DSGE Models in Central Banks: Progress and Prospects

From empirical fit to internally consistent counterfactual analysis and richer transition mechanisms

Non-Technical Summary

Central banks need models not only to forecast the economy, but also to assess how economic outcomes may differ across alternative policy and economic scenarios. This requires counterfactual analysis: understanding how the economy might evolve under a different path for interest rates, a change in fiscal policy, or an intervention in financial markets. This article explains why Dynamic Stochastic General Equilibrium (DGSE) models have become an important tool for conducting such analyses. By describing how households, firms, financial markets, and policymakers interact, these models provide a coherent framework for understanding how economic shocks and policy decisions affect the economy.

DSGE models emerged partly in response to the limitations of the large macroeconomic models used in the 1960s and 1970s. Those models could reproduce historical relationships reasonably well, but these relationships could change when policy itself changed. Modern DSGE models instead make explicit how households and firms respond to their economic environment and to policy. This makes them particularly useful for policy experiments in which past experience alone may provide limited guidance. The timeline below illustrates how the framework subsequently evolved as new economic and policy challenges emerged.

Over time, the framework has evolved with the questions faced by central banks. Models have been extended to incorporate, among others, unemployment, international linkages, financial markets, fiscal policy, and unconventional monetary policies. The global financial crisis, for example, highlighted the importance of banks, credit conditions and financial balance sheets, leading these mechanisms to play a much larger role in subsequent models.

More recently, heterogeneous-agent models have addressed another important limitation of traditional DSGE models: households are not all affected by economic policies in the same way. Borrowers and savers, wealthy and liquidity-constrained households, or workers with different income risks can respond very differently to changes in interest rates, taxes or transfers. Heterogeneous-agent New Keynesian (HANK) models incorporate these differences and therefore allow policymakers to study not only the aggregate effects of a policy, but also how those effects depend on who gains income, who loses it, and who adjusts spending the most.

The main lesson is therefore not that one generation of models should simply replace another. DSGE models remain valuable because they provide a common language for organizing economic arguments and checking that policy scenarios are internally consistent. But their conclusions inevitably depend on the assumptions, data and mechanisms built into them. In practice, central banks therefore benefit from using a portfolio of models: relatively simple and fast models for routine analysis, richer models when financial or distributional mechanisms are central, and other empirical tools and institutional judgement alongside them.

Keywords: DSGE Models; Central Banks; New Neoclassical Synthesis Models; HANK; Monetary Policy; Policy Analysis
Codes JEL : E32, E52, E58, C11, D14, D31