|M.Sc Student||Cohen Doron|
|Subject||Risk Compensation, Over-Regulation, and the Effect of|
|Department||Department of Industrial Engineering and Management||Supervisors||Professor Ido Erev|
|Professor Eitan Naveh|
|Full Thesis text|
Empirical research suggests that the effort to enhance safety can backfire. New technologies that make the environment more forgiving can trigger "risk compensation", and new regulations that punish reckless behavior can lead to "over-regulation”. The leading explanations assume agents are generally maximizers, but they err in certain cases. This implies that safety interventions should either protect people from their errors, or regulate the environment to reduce the temptation to take risks that are beneficial to the agents but detrimental to others. The current analysis compares two abstractions of this assumption: a “fixed error” abstraction that assumes a small but relatively fixed probability of error, and a “small samples” abstraction that assumes agents rely on a small sample of similar past experiences. The two abstractions provide similar predictions in many settings. In particular, both abstractions agree in predicting "relatively boring" risk compensation and over-regulation effects. This agreement is clarified in two sets of examples. In Example Set 1, both abstractions suggest that safety interventions that make the work environment safer may increase workers expected return but can also increase risk taking rates. In Example Set 2, both abstractions demonstrate an over-regulation effect (a non-monotonic effect of the strength of the regulation) that does not reflect a causal relationship (e.g. a third variable is responsible). The two abstractions differ with respect to the possibility of more interesting risk compensation and over-regulation effects. In Example Set 3, only the small sample abstraction predicts that an intervention that makes the work environment safer can impair worker’s expected return. In Example Set 4, only the small sample abstraction predicts a causal relationship between tougher regulation and an increase in risk seeking. Two experimental studies were conducted to compare the distinct predictions. Study 1 examined human behavior in Example Set 3. In this experiment, two repeated choice tasks were studied using the clicking paradigm. One task simulated a "before safety intervention" setting, and the second simulated a variant of this setting with a safety intervention that improves the payoff from risky choices. The results support the small samples hypothesis: they demonstrate that the safety intervention increased risk seeking and impaired expected return. Study 2 used a similar method to examine human behavior in Example Set 4. In support of the small samples hypothesis, results reveal an over-regulation effect that reflects a causal relationship. Tougher regulations increased risk seeking. Implications for policy makers and safety engineers are discussed.