Beshears and Katherine L. Milkman of the Wharton School at the University of Pennsylvania looked at how so-called "escalation bias" affects analysts' forecasts. Their paper, "Do Sell-Side Stock Analysts Exhibit Escalation of Commitment?" (PDF, 39 pages) was published in March by the Journal of Economic Behavior and Organization.
Beshears and Milkman studied the Institutional Brokers' Estimate System I/B/E/S database with more than 6,200 analysts' quarterly forecasts on about 3,500 companies over more than 18 years, from 1990 to 2008.
Among their specific findings:
- As analysts got more and more extreme, or "out-of-consensus," they became less and less responsive to the new earnings information when revising their full-year forecasts.
- Stubbornness hurts forecasting accuracy. Revised full-year forecasts from extreme incorrect analysts were further off the mark from actual earnings than they would have been had the analysts' updating behavior been like the behavior of analysts who started at the consensus point.
- Analysts are punished for stubbornness. The more extreme, incorrect, and stubborn an analyst was, the less likely that he or she would rank among Institutional Investor magazine's "All-American" list of top analysts.