We jointly estimate the preceding system of equations (including the selection model) using Roodman's (2009) conditional mixed-process regression procedure. This enables a simultaneous estimation of all five equations and uses a simulated maximum likelihood algorithm (Geweke 1989; Hajivassiliou and McFadden 1998; Keane 1994) to directly estimate the cumulative higher-order likelihood function (for a similar approach and additional details on the method, see Kashyap, Antia, and Frazier 2012) using a seemingly unrelated regression estimator.