We estimate the return to education using a sample drawn from the National Longitudinal Survey of Youth 1979 (NLSY79). Rather than accounting for the endogeneity of schooling through the use of instrumental variables we employ a parametric version of the Klein and Vella (2006a) estimator. This estimator bypasses the need for instruments by exploiting features of the conditional second moments of the errors. As the Klein and Vella (2006a) procedure is semi-parametric it is computationally demanding. We illustrate how to greatly reduce the required computation by parameterizing the second moments. Accounting for endogeneity increases the estimate of the return to education by 5 percentage points, from 7.6% to 12.7%.