Sampling distribution of the relative risk aversion estimator: Theory and applications