A Heavy Accusation

Do Senators routinely profit on short-term stock trades, based on greater knowledge than the average investor? That's the implied charge in an article about an academic study linked to in a number of good blogs this week. (Kevin Drum, Tyler Cowen -- via Daniel Drezner, Jane Galt)

The study found that, in the period 1993-1998, based on extrapolations from financial disclosure reports, Senators buying and selling common stocks beat the market by 12%, while ordinary investors lag the market and corporate executives beat the market by only 5%. The study cited concludes that "These results suggest that Senators knew appropriate times to both buy and sell their common stocks."

That's a heavy accusation, and it piqued my curiosity, so I got a copy of the study by Alan Ziobrowski of the Real Estate Department at the J. Mack Robinson School of Business at Georgia State. (It's not yet published.) I was a little skeptical, not because I doubt that many Senators are corrupt and take advantage of their positions, but because it's a little hard to understand exactly how most Senators would come to have the knowledge needed to time stock transactions. 90% of Senate business is unlikely to have any predictable effect on any specific company's stock price. And of the things that do -- such as the Telecommunications Act of 1996 -- it's hard to see when the Senators have a real information advantage. If, for example, you knew that an amendment would pass on the telecommunications bill that would give an advantage to, say, AT&T, you could theoretically make some quick money buying the stock before the vote. But if you as a Senator know the amendment will pass, chances are, so do the hundreds of specialized reporters and lobbyists that are feeding information back into Wall Street. You might have a few hours advance notice that a company in your state is getting a federal contract. But if you're like most Senators, your #1 priority is to get credit for announcing it, so you don't really have time for a quick call to your broker before you hit the Senate press gallery or fax out a press release.

On the other hand, Senators might obtain profitable inside information or access to intitial public offerings of stocks from lobbyists, friends or contributors as a kind of bribe, which is a more serious charge. During this period of time, at least two Senators had been investigated for making quick profits on initial public offerings and penny stocks: Al D'Amato and Bob Toricelli. So I was curious if those two had distorted the study. In a footnote, the study says that IPOs were not included at all (why not?), so neither D'Amato nor Toricelli, nor the most identifiable type of actual corruption, figure. The study does note that of the 6,000 stock transactions he tallied in the six-year period, more than half were accounted for by only four Senators: the late Claiborne Pell of Rhode Island, John Danforth of Missouri, John Warner of Virginia, and Barbara Boxer of California. The first two retired in the mid-1990s, well before the period covered in this outdated study ended. Danforth, Pell and Warner are also all heirs to great estates (Warner's first wife was a Mellon heiress), and Pell and Danforth were Senators of absolute, unquestioned integrity. (Not to denigrate the other two.) Boxer, as it happens, was a stockbroker by profession before entering politics, so perhaps knows what she's doing.

The Ziobrowski study recalculates the results by another statistical method (which I am not qualified to evaluate), that he says weights the Senators equally and thus eliminates the distortion of the four heavy traders, finding by this method a smaller but still statistically significant advantage in the Senators' stock purchases, though not their sales. Ziobrowski concludes based on this method that "trading with an informational advantage is common among Senators." Yet in the first years of the study, which is when the advantage appears, only 25-27 of the hundred Senators traded stocks at all. So it is not "common," since it is actually not common for Senators to trade stocks at all. In the last two years of the study, the number of Senators actively trading stocks rose into the high 30s, but in those later years, the Senators' advantage over the market also disappeared.

The study performs various other statistical breakdowns of the data, by party, by committee assignment and by seniority, but with each breakdown, the sub-sample of the 25 actively trading Senators naturally gets even smaller. He concludes that first-term Senators do better than Senators with more seniority, but that's a counterintuitive result: If Senators were trading on either information about congressional action, or on information provided by lobbyists, in either case Senators with greater seniority, and hence greater power, should be at an advantage. Counter-intuitive results such as this should suggest that the results are mostly statistical anomalies resulting from the small samples. And it also seems to me that, before concluding that Senators trade on an information advantage, there should be some effort to determine what the information advantage is and whether the stocks in question were those on which a Senator might have some information.

It is a provocative study, and should not be written off altogether. And IPO profits should certainly be included. But for the most part, this seems like public choice theory run amuck.