111 510 510 libonline@riphah.edu.pk Contact

Measuring banking sector competition – I

The following are excerpts from a Working Paper “Measuring Competition in the Banking Sector of Pakistan: An Application of Boone Indicator” authored by State Bank of Pakistan’s Mahmood ul Hasan Khan and Muhammad Nadim Hanif

The banking sector of Pakistan has witnessed a notable transformation in its structure and business activities following the implementation of financial sector reforms since the early 1990s. Specifically, the reforms helped transform a repressed financial sector into a market oriented and sound financial sector, predominantly owned and managed by the private sector. How these developments have impacted competition among the banks is still an open question. This study attempts to answer this question with the application of a new approach to measure competition: Boone Indicator of competitiveness. This measure postulates that inefficient firms (banks) in a competitive environment are punished harshly, and there is an output reallocation from inefficient to efficient firms/banks. We have estimated elasticity of market share to marginal costs for 24 banks in Pakistan, using a balanced panel of bank level (annual) data for the year 1996 to 2015. Marginal costs are obtained indirectly by first estimating a translog cost function using earning assets as an output, and cost of financial capital, physical capital and labor as inputs. The estimated Boone Indicator value of negative 0.31 is significant and suggests that inefficient banks have been losing their market share to efficient banks over the estimation period: a reflection of underlying competitive environment. Increasing value of Boone Indicator (in absolute terms) over the period of study suggests that competition among the banks in Pakistan has increased over time.

The authors would like to thank two anonymous referees of this paper and the participants of the Seminar at SZABIST for their useful comments.

From the early 1970s to 1990, the banking business in Pakistan was subject to credit ceilings, directed and subsidized credit, control on deposit and lending rates, etc. These policies together with weak control on the management rendered the banking services inefficient, and resulted in lack of healthy competition among the banks. Financial sector reforms initiated in early 1990s aimed at developing a market oriented, efficient and sound banking system under the private sector management. Has the policy measures like privatization of banks, entry of new banks/services, restructuring of pricing mechanisms, and the development of regulatory infrastructure, been successful in creating healthy competition among the banks in Pakistan is an open question. To answer this question, we have made temporal analysis of competition in the banking sector of Pakistan by using Boone Indicator approach based upon balanced panel data (from the year 1996 to 2015) of 24 banks in Pakistan. Boone Indicator is a measure of degree of competition amongst firms (banks in this case study). This indicator is based on the idea that competition reallocates output from less efficient to more efficient firms/banks. This measure of competition is thus essentially an elasticity of market shares to marginal costs. To calculate such elasticity, we estimate a (log linear) regression of market share of a bank upon its marginal cost over the period of study. Higher the (absolute) estimated elasticity of market share stronger is the competition as the inefficient banks are punished more heavily.

A key hurdle in the estimation of above elasticity is the non-availability of information on marginal cost directly from the financial statements of banks. In this study, we first estimated a (translog) cost function upon bank-by-bank data using ‘earning assets’ as an output and cost of financial capital, physical capital and labor as inputs. Marginal costs are obtained indirectly from (differentiated) translog cost function. The estimated Boone Indicator value is found to be-0.31 and is statistically significant. It implies that a 100 bps rise in marginal cost of a bank in Pakistan has been associated with 31 bps loss of market share over the period of this study. Lost market share will obviously be captured by relatively efficient (lower marginal cost) banks. Temporal values of the Boone Indictor reveal that the underlying competition in the banking sector of Pakistan has strengthened over the period of analysis in this study.

Operating environment for banks in Pakistan has witnessed substantial changes since the implementation of financial sector reforms in early 1990s. Specifically, privatization of state-owned commercial banks was initiated, and private sector was encouraged to open new banks; direct credit schemes were gradually discontinued; cap on lending rate was abolished; branch licensing policy was liberalized; and use of information technology for the provision of financial services was facilitated. Moreover, State Bank of Pakistan (SBP) has been actively facilitating Islamic banking, branchless operations, and micro financing, to promote access to financial services. All these changes were designed to instill a healthy competition, and create a sound and an efficient banking system capable of supporting the growing economic activity. Understating of degree and evolution of bank competition is also important as it has strong implications for the way changes in monetary policy stance impact the ultimate underlying objectives.1

Above developments have led to significant changes in the structure of the banking sector in Pakistan: the share of big 5 banks has declined from over 90 percent in early 1990s to 51.5 percent by end 2015; the ownership structure has changed from the public to the private sector as the later controls over 80 percent of banking assets; Islamic banks have emerged in banking arena, which control more than 10 percent of banking assets; and a number of foreign banks have switched their operations from a branch mode to full-fledged locally incorporated subsidiaries to expand their businesses.2 All this suggests that the level of competition must have been changing over time.

Literature on measuring competition has also been evolving along with developments in the banking sector around the globe as well as in Pakistan. It started exploring banking sector competition on the lines of industrial organization using ‘structure-conduct-performance’ and ‘efficient structure’ paradigms as we discussed in Khan and Hanif (2017a). The deficiencies in the structural measures of competition encouraged the use of formal tests, like Panzar-Rose (1987) H-statistic (as we used in Khan and Hanif, 2017b) to assess the underlying competitive environment of the banking sector. While PR-H statistic is useful to know the degree of competition among the banks in a country, it does not give any clue on how competition is evolving in the banking sector.

This issue was addressed by J. Boone in a series of studies conducted during 2000 and 2008 to develop a new measure of competition: ‘performance-conduct-structure’. This was later named as Boone Indicator by the researchers who used this approach in their empirical studies on competition. This measure is based upon a simple intuition that an increase in competition reallocates output from less efficient to more efficient firms (banks). Griffith et al. (2005) validated the idea of Boone by using simulated dataset. It was found that Boone Indicator outperforms traditional measures of competition including concentration ratio.

Given the substantial changes as a result of financial sector reforms in Pakistan, there is need to study how competition has evolved over time in the country during the last two decades. In Khan and Hanif (2017a and 2017b), we have reviewed studies on competition in the banking sector of Pakistan including Arby (2003), Khan (2009), and Bhatti and Hussain (2010). None of these studies had used this new indicator of measuring competition in the banking sector of Pakistan. Mirza et al. (2016), however, assessed the competition in the banking industry of Pakistan over the period of 2004-2012 by using the Boone Indictor (along with other indicators), but the authors did not analyze the evolution of competitive environment in banking industry over the period of analysis. Objective of our study is to fill this research gap and measure (estimate) the level of competition in the banking sector of Pakistan, with a view to explore how competition has evolved over time in the country during the last two decades.

In these settings, this paper presents temporal analysis of competition in the banking sector of Pakistan by using a new approach developed by Boone (2000, and 2008). This is the first ever research study applying Boone Indicator approach to measure competition in the banking industry of Pakistan, based on a large panel data of 24 commercial banks from the year 1996 to 2015, which uses estimated marginal cost (rather than proxy it with average cost).

This paper is organized into six sections. Introduction in this section is followed by a brief review of literature in section 2. Section 3 presents theoretical underpinnings of the new approach to measuring competition used in this study. Data description and selection of variables is the subject of Section 4, which is followed by the estimation and interpretation of empirical results in section 5. The final section concludes the paper.

Lack of any direct measure of competition rendered the policy makers and researchers around the globe to explore different indicators which can exhibit the underlying level of competition. Traditional theory of industrial organization focused on the market structure to analyze competition,3 which led to the development of various indicators related to the market structure including Concentration Ratio, Herfindahl-Hirschman Index (HHI), Hall-Tideman Index (HTI) etc. All these measures assign weights according to the size of firms (banks) in the sample to understand the market structure.4

A credible challenge to structure-conduct-performance (SCP) paradigm came from an efficiency-based competing explanation for higher concentration. The basic intuition is that the efficient banks/firms grow at a higher pace as compared to their peers. As a result, market share of the efficient banks/firms increases over time, which ultimately contributes towards market concentration. This is popularly known as an efficient structure (ES) paradigm.5

Given the strong theoretical underpinnings of both SCP and ES paradigms, the subsequent theoretical research on competition primarily focused on the conduct or behavior of firms/banks to understand competition in the market. The most notable development was the introduction of H-statistic to formally test the underlying market structure (Rosse and Panzar (1977), Panzar and Rosse (1987)). This test essentially measures the pass through of changes in input prices to the revenue under the standard assumption of market equilibrium.

While the PR-H model is a useful measure to classify the underlying market structure into three broad categories (perfect, monopoly and monopolistic competition), it is difficult to analyze the impact of policy interventions on the level of competition. Moreover, this model can only be applied to “investigate the competitive nature of the all banking activities” (Leuvensteijn et al. 2011). These issues are largely addressed in a new measure of competition, popularly known as Boone Indicator or relative profit differences (RPD) or performance conduct structure (PCS).6 In a series of papers, Boone (2000, 2004 and 2008) and Griffith et al. (2005), a new measure of competition was developed based on a widely held intuition that an increase in “competition reallocates output from less efficient to more efficient firms”.

Empirical evidence in favor of this new measure is provided in Griffith et al. (2005), which used simulated data to show that the new measure based on relative profits outperforms the traditional measures of competition (the HHI, concentration ratios, market shares, and price-cost-margin), especially in markets where the marginal costs and products are symmetrically differentiable.

Schaeck and Cihak (2010) explained the philosophy behind Boone Indicator as a measure of the degree of competition and estimated the Boone Indicator for US and a large number of European countries in order to assess ‘how competition enhances banking soundness’. Similarly, Mirzaei and Moore (2014) studied the determinants of banking competition across different income groups by using Boone Indicator (along with Lerner Index) as measure of competition for 146 countries over the period of 1999-2011. More recently, Clerides et al. (2015) measured the degree of banking competition in terms of Boone Indicator (along with Lerner Index and Adjusted Lerner Index) for 148 countries using bank-level data for 1997 to 2010. Studies like Schaeck and Cihak (2010), however, used average cost to proxy marginal cost in estimation process. World Bank (2017) attempted to overcome this shortcoming and used marginal cost to estimate the Boone Indicator pertaining to banking market for large number of countries using Bankscope dataset, including for Pakistan, over 1999-2014 period only.

The estimates of Boone Indicator for individual countries in cross countries studies like Mirzaei and Moore (2014) are less likely to be subject to rigorous analysis because the same specification is utilized for all the countries and any single country is not focused. The country-specific detailed information, especially for developing countries like Pakistan, is rarely used in cross-country studies due to lower weight and broad focus of the studies. Even in case of World Bank (2017), the Bankscope dataset is largely confined to big banks.

As far as country study on Pakistan is concerned, Mirza et al. (2016) is the first one which attempted to assess competition among banks in Pakistan (for the period of 2004-2012) using Boone Indicator (along with other indicators of measuring competition). In this study again, the authors used average cost to proxy the marginal cost. This approximation implicitly put a strong assumption on the analysis as the marginal cost would be equal to average cost only at the equilibrium point. A precise measure of marginal cost in line with economic theory can be derived from the cost function (as we discuss in the methodology section). We rely on a well-known translog cost function (TCF) and use a convenient approximation to estimate the marginal cost.

In the backdrop of above discussions, we study competition in banking sector of Pakistan by applying this new approach (Boone Indicator) while using (log of) market share as dependent variable in estimation of Boone Indicator. We use a measure of marginal cost instead of average cost (which has been used in Boone (2000, 2004 and 2008), Griffith et al. (2005), Schaeck and Cihak (2010) and Mirza et al. (2016)). We also improved upon the way variables used in this study are measured as discussed in following section in detail. – Courtesy State Bank of Pakistan

1. During the last couple of decades SBP has itself come a long way to become an independent monetary policy decision maker.

2. These issues are discussed at length in Khan and Hanif (2017a).

3. The SCP paradigm assumes one-way causation from the market structure to performance.

4. For details, see Hall-Tideman (1967); Adelman (1969); Davies (1979); Kwoka (1985); and Rhoades (1995).

5. For details, see Demsetz (1973) and Peltzman (1977)

6. Boone himself termed this new measure as performance-conduct-structure while other researchers named is as a Boone Indicator.

Recorder Report, "Measuring banking sector competition – I," Business Recorder. 2017-09-29.
Keywords: Economics , Commercial banks , Financial sector , Monetary policy , Management , Pakistan , SBP , PCS

Leave a Reply

Your email address will not be published. Required fields are marked *