Testing Economic Theory Using the Cointegrated Vector Autoregressive Model: New Keynesian Models and House Prices

Abstract

This PhD project investigates a method for taking theoretical models, mainly New Keynesian models for business cycle analysis, to the data. The cointegrated vector autoregressive (CVAR) model is used as a statistical framework, which provides the possibility of testing aspects of theoretical New Keynesian models. Additionally, it may be used as an alternative method for estimating these models, contrary to what is mainly used. In particular, there is no need to alter the non-stationary macroeconomic variables prior to estimation and the theoretical steady state of the model may be tested as restrictions on the long-run relationships of the CVAR model. In chapter two and three, properties of theoretical New Keynesian models are tested in the CVAR model. Chapter two includes a New Keynesian model which explains price- and wage setting, while the New Keynesian model in chapter three includes the housing market. The theoretical model in chapter two is mostly in line with what is found in the data, while there is a gap between the theoretical model in chapter three and the empirical observations. Hence, the CVAR framework is found to be a valuable tool in testing properties of the theoretical models, and may be used in order to assess the empirical relevance of the model before using it for policy analysis.The final chapter finds evidence in favor of a bubble in the US housing market prior to the financial crisis starting in 2008, due to explosiveness in house prices using a theoretical model for the housing market. This may explain the deviations between the model and the empirical observations in chapter three. It also implies that the CVAR model can be used in order to monitor the housing market