This is a slightly updated post from yesterday (but the main results are relatively unchanged).
As of 5 PM ET on Monday, February 29, prediction markets move the fundamental model upward about 1.3 percentage points in expected vote share for each contested Electoral College election. I have run my presidential fundamental model (the underlying academic paper is joint work with Patrick Hummel) and posted the results on the PredictWise Presidential General Election forecast page. The key takeaway is that the fundamental model predicts a narrow victory for the generic Republican candidate over the generic Democratic candidate. This was done running the fundamental model, as published, with my best estimation of the 2016 variables that have yet to be determined. But, the fundamental model is the generic Republican candidate versus the generic Democratic candidate and there is a lot we know about the actual likely Republican and Democratic candidates that supersedes the fundamental model. You will notice on the same page I hold by my prediction that the Democratic candidate is about 64% likely to win the general election based on a mix of: prediction markets and fundamental model.
The difference between the PredictWise probability and fundamental probability is an upward shift in the mean expected vote share for any state, by 1.3 percentage points. I do this because the prediction markets tell me that the flip state, the state that divides a Democratic and Republican victory, is at 64% as of today. I am assuming that this shift is national, that the difference between the specific Democratic and Republican candidates versus the generic Democratic and Republican candidates is 1.3 percentage points. That is enough to flip three states and make three more in serious play for the Democrats.
The fundamental model predicts the winner for each of the 51 Electoral College elections. A quick look at the elements of the fundamental model shows variables that are either national, do not vary by state, and state-specific.
National Variables that do not vary by state.
1) Presidential Approval: This year’s presidential approval is pretty good for an incumbent candidate in his 8th year. Obama is at 48% on Huffington Post’s Pollster average. But, I have serious doubts about whether this term means the same thing in 2016 that it did in 2000 or 1984. Obama’s presidential approval has been remarkably flat, indicating that an increasingly partisan population is unwilling to express true approval or disapproval across party lines.
2) Incumbency: This is a binary term that is turned on because one of the parties has held the presidency for 8+ years at the time of the election. This goes against the Democrats hard, because historically the third term is hard to win. In the post-WWII era the Republicans lost in 1960, Democrats lost in 1968, Republicans lost in 1976, Republicans won in 1988 (then lost in 1992), Democrats lost in 2000, and Republicans won 2008. George HW Bush in 1988 is the only one of six tries that a party held for a third consecutive term.
Mixed Variables that vary on both the national and state-level.
3) Income: This term is both national and state-by-state. Technically the variable is changes in state-level income over the last few quarters. But, this term has a lot of national level variation that drives most of the movement (i.e., state income in any given year is highly correlated with national movement in state income). This variable is relatively neutral this year in that the growth in small, but steady.
4) Past Election: This is the last two presidential elections. I demean the state returns from the national returns, but it still favors the Democrats who have won the Electoral College relatively easily in the last two cycles. While the average state is more Republican, more states than necessary to win the Electoral College have higher returns for the Democrats than the Republicans.
5) Changes in State Legislature: This is the change in Democratic representation in the lower house of the state legislature in the last election. This was relatively neutral this year.