Twisting Theories to Suit Facts : Revisiting the Effects of Technology Shocks

This paper addresses an issue that has long been neglected in the proxy structural vector autoregression (SVAR) literature, namely the need for model-implied shock realizations to align with historical records. To resolve this issue, we propose a solution that combines the proxy SVAR approach with narrative sign restrictions. Our simulation exercise attests that the combination would lead to more informative inferences on shock effects. Applying the new strategy to jointly identify anticipated and unanticipated technology shocks, our estimation reveals that technology shocks are not the main drivers of business cycle. Moreover, we find that anticipated technology shocks induce persistent hump-shaped comovements among macroeconomic variables, which aligns well with the prediction by a labor search and matching model