Association for Public Policy Analysis & Management

Election fraud is considered pervasive throughout developing countries, raising concerns it can facilitate corruption and inhibit economic growth by preventing voters from holding elected officials accountable. This paper uses data from the Philippines to examine whether reducing election fraud improves government performance. Despite the intuitive appeal of linking election fraud to government performance, there has been no evidence demonstrating the causal pathway. This is largely due to a lack of election fraud measures and settings in which there is demonstrable election fraud. This project demonstrates for the first time that reducing election fraud causes significant improvements in government performance.

To measure the type of government corruption and red tape that inhibits economic growth, I use data on building permit approvals in municipalities, since delays in granting approvals are often associated with requests for bribes. The annual data from the Philippine Statistics Authority cover the years 2006-2015. To measure election fraud, I examine municipal-level election returns during the period 2001-2013 from the Commission on Elections. I specifically examine the uniformity of the observed distribution of last digits of mayoral vote totals. Election fraud is indicated when particular digits occur more often than others. To identify historically high-fraud areas ex ante, I refer to a list from the Philippine National Police.

I exploit a switch to automated elections in 2010 that made election fraud much more difficult through the use of stronger ballot security measures, timely counting of ballots, and simultaneous transmission of votes to various servers. This enables me to use a difference-in-differences research design to identify the causal effect of election fraud on government performance. The idea is that if automated elections drive election fraud down regardless of a municipality’s historical level of election fraud, then historically high-fraud areas exogenously experience a greater reduction in election fraud than areas that were already low-fraud. Thus, if reducing election fraud improves government performance, outcomes should improve significantly more over time in the historically high-fraud areas relative to the historically low-fraud areas.

Digit-based tests indicate that automated elections significantly reduced election fraud. Estimates from the difference-in-differences research design indicate that the reduction in election fraud led to a sharp and sustained 20 percent increase in the number of building permits approved annually. This facilitates greater investment in the local economy. These findings have important implications for policy. First, I provide strong evidence that the automated election technology deployed in the Philippines reduced election fraud. In doing so, automated elections restored electoral accountability because candidates can no longer manipulate vote totals in their favor. Perhaps more importantly, results here demonstrate the positive effects of investing in credible elections on an important outcome. To the extent that results here generalize to other aspects of government performance, the findings indicate that reducing election fraud can result in meaningful differences in the type of government performance that directly affects economic growth.