The geography of Spanish bank branches

Abstract

This article analyzes the determinants of bank branch location in Spain taking the role of geography explicitly into account. After a long period of intense territorial expansion, especially by savings banks, many of these firms are now involved in merger processes triggered off by the financial crisis, most of which entail the closing of many branches. However, given the contributions of this type of banks to limit financial exclusion, this process might exacerbate the consequences of the crisis for some disadvantaged social groups. Related problems such as new banking regulation initiatives (Basel III), or the current excess capacity in the sector add further relevance to this problem. We address this issue from a Bayesian perspective, using a Poisson regression model within the framework of generalized linear mixed models. This proposal allows us to assess whether over-branching or under-branching has taken place. Our results suggest, among other findings, that both phenomena are present in the Spanish banking sector, although the implications for the three types of banks in the industry, namely commercial banks, savings banks or credit unions, vary a great deal.

Keywords:

• bank
• Bayesian statistics
• branch
• municipality

AMS Subject Classification::

• C11
• G21
• R1

Acknowledgments

All authors thank two anonymous referees and participants at the 4th IFABS Conference (Valéncia, Spain, 2012) for helpful comments, as well as the Ministerio de Educación y Ciencia (ECO2011-27227, MTM2010-19528 and MTM2013-42323-P, jointly financed with European Regional Development Fund), the Generalitat Valenciana (PROMETEOII/2014/046, ACOMP11/218, ACOMP/2013/152 and ACOMP/2014/283), the Universitat Jaume I (P1-1B2014-17, P1-1B2013-06) and the Ivie (Instituto Valenciano de Investigaciones Económicas) for funding this research initiative.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Real Decreto 1582/1988.

2. The recent study by Alamá and Tortosa-Ausina [1] provides an extension to the one by Bernad et al. [4] by enlarging the set of socioeconomic variables included in the analysis, considering a different approach. Both Bernad et al. [4] and Alamá and Tortosa-Ausina [1] use the same database, developed by the foundation of the Spanish savings bank La Caixa (http://obrasocial.lacaixa.es), which provides information at a municipal level. However, it does not take into consideration the exact geographical component of the location of bank branches. Therefore, it excludes the information on the precise location of each bank branch.

3. http://www.maestre-ediban.com.

4. Other authors such as Bernad et al. [4] were actually able to use per capita income itself because of confining the analysis to municipalities with population over 1000 – for which information is available. In our case, since we carry out the analysis for all Spanish municipalities, we were impelled to use unemployment given its accuracy as a proxy. However, it has some limitations, among which its inability to reflect the shadow economy is one of the gravest. This is particularly important in Spain, where the shadow economy has been traditionally high and has gained further importance during the crisis years.

5. Ministerio de Empleo y Seguridad Social. Tesorería General de la Seguridad Social. Servicio Público de Empleo Estatal.

6. http://internotes.cajaespana.es/pubweb/decyle.nsf/.

7. Another relevant issue is related to the dynamic interactions between unemployment, financial exclusion and the availability of financial services. Given the strong changes in the Spanish banking sector from a regulatory perspective which are affecting many banking firms and their presence in many marginalized locations, we consider this question deserves a specific analysis which lies beyond the boundaries of our study. Specifically, whereas a relevant literature has analyzed the links between financial development, in terms of more bank branches, and economic growth [8, 16, 17], the phenomenon of how a decreasing number of bank branches affects economic growth and economic development has never been dealt with by the literature.

8. Other information considered in the literature that can affect bank branch location is related to economic activity in general, and can be decomposed into different variables such as tourism, commercial activities (both retail and wholesale), and number of trucks. However, in the particular case of Spain, we are dealing with, this information is missing for municipalities with fewer than 1000 inhabitants.

9. NUTS stands for Nomenclature of Territorial Units for Statistics. See the web page http://epp.eurostat.ec.europa.eu/portal/page/portal/nuts_nomenclature/introduction for more information. Municipalities corresponded to NUTS level 5 until 2003. Since then, both levels 4 and 5 have been replaced by Local Administrative Units (LAUs). In the case of Spain, provinces (NUTS level 3) correspond to LAU level 1, and municipalities correspond to LAU level 2.

10. In the case of rival precedence, Hannan and McDowell [13] found that by adopting automatic teller machines, rivals increased the conditional probability of a banking firm applying the same innovation. Rational herding refers to situations where it is individually rational for agents/firms to mimic the actions of others, despite it might lead to suboptimal aggregate outcomes. Related to this, Chang et al. [7] found that the apparent clustering of bank branches in New York City between July 1990 and June 1995 might be partially attributable to rational herding by banks.