The image of a senator quietly dumping stock after a classified briefing has become a fixture of American political folklore. It fuels bestselling books, viral social media accounts, and a cottage industry of financial products built around copying congressional trades. As of March 2026, retail investors had poured at least $650 million into vehicles designed to mirror the portfolios of lawmakers, on the assumption that elected officials enjoy an informational edge the rest of us can only dream about.
A new analysis, released as an NBER working paper, takes a hard look at that assumption. The researchers find that from 2012 through 2023, members of Congress and their immediate family members, on average, did not beat the market. Even more striking, the timing of their trades looks a lot like the timing of ordinary retail investors scrolling financial social media.
The question behind the headlines
Haotian Chen of the University of California, Los Angeles and Bruce Sacerdote of Dartmouth College set out to revisit a debate that has kicked up repeatedly since the early days of the COVID-19 pandemic, when several senators faced scrutiny for stock sales made shortly after closed-door briefings about the looming outbreak. Prior academic work had already suggested that lawmakers’ portfolios are unremarkable. But earlier studies mostly asked whether legislators beat a market benchmark in aggregate, a question that can mask abnormal profits on individual trades and that ignores the element driving public outrage: timing.
A trade can be unprofitable yet still reflect the misuse of nonpublic information, and a profitable trade need not be informed. The authors wanted to separate these threads by looking at whether congressional trades show signs of anticipating future price movements, or whether they instead track signals anyone with an internet connection could see.
Building the dataset
Under the 2012 STOCK Act, members of Congress must disclose trades above $1,000 within 45 days. The researchers hand-collected and, for later years, scraped these Periodic Transaction Reports for every senator and House member from 2012 through 2023. After filtering to publicly listed securities and consolidating same-day trades, they assembled a dataset covering 334 legislators trading in 3,403 different public firms, roughly 71,000 transactions in all.
The distribution is lopsided. Representative Rohit Khanna alone accounted for about 17 percent of all trades, while 33 legislators reported just a single transaction over the entire period. Most trades are small: about three-quarters fall in the lowest disclosure bracket of $1,000 to $15,000.
What the returns look like
The researchers calculated buy-and-hold abnormal returns, which compare what a legislator earned on a specific trade against what a passive investor would have earned holding a benchmark over the same window. They ran these comparisons against five different benchmarks, including the broad market, industry-and-size matched portfolios, and standard academic factor models that adjust for known risk characteristics.
Across specifications, the average legislator trade slightly underperformed or roughly matched the market. Over a full year, trades lagged the broad market by about 26 basis points. When the authors weighted trades by dollar value, the result flipped positive but remained modest, implying an abnormal gain of roughly $115 per $10,000 invested over a year.
The authors then looked at groups one might expect to have the best access to private information: frequent traders, party leaders, and legislators trading in industries overseen by their committees. These groups did worse, not better. Party leaders’ trades underperformed by about 4.4 percentage points at the one-year horizon. Trades in industries matched to a legislator’s committee oversight underperformed by about 2.8 percentage points. Chen and Sacerdote write that this pattern is “difficult to reconcile with the insider-trading hypothesis.”
Following the analysts
Poor returns alone don’t settle the question, since even informed traders sometimes lose money. So the researchers turned to timing. Their first test compared each trade against professional analyst recommendations, using I/B/E/S, a database that standardizes brokerage ratings on a five-point scale from Strong Buy to Strong Sell.
The pattern is consistent and monotonic. When professional consensus was most bearish, legislators bought less often than they sold. When consensus was most bullish, they bought more often. Moving the average analyst recommendation one notch toward a more bearish view was associated with a 4.7 percentage point lower probability that a given trade was a purchase. Recent analyst downgrades in the 30 days before a trade were linked to a 3.4 percentage point drop in the likelihood of buying.
To check whether legislators might actually be anticipating analyst moves rather than following them, the researchers ran a lead-lag test. Future analyst revisions had little predictive power for trade direction, while past revisions did. An event study around the weeks of analyst upgrades and downgrades likewise showed no systematic pre-trade shift in buying or selling behavior.
Following the crowd
The second timing test drew on retail investor chatter. Using roughly 501 million posts from StockTwits, a social media platform where users tag posts with stock tickers, the authors built a measure of abnormal attention for each stock in each week. They then traced what that attention looked like around the weeks when legislators traded.
Attention on a given stock rose in the weeks before a congressional trade, peaked in the trade week itself at about 2.6 percent above the prior week’s level, and then declined. Legislators, in other words, tended to trade stocks that retail investors were already talking about, and they tended to do so near the top of the attention cycle rather than ahead of it.
A separate analysis using Twitter/X sentiment scores from Context Analytics showed a similar pattern for the emotional direction of that chatter. The sentiment score is normalized so that a value of 1 means sentiment is one standard deviation more bullish than that stock’s recent baseline. A one-unit increase in the score was associated with a 1.83 percentage point higher probability that a legislator’s trade was a purchase. When sentiment was unusually bullish, buys became more common; when it was unusually bearish, buys became less common. A placebo test showed that future sentiment did not predict current trade direction, while recent sentiment did.
Why the “insider” narrative may be off
Chen and Sacerdote offer two interpretations of why lawmakers’ trades look so ordinary. The first is deterrence: STOCK Act disclosure rules and the army of media trackers and social media accounts watching for suspicious activity make concealment difficult, while the reputational and electoral costs of getting caught are steep. The second is that much of the information legislators encounter through hearings, briefings, and lobbying conversations is noisy, contingent on legislative outcomes that may never happen, and often already priced in by the time any trade could be placed.
The authors also revisit a well-known earlier finding that lawmakers tended to overweight firms headquartered in their home states and firms whose political action committees had donated to them. That overweighting still shows up in their data: connected firms receive two to ten times the portfolio weight of unconnected firms. But trades in those connected firms did not produce abnormal returns. Trades in local firms actually underperformed the market.
Caveats worth keeping in mind
The analysis rests on self-reported disclosures, which are known to include late filings and occasional omissions. Disclosure brackets also obscure exact trade values; the authors assume trades take the lower bound of each bracket and show that alternative assumptions do not change their conclusions. The study covers stock trades and does not examine less regulated arenas like cryptocurrency or prediction markets, which the authors flag as areas for future work. And as they note, portfolios can still matter for governance even if they don’t generate trading profits, since holdings have been linked in prior research to roll-call votes, government contracting, and lobbying activity.
For anyone building an investment strategy around copying Capitol Hill, the findings raise a direct question. The premise of those $650 million in trade-mimicking products is that lawmakers know something the market doesn’t. Chen and Sacerdote’s data suggests that, on average and on the visible margin, what lawmakers know is roughly what you can learn from a brokerage report and a scroll through StockTwits.
