Leading inflation indicators in Finland : pairwise analysis of Granger-causality and cointegration

Abstract

We analyse a set of macroeconomic variables in order to evaluate their ability to (linearly) predict inflation.A series of tests is conducted in which the consumer price index is paired with a single macroeconomic variable, such as monetary or credit aggregate, an interest rate, an asset price, a survey variable and or some other nominal or real variables.Using bivariate autoregressive models, hypot- heses concerning cointegration and causality between price level and the particular macroeconomic variable are tested. Only in a few cases is cointegration found.We conclude that in general yield indices and cost variables cointegrate and predict inflation over a one-year horizon.Variables concerning economic activity, broad monetary aggregates and certain interest rates predict inflation about one year ahead.However, most of the variables of our dataset perform poorly as leading indicators of inflation.