10/22/2010

Stock prices and the probability that Republicans will take control of Congress

Data is from January 20, 2009 to October 15, 2010
Variables used in Regression
ProbIntradeHouseRep - Intrade Prob of Republicans taking control of House
ProbIntradeSenatetotal - Intrade Prob of Republicans or Neither party taking control of Senate
stronglydisapprove - from Rasmussen Daily Tracking Survey
UnempRate - Unemployment Rate from BLS
cpi - Consumer Price Index
GDP - Gross Domestic Product
consumerconfidence - University of Michigan's consumer confidence survey





There is a lot of correlation between the Intrade Prob of Republicans taking control of House, the Intrade Prob of Republicans or Neither party taking control of Senate, and the percent of likely voters who strongly disapprove of Obama in the Rasmussen Daily Tracking Survey.

Note: There is a lot of correlation between ProbIntradeHouseRep, ProbIntradeSenatetotal, and stronglydisapprove.

Correlation between ProbIntradeHouseRep ProbIntradeSenatetotal is 0.7642
Correlation between ProbIntradeHouseRep stronglydisapprove is 0.7447

Because of that collinearity, I also just reran the political variables by themselves.

Labels:

2 Comments:

Blogger OGV said...

A small suggestion: regress Probability that GOP takes house on probability that GOP takes senate, and save the residuals. these should be orthogonal on the probability that GOP takes house and can be interpreted as the effect of GOP taking senate on stocks.

Uraa
O

P.S. love your work

10/23/2010 7:34 PM  
Blogger John Lott said...

Thanks for the comments, Uraa. That is nice of you. While the probabilities for the House and the Senate are highly correlated, there is obviously some amount of the variation in the House probability that isn't explained by the Senate probability. As you suggested, I reran the regression and the residuals do indeed continue to explain some of the variation in stock prices.

10/31/2010 4:51 PM  

Post a Comment

Links to this post:

Create a Link

<< Home