In a recent post, Matthew wrote about the New York gubernatorial Democratic primary between incumbent Andrew Cuomo and self-proclaimed anticorruption candidate Zephyr Teachout. He laid out several reasons why even progressive voters who care about combating corruption could rationally cast a vote for Cuomo over Teachout. Since Matthew’s post, primary polls have closed and, indeed, unofficial results show Cuomo taking the Democratic nomination in a landslide (though not as sweeping a landslide as expected). Matthew’s predictions have been borne out, but as this post will explain, likely for different reasons than those he posited.
Matthew predicted that progressive anticorruption voters would vote for Cuomo because they believed that (1) eking out a victory in the primary with relatively low levels of support could hurt Cuomo’s chances in the general election, or (2) a win by the leftist Teachout could greatly lower the chances of a Democratic win on election day by causing centrist Democrats to stay home or defect to the Republican candidate.
Both of these reasons are compelling in theory, but make less sense when applied to New York’s political landscape (which, in fairness, Matthew qualified they might). The first reason is plausible in races where the general is as or more competitive than the primary. However, in a state as overwhelmingly blue as New York, the tough race for most Democratic candidates is the primary, not the general. Low levels of support in the primary might bear on a candidate’s national aspirations, but less so on the general election. And while it’s true that Cuomo is regarded by some as too right-leaning on hot-button issues like income inequality and budgeting, this suggests more votes for Teachout, not more defections to Rob Astorino, Cuomo’s Republican challenger. In fact, one could easily imagine Matthew’s first argument running the other way: lower levels of support for Cuomo in the primary could serve to mobilize the Democratic base and increase party unity going into the general election.
The second reason fails to take into account the credibility of the Republican challenger’s threat. Astorino is a socially conservative Catholic radio broadcaster with even less name recognition than Teachout. Though he pulled off a surprising win in the 2009 race for Westchester County Executive, an August poll showed that 53% of voters had formed no opinion of him or had never heard of him. Current polls show him trailing Cuomo by a 28-40 point margin and at least $20 million in campaign donations. New Yorkers have elected one Republican governor (George Pataki) in the last 40 years. So while it’s true that a Teachout/Astorino race is a less certain Democratic win than a Cuomo/Astorino face-off, it certainly wouldn’t be a lost cause. In fact, for a voter who really cares about corruption, New York’s voting patterns and Astorino’s lackluster campaign suggest that this primary would have been a particularly safe space to vote for an anticorruption candidate.
So why didn’t they? Putting campaign funding and name recognition gaps aside for a minute, let’s consider a few other possibilities.
(1) As has been previously suggested on this blog, it’s unclear how much weight corruption actually carries in the voting booth, even in a state with major corruption issues such as New York. Related, but slightly different versions of this argument are that (a) the longer Albany maintains its corrupt status quo, the more likely voters are to perceive corruption as an inevitable part of the political game, and (b) partially as a result of (a), voters don’t believe that anticorruption candidates will be able to actually decrease corruption if elected. This means that even voters for whom corruption is the number one issue may be swayed to vote for the “more corrupt” candidate in hopes of having some say on less important but more tractable issues. Some studies support this line of thinking, finding that anticorruption candidates, once in office, are less effective at battling corruption than their campaign promises suggest. This was demonstrated by India’s anticorruption party, which promptly dropped its anticorruption agenda after being elected, and more locally, by Cuomo’s failed Moreland Commission.
(2) A second possible reason anticorruption voters didn’t vote for Teachout is the power of endorsements. In New York – particularly in primary elections where voter turnout tends to be low – endorsements from political parties, public-employee unions, news outlets, and other elected officials hold significant sway. In this primary race, the endorsements were a mixed bag. Teachout scored some significant endorsements (from the Public Employees Federation) and some important non-endorsements of Cuomo (from The New York Times). However, Cuomo got an important boost from the Working Families Party, an influential progressive New York mainstay that endorsed Cuomo even after recruiting Teachout to run. He was also endorsed by two major unions, the Transport Workers Union and 1199 SEIU, which are able to mobilize their large member bases to vote the party line. Lastly, Cuomo received the endorsement of current New York City mayor Bill de Blasio, the very elected official whose leftist progressive wake Teachout was hoping to ride into office. Elections heavily swayed by endorsements would not seem to bode well for anticorruption candidates and agendas. Endorsements usually come from large civic and political organizations, which have more to gain by supporting an incumbent with whom they have a well-established, potentially profitable relationship than by rocking the status quo boat.
GAB | The Global Anticorruption Blog 
This is a terrific analysis. On reflection, I think you’re (mostly) right and I was (largely) wrong, or at least overly simplistic. My biggest mistake was a somewhat careless overgeneralization of Myerson’s analysis; although I tried to acknowledge and adjust for the differences in electoral context between his model and the NY primary, I should have placed more emphasis on the fact that the primary system does allow anticorruption voters to cast a “protest” vote for Teachout without necessarily reducing the probability of a Democratic Party win in the general election (though I do think, if her candidacy had become more viable in the primary, concerns about her performance in the general might have loomed larger for some voters — even if she would have been favored, an increase in Astorino’s odds of winning from, say, 1% to 10% might have been a scary prospect for progressive New Yorkers). But you’re entirely right that I greatly overstated the concern that votes for Teachout might hurt Cuomo in the general, precisely because Teachout voters are overwhelmingly likely to turn out for Cuomo in the general. Indeed, if I’d been thinking clearly, that fact should have led me to predict that Teachout would actually get a fairly high percentage (though well short of a majority)–which is exactly what happened.
All that said, I do still think the coordination problem that Myerson identifies is still an issue for anticorruption voters in a context like this, and some of your proposed “alternative” explanations may in fact be understood as a manifestation of this problem. Take endorsements. Why did so many prominent individuals and organizations that might otherwise have favored Teachout endorse Cuomo? One reason they might have done so is that they expected Cuomo to win, and didn’t want to alienate him. (I suspect this is particularly true for Mayor de Blasio.) One of the reasons they expected him to win was that they expected other influential actors to endorse him. If they could have all collectively thrown their weight behind Teachout, it might have converted her from a symbol of dissatisfaction to a viable candidate who had a legitimate shot at winning. But the worse outcome for any of these endorsers might have been to endorse Teachout without a sufficient number of others doing so, alienating (or at least irritating) the current and likely future governor. And this does, I think point up a more general problem that anticorruption insurgents (or insurgent candidates of any stripe) face: especially in winner-take-all elections, even an insurgent that a majority of voters would prefer may have a hard time, because even sympathetic voters don’t want to shift their support unless a very large number of other voters do so as well.
Oh, one more thing: There’s a historical example from the developing world that seems to parallel the NY gubernatorial race fairly closely (but not perfectly): the 2006 Brazilian presidential elections.
Rather that using partisan primaries, Brazil uses (or used — I haven’t kept up to date with Brazilian politics) a run-off system for its elections: There’s a first round election, and if no one candidate gets a majority, there’s a second round that pits the top two candidates against one another.
In 2006, the incumbent President Luiz Inácio Lula da Silva (known as Lula) represented a left-leaning party and had been elected in 2002 with substantial support from labor unions and the rural poor. However, his administration was dogged with allegations of corruption (though Lula himself was not directly implicated). In the 2006 election, Lula faced a challenge not only from the right — Geraldo Alckmin — but also from two left-wing rivals (Heloisa Helena and Cristovam Buarque) who had broken with Lula and his party, in part over issues related to corruption, and in part over concerns that Lula was not sufficiently left-wing. (Sound familiar, NY voters?)
In the first round, surveys showed that “anticorruption voters” (voters who identified corruption as one of the most important problems facing the country) were statistically more likely than other voters to support Helena or Cristovam in the first round. In the second round, however, these anticorruption voters were more likely to support Lula over Alckmin in the second round (after Helena and Cristovam had been eliminated), and Lula won the election handily.
Of course, this is not quite the same as this year’s context for New York governor, because Brazil uses a two-round runoff system. But there do seem to be striking similarities that are consistent with your account of New York: left-wing anticorruption voters support a left-wing anticorruption candidate early, but once that candidate is eliminated, they shift their support to the incumbent left-wing politician, notwithstanding concerns about corruption.
(The source for the above discussion of the 2006 Brazilian elections, by the way, is Lucio Renno’s chapter, “Corruption and Voting,” in Power & Taylor eds., Corruption and Democracy in Brazil. Alas, I don’t think a version is available online.)