If labor markets operated entirely frictionless, productivity premiums associated with different worker characteristics would equal the wage premiums earned by workers possessing those characteristics. Using matched employer-employee data from the manufacturing sector of three sub-Saharan countries, we evaluate to what extent the two premiums differ for four characteristics that are clearly related to human capital: schooling, training, experience, and tenure. Equality holds strongly and even surprisingly well for firms in Zimbabwe (the most developed country in the sample), but not at all in Tanzania (the least developed country), while results in Kenya are intermediate. Where equality fails, the pattern is for general human capital characteristics (schooling, experience) to receive a wage return that exceeds the productivity return, while the reverse applies to more firm-specific human capital characteristics (training, tenure). Schooling tends to be over-rewarded, even though large productivity gains are consistently associated with formal employee training programs. Wages tend to rise with experience, while productivity gains are mostly associated with tenure. We demonstrate the remarkable robustness of the findings controlling, among other things, for sampling errors, nonlinear effects, and non-wage benefits. Localized labor markets and imperfect substitutability of different worker-types provide a partial explanations for the estimated gap between the wage and productivity premiums.