Résumé : Multidimensional auctions are a natural and practical solution when auctioneers pursue more than one objective in their PPP transactions. However, it is difficult to achieve auction efficiency with multiple award criteria. Using auction data from road and railway concessions in Latin America, the probability of renegotiation is estimated by a two-stage least squares technique with a binary selection in the first-stage regression. It is found that auctioneers tend to adopt the multidimensional format when the need for social considerations, such as alleviation of unemployment, is high. This implies that such political considerations could hinder efficiency and transparency in auctions. It is also shown that the renegotiation risk in infrastructure concessions increases when multidimensional auctions are used. Rather, good governance, particularly anticorruption policies, can mitigate the renegotiation problem.