Indonesian Regional Economic Development: A Neoclassical Growth Analysis

Indonesian Regional Economic Development: A Neoclassical Growth Analysis Haryanto ) Abstract This study examines the growth experience of 285 districts in Indonesia from a neoclassical perspective. Using real districts GDP per capita from 1983-1998 based on constant price 1983 excluding oil and gas, the neoclassical proposition of convergence was evaluated. The σ convergence analysis shows that the dispersion of personal income during the observed period decreased from 0.266 in 1983 to 0.242 in 1993, but then rose to 0.297 in the period of crisis in 1998. This result implies that income distribution tended to converge during the periods of high national economic growth. The empirical result for absolute convergence supports the hypothesis that poor districts grow faster than the rich ones, thus economies in the poorer districts have tended to catch up with the richer districts. Another finding suggests that the speed of convergence among 285 districts can be accelerated by enhancing capital accumulation in poor districts, providing its infrastructure, building up the quality of their workforce by investing more in education, ensuring transfer of technology to the local industry, and controlling the population growth rate. I. Introduction 1.1 Background Study The experience of Indonesian economic development since the past two decades has attracted a great deal of attention of the development economist. Indonesian development s history until the mid-1990s was one of remarkable economic success. From 1980 through 1996, Indonesia s GDP (Gross Domestic Product) rose about 6.5% per year, provincial GDPs increased, and so did economic equality of per capita provincial GDPs. During the same period, economic growth was accompanied by a spectacular reduction in the percentage of population living under the poverty line. From an estimated level of 42.3 million people (28.56% of total population) in 1980, people living under the poverty line decreased to 27.2 million people (15.08% of total population) in 1990, and it gradually declined to the level of 22.6 million people (11.2% of total population) in 1996 (Central Bureau of Statistic, 1997). However, the Indonesia s vibrant economic performance, which was painstakingly achieved during 1980-1996, has been wiped out by the Asian crisis since 1997. It is believed that the crisis has brought about a setback of Indonesian macroeconomic improvements, and has negative social impacts on the most vulnerable groups of population. Indonesia s GDP per capita reduced from US $ 1,118.8 in 1996 to US $ 375.2 Haryanto, SE, MA is staff of Education and Training Center for Development Planning, Bappenas; MA in International Development from International University of Japan in 2001-red. C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 1

and US $ 540 in 1997 and 1998 respectively. The growth of Indonesian GDP decreased from 7.82% in 1996 to 4.70% in 1997, and it faltered in 1998 to be 13.20% (Central Bureau of Statistic, 1999). In addition, the impacts of the crisis on poor people are obvious, but those living slightly above the designed poverty lines are severely hurt as well. People living under the poverty line increased tremendously from 22.6 million people (11.2% of total population) in 1996 to 49.5 million people (24.23% of total population) in 1998 (Central Bureau of Statistic, 1999). These economic changes incite an important issue about the Indonesian economic development: whether the changes in the social fabric and the macroeconomic variables lead to divergence or convergence among Indonesian regions in terms of economic equality. The Indonesian crisis seems to have significantly affected the distribution of income among the regions. From this point of view, then rises another important issue: Has the Indonesian regional distribution of income tended to diverge during the period of high national economic growth or during the period of crisis? Considering those two issues, the aims of this study are: (1) To understand whether the regions of Indonesia, during recent past years, have exhibited convergence or divergence in per capita GDP; (2) To investigate the trends of Indonesian regional distribution of income during the periods of high national economic growth and during the period of crisis/poor economic performance; (3) To identify some policy implications toward Indonesian regional development based on the findings of the data analysis. 1.2 Methodology Numerous studies have been conducted in many countries to understand whether income disparity among countries or among regions within countries exhibits a tendency to diverge or converge over time by using the neoclassical growth models (see for example, Barro (1991), Barro (1994), Barro and Sala i Martin (1995), Sala i Martin (1996), Dewhurst (1998), Garcia and Sulistianingsih (1998), and Heng and Siang (1999)). This study applies the neoclassical growth model to measure the Indonesian regional disparity course. The neoclassical growth model suggests that countries or regions with similar production technologies, resource endowment, time preference of consumption and comparable of saving and population growth rates should converge to similar steady state levels of per capita income. This convergence (β convergence) property implies that poor regions starting with a relatively low standard of living and a lower capital/labor ratio will grow faster during the transition as they catch up with the rich regions, but ultimately both groups will arrive at the same level of per capita income. Moreover, A standard deviation method will also be employed to investigate the dispersion of regional income per capita, σ convergence. β convergence has two aspects: absolute and conditional convergences. Absolute convergence refers to a situation in which poor countries or regions tend to grow faster in terms of per capita income than rich ones without conditioning on any other characteristics of the economies and therefore the poor tend to catch up or converge to the rich. If, after taking into account any other characteristics of the economies, there is a negative association between the rate of growth in per capita GDP and the initial per capita GDP, then there exists conditional convergence. C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 2

To deal with the problem, this paper uses this method to analyze the Indonesian regional development. In order to determine whether inequality per-capita income has tended toward convergence, this paper will first examine the absolute convergence by applying the following formula (Barro and Sala-i-Martin, 1995): 1 y ln Τ y it i0 1- e = a - T βt ln [ yi0 ] + U i0, T Where: ln is natural logarithm; y it is per-capita GDP in region i at period T; y i0 is the initial per-capita GDP in region i; a is the intercept of the equation; and U i0,t is average of the errors terms between periods 0 and T. If there is a negative association between the initial per capita GDP and the growth rate per capita GDP, the β convergence exists. This paper will also examine σ convergence by computing the dispersion of provincial per-capita GDP of Indonesia regions. Barro and Sala-i-Martin (1995, p. 383) say that the dispersion can be measured by calculating the standard deviation of per-capita logarithm for each year. The following formula will be used to estimate the standard deviation for each year: S D 1 n 2 t = (ln y t ln yit ) n i= 1 where, SD t stands for standard deviation at period t, ln ỹ t and ln y it represent the logarithm of the average per capita GDP of Indonesian regions at period t and the logarithm of per capita GDP in region i at period t respectively, whereas n is the number of regions. If SD t-1 is less than SD t the σ convergence exists. However, if SD t-1 is more than SD t the σ convergence doesn t exist. Finally, to investigate whether the regional distribution of income tended toward divergent during the period of high national economic growth or during the period of crisis, we can plot the σ convergence for each period. The growth rate across regions will be convergent if the period of the crisis, which is represented by the lower of the aggregate GDP growth, exhibits the decline in the value of σ convergence, and vice versa. This paper is going to employ Gross Domestic Regional/District Level Product (GDRP) data excluding oil and gas from 1983 to 1998, and would be calculated based on constant 1983 price. Resources of the data are mainly taken from Indonesian statistical agency, namely Biro Pusat Statistik (BPS), and the National Development Planning Agency (Bappenas). II. Results of Analysis 2.1 σ Convergence Figure 2.1 depicts the standard deviation of the log of per capita income for the 285 district regions from 1983 to 1998. As presented in the figure, the dispersion of personal income decreased from 0.266 in 1983 to 0.242 in 1993, but then rose to 0.297 in 1998. It is estimated that the decline of the dispersion of personal income from 1983 to 1993 was closely associated with economic growth and structural changes in the industrial C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 3

sectors. Between 1983 and 1993 the economic growth was high in Indonesia at 7.1% per annum. Figure 2.1 Dispersion of per Capita Income, 1983-1998 0.300 0.290 0.280 0.270 0.260 0.250 0.240 0.230 0.220 0.210 0.200 1983 1985 1986 1990 1993 1996 1998 STD of Log per capita GDP Using GDP provincial data including oil and gas in Indonesia from 1975 to 1993, Garcia and Sulistianingsih (1998) found similar results: the dispersion of per capita GDP across provinces declined from 0.39 in 1975 to 0.28 in 1993. Barro and Sala-i-Martin (1995) studied the dispersion of per capita income across regions for many countries (U.S., Japan, Italy, Spain, Germany, and the United Kingdom). They found that the dispersion across the states in the U.S. declined from 0.5 in 1880 to 0.2 in the late 1980s whereas for the Japanese prefectures the dispersion reduced sharply from 0.5 in the 1930s to 0.15 in the 1980s. Engelbrecht and Kelsen (1999) also analyzed σ convergence across APEC region, and they found that the dispersion of per capita income reduced from 0.95 in 1965 to 0.91 in 1990. 2.2. β Convergence We generate the following model to estimate the unconditional β convergence: LY 0_t = α 0 + α 1 LY 0 Where: LY 0_t y it y it y i0 LY 0 α 0 α 1 β = per capita growth rate, or 1 y it = log T y i 0 = per capita GDP in region i at period t = per capita GDP in region i at period t = the initial per capita GDP in region i = log y i0 = the intercept of equation = the estimated coefficient of LY 0, or βt 1- e = T = the speed of convergence C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 4

The other three regressions are presented to test whether the initial level of per capita income and other explanatory variables have a negative effect on the rate of growth of per capita income (conditional convergence). To estimate the conditional convergence, this research uses two dummy variables, to account for districts with oil and non-oil resources, and between regencies and municipalities, and growth of population as another explanatory variable. The estimations cover the periods of 1983-1998, 1986-1998, and 1990-1998. Per capita income in 1983, 1986, and 1990 represents the initial income level for the regressions for 1983-1998, 1986-1998, and 1990-1998 respectively. This research estimates β convergence across 285 districts in Indonesia. The following model is used to estimate the conditional convergence: LY 0_t = α 0 + α 1 LY 0 + α 2 MUNIP + α 3 OIL + α 4 P_GROWTH i Where: LY 0_t = per capita growth rate, or 1 yit = log T yi0 y it = per capita GDP in region i at period t y i0 = the initial per capita GDP in region i LY 0 = log y i0 MUNIP = dummy variable (1 for municipality and 0 for regency) OIL = dummy variable (1 for district with oil resource and 0 for district without oil resource) P_GROWTH i = the average population growth rate in region i from 1983-1998 α 0 = the intercept of equation = the estimated coefficient of LY 0, or α 1 α 2,3,4 β βt 1- e = T = the estimated coefficient of MUNIP, OIL and P_GROWTH = the speed of convergence Before estimating the regression models based on equations presented in the Figure 4.2 Per Capita Growth Rate Versus 1983 Per Capita Income Across 285 Districts in Indonesia Per Capita Growth R 1983-1998 30.0 25.0 20.0 15.0 10.0 5.0 0.0-5.0 4.5 5.0 5.5 6.0 6.5 7.0-10.0 Log of Per Capita Income 1983 C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 5

previous chapter, we provide a chart as depicted in figure 2.2 to estimate the relationship between the initial per capita income 1983 and the subsequent growth of per capita income from 1983 to 1998. As shown in Figure 2.2 there is a downward slope that links the initial income and the rate of growth indicating that the absolute convergence occurs across 285 district in Indonesia. The negative association between those two variables suggests that poor districts grow faster than the rich ones; thus districts with lower income per capita tend to catch up the districts with the higher per capita income. The line was also statistically significant which concludes that there is absolute convergence. 2.2.1 Unconditional β Convergence This research estimates β convergence for the periods of 1983-1998, 1986-1998, and 1990-1998. Per capita income in 1983, 1986, and 1990 constitute the initial income level for the regressions for 1983-1998, 1986-1998, and 1990-1998 respectively. Regression estimates are presented in Table 2.1 below: As shown in tables 2.1, the estimated parameters of all regression equations, - 0.0203 for period 1983-1998, -0.0208 for period 1986-1998, and 0.0201 for period 1990-1998, support the hypothesis of absolute convergence according to which poor districts Table 2.1. Unconditional Converg ence Variable Periods 1983-1998 1986-1998 1990-1998 Constant 0,3025 0,3159 0,3119 (6.921) (5.395) -3,7139 ln Y 83-0,0203 - - (-5.842) - - ln Y 86 - -0,0208 - - (-4.5666) - ln Y 90 - - -0,0201 - - (-3.1220) R 2 0,1076 0,0686 0,0333 Estim ation M ethod OLS OLS OLS Im plied β 4,3 3,4 2,22 Notes: 1. Dependent variables are 1/T*ln (yit/yi0) 2. Figures beneath the estimated coefficients are the associated t-statistics tend to grow faster than the rich regions; thus the poorer tend to catch up with the richer districts. From the findings, however, it has been noted that the relation between the initial per capita income and the rate of growth of per capita income was slightly low, which is shown by the coefficient correlations, 0.105, 0.069 and 0.033 for periods 1983-1998, 1986-1998 and 1990-1998 respectively. However, all the estimated slopes are statistically significant, which are shown by the values of t-statistic, which are higher than the critical value 5% significance level. The estimated β coefficients, which measure the speed of convergence of per capita income among 285 districts in Indonesia, are 4.3% for period 1983-1998, 3.4% for period 1986-1998 and 2.22% for period 1990-1998. These finding implies that if these conditions continued as during the periods of observations, i.e. 1993-1998, 1986-1998, and 1990- C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 6

1998, it would take 16, 20 and 32 years respectively to reduce the dispersion of per capita income among district regions by half. 2.2.2 Conditional β Convergence According to Barro and Sala-i-Martin (1995) the concept of conditional β convergence assumes that all economies are different in their parameters; thus, they are also different in the steady states. The main idea is that if a rich economy has a higher per capita income, then in turn, it also has a higher saving rate than a poor economy, the rich economy may proportionally further from its steady state position. In this case, β convergence will exist in the sense that a lower initial per capita income tends to generate a higher per capita growth rate, once we control the explanatory variables of the steady state. Moreover, as mentioned in the neo-classical growth model, income per capita depends on the capital stock, population and the human capital. The capital stock and the human capital are expected to have a positive effect on the growth rate of per capita income; while population is negatively affecting the rate of per capita income. Due to a lack of data, this research uses two dummy variables, for municipality and regency as proxy for human resource and health services, and for oil/non-oil as proxy for capital stock. Assuming that municipality region is economically better than regency, which in turn is better in quality of human resource and health services, therefore, the estimated coefficient for this dummy variable is expected to be positive. Moreover, a region with oil resources is assumed to be higher in capital stock than the region without oil resource; thus this dummy variable is also estimated to have a positive effect on the growth rate of per capita income. Results of the estimated regression are presented in the table 2.2. As shown in Tables 2.2 the estimated coefficient for log initial income per capita and other variables, except for OIL, have the expected sign and are statistically significant Table 2.2. Conditional Convergence Variable Periods 1983-1998 1986-1998 1990-1998 Constant 0.2944 0.375 0.3647 (6.163) (5.903) (3.891) ln Y 83-0.0188 - - (-4.891) - - ln Y 86 - -0.025 - - (-4.969) - ln Y 90 - - -0.0234 - - (-3.203) M U N IP 0.0267 0.0347 0.0385 (4.827) (4.898) (3.782) OIL -0.0141-0.0101-0.0259 (-1.974) (-1.132) (2.037) P_GROW TH -0.0062-0.006-0.008 (-6.8622) (-6.292) (5.036) R 2 0.2944 0.2131 0.0333 Estimation M ethod O LS O LS O LS Im plied β 4 4.4 2.6 Notes: 1. D ependent variables are 1/T*ln (yit/yi0) 2. Figures beneath the estimated coefficients are the associated t-statistics C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 7

at less than 5% significance level (except for OIL in 1986-1998). It is estimated that the negative sign of OIL coefficient is attributed to the exploitation from the central government to regional oil resources. In the case of Indonesia, the central government absolutely controls all regional oil productions and equally distributes the oil income to all Indonesian districts. The average growth rate of real income per capita for the 27 oil producing districts during the period of observation, 1984-1998, was only 3.4% per year; the other 258 districts had around 9% in average. This economic gap leads to social adversity from oil regions, and in turn, to potential rise in social tension and disintegration. Thus, it is suggested that the priority of policy implementation should be aimed at promoting economies of oil districts, such as expediting the implementation of regional autonomy. Results also indicate that the speeds of convergence are 4%, 4.4% and 2.6% per year respectively for the period 1983-1998, 1986-1998, and 1990-1998. The speeds of conditional convergence in 1986-1998 and 1990-1998 are a little faster than the speed of unconditional convergence for the same periods. This condition implies that a lower initial per capita income tends to generate a higher per capita growth rate, once we control for the explanatory variables of the steady state. Moreover, by controlling the determinant variables, such as promoting the quality of human resources as well as reduction in the population growth rate lead to acceleration of the speed of convergence, which in turn, reduces the economy gaps among Indonesian districts. To estimate the conditional convergence this research employs regional dummy variables as proxy of human resource. Generally speaking, the quality of human resources in municipality regions is better than that in regency regions. Hence, higher quality of human resources tends to foster the growth rate of per capita income. As presented in the tables the estimated signs of regional dummy variables in all periods of observation are positive indicating that increases in the initial per capita income that are accompanied by the typical increase in quality of human resources are systematically related to subsequent growth rate of per capita income. Many researchers find similar evidences; see for example, Barro and Sala-i-Martin (1995), Garcia & Sulistianingsih (1998), Engelbrecht and Kelsen (1999); and Heng and Siang (1999). In the neoclassical models exogenously higher population growth rate tends to reduce the growth rate of per capita output for given values of the state variables. Results shown in table 4.2 depict the negative association between the population growth rate and the rate of growth of per capita income. The estimated coefficients for the population growth rate are 0.006, 0.006, and 0.008 for the periods of 1983-1999, 1986-1998, and 1990-1998 respectively, which are quite significant at less than 5% significance level. It implies that if this condition continues as during the period of observation, for example 1993-1998, increasing 1% of population rate will reduce the growth rate of income per capita by 0.6%. The empirical results do support the hypothesis of conditional convergence among 285 districts in Indonesia. The most important finding is that the gap in per capita income among districts can fall more rapidly if the districts have better quality of human resources. Another important finding is that the higher population growth leads to reduction in the growth rate of per capita income. C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 8

III. Conclusion and Policy Implications This research has shed some light on the question of catching-up amongst 285 districts in Indonesia for the periods 1983-1998, 1986-1998, and 1990-1998. The estimates of σ convergence, absolute convergence, and conditional convergence give favorable support for the neoclassical growth model predictions. The σ convergence analysis shows that the dispersion of personal income during the observed period decreased from 0.266 in 1983 to 0.242 in 1993, but then rose to 0.297 in the period of crisis in 1998. This result supports the hypothesis that income distribution tended to converge during the periods of high national economic growth. After 1993, however, the dispersion amongst districts has been increasing to 0.280 in 1997 and 0.297 in 1998. The empirical result for absolute convergence also supports the hypothesis that poor districts grow faster than the rich ones, thus the economy in the poorer districts tend to get closer to the richer. The coefficient correlation of initial income per capita for all sub-periods are less than 15%, which are quite low, but statistically significant at less than 5% confidence level. In addition, the neo-classical growth model tells us the importance of human resources, population growth and capital stock as determinants of economic growth. Thus, using some proxies for these explanatory variables, the conditional convergence is analyzed to estimate the speed of convergence among 285 districts. Taking into account these explanatory variables in the equations, the evidence indicates that the speed of convergence has been increasing for the period of 1986-1998 and 1990-1998. Other findings show that the regional dummy variable, municipality/regency, which is the proxy for quality of human resources, has positive effect on growth whereas population growth rate has a negative effect on growth. However, another regional dummy variable, region with/without-oil, which is the proxy for capital stock, has a reverse sign. The reverse sign of this dummy variable is estimated due to the lack of regions/districts, which control and utilize oil production. The central government absolutely takes over all oil resources and productions; then they distribute the oil income relatively equally to all districts. Thus, the oil resources have no significant effect on the oil-producing district. Based on the findings, the policy implementations should be aimed to accelerate the speed of convergence, or in other words to reduce the gap of income distribution amongst districts. Central as well as local government policies can encourage or discourage the economic growth. With regard to regional endowments, it is suggested that the government distributes proportionally to each regions based on the regional resources. Otherwise, it leads to social adversity and national political instability, which in turn, slows down regional economy. In this regards, expediting the implementation of regional autonomy constitutes the effective way to stimulate the regional growth. Results from conditional convergence show that higher quality in human resources has a positive effect on growth; and higher population tends to reduce the economic growth. Most evidence indicates that rural districts have lower per capita income than urban areas. Moreover, the poorer districts have the lower literacy and skilled labor forced and higher population growth rate in terms of fertility than in urban districts. The implication, therefore, is that in order to speed up the convergence, more regional policies should be aimed at increasing the quality of human resources and reducing the fertility rates in rural districts rather than in urban areas. C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 9

Bibliography Akita, T. (1988). Regional development and income disparities in Indonesia. Asian Economic Journal, 2, 165-191. Akita, T. & Lukman, R. A. (1995). Interregional inequalities in Indonesia: A sectoral decomposition analysis for 1975-92. Bulletin of Indonesian Economic Studies, 9, 61-68. Akita, T., Lukman, R. A. & Yamada, Y. (1999, June). Inequality and the distribution of household expenditure in Indonesia: A theil decomposition analysis. The Developing Economies, 37, 197-221 Akita, T. & Lukman, R. A. (1999, August). Spatial patterns of expenditure inequalities in Indonesia: 1987, 1990, and 1993. Bulletin of Indonesian Economic Studies, 35, 67-90. Anand, S. (1983). Inequality and Poverty in Malaysia: Measurement and Decomposisition. Malaysia: Oxford University Press. Bank Indonesia (The Central Bank of Indonesia) (2000). Selected foreign exchange middle rates against Rupiah at Bank Indonesia. Bank Indonesia, 1 page. Online. Available: http://www.bi.go.id/ind/ datastatistik/t30515.rep (2000, February 5). Bappenas (National Development Planning Agency) (1999). Data Dasar Pembangunan Daerah Kabupaten/Kota (The Basic Figure for Districts and Municipalities Development). Jakarta: Bappenas, 1999. Barro, Robert J. (1991, May). Economic growth in a cross section of countries. The Quarterly Journal of Economics, 407-440. Barro, Robert J. (1994, October). Democracy and growth: working paper no. 4909. National Bureau of Economic Research, Massachusetts, USA. Barro, R. J. & Sala-i-Martin, X. (1995). Economic Growth. New York: McGraw Hill. Blanchard, O. (1997). Macroeconomics. New Jersey: Prentice-Hall, Inc. Daimon, T. & E. Thorbecke (1999, forthcoming). Mitigating the Social Impacts of the Indonesian Crisis: Lesson from the IDT Experience, May 1999. Dewhurst, J. H. L. (1998). Convergence and divergence in regional household incomes per head in the United Kingdom, 1984-93. Applied Economics, 30, 31-35. Engelbrecht, H. J. & Kelsen, B. (1999). Economic growth and convergence amongst the APEC economies, 1965-1990. Asian Economic Journal, 13, 1-17. Fingleton, B. (1999, April). Estimates of time to economic convergence: An analysis of regions of the European Union, International Regional Science Review, 22, 1-34. Garcia, J. G. & Soelistianingsih, L. (1998, April). Why do differences in provincial income persist in Indonesia? Bulletin of Indonesian Economic Studies, 34, 95-120. C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 10

Gylfason, T. (1999). Principle of Economic Growth. London: Oxford University Press. Heng, T. M., & Siang, T. C. (1999, August). A neoclassical analysis of the ASEAN and East Asian growth experience. ASEAN Economic Bulletin, 16, 149-165. Kangasharju, A. (1998). β convergence in Finland: regional differences in speed of convergence. Applied Economics, 30, 679-687. Lee, M., Longmire, R., Matyas, L., and Harris, M. (1998). Growth convergence: some panel data evidence. Applied Economics, 30, 907-912. Lopez-Bazo, E., Vaya, E., Mora, A. J., and Surinach, J. (1999). Regional economic dynamics and convergence in the European Union. The Annals of Regional Science, 33, 343-370. Mankiw, G., Romer, D., and Weil, D. (1992). A contribution to the empirics of economic growth. Quarterly Journal of Economics, 107, 407-437. Nombo, R. L. (2000). Regional autonomy program in Indonesia: What issues complicate its implementation? The Indonesian Quarterly, 28, 285-294. Premus, R. & Liu, L. G. (2000). Global Economic Growth. London: Greenwood Press. Pritchett, L. (1997, Summer). Divergence, big time. Journal of Economic Perspective, 11, 3-17. Ray, D. (1998). Development Economics. Princeton, New Jersey: Princeton University Press. Sala-i-Martin, X. (1996). The classical approach to convergence analysis. The Economics Journal, 106, 1019-1036. Statistics Indonesia. Gross domestic product statistics-selected table. Biro Pusat Statistics, 2 pages. Online. Available: http://www.bps.go.id/statbysector/natreg/ gdp/table3.shtml (2000, February 5). Solow, R. M. (1956). A contribution to the theory of economic growth. Quarterly Journal of Economics, 70, 65-94. Thirlwall, A. P. (1999). Growth and Development. London: Macmillan Education, Ltd. Wan, Henry Y. (1971). Economic Growth. New York: Harcourt Brace Jovanovich, Inc. World Bank (1995). World Development Report: Poverty. New York: Oxford University Press. C:\WINDOWS\Desktop\Majalah Perencaan Pembangunan\Edisi 24 Th 2001\Haryanto.doc # 11