A Bayesian Efficiency Analysis of Angolan Banks

• DOI: http://doi.org/10.1111/saje.12124
• Author: Carlos Pestana Barros,Emanuel Reis Leão,Zorro Mendes,Nkanga Pedro João Macanda
• Date: 01 September 2016
• Published date: 01 September 2016

A BAYESIAN EFFICIENCY ANALYSIS OF ANGOLAN BANKS

CARLOS PESTANA BARROS

†

,EMANUEL REIS LEA

˜O*

‡

,NKANGA PEDRO JOA

˜O MACANDA

‡

AND

ZORRO MENDES

†

Abstract

This article studied the technical eff‌iciency of Angolan banks from 2005 to 2012 using a Bayes-

ian stochastic frontier model. The intermediation approach to banking was adopted. The results

revealed that Angolan banks were very eff‌icient and that eff‌iciency varies little among the banks

analysed. Furthermore, the differences in eff‌iciency between foreign banks, public banks, large-

sized banks and banks that belong to a local conglomerate were examined. It was concluded that

the greatest eff‌iciency was to be found in the case of foreign banks. Since size and conglomerate

membership do not seem to lead to greater bank eff‌iciency, it was proposed that Angolan policy-

makers should promote competition in the banking sector.

JEL Classif‌ication: C11, G21, D22

Keywords: Angola, banks, stochastic frontier

1. INTRODUCTION

This article studies the eff‌iciency of Angolan banks from 2005 to 2012 using a Bayesian

stochastic frontier model (Assaf et al., 2011, 2012). The issue of banking eff‌iciency is a

traditional theme in banking research, and successive innovations have occurred with the

use of frontier models (Berger and Humphrey, 1992). Research using stochastic frontier

models includes, among others, the studies made by Berger (1995), Berger and Hum-

phrey (1997), Berger and Mester (1997), Altunbas et al. (2001), Goddard et al. (2001),

Maudos et al. (2002), Yildirim (2002), Al Shamsi et al. (2009), Drake (2010), Chiu

et al. (2011), Hall and Simper (2013). The focus has, however, been mainly on the

United States and on European countries. Studies focusing on emerging developing

countries, such as Angola, are less common.

Eff‌iciency analysis in the African context has spanned many sectors, e.g. Pauw et al.

(2007), Christian and Crisp (2012), Halkos and Tzeremes (2012), Koch and Slabbert

(2012), and Lemba et al. (2012).

As far as the Angolan f‌inancial sector is concerned, Barros et al. (2014a) examined

insurance companies, while Barros et al. (2014b) studied eff‌iciency in the banking sector

using the B-convexity model. Furthermore, Barros et al. (forthcoming in 2016) analyse

eff‌iciency in the banking sector with a two-stage TOPSIS and neural networks approach.

The scope of these last two studies is similar to ours, but the methodology and the data

are different, and the results we obtained also show some disparities. Barros et al.

(2014b) use the non-parametric B-Convexity model – a model based on Data Envelop-

ment Analysis (DEA) – to compute the eff‌iciency scores for a number of Angolan banks.

* Corresponding author: Assistant Professor Instituto Universit~ario de Lisboa (ISCTE-IUL),

Political Economy, Lisbon, Portugal. E-mail: emanuel.leao@iscte.pt

†

Instituto Superior de Economia e Gest~ao, University of Lisbon, Lisboa, Portugal

‡

Faculdade de Economia da Universidade Agostinho Neto, Luanda, Angola

V

C2016 Economic Society of South Africa. doi: 10.1111/saje.12124

484

South African Journal of Economics Vol. 84:3 September 2016

South African Journal

of Economics