Measuring international risk-sharing : theoretical issues and empirical evidence from OECD countries

Abstract

Whether financial market integration raised global insurance is a crucial, still open issue. All empirical methods to measure cross-border risk-sharing are based on the implicit assumption that international prices do not fluctuate in response to business cycle shocks. This paper shows that these methods can be completely misleading in the presence of large fluctuations in international prices as those observed in the data. I then propose a new empirical method that is immune from this issue. The risk-sharing inefficiency between two countries is measured by the wedge between their Stochastic Discount Factors (SDFs). This measure is a proxy for the welfare losses created by imperfect insurance. Welfare losses can be attributed either to the strength of uninsurable shocks (the extent of risk to be pooled) or to the degree of insurance against different sources of risk. The method is applied to study the evolution of risk-sharing between the US and OECD countries, assuming either constant or time-varying risk-aversion. The degree of insurance is found to have improved over time only for some countries and only if SDFs are estimated assuming time-varying risk-aversion. The results are also informative on the implications of different macro models for international risk. When confronted with the data, standard open-macro models (featuring constant risk-aversion) imply that nominal exchange rate fluctuations do not contain wealth divergences across countries, but rather represent an important source of risk. Time-varying risk-aversion instead implies that limiting welfare losses from imperfect risk-sharing requires reducing the volatility of macro fundamentals.