Microsoft Word Thesis Tran Luu Quoc Vuong docx UNIVERSITY OF ECONOMICS HO CHI MINH CITY VIET NAM INSTITUTE OF SOCIAL STUDIES THE HAGUE THE NETHERLANDS VIET NAM –THE NETHERLANDS PROGRAMME FOR MASTER OF[.]
UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES HO CHI MINH CITY THE HAGUE VIET NAM THE NETHERLANDS VIET NAM –THE NETHERLANDS PROGRAMME FOR MASTER OF DEVELOPMENT ECONOMICS DETERMINANTS OF ECONOMIC GROWTH IN A PANEL OF ASIA PACIFIC COUNTRIES FINAL THESIS Supervisor: Nguyen Hoang Bao Student: Tran Luu Quoc Vuong Class: MDE-16 -HO CHI MINH CITY, DECEMBER 2013 APPENDICES A: List of chosen countries and territories in Asia Pacific Region 51 B: Scatterplot matrix of variables 52 C: Tests 53 LIST OF FIGURES Figure 3.1 Conceptual framework of the study 14 Figure 4.1 Growth rate versus Level of Per capita GDP (simple relation) 27 LIST OF TABLES Table 3.1 Summary of variables 20 Table 4.1 Summary Statistic on the sample observations 29 Table 4.2 Correlation on the sample observations 30 Table 4.3 Basic Cross-Country Growth Regression 31 Table 4.4 Heteroskedasticity – White test 32 Table 4.5 Basic Cross-Country Growth Regression (robust) 33 Table 4.6 Multicollinearity testing 34 Table 4.7 Restricted Model (1) 35 Table 4.8 Restricted Model (2) 36 Table 4.9 Model Comparison 37 Table 4.10 Extended models 43 ABBREVIATIONS IMF : International Monetary Fund WB : World Bank UN : United Nations GDP : Gross Domestic Products GNP : Gross National Products R&D : Research and Development OECD : The Organization for Economic Co-operation and Development Table of Contents CHAPTER 1: INTRODUCTION 1.1 Problem statement 1.2 Objectives of study 1.3 Research questions 1.4 Scope of the research 1.5 Structure of the thesis CHAPTER 2: THEORETICAL AND EMPIRICAL BACKGROUND 2.1 Theoretical background: 2.2 Empirical review 2.3 Summary 12 CHAPTER 3: RESEARCH METHODOLOGY 13 3.1 Conceptual framework 13 3.2 Variables 14 3.2.1 Dependent variables 14 3.2.2 Explanatory variables 15 3.3 Data 22 3.4 Econometric model 22 3.5 Summary 24 CHAPTER 4: DATA ANALYSIS .26 4.1 Growth rate and convergence 26 4.2 Descriptive analysis 27 4.3 Regression Results for Growth rates 30 4.3.1 Basic Regression 30 4.3.2 Hypothesis tests 32 4.3.3 The Determinants of Growth rate 39 4.3.4 The extended models 42 4.4 Summary 44 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal CHAPTER 5: CONCLUSION AND POLICY IMPLICATIONS 45 5.1 Main findings and policy implications 45 5.2 Limitation of the research and Recommendation for further studies 45 REFERENCES 48 Tran Luu Quoc Vuong – MDEk16 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal CHAPTER 1: INTRODUCTION 1.1 Problem statement Nowadays, there are over 200 countries on over the world with different institutions, with different growth rates These growth rates vary hugely across countries and between developing countries to developed countries According to recent surveys of many international organizations (IMF, WB, UN and so on), developing countries and economies in transition continue to register much stronger growth than developed economies After the end of the World War II, the world started to recover economic and the growth became an important part of economic science Economists have tried to apply other theories with a lot of empirical model to understand the process of economic growth and explain enormous differences in economic performance across countries In the early, economists used Keynesian theories to discuss about the stability of the growth path Next, they focused on steady state growth rates exogenously determined by technological progress After that, Solow (1956) and Swan (1956) became the pioneers who enriched neoclassical economic growth model, and followed by a lot of others From 1980s, economists developed growth theory, which use endogenous economic growth model to evaluate the roles of knowledge, innovation and human capital Improvements of economic theories and hundreds of researches have provided acceptable explanation for a particular phenomenon of recent decades: the great divergence in economic performance But, indeed, the income of underdeveloped economies, as a whole, are failing to catch up the income of the developed countries Realizing this inadequacy, economic growth studies have shifted their focus to find out the reason of this problems There were a lot of researches discuss about this problem, such as: Solow (2000), Aghion and Howitt (1998), Kenny and Williams (2001), Temple (1999), Tran Luu Quoc Vuong – MDEk16 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal Bosworth and Collins (2003) and Rogers (2003) They identified main contributions to the recent growth literature, especially relevant institution factors The main objects these economists aimed were developing countries in the Asian Pacific area They wanted to understand how recent developments in the growth literature contribute to economic development of Asian Pacific countries In this study, the research examine the relationship between per capita growth rate to a set of quantifiable explanatory variables with a conditional convergence term: the growth rate rise when the initial level of real per capita GDP is low relative to the staring human capital and for given values of other policies, institutions and national characteristic variables This research will use panel data of 25 countries and territories of Asian Pacific region for the period of 1990 - 2010 This study use panel data approach and apply three-stage less squared regression to examine the determinants of the growth rate The regression, which uses panel data, has many advantages with regards to typical cross-country regression that was advanced by Raftery (1995), Sala-i-Martin et al (2004) First, using panel data, we will eliminate constraint by limited number of countries available Second, this method allows addressing the inconsistency of empirical estimates This inconsistency appears with omitted country-specific effects correlated with other regressors, or with endogenous variables which assumed to be exogenous 1.2 Objectives of study This research will examine the relationship between per capita growth rate and a set of quantifiable explanatory variables: human capital (education and health), policies, institution and national characteristics From empirical studies, for given initial levels of state variables (including physical and human capital), the economic growth rates have positive relation to the rule of law, the investment ratio and negative relation to the fertility rate, the ratio of government consumption to GDP and the inflation rate In the case of the terms of trade and the extent of Tran Luu Quoc Vuong – MDEk16 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal international openness, with fair movements, the growth will increase but the latter effect is surprisingly weak The research will define which factors affect to the growth rate of Asian Pacific developing countries, including Vietnam It also extend to evaluate some special countries, which have huge population size, such as: China, India; or developed countries, such as Japan, Australia, Singapore and so on 1.3 Research questions This research aims at answering the following questions: The relationship between per capita growth rate and a set of quantifiable explanatory variables? Which variables are significant ones to economic growth? Intensity of impact of these factors to economic growth of the Asia Pacific region? 1.4 Scope of the research The research focuses on examine the convergence hypothesis implies that the growth rate of real per capita GDP would tend to be inversely related to the level of initial real per capita GDP, and the role of a set quantifiable explanatory variables The chosen countries are 25 countries in Asian Pacific region, and some special countries which have huge population size (China, India) or specific characteristics (Japan, Singapore) This research only examines in the period of 1985 2010 because lacking of GDP data before 1980 for all countries in this region 1.5 Structure of the thesis The research will be organized with five chapters The chapter one has structure as above: explain the reason why we choose this topic, identify the problems that the research need to examine with some major research questions and the scope of the research The rest of this research has four chapters Chapter two Tran Luu Quoc Vuong – MDEk16 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal shows a brief view of theoretical background of growth model, from the neoclassical growth models to the endogenous growth models, and the role of trade and institutions on the growth rate as well A review of empirical studies makes clearly about results of different models when applying to explain recent developments In chapter three, we will highlight the research methodology, including conceptual framework and variable descriptions We will have a briefly discussion about the reason why we choose these variables, how to enter them into the equations after introducing the general econometrics model Besides, the process outline which we use to analyze data, is also mentioned in this chapter Chapter four will show the regression results of the models We will analyze and get findings to answer the questions in the chapter We also compare findings to results of other empirical studies And at chapter five, we resumes main findings and reevaluate factors which have significant influence to growth rate of an economy This chapter also identifies the limitation and implications for further study Tran Luu Quoc Vuong – MDEk16 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal CHAPTER 2: THEORETICAL AND EMPIRICAL BACKGROUND 2.1 Theoretical background: In the 1950s, growth theory was composed mainly of the neoclassical model, developed by Ramsey (1928), Solow (1956), Swan (1956), Cass (1965) and Koopmans (1965) One feature of this model is the convergence property with the characteristic of diminishing returns to capita in a closed economy, where technological progress is assumed to be exogenous and there are same technical opportunities across countries This theory said that, the predicted growth rate will higher with the lower starting level of real per capita gross domestic product If all economies were naturally the same, except for their starting capital intensities, the convergence would apply in an absolute sense; it means that, poorer economies typically grow faster per capita and tend thereby to catch up to the richer economies However, if they differ from a lot of factors, such as rate of birth, willingness to work, technology accessing, government policies, the convergence force applies only in a conditional sense The predicted growth rate will be higher if the starting level of per capita GDP is low in relation to its long–run (steady state) position For example, a poor country would not tend to grow rapidly if it also has a low long–term position which possibly because its public policies are unfavorable or its saving rate is low But along the time has been changing, this model showed its own limitation So, a new wave rose in economic growth theory Growth theorists of 1960s assumed that technological progress is exogenous to the growth rate, will make the economy temporarily grows in short term but not in long term But, unfortunately, while endogenous factors, like as the saving rate and population growth, determine the steady state or ceteris paribus of an economy, the technological progress decides the growth rate of the steady state Thus, we can use model of growth to explain everything but long-run growth Tran Luu Quoc Vuong – MDEk16 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal mechanism with regards to capital accumulation They showed that increase in inflation rate leads the level of output permanently fall In endogenous growth model, the growth rate depends on the rate of return on capital So, inflation decrease that rate, which in turn reduces capital accumulation and decrease the growth rate And, when this model is set within moneytary exchange framework, the inflation rate lowered both the return on all capital and the growth rate (Lucas 1980, Lucas and Stokey 1987, McCallum and Goodfriend 1987) In this research, we predict that the inflation has a negative relationship to the growth rate Inflation rate will be enters into the basic system in the form of the average inflation rate, and used as a measure of macroeconomic stability Table 3.1 Summary of variables: ExVariables Description pected Main empirical reference sign GDP_Growth Growth of per capita Init_GDP Barro (1995, 1997, 2003), Bülent GDP Ulaşan (2012) The initial level of per Lucas (1988), Rebelo (1991), Caballe capita GDP and Santos (1993), Mulligan and - Sala-i-Martin (1993), Barro and Salai-Martin (1995), Barro (2003), Nelson and Phelps (1966), Ben habib and Spiegel (1994) Education at- - Primary-school entainment rollment rate Tran Luu Quoc Vuong – MDEk16 + Barro (1991, 1995), Hanushek and Kimko (2000) 20 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal - Secondary-school enrollment rate Life Expec- the reciprocal of life tancy expectancy at age one Barro (2003),Daron Acemoglu and +/- Simon Johnson (2006), David de la Croix and Omar Licandro (1999) Fertility The fertility rate Barro and Sala-i-Martin (1995), - Barro (2003), Quamrul H Ahraf, David N Weil and Joshua Wilde (2012) Gov_share the ratio of govern- Barro (1991, 1995), Sala-i-Martin ment consumption to (1995), Landau (1983), Grier and GDP - Tullock (1987), Kormedi and Meguire (1985), Barth and Bradley (1987) Openness the extent of interna- Yanikkaya (2003), Barro (1995), Lu- tional openness cas (1988), Young (1991), Rivera+/- Batiz and Romer (1991), Grossman and Helpman (1991), Coe and Helpman (1995), Bülent Ulaşan (2012) Investment the ratio of real gross Sala-i-Martin (1995), Barro (1995, domestic investment 1997, 2003), M Knight, N Loayza to real GDP + and D Villanueva (1992), Kahn and Kumar (1997), Kahn and Reinhart (1990), Serven and Solimano (1992), Greene and Villanueva (1991) Tran Luu Quoc Vuong – MDEk16 21 Determinants of economic growth in a panel of Asia Pacific countries Inflation the inflation rate Thesis Proposal Fisher (1993), Barro (1995), Lucas - (1980), Lucas and Stokey (1987), McCallum and Goodfriend (1987) 3.3 Data The dataset consists of 25 Asian Pacific countries and territories We collect their data in the period of 1990 - 2010 The panel data from various sources but mainly draws from World Bank’s World Development Indicators (WDI, 2012) The definition and exact calculated measurement present in the next part 3.4 below 3.4 Econometric model Based on neoclassical growth model, a country’s per capita growth rate in period t, Dyt, can be express as: Dyt = f(yt-1,ht-1) (1) Where: Dyt is the growth rate of per capita output at time t, is diminishing in the output, y, for given y* yt is the level of per capita output at time t, the years end of each period y* is the long-run or steady state level of per capita output The target value y* depends on an array of choice and environmental variables, include saving rates, labor supply, fertility rates, government spending, tax rates, the extent of distortions of markets and business decisions, maintenance of the rule of law and property rights, the degree of political freedom, terms of trade and so on Tran Luu Quoc Vuong – MDEk16 22 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal From the equation (1) and review of empirical study, the most general econometric form testing the determinants of economic growth can be expressed as following: Growth = α + β1*log(Init_GDP) + β2*pri_school + β3*sec_school + β4/expectancy + β5*log(fertility) + β6*gov_share + β7*openness + β8*investment + β9*inflation Where: Growth: the represent the growth rate of per capita GDP for 1990-2000 and 2000-2010, plays the role as dependent variable Init_GDP: is an observation of the log of real per capita GDP in 1990 and 2000 In addition, the value of the log of per capita GDP in 1985 and 1995 are also included into the model as instruments Pri_school: is the ratio of total enrollment in primary education in 1990 and 2000, regardless of age, expressed as a percentage of the population of official primary education age Sec_school: is the ratio of total enrollment in secondary education in 1990 and 2000, regardless of age, expressed as a percentage of the population of official secondary education age Expectancy: is the life expectancy at birth This variable is used as an instrument and uses data of the year 1985 and 1995 Fertility: total lifetime live births for the typical woman over her expected lifetime This variable appears in instrument list and uses data of the year 1985 and 1995 Gov_share: are the averages for 1990-1999 and 2000-2009 of the government consumption This variable also uses average values for 19851989 and 1995-1999 as instruments Tran Luu Quoc Vuong – MDEk16 23 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal Openness: The degree of international openness is measured by the ratio of exports plus imports to GDP for 1990-1999 and 2000-2009 This variable is also includes as an instrument by using lagged value, average value for 1985-1989 and 1995-1999 Investment: The ratio of real gross domestic investment (private plus public) to real GDP enters into the regressions as averages for tenyear periods (1990-1999 and 2000-2009) This variable is also included as an instrument by using lagged value, average value for 1985-1989 and 1995-1999 Inflation: is the average rate of retail price inflation over each of the ten-year periods (1990-2000 and 2000-2010) This research uses three-stage least squares to estimate the determinants of the growth rate with a list of instruments To illustrate the objectives of the study, we will focus on some tasks as following First, we will find out the relation between variables and its instruments, such as: the initial level of per capita GDP, the investment rate, the government consumption ratio and the extent of international openness by regressing the endogenous predictors on the exogenous variables Next, we predict the values of these variables based on the results of those regressions And the third step, we use the predicted variable as the instrumented variables in our model During analysis process, we will apply White test and Wald test to checking 3.5 Summary In this chapter, conceptual framework establishes based on the empirical studies and theoretical background of convergence hypothesis in growth model The growth rate of per capita GDP is chosen as dependent variables and a set of quantifiable variables is used as explanatory variables We discussed about variables and their empirical studies’ results, identified their roles and impacts of them to the growth rate A summary of expected sign and empirical Tran Luu Quoc Vuong – MDEk16 24 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal reference for all variables showed in table 3.1 above In the end of chapter, we have a brief introduction about what we next and some test we will use Tran Luu Quoc Vuong – MDEk16 25 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal CHAPTER 4: DATA ANALYSIS 4.1 Growth rate and convergence In the neoclassical growth model of Ramsey (1928), Solow (1956), Swan (1956), Koopmans (1965) and Cass (1965), there was a hypothesis called absolute convergence This hypothesis said that “poorer economies typically grow faster per capita and tend there by to catch up to the richer economies” It mean that the growth rate of real per capita GDP from 1985 to 2010 would have an inverse relationship to the level of real per capita GDP in 1985 The figure 4.1 below expresses the relation between the growth rate of per capita GDP from 1985 to 2010 and the log of per capita GDP in 1985 With the correlation -0.327, we can see that, for the 25 countries of Asia Pacific region, the absolute convergence is available However, with the extent of neoclassical growth model and the endogenous growth model, growth theorists initiated another concept: conditional convergence The relation between the growth rate and the starting position has to be examined under the situation of holding constant a set of variables that discriminate the countries Tran Luu Quoc Vuong – MDEk16 26 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal Figure 4.1 Growth rate versus Level of Per Capita GDP (simple relation) Grơth rate of per capita GDP 1985-2010 0.12 0.1 0.08 0.06 0.04 0.02 2.5 3.5 4.5 log(per capita GDP, 1985) 4.2 Descriptive analysis In this research, variables used to set up a framework can be split into two groups: group of initial levels of state variables (the stock of physical variables and the stock of human capital) and group of policy variables and national characteristics State variables used in this paper are the initial levels of per capita GDP in the form of log, ratio of primary and secondary schooling enrollment and the reciprocal of life expectancy at age one The second category comprises the ratio of domestic investment to GDP, the ratio of government consumption to GDP, the extent of international openness, the fertility rate and inflation There are total 50 observations in the dataset, in which 25 Asia Pacific countries with two 10-year intervals Within the period 1990-2010, on average, the growth rate of this region is about 5.2% The country has fastest growth rate is China with average growth rate 12.5% during 2000s while the lowest one is Mongolia with average growth rate 0.94% during 1990s The mean over sample years of GDP per capita is around 10,600 US Dollars The territory of Tran Luu Quoc Vuong – MDEk16 27 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal Macao reached max nearly 64,000 US Dollars in 2010 while Nepal only got 521 US Dollars in 1990 The life expectancy of this region during period 19902010 is rather high, about 69.8 The countries has lowest value is Papua New Guinea with value of this measure about 59, and increased 6.7 only durng 20 years The countries has highest figure is Japan with the figure of 80.96, and increased 4.1 during this period This variable appears in instrument lists and in the form of reciprocal of life expectancy The measures of education, primary and secondary school enrollment, have average values at 99.89% and 61.91% respectively But, in comparison with primary school enrollment, the gap between values and max value of secondary school enrollment variables is very large, 10.68 versus 161.74 In addition, standard deviation of this variable is very high, three times the value The most important thing is the number of observations of this variable, only 35 while this figure of other variables is 50 Besides, when calculating the correlation between this variable to others, we found that these figures are so high and have negative impact to regression This problem will be discuss further later Similarly, the gap between countries in the extent of international openness is very high, 18.171 for lowest (Japan in 1990s) and 404.588 for highest (Singapore in 2000s) while value of standard deviation are so high, about 7.5 time of value This gap will make some limitations when we examine the model, the result may be not precise and get bias We will discuss problems of these variables later About investment and consumption, the average values are 26.04 and 13.48 respectively The country have highest ratio of investment is Bhutan (49.05% of GDP in 2000s in average), and the country with lowest figure is Brunei (14.33% of GDP in 2000s in average) Bangladesh is the country has the least ratio of real government consumption to real GDP, only 4.5% during Tran Luu Quoc Vuong – MDEk16 28 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal 1990s and 5.2% during 2000s; while Brunei is the country consumes most (25.64% GDP during 1990s) About macroeconomic stability, Japan is the country has negative inflation rate (-2.59% during 2000s) although this figure was 1.2% for previous decade Mongolia has faced with 2-digit inflation (in average and highest at 268% in 1993) during 1990s, but economy has become better with inflation rate at 8.86% in next decade The detail descriptive statistics of sample shows in the table 4.1 following Table 4.1 Summary Statistic on the sample observations Variable Obs Mean Std Dev Min Max growth 50 0.052 0.025 0.009 0.125 log_init_gdp 50 3.549 0.568 2.709 4.634 pri_school 42 99.896 15.049 55.298 132.509 sec_school 35 61.911 32.433 10.684 161.739 expectancy 50 0.015 0.002 0.013 0.020 log_fertility 50 0.503 0.206 0.094 0.807 gov_share 49 13.479 5.094 4.534 25.640 openness 50 101.93 82.842 investment 49 26.039 7.028 14.331 49.049 inflation 50 0.067 0.078 -0.003 0.490 18.171 404.588 Table 4.2 provides the simple correlation among variables The correlation between variables: the initial level of per capita GDP, secondary school enrollment, expectancy and fertility rate, are so high, from 71.3 to 83.1 percent It can make us make a question about the relation between these variables If the multicollinearity exists between them, we will have to drop some of them Tran Luu Quoc Vuong – MDEk16 29 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal In relation between the dependent variable, the growth rate, and a set of explanatory variables, the correlation is good, all the signs are as expected But, the signs in relation between the growth rate and secondary school enrollment rate is negative As above discussion, the number of observations of this variable are too small, so it may affect the precision of the model Table 4.2 Correlation on the sample observations growth growth log_in~p pri_sc~l sec_sc~l expect~y log_fe~y gov_sh~e openness invest~t log_init_gdp -0.332 pri_school 0.098 0.081 sec_school -0.307 0.716 0.257 expectancy 0.119 -0.831 -0.303 -0.713 log_fertil~y -0.120 -0.757 -0.253 -0.566 0.812 gov_share -0.362 0.509 -0.112 0.439 -0.294 -0.122 openness 0.144 0.209 -0.015 -0.016 -0.186 -0.088 -0.008 investment 0.562 -0.183 -0.015 -0.149 0.077 -0.088 -0.064 0.109 -0.190 -0.369 -0.020 -0.058 0.406 0.394 -0.055 -0.021 0.065 inflation inflat~n 4.3 Regression Results for Growth rates 4.3.1 Basic Regression Our data consists of 25 Asia Pacific countries for two period: 1990-2000 and 2000-2010 with a set of 10 variables, including dependent variable But in this model, from result of descriptive analysis above, I decided to drop the secondary school enrollment variable Moreover, we will not use the log of initial level of per capita GDP, government consumption ratio, investment rate and the extent of international openness to regression Instead of them, we will use Tran Luu Quoc Vuong – MDEk16 30 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal their instrumented variables in our model Table 4.3 contains result of regression for the growth rate of real per capita GDP This model also allow the errors terms to be correlated across the time periods and to be assumed to be independent across countries The error variances are allowed different for each period but not allowed to vary across countries We use dummy variable to distinguish two time periods As above short analysis, there are convergence in this paper and we can found that from the table 4.3: the estimated coefficient on log_init_gdp is significant (P-value at 0.001) and at a value of -0.05 It means that the growth will be higher in response to lower initial level of per capita GDP if the other explanatory variables are held constant According to this coefficient, convergence occurs at the rate of percent per year that is the growth rate will increase 0.05 if a one-standard-deviation decline in the log of per capita GDP Table 4.3 Basic Cross-Country Growth Regression Variable Coefficient p_log_init_gdp -0.0504206 0.0134595 -3.75 0.001 -0.0778714 -0.0229697 pri_school -0.0000755 0.0002156 -0.35 0.729 -0.0005151 0.0003642 3.528819 0.1 0.917 -6.82775 7.566396 log_fertility -0.1114896 0.0331247 -3.37 0.002 -0.1790479 -0.0439313 p_gov_share 0.0003242 0.0007024 0.46 0.648 -0.0011084 0.0017568 p_openness 0.0000188 0.0000571 0.33 0.745 -0.0000977 0.0001353 0.000171 0.0005128 0.33 0.741 -0.0008749 0.0012168 inflation -0.041935 0.0356694 -1.18 0.249 -0.1146832 0.0308132 cyear 0.0104734 0.0062874 1.67 0.106 -0.0023498 0.0232966 2.75 0.01 0.0691852 0.4686769 expectancy p_investment _cons 0.3693233 0.268931 Number of Obs Prob>F Std Err 0.097938 t P>t [95% Conf Interval] 41 0.0024 R-squared 52.68% Adj R-squared 38.94% (We can observe details in Appendix C.1) Tran Luu Quoc Vuong – MDEk16 31 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal But, we can see that, almost variables are not significant and have very high value of P-value, except the initial level of GDP, log of fertility Besides, the sign of some variables are not like expected: primary school enrollment and ratio of government consumption We need to some hypothesis test to improve this model 4.3.2 Hypothesis tests Tải FULL (69 trang): https://bit.ly/3Yd2vBh Dự phòng: fb.com/TaiHo123doc.net The first problem we have to face is heteroskedasticity Heteroscedasticity does not cause estimated coefficient by OLS to be biased, but it can cause the variance and standard errors of the coefficients to be biased So, we still can use regression result from heteroscedastic data to estimate the relationship between the dependent variable and the explanatory variables, but results of hypothesis tests are possibly wrong We use White test to test heteroskedasticity for this model The following is the result: Table 4.4 Heteroskedasticity – White test White's test for Ho: homoskedasticity against Ha: unrestricted heteroskedasticity chi2(40) = Prob > chi2 = 41.00 0.4265 Cameron & Trivedi's decomposition of IM-test Source Heteroskedasticity Chi2 df p 41 40 0.4265 Skewness 10.62 0.3028 Kurtosis 0.35 0.5547 51.97 50 0.3972 Total Tran Luu Quoc Vuong – MDEk16 32 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal We can see that, the chi-square value is very high It indicates that heteroskedasticity was present in this model And the table 4.5 following is the result after dealing heteroskedasticity Tải FULL (69 trang): https://bit.ly/3Yd2vBh Dự phòng: fb.com/TaiHo123doc.net Table 4.5 Basic Cross-Country Growth Regression (robust) Variable p_log_init_gdp Coef Robust Std Err t P>t [95% Conf.Interval] -0.052423 0.0137314 -3.82 0.001 -0.08043 -0.02442 pri_school -0.0000761 0.0001879 -0.41 0.688 -0.00046 0.000307 expectancy 0.3806465 3.509894 0.11 0.914 -6.77783 7.539123 log_fertility -0.1121092 0.0371288 -3.02 0.005 -0.18783 p_gov_share 0.0004222 0.0007607 0.56 0.583 -0.00113 0.001974 p_openness 0.0000179 0.0000468 0.38 0.704 -7.8E-05 0.000113 p_investment 0.0002587 0.001414 0.18 0.856 -0.00263 0.003143 inflation -0.052122 0.0552661 -0.94 0.353 -0.16484 0.060594 cyear 0.0105059 0.0053372 1.97 0.058 -0.00038 0.021391 _cons 0.2784087 0.0996637 2.79 0.009 0.075143 0.481674 -0.03638 (We can observe details in Appendix C.2) Heteroscedasticity does not cause estimated coefficient by OLS to be biased, so the coefficient of variables not changes It means that almost the variables are not significant and the signs of some variables are different from expected The model is not good enough The variables may be not significant or there are relation between them that make the model not have statistical meanings Tran Luu Quoc Vuong – MDEk16 33 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal The next hypothesis test we have to is the multicollinearity test Not like heteroskedasticity, the impact of an explanatory variable on the dependent variable while holding other variables constant will be less precise One of the specialty of multicollinearity is that the standard errors of the affected coefficients tend to be large As a consequence, the test of the hypothesis that the coefficient is equal to zero may lead to a failure in rejecting a false null hypothesis of the explanatory variable which has no effect As above mentioned (in Chapter 4, part 2, Descriptive Analysis), there are some variables in this research have this characteristic, such as secondary school enrollment and international openness The result of multicollinearity test is described in the table 4.6 below We can see that, the VIF value of all variables of the model are smaller than 10, so, there is not the present of multicollinearity in this model Table 4.6 Multicollinearity testing Variable VIF 1/VIF p_log_init~p 7.22 0.138458 expectancy 5.84 0.171111 log_fertil~y 5.75 0.173822 p_gov_share 2.37 0.421843 inflation 1.9 0.526855 p_investment 1.85 0.540728 p_openness 1.45 pri_school 1.33 0.754336 cyear 1.26 0.791249 Mean VIF 3.22 Tran Luu Quoc Vuong – MDEk16 0.69091 34 6679420 ... 15 Determinants of economic growth in a panel of Asia Pacific countries Thesis Proposal In this research, we use education attainment and health to measure human capital, one of two state variables... variables and the stock of human capital) and group of policy variables and national characteristics State variables used in this paper are the initial levels of per capita GDP in the form of. .. initial level of real per capita GDP is low and depend on initial state variables like human capitals in the forms of education and health, other institution and national characteristic variables