New jobs or recalls?

This study demonstrates the importance of explicitly accounting for the possibility of recalls in the analysis of unemployment composition and the determinants of unemployment spell durations. Using Austrian administrative data we estimate that recalls accounted for nearly one half of the employment to unemployment to re-employment transitions in 1985 with the probability of recall being mainly dependent on industry and job characteristics related to seasonal work. We then analyse unemployment spell durations in a competing risks framework and, indeed, find significantly different duration patterns across new jobs and recalls.This paper draws on research in which the first author has collaborated with Georg Fischer. We would like to thank seminar participants at the ZEW, Mannheim, the Institute for Advanced Studies Vienna, participants of the European Economic Association congress, Dublin, and two referees of this journal for valuable comments and criticisms.