Fiscal Transfer Policies and Road Infrastructure Reduce Income Inequality in Rural-Urban Areas

Fiscal transfer policies and road infrastructure are essential in reducing inequality in Indonesia. However, previous research examining the effect of fiscal transfer policies and road infrastructure still came up with inconclusive findings, thus making it necessary to conduct further research on rural, urban, and sub-national areas in Indonesia. This study examines the impact of fiscal transfer policies and road infrastructure on reducing income inequality in rural, urban, and sub-national areas. The authors utilized time-series data from 2012 to 2021 and 34 provinces. The fixed effect GLS model showed that Kuznets' hypothesis existed at rural, urban, and sub-national levels. The results also showed that the special allocations fund significantly reduced income inequality in rural, urban, and sub-national areas. However, road infrastructure was significant only in urban areas. The findings suggest that the special allocation fund policy can be expanded in scope and increased in number to accelerate the reduction in income inequality.


INTRODUCTION
The extensive economic inequality, to some extent, is hard to avoid and always becomes an issue and a topic of debate among researchers and policymakers.The government must impose community regulations to minimize income inequality (Chancel et al., 2022).Considering the critical issue of the impact of income inequality on individuals, politics, social, and the economy (McGregor et al., 2019), it is deemed critical that policymakers take preventive measures to avoid a far more severe impact (Shukla & Mishra, 2019).Notably, there has been a huge income inequality gap between urban and rural areas.As stated by Imai & Malaeb (2018), income inequality occurs between urban and rural areas and in several ASEAN countries such as Thailand and Vietnam.
In Indonesia, the UNDP report predicts that the population in the top 10% of income oscillates around 40-50%.Income inequality has increased slightly since 1980, but in the late 1990s and early 2000s, there had been short-run fluctuations (Chancel et al., 2022).Urban inequality in Indonesia contributes significantly to national income inequality.In March 2020, the level of income inequality, as measured by the Gini Ratio, was 0.393, and in rural areas, it was 0.317 (Badan Pusat Statistik, 2021).This condition illustrates that inequality in urban areas is higher than in rural areas.Compared to the previous period, income inequality in urban areas in September 2019 was 0.391 or 0.001 points lower than in urban areas in March 2019, 0.092.Meanwhile, in rural areas in March 2019, it was 0.317 and decreased by 0.002 points compared to September 2019, i.e., 0.315 (Badan Pusat Statistik, 2020).
Many previous studies using the Gini Coefficient or Theil Index have measured inequality, particularly by analyzing income and consumption expenditure metrics using single-dimensional indices (Lazaridis, 2000;Lin, 2007).A pioneer in the study of inequality, Sen (1997) stated that there is nothing mathematically wrong with measuring inequality coming from income.Thus far, the evidence gathered shows that wealth inequality has increased in many countries over the last few decades, albeit at different speeds, highlighting the crucial roles of domestic institutions.These findings have been a concern among academicians, policymakers, and the public and have contributed to renewing the interest in wealth distribution theory (Benhabib & Bisin, 2018).
The roles of government in reducing inequality through fiscal decentralization policy are deemed critical.The essence of the policy is to divert the development budget from the central to the regional government.Fiscal policy directly affects inequality through tax progressivity, well-organized transfer, and the quality of public spending, indirectly affecting other factors determining income and wealth inequality (Odusola, 2019).The same result clearly showed that fiscal decentralization has significantly reduced income inequality (Feld et al., 2021;Goerl & Seiferling, 2014;Makreshanska-Mladenovska & Petrevski, 2019;Su et al., 2019), including rural and urban inequality (Chen et al., 2020;Deyin et al., 2017) On the other hand, some studies did not support the findings of previous studies, which have found that decentralization increased inequality (Clifton et al., 2017;Saputra, 2012;Sepulveda & Martinez-Vazquez, 2011;Siddique et al., 2008).Liu et al. (2017) also found that fiscal decentralization tends to increase inequality due to variability in levels of economic development, equal regional distributive policies, natural resources, and infrastructure imbalances.The same result revealed by Fan et al. (2020) highlighted that high central-to-regional transfers have not been able to reduce inter-regional inequality in per capita GDP and per capita income; in fact, the equalization effect was only found in urban income.Recent studies have also found that fiscal decentralization or policy increased income inequality (Cevik & Correa-Caro, 2020;Kyriacou et al., 2017;Liu & Long, 2021;Ong et al., 2023) and even tend to be weak (Canare et al., 2020).
Meanwhile, the provision of public capital, such as infrastructure, is deemed critical for the activities of households and companies (Zolfaghari et al., 2020) as well as production and consumption (Zou et al., 2008).Public capital becomes an engine of growth and a determinant of the distribution of wealth, income, and welfare (Chatterjee & Turnovsky, 2012).Infrastructure is a type of capital goods, institutions, and services that serve the activities of many industries in the market (Jain, 2012).It can be divided into two -economic and social infrastructure.The economic infrastructure refers to public utilities such as electricity, telecommunications, water supply, sanitation, and drainage; public engineering constructions such as dams and irrigation systems; and transportation facilities such as railways, ports, and airports.
In contrast, social infrastructure refers to education, health care, and health services (Zou et al., 2008).In addition, infrastructure can support economic growth by increasing labor and capital productivity, thereby reducing production costs and increasing profitability, production, income, and employment (Nallathiga, 2015).This study focuses on road infrastructure as the primary variable, while water supply, sanitation, and drainage become the control variables.
Road infrastructure has been proven to be capable of reducing income inequality.A recent study by Lu et al. (2022) found that road infrastructure is more important for rural residents to increase their income.Even a study has found that public investment in road construction has prompted economic growth and reduced poverty (Churchill et al., 2021;Zou et al., 2008).Other studies have confirmed previous findings that government spending for public capital can increase wealth inequality over time (Chatterjee & Turnovsky, 2012;Zolfaghari et al., 2020).Similarly, previous studies found that road transportation infrastructure can reduce income inequality (Huang et al., 2020;Mishra & Agarwal, 2019;Nugraha et al., 2020;Zhang & Zhang, 2021).On the other hand, some researchers have found less impact of infrastructure development on reducing income inequality (Fullerton Jr. et al., 2014).Other researchers have also shown that transportation infrastructure does not reduce income inequality (Fageda & Olivieri, 2019;Makmuri, 2017).
The differences in the research findings might be related to the differences in the research sites, variable measurement, and the level of economic development as another determining factor (Rodríguez-Pose & Tselios, 2010;Tselios, 2014), government size (Goerl & Seiferling, 2014;Sepulveda & Martinez-Vazquez, 2011), educational expansion (Shukla & Mishra, 2020;Tselios, 2014), poverty (Afandi et al., 2017) and other economic sectors (Zolfaghari et al., 2020).Even though this is not the first study to investigate the impacts of fiscal transfers and road infrastructure on income inequality, it is deemed crucial given the domestic perspective of Indonesia's context, which certainly requires another examination.Evaluating the consistency of the Kuznets curve with rural, urban, and rural-urban analysis units in Indonesia has been carried out previously with individual data (Tadjoeddin, 2013), international (Aktas & Iyidogan, 2022;Sidek, 2021), national (Yusuf et al., 2014), and sectorial district (Yusuf et al., 2021).
Several previous studies that have tested the relationship between fiscal transfers to regions and road infrastructure with income inequality still came up with inconsistent findings.In general, the interrelationships of these variables are analyzed separately, and few combine them in one analysis.In addition, in one study, a few previous researchers have yet to test the determinants of income inequality in rural, urban, and rural-urban areas.Thus, this study aims to examine the effect of fiscal transfer policies, road infrastructure, and control variables on rural, urban, and rural-urban income inequality to complement the existing literature on this issue, especially in the Indonesian context.The researchers used the Generalized Least Square (GLS) panel data regression model to find reliable evidence.This study adds to previous works, especially on the aspect of fiscal transfers in the form of general allocation funds and expansion of the use of special allocation funds (physical and non-physical) by the provincial government and road infrastructure in reducing various aspects of income inequality by taking into account rural-urban income inequality.Our empirical results corroborate that fiscal transfers and road infrastructure reduce income inequality.Therefore, this study provides results that policymakers can use to formulate policies that focus on strengthening fiscal transfers and building road infrastructure and how plans and interventions can be used as effective instruments to achieve income distribution in urban, rural, and national areas.

METHODS
This study used a quantitative method with a correlational design to examine the effects of fiscal transfer policy, road infrastructure, and control variables on income inequality using static and dynamic panel data regression.The secondary data refers to data obtained from the processing of a second party or data obtained from BPS (Central Bureau of Statistics) publications, academic literature, or journals, and those relevant to this research.The research used secondary data from a time series between 2012 and 2021 and cross-sectional data for 34 provinces (individuals) in Indonesia.
This research used the Gini Inequality Index (GI) dependent variable to measure income inequality.The data were used to estimate the Gini coefficient of household income inequality at the provincial level.The Gini coefficient is a commonly used measure of inequality (Allison, 1978;Badan Pusat Statistik, 2021).Gini inequality index variables consist of the rural (GIR), urban (GIU), and rural-urban (GI) inequality index as the dependent variable.The Gini coefficient value is 0 -1 (perfect equality to perfect inequality).Since this research had to use household income data, the Gini coefficient captured the distribution of household income in all household populations rather than the distribution of personal income among individuals.This research also used the inequality approach to income between rural-urban areas over time (Fuguitt et al., 1988;Imai & Malaeb, 2018;Jamal, 2019;Johnson & Lichter, 2020).
This study used the panel data regression model to predict the effects of fiscal transfer policy and road infrastructure on income inequality in rural, urban, and rural-urban areas.This model comprises ordinary least squares (OLS), fixed, and random effects models.This study used the Hausman Test to choose between the fixed and random effects models (Baltagi, 2005), which maintains that the fixed effects model remains applicable.Before applying the fixed effect panel regression method, this study tested the classical assumptions as an OLS linear regression to choose the right method to use in the data.In general, the panel data model can be written as follows: (1) Based on equation ( 1), the specifications of the general econometric model of this study are: where g = Gini inequality index (rural, urban, and rural-urban), ln(gdrpcap) = log natural of per capita income, lngaf = log natural of general allocation funds, lnsaf = log natural of special allocation funds, lnroad = log natural of road length province with an asphalt surface, i = provinces 1, 2, …, 34 and t = 2012 to 2021, Z = control variable consisting of mean years of schooling/mys, poverty/pov, unemployment/unemp, economic openness/openness, health infrastructure/inhealth, proper sanitation/sanit, clean water sources/water, and index gender empowerment/gei.γ = estimated parameter and e = error term.
Therefore, the use of random effects and generalized leans square (GLS) models is able to cope with the heteroscedasticity dan autocorrelation problem (Greene, 2003) because the OLS assumption is not fulfilled.The use of OLS may result in parameter estimation that is no longer efficient.The use of the REM equation model in a simple form can be written as follows: (2) Then equation ( 2) can be written as an error (Ɛ) equation: (3) To obtain an estimate of β with Generalized Least Squares is to minimize the squared error function (Ɛ). (4) In order for the value of Ɛ'Ɛ to be a minimum, a first-degree derivation of β is needed and equated to zero, as follows: so, (5) since β is equal with β GLS , then where ɸ is the weight

RESULT AND DISCUSSION
A descriptive analysis of the condition and development of income inequality at the regional, provincial, and national levels is presented before elaborating on the fiscal transfer policies and infrastructure's effects on income inequality.In 2013 and 2015, Etikonomi Volume 22(2), 2023: 333 -356 the Gini inequality index increased, indicating that the distribution of the population's expenditure in that period was getting worse.Meanwhile, since 2016, the expenditure disparity has continued to decline, showing the improvement in the distribution of population expenditure, as shown in Figure 1.The development of the Gini index between 2012 and 2021 is presented to identify the decline rate in expenditure inequality with the Gini index during the research period.The calculation results showed that the highest decline in expenditure inequality was found in Papua, with 11.18%, followed by Kalimantan, with 10.88%.However, in 2012 and 2021, the expenditure disparity in the Nusa Tenggara region experienced the opposite condition compared to other regions, which increased by 5.34%.The changes in expenditure inequality among urban, rural, and urban-rural by region are presented in Table 1.Table 1 shows that the changes in urban expenditure inequality in 2012 and 2021 occurred in all regions.In urban areas, the fastest rate of decline was found in Papua, with 15.42%, while the slowest was found in Java and Bali, with only 2.46%.However, two regions in rural areas experienced a rise: Nusa Tenggara and Papua.Conversely, other regions experienced a decrease in income inequality.This condition indicated that the income inequality in rural areas during this period was worse than in urban areas.Furthermore, by comparing the average acceleration of decrease or increase in the disparity in expenditure/income among residents at the provincial level, such as in Eastern Indonesia and Western Indonesia, this study revealed that the decline in the average expenditure inequality in Eastern Indonesia was lower than that of Western Indonesia so that by the end of 2021 the Gini index level in Eastern Indonesia had been higher, with an average of 0.358 (see Table 2).areas, with an average decline of 8.01%.In contrast, income inequality was getting worse in rural areas in Eastern Indonesia (up 0.57%), and the overall Gini index in rural areas decreased by 4.14%.The provincial development target, as recorded in the National Medium-Term Development Plan (Indonesian, RPJMN) 2020-2024, states that "inequality between provinces was reduced by encouraging the transformation and accelerating development in the Eastern Region of Indonesia, including Kalimantan, Nusa Tenggara, Sulawesi, Maluku, and Papua and by maintaining the momentum of growth in Java, Bali and Sumatera Regions" (Appendix I of the Presidential Regulation, 2020).However, this effort would be hard to realize because differences in resources and characteristics in each region are the primary factors of the increasing expenditure and development inequality (Heinrich Mora et al., 2021;Kamaruddin & Alam, 2019), especially in rural areas.Therefore, policymakers at the provincial level should pay more concern to the economic sectors that absorb more labor to make a real contribution to increase economic growth and empower people with lower incomes (Raharti et al., 2021).
A summary of the research data statistics should also be presented to control the research data before testing the relationship between variables.Statistical summary includes the average values, maximum values, and minimum values.The average Gini index value during the study period (2012-2021) in urban areas was higher than in the sub-national and rural areas.Table 3 presents the aforementioned statistical summary.Table 3 shows that the average provincial per capita income in 2010 showed constant values at the national level, with IDR 39,712,220 per population per year.The average value of general allocation funds (Indonesian, DAU) and special allocations (Indonesian, DAK) was IDR 1,241.41 billion and IDR 695.22 billion, respectively.The average value of the length of paved roads under the province's authority was 995.41 kilometers.The average poverty score for the province was 11.16%, and the mean years of schooling was 8.22 years.The average unemployment rate per province per year was 5.29%.
Meanwhile, the average ratio of imports to exports was 5.72% per year, and health facilities in hospitals, health centers, and auxiliary health centers were 277.97 units.The average percentage of the population accessing proper sanitation and clean water was 28.96% and 36.37%,respectively.The average women's empowerment index was 76.71%.values were 336.324 -582.525, and the probability value was 0.000 < 0.005; thus, the FE model was selected.Furthermore, the researchers chose between the Fixed Effect and the Random Effect (RE) models utilized with the Hausman test (Hausman, 1978).Both models were potentially valid in estimating the panel model with the unobserved provincial heterogeneity.Hence, the FE or RE model could be a valid model based on the results of the Hausman specification test.A formal test of the Hausman specification was estimated to generate the right decision between the two models: FE and model and RE model.The test results for all models showed that the Chi-square statistic of the cross-sectional random was 12.249 -66.054 while the Probability value was 0.000 -0.002 < 0.05.Thus, the effects model remained consistently applicable for the analysis.
The results of the residual normality of all analytical models with Jarque-Berra resulted in a Probability value of 0.3507 -0.9368 > 0.05, indicating the normal distribution of the data.The correlation value between independent variables could be used for multicollinearity.The correlation calculation result showed that r = -0.3151-0.6059 < 0.80, indicating that the model is free from multicollinearity.Also, since the panel data have cross-sectional and time series dimensions, data analysis commonly faces heteroscedasticity and serial correlation (Beck, 2001;Hsiao, 2007).Some of the causes of this problem have been extensively discussed by Beck (2001), who revealed the following results.First, provincial governments can have their variants of error related to their unique regional characteristics; second, there might be a spillover of the errors (surprise) in one provincial government to other regions within the same region.This condition may lead to serial autocorrelation if a shock from one provincial government is correlated to the shocks from other provincial governments in the same year; third, within the same provincial government, shocks or errors in the current years might be correlated with errors in the previous period or periods.
Based upon the problems related to the nature of the data in the panel model, the researchers tested the results of the autocorrelation test using the Durbin-Watson (DW) value criterion, i.e., the DW value between 1 and 3 (Field, 2009).Thus, conservatively, there would be no autocorrelation problem (though the value around two also becomes a concern).In this model, the value of DW was 1.517 -1.885; hence, the model is free from the autocorrelation problem.Finally, the heteroscedasticity test was carried out using the White-Heteroscedasticity test, namely the Obs*R-squared of 9.675 -161.493 and the probability value for all models of 0.000 -0.022 < 0.05.On this basis, it was found that there was a heteroscedasticity problem in the regression model.
Further, the relationship between fiscal transfer policy road infrastructure and income inequality was examined.Here, several income inequality models used a panel estimation analysis of Generalized Least Squares (GLS).This estimator produced not optimally biased results for considering heteroscedasticity and serial correlation in panel data (Hansen, 2007;Musau et al., 2015).Table 4 below presents the results of the GLS panel containing coefficients, standard error, probability values, R2, and F-stat.Tables 4 and 5 show that Kuznets' hypothesis showing the relationship between per capita income and income inequality has existed in rural, urban, and rural-urban areas for all provinces in Indonesia (with control and uncontrol variables), except in urban areas (Urban Model 1) that was without control variables.Hence, it is clear that the increase in people's income at the beginning was proven to increase income inequality significantly, but when reaching the threshold of certain income per capita, increasing income per capita was able to reduce income inequality in rural, urban, and rural-urban areas (with control variables).On the other hand, the coefficient value of fiscal transfer Volume 22(2), 2023: 333 -356 as measured by the general allocation fund was 0.001 -0.004, which was statistically insignificant in the rural, urban, and rural-urban models (see Tables 4 and 5).This result means that the general allocation fund increased the average income inequality in rural, urban, and rural-urban, although it was insignificant.

Etikonomi
The coefficient value of the special allocation fund was -0.003 to -0.004 (rural, urban, and all models 2 and 3).This result indicates that special allocation funds could reduce the average income inequality in Indonesia's rural, urban, and rural-urban areas.The road infrastructure coefficient value was -0.021 -0.031.It was statistically significant by 1% and 5% in the rural and urban models, but insignificant in the urban model 3 and the rural-urban model (see Table 4 and Table 5).This pinpoints that road infrastructure could reduce the average income inequality in urban areas but not in rural and ruralurban.Income per capita as a control variable consistently reduces income inequality in rural, urban, and rural-urban areas.
Furthermore, the control variable that significantly reduced income inequality in urban areas and throughout the province was health infrastructure-the mean years of schooling variable only had a significant effect at the whole province level.The variables of women's empowerment and economic openness only had a significant effect in urban areas.
The results of evaluating the Kuznets' curve model of income inequality in Indonesia by provincial level and rural existed without control variables.Also, they existed with control variables for rural, urban, and rural-urban models.The income threshold could be achieved at IDR 27.80 million in rural, IDR 38.62 million in urban, and IDR 37.72 million in the sub-national average level in achieving welfare when income inequality started to decrease.It highlights that the level of income distribution started to improve when the average level of per capita income in each region reached the threshold level during the study period.This condition also illustrates opportunities to improve community welfare through regional economic activities.
The results of Kuznets' hypothesis testing showed a number of differences between rural, urban, and provincial averages.The differences in the evaluation results between rural and urban areas were caused by higher income inequality in urban areas compared to rural areas in 2012 -2021.In addition, the Kuznets curve inequality model estimation in rural areas did not include the inequality data for Jakarta Province since it did not have rural areas.Another reason, as stated by Kuznet (1955), was the alteration from low-productivity agriculture to more productive sectors such as manufacturing and services during the growth process.Consequently, inequality widened in urban areas and decreased in rural areas.The findings of this study were in line with the research by (Aktas & Iyidogan, 2022;Sidek, 2021;Tadjoeddin, 2013).
We showed that fiscal transfers significantly reduce income inequality in rural, urban, and rural-urban areas.The instruments for fiscal transfers to the regions were in the form of general allocation funds and special allocation funds, where special allocation funds to the province are strongly related to the declining income inequality at the regional level, while general allocation funds do not, although there was a positive relationship between them.This condition illustrates that the general allocation funds transferred to the province cannot redistribute income directly to the community, given the fact.Bearing in mind that the general allocation funds implemented by the government are for equal distribution of provincial financial capabilities, thereby reducing gaps in provincial funding sources and fiscal gaps between provinces in the context of meeting regional needs in decentralization.It means that general allocation funds could not increase people's income in rural, urban, and rural-urban areas.This result differs from previous studies, which found that fiscal transfer policies can reduce income inequality among residents in rural, urban, and rural-urban areas (Kyriacou et al., 2017;Ong et al., 2023).Another study that did not confirm these results was conducted in South America by Clifton et al. (2017), which showed that fiscal policy has been proven to reduce income inequality in the region.However, these results confirmed previous research findings that the relationship between fiscal decentralization policies and inequality was weak (Canare et al., 2020).
The special allocation fund significantly reduced income inequality in rural, urban, and rural-urban areas.In other words, there was an increase in special allocation funds, thereby reducing income inequality.This study has proven that one of the keys to reducing income inequality was increasing government budget allocations through the mechanism of special allocation funds.The results of this study are consistent with the study of Enami et al. (2019), which found that the fiscal transfer system was both directly and indirectly effective in reducing the Gini income inequality index in Iran.Also, the findings of this study confirmed previous findings revealing that the decentralization of fiscal revenues is negatively correlated with income inequality in urban, rural, and rural-urban areas (Chen et al., 2020;Deyin et al., 2017).Qomariyah et al. (2019) specifically emphasized that special allocation funds may reduce income inequality.In addition, Su et al. (2019) concluded that fiscal decentralization impacted increasing population income and fiscal spending, encouraging income growth for urban and rural residents.However, the growth rate slowed over time.Meanwhile, other researchers have also emphasized that fiscal decentralization is an important determinant that significantly reduces inequality (Makreshanska-Mladenovska & Petrevski, 2019).
However, the empirical results of this study did not confirm any research findings highlighting that decentralization increases inequality (Saputra, 2012;Sepulveda & Martinez-Vazquez, 2011).A similar research finding also articulated that fiscal decentralization did not impact income inequality (Cevik & Correa-Caro, 2020;Fan et al., 2020).Likewise, Mudayen & Maridjo (2017) also found an inverse relationship between special allocation funds and income inequality between provinces in Indonesia.The special allocation fund for assignment supports accelerating the development of basic service infrastructure focused on priority locations (sub-districts) in the district/city.The provincial government has a vital role in managing transfer funds from the central government through special allocation funds to accelerate area-based development, namely Etikonomi Volume 22(2), 2023: 333 -356 rural and urban areas.Following its designation, special allocation funds aim to accelerate infrastructure development for education, health, road transportation, drinking water, sanitation, housing, and settlements to reduce income inequality in rural, urban, and rural-urban areas.After this, it directly impacts improving the economy and community welfare.
Transportation development is expected to increase public access to ensure the smooth operation of economic activities to increase people's income.Road infrastructure has become the center of attention at the urban, rural, and national levels because of its very influential role in ensuring the mobility of goods and services in the regions.However, the results of this study show different results regarding the effect of road infrastructure on income inequality by region.The research findings have shown that road infrastructure significantly reduced income inequality in urban areas.In other words, improving road infrastructure could reduce income inequality in urban areas.Hence, road connectivity in urban areas was better than in rural areas.The empirical test has confirmed previous findings that transportation infrastructure significantly affected income inequality (Z.Huang et al., 2020;Mishra & Agarwal, 2019).
On the other hand, the results of empirical testing showed that road infrastructure did not significantly reduce income inequality in rural and rural-urban areas.This result means that the improvement of road infrastructure increases income inequality in rural and rural-urban areas.This finding corroborates previous research, which found that road infrastructure improvement does not impact income distribution (Fullerton Jr. et al., 2014).The wealth generated by rural people working in cities is poorly distributed to themselves, exacerbating income inequality within cities (Zhang & Zhang, 2021).Likewise, (Fageda & Olivieri, 2019) showed that massive transportation infrastructure investment reduces the income gap between regions.The results of previous studies in Indonesia with provincial-level data show that an increase in the number of roads widens the income gap in Indonesia (Makmuri, 2017).
Road infrastructure for rural communities is critical to increase their income (Lu et al., 2022).The income received by people in rural areas is starting to approach that in urban areas, so rural residents are also starting to benefit from the provision of transportation infrastructure.On the one hand, the income of the rural population can increase when the excess rural workforce has the option of working in the secondary and tertiary sectors in urban areas (Liu & He, 2019).Hence, it is necessary to improve the quality and quantity of road infrastructure development because it has proven to impact increasing income distribution in urban areas.Road infrastructure development is not merely intended for luxury but serves the basic needs of the community that must be met.Therefore, income inequality between residents shall be reduced through the improvement and addition of proper roads in urban areas because it can increase the income and welfare of people in urban areas.Increasing income may reduce overall income inequality (Chatterjee & Turnovsky, 2012;Liu & He, 2019;Zhang & Zhang, 2021;Zolfaghari et al., 2020).Since Indonesia is an archipelagic country, there needs to be better connectivity between regions in rural areas, thus hindering community access to economic activities that ultimately lead to longer time for distribution and expensive transportation costs.
As a consequence, this condition has exacerbated income distribution in rural areas.In contrast, urban areas have better connectivity with better, flexible, and accessible road infrastructure to increase income distribution.In addition, the positive impact of urban roads on the Gini coefficient lies in the development of the road network and the urban economy, leading to large numbers of rural people migrating to cities for work.In this case, those living in rural areas become increasingly marginalized, increasing income inequality.
Although research findings have proven that fiscal transfer policy and transportation development reduce inequality in rural, urban, and rural-urban areas, they also added to the consistency of previous findings, indicating some weaknesses.The drawback of this research mainly lies in the measurement of fiscal transfer, which needs to include the tax and non-tax revenue sharing, which is included in the transfer fund.Although the revenue-sharing funds are not as significant as the general allocation funds and special allocation funds, the use of these funds by the regions can be directed at increasing income distribution.In addition, this study also did not include roads under the authority of the central and district/city governments as well as railways.However, the total number of roads in all regions, either both national, provincial, and city districts, as well as railways, has proven to be significant in reducing income inequality (Zhang & Zhang, 2021) since it can increase the economic activities without disturbing people's connectivity in the form of roads.Therefore, it is unsurprising that this research has shown that road infrastructure only contributes to reducing income inequality in urban areas.Another weakness of this study is related to the limited length of the series.Thus, as a consequence, it was not easy to do modeling for data analysis to produce relevant analysis.

CONCLUSION
This study uses the GLS fixed effect panel based on time series data from 2012 to 2021 in 34 provinces to examine the relationship between fiscal transfer policy and road infrastructure and income inequality in Indonesia.The researchers concluded that this research contributes to the development of the existing body of knowledge and inputs of policies to reduce income inequality.The study results have shown that the Kuznets' hypothesis existed in rural and rural-urban areas (without control variables) and all models (with control variables).The main results show that the special allocation fund significantly reduces income inequality in rural, urban, and rural-urban areas.However, the general allocation fund does not significantly reduce income inequality.The government plays an essential role in determining policies to decrease income inequality.In addition, it was also revealed that road infrastructure is a determinant factor in reducing income inequality in urban areas but only in rural areas and rural-urban for some provinces.Developing and improving road transport Etikonomi Volume 22(2), 2023: 333 -356 infrastructure does not automatically lead to decreasing income inequality.Hence, the government is suggested to improve the ability of low-income groups in urban areas to utilize transportation infrastructure to increase their income, such as e-commerce and logistics transportation-based economic activities.The government can also minimize the increasing flow of urbanization from rural areas and move the orientation of rural economic infrastructure development to increase people's incomes in rural areas to create an even distribution of economic development.
Referring to the study's limitations in which infrastructure was not significantly correlated with income inequality in rural areas and on a national average, it is necessary to retest by adding different data and analytical models (e.g., System Dynamic Panel).Also, road transportation has ignored rail transportation, which all levels of society have widely used.In addition, the following studies must include land transportation variables in the form of trains in an analytical model to generate more comprehensive analytical results.The regional fiscal transfer policy also needs to address the existence of tax and non-tax revenue-sharing funds to the regions not included in the model in predicting income inequality; thus, future researchers should add tax and non-tax revenue-sharing variables.

Figure 1 .
Figure 1.The Development of Indonesia Gini Index, Urban, Rural and the Difference of Urban and Rural, 2010 -2021

Table 2 . The development of the Gini Index based upon the West-East Indonesia area, in 2012 and 2021
Table2shows that the decline rate in expenditure inequality in urban areas from 2012 to 2021 in Eastern Indonesia was more rapid than in Western Indonesia's urban

Table 4 . The Results of the Regression of Fixed Fffect of GLS Rural and Urban
First, the researchers conducted the Chow test to select the Common Effect (CE) or Fixed Effect (FE) model.The test results showed that all cross-section Chi-square