The Analysis of Absolute Convergency of Human Development Inter Provinces in Indonesia

Economics Development Department, Economics Faculty, Universitas Negeri Semarang Article Info ________________ Article History : Received July 2016 Accepted September 2016 Published November 2016 ________________


INTRODUCTION
The Human Development Index (HDI) is a composite indicator developed by UNDP in order to assess development success.HDI includes three basic dimensions of welfare: health, education and decent living standards.However, HDI does not describe human development as a whole, yet it is a better measurement than income.One of the most serious disadvantages of human development index (HDI) is that it does not consider the distribution of human development within a country (Harttgen, 2012).
Indonesia is one of the developing countries with HDI of 57.2 in 2005 and increased to 68.4 in 2013 (UNDP, 2014).Indonesia's HDI is the accumulation of total HDI value of every province in Indonesia.The diversity of potential natural and human resources between regions makes the achievement of HDI in each province varies (Central Bureau of Statistics, 2013).During the period of 2005-2013 the achievement of HDI value in each province in Indonesia experienced an increasing trend as shown in figure 1.The increase of HDI does not rule out the possibility of inter-provincial HDI gap in Indonesia.It can be seen on the highest achievement of HDI in 2005 that is DKI Jakarta with value equal to 76.07 and the lowest HDI that is Papua with HDI value equal to 62.08.Meanwhile, in 2013 the two provinces were still in the same position but with higher HDI value.In 2013, the value of Jakarta's HDI was 78.59 and the HDI value of Papua Province was 66.25.
In 2005, the difference between the highest and the lowest HDI was 13.99 points, while in 2013 the difference was at 12.34 points (Central Bureau of Statistics, 2013).It can be concluded that the disparity of human development at the provincial level is relatively decline.Therefore, the decline in that disparity indicates that the value of inter-provincial HDI in Indonesia moves into one point.
Above all, low HDI achievement is not a barrier for human development improvement.Provinces with low HDI achievement have an opportunity to grow faster than provinces with high one.For example, regions in Papua tend to grow faster than those in DKI Jakarta which has high HDI (Central Bureau of Statistics, 2013).The process of self-chasing which is done by provinces that have low HDI is named as convergence.
According to Barro and Sala I (2004: 45), convergence is a phenomenon that leads to a single point.The convergence process is related to the development process of a region.In the context of economic growth, convergence occurs when poor areas with low incomes will grow faster than developed regions with high incomes, so by the time those regions will have the same growth (convergent) (Mankiw 2004: 216).That explanation is based on the economy of a region will lead to steady state and if the economy has reached that condition then it will slowly run.Furthermore, there are two convergence concepts, namely sigma convergence and beta convergence.Sigma convergence occurs when the dispersion as measured by the standard of cross-regional HDM logarithmic deviation tends to decrease.While beta convergence occurs when regions with low HDI tend to grow faster than areas with high HDI so low HDI areas tend to pursue high HDI areas.
In addition, beta convergence can be divided into two, namely absolute convergence and conditional convergence.Absolute beta convergence occurs when regions with low HDI grow faster than high HDI areas without any other influence from economy.While the conditional convergence measurement is done by adding control variables.
From the description above, this study aims to identify the sigma convergence and absolute beta of human development among provinces in Indonesia and identify the absolute beta convergence rate.

RESEARCH METHOD
This study used secondary data that is data of HDI.The data is obtained from the publication of the Central Bureau of Statistics.The type of data used is panel data namely combination of time series and cross section.The time series data are from 2005 to 2013 while cross section data are 33 provinces in Indonesia.
Sigma convergence was measured by HDI dispersion logarithmic.While absolute beta convergence analysis was performed by panel data regression analysis.Besides, model selection was done to determine the best model between common, fixed and random effect in order to be used in panel data estimation.Model selection was done by using chow test and hausman test.
The absolute convergence model used in this study is a model developed by Barro and Sala I (1992), as follows: i is the state or region, α is the intercept, (y it+T /y i,to ) is the growth rate of GDP per capita, y it+T and y i,to is the per capita income at the end and the beginning of the period, while (1-e -βT ) is the initial revenue coefficient that decreases with the length of the time interval.To obtain the HDI convergence regression model, Y it + T variable is replaced by HDI it variable, while Y ito is replaced by using HDI t-1 variable.Further, the measurement of HDI convergence is done every year so T is equal to one then the absolute convergence model of HDI can be written as: HDI it is the provincial HDI i in year t, HDI it-1 is the initial HDI and Log (HDI it / HDI it-1 ) represents the annual growth rate of HDI.While e it is an error term.Consequently, if the initial regression coefficient on HDI (β1) is negative, it means that if provinces with low HDI grow faster than provinces with the high one, absolute beta convergence (absolute β convergence) occurs.
According to Barro and Sala I (2004: 56), knowing the speed of convergence is important because if convergence occurs faster so economy will be getting closer to be steady state.The speed of convergence is calculated by finding β values first with the following formula: β is the convergence beta or average convergence rate, β1 is the initial HDI variable regression coefficient, and T is the number of time periods of analysis.If the value β1 <0 then the parameter β will be positive and the higher β value indicates faster convergence (Wau, 2015: 110).After knowing absolute and conditional convergence, the speed of convergence is calculated by the following formula:

Sigma convergence
The concept of sigma convergence explains that human development between provinces in Indonesia is convergence if it decreases over time.In this study, the calculation of sigma convergence uses standard deviation of HDI logarithmic.The result of calculation of sigma convergence can be seen in figure 2.
Source: Processed data Figure 2. Sigma Convergence Figure 2 shows that at the beginning of the analysis year, the level of HDI gap between provinces in Indonesia is quite high.HDI dispersion trends were declining from 2005-2013.The value of HDI dispersion in 2005 was 0.048683 and decreased to 0.039037 in 2013.These results show that there is HDI sigma convergence among provinces in Indonesia.
This sigma convergence notes that the intergovernmental HDI gap narrows from 2005 to 2013, yet the decline value of HDI dispersion is quite small.Thus, the result of this sigma convergence indicates that reducing the gap rate could not quickly be done (Malik, 2014.In fact, the decline takes a very long time to get equity.
Moreover, changes or increases in HDI could not quickly happen.Human development is a process and could not be measured in a short time because the results of investment or development in education and health could not be felt in the short time.Yet, it takes a long time to take the benefit of investment or development in education and health.
Beta convergence explains that regions with low HDI are able to grow faster than the high ones.The calculation of beta convergence which was conducted in this study is absolute beta convergence.This analysis was done by using panel data regression by referencing to the model proposed by Barro and Sala I (1992) as in equation 1.To get the absolute beta convergence model of HDI, the model is modified, so it becomes equation model 2. The method used for the estimation in this study is Generalized Least Square (GLS) method.It is to solve the problem of heterocedasticities among cross section units.
Furthermore, in estimating the absolute convergence is only use one independent variable that is the initial HDI log without entering other independent variables.The negative sign of the resulting regression coefficient will indicate the direction of convergence or divergence.The result of absolute convergence estimation can be seen in table 1.
Based on Table 1, it can be concluded that the initial HDI regression coefficient is -0.070076.The regression coefficient has a negative sign which means that there is a negative relationship between the growth of HDI and initial HDI.This indicates that there is convergence in interprovincial HDI in Indonesia in 2005 -2013.In  As for provinces with low HDI will catch up so that the growth of HDI in all provinces is convergent or same.These results are supported by researches conducted by Konya and Guisan (2008), Hendrani (2012), Jorda and Sarabia (2015) and Yang et al (2016).Those researches explain that there is absolute beta convergence of HDI.The value of determination coefficient of 0.481349 shows that 48.13 percent of HDI growth diversity in provincial units can be explained by the model, while the rest is explained by other variables outside the model.Besides, the constant value will be 0.306346 if the independent variable is remain constant, so the growth rate of the provincial HDI will be 0.306346.
The result of the fix effect model is assumed that there is a difference of intercept between cross section but the intercept is same time to time.It shows that every province analyzed in this study has uniqueness or heterogeneity that is the growth of basic HDI which is different between a province to another.The individual effects of each province are obtained from constant values plus the constants of each province.This difference can be seen from the intercept of each province as in table 2. The speed of convergence indicates how rapidly HDI resulted can establish HDI conditions.As the higher the speed of HDI convergence, so the established HDI conditions will be quickly reached.Furthermore, the absolute beta convergence rate of HDI is calculated based on β1 coefficient value as the result of the absolute convergence estimation.
Based on Table 1, it can be noted that the absolute convergence speed of the average HDI in 2005-2013 is 0.807 percent per year.These results indicate that the inter-provincial HDI gap in Indonesia will decline by 0.807 percent per year.Thus, the speed of absolute convergence is relatively slow so the time required to achieve a steady-state HDI becomes quite long.

CONCLUSION
Regarding to results above, it can be concluded that there is convergence of sigma HDI and convergence of absolute beta of HDI between provinces in Indonesia in 2005-2013.The HDI sigma convergence shows that the interprovincial HDI gap is decline while the absolute beta convergence of HDI shows that provinces with low HDI are growing faster than provinces high HDI.Forasmuch as provinces with low HDI will catch up so the growth of HDI in all provinces will be convergent or equal.Meanwhile, the speed of absolute convergence of inter-provincial HDI in Indonesia is 0.807 percent per year.That speed is relatively slow.As the result, the time required to achieve established HDI conditions will be quite long.

Figure 1 .
Figure 1.Development of Provincial HDI in Indonesia in 2005-2013 Source: Central Bureau of Statistics.

Table 1 .
Absolute Convergence Result by Fixed Effect of GLS Method Approach

Table 2 .
Absolute Convergence Individual EffectBased on Table2, it can be seen that the province that has the largest constant is Riau Islands 0.310208 which is the sum of the average constant of 0.306346 and the intercept of Riau Islands 0.003862.While the province that has the smallest constant is Papua that is equal to 0.300509.If other variables are considered zero, the growth of HDI Riau Islands is 0.310208 while Papua is equal to 0.300509.