A flexible cost function model with risk

Subal Kumbhakar, Ana Lozano-Vivas, Kai Sun

Research output: Preprint or Working paperWorking paper

Abstract

In examining bank cost efficiency in banking inclusion of risk-taking of banks is very important. In this paper we depart from the standard modeling approach and view risk intimately related to the technology. Thus, instead of controlling for risk by viewing them as covariates in the standard cost function we argue that the technology differs with risk, thereby meaning that the parameters of the parametric cost function changes with risk in a fully flexible manner.
This is accomplished by viewing the parameters of the cost function as nonparametric functions of risk. We also control for country-specific effects in a fully flexible manner by using them as arguments of the nonparametric functions along with the risk variable. The resulting cost function then becomes semiparametric. The standard parametric model becomes a special case of our semiparametric model.
We use the above modeling approach for banks in the EU countries. Actually, European financial integration is seen as a stepping stone for the development of a competitive single EU market that promotes efficiency and increases consumer welfare, changing the risk profile of the European banks. Particularly, financial integration allows more risk diversification and permits banks to use more advanced risk management instruments and systems, however it has at the same time increased the probability of systematic risks. Financial integration has increased the risk of contagion and changed its nature and scope. Consequently the bank’s risk seems to be an important issue to be investigated.
Original languageEnglish
Place of PublicationBirmingham
PublisherAston University
Number of pages9
Publication statusUnpublished - 2013

Keywords

  • cost function
  • semiparametric model
  • parameter heterogeneity
  • nonparametric model

Fingerprint Dive into the research topics of 'A flexible cost function model with risk'. Together they form a unique fingerprint.

Cite this