The need for a diverse approach to macroeconomic modelling
Haldane and Turrell argue that the real problem with the DSGE models is not that they did not forecast the Global Financial Crisis but that they didn’t even signal the possibility that a crisis could arise endogenously.
Drawing on the last seven decades of simulation, macroeconomic agent-based models are one approach with promise. In these models, aggregate behaviour emerges from the behaviours of individual agents and, in this sense, they are very much in the spirit of the “microfoundations revolution”, even if the modelling philosophy is different to that of the dominant macroeconomic models. Having already proven their worth in partial (dis)-equilibrium settings both in finance and the natural sciences, agent-based models are well placed to allow researchers to study heterogeneity, networks and crisis dynamics within a macroeconomic context. There is work to be done before macroeconomic ABMs enjoy the same standing in mainstream economics as their cousins.
Drawing on different disciplines: macroeconomic agent-based models
Authors: Andrew G. Haldane, Arthur E. Turrell
From: Bank of England
The calibration issue
Agent-based models face mounting criticism which mainly focuses on the rigor of current validation and calibration practices. In this paper, Donovan Platt compares a number of prominent ABM calibration methods and concludes that Bayesian estimation, though less popular in the literature, consistently outperforms frequentist.
A Comparison of Economic Agent-Based Model Calibration Methods
Author: Donovan Platt
From: Oxford University
