From scratch to efficiency gains after a financial crisis? A tale of a restructured banking system

Abstract

We study the efficiency of the Argentine banking system after the 2001–2002 crisis. The financial system had to be restructured from scratch and recovered jointly with the economy, but its productivity and average cost levels have been stagnant since 2007. The analysis includes efficiency frontier estimations for retail banks and a comparison of subsamples for different categories of banks for the period 2005–15. We try to determine whether public banks are more efficient than private ones, whether privatized are more efficient than always private, as well as national versus foreign entities. Our findings show a modest average efficiency of the system and quite similar efficiency rankings for the different groups of banks. On average, public tend to be slightly more cost efficient than private, and national are slightly more efficient than foreign.

Keywords:

• Banks
• costs
• efficiency

JEL CLASSIFICATIONS:

• G21
• L51

Acknowledgments

Sonia Leon thanks financial support she received from Fundación UADE.

Notes

1. The currency board system was named after the 1991 Convertibility Law, by the norm which established a pegged exchange regime and a 1/1 local currency backed with US dollar reserves in the Central Bank. The name ‘Convertibility’ recalled a quite similar crisis the country suffered a century ago and the instrument –pegging to Gold Standard and full convertibility of local currency- which in 1890 permitted the recovery and a quarter of century of prosperity.

2. 18 banks were privatized, but after privatization, some were merged into others or closed. Currently, remains 5 privatized banks.

3. The test is based on the calculus of –where L_R is the log-likelihood of the restricted model and L_U is one of the unrestricted models– which is distributed as a chi-square with degrees of freedom equal to the number of restrictions. The null hypothesis specifies the restricted model.

4. The Akaike and Bayesian indexes are information criteria to measure the model fit to data and are used to select from alternative models. Both criteria penalize the loss of degrees of freedom (or the increase in the number of parameters to estimate). The best model is the one with the lowest value for each criterion.