The iterative least squares method for estimating panel models with unobservable factor structure is extended to cover the case where the number of factors is unknown a priori. The proposed estimation algorithm optimizes a penalized least squares objective function to estimate the factor...