We investigate whether microfoundations might increase the predictive power of macroeconomic models of wage inflation. By comparing past predictions to observed values, we find that the Phillips curve with the average unemployment rate in districts with prevalently low unemployment rates delivers more accurate predictions of aggregated wage inflation than the Phillips curve with the overall unemployment rate. The identification of specific groups of districts is based on our estimates of the wage curve at the regional level, i.e. the relationship between the regional level of wages and regional unemployment. Real wages adjust to changes in local unemployment in districts with low unemployment rates, a low share of public sector employment, and for the short-term unemployed. On the other hand, the welfare system might represent a floor preventing downward wage adjustments in districts with high unemployment rates and for the long-term unemployed. In the public sector, wages are negotiated at the economy-wide level, while the variance in regional unemployment does not play a role.
Keywords: Panel data, partial adjustment model, Phillips curve, unemployment, wage curve
Issued: April 2003
Download CNB WP No. 1/2003 (pdf, 527 kB)