Neoclassical growth theories, with only capital and labour in the production function, predict decreasing returns to capital and that poor countries grow more easily than rich countries, but empirical studies attribute substantial growth effects to unexplained technological progress. `New' growth theories incorporate the effects of technological progress. For example, investment in human capital may raise not only trainees' productivity, but also that of their co-workers, if they become more productive through working in a productive environment.

In Discussion Paper No. 714, Research Affiliate Rudolf Winter- Ebmer tests for the `external effect' of education on wages using 1983 Census data for Austria. Defining groups of workers by two- digit industries, he separates the direct and external effects of education by estimating wage equations for individual workers, controlling for individual education, job experience and 38 other variables that might influence individual wages, and 25 dummies to capture industry-specific effects. He finds industry wage differentials with a standard deviation of 5.5%, which are explained by industry-specific variables such as concentration ratios, union density, injury rates and investment ratios, but the <MI>average<D> lengths of schooling and job experience in each industry consistently had a positive impact on its wage premiums. Winter-Ebmer tests two explanations on sub-samples of data for skilled and unskilled workers: first, that workers with high unobserved abilities migrate to industries where these are highly rewarded; second, that observed abilities entail different productivities in different industries. Neither test can refute the main result that education has external effects.