Efficiency Vs. Equity of State Electricity Distribution: Evidence from an Indonesia Village

The purpose of this paper is to analyse the determinants of household access on State electricity at the village level in Indonesia. Access to State electricity differs significantly between provinces in Indonesia, as well as between rural and urban areas. This study utilises data at the village level known as Potensi Desa or PODES, for the time period between 2005 and 2011. Panel data regression is performed to estimate the impact of village's remoteness as well as economy and social heterogeneity among communities in villages. By controlling village’s characteristics, this study finds that the number of households utilising State electricity are lower in villages where: (i) they have low population density, (2) are located in remotes areas, and inhabitants have limited access to economy-generating activities. Access to State electricity is significantly better in the region where potential economic activities are mostly non-agricultural. As compared to a previous study that focuses on ethnic heterogeneity, this study found that in villages where the community comprises of more than two ethnicities and more than two difference religions, households access on State electricity is higher than the villages where there is no social heterogeneity. This study concludes that State electricity distribution meets efficiency principle, but not equity or universal distribution.


Introduction
Electricity is an essential energy source for supporting daily activities of household, businesses, and industries. Domestic activities such as cooking, water pumping, washing and ironing will be easier if households utilise electricity and have sufficient wattage for supporting the appliances. Even though electricity access is important for daily activities, communities in rural areas of developing countries such as Indonesia still cannot enjoy access to electricity, universally.
Access to State electricity differs significantly among provinces. Gap in electricity access is also large between urban and rural areas. On an average, electrification ratio in Jawa and Sumatera islands is up to 95 per cent, while in other provinces the ratio is about 80 per cent, but in East Nusa Tenggara and Papua the ratio is still below 30 per cent (BPS, 2014).
In Indonesia, differences in access to electrification are also related with different rules in local institutions. Usually, it is organised by traditional 'adat' or civil law (Pal and Wahhaj, 2012). In regions where adat rule is dominant, average electricity consumption is lower than the regions where the pertaining rule is civil law. Electricity distribution not only varies by regions, but also by the level of economic activities. West Papua and Central Kalimantan provinces are richer in natural resources, but at the district level, these provinces still experience gaps in access to electricity. Communities in urban areas enjoy almost 90 per cent electricity, while in rural areas, household access to electricity is below 40 per cent (DettmanandPepinsky, 2014).
In geographically disadvantaged province like East Nusa Tenggara, percentage of rural household with State electricity access is only about 28 per cent.
Official Statistic data shows that the access to State electricity is concentrated in Jawa and Sumatera. On an average, number of villages in Jawa, Aceh and North Sumatera have almost three to four times higher access to PLN than villages in other regions or islands (BPS, 2012;. Demographic indicators show that total size of areas in Jawa and Sumatera are approximately 32 per cent of Indonesia, but the number of villages in these two islands is about 63 per cent of the total villages in Indonesia. In terms of regional per capita GDP, provinces in Jawa and Sumatera have relatively higher per capita GDP as compared to other provinces in the eastern part of Indonesia (BPS 2012(BPS , 2015.
In this case, larger share of households with electricity access in Jawa and Sumatera might be related to higher potential demand and cost efficient for electricity grid distribution.
There are a growing numbers of studies that investigate determinants of access to basic infrastructure in developing countries, especially access to electricity in rural areas.
Empirical studies consider various factors that might influence electricity distribution. Some studies consider geographical barriers, while other studies consider economic and social diversity, demand and supply sides or human capital and government funding. Studies that focus on geographical barriers in general conclude that there is a negative link between access to electricity with the degree of village's remoteness. Villages with mountainous or hilly landscapes, separated by river, lake or sea require higher cost of installation (Bazilian et al., 2012;Cook, 2013aCook, , 2013bFoster and Caramanis, 2013;Lahimer et al., 2013;Zvoleff, Kocaman, Huh, and Modi, 2009).
In terms of the supply side, economic and social diversity are considered as an obstacle to governments in providing infrastructure and access to public goods like electricity.
Social diversity can influence preference of individual and the community on consuming types of public goods. On the demand side, economic inequality can reduce the potential demand for electricity, as this is related to price affordability from low income consumer.
Studies that focus on diversity reported that both economic and social diversity have negative association with infrastructure and access to public goods distribution, including electricity (Alesina, Baqir and Easterly, 1999;Alesina and Ferrara, 2004;Banerjee, Iyer and Somanathan, 2008;Banerjee and Somantahan, 2004;Banerjee., Iyer and Somanathan, 2005).
Access on electricity distribution is negatively related with the role of institution, the quality of regulation and the availability of technician for maintaining the electricity network. In some developing countries, power loss becomes another issue in-line with bad institutions and low gridline maintenance (Agbemabiese, Nkomo and Sokona, 2012;Chaurey, Krithika, Palit, Rakesh and Sovacool, 2012;Javadi, Saidur and Kamalisarvestani, 2013;Karekezi and Kimani, 2004;Yadoo and Cruickshank, 2010). in Papua. This data implies that diversity in ethnicity is larger in the eastern part as compared to the western part of Indonesia (Jawa and Sumatera). Empirical studies reported that higher the variety in group characteristics, lower will be the probability of that group to get public goods allocation (Banerjee et al., 2008;Esteban and Ray, 1999). This finding is supported by many studies such as Alesina, Baqir, and Easterly (1999), Banerjee (2004), Banerjee and Somanathan (2007), Baldwin and Huber (2010) and Habyarimana et al., (2007).
The issue to be addressed in this study is estimating the determinant of electricity access at village level in Indonesia. Following previous empirical studies which investigate the role of geographic and diversity as determinant of electricity distribution, this study focus on testing two points. Firstly, does geography or social diversity undermine the State electricity distribution in rural areas of Indonesia? Secondly, can villages with better access of electricity generate a variety of small household business? This study will utilise micro data at village level, known as Potensi Desa (PODES) for the data surveys in 2005 and 2011. PODES is a nationwide survey; therefore, utilising this data will benefit in terms of national coverage. As far as the authors are concerned, among different studies referred in this study, only the study of Oda and Tsujita (2010) apply village level data for the case in India. The authors believe that by applying micro data at village level, deeper empirical evidence can be presented as compared to investigate data at macro level.
The rest of the paper is organised as follows: The next session will present brief literature review and hypothesis. This session is followed by presenting model and data, and subsequently followed by results and discussion. The last two sessions are policy recommendations and conclusion. shorter studying time and low quality of food intake will deteriorate human capital investment in the long run. This condition would further result in poor condition/s that will be carried into the next generation, and in aggregate, this condition results in low income and output. Based on these considerations, access on electricity becomes one of the main points for achieving Sustainable Development Goals (SDGs), especially for reducing poverty rate (Attigah andTasch, 2013).

Empirical Studies of Electricity Access
Empirical studies report positive impact of electrification access on education, health, fertility rate, income and productivity.
An improvement of rural areas for access on electricity has significant positive impact on educational outcomes (Brodman, 1982;Dasso, Fernandez and Nopo, 2015;Squires, 2015).
Other studies find that crime rate has changed in specific regions where street lighting is supported by access from State electricity (Clarke, 2008;Wright et al., 1974). Comparing score test for math and reading, this study also finds no significant difference between threated and non-threated villages.
However, among the threated villages, reading test score for girl is higher when compare to boys, but the opposite is true for math test score. Squires (2015) applies longitudinal data (1991, 1993, 2001, and 2005) to investigate the impact of electrification programme on education in Honduras. In general, this study concludes that there is no significant impact of electrification on children school enrollment, school attendance and potential dropouts.
Surprisingly, this study reported that there is a negative impact of rural electrification on school enrollment, attendance and risk of dropouts in a village. The author argues that this finding might reflect on short run adjustment.
Access on electricity was responded by higher participation in labour market. Household wife can substitute their domestic time into labour Efficiency vs. Equity of State Electricity Distribution: Evidence from an Indonesia Village market, but consequently, the first son or daughter have to sacrifice their schooling for looking after their younger siblings. Therefore, the impact of electricity on school attendance is not significant, but it is transmitted in female participation in labour market.
Short and long run impacts of electrification programme on fertility rate are reported by Herrin (1979) and Grimm et al., (2015). The study of Herrin (1979) was conducted in Misamis province, Northern part of Mindanao Island, Philippines. This study investigated 43,725 villages in Misamis, for the period 1970-1975. Based on the empirical investigations, the authors conclude that relatively in short run from 1970 to 1975, the fertility rate in Misamis decreased significantly. The authors interpreted this result carefully due to mixed and indirect evidences regarding impact of electricity access on fertility rate. Combinations of follow-up programmes for rural electricity were recorded, which includes agricultural, support for small business, road improvement as well as family planning programme. This study found business and labour market confidence increased and it was followed by higher participation in female labour. This finding implies that bringing up a child becomes expensive, and as result significant decline in fertility rate. The study of Grim et al., (2015) focuses on Indonesia. This study utilises a combination of four sources of micro data, including: annual Indonesian National Socio-economic Household Survey (SUSENAS), longitudinal data of villages survey and Vaona and Magnani(2014). The effect of diversity on economic, social identity and geographic aspects of electricity access and consumption has become a growing concern recently (Glennerster, Miguel and Rothenberg, 2010;Habyarimana et al., 2007;and McQuoid, 2011).

Hypothesis
Hypothesis to be tested in this study will be related to the characteristics of Indonesia Three hypotheses will be tested in this study. The first two hypotheses are regarding the determinant of electricity access. This study hypothesises that electricity access is: (1) negatively related with geography remoteness and (2) negatively related with ethnic and religious diversity. The first hypothesis follows the study of Torero (2014) and the study of Alex et al., (2006). Remote areas have difficult topography and typically low population density. Low population means low economic activities and consumption. On the other hand, empirically, it is also reported that there exists a negative relationship between high ethnic and religious diversities with access to public goods, including electricity (Alesina, Baqir and Easterly, 1997;Alesina and Ferrara, 2004;Baldwin and Huber, 2010;Banerjee, 2004;Banerjee. et al., 2005). (1) Notations in equation (1) Table 1 indicates that communities in the villages slowly get mixed in ethnicity, but do not in terms of religion.   By controlling the village characteristic, Table 2 indicates less assessability or remote location of a village and lower number of households utilising State electricity. Remoteness is indicated by villages located inside the forest, only assesses through water transportation, no asphalt road, no regular transportation and further away from city centre. Combining village's characteristics and village's heterogeneity does not change the sign of coefficient of village's remoteness, even though the magnitude gets lower. Table 3 presents estimation results of association between village's characteristics and heterogeneity. Heterogeneity in social identity is specified as heterogeneity in ethnicity and religion. Village inhabited by more than two different ethnicities and inhabitants pertains to more than two different religions.  The estimation result in Table 3 indicates that both diversity in ethnicity and religion do not determine access on electricity. Villages where inhabitants belong to difference ethnics and religions do have more households utilising State electricity, as compared to villages with homogeneous inhabitants. Indonesia is inhabited by 1200 different ethnics and the State declares formally five religions in Indonesia, namely: Islam, Catholicism, Protestantism, Hinduism and Buddhism. Development and cultural assimilation result in heterogeneity of inhabitant in each village. Data description in Table 1 indicates that in about 62 per cent of the villages in 2005 and about 72 per cent of villages in 2011, they have two or more ethnicities co-existing in the same village. No changes were recorded for religious diversity between 2005 and 2011.

Estimation Result
The first two hypotheses to be tested are related with Table 2 and Table 3. The first null hypothesis states that there is no association between remoteness and access of transportation to the number of households utilising State electricity. The second null hypothesis points that there is no association between heterogeneity conditions in the village with the number of households utilising State electricity. Regarding the first hypothesis, estimation results indicate that villages located in or near the forest, only can be reached through water transportation. Village roads are not covered by asphalt and the further roads from villages to sub-district cities have lower households that utilise State electricity. The sign of coefficients estimation for geography and remoteness have not changed, even with slight depression in the magnitude when relaxing the control variable i.e., characteristics of the village such as population density.
Regarding estimation for the second hypotheses, this study finds different results as compared to the previous study by Alesina et al.(1999), Alesina and Ferrara (2004), Banerjee and Somantahan (2004), Baldwin and Huber (2010) and Balasubramaniam, Chatterjee and Mustard (2014). These studies claim that the higher the index of fractionalization for ethnicity, lower the distribution of public Efficiency vs. Equity of State Electricity Distribution: Evidence from an Indonesia Village goods. The study of Alesina et al. (1999) focuses on America, while Alesina and Ferrara (2004) conducted cross-countries study. The two studies found that ethnic fractionalisation do undermine the effectiveness of public spending and further lowering access to public goods such as health facilities, school infrastructure, electricity, water and sanitation among various States in America.
The study of Banerjee andSomanathan(2004) andBalasubramaniam et al., (2014) Baldwin and Huber (2010) updates the study of Alesina and Ferrara (2004)  According to the study of Baldwin and Huber (2010), Between-Group Income Inequality or BGI in Indonesia is 0.32 and it is ranked 23 among 46 countries. On the other hand, the Ethno Linguistic Fractionalisation (ELF) is about 0.76, and Indonesia is ranked number 7 among 46 countries. Cultural Fractionalisation (CF) of Indonesia is quite high, close to 0.6. These figures indicate that social heterogeneity (ethnic, religion, language, and cultural) is quite high as compared to the index of income inequality. Figures 1, 2 and 3