Frontier estimation with local polynomials and high power-transformed data

We present a new method for estimating the frontier of a sample. The estimator is based on a local polynomial regression on the power-transformed data. We assume that the exponent of the transformation goes to infinity while the bandwidth goes to zero. We give conditions on these two parameters to obtain almost complete convergence. The asymptotic conditional bias and variance of the estimator are provided and its good performance is illustrated on some finite sample situations.