Monday, November 16, 2020

How much value should you place in macroeconomic forecasts? - The under and overreaction of forecasters


How much stock should you place in the consensus macro forecasts generated by the experts? How much value should you place on the forecasts of an individual macro forecast from an expert? These are the core questions address in the paper, Overreaction in Macroeconomic Expectations by Pedro Bordalo, Nicola Gennaioli, Yueran Ma, and Andrei Shleifer. 

Their extensive work is focused on the Survey of Professional Forecasters (SPF) and the Blue Chip Survey and provides analysis on 20 time series expectations in total. They find that there is consistent under-reaction in the expectations of the consensus forecasts. The aggregated expectations and the slow revisions of economist means their forecasts are playing catch-up to reality. This conclusion is not new albeit still inconsistent with a rational expectations view of the world. Given this under-reaction to news, the consensus forecast revisions are positively correlated to forecast errors. The conclusion is that investors should discount these forecasts. 

While the consensus may seem slow to react to news and information when forming a collective expectation, the authors find that there is an over-reaction in the behavior of the individual economic forecasters. This conclusion seems odd when compared with the consensus result.

The authors study closely this individual and consensus behavior of over and under-reaction within the broad set of macro variables and conclude that these results are based on the process of aggregation for the consensus forecast and on diagnostic expectations for the individual forecasts. 

The diagnostic expectations view states that there is a specific Bayesian updating process that is often used by forecasters where there seems to be more weight placed on the posterior expectation or new information. This can occur when it is difficult to assess new information as permanent or transitory. The diagnostic expectations can lead to a representative or recency bias which will produce an over-reaction in forecasting based on new information. Following the forecasts of individual economists can lead to errors from over-reaction.

While individuals may over-reaction, the use of different models, different information, and slow updating will lead to an under-reaction in aggregate. So, the investor should fade the forecast of the individual but assume that the group is slow to respond to the market realty. 

This research reinforces common themes which we have been presenting for years - follow trends in prices  as a core base for any market judgment because following the forecasts of economists may not give you an unbiased forecast, but the potential for both under and over-reactions.  

No comments: