How Much Are SVARs with Long-Run Restrictions Missing without Cyclically Moving Factor Shares? Ra¨ ul Santaeul` alia-Llopis ∗ Washington University in St. Louis Federal Reserve Bank of St. Louis February 3, 2012 Abstract I incorporate into the standard structural vector autoregression (SVAR) framework with long-run restrictions the systematic ﬂuctuations of labor share under the identifying as- sumption that labor share innovations are purely redistributive—that is, without long-run eﬀects on labor productivity. This leaves the long-run identiﬁcation of productivity shocks (i.e., Gal´ ı (1999)) and investment shocks (i.e., Fisher (2006)) fully untouched. In contrast to previous assessments, I ﬁnd that productivity shocks account for a signiﬁcant 47% of the variance of hours, while investment shocks for a smaller 13%. Further, I ﬁnd that these results are not due to diﬀerences in the SVAR-estimated shocks but to the presence of dynamic eﬀects of productivity on labor share that substantially propagate the eﬀects of productivity on hours. ∗ I thank Fabio Canova, Jes´ us Fern´ andez-Villaverde, Bart Hobijn, Claudio Michelacci, Jos´ e-V´ ıctor R´ ıos-Rull, Juan Rubio-Ram´ ırez, Harald Uhlig, and Tao Zha for useful comments. I also thank the seminar participants at the European Meetings of the Econometric Society, Oslo, August 2011; the Asian Meetings of the Econometric Society, Seoul, August 2011; the Society for Economic Dynamics Meetings, Ghent, July 2011; the North American Meetings of the Econometric Society, St. Louis, June 2011; the Washington University in St.Louis and Federal Reserve Bank of St. Louis Macro Lunch Workshop; the Federal Reserve Bank of Atlanta; the Labor Markets Workshop organized by Universidad Pablo Olavide in Huelva, Spain, in September 2009; and the Midwest Macro Meetings in Bloomington, Indiana, in May 2009. I also thank Irina Panovska for her excellent research assistance.