Recent work examining a constructed hourly wage measure indicates gender earnings equality among US physicians. Yet, physicians are not typically paid by the hour and the constructed measure fails to capture the true influence of hours worked and presents a fallacy of composition. When the earnings specification replaces the constructed wage measure with annual earnings or is modified to properly include hours worked, earnings equality vanishes and large gender differentials reappear.

Purpose -This paper aims to determine the characteristics of board certification among US physicians and to test whether accounting for the expected gains to certification alters the pattern of the determinants of board certification. Design/methodology/approach -Splitting the sample into sub-samples by characteristics associated with certification/non-certification identified in a probit, the incremental gain to certification from log-earnings equations is identified. Realizing that these methods are susceptible to sample selection, correction is made for it using the Heckman approach. Using the sample selection corrected equations, the expected gain to certification among those who certify is then predicted and those who do not certify is then predicted and this difference is included as a proxy for the expected gain in the original probit to ascertain whether including the expected gain alters the determinants of certification. Findings -Accounting for the expected gain alters the pattern of the determinants of certification. Although some groups such as blacks appear less likely to certify, after accounting for their expected return to certification, they are not as less likely. This is explained in terms of the expected marginal return to certification, market structure and practice setting.Research limitations/implications -The data used in the analysis apply only to young physicians in the USA. Also, these results may be applicable only to the particular cohort used in this analysis. Practical implications -The findings help to explain the absence of minority board certified physicians within the USA. Originality/value -This paper is the first to simultaneously estimate the returns to physician board certification and the decision to obtain certification. IntroductionThis paper simultaneously estimates the returns to physician board certification and the decision to obtain certification. We confirm that those physicians with the highest expected returns to certification are more likely to obtain certification. These physicians operate in competitive environments, are married, white, experienced and within a specific subset of specialties. Critically, we identify the importance of sample selection on certification in estimating physician earnings and show that accounting