ANALYSIS OF INCLUSIVE GROWTH IN POVERTY, UNEMPLOYMENT AND INCOME INEQUALITY IN WEST SUMATERA PROVINCE: PANEL ERROR CORRECTION MODEL APPROACH

This study investigates the determinants of inclusive growth in poverty, unemployment and income inequality in West Sumatera Province. This study uses time series from 2011-2017, while the cross section is 12 districts and 7 cities, which were analyzed by the panel error correction model. There are several important findings in this study. First, the results of the analysis for the long term are that inclusive growth in poverty is positively and significantly affected by health, education, investment and government spending. Meanwhile, inclusive growth in unemployment and income inequality is negatively and significantly affected by health, education, investment and government spending. Second, the results of the analysis for the short term are investment and government spending disrupts the balance of inclusive growth in poverty. Meanwhile, education and investment disrupt the balance of inclusive growth in unemployment and income inequality. Third, there is a long term balance on inclusive growth in poverty, unemployment and income inequality. Therefore, it is highly recommended for districts/cities governments in West Sumatera Province to intervene in the independent variables that have significant and disturbing effects in the short term, so that conditions of inclusive growth in poverty, unemployment and income inequality can be achieved optimally.


INTRODUCTION
The aim of economic development is not only to achieve high growth because it will increase poverty, unemployment and widen income inequality, which is caused by the benefits of economic growth is only enjoyed by a part of the community, so that it will cause various social problems (Anyanwu, Kayizzi-mugerwa, & John, 2013). Current economic development should be development that prioritizes various aspects, namely quality economic growth Source: Central Bureau of Statistics of West Sumatera Province.

Picture 1 Inclusive Growth Index in Poverty, Unemployment and Income Inequality in West
Sumatera Province.
Based on the information in Figure 1, that in aggregate the conditions of poverty, unemployment and income inequality in West Sumatera Province have fluctuated over the past 7 years, which phenomenon creates a question whether the ongoing development is achieving inclusive growth, so it is important to do an analysis using an index of inclusiveness in poverty, unemployment and income inequality in West Sumatera Province. Studies on inclusive growth in poverty, unemployment and income are still very few because this topic is a new study in the field of economics, so this study adopts relevant study that only analyzes the determinants of poverty, unemployment and income inequality without analyzing the conditions of inclusiveness. Based on relevant study that to analyze the conditions of poverty, unemployment and income inequality can use several determinants, which include, health, education, investment and government spending (Higgins, 2010;Pandiella, 2016;Škare & Družeta, 2016).
Based on the phenomena that have been explained, it is necessary to analyze the inclusive growth in poverty, unemployment and income inequality in the long term and short term by using various determinants whether the conditions of poverty, unemployment and income inequality that have taken place in West Sumatera Province so far have achieved growth inclusive.
Poverty, unemployment and income inequality are problems faced by all countries. Economic growth as one indicator in overcoming social problems if economic growth is a concept of economic development (Ribeiro & Marinho, 2012). Economic growth, government spending and education in the long term and short term have a positive and significant relationship to poverty. Conversely, inflation and population have a negative relationship in the short run but in the long run are positive (Munir & Ullah, 2018). Macroeconomic stability and structural change are the basis for achieving inclusive growth. Other indicators included in the analysis also have several important implications, the role of the external sector can also be positive with trade provisions that encourage greater inclusiveness (Khan, Khan, Safdar, Munir, & Andleeb, 2016). Macroeconomic conditions and social financial deepening will increase inclusiveness, and reduce unemployment poverty and income inequality, while reforms in the trade sector are needed to increase efficiency in terms of inclusiveness (Yusuf, 2017). Poverty reduction measured by the national poverty line is compared with education and health. Multidimensional poverty is used to assess the inclusiveness of growth which draws Poverty Unemployment Income Inequality attention to the success of the government in providing public goods and providing sufficient employment opportunities (Danaan, 2018). Poverty is a major obstacle to socio-economic development. Poverty has survived and several interventions have failed to produce a significant increase in the Human Development Index. The policy that must be carried out is to design programs that favor the poor effectively and progressively. Poverty reduction can be realized by empowering society to develop their productivity (Sherriffdeen & Olorunfemi, 2016). The health sector is very important to encourage the growth of the poor population. In addition, there needs to be government involvement in funding the health sector by providing adequate health care facilities (Djamaluddin, 2014). The identification of demographic, economic, social and government policies is significant for poverty. High poverty is caused by a large number of household members who work in the agricultural sector or work as laborers and have a low level of education and live in houses with dirt/bamboo floors. The main determinants of poverty are employment, marital status, education level, religion, financial discipline, use of products and services to banks, distance, household size, access to political contracts, gender, income level, and age (Sanya & Olumide, 2017). Population is a blessing and not a threat, so that growth in favor of the poor can be achieved. In addition, government involvement is needed in funding the health sector by providing adequate health care facilities (Chibba, 2009). Improvements to education, health and investment will reduce unemployment (Riaz & Jafar, 2018). Increasing government spending in the productive sector will reduce unemployment and income inequality (Tangtipongkul & Wangmo, 2017). The conditions of health and education will reduce income inequality (Vo et. al., 2019). Increasing investment and spending in productive sectors will create an even distribution of income (Herzer & Sebastian, 2012). Based on relevant studies, it can be concluded in general that the determinants of poverty, unemployment and income inequality consist of health, education, investment and government spending. The difference between this study and previous study is the focus of the analysis, which this studies analyzing inclusive growth in poverty, unemployment and income inequality in West Sumatera Province in the long term and short term.

RESEACH METHODS Study Types and Variables
This study uses secondary data from the publication of West Sumatera Province Central Statistics Agency and the Indonesian Directorate General of Fiscal Balance. The object of study is all districts / cities in West Sumatera Province, which consists of 12 districts and 7 cities. The observation year used was from 2011-2017. Determinants of inclusive growth in poverty, unemployment and income inequality used in this study are health, education, investment and government spending.

Definition of Operational Variables Inclusive Growth in Poverty
An index of inclusive growth in poverty was formed using per capita spending and the number of poor people. Per capita expenditure is used as a proxy for calculating income, which uses the assumption of expenditure equal to income.

Inclusive Growth in Unemployment
An index of inclusive growth in unemployment was formed using number of labor force and number of people employed. Where : = coefficient of inclusive growth in unemployment = elasticity of employment for the number of workers = elasticity of employment for economic growth = economic growth

Inclusive Growth in Income Inequality
An index of inclusive growth in income inequality was formed using the Gini index and share 60 percent of the lowest population expenditure. Where: = coefficient of inclusive growth in income inequality = elasticity of income inequality to share 60 percent of the lowest population expenditure against total population expenditure = elasticity of income inequality on economic growth = economic growth

Health
The health is measured using a percentage of the population that is sufficiently nutritional, ie the number of people who meet the minimum nutritional adequacy standard compared to the total population.

Education
The education used is the logarithm of the average length of schooling in districts/cities in West Sumatera Province, which is in the form of the length of time a society attends education in a certain period.

Investment
The investment used is the logarithm of the formation of gross fixed capital on the expenditure of production units to add fixed assets minus the reduction of fixed assets of former districts/cities in West Sumatera Province in a certain period.

Government Spending
The government spending used is the logarithm of the government spending for district/city infrastructure in West Sumatera Province in a certain period. The scope of government spending on infrastructure consists of road, irrigation, networks, buildings, and buildings.

Inclusiveness of Economic Growth in Poverty, Unemployment, Income Inequality
The measurement of the inclusiveness of economic growth in poverty is concluded by comparing the value of IGp; IGem; IGin with Gg, among which are.
IGp; IGem; IGin = Gg Growth is neutral, so everyone receives the same benefits from growth.

IGp; IGem; IGin > Gg
Growth is pro-poor, so this condition reduces poverty, unemployment and income inequality 0 < IGp; IGem; IGin < Gg Growth is not pro-poor, but this condition is still happening poverty reduction, unemployment and income inequality still occur IGp; IGem; IGin < 0 < Gg Growth does not favor the poor, this condition results in increased poverty, unemployment and income inequality

Model Specifications
This study aims to analyze of inclusive growth in poverty, unemployment and income inequality in the long term and short term, so the approach used is panel error correction model, which approach has an independent variable as a determinant of the dependent variable in the long term and short term. The relationship between these variables is summarized in Figure 2 below.

Inclusiveness of Economic Growth in Poverty
Declining poverty is one measure of economic growth as inclusive growth. Inclusive growth occurs when the coefficient of inclusive growth in poverty (IGp) has a value greater than economic growth (Gg). The coefficient of inclusive growth in poverty (IGp) is summarized in Table 1 below.  LLDIKTI Wilayah X 27 The conclusions in Table 1 are summarized in Table 2 below. Based on the information in Table 1 which is summarized in Table 2, the conditions of inclusive growth in poverty in the Province of West Sumatera in the aggregate are good enough. During the study period, only 2014 did not experience inclusive growth in poverty. If seen from the coefficients in Table 1, which IGp in West Sumatera Province since 2011-2017 is always positive. This shows that poverty that occurred in West Sumatera Province decreased along with the occurrence of economic growth. However, only in 2014 the value of IGp in West Sumatera Province was lower than Gg, which means that despite the decline in poverty, the results of growth were not evenly distributed and the benefits of economic growth were received more by society who were not classified as poor.

Inclusiveness of Economic Growth in Unemployment
Declining unemployment is one measure of economic growth as inclusive growth. Inclusive growth occurs when the coefficient of inclusive growth in unemployment (IGem) has a value greater than economic growth (Gg). The coefficient of inclusive growth in unemployment (IGp) is summarized in Table 3 below.   Table 3 are summarized in Table 4 below. Based on the information in Table 3 which is summarized in Table 4, the conditions of inclusive growth in unemployment in West Sumatera Province that in aggregate still need to be improved. During the study period, only in 2016 and 2017 experienced an inclusive growth in unemployment, this condition means that the benefits of economic growth were received more by the middle to upper society thereby increasing unemployment.

Inclusiveness of Economic Growth in Income Inequality
Declining income inequality is one measure of economic growth as inclusive growth. Inclusive growth occurs when the coefficient of inclusive growth in income inequality (IGin) has a value greater than economic growth (Gg). The coefficient of inclusive growth in income inequality (IGin) is summarized in Table 5 below. The conclusions in Table 5 are summarized in Table 6 below.

Analysis Results for Panel Error Correction Model Unit Root Test
The unit root test in this study was carried out using the ADF test, which is summarized in Table 7 below. Based on the information in Table 7, that only inclusive growth in poverty and income inequality which is stationary at the level. Whereas inclusive growth in unemployment, health, education, investment and government spending are at the first difference, so that on average all variables in this study are stationary at first difference.

Selection of Model for Panel Regression
The model selection for panel is done using the Chow test and the Hausman test. The probability result of the Chow test for inclusive growth in poverty is 0.0149; inclusive growth in unemployment is 0.0235; Inclusive growth in income inequality is 0.0462, where the overall probability in the Chow test is less than 0.05, so the fixed effect model is better than the common effect model.Furthermore, the probability result of the Hausman test for inclusive growth in poverty is 0.4817; inclusive growth in unemployment is 0.2027; Inclusive growth in income inequality is 0.3562, where the overall probability in the Hausman test is greater than 0.05, so the random effect model is better than the fixed effect model.

Classic Assumption Test
The assumption test in the random effect model was not carried out because the random effect model was Generalized Least Square (GLS) which involved the variance covariance structure in the estimation process, so that the conditions had been generalized (Gujarati & Porter, 2009).

Equation for the Long Term
The random effect model is used to analyze inclusive growth that reduces poverty, unemployment and income inequality in West Sumatera Province in the long term. The estimated results of the long term equation are summarized in Table 8 below.  F-statistic) 0.0479 0.0427 0.0447 ***, **, * statistic relationships are significant at 1%, 5%, 10%. Source: own calculations.
Based on the information in Table 8, the long term equation conditions for inclusive growth in poverty (IGp), inclusive growth in unemployment (IGem) and inclusive growth in income inequality (IGin) can be interpreted as follows.
The average condition of inclusive growth in poverty in West Sumatera Province has increased by 0.0971 percent every year. R-Squared is 0.2653, which means that all determinants used in the model are able to explain the proportion of variance from inclusive growth to poverty by 26.53 percent, while the rest is explained by other variables outside the model. Prob (F-statistic) is 0.0479 which is smaller than 0.05 and F-statistic is 2.9378 which is greater than F-table is 2.88, so that simultaneously the determinant used has a significant effect on inclusive growth in poverty in the long term. The conditions in the long term if partially that the effects of health and education are positive and significant at the 1 percent significance level, which is consistent with study (Ageme, 2018;Tello, 2015). Meanwhile, investment and government spending have a positive and significanteffect at the 5 percent significance level, which is consistent with study (Abimbola, Olokoyo, Babalola, & Farouk, 2018;Bakari et al., 2019).
The average condition of inclusive growth in unemployment in West Sumatera Province has decreased by -0.0171 percent every year. R-Squared is 0.2893, which means that all determinants used in the model are able to explain the proportion of variance from inclusive growth to unemployment by 28.93 percent, while the rest is explained by other variables outside the model. Prob (F-statistic) is 0.0427 which is smaller than 0.05 and F-statistic is 2.8940 which is greater than F-table is 2.88, so that simultaneously the determinant used has a significant effect on inclusive growth in unemployment in the long term. The conditions in the long term if partially that the effects of education and investment are negative and significant at the 1 percent significance level, which is consistent with study (Oluseye & Gabriel, 2017). Meanwhile, helath has a negative and significanteffect at the 5 percent significance level, which is consistent with study (Maqbool et. al., 2013) and government spending has a negative and significanteffect at the 10 percent significance level, which is consistent with study (Alrabba, 2017).
The average condition of inclusive growth in income inequality in West Sumatera Province has decreased by -0.0037 percent every year. R-Squared is 0.1919, which means that all determinants used in the model are able to explain the proportion of variance from inclusive growth to income inequality by 19.19 percent, while the rest is explained by other variables outside the model. Prob (F-statistic) is 0.0447 which is smaller than 0.05 and F-statistic is 2.9261 which is greater than F-table is 2.88, so that simultaneously the determinant used has a significant effect on inclusive growth in income inequality in the long term. The conditions in the long term if partially that the effects of health, education and government spending are negative and significant at the 5 percent significance level, which is consistent with study(Charles-Coll, 2011). Meanwhile, investment has a negative and significanteffect at the 10 percent significance level, which is consistent with study (Niyimbanira, 2017).

Cointegration
The balance for long termof inclusive growth in poverty, unemployment, income inequality can be determined through the results of the cointegration test, which are summarized in the following Table 9 below. Based on the information in Table 9, it can be concluded that inclusive growth in poverty, unemployment and income inequality has a cointegration relationship with its determinants in the long term because the residual value of the basic equation is stationary at the level. In addition, cointegration is a requirement for using the error correction model.

Equations for the Short Term
The random effect model is used to analyze inclusive growth that reduces poverty, unemployment, income inequality in West Sumatera Province in the short term. The estimated results of the short term equation are summarized in Table 10 below. 0.0000 0.0000 0.0000 ***, **, * statistic relationships are significant at 1%, 5%, 10%. Source: own calculations.
Based on the information in Table 10, the short term equation conditions for inclusive growth in poverty (IGp), inclusive growth in unemployment (IGem) and inclusive growth in income inequality (IGin) can be interpreted as follows.
The average condition of inclusive growth in poverty in West Sumatera Province has increased by 0.0384 percent every year. R-Squared is 0.7293, which means that all determinants used in the model are able to explain the proportion of variance from inclusive growth to poverty by 72.93 percent, while the rest is explained by other variables outside the model. Prob (F-statistic) is 0.0000 which is smaller than 0.05 and F-statistic is 48.0566 which is greater than F-table is 2.88, so that simultaneously the determinant used has a significant effect on inclusive growth in poverty in the short term. There is a long term balance towards inclusive growth in poverty because the probability of Res (-1) is smaller than the significance at 5 percent. The conditions that partially disrupt the balance in the short term is investment and government spending, which is consistent with study (Anwar, Uppun, Tri, & Reviani, 2016;Thomas & Chittedi, 2015). Meanwhile, health has a positive and significant effect at the 10 percent significance level, which is consistent with study (Akerele & Adewuyi, 2011;Lee & Sissons, 2016). In addition education has a positive and significant effect at the 5 percentsignificance level, which is consistent with study (Park & Mercado, 2018;Singh, 2017).
The average condition of inclusive growth in unemployment in West Sumatera Province has decreased by -0.0644 percent every year. R-Squared is 0.4467, which means that all determinants used in the model are able to explain the proportion of variance from inclusive growth to unemployment by 44.67 percent, while the rest is explained by other variables outside the model. Prob (F-statistic) is 0.0000 which is smaller than 0.05 and F-statistic is 17.4353 which is greater than F-table is 2.88, so that simultaneously the determinant used has a significant effect on inclusive growth in unemployment in the short term. There is a long term balance towards inclusive growth in unemployment because the probability of Res (-1) is smaller than the significance at 5 percent. The conditions that partially disrupt the balance in the short term is education and investment, which is consistent with study (Aqil et. al., 2014).Meanwhile, government spending and health has a negative and significant effect at the 10 percent significance level, which is consistent with study (Chowdhury & Hossain, 2014).
The average condition of inclusive growth in income inequality in West Sumatera Province has decreased by -0.0135 percent every year. R-Squared is 0.2182, which means that all determinants used in the model are able to explain the proportion of variance from inclusive growth to income inequality by 21.82 percent, while the rest is explained by other variables outside the model. Prob (F-statistic) is 0.0000 which is smaller than 0.05 and F-statistic is 6.0270 which is greater than F-table is 2.88, so that simultaneously the determinant used has a significant effect on inclusive growth in income inequality in the short term. There is a long term balance towards inclusive growth in income inequality because the probability of Res (-1) is smaller than the significance at 5 percent. The conditions that partially disrupt the balance in the short term is education and investment, which is consistent with study (Afandi, 2017).Meanwhile, government spending and health has a negative and significant effect at the5 percent significance level, which is consistent with study (Var Arnum & Naples, 2013).

CONCLUSION
Based on the estimation results, it can be concluded that the analysis for the long term is that inclusive growth in poverty is positively and significantly affected by health, education, investment and government spending.Meanwhile, inclusive growth in unemployment and income inequality is negatively and significantly affected by health, education, investment and government spending.The analysis for the short term is that investment and government spending disrupt the balance of inclusive growth in poverty. Meanwhile, education and investment disrupt the balance of inclusive growth in unemployment and income inequality. There is a long term balance towards inclusive growth in poverty, unemployment and income inequality even though in the short term there is no balance.