Panel Paper:
The Policy Consequences of Election Timing
*Names in bold indicate Presenter
In this paper, we combine a credible research design to isolate the causal effect of timing on election outcomes with micro-targeting voter data, which allow us to characterize the composition of voters participating in each election. In particular, we examine school bond and tax referenda in California, Michigan, Ohio and Texas from 2000 to 2014. We focus on these types of ballot measures because they concern policies over which the relevant interest groups (e.g., public school employees) and conservative voters are expected to have divergent preferences. During this period, we observe numerous referenda put on the ballot by the same school districts, allowing us to examine how election outcomes differ within districts over time, depending on the date their measures appear on the ballot.
Our preliminary analysis shows that bond and tax measures are more likely to pass during on-cycle, high-turnout elections. Turning to the micro-targeting data, we find that this occurs because education employees never make up a large share of the electorate, even during low-turnout, off-cycle election. By contrast, the overall partisan and ideological composition of the electorate changes substantially between elections, resulting in a much more conservative mix of voters on off-cycle election dates, which reduces the probability of passage for measures appearing on the ballot during these elections.