The transition to direct mayoral elections in clientelistic environments: Causal public spending and service delivery effects

We examine the impact of the transition to direct mayoral elections on district spending and household public service access in Indonesia during a period of momentous national democratic reform. We leverage the arguably exogenous timing of direct local elections to specify a staggered difference-in-differences model, which we estimate using the latest methods to plausibly identify causal effects. We find that the transition to direct elections led to a consistent and large decline in capital spending in both pre- and post-election years. We also determine that the transition resulted in a moderate decrease in household service access in the post-election period. Pre-election capital spending impacts are a function of both general disruptions associated with the transition and emerging clientelism. Service access effects are completely explained by the relative extent of clientelism across districts. We conclude that the local democratic transition in Indonesia had a mostly negative impact on key spending and service outcomes, at least in the short-run and for those districts in which clientelistic practices were especially pronounced. • We study the impact of direct mayoral elections in Indonesia. • We find that direct elections led to a large decrease in local capital spending. • We find that direct elections led to a small decrease in service access. • We demonstrate that clientelism explains most important effects.