By Jesper Lindé, Frank Smets and Rafael Wouters
http://d.repec.org/n?u=RePEc:hhs:rbnkwp:0323&r=dge
In this paper we discuss a number of challenges for structural macroeconomic models in the light of the Great Recession and its aftermath. It shows that a benchmark DSGE model that shares many features with models currently used by central banks and large international institutions has difficulty explaining both the depth and the slow recovery of the Great Recession. In order to better account for these observations, the paper analyses three extensions of the benchmark model. First, we estimate the model allowing explicitly for the zero lower bound constraint on nominal interest rates. Second, we introduce time-variation in the volatility of the exogenous disturbances to account for the non-Gaussian nature of some of the shocks. Third and finally, we extend the model with a financial accelerator and allow for time-variation in the endogenous propagation of financial shocks. All three extensions require that we go beyond the linear Gaussian assumptions that are standard in most policy models. We conclude that these extensions go some way in accounting for features of the Great Recession and its aftermath, but they do not suffice to address some of the major policy challenges associated with the use of non-standard monetary policy and macroprudential policies.
The models of Smets and Wouters are the foundation of many of the models currently used in policy circles. Given that these models have been heavily criticized by some that they are lacking the very features that were relevant in the last crisis, it is interesting to see what Smets and Wouters (and Lindé) have come up with as a response. Given all the other work that has been done by others, and that has in part been highlighted on this blog, the response is disappointing, though. One could have in particular expected a more explicit modeling of the financial sector, for example, or there still seems to be a lot of linearity in the model even with the ZLB. My assessment, however, is based on a very quick reading of the paper, it is 95 pages long after all.
NEP-DGE Blog 