DSGE and Agent-Based Models in Monetary Policy Analysis: A Comparative Perspective

This paper provides a systematic comparative assessment of New Keynesian Dynamic Stochastic General Equilibrium (NK-DSGE) models and Agent-Based Models (ABMs) as tools for macroeconomic and monetary policy analysis. It addresses three measurable research questions: (i) how the two frameworks differ in their theoretical foundations and behavioral assumptions; (ii) whether ABMs demonstrate empirical or forecasting advantages relative to DSGE models during crisis periods; and (iii) how central banks can integrate both approaches within a complementary modelling toolkit. The methodology combines a structured comparative review of post-2008 central bank applications with an illustrative quantitative simulation of a 100-basis-point monetary tightening shock using a calibrated three-equation NK-DSGE model and a stylized heterogeneous-agent ABM. The simulation results show that the ABM generates a recession nearly twice as deep (−4.8% versus −3.0%) and significantly more persistent than the DSGE counterpart, reflecting nonlinear amplification mechanisms absent in representative-agent structures. Surveyed empirical evidence further indicates competitive or superior forecasting performance of large-scale ABMs during major crises; for example, the Austrian model of Poledna et al. (2023) anticipated a six percent GDP contraction during the COVID-19 crisis and outperformed DSGE and VAR benchmarks in crisis forecasting. While NK-DSGE models retain advantages in normative and welfare-based policy evaluation due to analytical discipline and microfoundations, ABMs provide greater realism in modelling heterogeneity, financial frictions, and systemic instability. The findings support a dual-toolkit strategy in which reformed DSGE models and empirically grounded ABMs are used complementarily within modern central banking practice.