are the real challenges that are facing in economic development. The percentage of service output and GDP per capita is negatively related. The bigger the service sector actually slows growth. This could be a result of inefficiency. Therefore, the resources that are used in the sector are not being utilized to their optimal use. This inefficiency could point to the need for better training for labor force. 55 ? The political rights variable also shows that the freer the country, the higher the growth. However, problems such as corruption are not easy to correct. The data show that a free nation is good for growth. The practical question then is: how can a country get achieve this goal? A country needs a fair election with transparency. Every citizen, despite of their ethnic or religious background, can freely exercise the right and participate in the political process. There needs to be an effective opposition to those in power to keep the system in check. Voters also need to be educated in order to make a better decision and stay clear of political propaganda. These are some of the issues face by developing countries. The result that inflation is negatively related to growth as expected. Foreign direct investment and openness are not significant in the original model. However, when interacting FDI with education, the new interaction term is found to be positive and statistically significant. Past literatures have found these variables to be related to economic growth. In the case of FDI, it was found that the level of education affect the efficiency of foreign direct investment in a country. Again and again, many of these problems can be traced back to the lack of proper training or education. Education is the root of many things. Better training makes people more productive. Therefore, the resources at their disposal become better utilized. The same can be said about foreign direct investment. In short, highly skilled labors use resources to their optimal use. Education is also important in the political process. Although some corruption is unpreventable, educated voters alleviate the problem to a 56 ? degree. Therefore, these developing countries should focus on better training and education for their population. Will developing countries ever catch up in term of GDP and share the world?s wealth, which is enjoyed by developed nations? If they will, how will that happen? To catch up in terms of their GDP share, these developing countries have to grow at a faster rate. How will that happen? Much more research can be done in this area. The world today is different from two hundred years ago when the Western nations first took off in growth during the Industrial Revolution. Developing countries are growing in different environments. Different factors contribute to economic growth. It is an interesting topic. The tasks of this study were to find out what can allow these nations to catch up with developed nations. Economic development is more than just the sum of total outputs. Economic development means a more equal distribution of income and an improvement in living conditions for everybody. Above all, when basic needs are met, people can lead a more fulfilling life. 57 ? REFERENCE Barro, Robert J. 1991. ?Economic Growth in a Cross Section of Countries.? The Quarterly Journal of Economics, 106(2): 407-433. Barro, Robert J. 1997. Determinants of Economic Growth: A Cross-Country Empirical Study. Cambridge: MIT Press. Barro, Robert J., and Xavier Sala i Martin. 1992. ?Convergence.? Journal of Political Economy, 100(2): 223-251. Borensztein E, J. De Gregorio, and J.W. Lee. 1998. ?How does foreign direct investment affect economic growth?? Journal of International Economics, 45: 115-135. Domar, Evsey D. 1947. ?Expansion and Employment.? American Economic Review, 37(1): 343-355. Gujarati,Damodar N. 2003. Basic Econometrics. 4 th ed. Singapore: McGraw Hill. Eswaran, Mukesh and Ashok Kotwal. 2002. ?The role of service sector in the process of industrialization,? Journal of Development Economics. 68: 401-422. Harrod, Ray. 1948. Towards a Dynamic Economics. London: MacMillan. 58 ? Lucas, Robert E. Jr. 1998. ?On the Mechanics of Economic Development,? Journal of Monetary Ecnomics. 22(1): 3-42. Matsuyama, K. 1992. ?Agricultural Productivity, Comparative Advantage, and Economic Growth.? Journal of Economic Theory. 58: 317-334. Romer, Paul M. 1986. ?Increasing Returns and Long-Run Growth.? Journal of Political Economy. 94(5): 1002-1037. Romer, Paul M. 1994. ?The Origins of Endogenous Growth.? Journal of Economic Perspectives. 8(1): 3-22. Segerstrom, Paul. 1991. ?Innovation, Imitation, and Economic Growth.? Journal of Political Economy. 99(4): 807-827. Solow, Robert M. 1956. ?A Contribution to the Theory of Economic Growth.? The Quarterly Journal of Economic. 70(1): 65-94. Solow, Robert M. 1994. ?Perspectives on Growth Theory.? Journal of Economic Perspectives. 8(1): 45-54. Yanikkaya, Halit. 2003. ?Trade openness and economic growth: a cross-country empirical investigation.? Journal of Development Economics 72: 57-89. ? ? ?