We argue that promoting education may be a means to reduceincome inequality. When workers of different skill levels areimperfect substitutes in production, an increase in the level ofhuman capital in the economy reduces the return to education.Hence, a given compression of after-tax incomes can be achievedat lower marginal tax rates. Optimal redistribution policy faces atrade-off between the distortions of taxes on effort and the distortionsof education subsidies on the investment in human capital.We discuss empirical evidence on three crucial elasticities. Ourargument explains the actual pattern of education subsidies inOECD countries quite well.However, there is an offsetting effect. When education andability are complements, high ability types take up more education.A subsidy to education will then favor these types. Wediscuss the condition for the net effect of education subsidies tobe progressive. The empirical evidence suggest that this conditionis critical for a simple education subsidy. We consider somemore elaborate schemes for education subsidies.This discussion paper has resulted in ch. 4 of href=http://books.google.nl/books?hl=nl&lr=&id=edksC0nRPZYC&oi=fnd&pg=PR7&dq=allesintitel:+%22Labor+Market+Institutions+and+Public+Regulation%22&ots=-umxBkjdnT&sig=oeFDTjhnSeo-w-4nuTjTJpuaUr8@v=onepage&q=&f=false>'Labor Market Institutions and Public Regulation', pp. 123-61, (Jonas Agell, Michael Keen, Alfons Weichenrieder (eds.)), 2004, MIT Press, 228 p.