The objective of this study is to measure and explain the measured variation in the performance and productive efficiency of Mauritian commercial banks in the post-financial liberalisation period. A wide range of financial reforms have been instituted since the 1980s which included measures such as liberalisation of interest rates, removal of quantitative controls on credit, lifting of barriers on competition, privatisation of public financial institutions and introduction of market-based securities. The major aims of the reforms have generally been to raise both the level of investment and the efficiency of its allocation and in addition, to enhance the provision of financial services to all sectors of the economy. These reforms were geared towards the liberalisation of the financial system in order to enhance efficiency in the mobilisation and allocation of financial resources (Jankee 2001, World Bank 2003).
The main impetus for this study lies on the objectives of policymakers to increase efficiency in the financial system and develop Mauritius into a regional financial center. Given the importance of the banking sector in the Mauritian economy, an examination of efficiency in the banking sector will provide useful guidance to policy makers towards understanding and assessing the process of the banking sector reform. Such a study has been motivated by a number of factors. First, in contrast to developed countries, very limited evidence has been obtained on these issues particularly in the case of developing countries (see Humphrey, 1993, 1997 for an extensive survey). Moreover, a wide range of methods has been developed, broadly classified under parametric and non-parametric, to study bank efficiency and productivity (Bauer et al., 1998; Avikaran, 1999).
Given that no such study has been undertaken for Mauritius, such a work attempts to extend the literature in various ways. First, an examination of the impact of financial deregulation on the efficiency (allocative and technical) and productivity (TFP) of banks would contribute to the literature on a small and deregulated open economy in the African region. Moreover, we also incorporate different objectives of the banking industry in our analysis of efficiency and productivity. Following Leightner and Lovell (1998), two aspects are examined. Firstly, we look at banks as profit maximisers and secondly, as banks pursuing the objectives of the Central Bank in terms of financial stability and economic performance. We apply the non-parametric approach Data Envelopment Analysis (DEA) (see Coelli, 1996), a widely applied technique to estimate efficiency scores and productivity of banks for the period 1994-2004. These estimates can be used to compare Mauritius with other countries in terms of efficiency of the banking system.
The outline of this paper is as follows: in section 2, we give an overview of the Mauritian banking sector. Section 3 briefly reviews the literature on issues relating to bank efficiency, productivity and financial deregulation. Section 4 discusses the Data Envelopment Analysis (DEA) and the methodology that is used to compute efficiency and productivity of banks for the period 1994-2004. The estimated efficiency scores are discussed in section 5. Section 6 concludes the paper and highlights the policy implications.
II.An overview of the Mauritian
Before discussing the efficiency issues, in this section I will discuss the main features of the Mauritian banking sector (see Bundoo and Dabee, 1998; Junglee 2001; Jankee 1999; Worldbank 2003). Mauritius is a small island economy in the Indian Ocean and inherited a bank-dominated financial system at the time of independence in 1968. As compared to many developing countries especially in the African region, the Mauritian financial system was quite developed with 11 banks. The first set of control has been the regulation of banks’ interest rates by monetary authorities throughout the 1970s until late 1980s. Other repression policies consisted of the imposition of cash ratio and liquid asset ratio that were gradually tightened over the years as well as the exchange control on current and capital transactions. In the mid 1970s, the monetary authorities tightened their control over the financial system in an attempt to regulate credit expansion and allocate it to productive sectors.
In the early 1980s, the control over the overall credit was modified whereby sectors were categorised into priority and non-priority and ceilings were imposed respectively on both types of sectors. Furthermore, banks were individually subject to a certain quantum of credit depending upon their extent of deposit mobilisation and credit creation. The early 1980s were marked by the beginning of the process of gradual liberalisation of the financial system. Controls over interest rates were gradually lifted. Exchange control on current transactions was no longer imposed as from mid 1980s. By late 1980s, interest rates were fully liberalised. However, quantitative controls in the form of reserve requirements and credit ceilings continued to be imposed. The 1990s were marked by the relaxation of most of the remaining banking sector controls. Credit ceilings were gradually abolished and the exchange control act was suspended by mid 1990s. The cash ratio and liquid asset ratio were gradually lowered and the liquid asset ratio was brought down to zero in 1997.
The financial liberalisation programme was also accompanied by other market-oriented reforms such as a free float exchange rate, the auctioning of Treasury bills and the setting up of a secondary market for government securities amongst others. Most recently, transactions involving the repurchase of bank reserves and foreign currency swaps have increased enormously. In addition, bank branches expansion has also contributed largely to the institutional development of the banking sector. From 32 in the 1970, the number of bank branches expanded significantly to reach 117 in 1990 and 163 in 2003. Some other specific details on banking sector developments are given in the appendix.
III. Overview of Literature
The banking sector has attracted considerable theoretical and empirical research during the preceding decades. Studies have involved a number of issues including the role of banks in financial development, bank efficiency, pricing behaviour of banks and banking regulation. Prior studies on bank efficiencies concentrated on estimating cost functions and measuring economies of scale and scope with the implicit assumption that banks being studied operate efficiently (Gilbert, 1984). Many researchers who have claimed the importance of investigating inefficiencies in the banking units have questioned this assumption. Since then, this issue has led to considerable research. However, one issue of recent interest has been the effects of deregulation on the performance of banks (see Berger, 1993, 1997 for a review). The literature distinguishes two main types of bank efficiency. The first is operational efficiency as introduced by Farell (1957) to measure efficiency and the second is X-efficiency as introduced by Leibenstein (1966) to explain differences in efficiency between banks. The concept of operational efficiency is purely technical and can be defined as the product of technical efficiency and allocative efficiency (see Coelli, 1996). While technical efficiency tells us how far the bank output is from the bank’s isoquant, allocative efficiency captures inefficiencies due to the fact that the bank has picked up a suboptimal input combination given input prices.
A number of factors have motivated research on bank efficiency (Berger et al., 1993, 1997; Hardy et al., 2001). First, there is the mainstream economic thinking that improving the efficiency of financial systems is better implemented through deregulatory measures aiming at increasing bank competition on price, product, services and territorial rivalry (Smith, 1997; Fry, 1995). However, empirical evidence on the impact of financial deregulation on bank efficiency has been mixed. Berger and Humphrey (1997) stated that the consequences of deregulation might essentially depend on industry conditions prior to the deregulation process as well as on the type of deregulation measures implemented. The deregulation on the asset side of the balance sheets that focused on the liberalisation of the volume and the interest rates of bank lending resulted in the improvement of both efficiency and productivity of Norwegian banks (Berg et al., 1992). Turkish banks had a similarly experience (Zaim, 1995). But the impact of liberalisation on Indian banks resulted in varied productivity efficiency depending on the type of ownership (Battarcharyay et al., 1997).
Berger and Humphrey (1997) undertook a comprehensive survey of 130 studies that apply the parametric and non-parametric frontier efficiency analysis to financial institutions in 21 countries.
A number of issues had been raised and tested relating to bank efficiency and financial deregulation. These issues mainly included the alternative methodologies used to estimate different types of efficiencies, namely technical efficiency, allocative efficiency, scale efficiency, pure technical efficiency, cost efficiency and change in factor productivity (see Coelli, 1996). Moreover, researchers have also tested empirically the extent to which factors such as market share, total assets, credit risk, technology and scale of production, bank branches, ownership and location, quality of bank services and diversity of banking products, financial deregulation and managerial objectives determine bank efficiencies.
Das (2002) examined the effects of financial deregulation on risk and productivity change of public sector banks in India for the period 1995-2001. They found evidence that capital; non-performing loans and productivity were entwined, with each reinforcing and to a certain degree complementing the other. They also found that higher capital led to a rise in productivity whilst higher loan growth reduced productivity. Moreover, increased government ownership tended to increase productivity.
Leightner and Lovell (1998) using the best practice production frontiers, constructed the Malmquist growth indexes and productivity indexes for each Thai bank, for 1989-1994, incorporating two different specifications of the services that the bank provides, one derived from the objectives of the bank itself and the other derived from the objectives of the Bank of Thailand. They found higher productivity growth of banks when the bank objective of profit-maximisation was pursued as compared with the model when the Bank of Thailand's objective was achieved.
Laeven (1999) used DEA to estimate the efficiencies of the commercial banks in Indonesia, Korea, Malaysia, Philippines and Thailand for the years 1992-1996. They also included risk when analysing the performance of banks and found that foreign banks took lower risk as compared with family-owned banks.
Battacharyay et al. (1997) examined productive efficiency of 70 Indian commercial banks during the early stages of the on-going liberalisation process. They estimated the technical efficiency scores using DEA and then used stochastic frontier analysis to attribute variation in the calculated efficiency scores to three sources, temporal, component, ownership component and random noise component. They found public owned banks to be the most efficient followed by foreign banks and privately owned banks. Hardy et al. (2001) estimated the effects of banking reform on the profitability and efficiency of the Pakistani banking system. They estimated the profitability, cost and revenue frontiers to derive measures of efficiency of the banking system relative to the best available practice. They found that financial market reform has increased both revenues and costs but did not increase overall profitability and led to convergence in efficiency.
Jagtiani and Khantavit (1996) examined the impact of risk-based capital requirements on bank cost efficiencies in the US banking industry. They found that the introduction of risk-based capital requirements led to a significant structural change in the banking industry both in terms of efficient size and optimal product mixes. Their results implied that regulations encouraging large banks to expand production and product mixes resulted in a less efficient banking industry.
Sathye (2001) empirically investigated the X-efficiency, both technical and allocative, in Australia. He used the non-parametric method of DEA to estimate the efficiency scores. He found that banks in the sample had low levels of efficiency as compared with the banks in the European countries and in US. Efficiency in Australia came mainly from the waste of inputs (technical efficiency) rather than choosing the incorrect input combinations (allocative efficiency). Moreover, domestic banks were found to be more efficient than foreign owned banks.
Overtime, a number of methods have been used to measure the performance of banks. The use of financial ratios has been criticised because of its reliance on benchmark ratios (see Yeh 1996). These benchmarks could be arbitrary and misleading. Further, Sherman and Gold (1985) noted that financial ratios do not capture the long term performance and aggregate many aspects of performance such as operations, marketing and financing. In recent years, there has been an increasing use of frontier analysis methods to measure bank performance. In the frontier analysis methods, the institutions that perform better relative to a particular standard are separated from those that perform poorly. Applying a non-parametric or parametric frontier analysis does such separation to firms within the financial services industry.
The parametric approach includes stochastic frontier analysis, the free disposal hull, thick frontier and the distribution free approaches, while the non-parametric approach is the data envelopment analysis (DEA) (see Molyneux et al., 1996). Furthermore after Charnes et al. (1978) who coined the term DEA, “a large number of papers have extended and applied the DEA methodology” ( Coelli, 1996). In the present study, we employ the non-parametric method (DEA). This approach has been used since a lot of “recent research has suggested that the kind of mathematical programming procedure used by DEA for efficient frontier estimation is comparatively robust (Seiford and Thrall, 1990.)
IV.1 Data Envelopment Analysis (DEA)
The Data Envelopment Analysis is a linear programming technique initially developed by Charnes et al. (1978) to evaluate the efficiency of public sector non-profit organisations (see Molyneux et al., 1996). DEA involves the use of linear programming methods to construct a non-parametric piecewise frontier over the data so as to calculate efficiency relative to this frontier. Thus, DEA calculates the relative efficiency scores of various decision making units (DMU) in a particular sample. The DMUs can be banks or branches of banks. The DEA measure compares each of the banks/branches in that sample with the best practice in the sample. It tells the user which of the DMUs in the sample are efficient and which are not.
The ability of the DEA to identify possible peers or role models as well as simple efficiency scores gives it an edge over other methods. Fried and lovell (1994) have given a list of questions that DEA can help to answer. Details about the various frontier measurement techniques are found in the works of Bauer et al. (1989), Bauer (1990), and Leightner and Lovell (1998) etc. There are a number of software options for running DEA. This study uses the Software (DEAP) developed by Coelli (1996) to calculate the technical, allocative and cost efficiency scores of banks. Methodologically, the characteristics of DEA can be described through the original model developed by Charnes, Cooper and Rhodes. Consider N units (each is called a DMU) that convert I inputs into J outputs, where I can be larger, equal or smaller than J. To measure efficiency of this converting process for a DMU, Charles et al. propose the use of the maximum of a ratio of weighted outputs to weighted inputs for that unit, subject to the condition that the similar ratios for all other DMUs be less than or equal to one. That is,
where are positive known outputs and inputs of the nth DMU and are the variable weights to be determined by solving problem (1). The DMU being measured is indicated by the index 0, which is referred to as the base DMU. The maximum of the objective function given by the problem (1) is the DEA efficiency score assigned to . Since every DMU can be, this optimisation problem is well defined for every DMU. If the efficiency score, satisfies the necessary condition to be DEA efficient; otherwise it is DEA inefficient.
It is difficult to solve problem (1) as stated, because the objective function is non-linear and fractional. Charnes et al., however, transformed the above non-linear programming problem into a linear one as follows:
The variables defined in problem (2) are the same as those defined in problem (1). An arbitrarily small positive number, is introduced in problem (2) to ensure that all of the known inputs and outputs have positive weight values and that the optimal objective function of the dual problem to problem (2) is not affected by the values assigned to the dual slack variables in computing the DEA efficiency score for each DMU. The condition ensures that the base is DEA efficient; otherwise it is DEA inefficient, with respect to all other DMUs in the test. A complete DEA model involves the solution of N such problems, each for a base DMU, yielding N different weight sets. In each program, the constraints are held constant while the ratio to be maximised is changed. Finally, these DEA problems are solved in the paper using the computer software developed by Coelli (1996).