This paper develops and applies a new semi-parametric correction for sample-selection in the
context of a multi-market Roy model of mobility and earnings. Instead of workers choosing occupations
as in Roy's paper, this paper formulates a model where individuals choose which of the 50 states in the
U.S. (plus the District of Columbia) to live and work in. The new econometric methodology combines
Lee's (1982) parametric "maximum order statistic" approach to multi-choice selection models with Ahn
and Powell’s (1993) more recent work on "single-index" models. The resulting correction requires no
assumptions on the joint distribution of the error terms in the outcome and multiple selection equations
and can easily be adapted to a variety of other polychotomous choice problems. The empirical work,
which uses Census data for I980 and 1990, confirms the role of comparative advantage in mobility
decisions. The results suggest that self-selection of higher educated people to states with higher returns
to education generally leads to downward biases in the returns to education in state-specific labor
markets. I also find that state-to-state migration flows respond strongly to differences in the return to
education and amenities across states.