Does local government proliferation improve public service delivery? Evidence from Indonesia

ABSTRACT

Local government proliferation—the creation of new local governments via the splitting of administrative jurisdictions into smaller units—is a ubiquitous phenomenon in developing countries. Supporters of proliferation argue that it brings government closer to citizens and thereby helps to better match public service supply with demand. In Indonesia, the number of local governments has increased by about 80% since just prior to the initiation of the government’s decentralization program in 2001 until present. This investigation exploits the plausibly exogenous timing of local government splitting to identify the causal effects of proliferation on education and infrastructure service access in Indonesia. The examination finds that the establishment of new local governments has no impact on school enrollments but that it negatively affects water and sanitation access. The study offers some preliminary evidence to imply that the poor infrastructure service performance of newly created local governments is driven by the relatively corruptible nature of the sector and the comparatively more fragile governance environments that exist in new local jurisdictions.

Acknowledgments

The author thanks three reviewers for their very useful comments and suggestions.

Notes

1. There are currently 34 provinces and 529 districts (not including Jakarta), including 430 (mostly) rural districts (kabupaten) and 99 municipalities (kota). Provinces and districts are autonomous, nonoverlapping units of government, and both are directly accountable to the central government. Districts are not formally accountable to the provinces within which they are located, although the latter play a coordinating role among districts within their borders.

2. This study focuses on service access and not service quality, due to a lack of comprehensive data on the latter.

3. Government Regulation (GR) 129/2000 was in force from 2000 until 2007, at which time it was replaced by GR 78/2007.

4. The data indicate that five parent districts themselves split during the period, creating an additional six child districts. This study associates those newly created districts with the original split districts. An alternative would be to drop those districts from the analysis. This alternate strategy was also implemented and there were no qualitative changes to the derived conclusions.

5. Protected water as defined by BPS includes piped water, communal tap water, rainwater, spring water, and covered wells. Protected sanitation facilities include flush toilet, piped sewer system, flush/pour flush pit latrine, and ventilated pit latrine.

6. In fact, economic conditions during the period of study were quite stable. By 2001, recovery from the Asian financial crisis was well underway, and growth between 2001 and 2004 held consistently around 4.5% per year. In 2005, growth increased to around 5.0% per annum and remained relatively stable through 2010. See Tabor (2015) for the evidence and some discussion.

7. An alternative to difference-GMM is systems-GMM (Blundell & Bond, 1998). The latter augments difference-GMM by simultaneously estimating the levels equation (i.e., Equation 2), whereby moment conditions imply that lagged changes in the dependent variable can serve as instruments for the level of the lagged dependent variable. As usual, exogenous variables function as instruments for themselves. The additional instruments used in systems-GMM may result in more efficient parameter estimates, but the increased efficiency comes at the cost of an additional assumption: deviations from the long-run mean of the dependent variable, conditional upon any exogenous variables, must be unrelated to the individual-specific unobserved effect. Preliminary analysis in this study suggested that difference-GMM is the preferred model and so it is used in the empirical examinations here.

8. Overinstrumentation can result in two main problems. First, use of too many instruments may overfit the endogenous variables, thus inadequately expunging the endogeneity and biasing coefficient estimates in the process. Second, the employment of an excessive number of instruments may also create problems for the Hansen test of instrument exogeneity; that is, if the number of instruments becomes too large, the test is invalidated. See Roodman (2009) for elaboration.

9. The test of overidentifying restrictions can be seen as a test of the null hypothesis that the set of instruments is exogenous. If the null is not rejected, the test provides support for the claim that the instruments can be taken as exogenous. See Roodman (2007) for a discussion.

10. Data on DPRD political fragmentation have been provided by the General Elections Commission (KPU [Komisi Pemilihan Umum]) and on local government audit opinions have been supplied by the Supreme Audit Agency (BPK).

11. Patronage politics are common throughout the developing world. See Gordin (2002) and Arriola (2009) for reviews of the phenomenon for Latin America and Africa, respectively.

12. Political fragmentation (PFG) is defined as follows: , where the subscripts i and j refer to individual districts and political parties, respectively, and ν is a political party’s share of total votes in the district. See Laakso and Taagepera (1979) for the initial development of the index.

13. The expected (i.e., predicted) values of political fragmentation and audit outcomes conditional on the exogenous control variables used in this study are 5.8 and 1.8 for original districts and 6.5 and 1.4 for new districts. Differences across new and original districts are statistically significant at the .01 level.

Additional information

Notes on contributors

Blane D. Lewis

Blane D. Lewis is Senior Research Fellow and Associate Professor of Economics in the Arndt-Corden Department of Economics, Crawford School of Public Policy, The Australian National University. He is editor of the Bulletin of Indonesian Economic Studies. His research focuses on intergovernmental fiscal relations, local public finance, service delivery, and decentralization, especially in Indonesia. He is widely published on these topics, and his research has appeared in Economic Geography, Fiscal Studies, Journal of Development Studies, Journal of Urban Affairs, Public Administration and Development, Studies in Comparative International Development, Regional Studies, Urban Studies, and World Development, among others. He has worked for many years as a policy advisor in Indonesia with international agencies such as the Harvard Institute for International Development, Research Triangle Institute, and World Bank.