Trade Liberalization, Poverty and Income Distribution in Thailand: A Computable General Equilibrium-Microsimulation Analysis

Transcription

1 Trade Liberalization, Poverty and Income Distribution in Thailand: A Computable General Equilibrium-Microsimulation Analysis Natta Akapaiboon A dissertation submitted to the faculty of the University of North Carolina at Chapel Hill in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Economics. Chapel Hill 2010 Approved by: Patrick Conway, Advisor Alfred J. Field, Jr. David K. Guilkey John D. Kasarda Sudhanshu Handa

2 c 2010 Natta Akapaiboon ALL RIGHTS RESERVED ii

3 Abstract NATTA AKAPAIBOON: Trade Liberalization, Poverty and Income Distribution in Thailand: A Computable General Equilibrium-Microsimulation Analysis. (Under the direction of Patrick Conway.) Economic theory suggests that removing distortions such as tariffs and other trade barriers will cause an economy to operate more efficiently and to grow more rapidly. Since economic growth is generally found to be the main source of poverty reduction, trade reforms that increase growth have been the focus of trade negotiators, researchers, developing countries governments, and international development agencies as tools to reduce poverty. However, since the empirical evidence of trade-growth-poverty linkages is mixed, there is no guarantee that trade liberalization will benefit the poor. The analysis of the impact of trade liberalization on poverty and inequality is complex since trade reform is observed at the macro level while income distribution and poverty issues are observed and analyzed at the micro level. A general equilibrium model or an income generation model based on household survey data alone is therefore not able to fill this micro-macro gap that has characterized the trade-poverty literature to date. Two approaches, based on a macro-micro framework, are developed in this study in order to bridge this micro-macro gap. The framework centers on the communication between a CGE model, which is used to analyze the general equilibrium effects of trade reform, and household survey data, which are used for poverty analysis at the micro level. The simulation results show that trade liberalization has a positive effect on economic growth in Thailand. The manufacturing sector s output expands after trade liberalization, while the output of the agricultural sector declines. There is also a movement of labor out of agricultural sectors into the expanding manufacturing and service sectors after trade reform. At the micro level, household income is found to increase; this is mainly due to an increase in unskilled and skilled wages, which are the main source of income for the majority of Thai iii

4 households. Trade liberalization is found to reduce poverty; this reduction in poverty is greater when the more liberalized trade policy is implemented. However, policies aimed at skill advancement or skill training would ensure the smooth movement of workers from agricultural sectors to manufacturing and service sectors. iv

5 To Mom, Dad, Tai and Tua. v

6 Acknowledgments I am indebted to family, friends, and my dissertation committee for their generous support during my graduate studies. I would like to thank my advisor, Patrick Conway, for his time, invaluable guidance, and continuous encouragement throughout the course of my research. I would also like to thank my committee members: Alfred Field, Jr., David Guilkey, John Kasarda, and Sudhanshu Handa for their thoughtful comments. In addition, I would like to thank Teerawut Sripinit for helping me throughout this research experience and for his continuous support and friendship. I would also like to thank Mark Jensen for his support and for his hours spent on proofreading my drafts. I would not have been able to finish this work without them. I gratefully acknowledge the financial support from the Thai government and UNC Department of Economics during my tenure as a graduate student. I would also like to thank Jennifer Chung-I Li, Yos Vajragupta and Nuntaporn Methakunavut for providing data used in this study. This work would not have been possible without the love and support of my family and friends. I would like to thank Suvanasorn family especially my aunts Suvalai and Somlak Suvanasorn, my uncle and aunt Tum and Tam Siddhichai, my friends John and Araya Coffey, and Sirinee Supornpaibul for always making me feel like home during my stay in the U.S. I would also like to thank my friends in Chapel Hill and Raleigh, especially Pang, Mink, Teau, Joyce and Tanita for their friendship and emotional support. My time in Chapel Hill would not have been the same without them. Lastly, I would like to thank my parents, my sister and my brother for their unending love and support, and for always being there for me. I hope that they are proud of me. vi

7 Table of Contents Abstract iii List of Tables xi List of Figures xviii 1 Introduction 1 2 Poverty, Income Distribution, and Evolution of Earnings in Thailand Introduction Overview of economic growth in Thailand Poverty incidence and income distribution in Thailand Poverty line Poverty measures Poverty incidence Characteristics of the Poor Income distribution Household survey data Earning determination model and evolution of income The Model Estimation results Decomposition of changes in earnings Household demographic and within-household labor force allocation Conclusion vii

8 3 Trade liberalization, Poverty and Income Distribution in Thailand: A Macro- Micro Analysis Introduction Thailand s trade policies: An overview Review of the literature A top-down approach A model with multiple households A macro-micro model Methodology framework Computable General Equilibrium (CGE) model and Social Accounting Matrix (SAM) for Thailand Computable General Equilibrium (CGE) model Social Accounting Matrix (SAM) for Thailand and labor disaggregation Model calibration and CGE base-run Reconciling CGE base-run and household survey Sectoral premium adjustment Labor-type premium adjustment CGE Simulations and Results Simulation scenarios Simulation results Macroeconomic variables Output Exports and imports Factor returns Employment Institution incomes Household Survey Data 127 viii

9 6 A Macro-Micro Model with No Labor Reallocation Methodology framework Results Household income Poverty measures Inequality measures Poverty status before and after trade liberalization Winners and losers from the trade liberalization Conclusion A Macro-Micro Model with Endogenous Occupational Choice Methodology framework Labor reallocation process Results Household income Poverty and inequality measures Poverty status before and after trade liberalization Winners and losers from the trade liberalization Conclusion Conclusions and Directions for Future Research 197 A Characteristics of the Unemployed 202 B Production Sectors 208 C Model Equations 209 D Thailand SAM 223 E CGE Simulation Results 228 F First Approach 238 ix

10 G Second Approach 252 Bibliography 266 x

11 List of Tables 2.1 Thailand s macro economic indicators Exports classified by product group (unit: millions of Baht) Official poverty line from Mean per-capita household income Poverty incidence Poverty incidence Distribution of the poor and the population by geographical locations Distribution of the poor and the population by household characteristics Gini coefficient and GE inequality measures Inequality in per-capita household income distribution Growth and redistribution decomposition of poverty changes Descriptive statistics (mean) of individual and household characteristics Descriptive statistics (mean) of household income Descriptive statistics (mean) of individual and household characteristics - households in the bottom and the top quintiles Descriptive statistics (mean) of household income - households in the bottom and the top quintiles Distribution of labor force Descriptive statistics of workers and earners Descriptive statistics of workers - households in the bottom and the top quintiles Descriptive statistics of earners - households in the bottom and the top quintiles Distribution of earners Descriptive statistics of wage, farm and non-farm profits earners Descriptive statistics of wage, farm and non-farm profits earners - households in the bottom quintile Descriptive statistics of wage, farm and non-farm profits earners - households in the top quintile xi

12 2.24 Wage-earnings estimation (Full sample) Farm profits estimation (Full sample) Non-farm profit estimation (Full sample) Wage-earnings estimation (Subsample) Farm profits estimation (Subsample) Non-farm profits estimation (Subsample) Decomposition of earnings - Wages Decomposition of earnings - Farm profits Decomposition of earnings - Non-farm profits Decomposition of earnings (wages) - Households in the bottom quintile Decomposition of earnings (farm profits) - Households in the bottom quintile Decomposition of earnings (non-farm profits) - Households in the bottom quintile Decomposition of earnings (wages) - Households in the top quintile Decomposition of earnings (farm profits) - Households in the top quintile Decomposition of earnings (non-farm profits) - Households in the top quintile Household s demographic and within-household labor force allocation Household s demographic and within-household labor force allocation - households in the top and bottom quintiles Thailand s key economic indicators Sectoral share of exports and imports in 1998 (percent) Tariff rate in 1998 (percent) Basic SAM structure CGE base-run and the national account - GDP (million baht) CGE base-run and the national account - Sectoral output at factor cost (million baht) Linkage variables from CGE base-run and household survey Sectoral premium Labor allocation from CGE base-run before and after adjusted for the sectoral premium xii

13 3.10 Labor-type premium (β k(j),l ) Labor allocation from CGE base-run before and after adjusted for the sectoral premium (α j ) and labor-type premium (β k(j),l ) Government revenue (billion baht) Simulation results - GDP (1,000 million baht) Simulation results - Real GDP (1,000 million baht) Simulation results - Share of GDP (%) Simulation results - Domestic output (1,000 million baht) Simulation results - Exports (1,000 million baht) Simulation results - Imports (1,000 million baht) Simulation results - Factor returns Simulation results - Employment Simulation results - Income of institutions (1,000 million baht) Descriptive statistics (mean and standard deviation) of household characteristics - all households, poor and non-poor households Descriptive statistic (mean) of household characteristics by regions Descriptive statistic (mean) of household characteristics by regions - poor and non-poor households Simulation results - Household income Simulation results - Household income and household characteristics among the poor Simulation results - Household income and household characteristics among the non-poor Simulation results - Poverty measures Simulation results - Inequality measures Simulation results - Allocation of household by poverty status before and after trade liberalization Simulation results (SIM I) - Household characteristics by poverty status before and after trade liberalization xiii

14 6.8 Simulation results (SIM II) - Household characteristics by poverty status before and after trade liberalization Simulation results (SIM III) - Household characteristics by poverty status before and after trade liberalization Simulation results (SIM IV) - Household characteristics by poverty status before and after trade liberalization Simulation results - Distribution of households by poverty status and households who win or lose from trade liberalization Simulation results (SIM I) - Household characteristics by poverty and winning status before and after trade liberalization Simulation results (SIM II) - Household characteristics by poverty and winning status before and after trade liberalization Simulation results (SIM III) - Household characteristics by poverty and winning status before and after trade liberalization Simulation results (SIM IV) - Household characteristics by poverty and winning status before and after trade liberalization Simulation results - Household income Simulation results - Household income and household characteristics among the poor Simulation results - Household income and household characteristics among the non-poor Simulation results - Poverty measures Simulation results - Inequality measures Simulation results - Allocation of household by poverty status before and after trade liberalization Simulation results (SIM I) - Household characteristics by poverty status before and after trade liberalization Simulation results (SIM II) - Household characteristics by poverty status before and after trade liberalization Simulation results (SIM III) - Household characteristics by poverty status before and after trade liberalization Simulation results (SIM IV) - Household characteristics by poverty status before and after trade liberalization xiv

15 7.11 Simulation results - Distribution of households by poverty status and households who win or lose from trade liberalization Simulation results (SIM I) - Household characteristics by poverty and winning status before and after trade liberalization Simulation results (SIM II) - Household characteristics by poverty and winning status before and after trade liberalization Simulation results (SIM III) - Household characteristics by poverty and winning status before and after trade liberalization Simulation results (SIM IV) - Household characteristics by poverty and winning status before and after trade liberalization A.1 Descriptive statistics of labor force A.2 Descriptive statistics of employed and unemployed workers A.3 Descriptive statistics of labor force - households in the bottom quintile A.4 Descriptive statistics of employed and unemployed workers - households in the bottom quintile A.5 Descriptive statistics of labor force - households in the top quintile A.6 Descriptive statistics of employed and unemployed workers - households in the top quintile D.1 Thailand SAM (1,000 million baht) E.1 Simulation results - CPI consumer price index (PQ-based) E.2 Simulation results - DPI index for domestic producer prices (PDS-based) E.3 Simulation results - EXR exchange rate E.4 Simulation results - PA output price of activity a (% change) E.5 Simulation results - QA level of domestic activity (% change) E.6 Simulation results - PDD demand price for commodity c produced and sold domestically (% change) E.7 Simulation results - QD quantity of domestic sales (% change) E.8 Simulation results - PE price of exports E.9 Simulation results - QE quantity of exports (% change) E.10 Simulation results - PM price of imports (% change) xv

16 E.11 Simulation results - QM quantity of imports(% change) E.12 Simulation results - PQ price of composite good c (% change) E.13 Simulation results - QQ quantity of composite goods supply (% change) E.14 Simulation results - PVA value added price (% change) E.15 Simulation results - QVA quantity of aggregate value added (% change) E.16 Simulation results - PINTA price of intermediate aggregate (%change) E.17 Simulation results - QINTA quantity of aggregate intermediate input (% change)235 E.18 Simulation results - PX average output price (% change) E.19 Simulation results - QX quantity of aggregate marketed commodity output (% change) E.20 Simulation results - PWE world price of exports E.21 Simulation results - PWM world price of imports F.1 Simulation results (SIM I) - Descriptive statistics (mean and standard deviation) of household characteristics - all households, poor and non-poor households F.2 Simulation results (SIM II) - Descriptive statistics (mean and standard deviation) of household characteristics - all households, poor and non-poor households240 F.3 Simulation results (SIM III) - Descriptive statistics (mean and standard deviation) of household characteristics - all households, poor and non-poor households241 F.4 Simulation results (SIM IV) - Descriptive statistics (mean and standard deviation) of household characteristics - all households, poor and non-poor households242 F.5 Simulation results (SIM I) - Descriptive statistic (mean) of household characteristics by regions F.6 Simulation results (SIM I) - Descriptive statistic (mean) of household characteristics by regions - poor and non-poor households F.7 Simulation results (SIM II) - Descriptive statistic (mean) of household characteristics by regions F.8 Simulation results (SIM II) - Descriptive statistic (mean) of household characteristics by regions - poor and non-poor households F.9 Simulation results (SIM III) - Descriptive statistic (mean) of household characteristics by regions F.10 Simulation results (SIM III) - Descriptive statistic (mean) of household characteristics by regions - poor and non-poor households xvi

17 F.11 Simulation results (SIM IV) - Descriptive statistic (mean) of household characteristics by regions F.12 Simulation results (SIM IV) - Descriptive statistic (mean) of household characteristics by regions - poor and non-poor households G.1 Simulation results (SIM I) - Descriptive statistics (mean) of household characteristics - all households, poor and non-poor households G.2 Simulation results (SIM II) - Descriptive statistics (mean) of household characteristics - all households, poor and non-poor households G.3 Simulation results (SIM III) - Descriptive statistics (mean) of household characteristics - all households, poor and non-poor households G.4 Simulation results (SIM IV) - Descriptive statistics (mean) of household characteristics - all households, poor and non-poor households G.5 Simulation results (SIM I) - Descriptive statistic (mean) of household characteristics by regions G.6 Simulation results (SIM I) - Descriptive statistic (mean) of household characteristics by regions - poor and non-poor households G.7 Simulation results (SIM II) - Descriptive statistic (mean) of household characteristics by regions G.8 Simulation results (SIM II) - Descriptive statistic (mean) of household characteristics by regions - poor and non-poor households G.9 Simulation results (SIM III) - Descriptive statistic (mean) of household characteristics by regions G.10 Simulation results (SIM III) - Descriptive statistic (mean) of household characteristics by regions - poor and non-poor households G.11 Simulation results (SIM IV) - Descriptive statistic (mean) of household characteristics by regions G.12 Simulation results (SIM IV) - Descriptive statistic (mean) of household characteristics by regions - poor and non-poor households xvii

18 List of Figures 2.1 Earnings-experience partial relationship: wage earnings Earnings-experience partial relationship: farm profits Earnings-experience partial relationship: non-farm profits Earnings-experience partial relationship Production technology structure Commodity flows in the CGE model xviii

19 Chapter 1 Introduction The increased importance of trade liberalization, particularly as advanced under WTO s Doha agenda, has been the focus of trade negotiators, researchers, governments in developing countries, and international development agencies as a tool to reduce poverty. Economic theory suggests that removing distortions such as tariffs and other trade barriers will cause an economy to operate more efficiently and to grow more rapidly. Since economic growth is generally found to be the main source of poverty reduction, trade reforms that increase growth seem likely to also benefit the poor. Thailand has gradually become a more open economy in recent years and has been continually opening its economy to foreign trade. However, since the empirical evidence of tradegrowth-poverty linkages is mixed, there is no guarantee that Thailand s trade liberalization will benefit the poor. In addition, the extent to which growth translates into poverty reduction depends on both the existing distribution of income and the distributional shares of benefits generated from growth. The links between economic growth, which is observed at the macro level, and poverty, which is observed at the micro level, consist of a complicated interaction of labor market, demographic, and education dynamics which is often overlooked when poverty conclusions are drawn from aggregate measures such as the poverty headcount index or a country s economic growth rate. In Chapter 2, I investigate the evolution of income in Thailand over the period of in order to gain a better understanding of how labor market, demographic, and educational

20 factors affect different types of earnings, and thus the process of poverty reduction in Thailand. An earnings determination model is applied to each type of an individual s earnings directly. I also employ a decomposition methodology to identify factors that contribute to the observed changes in earnings in Thailand. Chapter 3 and the rest of the study evaluate the impact of trade liberalization on poverty and income distribution in Thailand. The analysis is based on the Thai economy and the tariff rate observed in 1998 and evaluate what would have happened to income, poverty and income distribution in Thailand if the tariffs were completely removed. Therefore, while Chapter 2 represents the dynamics analysis of income over the period of , the analysis in the rest of the study is static and counterfactual. The analysis of the impact of trade liberalization on poverty and inequality is complex since trade reform is observed at the macro level while income distribution and poverty issues are observed and analyzed at the household (micro) level. Two approaches based on a macro-micro framework are developed in this study to evaluate the impact of trade reform on poverty and income distribution in Thailand. The framework centers on the communication between a CGE model, which is used to analyze the general equilibrium effects of trade reform, and household survey data, which are used for poverty analysis at the micro level. Chapter 3 reviews the relevant literature and presents the methodology used in this study. It also describes the CGE model and the Social Accounting Matrix used to analyze the general equilibrium effects of trade liberalization. In order to implement successfully both of the macro-micro approaches, consistency between the CGE base-run outcomes and the household survey data needs to be achieved first; the methods used to achieve this consistency are also described in Chapter 3. Once the reconciliation is achieved, the CGE model can be used to perform simulations for different trade scenarios. Chapter 4 reports the general equilibrium results of trade liberalization from CGE simulations. Trade liberalization is found to have a positive effect on economic growth. The results from CGE simulations show an increase in GDP ranging from 0.07% to 0.16%. The manufacturing sector s output expands after trade liberalization, while the output of the agricultural sector declines. Wages for both skilled and unskilled labor are found to increase, with skilled labor benefiting slightly more from trade reform. Returns to 2

21 capital in manufacturing sectors increase more from trade reforms than in agricultural sectors. Consistent with the simulation results on output, employment in manufacturing and service sectors expand and absorb labor out of agricultural sectors. This is true for both unskilled and skilled employment. Chapter 5 describes in more detail the household survey data that is linked with the outcomes from the CGE simulations for poverty analysis. It also describes important characteristics of Thai households. Chapter 6 describes the methodology of the first approach in which the linkage from the CGE model to the household survey data is done through wages and profits, which are the main sources of individual and household income. Once individual income data is updated by the CGE simulation outcomes, household income can be calculated and an analysis on poverty and income distribution impact of trade liberalization can be conducted. Chapter 6 also reports the results on household income, poverty, and income distribution which is done based on the updated income data. Chapter 7 describes the methodology and the results of the second approach in which, as in the first approach, the CGE model is linked to the household survey data through wages and profits but the assumption of exogenous labor occupational choice in the first approach is relaxed. In other words, the model allows labor to move across production sectors to match the employment outcomes predicted by the CGE model. Given the different initial tariff levels experienced by different production sectors, it is likely that, after removing all tariffs, some sectors will expand while other sectors will shrink. This will affect employment and factor returns, which are the main sources of individual and household income. Thus, by removing the assumption that workers have to remain in the same sector after trade liberalization, the model is able to provide more realistic results. Both approaches are used to examine different scenarios of trade reforms and to evaluate their effects on household income, poverty and income distribution. These models are also used to examine if the poor share in the gains from the freer trade and if so, whether their gain is greater or less than the gain of the non-poor. In addition, the analysis is performed on various demographic groups in Thailand to see if any of these groups are particularly impacted by trade liberalization. These results should benefit policymakers by allowing them to develop 3

22 policies that can mitigate any negative effects trade liberalization may have on particular demographic groups. The results from the analysis are shown in both Chapter 6 and Chapter 7. Under both approaches, household income is found to be increasing after trade reform, especially through an increase in wage income. The formal labor market proves to be an important channel that transmits the growth effect of trade reform to households. Trade liberalization is found to reduce poverty in Thailand and the poverty effect is greater when the more liberalized trade policy is implemented. Income inequality also declines after trade liberalization under both approaches. The results from the first approach, in which labor mobility is assumed to be fixed, show that trade reforms are desirable in terms of poverty incidence. Free trade allows some of before poor households to escape poverty and none of the before non-poor households fall back below the poverty line. Under the second approach, the number of the poor is found to decrease more but the depth of poverty, even if improving, is found to be greater than in the first approach. When labor is allowed to reallocate after trade reform, we also see an importance of non-farm self-employed jobs. However, not everyone will enjoy the higher returns in manufacturing and service sectors unless he/she is equipped with the skill required by the new industry. Therefore, policies aiming at skill advancement or skill training should be implemented in order to insure the smooth movement of workers from agricultural sectors to manufacturing and service sectors. Chapter 8 concludes and discusses the direction for future research. 4

23 Chapter 2 Poverty, Income Distribution, and Evolution of Earnings in Thailand 2.1 Introduction Thailand has experienced impressive economic growth in recent decades. Its gross domestic product (GDP) grew at an annual rate of almost 8 percent during the 1980s through the late 1990s. Economic growth has clearly been a powerful force in reducing poverty in Thailand. Poverty incidence (measured by the poverty headcount index) fell from 51.9 percent in 1990 to 26.2 percent in As the reduction of poverty did not come about as a result of any specific poverty alleviation plan Krongkaew (2002), this poverty reduction seems to support much of the previous literature that found that growth is the main source of poverty reduction (Deolalikar (2002), Kakwani and Krongkaew (1997), Paitoonpong et al. (2001) for Thailand; Dollar and Kraay (2002), Bruno et al. (1998) for surveys). Although the Thai government has been successful in generating economic growth and reducing the incidence of poverty, higher income inequality is also observed during the fast growth period. During 1990 to 2002, income inequality, as measured by the Gini coefficient, remained high around The households in the bottom quintile earned about 20 times less than households in the richest quintile during this period. Households in the bottom 1 The inequality in Thailand is relatively high compared to the rest of the world. In 2002, the highest inequality in the world was found in Sierra Leone with the Gini index of 0.63 (World Bank (2002)).

24 quintile share only about 3 percent of the country s income while 58 percent of the country s income goes to households in the top quintile. The extent to which growth translates into poverty reduction depends on both the existing distribution of income and the distributional shares of benefits generated from growth. Ravallion (2001) found that given existing inequality and assuming distribution-neutral growth, the income gains to the richest decile in India, a country where the Gini coefficient is under 0.30, will be about four times higher than the gain to the poorest quintile, while it will be 19 times higher in Brazil, where the Gini index stays high at around With the existing high level of income inequality in Thailand, it is clear that the gains of economic growth are disproportionately distributed to the rich. In addition, the evolving distribution of income during the growth process matters for poverty reduction. If inequality increases with economic growth, the benefits accruing to the poor will be less than what could have been achieved under distribution-neutral or pro-poor growth. On the other hand, if distributional shares of income during the growth process are pro-poor, it will improve the existing distribution of income and benefit even more to the poor. Therefore, it is not only a question of how much growth can reduce poverty that we are interested in, but also how inequality changes during the process of economic growth. The links from economic growth (which are observed at the macro level) to poverty reduction (observed at the micro or individual level) often occur through a combination of labor market, demographic, and educational dynamics. Even though poverty analysis based on aggregate poverty measures (such as the poverty headcount index or the Gini coefficient) can give us a broad picture of the growth-poverty relationship, its use to identify labor market or educational factors that are keys for poverty reduction is rather limited. For instance, comparing incomes of the households after the 1998 financial crisis in Russia with incomes of the same households in 1996, Lokshin and Ravallion (2000) found that poverty increased only two percentage points. However, what lies underneath the averages is the fact that a large proportion of the population (18%) fell into poverty, while a slightly smaller proportion (16%) escaped poverty over the same period. Eighteen percent of the Russian population could be ignored if the government thought that a-two-percentage-point increase in poverty was not 6

25 a serious problem and deserved little attention. Furthermore, inequality measures such as the Gini index may seem to change relatively little over time. Even small changes in overall distribution, however, can matter greatly to how much the poor share in growth. Given the complicated interaction between labor market, other demographic factors, and poverty reduction, it is important to look beyond those aggregate measures in order to understand better the process of poverty reduction. Since household income is used as a welfare measure for poverty analysis, understanding how it changes over time should give us a better picture of the factors determining income and how they affect poverty. However, a household income determination model is likely to be explained by average characteristics of all of its employed members, along with some household-level demographic variables such as household size or dependency ratio. As we all know, household income is composed of different types of earnings. A variable positively affecting one type of earnings could work in the opposite direction for another earnings type. An aggregate household income model will fail to provide such detailed information since what is captured in the aggregate household income model is merely the end result (after the canceling out). Thus, to move beyond the aggregate measures, in this chapter, the distribution of each type of individual s earnings is analyzed directly. An earnings determination model will be employed to identify the impact of demographic and educational factors that affect each type of earnings and its distribution over the period of in Thailand. This will help us understand better the evolution of household income and thus the process of poverty reduction. The model is based on seven years of nationally representative household surveys which have rich information on individual s income, occupation, education, other demographic variables (such as age and sex), and household-level demographic variables (such as household size and household location). Based on the estimation results of the earnings model, I also employ a decomposition methodology to identify factors contributing to the observed changes in earnings. It is found that, in general, education is the most important determinant of wage earnings while it is not as significant for income of the self-employed. Profits depend on the productivity of unpaid family members which is consistent with the fact that businesses in Thailand are often small family businesses that employ family members. It is also found that an increase in wage and non-farm profits during the early 1990s was 7

26 explained mostly by the returns to education and experience of the workers while in late 1990s and early 2000s, the returns to such characteristics had become less powerful. During the late 1990s and early 2000s, education attainment among wage and non-farm self-employed workers drastically improved and it became the most important contributing factor to the increase in those earnings. However, the improvement in education among farm workers is insignificant. Most of the increase in farm income during 1990s and early 2000s was explained by the returns to farmer s experience and unpaid family workers. We know that some characteristics of the rich and the poor can be so different that they cause much of the income disparity between them over the period. However, what is so specific about the poor that prevents them from fully taking advantage from growth? Can those characteristics be identified so that special programs or some growth patterns should be encouraged over the others? To answer these questions, two subsamples of the population are compared: 1) earners who reside in a household whose income is in the bottom quintile and 2) those from the richest quintile. The results from the model estimation and decomposition show that education plays a less significant role in determining income of the poor. In addition, the majority of their income is from farm profits which are relatively low. The poor also live in bigger households with high dependency ratios that put more pressure on poverty reduction process. The chapter is organized as follows. The next section provides a brief overview of growth in Thailand, while Section 2.3 presents more detailed background information of changes in poverty and income distribution, and characteristics of the poor in Thailand over the study period. Section 2.4 describes the data used in the study. Section 2.5 outlines the basic model and presents the empirical results. Section 2.6 describes the decomposition methodology and decomposes the observed changes in Thailand s earnings distribution between 1990 and 2002 drawing on the basis of the models estimated in Section 2.5. Section 2.7 concludes. 8

27 2.2 Overview of economic growth in Thailand Thailand experienced an impressive annual growth rate of more than 8 percent on average from 1986 to The Asian economic crisis hit the country in It took the country two years to recover from economic crisis. The economic growth rates during the period of 1999 to 2006 averaged almost 5 percent per annum. Real GNP per capita at constant 2002 prices increased from 39,994 baht in 1986 to 100,610 baht in Table 2.1 reports the basic economic indicators of Thailand from 1986 to Thailand s high economic growth rate was mainly a result of several National Economic and Social Development plans that set economic growth as a primary goal during the period from 1960s to 1990s. From time to time, poverty alleviation and more equal income distribution, along with energy concerns and environmental degradation, have been included in the Plan. The main objective of the national development plan during this period, though, centers on economic growth. Before 1960, Thailand was a low-income country with 85 percent of its total population engaged in the agricultural sector. Its main exports consisted of rice, rubber, teak and tin (Krongkaew (2002)). With help from the World Bank, the national development plan was introduced for the first time in 1961 with its main goal of generating economic growth and expansion. During the 1960s and 1970s, import substitution and promotion of capital-intensive manufacturing products were the main development strategies. As most of import substitution was in capital-intensive products, an expansion of capital-intensive sectors created demand for skilled labor which benefited the rich who had the right skill sets. The policies had limited impact on agriculture from which most of the poor earn their income. Therefore, the policies likely contributed to rising income inequality in Thailand since 1970s. On the other hand, export promotion, which is mostly on labor-intensive products, is a good strategy for poverty reduction since it creates jobs for the poor. In the mid-1980s, import substitution was replaced by export promotion. Exports have been the main driving force for economic growth in Thailand since then. 9

28 Table 2.1: Thailand s macro economic indicators Population (Million persons) GDP 2.1 GDP at constant 1988 price (Billions of Baht) 4, , , , , , , , , , , , , , , , , , , , , (% change) Agriculture (Billions of Baht) 1/ (% change) (share of GDP (%)) Non-agriculture (Billions of Baht) 1/ 3, , , , , , , , , , , , , , , , , , , , , (% change) (share of GDP (%)) GDP at current price (Billions of Baht) 7, , , , , , , , , , , , , , , , , , , , , (% change) GNP per capita (Baht : Person) 115, , , , , , , , , , , , , , , , , , , , , Real GNP per capita (Baht : Person) 100, , , , , , , , , , , , , , , , , , , , , Inflation 3.1 Headline Consumer Price Index (2002=100) (% change) Core Consumer Price Index (2002=100) 2/ (% change) Notes: 1/ The NESDB has reclassified the GDP by industry to be followed the Thailand Standard Industrial Classification (TSIC) 2001 version since (Formerly used TSIC 1972 version) 2/ Exclude raw food and energy items from the consumer price index basket. Source: Bank of Thailand Table 2.2: Exports classified by product group (unit: millions of Baht) Agriculture 475, , , , , , , , , , , , (share of total exports (%)) Manufacturing 4,396, ,978, ,411, ,895, ,536, ,481, ,409, ,885, ,872, ,508, ,164, ,161, (share of total exports (%)) Samples & other unclassified goods 58, , , , , , , , , , , , Re-exports 7, , , , , , , , , , , , Total exports 4,937, ,438, ,873, ,325, ,923, ,884, ,773, ,215, ,248, ,806, ,412, ,406, Source: Bank of Thailand 10

29 During the 1990s, as the country experienced fast growth, tax reform and liberalization of the domestic financial sector also facilitated investment from abroad. With low interest rates prevailing in developed economies and the country s exchange rate fixed at the time, Thailand attracted capital inflow due to its high interest rates. Easier access to foreign resources had spilled over from tradable sectors (especially those producing goods for export) into nontradable assets, particularly stocks and real estate, creating an asset price bubble. At the same time, on the external side, due to slowing exports, current account deficits had been persistent from the mid 1990s. This put pressure on the central bank to keep the exchange rate fixed by incurring a large foreign debt. The economic growth that relied on unproductive sectors such as those in real estate came to an end in mid The asset price bubble burst with falling property prices and thus a sudden outflow of foreign capital. The outflow of capital, rising current account deficits and large foreign debt forced the Thai government to float the baht in July After the economic crisis in 1997, learning that the fast economic growth which depends solely on foreign capital is not sustainable, policymakers introduced new development plans which focus instead on the concept of human capital development with a moderate and sustainable level of economic growth. It was the first time that fast economic growth was no longer the main goal of the development plan. At present, Thailand is back on its long-term growth path. Annual growth rates from 1999 to 2006 averaged more than 4.5 percent. The national development plans and development strategies during the past three decades have played an important role in transforming Thailand from a poor, agrarian economy to a middle-income, newly industrialized country. Substantial structural changes took place with domestic production moving away from agriculture to manufacturing and towards services. From Table 2.1, agricultural production s share of GDP decreased from almost 20 percent in 1986 to less than 9 percent in 2006, while the share of manufacturing and service sectors increased commensurately. Thailand s exports during the past two decades experienced a dramatic structural change as well. Exports of manufactured products increased significantly from 6.1 percent of total exports in 1970 to 82.6 percent and 89.1 percent in 1995 and 2006, 11