Is agent-based computational economics (ACE) the future of macroeconomics?

Question

DSGE models are subject to a lot of criticism, and I wonder if ACE is an interesting candidate to answer some of this criticism?

In particular, the bottom-up approach of ACE seems more promising and "realistic" than the representative agent. It's also a methodology where knowing the individual characteristics of the agents does not necessarily translate in knowing the characteristics of the system (emergence).

That being said, the main drawback of ACE seems to be its difficulty to get empirically tested.

Answer 1

I would say yes, but expect some pushback initially as always. Many phenomena are emergence characteristics and those can't be tackle with representative agent models. An application I used agent-based modeling for was for comparative statistics of individual learning under different recommender system institutions. I found the method quite useful. For DSGE, a few behavioral patches can close some gaps, but leave quite a bit of space for ACE.

Comments

A "critic" of ACE I always found baffling is that "it's not economics because it's not tractable". I mean, as long as the person using this methodology is clear about its strengths and limitations and the topic is somewhat related to economics, I really don't get this criticism.

I usually understand this criticism as "you're doing something I can't, and I'll try to undermine your methodology/work by attacking its legitimacy as scientific work, instead of trying to honestly improving it" (or why I won't stay in academia after my graduation, episode 3847493).