Firm-financed training: Firm-specific or general skills?

In this article we analyse the specificity and generality of firm-financed training in Norway. Compared to most other OECD countries Norway has a compressed wage structure. According to non-competitive theories of training we should expect to find much firm-sponsored training in such an economy, and furthermore we should expect to find relatively much firm-sponsored general training. The results in this paper suggest that firm-financed training in Norway contain much general skills. We find that training paid for by previous employers has a positive effect on current wages, and the effect is at least on par with the impact on wages from training paid for by the present employer. We use two methods to control for selection bias in training; an instrument variable (IV)-approach and a fixed-effect approach. The IV-approach suggests that the original training estimate is biased downward. However, our training variable may be subject to measurement error, and recent research has shown that the IV-estimate will be biased upward when a mismeasured variable is binary (as in our case). This finding receives support when using a fixed-effect approach. The IV-estimate for training considerably exceeds the fixed-effect estimate. The fixed-effect estimate is also lower than the original OLS-estimate indicating that some selection bias in training is present. Copyright Springer-Verlag 2004