Regional welfare disparities and regional economic growth in Vietnam

This study had multiple objectives. First, it aimed at examining regional inequality in several welfare variables, such as health and education during the 1998-2004 period, and per capita income over time from 1990 to 2006. The first objective of the study was to gain a better understanding about the magnitude and evolution of regional inequality in health, education and per capita income. In addition, the study paid attention to the levels and trends of between- and within-region differences in these welfare variables separately. The second objective of the study was to explore the relationship between regional income inequality and regional growth during the 1990-2006 period in order to provide knowledge about empirical evidence of the relation between regional income inequality and regional economic growth in Vietnam. The third and final objective of the study was to examine the regional growth patterns in the country for the growth episode stretching from 1990 to 2006. Also in the third objective of the study, the role of space and spatial autocorrelations were considered in order to provide a better understanding about the spatial interaction between regions in Vietnam and its impact on regional economic growth. To attain these objectives, the study employed various methodological methods in order to measure and analyze regional inequality in welfare variables, to explore the empirical relationship between inequality and growth, and to test for growth patterns in the Vietnam economy. The study was subdivided into eight chapters. Besides the introduction provided in chapter one, an overview, conclusions, discussion and recommendations given in chapter eight, the other six chapters, from chapter two to chapter seven, covered all issues relating to characteristics of reform and welfare in Vietnam, the literature, methodological approach and empirical analysis. Economic reforms and changes in welfare in Vietnam were elaborated and discussed in chapter two. The main focus of this chapter was to discuss the macroeconomic situation, as well as the implementation of the economic reforms called doi moi, and its impact on people’s welfare. As discussed in the chapter, before the doi moi, the welfare system in Vietnam functioned quite well. Most Vietnamese people had equal access to most welfare services provided by the government, irrespective of religion, ethnicity or social status. Similar to other communist countries, welfare distribution in Vietnam before the doi moi was relatively equal, although the quality of the welfare system with respect to services offered was still questionable in terms of catching up to the international standard. At the time, most Vietnamese people enjoyed free access to the healthcare system. Pupils and students got free education at all levels. As a result, the free systems culminated in good health records attained throughout the country, high enrolment rates and a high rate of literacy among the Vietnamese people compared with other developing countries with the same or slightly higher level of development. The situation and condition of welfare, such as education and health after the doi moi, were also discussed and compared with the condition of welfare before doi moi to evaluate the impact of high and relatively high rates of economic growth of 6-9% per annum on a changing welfare system, by explanatorily analyzing the welfare system, using Gini coefficients. This showed that regional inequality in the country had generally changed. The methodological issue regarding measuring and analyzing regional welfare inequality was dealt with in chapter three. Apart from discussing the theoretical approach and empirical measurement models, the chapter also provided a short review of the literature on multidimensional regional inequality. From the various approaches to welfare inequality measurement and analysis, the present study chose the multidimensional approach, using Theil’s second index as the inequality measure. The rationale for using Theil’s second index (measure) and the valuable properties of the Theil’s second index were also discussed in this chapter. Theil’s second measure satisfies the property of additive decomposability and is very informative about the status of the distribution of well-being. Additive decomposability property meant that overall inequality could be decomposed into within- and between-region inequality, which would be very meaningful for understanding magnitudes and trends of regional inequality. Theil’s second measure also satisfied several desirable properties as measuring regional income/welfare inequality, such as mean independence, population-size independence, and the Pigou-Dalton principle of transfers. In order to compute Theil’s second measure of inequality, there is a need for estimating a composite index, based on the aggregate function by Maasoumi, in which the indicators’ coefficients and their associated weights are to be estimated. To do so, the chapter also developed the measurement models based on the principal component models (PCA) and partial common principal component models (PCPC), developed and used by Flury. These two types of models were discussed in detail in order to help the readers to comprehend the subsequent empirical parts of the thesis. The measurement and analysis of regional inequality for two welfare variables, namely health and education, were given in chapter four and chapter five, respectively. In chapter four, we proceeded with the measurement and analysis for health facilities and health status, and also examined spillover effects, for example, that health facilities in one region would affect people’s health status in the first-order adjacent province(s). We carried out several tests for spatial dependence among these two subcomponents of health. The results in chapter four revealed that total regional inequality had different patterns for healthcare facilities and health status during the 1998-2004 period. Generally, regional inequality in both the healthcare facilities and health status moderately decreased in Vietnam over the 1998-2004 period. However, there was a sign of increase during this stretch of time, as the results found in chapter four showed a slight increase in inequality in both components of health over the 1998-2002 period, and then it decreased in the next period. Empirical findings in chapter four also proved the importance of spillover effects prevailing in the case of the health component. Healthcare services seemed to be more equally distributed in the case of contiguity than in the case of non-contiguity, as Theil’s second measure in the former was lower than that in the latter. This finding indicated that if we ignored the existence of spillover effects, the regional inequality would be exaggerated. The tests for spatial dependence among health components were also conducted in chapter four for several specifications regarding the relationship between health status and health facilities. The results of the tests confirmed the existence of spatial effects between the two dimensions of health. The test also proved that the spatial error model was an appropriate specification for analysis of the relationship between the composite index with respect to health status and the composite index with respect to health facilities, while the spatial lagged model was a more appropriate specification to analyze the relationship between the composite index with respect to health status and the indicators of health facilities. Following chapter four, the measurement and analysis of multidimensional regional inequality for the education component were carried out in chapter five. However, different from chapter four, contiguity effects were not considered in this chapter, because we argued that due to mandatory education applied in Vietnam, pupils were not allowed to attend school in other regions (provinces) than they were registered to live (even though in practice pupils would seek education in other regions other than the one they lived, but this phenomenon was not common, so we did not have statistics for analysis). Thus, as a result, we did not take spatial dependence into consideration. For regional inequality in education facilities, the research showed that total inequality, its between-region term and within-region components all go into opposite directions between the two periods. However, the change in the within-region inequality was in contrast with the changes in total regional inequality. The between-region component as the total regional inequality and the between-region component in education facilities first decreased during the period 1998-2002 and then increased during the period 2002-2004, while the within-region inequality increased during the 1998-2002 period, then decreased during the 2002-2004 period. As research findings revealed, the distribution of education facilities among the regions in Vietnam was a bit more unequal over time. Meanwhile, the change over time of the within-region component of inequality was very moderate. For regional inequality in enrolments, our findings showed that the pattern of total regional inequality and its decomposed terms were the same as all of them increased in the first period (1998-2002), and declined in the next period (2002-2004). The structural shares of between-region and within-region inequality in the total inequality told us that in all three years under study the between-region term made up a larger part in the total inequality. These implied that the differences in enrolments between regions were more significant to the total inequality than the differences in enrolments within regions. The comparative composite index with respect to education facilities and the composite index with respect to enrolments presented in chapter five revealed that there was a sharp contrast between the two dimensions of education, as the most-favored regions with respect to enrolments were those in the lowland areas with least and medium-favored education facilities. The least-favored regions with respect to the education enrolments were those in the upland areas, which were very high-favored with respect to education facilities in terms of the composite index. Chapter six presented the measurement and analysis of the regional income inequality, and explored the evolution of income disparities, its decomposed terms as well as its relationship with economic growth among regions in Vietnam in the period from 1990 to 2006. The findings confirmed that generally the regional income inequality in the country moderately increased over time. The decomposition of the total inequality revealed that within-region inequality decreased, while between-region increased over time. However, the size of within-region income inequality differed among regions in the country, as within-region Theil’s second inequality measure showed different trends. The empirical exploration of the relationship between regional inequality and regional growth found revealed different patterns of the relationship. The estimates from the panel data regression have found a positive relationship between the rate of economic growth and the initial level of per capita income in all the cases studied. Meanwhile, using the pooled OLS technique, we found negative relationships between growth and inequality for all the cases studied: without both controls and regional dummies, with the controls and no regional dummies, and with both controls and regional dummies. The test for the validity of the method (either the fixed effects or random effects is more appropriate and preferred) revealed that the fixed effects method was appropriate and preferred for the cases: without both the controls and dummies and with controls but no dummies in the growth model, while the random effects technique was more appropriate and more preferred for the case with both the controls, regional and time dummies. Interestingly, estimates by both the fixed effects and random effects methods revealed that there was both a positive and negative relationship between regional growth rate and initial level of per capita income. The empirical results of the relationship between regional inequality and regional growth show that the estimated relationship between inequality and growth changes when using different estimation techniques. However, the estimation using panel data regression, the fixed effects and most cases of random effects methods seem to support the theoretical argument and empirical findings that there is a positive correlation between growth and inequality for developing economies like Vietnam’s, while pooled OLS regression yields the results of a negative relationship between the two variables: inequality and growth. Examination of the growth patterns or tests (absolute β-convergence test and ADF panel IPS t-bar test based on unit root test) for convergence in regional income was conducted in chapter seven. All the tests for convergence were rejected, which meant that regional income in Vietnam did not converge towards one another. Based on an informal exploratory data analysis we may conclude that divergence of regional income took place in Vietnam, given the size of beta and its high statistical significance. Parallel with the test for the per capita GDP convergence, the chapter did also attempt to test for the spatial dependence among provinces, in order to explain the growth pattern of one province in relation with that of others. However, the Moran’s I and both the LM statistics did not support our prior assumption of the spatial dependence among provinces in the growth progress. Lastly, chapter eight provided an overview, the conclusions, discussion and recommendations based on the entire research conducted in this study. The chapter drew several key conclusions. First, regarding the regional inequality in health and education, the key point was that the results from the multidimensional approach to measurement and analysis were satisfactory, because generally, welfare inequality in Vietnam, according to the World Bank reports, increased over time, as inequality in income per capita increased. However, the multidimensional approach found that inequality in education and health decreased over the period studied. Additionally, the chapter concluded that spillover effects were important in the case of measuring and analyzing regional inequality for health, and they should be taken into consideration. For education, analyzing the comparative composite indices with respect to education facilities and education enrolments revealed that in some regions these two indices did not coincide, which meant that regions, which were favored in terms of the composite index with respect to education facilities, were not the ones that were favored in terms of the composite index with respect to enrolments and vice versa. This conclusion would be very useful for policymakers in considering the effective way to invest in improving general education in the country. Chapter eight also described several limitations of the study, regarding the approaches used in the study, the number of indicators used, the sources and duration of the data. However, the limitations were not critical, but if we overcame these setbacks, the results of the study would be more impressive, empirically speaking. In addition, several recommendations for policy were given in chapter eight, regarding policy on health and education, and policy on regional development and sustainable development.