How well do analysts forecast growth at different time horizons?

We have long been interested in the science of forecasting. Investing, after all, is about making bets on predictions, so it’s worth knowing as much as we can about how to be good at making predictions—or, on a more meta-level, the limits of human forecasting ability.

An interesting new study adds to our knowledge about economic forecasting by showing that predictive accuracy in economic forecasting appears to deteriorate the further into the future is being forecasted. The staff of the Federal Reserve Bank of New York studied the Philadelphia Fed’s survey of professional forecasters’ predictions about future levels of inflation and output growth. The authors found that errors increased considerably for horizons beyond a year. Forecasters are underconfident about their predictions in the short-term but overconfident about those in the long-term, as pictured in the graph below.

Figure 1: Forecast Overconfidence by Time Horizon

Source: Bassetti et al. Note: The dots show the average ratio of squared forecast errors over subjective uncertainty for estimates made eight to one quarters ahead. The whiskers indicate 90 percent posterior coverage intervals based on Driscoll-Kraay standard errors.

The same finding on short-horizon predictability seems to hold for equity analysts. A 2021 paper found that one-year cash flow growth estimates from equity analysts are highly volatile, with a standard deviation of 28%, yet they have a high correlation of 0.48 with realized future cash flow growth.

Figure 2: Expected and Realized Short-Term Expectations

Source: De la O & Myers

The researchers found that analyst forecasts typically fail to predict future recessions, unsurprisingly, but they “do tend to predict recoveries and track future earnings growth reasonably well during normal times.”

Although the short-term forecasts, outside of recessions, are decent, the longer-term forecasts are so bad as to be a contrarian indicator. A paper by the powerhouse team of Bordalo, Gennaioli, La Porta, O’Brien, and Shleifer found that higher than average long-term growth expectations (3-5 years out) consistently predict low future stock returns, as shown in the graph below.

Figure 3: Response of Cumulative Investment Growth to Predictable Forecast Errors

Source: Bordalo et al. Note: The figure shows the cumulative impact of a 1 standard deviation change in forecast error on the change in investment-to-capital over the next h quarters.

Long-term forecasts seem to reflect animal spirits, irrational optimism and pessimism about future events that no one can accurately predict. In contrast, short-term forecasts seem to provide real and relatively accurate information about future cash flows.

Acknowledgment: Luke Burke is a senior at the University of Pennsylvania, where he studies English. He is a member of the chess club, plays on the club tennis team, and writes for the university’s satire publication, Under the Button. After graduation, he plans to pursue a career in investing.

back to library

Archive

Short-Term vs. Long-Term Forecasts