Inclusive Growth and Its Determinants Recent Evidence from Indonesia with Provincial Data

This study evaluates Indonesia’s achievement in inclusive economic growth by analyzing regional data from 34 provinces for the period 2016-2018. For this purpose, the study used three indicators of achieving inclusive economic growth, namely economic growth that reduces inequality, poverty, and unemployment (or increases employment), the Poverty-Equivalent Growth Rate (PEGR) method, and the technique of multiple linear regression analysis (i.e. the fixed effect model). The results indicate that economic growth in Indonesia is not yet fully inclusive. Only a few provinces that have achieved inclusive growth. It was found that access to technology represented by the percentage of households owning a computer and access to energy represented by the percentage of households using LPG as the main fuel for cooking have positive effects on the acceleration of inclusive economic growth in Indonesia.


Introduction
According to Ali, Zhuang [1], Ali, Son [2], and Rauniyar and Kanbur [3], the term "inclusive economic development" has no widely accepted definition. The concept clearly encompasses inclusion and economic development, and views inclusion as a process as well as a goal. Such as Sen [4], Sachs [5], Ali, Son [2], Rauniyar and Kanbur [3], and McKinley [6] stress that inclusive economic development is economic growth coupled with equal economic opportunities. It focuses on creating economic opportunities and making them accessible to everyone in society at all levels, not just to the poor. In the same way, inclusive economic growth is one that emphasizes economic opportunities created by economic growth are freely available to all, particularly the poor [7,8]. Inclusive economic growth has a number of elements, which include poverty reduction, employment generation, improvement in quality of employment, agriculture development, industrial development, social sector development, reduction in regional disparities, environment protection, and equal distribution of income. Among these elements, poverty reduction, employment creation and equal distribution of income have received the most attention in empirical studies of, explicitly or implicitly, inclusive economic growth [9][10][11][12][13][14][15]. In Indonesia, many reforms have been carried out since the end of the 1997-98 Asian financial crisis. The government has embarked upon institutional transformation, making the country one of the region's most vibrant democracies. In the social, economic, and political fields, Indonesia has seen much progress. Wide reforms have been carried out in all areas of governance, including in the financial sector, and a new development strategy has been adopted for "inclusive" economic development [7,8].

Aim and research problems
This study is part of ongoing research project on "Inclusive Development in Indonesia". The aim of this study is to evaluate Indonesia's achievement in inclusive economic growth by analyzing regional data from 34 provinces. The definition of inclusive economic growth used in this study is an economic growth that reduces poverty, income distribution inequality and unemployment.
This research focuses on the following three questions: i.
is Indonesia successful in achieving inclusive economic growth? ii.
Are there differences in achieving inclusive economic growth between provinces? iii.
What factors that most determine the achievement of inclusive economic growth in 34 provinces?

PEGR
In general, regions with an agricultural basis tended to have pro-poor growth in reducing inequality but had anti-pro-poor growth in labor absorption. Meanwhile, areas with high tourism potential ever, it has pro-poor growth in employment.

Theoretical framework and hypotheses
Theoretically, there are two key channels through which economic growth affects employment and hence poverty and inequality, namely stronger output growth and increasing labor productivity in labor-intensive sectors [56,57]. In developing countries including Indonesia these sectors are agriculture, middle to low technology-based industries such as textile and garments, footwear, leather, furniture, tobacco, electronics, and food and beverages, and trade. However, from the literature review it revealed that an economic growth to be inclusive is influenced by many factors. In this study, as illustrated in Figure 1, 13 factors are included in the analysis of inclusive growth (IG), namely school participation rate (APS), number of community health centers (PKM), life expectancy (AHH), regional health insurance (JKD), credit outstanding of micro, small and medium enterprise (MSME), percentage of households that own a computer (KMP), percentage of households accessing the internet (INT), number of local traditional markets (PSR), length of national roads (PES), percentage of households that have access to proper sanitation (SNT), percentage of households that have access to clean water (AML), electricity (DLT), and percentage of households that use liquified petroleum gas as main fuel for cooking (LPG) ).

Research Method and Data Sources
The model and techniques of analysis used in this study differ according to the hypothesis being tested. For H1, the analysis model used was the Poverty-Equivalent Growth Rate (PEGR) formula adopted from several previous similar studies conducted by, among others, Klasen [58], Sholihah [47], Azwar [33], and Prabandari [49]. PEGR is often used to measure the benefits of economic growth for the poor. By adopting the PEGR concept, inclusive growth can be measured by the following formula: where: By describing i from equation (1) as poverty (p), inequality (in) and labor (em), and j refers to indicators of economic growth (g), then by adopting the equation, inclusive growth can be measured by the following formula:

a)
Inclusive growth index that reduces poverty ( ) IGp with the following formula:

Annals of Social Sciences & Management studies b)
Inclusive growth index that reduces inequality ( ) IGin with the following formula: Inclusive economic growth index in absorbing labor

( )
IGem with the following formula: An economic growth can be said to be inclusive if the value of For H2, the technique of multiple linear regression analysis was used. In this determinant analysis, the dependent variable is inclusive growth using index values or inclusive growth coefficients and the 13 determinants as independent variables. In a multiple linear regression classic assumption tests were performed which aimed to obtain the results of a regression estimate that meets the best linear unlimited estimator requirements, which are linear, unbiased, and minimum variables, namely, normality test, heteroscedasticity test, autocorrelation test and multicollinearity test. In accordance with the variables and research objectives, an empirical linear regression regression model can be formulated according to the inclusive growth indicator approach (the results of the PEGR analysis), as shown by equation 6, and definition of operational variables is given in Table 2.

Economic growth, poverty, inequality, unemployment
Since the end of the 1998 Asian financial crisis that forced the Indonesian economy to drop sharply with a growth rate of minus 13 per cent, Indonesia's economic growth rate has never exceeded 5.5 per cent compared to an average of 7 to 8 per cent that ever achieved before the crisis. However, looking at the development of the three main components of inclusive growth, it seems that Indonesia is on the right track towards inclusive growth. As can be seen in Figure 2, the level of inequality tends to decline; although it is not significant and still falls into the category of moderate inequality. Indeed, reducing inequality remains a serious problem in Indonesia which has not been easy to overcome. This may suggest that more government efforts are still needed to achieve inclusive growth. The number of poor people and the level of unemployment also shows a declining trend. Economic growth in a region can be said to be inclusive if its inclusive growth index ( ) IG is greater or equal to its economic growth ( ) Gg  . The results show that in 2017 and 2018 IG is below the economic growth rate although the inclusive economic index grew much faster (39%) than the increased economic growth rate (2%) during that period (Figure 3). This means that Indonesia's economic growth is not inclusive yet because only a few provinces have achieved inclusive growth during that period. Of the 68 observations (34 provinces in 2017 and 2018), only 8 observations that achieved inclusive growth. Meanwhile, as shown in Table 3, economic growth, poverty, open unemployment and inequality vary by province. To some degree, these variations reflect differences in many growth factors between provinces including the availability of resources, the average level of education of the workforce, economic structure, and infrastructure development. In 2018, the highest economic growth was achieved by Papua and West Papua ranked second. Their high growth rates were influenced greatly by the wealth of their natural resources, especially mining such as copper, gold, oil and gas.

Inclusive growth index
As already explained in the metodology, in this study three coefficients of the inclusive growth index were analyzed using provincial data. First, the coefficient of economic growth that reduces poverty. Economic growth is said to be inclusive if the IGp coefficient is greater or equal to the Gg coefficient. Second, the coefficient of economic growth that reduces inequality. Economic growth is inclusive if the IG in coefficient is greater or equal to the Gg coefficient. Third, the coefficient of economic growth that increases labor absorption. Economic growth is considered inclusive if the IG em coefficient is greater or equal to the g G coefficient. With respect to poverty. the results show that in 2017 only four provinces had achieved inclusive growth and increased to seven provinces in 2018 (Table 4). It is obvious that most provinces in Indonesia have economic growth that is not yet inclusive in reducing poverty. Their IGp coefficient is positive but smaller than their g G coefficient, meaning that poverty continued to decrease, but only a small portion of the poor did benefit from the growth. North Kalimantan, the newest province in Indonesia, has negative coefficients in both years. This means that economic growth was enjoyed by people who were not poor (anti poor). A negative coefficient also indicates that economic growth cannot explain its role in reducing poverty, and even tends to exacerbate poverty [58][59][60]. With respect inequality, in 2017 there were four provinces that had achieved inclusive growth, i.e. Kep province. Riau, West Nusa Tenggara, West Papua and Papua and increased in 2018 to six provinces, i.e. Riau, Jambi, Kep. Bangka Belitung, West Java, West Nusa Tenggara and East Kalimantan. Kep Riau, Papua and West Papua failed to maintain their inclusive growth in 2018. Whereas the province of West Nusa Tenggara was able to maintain its inclusive growth in reducing inequality for two years in a row. Next, Chow test and Hausman test were performed to determine the panel model to be used, and the results show that the probability value of Chi-Square is smaller than 0.05. This means that a more appropriate model to be used to estimate the effect of independent variables on IG is the fixed effect model. The estimation results are as follows:  (7) Finally, two tests were carried out, namely the individual parameter significance test (t-test) to see whether each of these independent variables individually affected IG significantly, and the simultaneous significance test (F-test) to see whether all of these independent variables together affected IG significantly ( Table 5). The results of the t-test show that there are only two variables whose probability value is smaller than 0.05, which means that each of them significantly affect the IG , namely the percentage of households that own computer and the percentage of households that use LPG as fuel for cooking. While the F-test results show a probability value smaller than 0.05, which means that together all these independent variables have a significant effect on IG .

Conclusion and Research Limitation
There are two important findings from this research. First, although at the national level poverty, inequality and unemployment continued to decline, the economic growth in Indonesia is not yet fully inclusive. The average inclusive growth index in Indonesia is still below the average economic growth. Meanwhile the achievement of inclusiveness at the regional level shows different results between provinces. However, looking at the average index value of inclusive growth, at least during the 2017-2018 period, IG Indonesia shows an upward trend.
In 2017 the provinces that had experienced inclusive growth were Kep Riau and West Nusa Tenggara, and in 2018 were Kep. Bangka Belitung, West Nusa Tenggara, East Kalimantan and Riau. So, it can be concluded that Indonesia is well in a good track. Second, the percentage of households that own computer that represents households' access to technology and the percentage of households that use LPG gas as the main fuel for cooking that represents their access to energy are two factors that have strong influences in accelerating the realization of inclusive growth in Indonesia. However, this research has some limitations which are the followings: i) this study only used thirteen variables, i.e school participation rate, number of Puskesmas, life expectancy, regional health insurance, road length, number of traditional markets, number of households that own computers, users of the Internet, access to proper drinking water, access to proper sanitation, electricity distribution, and the use of LPG fuel for cooking. It is most likely that there are still many other determinants of inclusive economic growth at the provincial level but not included in this study due to data problem, ii) the time frame used is only 2017-2018. Ideally, a dynamic model over a longer time period would be more appropriate. However, for other years, not all provinces have data for all variables, iii) the data used is provincial data. If the study population is regencies or cities throughout Indonesia (if data are available), the results will be different.