Aggregation bias in wage rigidity estimation

I argue in this paper that the estimation of wage rigidity using country level data suffers from aggregation bias. Using European data for the years 2000-2017, I find that wages respond less flexibly to changes in unemployment at the regional level, compared to estimation using the same data aggregated at the country level. A possible explanation is that in the European data changes in aggregate unemployment tend to be driven by regions with low unemployment rates, while unemployment in regions with high unemployment rates is less variable and less responsive to aggregate shocks. The relationship between unemployment and wages -the wage curve- is downward sloping and convex. Due to this nonlinearity, the higher variability in lower regional unemployment rates implies higher observed wage flexibility at the aggregate country level, and biased inference. The implication is that wages are even less responsive to changes in unemployment than is observed in aggregate data and commonly assumed in macro-economic models, such that for example fiscal stimulus would lead to less wage inflation than anticipated.