ABM

The agent-based model (ABM) is used for large-scale simulation of the Austrian economy, modelling the dynamic interactions of companies and households at the individual level. Unlike traditional equilibrium models, it captures how economic shocks propagate through networks of heterogeneous households and companies, generating endogenous feedback between individual decisions and macroeconomic outcomes. The high granularity of the ABM enables policy analysis that accounts for distributional consequences of economic policy.

The ABM is a macroeconomic model calibrated using microeconomic data, such as firm balance sheets and household surveys. Macroeconomic ABMs relax key assumptions of models with explicit equilibria, such as representative agents and market clearing, by simulating dynamics arising from interactions among many heterogeneous agents. The macroeconomic ABM developed by Poledna et al. (2023) offers a more detailed representation of the economy than sectoral or aggregated CGE, DSGE, or HANK models, while matching DSGE and other statistical approaches in forecasting accuracy. This combination of detail and predictive performance makes ABMs well-suited for simulations in areas such as monetary policy and inflation, fiscal and tax policy, as well as labour market dynamics.

Macroeconomic agent-based modelling grounded in microeconomic data represents a recent methodological advance in policy simulation, enabled by the growing availability of comprehensive microdata. The global trend toward opening administrative data for research, together with advances in computing power and statistical methods, is likely to increase interest in large-scale simulation. These developments also respond to growing demand from policymakers, who increasingly require distributional impact assessments as a standard component of policy evaluation.

The efforts at WIFO and collaborating institutions have significantly boosted the international visibility of ABM research, with Vienna emerging as a leading hub. The Bank of Canada became the first central bank to integrate an agent-based model into its suite of policy analysis tools, alongside traditional DSGE frameworks (Hommes et al., 2024) and other econometric models. Following this example, central banks are showing increasing interest in adding macroeconomic ABMs to their modelling portfolios.