Understanding the effect of climate change on the global economy and capital markets is critical to institutional money managers, chief risk officers, and policymakers. In their analysis, the authors endeavor to gauge asset managers’ behavior in the face of climate disaster risk as a function of their proximity to the disaster zone.

More specifically, the authors evaluate how portfolio managers of US equity mutual funds could misread the effect of an extreme climate event by overestimating the likelihood of a future natural disaster and consequently reducing shares of companies in their portfolios recently subjected to such events. At issue is how decisions made in this manner could negatively affect portfolio performance.

The authors investigate how the salience hypothesis and the information hypothesis explain the risk aversion of portfolio managers who operate within 100 miles of a disaster zone relative to those based a greater distance away. The salience hypothesis or bias refers to an inclination to overweight the probability of an occurrence as a function of how easily one may recall an event based on its intensity, physical proximity, or emotive effects. The information hypothesis assumes that because professional asset managers are privy to better quality information than retail investors, they should make more sound portfolio choices as a result of better evaluation of how natural disasters could affect their holdings.

The authors employ a difference-in-differences approach to compare the portfolio decisions of US equity mutual fund managers based within 100 miles of firms headquartered in a disaster region (the treatment group) with those of managers located farther away (the control group). The data comprise actively managed, open-ended, and diversified US equity mutual funds from the CRSP Survivor-Bias-Free US Mutual Fund database during the years 1995 to 2016. The authors exclude exchange-traded funds and index funds, value-weighting fund-class returns to avoid multiple-counting of funds with multiple share classes and value-weighting expense and turnover ratios. The Thomson Reuters mutual fund database provides mutual fund quarterly holdings snapshots. Disaster areas are identified through SHELDUS, a database of major climatic disasters. The final sample consists of 3,268 funds, of which 1,700 are based within 100 miles of a disaster zone.

Managers in the treatment group sell shares of firms headquartered in a disaster area more than managers in the control group. These findings hold when controlling for such firm-level traits as price momentum, leverage, and sales growth. Results also persist for fund characteristics such as expense and turnover ratios, fund size, returns, and net flows.

The authors also find that salience affects managers less with the passage of time and as their distance from a disaster area increases. Finally, managers exhibit less salience bias as they gain experience managing portfolios in the context of natural disasters. The authors find no support for the information hypothesis.

The authors examine and reject several alternative explanations. Funds in proximity to a disaster region decrease their share of disaster zone equities more than those at a distance from the zone, irrespective of any temporary decline in disaster zone firms’ stock prices. Moreover, the aforementioned results hold up regardless of such fund characteristics as size, age, and manager experience. Finally, the so-called clientele effect has no bearing on managers’ portfolio decisions. That is, managers’ inclination to sell company shares close to the disaster zone does not result from investors’ decision to sell.

In the two years after a natural disaster, funds most underweight in disaster zone shares underperformed significantly, consistent with manager misestimation of disaster zone shares’ economic risk. Investment managers’ overreaction thus negatively affects investors’ performance. Finally, the authors’ hypotheses and conclusions are robust to numerous tests for such things as a climate disaster’s degree of damage, floods, less prevalent disasters (e.g., wildfires, tornadoes), and different definitions of fund proximity.

“We find that climatic disaster risk misestimation is costly to the fund investors as it adversely affects portfolio returns.”

While the authors highlight the effects of risk misperception in a particular domain, their research has important implications for the effects of risk misperception on asset pricing in general.