Worker Turnover,Capital Dispersion,and Matching

Abstract. A model acknowledging technology and wage dispersion, search frictions, and costly worker turnover is used for testing the notion of random matching. Using a linked employer–employee data set on roughly 9,000 Norwegian establishments and 200,000 jobs during the period 1989–95, I show that establishments investing more in capital, pay more, and experience lower worker turnover rate. Strictly convex turnover costs are identified. High‐wage establishments post on average less intensively than low‐wage establishments. Positive relationships between wages and posting are observed for high‐tech industries and in the capital and surroundings. Thus, the notion of random matching is generally rejected.