high level of international risk sharing when the productivity growth contains long run risk
This theoretical paper investigates international risk sharing and its implications for equity home bias. A general equilibrium model, featuring two closed economies with nontrivial production sectors, is developed. Moreover, productivity contains a small but persistent highly correlated long run risk that becomes the major determinant of the intertemporal marginal rate of substitution (IMRS) in a model with the recursive preferences. Despite adopting the model of closed economies and autarkic asset holdings—a scenario leading to the lowest level of international risk sharing under the same conditions—our model is still able to generate international risk sharing indexes always over 96% for a broad range of parameter values, excepting two cases: where the elasticity of intertemporal substitution (EIS) is the reciprocal of the relative risk aversion (RRA); and where EIS is around 0.7. In those cases, the risk sharing index drops sharply to about 30%. This result sheds light on why the benchmark model, featuring a power utility whereby EIS is the reciprocal of RRA, generates international risk sharing as low as 30%. However, when EIS takes these values, our model’s results cannot be reconciled with asset market data-model yields low volatility of the logarithms of IMRS, even lower than Hansen-Jagannathan lower bound. The implication is that the low proportion of foreign assets in a domestic agent’s portfolio, a phenomenon observed in the data, might not be a puzzle or a departure from the agent's optimality condition. After all, risk has already been well shared internationally due to the high correlations across countries of the long run productivity shocks. Hence, there is not much incentive left for an agent to hold foreign assets in her portfolio to further share the risk internationally. Therefore, equity home bias might not be a puzzle as claimed by the benchmark model, in the sense that it can be adequately reconciled with our theoretical result
