Inaccuracy of Loglinear Approximation in Welfare Calculations: the Case of International Risk Sharing

This paper investigates the accuracy of the log-linear approximation method in welfare calculations, especially in measuring welfare gains of international risk sharing. We derive closed-form solutions for a two-country complete market economy using log-linearization and a nonlinear solution method and compare risk-sharing gains over financial autarky. We document that the loglinearized model underestimates risk-sharing gains by up to 20% of world consumption under certain parameter values with endogenous labor supply. While the nonlinear solution generates 4% risk-sharing gains, the loglinear approximation results in a loss of 16%. Loglinear approximation errors are large enough to generate welfare reversal between autarky and complete market economies, a violation of the first welfare theorem. This result can be crucial because a large number of papers adopt loglinearization method in calculating welfare.