A dynamic stochastic general equilibrium (DSGE) model is a type of macroeconomic model that describes the whole economy as the outcome of decisions by households, firms and policymakers who plan over time and form expectations about the future. "Dynamic" means the model tracks how the economy evolves period by period; "stochastic" means it is driven by random shocks to technology, demand, prices or policy; and "general equilibrium" means all markets are solved together, so a change in one feeds through to the others.
DSGE models became the main framework of academic macroeconomics and a standard tool at central banks from the late 1990s onwards. They sit beside older large-scale econometric models and simpler statistical forecasting tools, and they are closely tied to the ideas of rational expectations, the Taylor rule and inflation targeting.
Origins in the Lucas critique
The approach grew out of Robert Lucas's 1976 critique of the large macroeconometric models of the time. Lucas argued that relationships estimated from past data, such as a stable trade-off between inflation and unemployment, would shift when policy changed, because people adjust their behaviour to the new policy. A model useful for policy analysis therefore had to be built from "deep" parameters, such as preferences and technology, that do not change with policy.
Finn Kydland and Edward Prescott answered this in 1982 with the first real business cycle model, in which fluctuations came from technology shocks hitting a perfectly competitive economy. They shared the 2004 Nobel prize in economics partly for this work. The real business cycle model left no role for monetary policy, which led economists to add frictions.
New Keynesian DSGE models
The New Keynesian versions that dominate policy use add sticky prices and wages, monopolistic competition among firms and a central bank that sets the policy interest rate. Because prices adjust slowly, changes in the nominal rate move the real rate and so affect spending and output. The small textbook version reduces to three equations: a demand (IS) curve linking the output gap to the expected real interest rate, a New Keynesian Phillips curve and a policy rule. The Phillips curve is:
- is inflation in period , and is the inflation expected today for the next period.
- is the household discount factor, a number slightly below 1.
- is the output gap, and measures how strongly inflation responds to it.
With , expected inflation of 2%, and an output gap of 1%, inflation is . The equation shows why expectations matter: firms that cannot reprice every period set prices with future inflation in mind. Medium-scale models, such as those built by Smets and Wouters for the euro area, add investment, habit formation, wage stickiness and many more shocks, and are estimated with Bayesian methods.
Use at central banks
Central banks use DSGE models for forecasting, for telling a consistent story about why the economy is moving, and above all for policy experiments: comparing interest-rate paths, studying forward guidance or tracing the effects of a supply shock or demand shock. The European Central Bank (ECB), the Bank of England and the Federal Reserve system have all developed DSGE models, usually alongside semi-structural and statistical models rather than as their only tool. They form one part of the wider toolkit of macroeconometric models.
Criticism after 2008
The Global Financial Crisis (GFC) exposed weaknesses. Most models in use before 2008 had no meaningful banking sector, no defaults and no role for asset-price bubbles, so they could not anticipate or explain a crisis that began in the financial system. Critics also pointed to:
- The representative agent. Many models assume one typical household, which hides inequality, borrowing limits and the way different groups respond to policy.
- Linearisation. Models are usually solved around a steady state, which handles small shocks well but large ones, or the zero lower bound, poorly.
- Many free shocks. Estimated models may need numerous unexplained shocks to fit the data, which weakens their claim to explain the economy from first principles.
The response has been to extend rather than abandon the framework. Newer models add financial frictions and bank balance sheets, heterogeneous households (HANK models) and occasionally binding constraints. Supporters argue that DSGE models remain the clearest way to keep a policy analysis internally consistent; sceptics argue that simpler, more data-driven models forecast as well and are easier to explain.
Ref 1100001778