Oliver Blanchard of the Peterson Institute for International Economics had an interesting hot take on the future of DSGE models. In essence, his argument was that while DSGE models have a number of unappealing features (they’re based on shaky assumptions, estimation as a system makes it difficult to pinpoint flaws in the models, they don’t provide sufficient analysis of distributional effects for contemporary economic problems, and they are bad at communicating results to a lay audience), DSGE models are ultimately redeemable. As such, Dr. Blanchard lays out two proposals to revitalize the use of DSGE models: (1) DSGE models need to incorporate the work of other sub-disciplines (such as behavioral economics, big data, etc.) instead of relying simply on micro-foundations and (2) DSGE models are not a one-size-fits-all tool and economists must recognize their limitations.
While I agree with many of Dr. Blanchard’s diagnoses, I am not so sure about his proposed solutions. Moving away from exclusive reliance on micro-foundations certainly sounds desirable in theory, but it seems to me that adding the work of other disciplines would increase, rather than decrease, the complexity of DSGE models which would (1) make it more difficult to solve these models equation-by-equation and (2) raise the barriers to entry for understanding the assumptions of any given DSGE model.