Assessing the Empirical Performance of the DSGE models in the lead up to the Crisis

The global financial crisis has sparked renewed debate over the state of macroeconomic modeling, particularly in the lead up to the 2008/2009 Great Recession. The standard workhorse of macroeconomic modeling, the Dynamic Stochastic General Equilibrium (DSGE) model, has been subject to intensive scrutiny. Over the past decade, there has been significant increase in the use of DSGE models by central banks for policy analysis, forecasting and prescriptions. The majority of central banks from developed countries have established DSGE models, including the Federal Reserve Bank, the European Central Bank, the IMF and the Bank of England. Given their prevalence among central banks coupled with their use by policy makers for analysis and forecasting, the objective of this research paper is to assess the behavior and forecasts made by these DSGE models in the run up to a financial crisis. A DSGE model is estimated for the United States for the pre-crisis period Q1.1947 to Q4.2007. An empirical verification of the data is undertaken, whereby forecasts made by the DSGE models are compared with the observed post-crisis data. We find that the DSGE model does a poor job of forecasting the Great Recession, and gives no indication that a downturn is imminent in the economy. Within the current paradigm, there is no role for financial frictions. As such, we suggest that the building blocks of DSGE models are too simplistic to effectively model key dynamics within the economy. We use the role of debt accumulation by US households as a means of illustrating this. The global financial crisis has sparked renewed debate over the state of macroeconomic modeling, particularly in the lead up to the 2008/2009 Great Recession. The standard workhorse of macroeconomic modeling, the Dynamic Stochastic General Equilibrium (DSGE) model, has been subject to intensive scrutiny. Over the past decade, there has been significant increase in the use of DSGE models by central banks for policy analysis, forecasting and prescriptions. The majority of central banks from developed countries have established DSGE models, including the Federal Reserve Bank, the European Central Bank, the IMF and the Bank of England. Given their prevalence among central banks coupled with their use by policy makers for analysis and forecasting, the objective of this research paper is to assess the behavior and forecasts made by these DSGE models in the run up to a financial crisis. A DSGE model is estimated for the United States for the pre-crisis period Q1.1947 to Q4.2007. An empirical verification of the data is undertaken, whereby forecasts made by the DSGE models are compared with the observed post-crisis data. We find that the DSGE model does a poor job of forecasting the Great Recession, and gives no indication that a downturn is imminent in the economy. Within the current paradigm, there is no role for financial frictions. As such, we suggest that the building blocks of DSGE models are too simplistic to effectively model key dynamics within the economy. We use the role of debt accumulation by US households as a means of illustrating this.non-peer-reviewe