Thursday, August 6, 2009

Obama's Polls and the Stock Market: Correlation vs. Causation

To celebrate the President's first 200 days in office, Quinnipiac released a new poll showing the President's job approval rating at a new low - 50%, down from 57% only a month earlier. The poll was just the latest in a brutal string of polls showing the President's approval rating approaching or surpassing the dreaded 50% line. And according to the RealClearPolitics average of polls, the President's approval ratings have steadily declined, while his "disapproval ratings" have steadily risen, since the Spring.

At the same time, the stock market is on fire. Reuters reports today that Goldman Sachs' Abby Joseph Cohen, chairwoman of the firm's investment policy committee, believes we're in a "new bull market" that probably began back in March. Many notable bulls, like Larry Kudlow, agree. Other experts are more cautious or are downright hostile to the bull market idea, but the indisputable fact remains: the Dow has been on tear since it cratered last winter.

A few partisans have suggested that it's actually Obama's plummeting approval ratings that have emboldened the stock market because the decline decreases the chances that his anti-market, high tax policies will be enacted. With this in mind, I've decided to put the "Obama vs. The Market" theory to the test. I gathered weekly data (every Monday since the inauguration and today) on the following:

1) The Dow Jones Industrial Average close;
2) The S&P 500 close;
3) The RealClearPolitics average spread between "approves" and "disapproves"; and
4) Rasmussen's "intensity index."

Now, because basic approval numbers alone don't move - or thwart - policy, I decided to look to the "spread" between general "approves" and "disapproves" and between "strongly approves" and "strongly disapproves" (the so-called "intensity index"). A lot of pollsters and pundits prefer this metric because politicians can be elected - and sadly they often are - with less than 50% approval if their opposition is non-existent. And intense constituent opposition - like that at the August healthcare town halls - can definitely undermine a policy, regardless of whether that policy engenders a lot lukewarm support. (I also see this a lot in trade policy with ambivalent free trade consumers against hardcore protectionist union members.)

I used the RCP average to smoothe out any outlier polls, but because I'm only interested in the trends, not the actual poll numbers, methodological differences between polls (like question phrasing, sample sizes or breakdowns) shouldn't matter anyway. I also used both the DJIA and the S&P 500 in case there was any difference between the two indexes.

Here are my results for the Dow, with trendlines in black (click to enlarge):

Hmm, the trendlines sure appear to show quite the inverse correlation. Now let's try the S&P, again with trendlines in black (click to enlarge):

Pretty similar results.

Next I used Excel to check the "strength" of the correlation between the poll spreads and the market indexes, starting from when the "bull rally" began in mid-March (Monday, March 9, in my dataset). Data are available upon request. Just leave your email in the comments, and I'll send you the whole excel file.

Here's what I found (with size of the correlation, according to this guide):

RCP v. DJIA = -0.719377196 (large negative)
RCP v. S&P = -0.702647419 (large negative)
Rasmussen v. DJIA = -0.567438678 (large negative)
Rasmussen v. S&P = -0.569744972 (large negative)

I fully understand the limitations of calling the correlations above "large" or "strong," especially with the limited dataset and volatility of the polling numbers. However, one must admit that the trendlines are very interesting, and the correlation is certainly there to a pretty significant degree.

Of course, the unanswerable - and far more important - question is whether Obama's imploding poll numbers caused the exploding stock market. I think it could contribute to the market rally (especially in health care and energy stocks), but that's far more a biased free-marketer's hunch than scientific fact.

But that won't stop me from showing off those great charts, of course.

OBVIOUS DISCLAIMER: I'm a lawyer, not a statistician. But I do know how to use Excel, and this is not rocket science.

No comments: