Risk Compensation Experiment

Mitigating large losses is the focus of risk management. This wide domain covers excessive financial losses resulting in bankruptcy, personal accidents causing grave injury or death, and catastrophes igniting localized or systematic collapse of complex systems. Describing individual choice in mitigating the risk of excessive loss, risk compensation refers to a negative effect of an external safety condition on the choice of effort. The hypothesis of risk compensation is a recurring argument in public debates but observational evidence is inconsistent and lacks a clear definition. Filling this gap, we conduct a combined field and laboratory experiment.

In the field component of the study, we observe the subjects' behavior before they begin the laboratory experiment. We implement two interventions: (i) half of the participants are randomly selected to wear certified masks to the experiment (N95 or equivalent), while others receive no mask requirement; and (ii) in half of the sessions, information about a HEPA filter that is installed in the room is displayed on a large TV monitor, while nothing is displayed in the other sessions (HEPA filter is installed in all sessions). We measure the effects of these interventions on outcome variables that are believed to be linked to the spread of COVID-19 and other infectious diseases: distance between the subject and the person in front of him/her in the queue (before entering the laboratory); how often each subject removes the mask, e.g., to drink water; and whether or not the subject uses a hand sanitizing wipe.

In the laboratory, the subjects participate in an incentivized individual choice experiment in which they receive an endowment they can lose in a lottery. The chance of a loss depends on an exogenous parameter (which are not directly revealed to the subjects, but the subjects can observe outcomes from a small sample of simulations) and the costly insurance that subjects can choose to purchase. We vary the exogenous parameter, and observe how the subjects' insurance choices change in response.

We have three overarching goals from this study. First, in the field component of the study, we examine whether the two interventions -- HEPA filter information and the requirement to wear a high-grade mask -- lead to risk compensating behavior in the real world. Second, in the lab experiment, the setup allows us to investigate whether the subjects have a tendency to under-, over-, or correctly compensate for the exogenous changes in risk. Finally, we examine the extent to which the subjects' compensating behaviors in the lab correlate with their behaviors in the field.