The random parameters stochastic frontier cost function and the effectiveness of public policy: Evidence from bank restructuring in Mexico

•We use random parameters stochastic frontier models to estimate cost efficiency.•These models tackle the problem of confounding heterogeneity with inefficiency.•Rank order efficiencies are consistent across alternative models.•Foreign acquisition does not yield significant changes in bank cost efficiency.•Consolidation of domestic banks creates significant long-term efficiency gains.

We apply specifications of the random parameters stochastic frontier cost function model to estimate bank efficiency. This class of model appears to resolve the long standing problem of confounding inefficiency and heterogeneity. Mean cost efficiencies from random models are higher by as much as eleven percentage points compared to pooled OLS estimates. Whilst tests show efficiencies are not drawn from the same population, rank order efficiencies are strongly associated. In a second step, we employ the estimated efficiencies to determine the effect of foreign acquisitions on bank cost efficiency following legislative reforms made as part of Mexico's bank restructuring programme in 1995. Foreign bank acquisition does not significantly affect efficiency whereas consolidation of local banks yields significant long-term improvements in efficiency. We recommend random parameters stochastic frontier models since they better accommodate heterogeneity and produce more precise estimated efficiencies.