Extrapolation Bias
Touch a hot pan, feel the pain burn through your fingers, and you quickly learn not to touch a hot pan again. This is a basic survival instinct: avoid things that have been painful.
Unfortunately, this basic survival instinct leads to one of the most common investment mistakes. Investors tend to sell after down markets, running away from the stimulus that brought them pain in the past.
Andrei Shleifer, who pioneered behavioral finance and is one of the best financial researchers of the past century, recently published exciting new research that shows that investors extrapolate from the recent past in forming their return forecasts for the future—and that they act on these backward-looking forecasts.
Shleifer looked at surveys of professional investors and public company CFOs and found that their forecasts for the next-12-month returns were highly correlated with the last-12-month returns.
Below, I show two tables from Shleifer’s paper that compare surveyed expectations about next 12m stock performance to past 12m performance from investors (left-hand side) and CFOs (right-hand side).
Figure 1: Survey Expectations vs. Past 12m Returns
Surveyed expectations are nearly perfectly correlated with past 12m returns, both for investors and CFOs. When recent stock market returns have been bad, investors fear the pain will continue, and when returns have been good, they expect good times to roll on forever.
Shleifer also shows that corporate executives base investment decisions on these extrapolative forecasts. When they expect low returns in the future—driven by negative returns in the recent past—they invest less in their own businesses.
Figure 2: Expectations vs. Investment
There is a 0.78 correlation between investor expectations, investment plans, and actual capital spending.
Professional investors make the same mistake: they tend to sell stocks after a period of bad returns and to neglect risk after periods of good returns. They have an extrapolative bias.
Unfortunately, this mental model is not only irrational—it is actually negatively correlated with rational models (a rational model, for example, would predict higher returns when prices are low and lower returns when prices are high). The big revelation of Shleifer’s new paper is that these correlations represent a systematic and recurring error: “In particular, they are consistent with the presence of excessive optimism in good times and excessive pessimism in bad times: future realized earnings growth systematically falls short of expectations when past earnings are high, and exceeds expectations when past earnings are low.”
The biggest errors in expectations—and thus investment decisions—come after very negative recent experiences, when investors are far too pessimistic and make massive forecasting errors. The next two tables show aggregate CFO expectations about next-12-month earnings growth against past-year corporate profitability on the left, and the same series using analysts’ expectations data on the right.
Figure 3: Past Corporate Profitability vs. Expectations Errors in Earnings Growth
Shleifer’s findings reveal a predictable, systematic, and exploitable error in investor expectations. “Expectational errors are consistently predictable from highly relevant publicly available information, such as past profitability,” he says.
To profit from this insight, investors need to be aware of and combat these pervasive expectation errors, following Warren Buffett’s famous saying, “be fearful when others are greedy and greedy when others are fearful.”