A New Yorker article citing studies got me interested. Three hundred members of Congress had reportedly outperformed benchmark indices by approximately 6% per year across 16,000 trades from 1985 through 2001, and US Senators had outperformed by 10% annually from 1993 through 1998 per a prior paper. Although the outperformance was extreme, I did not initially question the numbers. A Morgan Stanley colleague thought the legislators were probably accessing better money managers. This colleague tended to focus her practice on finding the best blends of managers for her clients. I favored the more conventional hunch that our legislators were corruptly trading on illegal inside information. But I wanted to know more.

All US legislator transactions can be viewed online at www.opensecrets.org. As a data cleaning matter, we noted quickly that there was no standardization of investment names within the filings. A legislator might disclose that they had bought Coke, Coke Inc, Coca Cola, Coca Cola Inc, Coca Cola Incorporated, or other variations on the name. All of this of this needed to be standardized across lists of tens of thousands of companies. Fortunately, a relevant scrubbing package existed in Python, our coding language for the project.

Another difficulty was seeing everything a legislator bought or sold, not just publicly traded stocks. We learned, as an example, that Nancy Pelosi had an ownership stake in a soccer team in Sacramento, but our desired screen was for only for stocks we could buy.

After resolving these problems, we came to the real challenge: there was no way of knowing whether a Congressperson or Senator had personally initiated the trade, or whether a money manager had acted on discretion with no legislator input. Breaking this apart was necessary, because when we looked at all trades against benchmark indices (using sector ETF’s), US Congresspeople and Senators displayed no information edge. I was not discouraged, as I had expected this result. The number of trades executed by most legislators suggested widespread use of money managers, and money managers tend to underperform benchmarks after fees and expenses.

Things got interesting when we attempted to isolate trades more likely to have been directed personally by a legislator, starting with any trades bearing a geographic connection to that individual. An example would be tagging any Coca Cola trades executed by the US Congressperson in Georgia District 5 and both of Georgia’s Senators. We had difficulty imagining how legislators visiting companies in their home district could fail to gain actionable information. When we ran the numbers against benchmark indexes, however, we found that US Congresspeople and Senators showed no outperformance.

We then looked at all companies where there might be a committee relationship to a Congressperson or Senator. In seeking out such relationships, we considered whether the committee had an ability to impact a particular industry, and to what degree an industry might be in a committee’s crosshairs. Mapping approximately 150 industries against Congressional and Senate committees, we postulated, for example, that Homeland Security might have a high degree of interest in immigration that could affect high tech or agriculture or restaurants but would probably have less interest or the ability to affect industries such as cosmetics or television and broadcasting. To tighten our suppositions, we sat down with US State Department and other federal officials across agencies friendly to us and curious about our project, and we adjusted weightings after reviewing with these government insiders. Finally we ran the numbers. As with geography, committee linkages had no bearing on performance.

Our remaining option for attempting to isolate trades initiated personally by a legislator was unusually large transactions for that individual. If a legislator normally making stock purchases in the $25,000 to $50,000 range put $1 million into one or two acquisitions, the legislator and not a money manager probably chose the larger trades, so we isolated these larger trades. Once again, it was proverbial stock picking by monkeys showing no outperformance.

This exercise took considerable time for our team to execute, and I had mixed feelings about the results. From the standpoint of trying to earn returns for our clients, it was disappointing. But as someone living in a democratic society, the outcome was reassuring. A better paper on the topic eventually picked apart shortcomings in the previously academic work and found significant legislator underperformance in concluding that “widespread ‘insider trading’ in Congress is more myth than reality.” Before reading that paper, however, we were glad to have done our own homework in reaching the same skeptical conclusion, and we were glad never to have risked a dime on the strategy. Sometimes an investment strategy “ought” to work, but it simply doesn’t.