Estimating the Returns to Insider Trading: A Performance-Evaluation Perspective
Jeng, Metrick, and Zeckhauser cleanly separate insiders’ own purchase returns from outsider mimicry and find no significant abnormal return for the sale portfolio.
Jeng, Metrick, and Zeckhauser reframe the question around the returns earned by insiders themselves. Their performance-evaluation approach constructs purchase and sale portfolios from reported trades, controls for common stock characteristics, and measures the economic cost imposed on non-insider counterparties. This differs deliberately from asking whether an outsider can imitate a public filing.
The purchase portfolio earns abnormal returns of more than six percent per year in the authors’ framework. Roughly one-third of the raw difference is associated with insiders’ tendency to buy small, value, and higher-beta stocks; the remaining monthly estimates vary with the performance model. The sale portfolio does not earn significant abnormal returns. Because insider transactions are a tiny share of market volume, the estimated adverse-selection cost to outsiders is about ten cents on a $10,000 transaction.
These findings are the clearest quantitative basis for the site’s purchase–sale asymmetry and dead-cat-bounce caveat. A sale can fund taxes, consumption, diversification, or an option exercise; observing one does not establish a negative view. The paper also prevents a common category error: insiders’ own historical portfolio returns are not a public copycat result. Insider Atlas therefore describes transaction direction and reported value but does not display the six-percent figure as a company-level expectation. Its sales pages carry the caveat even when the reported dollar total is visually large.
Abstract excerpt
“This paper uses performance-evaluation methodology to estimate the returns earned by insiders when they trade their company’s stock.”
Connections in Insider Atlas
The NTSK purchase event and DELL sales event provide a descriptive direction contrast. The DELL company page keeps the sales caveat beside the history.