It is argued that the neoclassical theory of wages has better predictive power than is customarily argued. Wages should closely approximate productivity in analyses involving large samples of workers, and in longitudinal analyses involving extended periods of time. Time and aggregation eliminate many of the market imperfections that make wages unequal to productivity in cross-sectional analyses of individuals. The article also develops a pseudo-market model derived from the work of Nelson and Winters that would allow wages to equal productivity without invoking the marginalist assumptions of continuous perfectly rational wage optimization on the part of firms. Empirical support for the wage productivity link comes from an analysis of the wages of French coal miners from 1900 to 1935.