Association for Public Policy Analysis & Management

In this paper, we study voters’ response to government-provided monetary incentives for turnout in Peru along several margins, including turnout and registration, settlement of outstanding fines and information acquisition. Voting has been compulsory in Peru for many decades. Until 2006, the electoral abstention fine was homogeneous. Following that year's national election, a reform was implemented. Districts were classified into three categories according to their poverty level: non-poor, poor and extreme poor. The value of the fine was reduced in all districts, but the magnitude of the reduction varied depending on the category, with poorer districts enjoying a larger decrease. This reform provides plausibly exogenous variation in the value of the fine, which we combine with administrative data from national elections over a 16-year period.

Using a difference-in-difference strategy, we find that a larger fine for abstention leads to higher voter turnout. We estimate an average marginal increase of 0.5 percentage points for a 10 Peruvian Sol [S/] hike (approximately US$3), but this average masks important dimensions of heterogeneity. We observe that the marginal effect of the fine is more than three times as large in 2016 than in 2011, indicating gradual and substantial adaptation over time. We also observe that the effect of a larger fine is not uniform across different types of elections.

The observed increase in voter turnout in response to a larger fine can be decomposed into potentially separate effects on the decision to vote (the numerator) and the decision to register to vote (the denominator). Voter registration is automatic in Peru and people are legally required to vote in their district of residence, which they can adjust by changing the address on their national identification card (DNI). We show that a S/10 increase in the value of the fine leads to a 1% decrease in voter registration. We argue that voters are strategically changing their address of registration in order to avoid paying a larger fine. We provide further evidence of irregularities by showing that the rise in registration is only present for younger voters with ages between 18 and 20, who must obtain a DNI for the first time and can report a fraudulent address at a very low cost.

To gauge the contribution of the value of the fine to the aggregate effect of compulsory voting, we use rich data at the `voting-table' level and leverage idiosyncratic variation at this level in the age composition of the electorate in the 2016 elections. Our novel empirical strategy exploits the fact that citizens above the age of 70 are exempt from the mandate to vote, and so we compare turnout in voting tables with a higher share of 70 year-old voters to those with a higher share of people aged 69. A back-of-the-envelope calculation using our elasticity estimates for the same year yields that a 100% reduction to the monetary incentive would only explain 20% of the drop in turnout observed between ages 69 and 72.