Peter Levine on Super PAC game theory

Here’s the post: Game theory and the Super PACs. Levine points to the recent shift in campaign finance focus from the presidency to Congress, and adds nuance to a debate that is frequently overrun by absolutist intuitions:

No wonder Karl Rove is spending his money on behalf of Senate Republicans. The Center for Responsive Politics reports that conservative super-PACS were spending $10 million/week on behalf of Mitt Romney until a few weeks ago, but they are down to just $2.07 million in the last week.  CRP also calculates that Restore Our Future has spent $84 million on congressional races, American crossroads has spent $34 million, and Americans for prosperity has spent $31 million.  Meanwhile, an industry like financial services (including real estate and insurance) demonstrates how to distribute your cash if you are mainly concerned about your own after-tax profits plus mollifying the winner. They’ve given $221 million to Republicans, of which only $29 million had gone to Romney. They have also given $116 million to Democrats, including an ingratiating $12 million to Obama.

I’d only add two points. First, if there’s any influence to be bought in an election, it’s likely that it is in the form of early money. So even though I’ve taken this article as evidence of the rent-seeking hypothesis of campaign finance (that the money follows the winner rather than determining the winner) I also think that politicans and funders have long understood the wisdom built into the acronym of EMILY’s List: “Early Money is Like Yeast; It Makes the Dough Grow.” Another way game theory impacts campaign finance is through signaling theory: large donors will want to signal their loyalty to a party and a candidate early and in a way that involves real sacrifice if they’re to have any hope of having real influence. So when we think about the intersection of affluence and influence, or when we evaluate the effects of Citizen’s United, we might want to ignore the massive gross receipts and focus on how early donors leverage their crucial support. Money given during moments of uncertainty can come with many more strings than money given to a candidate who thinks she’s going to win. This would also be more evidence that financing limits are less effective than we might hope: the real benefits of campaign finance would be from throwing in relatively cheap support early.

Here’s the second thing. I’ve been brooding for a while on exactly what the current evidence suggests money can do in an election. For instance, there’s a lot of conflicting evidence on negative campaigning, but on balance negative ads look likely to expand voter turnout rather than discourage voters. Consider the Levitt paper using repeat challengers for the House of Representatives. Levitt’s data showed that it would cost about $100,000 to move the electorate 0.3% in 1994. Since the national election is likely to be won by about a percentage point, that’s actually a pretty scary number; but since it’s a national election, we might want to discount the cost of moving the electorate simply because of the size of the target. You can’t run this kind of regression on a presidential race, unfortunately, because it requires repeat challengers, but extrapolating that data to figure out what it costs to move the electorate in a national race, I’ve made a quick guess that it would be something like 435x as much (for each House district): $43.5 million dollars per 0.3% change in 1994. To account for inflation, we would add 54% to get the contemporary numbers: $67 million. I’d guess at the national level it will be much more expensive, but it also might be cheaper if you only focused on close districts and take maximum advantage of the electoral college.

The Super PAC expenditure difference between Democratic and Republican PACs was just a little under that: $60 million. So on one scenario, we’d expect Romney to do 1/3 of a percent better than he would have done under conditions of more equal spending. Of course, if there are key districts, then the Super PACs could devote all their resources just to those, and they wouldn’t have to deflate their funding quite so broadly. This is what Levine suggests they have done. Yet 538 lists the Democratic likelihood of keeping the Senate at 79.9%, twice what it was at the end of August! That’s not definitive proof against the power of money in statewide races, of course. If this bounce was always coming, then the Super PACs may have blunted it: perhaps the Democrats would be looking at a likely 5 seat majority instead. But it does lend additional support to the claim that it’s not the finance totals that are doing the majority of the work.

In my view, the Republicans simply can’t generate enough money to buy this election. At the same time, I am quite sure that the financial services industry has spent enough money, in the right ways, to prevent rigorous and sensible regulation of their industry. Hmmm… there’s a thought: if Republican donors really want to win, they’d be better off skipping the Super PACs and spending their money to short the stock market to create a massive crash this October. But I’m betting they won’t.

Postscript-Levine’s two posts before this one were also pretty great. Levine is very often interesting, but he’s been on a roll again. (We’re friends on Facebook, but “liking” these doesn’t quite give them the appreciation they deserve.) :

  • Not a post, per se, but a New York Times article on non-college youth: Struggling Young Adults Pose a Challenge for CampaignsCIRCLE’s report was the impetus for the article, and Levine is quoted: “Extensive research shows that if you ask young people to volunteer or vote, they respond at high rates.” 60% of American young people will attempt college in some form, but only about half of them will attain a bachelors degree, so there’s good reason to worry that civic engagement is heavily correlated with educational attainment. What can we do to correct that trend? (CIRCLE has some ideas.)
  • Ideology in the Chicago Teacher’s Strike is a pretty phenomenal fact-checking on some of the broader ideological analyses of the CTU strike:
I am basically on the teachers’ side, but that is because I share many of their substantive views of testing, funding, and the curriculum. I do not find it helpful to describe them as progressive and the mayor as neoliberal and to read the strike as a showdown between those two movements. The questions should be taken one at a time: How should we assess teachers? How long should the school day be? How much do we need to spend per student? And how is the available money being allocated?






2 responses to “Peter Levine on Super PAC game theory”

  1. peterlevine (@peterlevine) Avatar

    Josh, you’re too kind, especially since my post didn’t really honor its promise to use game theory. I just gestured at the idea that game theory might be relevant.

    Relevant to your estimates of the cost of moving elections are some experiments conducted by Don Green and colleagues. They found that for about $9/head, you could convert non-voters into voters. Those are results for in-person, phone, and mail canvassing–not TV or radio advertising. But they suggest that you could get about 11,000 extra votes in a congressional district for $100,000. Considering that a winning candidate in a contested race can easily draw less than 100,000 voters, that’s a pretty big impact.

    1. Joshua Miller Avatar

      Well, I did mean to chastise you for not developing a payoff matrix…. 🙂

      The Don Green study you mention (a book!) sounds very relevant, but isn’t that strategy already exhausted, at least for partisan purposes? I’ll need to read the book to see how elastic the supply of convertable non-voters is, but $9/head isn’t going to last forever without diminishing returns: the first non-voter you convert is going to be a lot cheaper than the last. If a half-page newspaper ad produces 473 votes (at $5 a vote!) will a full-page ad produce 946 votes? That can’t be right… it’s the same newspaper, and some people don’t read any newspaper at all. (They might get radio ads… but again, diminishing returns, right?)

      In terms of civic engagement, there’s plenty of room to boost voter turnout, so I’m with you in spirit. But I suspect this doesn’t militate for (or against) any particular campaign finance scheme.

      It’s a little bit separate from the Citizen’s United discussion we’ve been having, since Get Out The Vote drives have never been regulated as campaign finance. The BCRA overturned by Citizens United didn’t prevent non-partisan GOTV (though the Levin Amendment did restrict party-directed GOTV) so we did and still do treat corporate funding for GOTV as an acceptable form of corporate citizenship.

      Still, it could be relevant since GOTV targeted at particular demographics known to vote in partisan ways has always been common. That’s part of why Republicans attacked ACORN, right? In practice, their GOTV efforts were disproportionately helping Democrats. Yet this was true even under BCRA.

      In Campaign Finance Reform terms, I suspect that militates in favor of further-restricting GOTV, which in civic engagement terms would be perverse. There’s something to be said for competitive GOTV: Republicans target their likely voters, Democrats target their likely voters, and the state and other non-partisan groups (League of Women Voters, etc.) target the likely non-voters who the partisans don’t reach. Together, they *should* be able to energize just about anybody who is persuadable to get out and vote.

      Something tells me that a lot of the objections I’ve just laid out are cogently dispatched in Green’s book, though, so I’m ready to be persuaded I’m wrong!

Second Opinions