While the United States does have its problems with democracy, one area where we shine is direct democracy. Rare at the federal level, at the state and local level direct democracy is quite common in the US, much more so than most other democracies (Switzerland also stands out). Almost half the states have some form of citizen initiative or referendum process, and it is used frequently in most of those states. But even more direct democracy takes place at the local level.
And much of that direct democracy at the local level takes place through what are called special elections. I’m not talking about elections to fill unexpected vacancies in office — though of course those do happen. I’m talking about actual voting on issues. Many of these issues revolve around questions of public finance: whether to raise a local sales tax, to approve a property tax millage, or to issue bonds for a capital project.
One very relevant example for me is an upcoming special election in my city of Conway, Arkansas. Citizens are being asked to approve the issuing of bonds to construct a community center, pool, soccer fields, and some other amenities. The bonds would be secured by a tax on restaurants. The tax already exists — city councils can put these in place without a public vote. But to issue bonds, the citizens must be asked. I wrote an op-ed about it in my local paper (if that is gated, try this blog post).
The key is that this is a special election. There are no other issues on the ballot. It takes place on February 8th, not a date that probably stands out in voters minds as an election date. What will this special election mean for voter turnout? A lot of academic research, including a paper that I wrote (currently under review, but summarized here), finds clear evidence that voter turnout will be much lower. Will the result be different? Again, a lot of evidence suggests yes. For example, property tax elections in Louisiana were less likely to pass with higher turnout, and less likely to pass in a general election (my research finds a similar result for sales tax elections in Arkansas).
But why are tax increases less likely to pass in special elections? On this question there are many theories, but they are hard to test. Is it because different kinds of voters show up at special elections, representing a different sample of the population? Possibly, but evidence is hard to find.
A new paper just published in the American Political Science Review sheds some light on these questions.
In their paper “Who Votes: City Election Timing and Voter Composition,” authors Hajnal, Kogan, and Markarian find they very different people are showing up at special elections in California. In short, voters at special elections are whiter, older, richer, and more conservative. When the same issues are voted on during general elections, the share of Latinos and Asians, young voters, poor voters, and liberal voters increases. How exactly these changes will affect the outcome of any particular special election is not clear, but the median voter at a special election is clearly different than one at a general election.
This question was challenging to answer in the previous literature because it is often hard to find out who voted at a special election. California, like most states, does have a voter file that is public information, but the demographics you can get from this file is limited. The authors combined this data with data from a private vendor that collects micro-data on voters to help candidates win elections (in this case, a firm that helps progressive candidates, but that shouldn’t bias the results — what matters is if the data is accurate). Big Data to the rescue! Through these sources, the authors are able to build a much richer database of the demographics of voters that voted in special and general elections.
What is the implication of this research? Does it mean that states should not allow local special elections? This depends on what a state is trying to accomplish. It’s still possible, also hypothesized in the literature, that the voters who show up for a special election are better informed. Certainly someone who is aware of the special election is paying better attention to politics. Maybe this makes them a better voter in some way. Unfortunately we don’t have much evidence either way — the data in this paper has no questions about voter knowledge.
What the paper does tell us is that voting on issues at special elections does give you an electorate that is not only smaller, it is less representative of the overall electorate. Without some of other compelling reason to hold it as a special election (such as an emergency or to fill a vacancy), it seems that holding voting on these issues during a general election is more democratic (it is cheaper too!).
Some states have indeed recognized this. California, the state studied in the paper, recently passed a law which requires local governments to vote on issues during general elections if turnout at special elections falls below a certain threshold. Arkansas also recently changed the rules about special local elections, though they are now essentially limited to four fixed dates per year — including the second Tuesday in February, which is when our special election will be held (there were bills proposed that would have limited the dates even more in Arkansas, and you can see me presenting my research at the blog post linked above to a legislative committee).
Special elections are indeed special.
Economist Writing Every Day 