Comparing the fit of New Keynesian DSGE models

by Jan ČAPEK


JEL classification

  • Multiple or Simultaneous Equation Models: Time-Series Models; Dynamic Quantile Regressions; Dynamic Treatment Effect Models; Diffusion Processes
  • Model Evaluation, Validation, and Selection
  • Forecasting Models; Simulation Methods


Forecast quality, global Sensitivity Analysis, model fit, Bayesian posterior odds ratio, parameter importance


The paper is focused on an analysis of model fit of Dynamic Stochastic General Equilibrium (DSGE) models following New Open Economy Macroeconomics (NOEM). Unlike most of the literature on the topic, this paper does not use Bayesian posterior odds ratio to analyze model fit to data; it uses alternative tools instead. In order to compare the results of the alternative tools to the standard posterior odds ratio, this paper uses the findings of Slanicay and Vašíček (2009), who compared model fit to data of several models with the tool Bayesian posterior odds ratio. The goal of the paper is to verify the results of Slanicay and Vašíček’s (2009) model variants with different criteria than posterior odds and to compare the results with findings of their paper. The tools for the analysis are criteria based on root mean squared error (RMSE) and tools from the Global Sensitivity Analysis toolbox. Conclusions of this paper are the following: Habit persistence in consumption is found to be important and price indexation unimportant as in Slanicay and Vašíček (2009). Furthermore, model variants with foreign economy modeled as AR1 processes always perform better than the ones with structurally modeled foreign economy. This finding is in contradiction to the results of Slanicay and Vašíček (2009).