The Impact of Monetary Policy on Unemployment: A Comparison of Local Projections and SVAR Analyses from Norway

Abstract

This thesis concerns the transmission of monetary policy shocks on the unemployment rate in Norway over the past 20 years. The study compares impulse response functions from local projections - estimated through OLS and IV regressions - and SVAR methodologies - identified with short-term restrictions and external instruments. While touching upon the identification problem, this study contributes to existing literature by providing a comparison framework for how identification strategies affect model results.

I note a delayed response of the unemployment rate to normalized policy shocks through both methodologies. Peak impact is observed after a period of 10 quarters, while quantitatively the response is of a small size. The response is similar for a local projection specified as in Jordà (2005) and for a two-variable SVAR model specified as in Sims (1980). The use of external instruments amplifies the response in local projections, but causes a loss of statistical significance in the SVAR responses. I extend the model to consider dependencies of the impact of monetary policy changes on the current level of interest rates or unemployment rates. As a measure of robustness, I provide estimates for polynomial local projections, and a five-variable SVAR model that includes GDP, the implicit price deflator, and exchange rates - allowing for various lag selections and variable ordering.