Sunday, August 17, 2014

Investors get their expectations wrong

Investors are supposed to form rational expectations. Think again. An extensive analysis of return expectations suggests that investors form their view through extrapolation of past performance. The work by Greenwood and Schleifer, Expectations of Returns and Expected Returns, is damning evidence against rational expectations, Do not follow what you see in surveys. It also tells us that systematic activity may be a better method for running money versus following the forecasts in herds which often just means following recent economic behavior.

Since the work of Shiller on excess stock market volatility, there has been a growing concern that the assumption of rational expectations and market efficiency have to be revised or certainly looked at more closely.

This work is an important addition because it reviews a number of sources for market expectations and gets the same answer. The author look at the Gallup poll, Graham and Havvey CFO survey, AAII survey, Investor's Intelligence survey, the Shiller survey, and Michigan survey. They are all correlated and all get it wrong.

Good news is extrapolated into strong market expectations which are often not realized. As important, investor expectations are negatively correlated with model-based expected returns derived from dividend/price, consumption patterns and market valuation.  Investors, no matter what the level of experience, do not seem to use the models that provide useful information on expected returns. Put differently, when expected return models forecast higher returns, they are usually correct. When the expectations of returns are high from surveys, the actual returns are low. These market expectations are correlated with mutual fund flows. The surveys show expectation that investors actually use, albeit incorrectly.

Use forecast data at your own risk. Survey show bad forecasting skill You may want to fade market expectations.This just adds to the list of evidence that discretionary forecasting does a poor job.  Models do better. Do not listen to the talking heads or the aggregated opinion of surveys. Use decision rules that do not rely on the opinions of others. 

No comments: