Rejoinder

Click to increase image sizeClick to decrease image size

Notes

1 Our analysis, available upon request, allows for endogenous loss probabilities via a linear function of effort, $(1-e)\mu$. The effort level, e, is in turn associated with a (potentially heterogeneous across agents) quadratic cost function. The analysis shows that for deductible levels as in our data, the choice of $200 in collision is not rationalizable, even in the presence of endogenous loss probabilities.