Learning to Price

Is monetary policy less effective in increasing real output during uncertain times? We argue that firm uncertainty adds flexibility to the aggregate price level and thus the effects of monetary policy are reduced in times of high uncertainty. To construct the argument we develop a price-setting model where firms have imperfect information about their nominal cost and must forecast it to set their prices. Our measure of uncertainty is firms' forecast variance and it is the result of an endogenous learning process. In normal times, the information friction delays the response of prices to a monetary shock and produces large and persistent real effects on output. In highly uncertain times, however, these real effects are reduced. The reason is that highly uncertain firms become more responsive to new information -which comes with noise- and adjust their prices more often.