ABSTRACT We examine the evolution of competition and efficiency in the U.S. banking industry over the past two decades by employing a novel two-step nonparametric method to estimate cost efficiency and the Lerner index for banks with multiple outputs. In the first step, data envelopment analysis is used to estimate the cost efficiency and the cost frontier. In the second step, the nonparametric local linear method is employed to smooth the estimated cost frontier, obtaining estimates of the marginal cost for each output and the Lerner index for each bank. Our innovative two-step nonparametric method has shed light on the evolving landscape of competition and efficiency that the ongoing consolidation in the U.S. banking sector is accompanied with the increasing trend of the concentration and the decreasing trend of efficiency. Further, small banks are more willing to set prices above their marginal costs but may not gain as much profit from a cost efficiency perspective. This phenomenon is a long-standing and enduring characteristic of the U.S. banking industry, reflecting the diverse strategies and priorities of banks operating in different market contexts.