DSGE Models Are Garbage

The crowning achievement of modern macroeconomics is an incredibly detailed dynamic model of the U.S. economy that is used by governments and big businesses alike to make all sorts of important economic policy decisions. 

I've felt this since I took 300 level macroeconomics in college back in the late 1980s that this model is basically bunk. It turns out that I was right.

It is really, really bad. Worse than even I would have guessed. Delete everything and start over from scratch bad. Deny macroeconomists the right to testify in court as expert witnesses because their models are junk science bad. As Marginal Revolution, a group blog of academic economists put it:

Imagine if you were trying to predict the motion of the planets but you accidentally substituted the mass of Jupiter for Venus and discovered that your model predicted better than the one fed the correct data. 

The paper and its abstract are as follows: 

Dynamic stochastic general equilibrium (DSGE) models have been an ubiquitous, and controversial, part of macroeconomics for decades. 

In this paper, we approach DSGEs purely as statstical models. We do this by applying two common model validation checks to the canonical Smets and Wouters 2007 DSGE: (1) we simulate the model and see how well it can be estimated from its own simulation output, and (2) we see how well it can seem to fit nonsense data. 

We find that (1) even with centuries' worth of data, the model remains poorly estimated, and (2) when we swap series at random, so that (e.g.) what the model gets as the inflation rate is really hours worked, what it gets as hours worked is really investment, etc., the fit is often only slightly impaired, and in a large percentage of cases actually improves (even out of sample). 

Taken together, these findings cast serious doubt on the meaningfulness of parameter estimates for this DSGE, and on whether this specification represents anything structural about the economy. 

Constructively, our approaches can be used for model validation by anyone working with macroeconomic time series.

Daniel J. McDonald, Cosma Rohilla Shalizi, "Empirical Macroeconomics and DSGE Modeling in Statistical Perspective" arXiv:2210.16224 (October 31, 2022).