DO AGGREGATE MEASURES OF MISMATCH MEASURE MISMATCH?A TIME SERIES ANALYSIS OF EXISTING CONCEPTS

This paper discusses the performance of popular mismatch indexes proposed by Lilien (1982), Jackman and Roper (1987), Jackman, Layard and Savouri (1991), and Lambert (1988). Results in the literature show that, with the exception of Lambert’s index, measures of mismatch reveal decreasing or at least stable mismatch for European labour markets. This result contradicts micro evidence which suggests declining mobility. Time series analysis performed in the paper tackles this puzzle, and shows that indices consisting of aggregate time series may lead to false conclusions. Measures of mismatch fail when time series shift upward: Without changing the relative structure between individual groups (regions, skills, occupation), this paper shows that trending time series imply decreasing mismatch for all but one index. The only exception is Lambert’s rho. Here stochastic trends lead to a spurious increase of mismatch. Keywords: Mismatch, labour mobility, time series analysis JEL classification: J60, C22, C43