On-the-job Training: Investment or Compensation?

I analyze the rm's decision of how much general but informal training to provide in the presence of on-the-job search. In particular, due to lack of formal measure and commitment technology on the training intensity, I assume that rms exibly adjust their training intensity at any time, but hardly charge the cost of training to their workers. Even without any contractual agreement on training, rms still provide training not only to improve productivity (investment motive) but also to keep their workers from defecting to other rms (compensation motive). Furthermore, better paying rms are encouraged to provide more intensive training by investment motive, while they are discouraged by compensation motive. The implied wage-training pattern among ex-ante identical rms is reverse U-shaped. That is, training intensity rises with wage payment at low wages, but declines at high wages. Also, as the search friction is mitigated, the investment motive is deteriorated, but the compensation motive is intensi ed. I also show that the training intensity per trainee declines with the accelerated o er arrival rates in the numerical experiment with reasonable parameter values. JEL Classi cation: D83, E24, J24, J33, J64 ∗I am indebted to Hidehiko Ichimura, John Kennan, Rasmus Lentz, Hideo Owan, and Chris Taber for their helpful comments and warm encouragement. In particular, I deeply thank Rasmus Lentz and Dale Mortensen for their useful guidance. All remaining errors are mine.