Thursday, September 14, 2006

The Return of Dr. Ray Fair

Some of you may remember Dr. Ray Fair, the Yale Economist who believed he had a formula to predict the share of the Popular Vote that a certain candidate (or party) would collect in an upcoming Presidential election. Dr. Fair’s model was based on key economic data, and some people found his formula a valid indicator. I am one of those people who find Dr. Fair’s work valid, although I doubt the likelihood of predicting actual shares of the Popular Vote to a fine degree. Dr. Fair’s system appeared to work; his 1978 paper produced results within 1.25% of President Carter’s actual results in the 1980 election.

In 1988 Dr. Fair evaluated the 1984 election, claiming accuracy within 1.7% of the actual result, and made his first forward prediction, promising that (depending on the rate of Inflation and GDP growth), Governor Michael Dukakis would take between 42.7% and 57.4% of the popular vote. Looking at the actual rates of Inflation (4.1%) and GDP growth (4.1%), Fair’s table predicted Dukakis would take 46.1% of the Popular Vote; in reality he took 45.7% of the Popular Vote, so Dr. Fair landed within 0.4% of his predicted target.

In 1990, pleased with his 1988 results, Dr. Fair prepared a paper (but did not publish) which, for 3.0% Inflation and 3.3% Real GDP growth, predicted Bill Clinton would claim 41.6% of the Popular Vote.

Dr. Fair saw those numbers and decided “Bush should have a fairly easy time winning re-election”. Dr. Fair failed to develop a factor for a third-party candidate, so while Clinton’s 43.0% of the Popular Vote made Fair only 1.4% off the target, Bush lost the election.

In June of 1996, Dr. Fair tried again, with an updated formula and a prediction which said that on the available data, President Clinton would claim 49.1% of the Popular Vote, which might mean a close loss. In the actual election, Clinton won, but his 49.2% result showed good results for Fair’s prediction.

In 2000, Dr. Fair released a series of updates, ultimately predicting in October that Vice President Al Gore would take 50.8% of the Popular Vote, but immediately warned that his formula had a 2.15 point margin of error, making the race too close to call. In the actual race, Gore only claimed 48.4% of the Popular Vote, so that Fair’s prediction was off by 2.4 points, outside his own margin of error.

2004 was therefore a natural point of interest for Fair-watchers like myself. Once again Dr. Fair issued a series of releases, with another release in late October 2004, which predicted President Bush would claim 57.7% of the Popular Vote. The actual results gave President Bush 51.6% of the Popular Vote, 6.1 points off Dr. Fair’s prediction, which led to a wry “post-mortem” by the good doctor.

These days Dr. Fair is – I think – retooling his formula yet again, but it seems there are certain lessons we can take from his work. First, obviously an exact predictor of elections is not yet within our grasp, for which I frankly am pleased. Man has a bad habit of fooling around trying to manipulate things, and if we understood what drove elections well enough to make it a pure science, I worry that the people would suffer for it. That said, I agree with Dr. Fair that the adage “people vote their pocketbooks” is true to enough of a degree that we can lay out the terrain for an election by watching the numbers. For a Republican, this is good news at the moment, with Inflation under 4% and Real GDP growth above 3%. It would be difficult indeed for a Democrat to win in 2008, unless the economy collapses, the Republicans nominated McCain or Chafee, or the Democrats nominated a mature, civil candidate who holds a working comprehension of National Security. Perhaps some combination of those possibilities would even be required. While I am thinking about the needs of the nation, let me remind Dr. Rice that as much as she would like to be Commissioner of the NFL, there is a serious need for her services as President of the USA.

But in his latest release, Dr. Fair has noticed something I knew all along, but has added his experience and wizardry to the matter; the fact that Presidential elections in the United States depend on the Electoral College, which in function means that one or a few states in almost every election make the difference in which candidate wins. In 2004 it was Ohio, and in 2000 it was Florida, for example. This is because there is a threshold in each state for a given candidate to win, and the states can be ranked by that demand threshold. Once you know that ranking, you can determine which state(s) will be the pivot point. The Republicans and Democrats have known this for many years, which is why the campaign money gets spent the way it does. What is interesting, is that there are sometimes subtle differences between which states the Democrats think will be pivot, and which states the Republicans think are pivotal. Dr. Fair has written another fascinating paper examining that process, which is well worth your time and effort to read.

While it is presumptuous for me to compare my thoughts to those of a Ivy League college professor, deference is not one of my strong suits, and I freely present my thoughts with the caveat to the reader that I am as yet unready to support my thesis with hard numbers, I would nonetheless suggest that while I agree with the demand ranking suggested by Dr. Fair, I have observed a certain set of tension points which restrict movement in party preference. For instance, while neither the Republicans or the Democrats have even begun their campaigns for the 2008 White House race, it would be safe to say that the Democrat will carry Massachusetts, and the Republican will carry Texas. The same can be said for an array of other states. Other states can be swayed, but begin in clear camps, like West Virginia or Michigan. Still others are very much in the middle, with no clear preference at this time, like Florida or Pennsylvania; they lean a bit and have history, but they are not at all locked into a position. I would compare these states, then, to balls on a number of adjoining pool tables; they can move about freely and appear to be headed for another table, but their impetus is not enough to change the table on which they sit – something genuinely unique would have to happen to move them to another table. This has happened before, of course – LBJ ran the table, basically, on Goldwater in 1964, Nixon did it to McGovern in 1972, and of course Reagan did it to Mondale in 1984. So I find it worth considering to ask, which states are on which tables, and what would it take to move them? I may post my conclusions on those questions if I can find evidence to back them up, but for now I submit them for consideration and discussion.

No comments: