At the JME-SNB-SCG conference, I discussed a paper by Karel Mertens (no relation) and Morten Ravn, estimating fiscal multipliers. Very neat! They embedd narrative data from the Romers in a SVAR. The point estimates suggest that tax multipliers are quite a bit higher than what has been found previously -- however, since there are only so few non-zero data points in the Romer data set, I raise some quibbles about the confidence intervals. My full slides are posted below.