To Encourage or Not to Encourage Seat Belt Use: Simultaneity Bias and Offsetting Behaviour.

Estimating the efficacy of seat belt use is important as compensating risk-taking or "offsetting behaviour" by drivers may attenuate benefits from vehicle safety regulation. However, evidence on partial-offsetting behaviour from previous studies may simply reflect simultaneity bias, as increasing deaths from traffic accidents results in tougher policies aimed at encouraging seat belt use. We find significant evidence of such measurement error. OLS estimates from Canadian provincial data between 1980 and 1996 imply partial offsetting behaviour as increased average seat belt use over the sample period resulted in a 13.42% decline in vehicle occupant fatalities, in contrast to the expected 39-46% drop. However, corresponding instrumental variables (IV) estimates imply a lack of significant risk compensation as increased seat belt use is associated with a 37% fall in occupant deaths.