Modelling technical efficiency of firms under one-step and two-step approaches (the case of commercial banks)

Over the past two decades, empirical research on firm cost efficiency employed two alternative approaches within stochastic frontier analysis (SFA), namely one-step and two-step algorithms. How do their results relate to each other on the same sample of firms? To answer this question, we compare cost efficiency estimations obtained via the two alternative methods applied to Russian bank quarterly data for 2005–2015. Banks were grouped ownership-wise into core public banks, other public banks, domestic private banks, and foreign subsidiary banks. Our results suggest that the average score of each group cost efficiency depends materially on the choice of the algorithm. While the two-step approach failed to reveal statistically significant differences across the four bank groups, the one-step alternative did reveal them. Core public banks lead the cost efficiency ranking, followed by other public banks and domestic private banks, while foreign subsidiary banks lag behind.