Inflation and Inflation Uncertainty in Sweden: a GARCH Modelling Approach

Abstract: This essay investigates inflation and inflation uncertainty in Sweden from 1970:Q1 to 2014:Q4. GARCH models are used to generate a measure of inflation uncertainty estimated under the distributional assumption of Student's t-distribution and GED. The preferable model found for Swedish inflation was a EGARCH(1,1) estimated with Student's t-distribution. The coefficient of the asymmetry in the conditional variance of the Swedish inflation series was found to be negative, thereby implying that a positive shock to the Swedish inflation leads to less uncertainty about inflation. Relating these results to the two competing hypotheses about the causal relationship between inflation and its uncertainty, the results in this essay does not support the Friedman-Ball hypothesis. With regards to the Cukierman-Meltzer hypothesis, the results seem to support a version of this hypothesis put forward by Holland (1995), referred to as the "stabilisation hypothesis"