We aim to improve upon the existing empirical literature on international risk sharing under three dimensions. First, we generalize dynamic multi-equation approaches to the estimation of risk sharing channels, by adopting a Heterogeneous Panel VAR model. Within this framework, the coefficients representing the extent of risk sharing achieved through the different mechanisms are allowed to vary across countries. Second, we introduce two new risk sharing channels – namely, government consumption and the real exchange rate (that we further decompose into relative prices and the nominal exchange rate) – which allow us to investigate the role of fiscal policy and international price adjustments in the absorption of macroeconomic shocks. Third, we establish a better link between the “channels” empirical model and a theoretical formulation of the risk sharing condition which allows for PPP violations. Our empirical analysis, for a set of 21 OECD countries over 1960-2016, contributes to identifying the geographical structure and dynamics of risk sharing channels and to describing their evolution in the latest half-century. For the OECD sample as a whole, we confirm through 2016 the strong smoothing role played by credit markets and the small degree of risk sharing achieved through factor incomes. Interestingly, government consumption tends to have a dis-smoothing effect, due to its counter-cyclical movements. Another noteworthy result is the negative risk sharing effect of the real exchange rate, driven by the dis-smoothing role played by the movements of the nominal exchange rate, only partially offset by relative price adjustments. The evolution of these risk sharing mechanisms is diverse, but the most important channels – namely credit markets and real exchange rate adjustments – exhibit slightly positive trends for the first half of the period, negative trends afterwards, and a recovery in more recent years. Our results demonstrate that the extent of risk sharing is strikingly different across countries, especially if we take into account valuation effects through the real exchange rate. Even considering only traditional risk sharing channels, the country-specific magnitude of risk sharing on impact ranges from around 15% to over 80%. In addition, dynamics are also quite diverse across countries; for example, risk sharing through credit markets, while quite effective on impact, provokes dis-smoothing for about two thirds of the countries from the second year onwards. Our approach is of particular interest for policy makers, as it allows identifying the strengths and the weaknesses of the institutional and behavioral risk sharing mechanisms at work in different countries.
