<p>This article proposes a semiparametric discrete choice model that incorporates a nonparametric specification for income effects. The model allows for the flexible estimation of demand curvature, which has significant implications for pricing and policy analysis in oligopolistic markets. Our estimation algorithm adopts a method of sieve approximation with shape restrictions in a nested fixed-point algorithm. Applying this framework to the Japanese automobile market, we conduct a pass-through analysis of feebates and merger simulations. Our model predicts a higher pass-through rate and more significant merger effects than parametric demand models, highlighting the importance of flexibly estimating demand curvature.</p>