By Federico Guglielmo Morelli, Michael Benzaquen, Jean-Philippe Bouchaud and Marco Tarzia
We study a self-reflexive DSGE model with heterogeneous households, aimed at characterising the impact of economic recessions on the different strata of the society. Our framework allows to analyse the combined effect of income inequalities and confidence feedback mediated by heterogeneous social networks. By varying the parameters of the model, we find different crisis typologies: loss of confidence may propagate mostly within high income households, or mostly within low income households, with a rather sharp crossover between the two. We find that crises are more severe for segregated networks (where confidence feedback is essentially mediated between agents of the same social class), for which cascading contagion effects are stronger. For the same reason, larger income inequalities tend to reduce, in our model, the probability of global crises. Finally, we are able to reproduce a perhaps counter-intuitive empirical finding: in countries with higher Gini coefficients, the consumption of the lowest income households tends to drop less than that of the highest incomes in crisis times.
If you also wondered what a self-reflexive DSGE model is, here is the lowdown. This is a mixture of HANK (Heterogeneous Agent New-Keynesian) models and ABM (Agent-Based Model). You have the general equilibrium, the heterogeneity and the dynamics of HANK, but some aspects of heterogeneity are fixed (say, skills, earnings, market structure) but new dimensions are added, such as social interaction. A household’s consumption preferences may be determined by the consumption of its neighbors. This is not quite like the habit formation models (or “catch-up-with-the-Joneses”), where the reference is aggregate consumption, here a network with local interactions is built. Of course, things are going to depend on how the network is set up, and for this particular paper, how rich and poor interact. Potentially crucially, these interactions do not change as the economy slides into a crisis. Also, the parameterization of the network appears to be impossible, thus results are provided for all possible values, showing that almost anything is possible. One would need to narrow this down to give some valuable conclusions.
I am surprised that you are promoting ABM here. As you mention, this literature has so many free degrees of freedom that it can conclude anything, hence no relevance whatsoever. Until ABM can get disciplined by data, it is a waste of resources.
I of course totally disagree that ABM is a waste of resources. I could equally say the same of GE models. Scientific progress is incremental and one must be allowed to venture into untreaded paths.
On the comment that “almost anything can happen” in ABMs this is exactly what I like about them. A good model is a model that does not spit out what you have put in, like GE models do.
The explosion of data will allow one to calibrate ABM better and prune down the garden of bifurcating paths. Until then, one should study ABMs in depth to learn how to use them, get accustomed to them — like with all new tools (think of Quantum Mechanics at its birth).
As a last note, ideas coming from systems biology will help pruning down such big models in a disciplined way — see e.g. our paper https://arxiv.org/pdf/2111.08654.pdf and refs. therein