This paper presents a statistical and economic interpretation of the low and often economically implausible risk aversion estimates obtained for fixed income assets throughout the finance literature. For a statistical interpretation, Monte Carlo simulations are used to demonstrate that specification errors introduce a serious downward bias in parameter estimates derived from the standard asset pricing model. For an economic interpretation, an international version of the asset pricing model is presented. The model suggests that by reducing the effect of country specific disturbances, an international measure of consumption growth yields more accurate risk aversion estimates than a national measure. The results of asset pricing tests suggest that risk aversion estimates derived from models constructed for the international measures are economically plausible and close to each other across eight industrialized economies. These results are robust for several asset returns.