Why is economics at once capable of being so dismal and yet so interesting? Because researchers, especially those working in for-profit private sector companies, get to work on interesting models designed to answer topical questions; but they also often over-engineer the toy created. This, occasionally inadvertent tactic, allows the creator to hedge his bets.

National City Corporation (NCC), a regional US bank worth keeping an investment eye on, has come up with this model (follow the link to "Election Prediction Model") for predicting the mid-west "battleground" states results. The standard error for some of the state results inhibits gamblers rushing out to bet Bush but the temptation is nonetheless large for the others. For example, Ohio, 20 Electoral College (EC) votes and one that when lost by the Republicans has always seen the Democrats win the White House, is shown going Bush. As is Michigan, with 17 EC votes.

The Ohio prediction is surprising given that the state has lost over 30,000 jobs over the last year to July. Democrats may be tempted on this basis to draw sustenance that NCC's model will be incorrect. There is comfort for them too in that Pennsylvania, the largest mid-west EC vote state with 21, is still shown as in the balance. However, Pennsylvania has added over 36,000 jobs since last year: its voting Kerry ought not to be assumed. In fact, given the last set of poor non-farm payroll numbers it is a shock that swing states such as Ohio and Pennsylvania are not already clearly leaning Democratic.

The 10,000 metre view is this: 270 EC votes are required to win. The Republicans appear at this writing to have 187 in the bag (ignoring all swing states mentioned in the piece); the Democrats have 179 with 172 to play for. (See the NY Times coverage page for more).

Should Bush carry two of either Ohio, Pennsylvania and Michigan - not outlandish if NCC are to be believed - he will have two key mid-west states and at least 37 EC votes. At this point the largest remaining swing state is Florida with 27 EC votes. Florida has added over 160,000 jobs since July 2003 to its well-diversified economy. A Bush governor and a Bush president - it is reasonable to put the 27 votes down in the republican column raising the President's tally to 251.

Do the remaining 19 EC votes required represent a mountain to climb? The five states most likely to go Bush in this situation appear to be West Virgina (5 EC votes), Missouri (11 EC votes), Colorado (9 EC votes), Arizona (10 EC votes) and Nevada (5 EC votes). All have improving job statistics and voted Bush last time - by comfortable majorities in the cases of W. Virginia, Colorado and Arizona. An incumbent president would count himself unlucky not to carry three of these five - and that, whatever the other results, would be enough to return Bush. On current trends.

But what does the gambling money reckon?

Currently IGsport.com are offering EC vote spreads on Kerry of 271-279. That is, a John Kerry presidency is predicted. Bush is quoted at 259-267. That looks a reasonable bet from this scribe's chair but not one to make you rich in a close election.

Potentially much more enriching (and a natural hedge if used with the Electoral College votes line) is the quote 50-54 for the Republicans to emerge as the party that will win the White House. That line is decided at 100 on a winner takes all basis. In other words, one risks 54 to win 100 or 1.85 to 1.

Bookmark and Share


Conclusion:
Bush to win.

Background:
This blog's preferred predictive model is based on the work of Margaret Fisk, author of the 1976 work The Gambler's Bible. With some tinkering to account for results post-1976 (thanks Jacques Black) what is clear is that the 2004 Presidential election is likely to be decided by three issues:
  1. Will the traditional Republican vote-getters of foreign policy and limited wars hold up?
  2. Will voters perceive the static unemployment rate in 2004 as a Bush failure?
  3. Will Kerry suffer from the Nader Effect?
Discussion:
For Kerry to win, it's an all or nothing task. If the model holds, Bush needs but one of the three issues to be resolved in his favour to win.

It is difficult to see Iraq providing votes for Bush. However, that's a non-American view influenced by the international press. It is not entirely obvious that wavering amongst early supporters of the war will translate into significant anti-war electoral opposition for Bush.

More complex is the economy. There is no question that the Bush administration applied great stimuli in 2003 to the economy. The effects have been mixed though positive, with the real beneficiaries being equity markets:

  • Real income per capita growth 2001 to 2003 is over 6.5%; and 2000 to 2003 (including Clinton's final year) is 8.4%
  • Stocks bounced over 20% in the calendar year 2003 but are treading water in 2004
  • Credit is still cheap (and the Federal Reserve signaled its intent to keep it that way)
  • Consumer spending is strong
  • On the other hand, job creation has been patchy, and it is the rate of unemployment that unsurprisingly correlates most with Electoral College votes won by incumbents in election years.

Developing that last point, unemployment has stood at 5.6% for every month this year bar one. It is well down from the June 2003 rate of 6.4%. Is it conceivable that US voters will punish the President for this, even if accompanied by a potentially skillfully waged "Two Americas" campaign? At worst the effect ought to be neutral but with no benefit for the Democrats.

Finally, the Nader effect. Only wild-eyed optimists would argue that this is anything but a potential vote-loser for the democrats. Can Nader be completely co-opted by them? Perhaps, but it is a net non vote-winning factor for Kerry even if accomplished: he'll only collect votes he ought to pick up anyway.

Bookmark and Share
Related Posts with Thumbnails