The effects of financial liberalization and new bank entry on market structure and competition in Turkey

____ __ __ _____ kNrf5 )~331 POLICY RESEARCHtf WORKING PAPER 1839 The Effects of Financial Promotingcompetition ln Turkey's newty liberaliea Liberalization and New financial market requires Bank Entry on Market striking a balancebetween potentiaiy conflicting Structure and Competition objectves and the risk entailed in financial in Turkey deregulation. Banking n,t be regulated and wel!- supervised, but 7 trKevy Cevdet Denizer bankina system also ne.c . become far more curnr)'m The World Bank Development Research Group S November 1997  PoL1cY RESEARCH WORKING PAPER 1839 Summary findings Until 1980, Turkey's financial system was shaped to rather than their efficiency. Deregulation and support state-oriented development. After the 1960s, the liberalization should be continued and strengthened. financial system, dominated by commercial banks, - The entry of small-scale firms alone is not enough to became an instrument of planned industrialization. increase competition, so new banks should probably not Turkey had an uncompetitive financial market and an be expected to alter the market structure. inefficient banking system. Controlled interest rates, * To promote competition will require addressing directed credit, high reserve requirements and other barriers to both entry and mobility. The main barrier to restrictions on financial intermediation, and restricted mobility seems to be the size of the large banks, which entry of new banks - plus the exit of many banks exerts a significant negative effect on competition. betwee 1960 and 1980 - created a concentrated market * Interbank rivalry among the leading banks can't be dominated by banks owned by industrial groups with facilitated without creating new banks of a certain size oversized branch networks and high overhead costs. with a reasonable number of branches. Breaking up Turkey since 1980 has seen a trend toward public banks (which hold 30 percent of sectoral assets, liberalization of its financial market. Reforms eliminated excluding the Agricultural Bank and three development interest rate controls, eased the entry of new financial banks) could help create 15 to 20 new banks with 40 to institutions, and allowed new types of instruments. 50 branches. This would reduce concentration and Regulatory barriers were relaxed, attracting many banks improve mobility in retail banking. (both Turkish and foreign) into the system, and Turkey's * Breaking up the public banks before privatization banking system became integrated with world markets. would probably also improve their governance structures Denizer examines how reform has changed the and efficiency. system, focusing on Turkey's commercial retail banking * Promoting the entry of nonbanks and local banks market. He finds that: would also increase the number of institutions competing - Although reform reduced concentration in the for deposits. Turkey lacks a healthy variety of credit industry, leading banks are still able to coordinate their institutions and should consider developing a mortgage pricing decisions overtly. High profitability appears to market and creating institutions for housing finance. have resulted from the banks' uncompetitive pricing This paper - a product of the Development Research Group - is part of a larger effort in the group to study financial reforms and financial markets. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Emily Khine, room N11-061, telephone 202-473-7471, fax 202-522-3518, Internet address kkhine@worldbank.org. November 1997. (51 pages) The Policy Researcb Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by the Policy Research Dissemination Center  The Effects of Financial Liberalization and New Bank Entry on Market Structure and Competition in Turkey Cevdet Denizer Macroeconomics and Growth Development Research Group Economist at the World Bank, Washington DC. The views expressed in this paper are mine and they do not reflect the views of the World Bank and its member countries. I am grateful to the Central Bank of Turkey, particularly to Hasan Ersel, and Alnila Tutuncuoglu for assisting with the data. I extend thanks to Pat Fishe, Tarhan Feyzioglu and Mark Wohar for helpful discussions on an earlier version of the paper.   Contents I. INTRODUCTION H. DEVELOPMENT POLICIES AND THE FINANCIAL SYSTEM IN RETROSPECT 11.1. The 1923-50 Period 11.2 The 1950-80 Period m. FINANCIAL REFORMS OF 1980 m. I Financial Liberalization 111.2 Changes in Market Structure and Performance of Commercial Banks IV. METHODOLOGICAL APPROACH IV. 1. Market Structure IV.2. Entry and Competition V. MODEL SPECIFICATIONS AND DATA V.1. Market Structure Model V.2. Competition Model V.3. Data VI. EMPIRICAL RESULTS VI. 1. Market Structure Equations VI.2. Competition Model Equations VII. POLICY IMPLICATIONS AND CONCLUSIONS Tables and References   THE EFFECTS OF FINANCIAL LIBERALIZATION AND NEW BANK ENTRY ON MARKET STRUCTURE AND COMPETITION IN TURKEY I. INTRODUCTION Until 1980, the Turkish financial system developed under an umbrella of monetary and regulatory policies aimed at supporting the state orchestrated development strategy. Particularly after the early 1960s, the commercial bank dominated financial system became an instrument of planned industrialization policies and operated under a framework characterized by controlled interest rates, directed credit programs, high reserve requirements, and other restrictions on financial intermediation, as well as restncted entry. While these financial and regulatory policies were not exclusive to Turkey and contributed to its industrialization, they had their costs on the banking system's competitiveness and efficiency.' Interest rate controls led to non-price competition in the form of branch network building by banks already in the system. This situation and restrictive entry policies, coupled with the exit of significant number of banks between 1960-80, gave rise to concentrated market dominated by public and private banks owned by industrial groups with excessively large branch networks and high overhead costs. In retrospect, it is generally thought that the combination of these factors created an uncompetitive market structure and an inefficient banking system. In contrast, the years since 1980 have seen a major trend toward the liberalization of financial markets in Turkey. Starting in June 1980, as part of a far reaching stabilization and structural adjustment program, the Government implemented financial liberalization and deregulation measures aimed at 1 As pointed out by Vives (1991), and Mayer and Vives (1993) until the advent of global financial deregulation in the 1970s most countries, both developed and developing, followed restrictive financial and regulatory policies. See also OECD (1992). However, it is worthwhile noting that financial restriction did not turn into financial repression in industrialized countries while it did in industrializing ones.  2 developing an efficient and competitive financial system that would support and facilitate the functioning of a liberal economy. To that end, reforms eliminated interest rate controls, eased the entry of new financial institutions, both bank and non-bank, and allowed new types of instruments. There were also policy measures to develop equity and bond markets. Although there were occasional setbacks and policy reversals in terms of interest rate controls, and a banking crisis in 1982, reforms have led to major changes in the sector. Relaxation of regulatory barriers has attracted a significant number of banks into the system, both Turkish and foreign. Reforms were also successful in halting the decline in financial intermediation observed prior to 1980 and contributed to financial deepening and a revitalization of the stock market. At the same time product variety increased and quality of financial services improved. Moreover, the Turkish banking system became more integrated with the external financial world and improved its financial technology and human capital. The objective of this paper is to examine several aspects of the banking market in Turkey to assess the nature of its structure and the state of competition. What kind of market structure exists in banking after the reforms? Was the entry of new banks sufficient to transform the market structure into a competitive one or did the distortions resulting from earlier financial and regulatory policies built endogenous constraints into the system thwarting competition regardless of new entry? Now that regulatory entry barriers are gone, are there mobility barriers in the system? These are some of the questions this study attempts to provide answers by drawing on market structure studies found in industrial organization literature and applying them to the Turkish banking market. The focus is on the commercial retail banking market since it is primarily through this channel that resources are mobilized and allocated. However, it must be noted at the outset that although recent developments in theory have improved our understanding of financial intermediation, there is still no  3 fully developed theoretical model of banking competition and that the quantitative results of this study must be interpreted with caution.2 The paper is organized as follows. Section II briefly examines the evolution of the banking market in Turkey and identifies the set of factors which shaped its structure since the establishment of the Republic in 1923 until the late 1970s. These issues are notjust of historical interest. They are reviewed to understand the relationship between overall development and financial policies, and more precisely, the cumulative impact of these policies on bank market structure and competition. Section III reviews the 1980 financial sector reforms and analyzes the developments in banking market structure, including a review of data on exit, entry, various measures of market concentration, and bank profitability. Section IV presents the methods used and hypotheses tested in this study as well as their underlying rationale(s) as well as the data. Results are discussed in section V. Finally section VI assesses the impact of reforms in light of results obtained, raises some fundamental issues and problems and discusses policy options to facilitate competition in the system. In the process, some international comparisons are also made. 'For a recent discussion of the theory of the banking firm and competition as well as the issues involved in assessing the benefits of increased competition in banking from a theoretical point of view, see Mayer and Vives (1993) and the articles therein.  4 II. DEVELOPMENT POLICIES AND THE FINANCIAL SYSTEM IN RETROSPECT H. 1. The 1923-50 Period At the time modern Turkey was established in 1923, the formal financial system comprised of 35 banks, of which 22 were Turkish owned and 13 were foreign with a total of 439 branches.3 Most of the foreign banks dealt with foreign trade and foreign companies operating in Turkey and their involvement with Turkish firms was limited. On the other hand, Turkish-owned banks were mostly small local banks and were too weak to support the newly emerging industry and commerce. During the first Economics Congress held in Izmir in 1923, it was emphasized that the country suffered from scarcity of capital, and that without establishing a national banking system the country would not industrialize (Akguc 1987). It was also argued that the banks should take the initiative in financing large industrial enterprises and the State should provide capital for new banks since the private sector was too weak to provide it or simply did not have capital. The conference had a significant impact on economic issues in general, and on banking and credit in particular, and influenced government policies in the following years. During 1923-32, in parallel to its broad strategy of industrialization via private sector encouragement, the Government's regulatory approach to banking and finance was quite liberal and aimed at developing a national banking system. While the Government provided the initial capital for 4 public banks which still exist today and lead development efforts, it allowed and actively encouraged the formation of private banks. As a result, about 29 new private banks, mostly single branch and local were established. There were practically no restrictions for entry. By 3For a more detailed account of the evolution of banking in Turkey which this section draws upon heavily, see Akguc (1987).  5 1932, the number of banks reached to 60, of which 45 were national, up from 13 in 1923, and 15 were foreign. However, the most important event of this period was the foundation of the Central Bank in 1930. These liberal economic policies did not last long however. In the early 1930s, partly due to worldwide depression and partly due to the realization that the private sector was too weak to be the engine of growth, the government adopted a new strategy. This new strategy, generally labelled as "etatist", emphasized state led development and assigned a secondary role for the pnvate sector. In order to accelerate industrialization, the government established state enterprises in key industries during the 1932-45 period which are still in operation. The important aspect of this period for banking was the creation of new public banks to provide support for the new state enterprises. Although there was no significant change in the Government's regulatory policy in the aforementioned period, there was no entry into the system. In fact, this period was characterized by the exit of most small private local banks due to the economic slowdown in Turkey resulting from the global economic crisis and the advent of the Second World War. In turn, these developments reduced the number of banks in the system and increased the dominance of public banks in the sector. The number of banks fell to 40 in 1945. During the same period the number of branches also fell to 411 from 483. II. 2. The 1950-80 Period The years following World War II to the 1950s on the other hand reflect attempts to reduce the role of the state in the economy and the expansion of the private sector. On the banking side, the period between 1944-1960 was characterized by the entry of 27 private banks and 3 public banks, including Akbank, Yapi ve Kredi Bankasi, Garanti Bankasi, and T. Sinai Kalkinma Bankasi. By 1958 there were 62 banks in the  6 system, a number which was not surpassed until 1989. The number of bank branches increased by about fourfold and reached to 1,759 by 1959 and the process of nationwide branching was well underway. However, most of the newly established banks did not stay in the system long and 10 small Turkish-owned banks, and 4 foreign banks were liquidated between 1945-59. With a net entry of 16 banks between 1944-60 and a small number of mergers among the existing banks, the total number of banks in the system at the end of 1960 was 59. The slowdown in economic activity towards the end of the 1950s, the 1958 recession, and the Government's stabilization program led to further failures. Between 1960-64, 15 more small banks ended their operations, some were liquidated and some were merged with others which brought the number of banks to 49 and the number of branches to 1909 by the end of 1964. The start of a planned development strategy in 1963, and to a lesser degree concerns over the failure of a large number of banks during the early 1960s, brought significant changes to banking and finance policy. In order to attain plan targets, the public sector increasingly assumed a larger role in the allocation and mobilization of resources through directed credit programs, subsidized lending to priority sectors and other constraints on financial intermediation. Cumulatively, these measures turned the financial system into an instrument of industrialization policy. While the efficiency of this arrangement in terms of directing credit according to plan targets was questioned by Akyuz (1984) it remained in effect until late 1980. There were no changes in interest rate policy however. Like before, interest rates were administratively set by the government and this policy was not specific to the planned era. Since 1940s deposit interest were controlled by the govermment and they were changed only 5 or 6 times between 1940 and 1978.5 In general, however, the impact of these policies was to increase the role of the State in financial markets. According to Hanson and Neal (1986), around only a quarter of total credit was free from government control as late as 1983. 'See the tables presented at the end of Fry's 1979 book entitled Money and Banling in Turkey.  7 Unsurprisingly, the adoption of planned development strategy in 1963 also brought significant changes to the regulatory policies which has been a major determinant of market structure prior to the 1980 reforms. Development plans in effect shaped regulatory policies of successive governments' and they became more conservative over time - reinforcing the larger role assigned to the plans by restricting entry which in turn made State control of financial resources easier. A common theme running in all the three plans Turkey implemented during the aforementioned period was that the country had enough commercial banks and that the smaller banks in the sector should be merged to reduce overheads so that stronger institutions could be created (Akguc 1987). Given this line of reasoning, the plans argued that the need and benefits for new commercial banks should be clearly demonstrated if they were to be established at all. Instead, the plans argued for the establishment of development and specialized service banks, mostly to support industry.6 In line with the recommendations of the plan the government pennitted the establishment of 4 new development and specialized banks between 1962-75 period which were not authorized to collect deposits. During the 1962-80 period only 3 new commercial banks were established which demonstrate the existence of strong regulatory entry bariers. On the other hand, as noted before 23 banks were either liquidated or merged during the 1960-80 period which reduced the number of banks to 43 by 1980 from 59 at the end of 1959. Another characteristic of the 1963-80 period, particularly after early 1970s, has been the emergence of private banks owned by industrial groups which Akguc (1987) refers as the beginning of holding banking. The reasons for this are straightforward. During the 1963 -1974 period Turkey followed a strongly growth oriented strategy led by both public and private sector investments, mostly in import competing sectors, infrastructure, and heavy industries. The public sector investments were financed by monetizing budget deficits, issuing low yield bonds mostly purchased by public pension funds and bank deposits. At the same time, after the mid 1960s and during the 1970s, the private sector, encouraged and supported by the 6See the discussion of banking and finance policy in 5 yearly plans in Akguc (1987) pp. 48-58.  8 government through high protection rates and a complicated incentive scheme for investments, was also expanding through a holding company structure and was in need of financing. While the government had access to capital for its large investments, the private sector did not. In the absence of capital markets they had to rely on bank loans to finance their investments (Fry 1988). Since the public banks were primarily financing public investments, the private sector had all the incentives to establish or acquire banks to finance their investments. Consequently, with restricted entry, major groups began to acquire banks established earlier and by the early 1970s alnost all major private banks belonged to holding groups (Akguc 1987). The period between 1963-80 also saw a rapid expansion of branches of banks already in the system. Under interest rate controls, the only mode of competition to collect deposits was non-price competition in the form of establishing branch network throughout the country. Rising inflation during the late 1960s and throughout the 1970s also provided another strong incentive banks to expand their branch networks. With interest rates becoming increasingly negative in real terms, opening new branches to collect deposits and investing them into real assets was highly profitable. In fact, as deposit rates became increasingly negative in real terms, the number of branches of both public and private banks increased. However, it must be noted that as long as deposit rates were controlled by the Government and inflation was rising, this made sense and was consistent with profit maximization.' Due to these factors the number of branches jumped to 5769 in 1980 from 1720 in 1960 despite the fact that there was a significant reduction in the number of banks. The important thing to note about this process is that it resulted in excessive investment in bank branches and contributed to bank sizes that are larger than they would be if the price of capital was not distorted. At the same it significantly contributed to concentration in the sector since there was very little entry which meant that the expanding banks were the same old ones. As long as the marginal cost of deposits, equal to the interest rate on deposits plus the cost of buildings and equipment was less than the inflation rate banks would expand their network to collect deposits. Hence, as the spread between the deposits rates and the inflation rate widened profit maximization would require more investment into bank branches, which is actually what happened in Turkey.  9 Cumulatively, the combination of mutually reinforcing financially and regulatory restrictive policies, coupled with the exit of 23 banks over the 1960-1980 have led to highly concentrated market structure, and an overbranched, inefficient banking system. By 1980, the top 5 banks controlled about 70 percent of deposits, 64 percent of assets and owned 60 percent of all branches, as well as controlled more than 10 percent of the number of deposit accounts (see table 1). Overhead costs in the sector reached to atound 7 percent of total assets, almost triple the OECD average by 1980. Hence, although the developnit strategy and its related financially and regulatory restrictive policies contributed to the industrialization of Turkey, they may have introduced distortions that are difficult to eliminate with respect to the systems' efficiency and competitiveness. Fry (1979) for example noted that even if all interest rate restrictions were abolished, " a minimum deposit rate might be needed to force Turkey's cartelized and oligopolistic banking system to achieve the competitive ideal solution".  10 11. FINANCIAL REFORMS Im. 1. Financial Liberalization In June 1980, simultaneously with the structural adjustment and broad liberalization policies that put an end to the import substitution era, the Government launched financial reforms.8 The goal was to develop a competitive and efficient financial system that would support a more liberal economy. This was to be achieved through deregulation and promoting entry into the system. Reforms eliminated interest rate restrictions on deposits and loans, and eased entry into the market and permitted new types of financial instruments and institutions. The initial phase of deregulation saw sharp increases in interest rates and attempts by the larger banks to hold them low through the so-called "gentlemen's agreement" which in essence was open collusion. However, this proved unsustainable. Faced with higher rates offered by the unregulated brokerage houses, larger banks increased their rates which resulted in fierce competition and extremely high real interest rates. This situation, combined with financial distress in real sectors led to the collapse of six banks during 1983 and 1984. These developments in turn has led to partial reversal of reforms and the Central Bank began to reregulate deposit interest rates, though at much higher levels relative to pre-1980 situation. However, as much as this was to restore financial stability it was also a measure to deal with collusive practices of banks. The Central Bank continued with the regulation of deposit rates until 1988 occasionally adjusting them to maintain positive real rates of return. In late 1988, deposit rates were again liberalized and this policy was maintained since then although there were a number of temporary interventions. Therefore, the 8For a more detailed review of financial liberalization experience of Turkey, see OECD (1988), Onis and Ozmucur (1988), Akyuz (1990), Atiyas (1990), Akkurt et.al (1992), and Atiyas and Ersel (1994).  11 switch to price competition was not complete until late 1988 although the reform process started in 1980. Nevertheless, despite occasional setbacks, higher levels of interest rates resulted in substantial growth of the financial system and contibuted to financial deepening. By the end of 1990, the stock of financial assets reached 47.7 percent of GDP from around 28 percent in 1980 while the M2/GDP ratio rose to 25.6 percent from about 21 percent in 1980 (table 2). In line with financial liberalization policies, most directed credit programs and preferential rates were eliminated contributing to more efficient allocation of resources during the past decade. Although reserve requirements were lowered, liquidity ratios were increased which in turn put a wedge between deposit and loan rates. 1.2. Changes in Market Structure and Performance of Commercial Banks Reforms were successful in attracting entry into the banking system, one of the key objectives. As a result of easing of entry restrictions, the number of banks increased from 43 to 66 between 1980-90. Out of the 43 banks in 1980, 8 banks were either liquidated or merged with other institutions Hence, there were 31 de novo entries into the system, of which 19 were foreign and 11 national during the 1980-90 period. However, almost all of the new entrants specialized in trade finance and wholesale corporate banking. None of the new banks, foreign and Turkish, established offices beyond the three largest cities, and by and large they eschewed the retail banking market despite the fact that there are no restrictions on the scope of their operations. At the end of 1990, they accounted for less than half of 1 percent of savings and commercial deposits. Hence, the new financial institutions filled certain profitable niches which in itself is a positive development. Their impact on the retail banking market level, however, have been quite limited. Nevertheless, as pointed out by Akkurt et al. (1992), the entry of new banks, particularly the foreign ones, has been instumental in improving the quality of human capital and financial technology of the sector.  12 As expected, reforms reduced concentration in the sector. Table 1 presents 3, 5, 8, and 10 firm concentration ratios in terms of deposits, savings deposits, loans, assets, and number of savings accounts. The declines were most pronounced in the 3 and 5 firm concentration ratios. This result has been mainly due to the top 5 banks, except one, losing their market shares, especially in total deposits. In fact, with the exception of the largest bank (Ziraat Bank), banks who ranked among the top 3 and 5 in deposits in 1980 all saw their market shares decline in varying magnitudes. The decline in share of second largest bank in total deposits has been particularly significant as its share fell from 20 percent in 1980 to around 12 percent in at the end of 1991. While the top 3 and 5 banks have lost market shares the second tier banks that existed before the 1980 reforms have increased theirs. It appears that they were the ones who benefited from dorequlation of interest rates to increase their market shares and probably came closer to their optimal scale in tern's of their operations. While the quantitative declines in some measures of concentration have been large it has been small in pome others considering the number of entries. For example, the 3 firm concentration ratio in terms of t9tat deposits declined from 53 percent in 1980 to 40 percent in 1990 while the 5 firm concentration ratio fell from 09 percent to 55 percent, also in the same period. However, when 8 and 10 firm concentration ratios are analyzed the declines are much less pronounced. The top 10 banks accounted for 88 percent of total deposits in 19$Q and 82 percent in 1990, a decline of 6 percentage points compared to a decline of 13 percentage poits in the 3 firm concentration ratio. Likewise, 10 firm asset and loan concentration ratios registered sma,ler declines as shown in table 1. This indicates that while there were changes in the market shares of banks following the reforms reflecting some interbank rivalry, these have been mostly among the top 10 or top 15 banks which were in the system before the 1980 refonns. This may suggest that a critical number of brapches is needed to be an effective competitor in the retail banking market.  13 Savings deposits are particularly important for the analysis of competition in retail banking since it is one of the main outputs of retail banks and it is the most basic financial asset people hold.9 At the end of 1991, of the 45.6 million bank accounts in Turkey, 36.7 million were savings accounts representing for more than half of the volume of total deposits in the system. During the first half of 1980s, there was a marked increase in concentration ratios for savings deposits as shown in table I and by 1986, the 3 firm concentration ratio reached to 63 percent and 10 firm concentration ratio reached to 92 percent. The process was reversed in 1987 and at the end of 1991, 3 and 10 firm concentration ratios stood at 42 percent and 83 percent respectively. As before, the decline in the magnitude of the 10 firm concentration ratio for savings deposits was less than the decline in the 3 or 5 firm concentration ratio. Another interesting statistic to evaluate is the number of savings accounts. During the 1980-91 period, the number of savings accounts increased from 26 million to 36.7 million. At the same time there was a marked increase in the number of accounts opened with the large banks. As shown in table 1, the 3 firm concentration ratio of number of accounts increased from 55 percent in 1980 to 62 percent in 1991 while the 10 firm concentration ratio increased from 89 percent to 94 percent, in the same period. While in volume terms the percentage of savings deposits placed with the leading banks declined as explained above, the increase in the number of deposit accounts by the large banks probably implies that large banks attracted mostly small depositors while the sophisticated depositors moved their funds to other banks or to exploit other profitable investments. This would however, suggest that the power of top 10 banks did not decrease with respect to the most basic item of retail banking, small savings accounts. 9Although there is disagreement over what banks produce, it seems reasonable and technically acceptable to viewmajor deposit and loan categories as bank outputs. See Berger and Humprey (1992b).  14 Performance of the banking sector in terms of profitability following the reforms have improved despite declining concentration ratios and new entry. As shown in table 3, and pointed out by Atiyas and Ersel (1994) profits in the banking sector increased substantially, particularly after the mid 1980s, and reached to levels about 5 times the OECD average by the end of 1990. At the same time, the declining trend in operating costs following the initial reform years were reversed in 1988 and since then these rose sharply and reached to more than double the OECD average (see table 3). This would imply that deregulation has not yet led to rationalization of the use of capital and labor and hence to improvements in productive efficiency as expected. The ability of banks to increase and maintain high profit rates under these circumstances would seem to suggest that the source of profits was market power or some other market imperfection rather than productive efficiency. Furthermore, as pointed out by Rhoades (1993) it would also mean that additional resources are not entering the market, implying the existence of non-regulatory entry barriers. IV. METHODOLOGICAL APPROACH The review of developments in the banking system suggests that market structure continues to have a significant impact on the conduct and performance of banks, and implicitly on competition. However, observations by themselves are not sufficient to establish a causal link between these parameters, which requires an empirical investigation. In order to do this, this study presents and tests a number of hypotheses by drawing upon the methods of industrial organization. The analysis is in two parts. The first part attempts to determine if there is a relationship between market structure and performance of banks using the structure- conduct-performance (SCP) paradigm as a framework of analysis. The study examines the two main hypotheses, namely the "traditional" and "efficient structure" hypotheses found in the general literature for the explanation of market structure-performance relationship. The second part focuses more directly on  15 competition in the retail banking market. In particular, it analyzes the impact of new bank entry and sunk investments in the system that resulted from pre-1980 interest rate and regulatory policies on competition. IV. 1. Market Structure The first hypothesis tested in this paper, the traditional SCP idea, emphasizes the market or industry structure when analyzing the pricing and output decisions of market participants. In this context, market structure refers to the number and size distribution of firms. The market is treated as the unit of analysis. If it is concentrated, say in assets, sales, deposits or some other measure of economic activity, then we are likely to observe non-competitive, collusive behavior and, in that case, equilibrium industry profitability will depend upon the degree and stability of collusion among firms. Therefore, the higher the share of the market controlled by a small number of large firms, or the higher the market concentration, the greater the possibility that market participants will agree to collude, either tacitly or overtly, and raise prices above costs, therefore earning supranormal profits. Hence, the existence of a positive relationship between some measure of concentration, proxying market structure, and profits, proxying performance, would imply that market structure is not competitive and market participants enjoy profits primarily because of their market power. The second and more recent hypothesis is known as the efficient structure hypothesis. It maintains that firm-specific efficiencies arising from superior management, use of new technology, etc., enable some firms to increase their market share at the expense of other relatively inefficient firms, leading to market concentration. The implicit assumption is that the differing efficiencies among firms lead to unequal market shares and high levels of concentration, and are causally due to factors that reduce costs. The leading firms will earn above-average profits even if they charge prices at the level of secondary firms. Therefore, we observe a positive relationship between market concentration and profits, but it is not due to collusion and  16 does not necessarily imply a causation between structure and performance. The efficient structure hypothesis implies that the causal link will be between market share, a measure of frm efficiency, and profits, but there will be no causal relationship between market concentration and profitability. Therefore, the positive relationship between concentration and profits found in some industry studies is spurious and simply reflects the correlation between market share and concentration. It is worthwhile noting that both hypotheses point to an observationally equivalent relationship between market structure and profits while differing as to the causal factors generating it. While it is possible that both hypotheses might be operative simultaneously it is nevertheless important to distinguish between the two as they have different public policy implications. If profitability is due to market structure, then a regulatory policy to reduce concentration and consolidation in the sector may be justified. On the other hand if performance is due to efficiency then such a regulatory policy may be welfare reducing. Weiss (1974) suggested that by estimating a profit function that takes both market share and concentration measure into account at the same time it may be possible to ascertain whether profitability is due to the efficiency of or to market structure, and hence the validity of the two hypotheses, in explaining the structure-performance relationship. Tests ofthis nature havebeenundertakenby Smirlock (1985) and Evanoffand Fortier (1988), Molyneaux (1992) and it is the approach adopted in this study.'