Norwegian inflation : an empirical study - 1667-2004

Abstract

This thesis contains an empirical investigation of several models of Norwegian inflation estimated on annual observations from 1667 to 2004. The approach taken is simple one equation models estimated by OLS or IV. The focus has been on presenting a variety of models instead of an in depth analysis of anyone particular model.

The analysis starts out with the sub sample covering the years prior to 1830 using the death/birth-ratio and temperatures as proxies for supply shocks. In addition we include English inflation, paper money and war dummies as explanatory variables. We find a very strong relationship between Norwegian and English inflation. One possible interpretation is that Norway can be viewed as a small open economy even prior to 1830. The death/birth-ratio is an indicator of the demographic conditions in the society, but is treated as a proxy for supply shocks in this thesis. This interpretation may sometimes be problematic, but arguments are given for its validity. This thesis shows that wars did not contribute to inflation until the mid 18th century and that the introduction of paper money did affect prices, especially during the Napoleonic War when the monetary regime collapsed.

The period post 1830 are covered by three types of models; the inverted money demand function, the P*-model and the Phillips Curve model. The period is divided into two sub samples, one prior to 1914 and one post 1914.

The inverted money demand function shows reasonable properties in both samples. We identify a positive effect of money growth, a negative effect of output growth, a positive effect of interest rates and a positive effect of imported inflation. The main difference between the two sample periods is that the effect of lagged inflation is increasing, the effect of money growth is decreasing and the effect of import prices is increasing. The direction of causality is discussed, and we find some evidence that indicates that it has changed. In the 19th century money affected prices, but in the 20th century the direction of causality seems to be from prices to money.

The P*-model views inflation as a function of lagged inflation and lagged deviation from the equilibrium price level. We find that the model fits data well on the 19th century sample, but estimated on the 20th century sample the model shows signs of misspecification.

The final model is the Phillips Curve. The main focus is on the so called hybrid version of the (Neo-Keynesian) Phillips Curve, including both lagged inflation and expected inflation next period in addition to the output gap which is an indicator of the activity level of the economy. We are not able to detect significant effects of the output gap. More favorable results are provided by using the unemployment rate as the indicator of economic activity. We show that the model can be improved even more by introducing the yield spread as a proxy for inflation expectations.