Blog - Investing Notes

June 13, 2005 – Public Sentiment and Stock Returns

In their March 2005 paper entitled "Measuring and Interpreting Expectations of Equity Returns", Jeff Dominitz and Charles Manski examine the results of two series of surveys measuring American public expectations of equity returns for the next 12 months. The surveys took place during July 1999-March 2001 (the Survey of Economic Expectations) and June 2002-August 2004 (the Michigan Survey of Consumers). The question posed asks respondents to state the percent chance that a diversified mutual fund will have a positive nominal return in the year ahead. They discover that:

• At any given time, expectations of equity returns vary considerably from person to person.
• Specific individuals interviewed twice at six-month intervals tend to give similar responses.
• Mean expectations for positive equity returns are ~70% during the (bubble/post-bubble) period 1999-2001 and only ~40%-50% during the (decline/post-decline) period 2002-2004.
• Males are more optimistic than females. Non-hispanic whites are more optimistic than other ethnic groups. Those with more formal education are more optimistic than those with less. The young are more optimistic than the old.
• There are no dominant types of market thinkers (random walkers, trend-followers, mean reverters).

We find the third point above to be the most interesting. The following chart, based on the Michigan survey data from Table 1 of the paper, compares the public's mean expectation for a positive nominal return from stocks with the performance of the S&P 500 index over the prior year. The Pearson correlation for these two series is a strong 87%, meaning that the two series move pretty reliably in the same direction. The correlation is about equally as strong if we use the S&P 500 index performance over the past two or three years. But how well does the public's mean expectation predict future S&P 500 index returns?

The next chart, again based on the Michigan survey data from Table 1 of the paper, compares the public's expectation for a positive nominal return from stocks with the performance of the S&P 500 index over the coming year. The Pearson correlation for these two series is a moderately strong -64%, meaning that the two series tend to move in opposite directions. In other words, low (high) sentiment suggests a high (low) future return.

Putting the prior two charts in perspective, the next one plots the correlations between the public's expectation for positive stock returns in the coming year and the performance of the S&P 500 index over various intervals before and after the expectation measurement. It shows that the past shapes public expectations about investing in equities and that the actual performance of stocks tend to "trick" the public.

In summary, human sentiment is a rear-view mirror measurement, subject to reversal.

Note that the sample used to generate the above charts is small (in terms of the number of independent market intervals considered), and results are therefore suggestive rather than reliably predictive.

For related research, see Blog Synthesis: Sentimental Journey, encompassing a broad range of equity market sentiment measures.