The Impact of the Decentralization Program on Poverty Reduction in Indonesia: A Macro and Micro Level Analysis
Ali Said ([email protected])
Accepted 2009, The Flinders University of South Australia, Adelaide
This thesis represents an effort to examine the impact of the implementation of regional autonomy on poverty reduction in Indonesia at both the macro and micro levels of analysis. At the macro level, the study addressed the relationship between decentralisation and poverty by focusing on districts in Java–Bali. At the micro level, it focused on comparing the performance of the decentralisation program, especially with respect to poverty reduction, in two selected districts (Purbalingga and Purworejo).
The macro level analysis shows that nearly half of the selected districts in Java–Bali experienced a more rapid reduction in poverty after than before decentralisation, while in a similar proportion the recorded reduction in poverty was slower than before decentralisation. The remaining districts (8.25%) recorded an increase in poverty after decentralisation. In other words, decentralisation has assisted poverty reduction in a large number of districts, but certainly not in all. The implementation of regional autonomy has also led to a considerable rise in real per capita public spending at the district level, suggesting that decentralisation may have helped poverty reduction by increasing the funds available for spending on programs that assist poor people.
The macro level analysis shows that there are three main challenges facing the success of the decentralisation program in reducing poverty more rapidly. First, in many districts, a high rate of economic growth has not benefited the poor. Second, the challenge facing local governments during the decentralisation era is how to target public spending so that it benefits the poor population and is effective in poverty reduction, since there has been a tendency for the poor to gain less from public spending, especially in health care and higher education, than the rich. Third, the widespread corruption that emerged during the first five years of decentralisation reduced the effort to improve public services in general and poverty in particular.
Because the macro level analysis failed to explain why poverty reduction improved under decentralisation in some districts but not in others, and could not identify the mechanisms through which the implementation of decentralisation has helped poverty reduction, a case study of two districts was undertaken. This micro level analysis reveals that decentralisation has had a positive impact on poverty reduction in Purbalingga, while the rate of poverty decline in Purworejo has slowed. Despite a higher economic growth rate in Purworejo than in Purbalingga since decentralisation, this growth has given less benefit to the poor in Purworejo than in Purbalingga. The significant progress achieved by the local government of Purbalingga in reducing poverty during the autonomy era, and the weaker impact of decentralisation on poverty reduction in Purworejo, can be explained by government performance in public spending and public service delivery, the performance of local economic development, and governance quality. Among these three main factors, good governance seems to be the key to successful implementation of decentralisation for poverty reduction.
© 2010 Ali Said
Three Essays on Econometric Evaluation of Public Health Interventions
Meliyanni Johar ([email protected])
Accepted 2009, University of New South Wales, Sydney
This dissertation consists of three independent essays evaluating the impact of public health interventions in Indonesia and Australia.
The first two essays concern the health card program in Indonesia. The program aims to allow poor households access to primary health care by providing a price subsidy for treatment at public health facilities. The data are derived from a longitudinal study of households, the Indonesian Family Life Survey (IFLS).
The first essay evaluates the effectiveness of the program. The quantitative approach combines the difference-in-differences (DID) estimator and propensity score matching (PSM–DID) to construct the unobserved counter-factuals: the outcomes of the program recipients had they not been exposed to the intervention. The health card program uses administrators to identify recipients on the basis of their welfare characteristics, as opposed to relying on the voluntary participation of households. This, together with the availability of many covariates, supports this strategy in providing accurate estimates of the program's effect on its recipients. The results indicate that the presence of a health card has only a limited impact on the utilisation of primary health care. The ineffectiveness of the health card program suggests the presence of other factors counteracting the demand incentive.
The second essay investigates whether public health facilities in a community are affected by the health card program. The mechanism through which this can occur is the program's weak means of compensating existing public health workers, who are paid on a salary basis, for the increased workload and other undesirable changes in the work environment that the program might cause. The outcomes of interest include infrastructure conditions and the availability of full-time health personnel in these facilities. Focus is placed on health centres that provide outpatient care and have no waiting list to ration demand. Identification of program effects is achieved through variations over time and in the intensity of health card distribution across communities. Using a regression DID method, the study finds some evidence that the program affects the availability of full-time GPs, and has no significant effect on other outcomes.
The third essay uses discrete choice experiments to elicit women's preferences for cervical screening. In 2007, Australian women saw both a screening promotion campaign and a vaccination program (HPV vaccine). As screening and vaccination are both preventive measures, the two interventions may support one another, but the vaccination program may also counteract the screening campaign's effectiveness, for example, through the misconception that vaccination can substitute for screening. A mixed logit model is used to capture heterogeneity in observed behaviour for women with similar observed characteristics. This model also accommodates the panel structure of the data, allowing for a respondent providing multiple responses and for unproportional substitution patterns across screening test alternatives. The results indicate that there was an overall reduction in preference for screening, with indications that this trend is associated with HPV-related events. The interventions have a minor impact on how women value various screening attributes (for example, price, test accuracy), however. A simulation exercise is performed to assess the plausibility of several strategies to increase the screening rate.
© 2010 Meliyanni Johar
Essays on Political and Fiscal Decentralization
Riatu M. Qibthiyyah ([email protected])
Accepted 2008, Georgia State University, Atlanta GA
Indonesia as a developing country in transition is a useful case study on how government might response to institutional change. The 1999 decentralisation policy in Indonesia introduced a more flexible procedure for creating new jurisdictions. At the same time, major changes occurred in the central–local fiscal system, and specifically in the intergovernmental transfer system. Implemented in 2001, Indonesian decentralisation policies have been characterised by rapid formation of new jurisdictions across the various levels of sub-national government. The objective of this dissertation is to examine the effect of intergovernmental transfers on the proliferation of political jurisdictions at the local level in Indonesia, and to investigate the impact of proliferation on outcomes in education and health, the two sectors that have been decentralised most extensively to local government level.
The first essay asks what determines local government proliferation and what is the impact of intergovernmental transfers on proliferation. In exploring the determinants of proliferation, the study provides a more elaborate empirical technique than exists in the literature, by employing panel binary outcomes, survival regressions and count analysis to capture the effect of variations in the level of intergovernmental transfers over time.
On the determinants of local government formation, the findings show that there are probably competing effects across fiscal transfer types on the decision to proliferate and on the extent of fragmentation. The results indicate that: (1) the lump-sum conditional grants positively influence the probability of proliferation; (2) a higher median share of equalisation grants in a province is associated with a higher number of local governments; (3) higher equalisation grants are associated with a longer duration to the proliferation event; and (4) higher tax sharing in proliferated local governments is associated with higher stability (a longer duration to a subsequent or sequential proliferation event). The findings suggest that tactical central–local behaviour may be present, but support for the hypothesis that proliferation is driven by rent-seeking should not be generalised to the overall system of transfers.
The second essay examines the impact of proliferation on service delivery outcomes, specifically in education and health. I apply panel difference-in-difference (DID) estimation and differentiate the units of observation into new and original local governments. The findings show that the impact of proliferation on education and health outcomes is not uniform.
The education estimations show support for the existence of improved education outcomes in new local governments as represented by a reduction in the drop-out rate but not by higher student test scores. This is despite relatively higher conditional grants being allocated to proliferated local governments. The findings may imply that the benefits of proliferation tend to cater to local preferences rather than to mandated national programs. Meanwhile, on the estimation of health outcomes, the study finds evidence of improvement in infant mortality only in originating local governments and not in new ones. Controlling for selectivity and production function covariates did not change the pattern of the impact.
© 2010 Riatu M. Qibthiyyah