Essays on the evaluation and estimation of the heterogeneity of price stickiness in a DSGE model

• Main Author: Jiao, Jing
• Published: Cardiff University 2012
• Subjects: HB Economic Theory
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.567301

The ‘New Keynesian’ model assumes that prices and wages are in an extreme ‘sticky’pattern. In this model, the ssumption that a lagged indexation scheme increases the persistence of inflation is in widespread used; however, in reality, this ad hoc indexation setup is inconsistent with the real data. Moreover, there is extensive evidence on micro price data indicates that heterogeneity in price stickiness is a commonly found feature of price setting throughout the Euro area. Therefore, this thesis aims at incorporating this micro price evidence in an elaborated New Keynesian Dynamic Stochastic General Equilibrium (DSGE) model by using a Generalised- Taylor-Economy (GTE) and Generalised-Calvo (GC) price settings. This thesis first presents the models, which are an extension of Smets-Wouters (SW) model (2003) which replaces Calvo with indexation price setting with heterogeneous price settings. In these new price settings, the micro evidence of heterogeneous price stickiness is directly emerged into macro DSGE models. The findings suggest that heterogeneous price stickiness can generate long-lived inflation and output persistence. Indirect inference is then used to evaluate the DSGE models of the French economy under different price settings. The results of the testing show that all models with differentprice settings are comprehensively rejected. The models are then estimated with Bayesian techniques as SW (2003) by using seven key macroeconomic observables. The results show that the GC model has the best performance. The rankings of the different price setting models are also proven to be robust to different priors and observables. Indirect inference evaluations are then conducted based on Bayesian estimated models, and all models are rejected. Indirect inference is then used as an estimation method. The testing results are improved on all models. The GC model is still considered to be the best performance model among all of the different price setting models.