Wednesday, May 07, 2008

Well, not quite THAT bad 

Paul Mirengoff says Obama is the favorite now, and by more than a little:

I consider Obama the favorite. One can usually predict the outcome of the general election, and come pretty close on the margin, by considering just a few variables: how the economy is doing, whether we're at war and how popular the war is, which party holds the White House and how long it has held it, and how popular the president is.

This year, these "fundamentals" point to a Democratic victory of at least 10 percentage points.

I actually know a thing or two about such models, having written a dissertation on political business cycles and a couple of papers regarding electoral behavior. A model many of us use as a reference is Ray Fair's Predicting Presidential Elections. It's written by an economist so it favors some of the same variables Paul is using, including measures for war. Fair assumes the war variable currently registers as a zero, so hang on there for a minute while we check everything else and then see whether or not the belief that America is in a war matters for the calculation.

Fair has a calculator you can use to test your own prediction. I plugged in 0.9% for GDP growth forecast for the first three quarters of this year, based on last week's announcement of a 0.6% first quarter growth rate, assuming zero for Q2 and 2.1% for Q3 (these are approximately the WSJ Economist survey data from April; it's possible some of the Q3 growth gets pulled into Q2 because the stimulus checks seemed to come out earlier than we thought, but for this calculation that's a wash.) Inflation over the second Bush administration has averaged 2.8% per year. I'm leaving it there as the baseline; we'll play with that in a bit. The last thing you need is a measure of "good news", or the number of quarters real per capita GDP growth was over 3.2% on an annual basis. I see three such quarters (2005q3, 2006q1 and 2007q3). Using those values, I get 48.61% as the Republican share of the two-party vote total. If you'll guess Barr, Nader and the other fringe candidates draw 1% of the total vote, that gives McCain 48.1% of the vote and Obama 50.9%, a difference of only 2.8%, not 10%.

Inflation doesn't matter too much to this, given that the equation calls for all 15 quarters to be factored in and 13 quarters are in the books. If I add an additional 1% to q2 and q3, it cuts McCain's vote share only by about 0.1%. Inflation this summer may make us feel crabby towards government, but I'm not inclined to believe voters will visit the sins of the Fed on McCain.

Could the war matter, and if so how much? Douglas Hibbs has long been the father of the "bread and peace model", and he reports a comfortable 6-8% Democrat margin. But his model only ascribes a loss of .75% to the cumulative fatalities in Iraq; the rest is his estimation of smaller impact of the Bush expansion than Fair's. Fair tested his model using shift variables for WW1, WW2, and Korea (not Vietnam, a story for some other time). I'm inclined to use that 3/4% adjustment from Hibbs.

I think thus that the margin is much less pessimistic than Paul has painted it. McCain probably starts, ceteris paribus, in a four-point hole, but not a ten. There's work to do, but given the unpredictability of the campaign so far, I wouldn't start heading for the exits just yet.

Labels: ,