Theses Doctoral

Disasters, Beliefs, and the Behavior of Investors

Xu, Xiao

This dissertation contains three essays in financial economics. The focus of the dissertation is to study how retail investors and the financial market react to the arrival or the possibility of disastrous events. In the first chapter, I explore the portfolio reaction to evidence of climate change by looking at how retail investors trade when they locally experience abnormal temperature. I test the hypothesis that retail investors will trade out of high emission stocks and trade into low emission stocks when experiencing abnormally high temperature through a channel of climate belief updating.

Using detailed administrative records of retail investors’ positions and trading activities from a large financial institution, I construct measures of trading imbalances at the zip code level for various types of stocks and study the impact from abnormal temperature. I do not find evidence that investors trade out of high emission stocks or trade into low emission stocks when experiencing abnormally high temperature. The estimated effects are neither economically nor statistically significant. Moreover, investors are not dynamically adjusting their portfolios in response to abnormal temperature. The nonresults are robust if I implement the estimations in quarterly or annual frequency. Focusing on only trading activities in the energy sector does not change the results.

Analyzing subsamples of investors with different levels of beliefs in climate change also produces nonresults. Although past literature has shown that local extreme temperature can induce changes in beliefs about climate change and related behavior, this paper shows that such belief updating does not translate into response in portfolio choice.In the second chapter, we model the contribution of a vaccine to the rebound in corporate earnings the year following the onset of COVID-19 while accounting for the role of fiscal and monetary measures. A vaccine that reopens the economy leads to a jump in earnings, while temporary fis- cal and monetary support for households and businesses leads to higher short-run earnings growth before a vaccine arrives. We show that our model can be consistently estimated using revisions of value-weighted industry-level consensus earnings forecasts. We first present reduced-form evidence that security analysts account for both effects. Our model estimates then suggest that the reopening effect is as important as the short-run growth effect in explaining the rebound in corpo- rate earnings.

The third chapter studies the partisan difference in trading behavior at the onset of the COVID-19 pandemic. Partisanship drives disagreement on the severity and persistence of the COVID-19 shock when it hit the US. Republicans were more optimistic than the Democrats when evaluat- ing the potential damage of COVID-19 to the economy. Using detailed administrative records of retail investors’ positions and trading activities from a large financial institution, I find that the partisan disagreement on COVID-19 is reflected in stock trading behavior: Republicans had more net flow into equity than the Democrats from March to May of 2020. Moreover, the difference is concentrated on industries with high face-to-face interactions and highly levered firms, which are expected to be more severely damaged by COVID-19. The results suggest that disagreement rooted in partisanship can have a real impact on household financial decisions and potentially on the overall financial market.


  • thumnail for Xu_columbia_0054D_17005.pdf Xu_columbia_0054D_17005.pdf application/pdf 1.78 MB Download File

More About This Work

Academic Units
Thesis Advisors
Hong, Harrison G.
Zeldes, Stephen P.
Ph.D., Columbia University
Published Here
January 12, 2022