IMF Working Papers

I study the implications of productivity shocks in a model where agents observe the aggregate level of productivity but not its permanent and transitory components separately. The model's predictions under learning differ substantially from those under full information and are in line with several empirical findings: (i) the response of investment to a permanent shock is sluggish and peaks with delay; (ii) permanent shocks generate positive rather than negative savings on impact; and (iii) saving and investment are highly correlated despite the assumption of capital mobility. Unlike other standard explanations of the Feldstein-Horioka puzzle, learning induces high correlations irrespective of the assumed persistence of shocks.