Hunting Alpha in Washington

In August, researchers from Harvard and Darthmouth published a very notable paper which studied the impact of federal legislation on U.S. corporations over the 20 years from 1989 to 2008. They found that by studying the deviations in voting behavior between ‘interested legislators’ and ‘uninterested legislators’ (these terms are defined below), signals could be derived about the impact of a given bill on the industry which the legislation implicated. Back-testing a simple strategy of going long (short) industries for which positive (negative) legislation passed in the previous month, generated excess returns of 76 basis points per month. The authors also examined various refinements of their strategy, which generated alphas as high as 200 basis points per month. Interestingly, the majority of these returns were earned in the short portfolio. This article will discuss the methodology, results and implications of this study.

The authors’ process was relatively simple – unfortunately describing it succinctly is not. To begin all firms which are headquartered in a given state (“represented industries”) are identified and assigned to one of 49 industry groupings. For each state, the authors then rank the represented industries according to the size of the entire industry, and define “important industries” as the three largest represented industries in each state (for example, the important industries for the state of Arkansas are the 3 largest industries by aggregated sales out of all industries with at least one firm headquartered in Arkansas).

Once the important industries have been identified for each state, the authors then analyze the text of each individual bill that was passed into law over the 20 year sample period and in order to assign them to one or more of the 49 industries (“affected industries”). Given that much of Washington’s legislation does not directly impact on business, only 20% of bills were assigned to an industry. Comparing the list of important industries for each state to the list of affected industries for each bill allows the authors to identify “interested legislators” for each bill: those legislators who represent states which domicile important industries which stand to be affected by the bill in question (for example, a Senator from Alaska would be an interested legislator for a bill which affected the oil and gas industry, whereas representatives from states which are not home to any oil and gas companies would be identified as “uninterested’ legislators”) . The difference in voting behavior between interested and uninterested legislators is used to determine whether a bill is positive or negative for the industries affected by the bill (i.e. a bill which received a higher proportion of yea votes from interested legislators than uninterested legislators was assumed to have a positive impact on the bill’s affected industries, and vice versa).

At the beginning of each month, the authors add a market-cap weighted portfolio of the firms in the affected industry group(s) to a long-short portfolio – long if a positive bill was passed, and short if a negative bill was passed. This portfolio is rebalanced monthly and was found to earn CAPM alphas of 76 basis points per month (controlling for other risk factors increases the alpha to this strategy as much as 16 additional basis points). Interestingly, the excess returns are driven by the short side of the portfolio – in the base case, 71 of the 76 basis points of alpha are generated by the short portfolio.

The authors then proceed to investigate other investment strategies in an attempt to improve their results, including: changing the timing and holding period of their investments; changing the definition of interested legislators; and changing the definition of affected industries. One such iteration increased alpha to 201 basis points per month! This best-in-class strategy included establishing positions in companies which are (1) members of the “most” affected industry for a given bill, and (2) actually operate in one of the interested legislators’ home states (other strategies tabulated returns to the entire industry, rather than individual firms). It is notable that throughout all of the iterations of the investment strategy, excess returns continued to be driven by the short side of the portfolio.

Although the authors are not this explicit, this commonality implies that the actionable strategy in this paper is to short affected industries where interested legislators are opposed to a bill that is passed into law. This conclusion makes intuitive sense: it would be very difficult for a small number of interested legislators to rally enough support to pass a bill which would blatantly benefit one industry (a few notable exceptions do come to mind); however, a bill which has popular support but is detrimental to certain industries would be strongly opposed by legislators with vested interests in the success of these industries. It is in such circumstances that investors are able to generate significant alpha.

Now that this paper has been widely circulated, the question becomes: will these excess returns continue to be available to investors? New evidence suggests that they will.

Posted in Blog.