Evaluation and indirect inference estimation of inattentive features in a New Keynesian framework

We test the standard New Keynesian (NK) Dynamic Stochastic General Equilibrium (DSGE) model under the condition with and without inattentive features, where inattentiveness is modelled in the form of sticky information and imperfect information data revision. All models are tested with the Indirect Inference method, and our test result based on real-time data suggests that the model with sticky information passes the test and consistently outperforms the baseline NK model with full information and rational expectation, while the model with imperfect information data revision fails to pass the test. Furthermore, we show that none of the models passes the test when Survey of Professional Forecaster data are used for model evaluation. Overall, our findings provide important implications on the modelling of expectation formation in the DSGE framework.