Competition Concentration and Market Structure of Consumer Banks in Pakistan

ABSTRACT

The paper focuses and explores the current market structure of banking industry of Pakistan. The structural approach of Panzar and Rosse has been used. The P-R H statistic which is the sum of elasticities of reduced form revenue equation is calculated and tested with the help of wald co-efficient test. The data was taken from 26 banks for the year 2001-2009. The paper concludes that the banks in the industry are earning their revenues in monopolistic kind of a structure. Moreover another important point is that the industry is heading towards perfect competition.

ACKNOWLEDGEMENTS

First of all I would like to thank Mr. Abdul Rehman Tahir, without his Continuous efforts I would never have been able to complete my work. I would also like to thank Mr. Zahoor Ahmad (Ex-GM, SNGPL) who also played a significant role in completion of my work. In the end I would like to thank Mr. Ahmed Junaid Raza (Chief Economist, SNGPL) and all those who gave me immense support throughout the course of my research, whether directly or indirectly and congratulate them on the completion of this research paper.

Table of Contents

DECLARATION II

CERTIFICATE OF APPROVAL IV

© [2011] [Zohaib Ahmad] V

ABSTRACT VII

ACKNOWLEDGEMENTS VIII

Table of Contents IX

Introduction 1

An overview of Topic 1

Research Objective 3

Related Definitions 3

Market Structure 3

Return on Assets 4

Cost of Labor 4

Cost of Funding 4

Literature Review 6

Herfindahl-Hirchman Index 18

Methodology 21

Research Type and Data 21

Panzar and Rosse Model 22

Market Structure and Characteristics 25

Theoretical Framework 26

Reference List of Variables under consideration 27

Theoretical Justification 27

Hypothesis 28

Estimation, Analysis & Results 30

Results of Parameter Tests of Significance 32

Limitations of the Study 33

Conclusion 33

References 34

Appendix-A i

Appendix-A ii

Appendix-A iii

Appendix-B iv

Appendix-C v

Appendix-D vi

Appendix-D vii

Introduction

An overview of Topic

The Banking sector of Pakistan is seen to have a potential for growth. During the past Decade it has experienced a very substantial growth which has attracted many investors from abroad. In fact it will be very true if we say that in Pakistan, the Banking industry is the only industry which is contributing very positively in the GDP of the country. Due to its profit potential many foreign investors have also been attracted and this results in increase in the total number of consumer banks. This increase eventually resulted in more innovative products at a cheaper price thereby facilitating the consumers. This increment in the consumer bank has also changed the overall structure of the industry. The industry that started off with Monopoly kind of structure is now have a very different picture altogether.

Market Structure refers to the number and characteristics of sellers and buyers in a market. Here we are concerned with the sellers only. By structure we mean Monopoly (when there is only one seller and who is the price setter), Perfect competition (when there are many sellers and all of them are price takers rather than price setters), Oligopoly (when there are small number of sellers each of whom has a very small degree of influence on price), and etc. Market Structure is very important factor/determinant in analyzing the competitiveness of the industry.

Banking institutions are one of the most important institutions in the progress of a country. During the seventies all the banks were nationalized. Normally it is hard to accept any real competition in such circumstances. However later in nineties, the banks were de-nationalized. Now most of the banks are working in private sector and as such there seems a lot of competition between all the banks. The banking sector in Pakistan is so lucrative that many of the foreign banks have also opened their branches in Pakistan. In recent past, lots of new services have been offered to general public and this retail/consumer banking has become very important. In such circumstances, competition is dependent on market structure.

Research Objective

The banking sector plays a very substantial role in the economy of any country. Especially in Pakistan, its role is even more important because the banking industry of Pakistan is amongst those few industries which are contributing significantly in the Country’s GDP. Therefore due to this very reason it is important to know that under what circumstances the banks are earning their revenues. The sole purpose of this study is to know the structure of banking industry of Pakistan. This structure may be Monopoly, a Perfectly Competitive or may be somewhere in between these two extremes. Thus, in short the main aim of this paper is to find out the market structure of consumer Banking industry of Pakistan. This will eventually help these banks to make their marketing strategies and offer new products accordingly.

Related Definitions

Market Structure

Market Structure refers to the number and characteristics of sellers and buyers in a market. In our particular case we are concerned with sellers only.

Return on Assets

The dependent variable in the study is Return on Assets (ROA) which is defined as the ratio of total net income after taxation to the total assets in the balance sheet. This Return on assets is influenced by various other factors (independent variables). These include three input variables and three bank specific control variables. The input variables include the three costs i.e. the labor cost, the cost of capital and the funding cost. These are defined as follows:

Cost of Labor

This refers to all the cost that is incurred directly or indirectly to pay the wages of workers, salaries of employees and other such items.

Cost of Capital: this refers to the cost of financing. It includes both the debt and equity.

Cost of Funding

These are the costs that are incurred in payments of interest on outstanding loans etc.

In addition to these there are three control variables which also have their impact on the dependent variable. These include ratio of total loans of the bank to total assets of bank, ratio of income from other sources to income earned from interest, and finally the ratio of non-interest earning assets to total assets in the balance sheet.

Literature Review

Hasan (2007) assess the market structure of consumer banks of Pakistan. The writer’s main aim is to know whether the banks of Pakistan are competing in a perfectly competitive situation, Monopoly or Monopolistic competition etc. for analysis twenty six banks is being taken. Further the article follows structural approach of Panzar and Rosse. A step by step process is being followed. Data is taken from 1997 till 2007. First of all M-Concentration ratio is calculated by the researcher. This ratio tells the market share of m big banks or in other words concentration of wealth in top m banks. In second step co-efficient of variation is calculated. This is calculated to overcome the shortcomings of M-Concentration ratio. After calculating Co-efficient of variation, Herfindahl index is calculated to know the degree of competition. After all these calculations H-statistic is calculated. Major focus of the paper is on this h-statistic. This H-statistic studies the impact of changes in cost on the total revenue that bank earns. The Paper basically studies the data from twenty six banks. Furthermore Panel data is being used in the research that means both the cross-sectional as well as the longitudinal data has been used. The dependent variables in the paper are “Interest income and total income. Independent variables include L-cost, F-cost and K-cost. L-cost is basically the cost incurred in paying salaries and wages and other fringe benefits to the employees. K-cost is the cost of deposits and other borrowings whereas F-cost is cost of capital” (Hasan, 2009). The paper concludes that the banking sector of Pakistan experiencing monopolistic competition and is moving towards perfectly competitive environment.

Rozas (2007) assess the market structure of Spanish Bankibg industry i.e to find out whether the Spanish banking industry is operating in perfect Competition, Monopoly or between these two extremes. Researcher has used P-R approach to arrive at the conclusion. The researcher has used panel data for the past twenty years till 2005. Only the commercial and saving Banks are taken for this purpose. The dependent variable includes “NITA which is ratio of net income to total assets” (Gutiérrez de Rozas, 2007). On the other hand there are some other independent variables which the researcher has taken. These independent variables are divided into two categories, input variables and Bank specific variables. Input variables comprises of three variables namely labor price(PL), price of loanable funds(PLF) and price of Capital expenditures(PCE). Further Bank specific variables include “EQTA which is defined as Equity to total assets, LOATA, defined as Loans to total assets, LFTA, defined as loanable funds to total assets and LDTLD which is proxied by individual market share to loans and deposits” (Gutiérrez de Rozas, 2007). The panzar and Ross model uses H-statistic which in this paper is defined as sum of intercepts of input variables. This is also called elasticities of function. In conclusion the researcher find that the banking system in spain is experiencing a competition which is result of concentration.

Bikker and Haaf (2000) and aims to find out the market structure of Banks, their concentration and the relationship between competition and concentration. For this purpose two approaches has been used i.e the structural approach and non-structural approach. The Panzar and Rosse approach has been used to assess the structure. The panel data from 23 European and non-European countries have been taken. The data belongs to year 1988 till 1998. The dependent variable taken is “INTR, which is the ratio of total interest revenue to total balance sheet. The independent variables include AFR- ratio of annual interest expense to total funds, PPE-ratio of personnel expense to total balance sheet, PCE-ratio of physical capital to fixed assets” (Bikker & Haaf, Competition, Concentration and their relationship, 2000).

Claessens and Laeven (2003 basically focuses on measuring the competitiveness among large group of countries. Initially the data from many countries was collected which was then narrowed down to just fifty countries. The researchers were mainly concerned to find out the factors that makes the differences in banks positions. For this purpose the role of different banks from foreign countries was analysed. The data used in the research belongs to 1994 till 2001. It is important to note that in this paper all the banks whether that b commercial, Saving, cooperative or any other type of banks were included. The P-R which stands for Panzar and Rosse approach was followed. The variables used in the article are categorised in twe groups namely input variables and bank specific control variables. The input variables include price of labor which is proxied by “ratio of personal expense to total assets” (Claessens & Laeven, 2003), Price of loans which is proxied by “Gross Interest revenue to total assets” (Claessens & Laeven, 2003) and price of Labor which is proxied by “operating expense to total assets” (Claessens & Laeven, 2003). On the other hand, as far as bank control variables are concerned they include “Y1 which is ratio of equity to total assets” (Claessens & Laeven, 2003), “Y2 which is ratio of net loans to total assets” (Claessens & Laeven, 2003) and “Y3 which is log of total assets” (Claessens & Laeven, 2003). H-Statistic which is the total sum of elasticities of Bank’s Revenue with respect to the bank’s input prices, ranges between 0 and 1 for almost all the banks which shows that banks are operating in monopolistic competition. The paper concludes that banks competition is strongly related with concentration and contestibility determines the competition for the banks.

Chun (2004) assess the market structure of banking industry of Korea. For this purpose two approaches was used. First Herfindahl index was calculated. The data set consists of data from year 1994 to 2001. The same was done using Pansar and Rosse approach. Basically the main reason behind using this approach was that the researcher wishes to know under what structure the banks are earning their revenues. The dependent variables which were chosen by the researcher includes “TREV, TREVIN and ROA” (Chun, 2004). These variables are proxied by ” total revenue to total assets, total interest revenue to total assets and net profits to total assets respectively” (Chun, 2004). Further independent variables include three input variables and two bank specific variables. The input variables include “price of Labor, Price of capital and Price of funds” (Chun, 2004) which are proxied by “personal expenses to employees, Capital expenses to fixed assets and capital expenses to fixed assets respectively” (Chun, 2004). Other bank specific variables include “RISKASS which is Provisions to total assets and BR which is no. of branches of each bank to total number of branches for all the banks” (Chun, 2004). The researcher has used two approaches in this paper. First the H-statistic is calculated by using year wise data of all the banks i.e the H-statistic is calculated by running the regression of all the banks for one year and so on. The second approach which is used is by taking data of three years at the same time. Both the results were then analyzed individually. The H-statistics ranges between 0 and 1 on the basis of which the paper concludes that the banks in Korean banking industry earn their revenues in a monopolistic competition.

Duncan and Langrin (2004) focuses on the market structure of banking sector of Jamaica. The main aim is to find out under what conditions the banks are earning their revenues. For this purpose two methodologies have been adopted namely Herfindahl Index which helps in analyzing the concentration and Panzar and Rosse model which gives the empirical analysis to arrive at the conclusion. This approach basically uses the H-Statistic which is the sum of elasticities of revenue function, to give an estimated value. This estimated value is then used to reach the actual value by using F test. The data used for the research is quarterly panel from 1989 to 2002. Moreover thirteen banks were taken “The PR model was basically used to test the market structure of industry for competitiveness whereas Herfindahl index tells the sum of all the Banks’ squared market shares where market share may be based on either deposits or assets. A quarterly HHI was calculated for both deposits and assets over the period of 1989 to 2002” (Duncan & Langrin, 2004). The dependent variable taken in the income earned from interest which is dependent on many factors. These include the price of capital, price of Labor, price of deposits, and some other control variables which include “LK (loan to capital), LA(loans to assets), DL (deposits to loans), KA (Capital to assets), LNA(non performing loans to assets) and LQA (liquid funds to assets) ratios” (Duncan & Langrin, 2004). The three variables namely Price of Capital, Labor and deposit are called input variables, sum of whose intercepts gives the estimated value of H-Statistic. The rest of independent variables are called the bank specific control variables. The results from both the methodologies were different. HHI index concludes that the banking industry has decline in competition due to the mergers of four banks in late 90s. on the other hand Panzar and Rosse approach concludes that the banking sector of Jamaica is earning their revenues in a monopolistic competition environment.

Bikker and Groeneveld (1998) focus on concentration as well as on the structure of banking industry in which the banks are operating. The paper basically analyzes the structure of all the banks of EU as a whole and at the same time all the steps are repeated for individual banks separately for all the countries. In this way the total of 892 banks were taken from 1989 till 1996 i.e. five years data has been taken. The Panzar and Rosse model is followed in order to arrive at the conclusion. The dependent variable taken is “INTR which is total interest revenue to total balance sheet” (Bikker & Groeneveld, Competition and Concentration in EU banking Industry, 1998). The other factors or variables that influence the dependent variables include “INTE which is proxied by interest expense to total funds, PE, proxied by ratio of personal expenses to total assets and CE which is proxied by Ratio of capital and other expense to total fixed assets” (Bikker & Groeneveld, Competition and Concentration in EU banking Industry, 1998). H-statistic which is the sum of all the intercepts of input variables comes out to be ranging between 0 and 1. Thus according to the definition of hypothesis it is a monopolistic competition. But this is the estimated value which is yet to be confirmed via the F-test. After doing all the required tests the paper concludes that the banking industry of EU countries has a monopolistic competition.

Draicos and Konstantinou (2005) to find out the market structure of Transition countries because they need special attention towards structure. The paper uses data of different economies/countries from year 1992 to 2002. There are total of 218 banks that has been taken for analysis. These banks belong to different countries whose economies have been through transition stage. These mainly include Poland, Hungary, Czech Republic, Latvia and many more. The methodology that has been used is very similar to those used by others i.e. Panzar and Rosse model is used for the analysis. The dependent variable used in the study includes Gross revenue and Return on assets which are dependent on certain independent variables which includes Leverage, net interest margin, total assets and operating expenses. H-Statistic is first calculated from reduced form equation. This gives the estimated value which then was further calculated using F-test. The results shows that the value of H lies within the range of zero and one which means that the structure is monopolistic. Thus the researcher concluded that the market structure of banking industries of those economies which have been through the transition phase earn their revenue in a monopolistic environment.

Hempell (2002) aims to find out the structure of banks in Germany. The researcher basically has used two different methodologies. The banks that were taken for analysis include Saving Banks, Co-operative Banks and Credit Banks. The data that has been used belong to year 1993 till 1998 i.e. five year data. The two methodologies that have been used in the paper are Herfindahl index and Panzar and Rosse approach. Herfandahl index is calculated using deposits, Loans and Securitized loans separately. Further the panzer and rosse approach is used to assess the market structure of these banks. The important point in the paper is that the researcher has separately calculated the H-statistic for all the three categories of Banks discussed earlier. The dependent variable that were used for the analysis include “IINC which is interest income from loans and money market transactions and from fixed interest, securities debentures, TINC1Dinterest income, current income from shares and other variable yield securities commission income” (Hempell, 2002). The independent variables include some factor prices for example “labor cost, funding cost, capital cost and other cost. These are proxied by ratio of annual personnel expenses to total assets, ratio of interest expenses to sum of deposits and other liabilities, ratio of capital expenses (depreciation, write-down on intangible and tangible assets) and fixed assets and ratio of other non-interest expenses to total assets respectively” (Hempell, 2002). In addition to these there are some other variables called bank specific variables. These include “OItTA- ratio of other income to the total balance sheet, EtTA- ratio of equity to total assets, TLtTA- ratio of total loans to total assets, LtTL- ratio of customer loans to total loans, LtTA- ratio of customer loans to total assets, CLtTL- ratio of commercial loans to total loans, IDtTD- ratio of interbank deposits to total deposits and TDtTDMM- ratio of deposits to deposits plus money market liabilities” (Hempell, 2002). The H-statistic was calculated separately for all the banks in one group and then after doing the F-test the paper concluded that credit banks are more competitive than the other two banks while if comparison is done for saving banks and cooperative banks, the saving banks are more competitive than cooperative banks.

Spierdijk et.al (2009) emphasizes that in panzer and rosse model it is not necessary to take a scaled revenue or price equation. The un-scaled equation gives the best result. To prove this point a panel data from 1986 to 2004 for 18000 different banks have been taken. The dependent variable taken is total revenue where as some other input variable which include Price of Labor, price of capital and price of funding. “The paper concludes that un-scaled PR model is one tailed and is inconsistent with any form of perfect competition” (Bikker, Shaffer, & Spierdijk, 2009).

Herfindahl-Hirchman Index

Herfindahl-Hirchman Index is a widely accepted measure of market concentration. Generally the HHI is calculated for Total Assets, Total Deposits and Total Advances (Loans) etc. of a Banks. In this Paper we have calculated it for all the above mentioned items. To Calculate the HHI, first calculate the market share of the individual bank. Then take square of this market share and sum it up. Finally multiply the sum by 10,000. Mathematically

HHI = *10,000

The value of HHI varies from 0 to 10,000. If it is 10,000 it means that there is only one bank in the market that has Monopoly power and it will charge whatever price it want. On the other extreme if the value is very close to zero, it means that the market is very concentrated with large number of banks and all the banks will be price takers in this situation. In between these two extremes there exists an oligopoly, monopolistic competition and etc. generally the banking industry all over the world is considered concentrated with monopolistic kind of a structure.

For our calculations of HHI we have taken Total Assets, Total Advances and Total Deposits for the calculation of HHI. The different values of HHI for each year using all the three above mentioned items have been calculated.

Herfindahl hirchman Index (HHI) for all the three indicators is showing a declining trend (See Appendix-A). On the basis of total Assets, the HHI for the year 2001 was 1572. As the time passes by its value falls continuously showing that the concentration among the industry is increasing. In the year 2009 this value falls to 927. This clearly shows that the competition among the banks is increasing and thus the industry is heading towards perfect competition.

Similarly, the HHI for advances and deposits is also showing the same results as that of total assets. The values for both of these items starts from 1547 & 1635 respectively. As the time passes by both these items show the declining trend and in the period of eight years falls to 999 & 978 respectively. This also shows that the market structure is heading toward the perfect competition. If we talk about the current values and make our decision on these values, we may conclude that presently the banks are earning their revenues under a monopolistic competition.

Methodology

Research Type and Data

As discussed earlier that knowledge about the market structure is very important for the banks that are operating in the industry. For this purpose we have used two approaches in order to know what is the current market structure of banking industry of Pakistan. The first approach which is being used is Herfindahl-Hirchman Index (HHI) which is a good measure of market concentration. The Second approach is also a widely used and accepted technique to measure the market structure of industry. This is known as Panzar & Rosse framework. This is known by P-R H Statistic. The main reason being using two different approaches is the nature and sensitivity of the topic. The results from both the approaches should be same and they should never contradict each other.

The research conducted is of Secondary nature. The data taken for both of these approaches is of 26 banks. For HHI, a panel data from year 2001to 2009 has been used for all the twenty six banks. As far as P-R approach is concerned, a panel data for the years 2001-09 for 26 banks has been used. All the data is collected from SBP publications. Some of the banks for which data was not available for the whole period or which have posted a very large sum of loss were not included with the view that these will be the outliers which will distort the analysis.

Panzar and Rosse Model

A Panzar & Rosse technique has been used around the world for analyzing the banking industry. Economists of the World Bank, Asian development Bank, European Banks and many other research industries has used this method for the in-depth analysis of banking industry. They have analyzed the data from many countries collected over a large period of time.

In the Past, there had been a lot of efforts to analyzed and estimate the structure of the market by analytical methods. However none of the techniques was found to provide an answer for the problem.

In 1986-87, J.N. Rosse and J.C. Panzar developed a estimation technique to analyze the degree of monopoly power being exercised in a particular market. They used firm’s profit maximizing first order condition to establish that the sum of elasticities of gross revenues with respect to each input price will be negative in the case of collusive equilibrium. This sum of elasticities will be less than one for the monopolistic competition. A detailed review is provided in a paper published by Rosse and Panzar in 1987 in the journal of industrial economics, vol. 35. For this analysis there are some key assumptions. For the positive value of P-R statistic to be reliable and capable of being interpreted as in the model, the model depends on the firm’s in question of being in long run equilibrium. The firms have stable costs and revenue functions. There are also some limitations in the model of Panzar and Rosse which affects the overall results. For example, if the bank in the sample have not completely adjusted to the market condition and they are in transition stage in this case the banks will be towards the spurious appearance of the market power. Another limitation of the test is that it cannot differentiate between competitive pricing and cost-plus pricing because the latter is not specifically associated with the particular level of market power, the implication for interpretation P-R statistic are not clear.

The P-R model is basically based on reduced form revenue equation as under:

Ln(Revenue)=αo+ α1(LnAPF)+ α2(LnAPK)+ α3(LnAPL)+ α4(LnLTAR)+ α5(LnOITII)+ α6(LnNEAS)…………………………………….(1)

Where:

Ln is the natural logarithm

APF is the average price of funds

APK is the average price of capital

APL is the average price of labor

LTAR is the ratio of Loans to assets

OITII is the ratio of other income to interest income

NEAS is the ratio of ratio of non earning assets

α is the coefficients

In the equation 1 the first three variables which include APK,APL and APF are the input variables where as the rest are bank specific variables. Further from the above equation 1, P-R H Statistic, which is the sum of elasticities of firm’s factor cost, is calculated by adding the coefficients of Input variables as under:

H-Statistic = α1+ α2+ α3

The interpretations of the values obtained are analyzed according to the principle below:

H≤0 Monopoly

H<0<1 Monopolistic competition

H=1 Perfect Competition

Below are some of the characteristics of different market structures which a particular banks can have:

Market Structure and Characteristics

Market structure

Concentration

Entry barrier

Product differentiation

No. of entities

Pricing control

Perfect competition

Low

Low

No

Massive

No

Monopolistic competition

Relatively low

Relatively low

Yes

Many

Moderate

Oligopolistic competition

Relatively high

Relatively high

Yes or no

Several

Certain extent

Monopoly

High

High

One

One

Large extent

Source: Wong Wing Chit (January, 2004).

Independent Variables

L-cost

F-cost

K-cost

Theoretical Framework

LTAR

OITII

NEASDependent Variable

Revenue

L-cost

OITII

NEAS

Revenue

K-cost

F-cost

LTAR

Reference List of Variables under consideration

“L-cost (Cost of Labor): Salaries, allowances and other benefits to employees” (Hasan, 2009)

“F-cost (Cost of Funding): Interest expense to average interest bearing liabilities” (Hasan, 2009)

“K-cost (Cost of Capital): Other operating cost to fixed asset ratio” (Hasan, 2009)

“LTAR: Loans to assets ratio” (Hasan, 2009)

“OITII: Other income to interest income ratio” (Hasan, 2009)

“NEAS: Non-interest earning assets to total asset ratio” (Hasan, 2009)

Theoretical Justification

The dependent variable in our model is the Revenue which is affected by various factors which includes three input variables and three bank specific variables. The three input variables include the average price of labor, average price of capital and average price of funds. The revenue is affected by average price of labor in a sense that if the price is increased it will reduce the total revenue. Thus we can say that both have inverse relation. Secondly the other two variables which include average price of fund and capital also have inverse relation with revenue. More over if we talk about the bank specific variables these include loan to asset ratio, other income to interest income ratio and non-earning assets to total assets ratio. Loan to asset ratio has a direct relation with the revenues because as the total loans increases the revenue is also increased. Moreover if we take other income to interest income ratio, it is also directly related to the total revenues. If the income raises it will eventually increase the revenues. Thus we can say that both are directly related with each other. In the end the relation between noninterest earning assets to total assets. This is inversely related with the revenues. If this is increased it will reduce the revenues.

Hypothesis

Perfect Competition Test: Generally, if the banks are working under perfect competition it is most likely to happen that if we change cost by just a certa