As some readers of this blog are likely aware, Fordham law professor Zephyr Teachout is challenging the Andrew Cuomo, New York’s incumbent governor, in the state’s Democratic primary, to be held tomorrow. One of her main campaign themes is corruption: Her campaign emphasizes corruption in the Cuomo administration both in the narrow sense of raising concerns about unethical and possibly unlawful conduct in New York state government (as well as Governor Cuomo’s controversial decision to disband the Moreland Commission, which had been looking into these issues), and also “corruption” in the broader sense of excessive influence of wealthy interests and the distorting effect this has on politics. Teachout herself concedes that if she wins it would be the “upset of the century,” and indeed most political prognosticators give her virtually no chance of winning. Why not?

It’s true, of course, that Teachout has no prior experience in electoral politics and is up against a savvy and well-funded incumbent. But there’s a bigger problem for her — and for any insurgent anticorruption candidate or party — that derives from the nature of the U.S. electoral system that Nobel Laureate Roger Myerson identified over two decades ago in a technical game theory paper on how electoral institutions affect the success or failure of insurgent anticorruption candidates. Although Myerson’s analysis does not correspond perfectly to the New York primary (for reasons I will explain in a moment), it is nonetheless enlightening–not only for the challenges faced by Teachout, but for anticorruption parties more generally.

Myerson asks the following question: If we assume (plausibly) that voters dislike corruption–regardless of their political ideology on other issues–then why don’t competitive elections drive corruption out of the system? If the incumbent is corrupt, then even if voters like the incumbent in other respects (such as her position on major public policy issues), one would think an insurgent candidate could adopt all those same policy positions, but promise not to tolerate corruption. Maybe those promises aren’t credible, or maybe the corrupt incumbent can manipulate the election — but suppose for the moment we assume away those problems. In a plurality, winner-take-all election, the voters still have the problem that they need to coordinate their votes, or else, by splitting their votes among the two candidates, they might hand the election to a third candidate who is even worse — perhaps one who is corrupt and has less appealing stances on the major policy issues. If voters could coordinate on the insurgent, everything would be fine, but if most voters anticipate that most other ideologically like-minded voters will support the corrupt incumbent, then it’s rational for each individual voter who supports the incumbent on policy grounds also to vote for thei incumbent, rather than “wasting” her vote on an anticorruption challenger. Things might look different in, say, a legislative election in a large multimember district that uses proportional representation (PR). In that setting, voters can rationally “defect” to the anticorruption party, because votes for that candidate are more likely to translate into actual seats; they’re not wasted. But in a first-past-the-post plurality voting system, rational anticorruption voters may not be able to coordinate on a shift of support to the anticorruption challenger, with the result that each such voter rationally supports the candidate whom she favors on policy grounds.

Now, although this analysis helps us understand the obstacles faced by an insurgent anticorruption party (or any other third party) in a first-past-the-post general election, it actually doesn’t apply directly to the New York state Democratic primary, because a vote for Teachout does not automatically hurt the Democratic candidate in the general gubernatorial election in November. Indeed, Meyerson’s analysis might even imply that rational progressive anticorruption voters should support Teachout in the primary (even if they would not do so if she were to try to run in the general election as a third-party candidate), because these voters can always vote for Governor Cuomo in the general election, should he win the primary. That’s plausible, and I think probably does explain why Teachout is getting as much support as she is. Nonetheless, I think there are two reasons why rational progressive anticorruption voters, who might prefer Teachout to Cuomo in the abstract, might still be tempted to vote for Cuomo–reasons that are consistent, though not perfectly congruent, with the spirit of Myerson’s analysis:

1. Assuming Cuomo wins the primary, then higher levels of support for Teachout in the primary might hurt him in the general. Though some Teachout supporters argue that a “protest” vote for Teachout will help shift Cuomo to the left, not everyone may be persuaded; many may instead conclude that a show of party unity would be more important heading into the general election.
2. If Teachout pulls off an upset win in the primary, then the chances that the Democrats will retain control of the governorship may drop precipitously, if centrist/conservative Democrats would either defect to the Republican candidate or stay home on election day.

To be clear, I don’t know enough about Cuomo, Teachout, or New York politics to have strong views about this particular race (though my sympathies usually lie with fellow law professors, and with those who make tackling corruption a high priority). Nor am arguing that Democratic anticorruption voters should vote for Cuomo rather than Teachout; the above discussion is meant to be predictive, rather than prescriptive. But I do think that the dilemma that Zehpyr Teachout faces — aside from all the issues related to name recognition, political experience, donor support, etc. — is one that poses a problem for anticorruption parties and candidates more generally.

It also connects directly to a question that commentators ask incessantly with respect to democracies in the developing world: How do corrupt politicians manage to get elected and re-elected, over and over, even though reams of evidence indicate that voters in the developing world don’t like corruption? One common, plausible answer is that voters feel like they don’t have any alternative. But that only raises the question, “Why not?”. Of course, sometimes we do see reasonably successful anticorruption parties in places like Eastern Europe and India. But it’s hard, particularly when–in addition to all the other hurdles faced by insurgent parties–the electoral system makes voters worry (rationally) that a vote for the anticorruption candidate would be “wasted,” and would only increase the chance that an even worse candidate prevails.

Of course, Myerson’s analysis might seem to suggest that electoral systems that use PR (with many candidates elected per district), and that are parliamentary (such that the legislature selects the executive), should have less corruption, because in that sort of system the legislative elections should allow anticorruption (or, more generally, non-corrupt) parties to compete effectively, and then these legislatures can in turn select the executive. As is often the case, however, things turn out to be a lot more complicated: The evidence on the anticorruption implications of PR vs. plurality electoral systems, and on parliamentarism vs. presidentialism, is decidedly mixed. (I’m trying to work through some of that material now, and if I may do a more general post on that topic in the coming months.)