- The Big Picture - http://www.ritholtz.com/blog -

Causation & Correlation: Gasoline vs. Incumbency

Posted By Barry Ritholtz On October 15, 2004 @ 12:03 pm In Finance | Comments Disabled

With the debates now behind us, investors can expect to hear a spate of “suspect” theories from political analysts assessing each candidate’s chances for electoral victory. Most of these will be rife with illogical, poorly reasoned, partisan rhetoric. Voters can do what they want, but we strongly advise investors to ignore them.

For any hypothesis to be logically persuasive, it must demonstrate two key principles: Correlation & Causation. Correlation is a threshold question referring to a verifiable relationship between two elements. Once correlation is shown, the logician’s next task is to establish causation, which we define as the responsibility of a source factor for a related consequence. Only when a theorist can conclusively demonstrate both of these elements, do we believe their analysis is worth an investor’s time, attention and capital.

This election year has been rife with numerous unpersuasive theories. We previously dismissed the claim that equity performance corresponds to the challenger’s polling numbers. That thesis ultimately failed due to causation problems. During the most correlated period (March 2004), the same underlying factors that dragged down the incumbent’s approval ratings (Abu Ghraib scandal, Iraqi insurgency) also spooked investors. These external factors impacted both the election race and the market. While observers noticed the correlation, they incorrectly assumed causation existed also. They were incorrect.

Consider the more recent obsession over the Political Futures Exchanges. We find it foolhardy to rely upon these lagging indicators, as they are minute in size and lack a compelling reason to believe they have any forecasting value. In our opinion, it is roughly the equivalent of predicting the future of the U.S. economy based upon a few bulletin-board micro-cap stocks.

Elections are too complex, and exhibit too much randomness, to rely upon flawed theories lacking in causation. However, we do not wish to leave readers without a methodology for handicapping the campaign. We therefore draw your attention to the chart nearby [1] showing both correlation between gas prices and incumbent approval ratings. Strong causation also exists: Every voting age driver feels the weekly bite of higher fuel prices. This is not a theoretical abstract (such as the Federal deficit), or an emotionally vague notion (which candidate to have a beer with?). Rather, it is visceral, negative and immediate.

As gasoline prices go, so go the challenger’s chances.


Article printed from The Big Picture: http://www.ritholtz.com/blog

URL to article: http://www.ritholtz.com/blog/2004/10/causation-correlation-gasoline-vs-incumbency/

URLs in this post:

[1] chart nearby: http://bigpicture.typepad.com/comments/2004/10/chart_of_the_we_1.html

Copyright © 2008 The Big Picture. All rights reserved.