Thursday, June 7, 2012

Using Wisconsin to test the effects of campaign money

The main complaint I'm hearing from the left with regards to the Wisconsin gubernatorial recall election is that Governor Scott Walker "bought" the election. He outspent Tom Barrett 7 to 1, and if it wasn't for that huge spending advantage, Walker would have been recalled. Observers seem to be using the recall as a textbook example of the idea that money determines elections. Note this exchange yesterday on NPR between Audie Cornish and Mara Liasson:
CORNISH: Now, supporters for Gov. Walker, Republicans outspent Barrett and the Democrats by about 7 to 1. And a lot of the GOP money came from large, out-of-state donors - you were talking about PACs there. And any lessons here for November? 
LIASSON: Well, this is the biggest story in the campaign so far: Money matters. Republicans have it.
Is that the right lesson to draw? What this election gave us is a rare and precious thing: a gubernatorial rematch. Walker and Barrett faced each other less than two years ago. Walker beat Barrett by five points back then, after raising $11 million to Barrett's $6 million. That is, Walker raised 65% of the funds raised by the Republican and Democratic candidates that year and he won 53% of the two-party vote. This week, Walker raised about 88% of the funds raised by the two candidates and he won -- wait for it -- 54% of the two-party vote.

So there's your money effect, folks. Go from a 2:1 money advantage to a 7:1 money advantage, and it could increase your vote share by a full percentage point! Woo hoo!

I don't mean to sound snide, but I'd say in general that if you pair the same candidates up against each other for the same office, you'll probably get similar results. And I'd say that the real lesson here is how little the electoral results changed after a vast change in financing. That is, the biggest story here is that money didn't matter all that much.


Update 1: Levitt (1994) used re-matches in congressional races to calculate the effects of spending on elections. The logic was that if you're just looking at the same candidates, you remove all the candidate-specific attributes of the race (notably, candidate quality) from the equation. He found that each additional $100,000 (in 1990 dollars) in challenger spending produced roughly 0.3 additional percentage points of the vote. Additional spending by incumbents produced no detectable effect.

If we apply that to this race, Barrett's spending fell by $2 million between 2010 and 2012. That's roughly $1.1 million in 1990 dollars. Following Levitt's findings, Barrett's vote share should have dropped by about 3 percentage points. In fact, his vote share only dropped by one point, suggesting that Barrett actually overperformed (at least by these two-decade old standards).
(h/t Brendan Nyhan)


Update 2: This post seems responsible for some degree of polarization, so allow me to revise and extend my remarks. I certainly wasn't asserting that nothing happened between the 2010 and 2012 elections in Wisconsin. Obviously, a great deal did. But it's not obvious whether those intervening events should have made things better or worse for Walker on balance. The union standoff probably hurt him, but that was a while ago, and the economy has improved there recently, which has probably helped him. The existence of the recall surely galvanized Walker's supporters as well as his opponents. Yes, Walker's approval ratings were in the low 40s last year and they've improved, but are we to believe this is all because of campaign spending?

My main point was that if you're going to claim that Walker bought the election or that his spending binge distorted the vote, you need to give us an idea of what his vote share should have been if not for all the spending. I don't really know what it should have been, but his vote share from less than two years ago against the same guy (plus an incumbency advantage?) struck me as a good place to start.

14 comments:

  1. PPP polling in February showed Walked with higher unfavorable numbers than favorables, and a potential rematch between Barrett and Walker was polling dead even. So the final result may have been the same as 2010, but isn't it worth noting that the support numbers didn't remain steady over those two years? Money surely had to have influenced enough voters since then, I would think.

    ReplyDelete
    Replies
    1. Well, that's the problem with measuring the effect of money on elections. There are always other factors.

      As this post points out, the Walker rematch really is about as close to a controlled experiment as you get with elections.

      Delete
    2. Yeah, there are lots of intervening factors. The union bill standoff may have hurt Walker on balance, but it's been a while, and the Wisconsin economy has improved, etc. But I figure if we're going to talk about how many votes Walker "should" have gotten if not for all the money, the 11/10 results are a pretty good baseline.

      Delete
    3. I strongly disagree. The very fact that something as rare as a recall election happened in the first place is strong evidence of a considerable difference between 11/10 and now. Simply dismissing "other factors" is insufficient, and justification needs to be provided for using a baseline for Wisconsin that's 2 years out of date.

      Delete
  2. Wow, so a 'serious' researcher completely fails to understand that there's a difference in effect between a few million and closer to FOURTY million.

    I thought this was a serious blog, discussing serious subjects. I guess not. That's too bad.

    ReplyDelete
  3. "So there's your money effect, folks. Go from a 2:1 money advantage to a 7:1 money advantage, and it could increase your vote share by a full percentage point! Woo hoo!"

    Thats nonsense. You pretend as though nothing happened between those two races. Your analysis has no scientific or logical basis. This type of reasoning wouldn't even apply to a tennis match, let alone an election! It's a re-match only to the extent that the two names on the ballot were the same. It took place in a different world at a different time from the previous election.

    Also, it is not equally likely that a "liberal billionaire" will spend as much money trying to preserve collective bargaining rights of public sector unions, as compared to a conservative billionaire. That you seem to consider the hypothetical "liberal billionaire" suggests that you see the workings of money in extremely simplistic ways. If you were investing a 100 dollars, would you be equally likely to invest it in the following two places? (1) A place where a 5% return was guaranteed (2) A place where a 15% return was guaranteed?

    Money favors right wing politics and right wing ideologies in general - which are more sympathetic to exclusion, concentrated private wealth and power than left wing ideological positions which are inclusive and less sympathetic to concentrated private wealth and power. When limits of the use of concentrated wealth are removed, it is not surprising that this favors the creation of a world that is friendlier towards the concentration of wealth.

    This is all fairly elementary and not controversial at all, but given the quality of your argument, it is worth restating here.

    ReplyDelete
    Replies
    1. Kartikeya, please see update 2 above.

      Also, I certainly didn't assert that money spent by liberal billionaires would have the same effect for the left that money spent by conservative billionaires would have for the right. I just don't think campaign spending has all that huge an impact on these kind of elections, and a lot of studies back that up. I'd note that Obama massively out-spent McCain four years ago, and he ended up getting just about the exact same percentage of the vote that all the economic forecast models predicted.

      Delete
  4. I think the real role the money played in this election was to reframe the debate in the public's mind as to whether the recall was an abuse of the electoral process. In that vein, the money was very influential. The exit polls indicate that nearly 70% of voters thought the recall was illegitimate (I agree), including among many who opposed Walker's policies, as I do.

    ReplyDelete
    Replies
    1. That may well be what happened. I'm always a bit suspicious, though, of voters' claims of why they voted the way they did. People may have voted for Walker because they opposed the concept of the recall, or they may have just been voting for Walker (because of partisanship, economic retrospection, or whatever) and rationalized their choice as an opposition to the concept of the recall.

      Delete
    2. Seth, the logic of partisanship does not match up with exit polls showing the same voters greatly favored Obama. Rob's point is right on. Targeting the mechanism of recall is exactly how you win a high-turnout election when partisanship is not on your side.

      It seems much more plausible to infer that (1) the political damage Walker's surprise actions caused took a bunch of money to fix than (2) the money had little to no effect. Walker funders are not idiots, they would not have put up so much money if they did not know it would matter. It seems the real story here is that money over-performed, giving Walker a wider margine than he needed.

      On election night, Democrats were ecstatic with the high turnout, thousands of new voters registered (new voters in WI almost always vote Dem), and indeed those voters favored Obama by a wide margine. What the Dems did not count on was how effective Walker's adds had been on independents and apparently even some Dems (if you dig in to the county-level numbers) in re-framing the election away from policy in a pretty blue state.

      Delete
  5. Wouldn't the Levitt study say that if Barrett had received out of state funding equivalent to that of Walker's and raised $28 million (Using your 7:1 ratio and the $4 million you state) that he would have gained 36 points? (a difference of $22 million from 2010 equivalent to $12 million in 1990 money / 100,000 * .3)

    ReplyDelete
    Replies
    1. This is one of the areas where the Levitt study breaks down. Campaign spending has increased much faster than inflation over the past few decades, and I seriously doubt that $100,000 carries the same bang that it did 20 years ago, even in constant dollars. Also, Levitt didn't account for outside spending (just not as much of an issue in 1994). If outside expenditures had the same effect that candidate expenditures did, then yes, your interpretation would be right. But it's quite possible that they have very different effects. Seems like we could use some new research here!

      Delete