Investor Confidence and Reaction to a Stock Market Crash

Abstract

This paper explores how investor overconfidence impacts reactions to stock market crashes. Using the 2018 Investor Survey dataset from FINRA we measure respondent’s self-perceived and actual investment knowledge and thus are able to identify overconfident investors from other investors. In our analysis, we find overconfident investors were significantly more likely to sell after a stock market crash than other investors, and thus more likely to lock-in the losses from the crash and miss subsequent upswings in the market. Moreover, these overconfident investors were significantly more likely than other investors to pursue risky investment strategies such as cryptocurrencies, margin accounts, options, and penny stocks that can also lead to large losses. Finally, we find that accurately aware investors, investors who have both high perceived and actual knowledge, buy significantly more after a crash.

Keywords:

• Investor survey
• Market crash
• Overconfidence

Notes

1 First Trust, 2018, September 28, 2018. “History of U.S. Bear & Bull Markets Since 1926.” https://www.cascadefs.com/history-of-us-bear-and-bull-markets

2 www.usfinancialcapability.org

3 www.bea.gov/data/prices-inflation/regional-price-parities-state-and-metro-area

4 www.census.gov/geographies/reference-files/2010/geo/state-area.html

5 Note that we also examined a similar survey question (question D20) that produced very similar results to those in the paper. Question D20 asks respondents how they reacted to an actual market drop of 10% in February of 2018. These results are available upon request.

6 The underlying NFCS variables A8, and A5_2015 that were used to form the Income and Education variables allowed responses of “don’t know” or “won’t say,” however no observations in our dataset included these responses.

7 The remaining 28.3% of observations represent those respondents with Perception equal to four that are not categorized as any of the four personality types.

8 While the regression tests presented throughout this paper are probit regressions, the results are similar when repeated using logit and OLS regression tests. The logit and OLS regression tests results are available upon request.

9 The key finding that overconfident investors are more likely to plan to sell after a market crash is based on the specification of Overconfident as Perception > 4 and Reality < 50%. We perform robustness tests for two alternative and more extreme specifications of the overconfident investor type: 1) Perception > 5 and Reality < 40% and 2) Perception > 6 and Reality < 30%. We find using these alternative specifications that our results as reported in the paper still hold and are statistically significant at the one percent level. These results are available upon request.

10 The key finding that overconfident investors are more likely to engage in risky investments are broadly robust to the more extreme specifications of Overconfident that are detailed in the previous footnote. For the Perception > 5 and Reality < 40% specification the coefficients associated with Overconfident remain significant at the one percent level for all dependent variables. For the Perception > 6 and Reality < 30% specification the coefficients remain significant for the dependent variables Margin, Options, and Penny Stock at the one or ten percent levels and are insignificant for the two cryptocurrency dependent variables. All of these significant coefficients are positive and larger that the corresponding coefficients associated with the Perception > 5 and Reality < 50% specification. These results are available upon request.

11 This is according the Investment Company Institute. Indeed, net outflows from mutual funds during the week that ended March 11, 2009, well after the worst part of the 2008-09 crash, were a record $21.65 billion. See Bases (2011), “Mutual fund outflows biggest since March 2009: ICI,” Reuters News, August 10, 2011.