We combine the augmented Solow model with the Mincer equation to derive a specification that identifies an education externality within a production function framework. The previous empirical literature has not reached a consensus about the size of the education externality, which is given by the difference between the microeconomic and the macroeconomic return to education. Relative to our benchmark value that is based on a parameterization of the derived specification, we find that the estimated education externality is too large when the empirical model is not properly restricted, and appears to be absent when all control variables of the empirical model are properly accounted for. We note that the absence of an education externality would be inconsistent with observed levels education subsidies.