[en] The problem to decide which patient-donor pairs in a kidney exchange program should
undergo a cross-match test is modelled as a two-stage stochastic optimization problem. We
give an integer programming formulation of this so-called selection problem, and describe
a solution method based on Benders decomposition. We extensively test various solution
methods, and observe that the solutions, when compared to solutions found by recourse
models, lead to an improvement in the expected number of transplants. We also investigate
the computational efficiency of our approach as a function of different parameters, such as
maximum cycle length and the presence of altruists.