Is the relative risk aversion parameter constant over time? A multi-country study

In this paper, an information matrix (IM)-based test is developed for testing the hypothesis of constant relative risk aversion parameter in the GARCH-M set up. A detailed Monte Carlo study is then carried out to evaluate the performance of this test in terms of size and power. Further, a bootstrap technique is suggested to correct the over-size problem found in small samples. The proposed test is then applied to the time series of returns on stock markets of five important countries to examine whether this important hypothesis holds or not, and it is found that the relative risk aversion parameter is not time invariant for all the five time series.