DECENTRALISATION, ACCOUNTABILITV AND LOCAL GOVERNMENT PERFORMANCE IN INDONESIA
MetadataShow full item record
What makes local governance working effective after decentralisation? Why some local governments are achieve better performance than others in managing decentralised public services and resources? How can malfunction local governments be reformed in order to perform effective and efficient? This article contributes to our understanding of these overarching questions by exploring the linkage between local government accountability, corruption, and citizens’ participation. Departing from critiques on fiscal federalism theory we propose hypothesised pathways through which decentralisation can lead better local public service performance. The basic hypotheses underlying this research is that decentralisation will result in better local public service performance only if mechanisms for strengthen political accountability are established within local governments. Without strong political accountability decentralisation only creates powerful incentives for political and bureaucrat agent to capture local political process and misailocate public resources. We test these hypotheses against evidence from 155 newly empowered local governments in Indonesia. Governance Decentrallsation Survey Data (GDS) is used to examine the linkage between decentralisation, accountability and local government perfonnance in the countly. The results of simple and multilevel regression model broadly support the hypotheses. Less corruption, higher local government accountability and citizen’s dii political participation are all associated with better local government ab’ performance. In contrast, poorly performing local public services are often mc deeply rooted in their political and social contexts. Local governments often be: fail to jovide better public service when political accountability is absent (iS due to weak checks and balances, lack of transparency, and weak electoral eli incentives. These effects remain statistically robust across all regression specifications.