Banking industry liberalisation in Ghana

Transcription

1 Available online at Review of Development Finance ) Short communication Banking industry liberalisation in Ghana Elvis A. Adjei, Shanti P. Chakravarty Bangor Business School, The University, Bangor, Gwynedd, LL57 2DG, UK Received 27 February 2011; accepted 11 January 2012 Available online 8 March 2012 Abstract Economic liberalisation is characterised by the entry of foreign companies and the emergence of new domestic institutions to compete with the existing institutions of lending, and there is a view that the process leads to greater efficiency, especially because foreign banks bring new expertise. Extracting information from accounts lodged with Ghana Central Bank, this paper ranks the cost efficiency of banks, Theil decomposition of the cost efficiency scores allows for a comparison of performance between banks under different types of ownership. There is pronounced differences in efficiency scores within and between groups by type of ownership, but foreign ownership, per se, is not the determining factor Production and hosting by Elsevier B.V. on behalf of Africagrowth Institute. JEL classification: G2; L8; N2; O2 Keywords: Africa; Ghana; Financial liberalisation; Foreign banks; Domestic banks; Public sector 1. Introduction Financial liberalisation entails diversity in ownership and governance structures of banks, and we are concerned with comparing the efficiency of different ownership structures. However, liberalisation has taken different forms in different countries because the model of perfect competition underlying the idea of full liberalisation does not apply to banking. There are different types of liberalisation requiring different ways of measuring the success of reform carried out within the constraints of the particular type of liberalisation under consideration. 1 Comparison of efficiency between different types of banks becomes country Corresponding author. Tel.: addresses: elvis908@yahoo.com E.A. Adjei), s.p.chakravarty@bangor.ac.uk S.P. Chakravarty). 1 For example, liberalisation in the United States had meant the repeal of the Glass-Steagal Act restricting the range of activities a bank could carry out Ibanez et al., 2010). Peer review under responsibility of Africagrowth Institute Production and hosting by Elsevier B.V. on behalf of Africagrowth Institute. doi: /j.rdf specific. We are concerned with measures of bank efficiency in post-liberalisation Ghana. Efficiency and productivity are problematic concepts Saraydar, 1989; Uddin and Hopper, 2003) and they are doubly problematic in the service sector. For example, an industry might be characterised by firms that make best use of inputs, within the constraints of extant knowledge, in production without making any further advances on new methods of production or delivery. The problem is further complicated in the service sector where inputs and outputs cannot always be clearly delineated, and this further difficulty is highlighted in the literature on bank performance. Assets of banks are the deposits and loans, loans are re-deposited, and deposits are lent out again. Notwithstanding these difficulties, there is consensus in the literature about certain particular indicators of efficiency which might help in the conduct of informed policy debate about the intermediation role of banks. Being concerned with the debate about financial sector liberalisation, this paper focuses on the banking industry. The intermediation approach to modelling bank production posits that banks mediate between depositors and borrowers. From this perspective, a bank purchases inputs deposits) in order to generate earning assets loans) as output Sealey and Lindley, 1977), and adding value in the process. The attainment of full allocative efficiency in a free market entails that risk-adjusted interest rates equalise risk-adjusted returns across all borrowers and lenders. Market efficiency, as that concept is understood above, is inadequate to capture gains and losses to the economy from liberalisation if either there are

2 94 E.A. Adjei, S.P. Chakravarty / Review of Development Finance ) interactions between industries and second best problems arise or there is information deficiency in market contracts. 2 Competition can increase even when reform of the financial sector is carried out within the context of an industrial strategy for the allocation of funds, as was done in South Korea Amsden and Euh, 1993). 3 To study the impact of liberalisation on the efficiency of banks, it may be fruitful to examine functional efficiency, [measured as] the productivity of the delivery of financial services Amsden and Euh, 1993:387). Increase in competition is a likely factor in reducing cost of delivery Mahesh and Bhide, 2008), but there may be institution-specific constrains on the outcome reflecting differences in ownership structure. 4 The process by which reforms in the banking system may contribute to cost efficiency is outlined by Jae Yoon Park 1988:13) narrating the Korean experience quoted in Amsden and Euh 1993:387): The managers and employees of financial institutions have acquired more of a competitive spirit. They have become very sensitive to the earnings and expenses of financial institutions...these may be considered the most important performance-effects that the first round of liberalization has achieved. Our paper examines the distribution of cost efficiency amongst Ghanaian banks during the period , when the initial impacts of the reform process that started in 1986 were bedded down. Data that are available for banks are pooled together and grouped by ownership criteria and also by a two category entry period, those that were already established in the pre-liberalisation period before 1986 and those that entered subsequently. 5 There is a view, for example the view informing the Narasimham Committee Narasimham Committee Report, 1991) constituted by the Reserve Bank of India to advise on financial sector liberalisation in that country, that the entry of foreign banks and the emergence of new domestic private banks introduce better management techniques in the banking industry. There is no consensus about how long it might take for these new techniques to be impounded within the entire industry and 2 The problem of externality is an important but neglected area of research in the literature on bank liberalisation with some notable exceptions e.g. Stein, 2010). Banks have a greater external impact on other sectors of the economy than most other firms, but this externality is not picked up in conventional measures of efficiency deriving from the literature in industrial economics. For example, Stein 2010) argues that privatised banks in some countries have withdrawn from providing services to the small and medium sized industries because the information cost is greater but the cost to the economy of depriving funds to these sectors is high. Thus the issue about how to measure bank efficiency is not a settled question. This is an important issue but it is outside the scope of this paper. 3 Amsden and Euh 1993) is a study of the impact of banking sector reform in South Korea where the state did not fully distance itself from lending decisions. 4 Increase in competition in banking which may reduce cost of delivery is not to be conflated with increased productivity elsewhere in the economy Valvarde et al., 2003). 5 Ownership criteria are explained in the notes following Table 1. opinion varies, especially, about the desirable pace of reduction in barriers to foreign competition Joseph and Nitsure, 2002). A way of examining the impact of liberalisation on efficiency is to examine whether the new entrants as a group have brought more cost efficient practices into the banking system which persist even after some years to distinguish the performance of these groups of banks from the rest. This is one aspect of our study. Another aspect is an examination of the importance of ownership structure, for example whether private ownership continues to lead to greater pressure for cost efficiency than public ownership even some years after liberalisation when public sector banks are granted greater commercial autonomy, and minority equity holdings in these banks are traded in the bourse. To examine the distribution of cost efficiency amongst banks, a cost efficiency frontier is estimated. The distance of a bank from this frontier can be regarded as a proxy for managerial slack, a measure of inefficiency, in that bank. Efficiency is measured as the corollary, closeness to the frontier. The paper is organised as follows. Section 2 describes the background to financial liberalisation in Ghana, and outlines the data. Section 3 explains how the cost efficient frontier is estimated and distance from the frontier is calculated for banks. Section 4 reports the results, and Section 5 concludes. 2. Background The shift in thinking in development economics in favour of greater emphasis on the market began to make a mark on the direction of economic policy in Ghana in the 1980s. The reform of the financial sector, which began in 1986, was an inevitable component of the policy of embracing the market. Ever since the reforms were implemented in the Ghanaian banking industry, the influx of privately-owned banks both domestic and foreign) has increased steadily. Before the reforms began in 1986, there were only 11 banks operating in the economy, of which four had foreign equity participation. Foreign participation was limited to 60% of total equity by an indigenisation decree passed in The government acquired 40% of equity stake in hitherto fully-owned subsidiaries of foreign banks. Since the start of reforms, the government has sold its minority shares in these banks. To distance government from financial markets further, government s equity stakes in the public sector banks have been diluted by sale to the public through listing in the stock exchange. 6 As of December 2008, there were 25 deposit taking banks in operation, of which 13 were foreign-owned and 9 were under domestic private ownership. Three banks remained in the public sector. Whilst the pressure of public opinion has exerted a restraint on privatising fully the remaining stake of government in the public sector banks, there has been considerable entry, exit and re-structuring of institutions. A number of banks, both in the public and private sector and some of them even 6 One of these institutions Social Security Bank) was eventually merged and acquired domestically and subsequently merger with and acquisition by a foreign bank.

3 E.A. Adjei, S.P. Chakravarty / Review of Development Finance ) Table 1 Bank type. Group Description Number of observations A Private sector banks under domestic 12 ownership in existence prior to liberalisation. B Private sector banks under foreign ownership 24 in existence prior to liberalisation. C Public sector banks. 36 D Domestically owned private sector banks 74 that have emerged since liberalisation. E Foreign banks that have entered the market by setting up on their own since liberalisation. 68 foreign owned, have either been recapitalised or liquidated. There has also been balance sheet restructuring and changes in management and operating procedures in distressed public sector banks for them to be able to survive under the new liberalised regime. This regime is characterised by greater autonomy for management coupled with greater responsibility for the outcome of commercial decisions. 7 The prudential regulation and supervisory systems have been strengthened. Unlike the experience of some of the former communist countries Dow et al., 2008), there has not been increase in the share of non-performing loans in lending decisions. The Financial Sector Adjustment Programme FINSAP) in Ghana was carried out in three phases across time, namely FINSAP-1 which covered the period ; FINSAP-2 for the period ; and FINSAP-3 which started in 1995, and it is an ongoing process. The first two periods witnessed radical change in direction of economic policy in Ghana. Many of the radical restructuring of the banking system had been completed by the start of our study period in The entry of banks under different types of ownership invites the question as to whether the different methods of cost control pursued by different types of banks rub into the management practices across types of ownership. For the purposes of our analysis, banks are categorised into the following groups Table 1) and their intra-group and inter-group differences in cost efficiency are examined. The number of pooled time series and cross section data available for each type of bank is also indicated in the table. The five types of banks are further re-grouped into domestic groups A plus D) and foreign groups B plus E) ownership for further analysis. 8 7 Mergers and acquisitions do not appear to have been the focus of restructuring the banking system. Only one domestic and cross-border mergers and acquisitions has been recorded since the reforms took place. 8 A bank is defined by the Bank of Ghana to be foreign owned when more than 50% of shares are held by overseas entities. Likewise, when more than 50% of outstanding equity of a bank is held in domestic private hands, the bank is classified as a domestic private sector bank. One of the banks, SG-SSB, in Group E is a foreign owned bank by the above definition, but the ownership was acquired by taking over a domestic entity. The capital structure of the state owned banks are in tradable equity, but with more than 50% in government hand. Ghana Commercial Bank is classified in our dataset as a state-owned bank although majority participation in equity were offered to the public in 1996, but re-estimation of the model with changed classification does not alter Data for this period on banks that existed before the reform process started and new entrants, both foreign and domestic, are collected through private communication from the Central Bank, but we do not have information for the entire period for each bank. The balance sheet and income statement of Ghanaian banks, on which our data are based, are not readily available from commercial sources. The dataset available to us is an unbalanced panel covering 25 banks over the period with a sample size of 214 observations. The bank-level data is unbalanced because new entry has taken place in the wake of reforms. Two state-owned banks were liquidated during our data period, and they are not included in the study. The number of banks included in our study varies from 14 in 1997 to 24 banks in These are all engaged in some form of commercial banking, amongst other activities, and can be regarded as universal banks. Other financial institutions engaged in lending activity, for example rural banks, have been excluded from this study because they operate under different legislation. All the development banks in Ghana undertake some form of commercial banking activities apart from their usual operations. Institutions that were operating as merchant banks prior to liberalisation had their license changed into universal banking. Since 2003, all the three-pillar banking model commercial, development and merchant have been replaced by universal banking. In light of this, data available for banks in the categories of commercial, development and merchant have all been included in the study. 3. Cost efficient frontier Much attention has been devoted in the literature over the past few decades to the study of the efficiency of financial institutions. Banking institutions, insofar as they pool risk, operate in an oligopolistic market which is also characterised by its own special kind of information constraints. There is potential for the existence of X-inefficiency, a term coined by Leibenstein 1966), in institutions where employees managers) cannot be fully monitored. This has given rise to research in industrial economics comparing individual firms against a best practice efficient frontier e.g. Berger and DeYoung, 1997; Fries and Taci, 2005). Managerial slack, or X-inefficiency, is measured by distance from the efficient frontier. Cost efficiency is measured by estimating a cost function which describes the relationship between a cost variable as the dependent variable, and a set of explanatory variables that include outputs), input prices, and proxies for differences in the economic and regulatory environment, random error, and inefficiency. Thus each bank is benchmarked against the best bank in the sample. We estimate cost efficiency using a stochastic frontier approach originally proposed by Aigner et al. 1977), Meeusen and van den Broeck 1977) and Battese and Corra 1977). the estimated coefficients of the model and efficiency scores do not change. Also, the composition of the board of directors did not alter after government divested itself of majority equity.

4 96 E.A. Adjei, S.P. Chakravarty / Review of Development Finance ) A general version of the cost function can be expressed in a logarithm form: ln C i = f Q i,p i,k i,z; β) + ln ε i, where i = 1,...,N 3.1) The variable C i is the observed variable cost of bank i; fq i, P i, K i, Z; β) is the cost frontier which maps the predicted cost of a cost-minimising bank. On the cost frontier, Q i represents the vector of banking output for bank i; P i denotes the vector of input prices that bank i pays; K i is the vector of bank i risk indicators. Z is the vector of economic and environmental variables that may influence cost performance; β is a vector of unknown parameters to be estimated; and ε i is a two-component error term that explains the deviation of a banking firm s observed cost from the cost-efficient frontier due to a random noise, v i, and possible inefficiency, u i. Thus, ε i = v i + u i 3.2) where both v i and u i are assumed to be independently and identically distributed. The term v i is an independent normal distribution with zero mean and a variance of σv 2. It captures the effects of measurement errors and random effects such as good and bad luck that are out of management control but capable of affecting bank performance. The second stochastic term u i is assumed to be non-negative, but it is otherwise independent and normally distributed with a positive mean and a variance of σu 2. This term captures the effects of inefficiency due to management error that are specific to the firm. The mean of the conditional distribution of the inefficiency term u i, given the composite error term ε i, is the X-inefficiency score in this model. Further manipulation of the algebra allows for an efficiency score to be estimated for each bank in the sample as a parameter in this model. The score is bounded by 0 and 1, where a score of unity represents best practice. The software Frontier 4.1 developed by Coelli 1996) is used here. Efficiency scores are estimated following an algorithm proposed earlier Battese and Coelli, 1992). Data for banks is available for the period , but not for the entire period for every bank. Estimates of the above efficiency scores are reported for each group of banks and expressed as percentage efficiency Table 4). 9 Further details are contained in Appendix A. 4. Results The results presented here are the efficiency scores that are computed for the five groups of banks categorised by ownership and listed in Section 2 above. As explained in Footnote 9, average efficiency scores, grouped by types of banks, are reported as percentage figures in Table 4 below. The closer is a group of banks to the cost minimisation frontier, the higher is the 9 The parameter measuring distance from the efficient frontier that is estimated lies between zero and unity. For example, if the estimate for a bank is unity, it would be reported as 100% efficient, but if the parameter estimate for this bank is only ½, it would be reported as 50% efficient. Table 2 Theil index of dispersion of cost efficiency scores and decomposition into intergroup and intra-group components for five different groups of banks. All five groups of banks Inter-group Intra-group Percentage contribution to the total index Table 3 The percentage contribution of inter-group and intra-group dispersions to the total Theil index of dispersion of efficiency scores for binary groups of banks. Groups Description Inter-group Intra-group A and B A and C A and D B and D B and E A + D) and B+E) Domestically owned private and foreign owned banks existing prior to liberalisation Domestically owned private banks existing prior to liberalisation and public banks Domestically owned private banks existing prior to liberalisation and new domestically owned entrants Foreign owned banks existing prior to liberalisation and new domestic entrants Foreign owned banks existing prior to liberalisation and new foreign entrants Domestic ownership versus foreign ownership efficiency score. The regression results used in this exercise are also contained in Appendix A. The Theil index for the dispersion of efficiency scores over the entire banking system is calculated and then additively decomposed into inter-group and intra-group components, as explained in Appendix A. The inter-and intragroup contributions to the total index of dispersion are listed in Tables 2 and 3 below. It appears that inter-group differences are over-shadowed by differences in efficiency within groups when the entire sample is examined as in Table 2. A more nuanced story emerges when a binary comparison is made between bank types as in Table 3. Whilst it is true that domestically owned private banks in existence prior to liberalisation do not do better or worse as a group than their newer counterparts which started operation in the wake of liberalisation, the story is not repeated when the comparison is made between foreign and domestic ownership, but only if data are partitioned by time of entry. For example, a comparison of A and B, banks that were in existence prior to liberalisation, suggest that the inter-group differences by foreign versus domestic ownership dominate the Theil index. Also, the average efficiency score of B is greater than that of A. Before jumping to the conclusion that foreign ownership, per se, leads to greater cost efficiency, based on data partitioned by

5 E.A. Adjei, S.P. Chakravarty / Review of Development Finance ) Table 4 Group efficiency scores. Bank type Efficiency score %) Theil index A Existing private domestic B Existing foreign owned C Public sector D New private domestic E New foreign owned All Entire sample A + D All private domestic B + E All foreign owned time of entry pre- and post-liberalisation, a comparison needs to be made using a larger dataset containing both the entry period. When all those that are foreign owned are grouped together B + E) and then compared with all those that are owned by the domestic private sector A + D), within group differences in efficiency scores come to dominate the total index of dispersion see Table 3). 10 Whatever the advantage that foreign ownership may confer, it is not impossible for it to be impounded by the domestic banks. It is also interesting to note that historical baggage notwithstanding, public sector banks can register high efficiency scores Table 4) some years after liberalisation and restructuring, and operating with greater commercial autonomy. 5. Conclusions Liberalisation of the banking sector can take different forms and the idea of what constitutes reform is not unique. For example, Amsden and Euh 1993) describes banking reform in Korea that did not dispense with an industrial policy directing investment funds to preferred sectors at lower interest rates. However, a common factor in liberalisation is the reduction of barriers to entry by both domestic and foreign players and the emergence of competition against public sector banks which are now allowed greater commercial autonomy. There is diversification in ownership and governance and the literature suggests that these developments have impact on bank efficiency Fries and Taci, 2005). For example, some authors suggest that foreign banks possess superior management practices and technological advantage over local banks, and as such, are expected to capitalise on their advantages and exhibit higher efficiency levels than their local peers Claessens et al., 2001). All banks taken together, intra-group variations contribute almost twice the amount to the dispersion in efficiency than inter-group variations when banks are grouped by ownership type. However, our evidence suggests that time may even out any advantage of ownership type. Foreign owned banks that existed prior to liberalisation have, on average, done better than those that came in the wake of liberalisation, but there is considerable variation in efficiency within banks under similar ownership type Table 3). 10 Note that the foreign owned banks as a group B + E) register higher efficiency scores than banks under domestic ownership A + D), the index of dispersion is substantially higher for the former. Appendix A. The cost function We use operating cost as dependent variable in our efficiency model and proxy this variable as the sum of personnel and other non-interest expenses. We specify two outputs loans and deposits) and two inputs labour cost and the price of deposits) and the frontier cost function model attempts to capture the least cost method of generating outputs from inputs. With respect to input price variables, we proxy the price of labour as the ratio of personnel expenses to total assets, 11 whilst the price of deposit is proxied as the ratio of interest expenses to the value of total deposits. This is an input price that is proxied by the ratio of interest payments to the amounts of funds purchased. In the case of banks considered here, funds purchased by a bank are the deposits placed in the bank by customers. The operating cost OC it ) in the regression equation below is deflated by the price of deposits. To capture the missing time dimension that is not explicitly modelled in the cost function, we include a time trend variable that accounts for the effects of technological factors, together with other factors such as regulatory changes. We also include three bank-specific indicators of credit, capital and liquidity risk to control for differences in risk profile. These indicators reflect the variation in the risk-taking strategies across banks. We measure credit risk as the ratio of provisions for bad and doubtful debts to gross loans; capital risk as the ratio of shareholders funds to total assets; and liquidity risk as the ratio of liquid assets to total bank liabilities. Based on the afore-mentioned variable definitions, we specify our preferred cost efficiency model on a two-output, two-input trans-log functional form. The efficiency estimate is derived from a cost function described by the equation below: ln OC it = α 0 + δ p ln w it + δ L ln LN it + δ D ln DP it δpp ln w it) δ LLln LN it ) δ DDln DP) 2 + δ LD ln LN it ln DP it + δ PL ln w it ln LN it + δ PD ln w it ln DP it + λ TP T ln w it + λ TL T ln LN it + λ TD T ln DP it + λ T T λ TT T 2 + θ R1 ln CRD it + θ R2 ln CAP it + θ R3 ln LIQ it + v it + u it A.1.1) where the subscripts i and t denote the ith bank in the tth time period. 11 Total assets comprise cash and short term funds, investments breakdown into components such as government securities, Bank of Ghana bills etc. is available only for some years), loans including overdrafts) and other assets. This last term comprise of investment in subsidiaries, plant and real estate including premises purchased for employee accommodation. Again, the data that has been made available to us does not contain breakdown into components for all banks.

6 98 E.A. Adjei, S.P. Chakravarty / Review of Development Finance ) Table A.1.1 Panel estimation of stochastic cost efficiency frontier. Variable Parameter Coefficient Standard error T-ratio Constant α ln w δ P ln LN δ L ln DP δ D /2ln w) 2 δ PP /2ln LN) 2 δ LL /2ln DP) 2 δ DD ln LN ln DP δ LD ln w ln LN δ PL ln w ln DP δ PD T λ T /2T 2 λ TT T ln w λ TP T ln LN λ TL T ln DP λ TD CRD θ R CAP θ R LIQ θ R The variables are explained earlier except that two changes are made in the way that they enter the regression equation above. The variables w it is the ratio of the price of labour to the price of deposits. Thus the two input prices are incorporated within the above variable. Secondly, all variables are transformed into their natural logarithm in specifying the regression equation. Variables CRD it, CRD it, CAP it, and LIQ it control for credit, capital and liquidity risk, respectively. The variables LN it and DP it are loans and deposits. The variables T and T 2 are linear and quadratic time trends, respectively. The random noise terms are discussed in Section 3 above. As explained in the text, the Coelli algorithm calculates the distance from the efficient cost frontier for all banks in the sample and the scores are aggregated by bank groups A to E. The scores, which lie within the closed set 0, 1) are expressed as percentages to lie within 0, 100) in the text Table 4) for different types of banks. Appendix B. Theil index of dispersion Suppose that there are n number of banks. The efficiency score is y {=y 1,..., y n }. Then the Theil index Theil, 1967), can be written as Cowell, 1977:55 62): T y, n) = 1 n ){ {[ yi ][ ln yi )]}} A.2.1) If the data are classified according to, say ownership type, we may wish to know how much of this index is due to differences in efficiency between types of ownership, say foreign and domestic inter-group contribution to the total Theil index of dispersion), and how much of the index can be attributed to dispersion of scores amongst each group intra-group contribution to the total Theil index of dispersion). Now think of a more general case. Suppose that the number of banks is n, having scores {y i } = {y 1, y 2,..., y n }. They are classified into m mutually exclusive and exhaustive groups. There are n k banks in each group, where k =1, 2,..., m. In Group k, the score of the banks is the set {x1 k,...,xk nk }, denoted for brevity as {x k }. The groups are mutually exclusive, i.e. that no individual is a member of more than one group. The groups are also exhaustive, i.e. that the sum of n k for all groups taken together is equal to the total number of banks, n. Exclusivity and exhaustiveness also imply that the union of the sets {x k } is the set {y i }. The MLD index, Ty, n), can be decomposed as follows: [ ) κ T υ, ν) = T χ κ νκ ) ],ν κ ) { κ + ) λν κ ν )) νκ ) } A.2.2) ν where the summation is over the index k =1, 2,..., m, and = average of the values {y 1, y 2,..., y n } =y 1 + y y n )/n, and k = average of the values {x1 k,...,xk nk }. 1 T x k,n k ) = n k ) { SUM over n k {[ x k i k ][ ln )]}} xi k k The MLD index can be broken down into two parts, A and B, as follows: T y, n) = A + B, where A is the term enclosed within [ ] and B is the expression enclosed in {}in Eq. A.2.2) above. T is the Mean Logarithmic Deviation measure of dispersion. The first part, A, can be interpreted as the within-group dispersion and the second part, B, can be regarded as the between-group contribution to the index of total dispersion. References Aigner, D., Lovell, C.A.K., Schmidt, P., Formulation and estimation of stochastic frontier production models. Journal of Econometrics 6, Amsden, A.H., Euh, Y.D., South Korea s 1980s financial reforms: goodbye financial repression maybe), hello new institutional restraints. World Development 21, Battese, G.E., Coelli, T.J., Frontier production functions, technical efficiency and panel data: with application to paddy farmers in India. Journal of Productivity Analysis 3, Battese, G.E., Corra, G.S., Estimation of a production frontier model: with application to the pastoral zone of eastern Australia. Australia Journal of Agricultural Economics 21, Berger, A.N., DeYoung, R., Problem loans and cost efficiency in commercial banks. Journal of Banking and Finance 21, Claessens, S., Demirgüç-Kunt, A., Huizinga, H., How does foreign entry affect domestic banking markets. Journal of Banking and Finance 25, Coelli, T.J., A guide to frontier version 4.1: a computer program for stochastic frontier production and cost function estimation. CEPA Working Paper 7/96, Department of Econometrics, University of New England, Armidale, Australia. Cowell, F.A., Measuring Inequality. Philip Allan, London. Dow, S., et al., A stages approach to banking development in transition economies. Journal of Post-Keynesian Economics 31 1), Fries, S., Taci, A., Cost efficiency of banks in transition: evidence from 289 banks in 15 post-communist countries. Journal of Banking and Finance 29, Ibanez, D., et al., Are universal banks better underwriters? SSRN ssrn.com/abstract=

7 E.A. Adjei, S.P. Chakravarty / Review of Development Finance ) Joseph, M., Nitsure, R.R., WTO and Indian banking sector: the road ahead. Economic and Political Weekly 37, Leibenstein, H., Allocative efficiency vs. x-efficiency. The American Economic Review 56, Mahesh, H.P., Bhide, S., Do financial sector reforms make commercial banks more efficient? A parametric exploration of the Indian case. Margin: The Journal of Applied Economic Research 2, doi: / Meeusen, W., van den Broeck, J., Efficiency estimation from Cobb- Douglas production functions with composed error. International Economic Review 18, Narasimham Committee Report, Report of the committee on the financial system. Government of India, New Delhi. Park, J.Y., Financial Liberalization in Korea. Korea Development Bank. Mimeo. Seoul National University May Cited in Amsden and Yuh 1993: 387). Saraydar, E., Conflation of productivity and efficiency in economics and economic history. Economics and Philosophy 5, Sealey Jr., C.W., Lindley, J.T, Inputs, outputs, and a theory of production and cost at depository financial institutions. Journal of Finance 32, Stein, H., Financial liberalisation, institutional transformation and credit allocation in developing countries: The World Bank and the internationalisation of banking. Cambridge Journal of Economics 34, Theil, H., Economics and Information Theory. North Holland, Amsterdam. Uddin, S., Hopper, T., Accounting for privatisation in Bangladesh: testing World Bank claims. Critical Perspectives on Accounting 14, , doi: /s ) Valvarde, S.C., et al., Deregulation, bank competition and regional growth. Regional Studies 37 3),