Bank behavior and the control of credit flows : lessons from the Sri Lanka experience

111 DOMESTIC FINANCE STUDY NO. 41 BANK BEHAVIOR AND THE CONTROL OF CREDIT FLOWS: LESSONS FROM THE SRI LANKA EXPERIENCE by Jamshed K. S. Ghandi June 1977 The views presented in this paper are solely those of the author and do not necessarily reflect the official opinions of views of the World Bank or its affiliates. Public and Private Finance Division Development Economics Department Development Policy Staff PREFACE The level and particularly the structure of saving and investment are shaped to a considerable extent by the nature of the financial system. The major part of the financial system in the LDCs is the commercial banking system and hence, it was but appro- priate that one of the major research projects of the Division should relate to the behaviour of the commercial banks and the extent to which it is modified by selective credit policies. This project em- braces three countries - Brazil, Sri Lanka and India. The Brazilian study which was carried out with the assistance of the Vargas Founda- tion was the first one to be completed. Its major findings are sum- marized in a paper by Shahid Yusuf (DFS No. 39, "The Commercial Banking Industry in Brazil 1964-1974, May 1977). The research on Sri Lanka was carried out by Professor Janshed Ghandhi, a consultant and Katrine Saito. Some of the problems encountered in estimating household savings were discussed by Saito in DFS No. 27 (An Estimation of Household Sector Savings of Sri Lanka). In this paper, Professor Ghandhi presents the results obtained from his detailed study of credit controls in Sri Lanka. The third study - on India - has recently been undertaken with the assistance of Dr. Rang Rajan of the Ahmedabad Institute of Management. It is much broader in scope than the other two and will concentrate on (i) the impact of branch expansion on resource mobili- zation and credit allocation to small farmers, small industry and small enterprises, (ii) economics of branch expansion, and (iii) the impact of credit guarantee scheme on lending to small industry and internedia- tion costs of the banks. It would also examine the relationship between transaction costs and the size of the loan. This study would be of relevance to the other LDCs; it would indicate the nature and charac- teristics of financial innovations required for the purpose of viable productive lending to small enterprises in agriculture as well as industry. V. V. Bhatt Introduction Few countries are so committed to a laissez-faire strategy that they make no attempt to influence the structure of credit flows in an effort to orient the structure of production and trade more consonant with their societal preferences. In developing countries, the potential role of selective credit controls in particularly great, since market failures are usually more widespread and more significant; sectoral imbalances are typically great and the rewards for eliminating these correspondingly significant. Most selective controls are indirect in that they do not attempt directly to influence the final "real" variables. They directly affect financial variables, the cost and availability of credit to the desired sectors. Recent work, primarily by Rao and Kaminow (1973), has examined the conditions determining the degree to which changes in credit conditions will serve to alter the composition of real variables. This is a very rich line of enquiry with considerable scope for further investiga- tion, both analytical and empirical. The purpose of this essay, however, is to investigate the other dimension of the process, of why some controls do succeed, and others fail, in redirecting the flow of credit. We will argue that such instruments, to be successful, should be devised with ex- plicit attention to the utility functions of the financial institutions so that an examination of the determinants of the portfolio behaviour of these institutions is an essential ingredient in the ordered development of effect- ive policies to restructure credit flows. Our empirical work will be restricted to the behaviour of camercial banks in Sri Lanka, but the general -2- principle applies to other financial institutions and to other countries. In section I we examine briefly the conventional model of bank asset allocation which, implicitly if not explicitly, underlies the design of the selective credit controls. This model does not appear to afford a realistic description of bank behaviour in Sri Lanka and we suggest an alternative process. In section II the workings of two selec- tive credit control schemes are examined in the context of the alternative hypotheses of bank behaviour developed earlier. Although, there is no evidence that either scheme was developed with a specific bank behaviour hypothesis in mind it is shown that the nature of the factors influencing bank behaviour significantly influences the relative success in shifting credit flows. In section III the Sri Lanka experience is generalised to argue that a necessary ingredient for the development of effective selective credit controls is the prior examination of objective functions of the institutions whose behaviour it is desired to modify. It is axiomatic that the objectives of a bank should influence the structure of its assets -- given other factors such as the cost and volatility of funds, the nature of the financial system of which the bank is a constituent and the nature of the regulations imposed by the authorities- 1/ See, for instance, Robinson (1962), Hester and Pierce (1975) or Ghandhi (1976). -3- There is much uncertainty, however, regarding the specific objectives which reasonably may be presumed to characterise banks in general -- and even greater doubt when we consider the specific case of the Sri Lanka banks. a. "Return-Risk" Objective Function Thaditional expositions of banking in terms of the familiar bank expansion multipliers implicitly assume profit maximisation on the part of the banks. This is clearly improper since pure profit maimisation implies asset specialisation,- which is not observed in practice. Follow- ing Markowitz (1952) (1959) and Tobin (1958) it is customary to hypothesize a more complex utility function, positive in the rate of return and nega- tive in risk, which suffices to yield portfolio diversification and is thus considered to be a more accurate depiction of the objective function of a bank. Although they may differ markedly in other aspects, contemporary bank portfolio models evidence a common reliance upon a risk-return utility 2/ function; the specific shape of the utility function posited may differ, but it is invariably described in terms of these two arguments. 1/ This conventional conclusion may not obtain if the financial institution is large relative to the market so that an effort to specialise in the highest yielding asset would drive down its rate of return to a degree that it might make the purchase of a hitherto inferior asset attractive. 2/ Normative attempts to devise optimal portfolios are sensitive to the form of the utility function, not merely its arguments; for analytical con- venience both Mrkowitz and Tobin assumed that it was quadratic, but we cannot say if this is a reasonably accurate, if stylised description of how banks do in fact act. For empirical work the generalised risk/rate of return objective function usually suffices. - ~ ~ ---...............---~.,- This approach to bank behaviour while it is clearly superior to simple profit maxrIAsation is not without limitations. It assumes a continuous tradeoff between risk and rate of return, that the banker will be willing to accept.any level of risk, however great, provided the asso- ciated return was "adequate". This assumption may be unwarranted,'even in a world without regulation. And the effect of much banking regulation undoubtedly is to reinforce such an inhibition, to truncate the utility function by discouraging and even by proscribing certain types of activities or assets. Further, commercial banks may be responsive to parameters other than risk and return in the determination of their portfolios -- they may place more or less emphasis upon factors such as the nature and size of the security, the maturity of the loan, the size of the firm to which the loan is made and the nature of the business in which it is engaged, the customer relationship, etc.- Commercial banks may attempt to manage their risks by altering the attributes of the customers that they serve and the loans that they make, while portfolio theory essentially accepts the rate of return and risk as exogenous parameters and demonstrates how to derive an optimum portfolio from such attributes. Further, once the (implicit) assumption of relatively perfect competition is abandoned more complex objective functions may be manifest. Recent developments in price theory have emphasized the alternatives available to oligopolistic enterprises -- the well known alternatives ranging from sales maximisation with a minimum profit constraint and maximisation 1/ These non-price dimensions of banking have been emphasized in the less formal, the more traditional, literature. See, for instance, Robinson (1962) or Hodgman (1963). of the rate of growth of the firm to more loosely formulated objectives 1/ in terms of "managerial discretion". Empirical investigations, primarily by Ecwards and Heggestad (1973) and Edwards (1977), indicate that oligo- polistic banks appear to have objective functions different from those of competitive ones. In the Sri Lanka context this may be particularly significant since the banking system is highly concentrated -- originally dominated by the foreign banks and more recently by the two publicly owned banks. Whether or not the conventional risk-return approach affords a realistic description of bank behaviour in a particular context is an empirical question and cannot be determined on a priori grounds. Unfort- unately, a thorough empirical examination of the Sri Lanka banks is not possible. Ideally, behaviour needs to be studied on an individual bank basis. But the published accounts of Sri Lanka banks do not provide the detail necessary for examination of the hypothesis, even for the alloca- tion of assets between very broad categories, such as cash items, govern- ment securities and commercial loans and advances. We have been compelled, accordingly, to rely upon aggregate data. The test used and ito predominantly negative results have been 2/ reported at greater length so that only a very brief summary is attempted 1/ See, for instance, Baumol (1967), Penrose (1959) and Williamson (1963) (1964). 2/ In Ghandhi (1976), pp. 125-135. -6- here. Essentially we conflate the inventory theoretic and the pure port- folio approaches to banking to obtain a reduced form test similar in substance though not in detail to that utilised by- Hinderliter and Rockoff (1976). The proportion of cash assets held is a function of the volume of total deposits, the structure of deposits, the interest rate on treasury- bills and a trend variable to reflect secular changes in the structure of the banking system. The proportion of assets held in government securities was hypothesised to be a function of the return on treasury bills relative to that on commercial loans, the structure of deposits and a trend variable. The results obtained, whether for the banking system as a whole or for Sri Lanka owned banks and foreign owned banks separately, are largely negative. The coefficients obtained were often of the sign opposite to that hypothesised by the model and, in any event, were rarely significant beyond the 10% level. Admittedly, the use of aggregate data, rather than micro-bank data, create problems of interpretation, but the results are so poor as to leave little doubt that the customary return-risk model does not afford a realistic description of the decision processes of the banks in Sri Lanka. This does not necessarily mean, however, that the Sri Lanka banks do not take into account risk and the rate of return in the determina- tion of their portfolios. The failure of the simple risk-return approach may be the consequence of the banks incorporating to a significant degree other variables, in addition to risk and return, into their decision pro- cess. Only if the covariance of these variables with those included in our regressions were zero would their neglect have no effect on the size and significance level of the estimated regression coefficients. Alterna- tively, the banks may evaluate risk and return in a different manner from that implied in our regressions so that the variables we utilised may be poor proxies of the variables actually examined by the banks in their decision process. We believe that the conventional risk-return approach, while it directs attention to important factors, fails to provide an adequately realistic description of the allocative process in Sri Lanka because it is too general and neglects essential structural features of the system. Sri Lanka bankers were nurtured in the traditions of British banking, particularly of British colonial banking, and these traditions did not disappear with the granting of self government -- or even with the estab- lishment of the (state owned) Peoples Bank and the nationalisation of the Bank of Ceylon Similarly, it is essential to recognise and incorporate into the analysis important attributes of the Sri Lanka economy which may influence asset choice by the banks. b. Portfolio Policies of Sri Lanka Banks: An Eligible Assets Approach Consonant with their British tradition, the Sri Lanka banks lend primarily by the purchase or discount of bills, by granting overdraft facilities- and by the granting of formal loans. Overdraft limits are reviewed periodically; formally, overdrafts are callable upon demand al- though the banks will customarily give notice of a required reduction. Loans are, in contrast, for a specified maturity. In practice, there may not be much difference between loans and overdrafts since most loans are 1/ Overdrafts dominated formal loans initially, but the ratio of loans to overdrafts has steadily increased so that by 1975 loans exceeded over- drafts by about 75%. -8- for short duration. Typically, loans and advances are secured, although the form of the acceptable security and, to a degree, its margin may vary between customers. The amount of security also varies somewhat with the form of the security since different assets differ in marketability and certainty of value -- it is reported that the loan is usually between 50% and 60%, rarely as much as 70%, of the value of the surety taken. The perponderant majority of bank loans are short termi and amply secured; as 2/ a consequence they are invariably liquid and safe. This emphasis upon liquidity and safety is reflective, in part, of an undue adherence to the (outmoded) principles of the Banking School,. which stipulated that a bank should prudently make loans only on the basis of short term self-liquidating commercial paper. More recent developments in banking theory, not to mention changes in the structure of funds, appear to have had little impact upon the conception of a suitable loan. To a degree, further, the heavy emphasis upon the finance of foreign trade also directed 1/ In practice, the actual term of a loan may be greater than the legal maturity. Particularly favoured customers often are given a tacit under- standing that overdrafts, usually to a limit, will not be called, or that short term loans will be renewed automically -- "rolled over" -- at maturity. Further, banks may in effect make longer term loans when they discover that a short term loan that they have made cannot be repaid at maturity. Although they may legally do so, it is unlikely that 'the bank will immediately foreclose on the loan or, in the event of the security being insufficient, compel reorganisation or liquidation. In the interests of a good long term relationship the customer often is "carried", the loan renewed, in the hope that it will eventually be repaid without the necessity of realising the security. 2/ This approach to banking is neither new nor restricted to Sri Lanka banks. For example, Keynes (1913) observed of Indian banks that they would not make loans a) for a duration greater than six months; b) upon personal security or upon the security of immovable property; or c) upon promissory notes un- less they were countersigned by at least two independent names. Loans were made primarily upon the security of inventory or of government bonds and shares of joint stock companies. the banks' attention towards short term, highly secured, loans; the decline in the relative importance of external trade did not result in a commensu- rate adjustment of loan attributes. It may be noted, more appropriately, that there are no widely acceptable and reliable credit rating agencies so that data on credit risk, pooled and systematised from a variety of sources, are not available. In addition, other thar the public joint stock companies and possibly some of the larger private companies, customers rarely main- tain accounts in a form, style and accuracy which would give outsiders an accurate picture of their operations. Finally, the problems of monitor- ing loans are very great, and expensive in terms of qualified manpower, so that banks seek to simplify operations and reduce costs and risks by re- 1/ cours6 to easily identifiable security. With the possible (partial) exception of their larger and more established customers banks seek their security for loans not in indicators of the capacity for repayment but rather in the quantity and quality of the collateral. And a prudent banker is one who attempts to ensure that the value of the asset so pledged is always 2/ greater than the amount of loan due. 1/ The Bank of Ceylon Commission (1968) observed that "even long standing cus- tomers of the bank have misused credit by diverting funds granted for one purpose to another without the knowledge or consent of the bank" (pare. 186) and attributes in part at least the bias towards security to "these tendencies to erratic behaviour" on the part of borrowers. 2/ N.U. Jayawardena, from the vantage point of a career in business, finance and governmental service, argues that the inherent tendencies towards risk aversion of the banks have been exacerbated by the legal process. Land titles are frequently in doubt and there are major limitations upon the collection of debts secured by immovable property. "The delays in the recovery of debt are proverbial; avoidance of summons and writs, and vexations pleas are the rule." See Jayawardena (1965), pp. 1-17-118. The consequence of such complicated, uncertain and time consuming procedures is not merely to raise the cost of credit; it has undoubtedly operated to orient the banks towards very safe customers and very safe types of loans. - 10 - The evidence in support of the assertion that the banks place heavy, almost exclusive eirphasis, upon the availability of suitable security in the making of commercial loans is fragmentary, but varied. For benchmark dates, the Central Bank of Ceylon has collected data on the distribution of bank advances outstanding, defined comprehensively to include bills purchased and discounted, overdrafts and loans. At Sept- ember 30, 1956, the first date for which data are available, unsecared loans accounted for 4.8% of total bank advances outstanding. This proportion increased fairly steadily until December 1963 when such loans accounted for 13.1% of total outstandings. However, since that date there has been a marked shift towards requiring security so that unsecured loans accounted for only 4.1% at the end of 1970 and were but 1.1% of outstandings at the end of 1975. Admittedly such data report only whether or not collateral was required for a specific loan, but there is no evidence that the value of the required security has altered so that it is reasonable to conclude that the role of collateral in the loan making process, always dominant, is now overwhelmingly important. The importance of the security syndrone has also been reported by other students of Sri Lanka banking. For instance, Jayawardena (1965), p. 113, notes that "the prerequisite for an advance is ample security and preferably tangible security ... besides being ample, the security must be prime and stable in value, and absolutely prime in title .... Bank lending has become so cautious as almost to eliminate the risk in lending."t Similarly, on the basis of a questionnaire survey of users, Balasuriya (1972), p. 227, concludes that usafety is the most important guiding principlent in the granting of a bank loan, and that such safety is sought in a large security margin. Even the official parliamentary Bank of Ceylon Commission (1968) para. 195 affirmed, IlWe find the main burden of the evidence tends to show that the Bank has erred on the side of caution in its lending," so that this view- point is not restricted to acedemic or outside critics of the banking system. Finally, the pattern of interest rates prevailing in Sri Lanka lends support to our hypothesis of the inordinate emphasis placed upon the availability of security. In a flexible and efficient credit market a diversity of interest rates are observed, reflecting differences both in the attributes of the loan as well as those of the borrower. Uniform and rigid interest rates imply the presence of credit rationing and/or a homo- geneity of attributes. Despite their limitations the available data in- dicate that in Sri Lanka the rates charged by the commercial banks in- sufficiently reflect the heterogeneity of potential attributes of loans or of borrowers. Official data on true actual rates charged on loans are not available and our attempts to examine a sample of loans were not successful. However, the Central Bnk collects data on maximum and minimum rates charged on advances by the banks, classified by differences in the form of security pledged. (Table 1). It is recognized that interest rates, if set at too low a leiel, may be avoided by various devices -- by requiring large compensating balances, by charging fees and commissions and even by requiring the purchase of insurance or other services which can be provided at a high charge, so that officially quoted interest rates may have a tenuous relationship to the true charge. Our investigations suggest that very little recourse, if at all, is had by the banks to such practices. offici als at the Central Bank and the commercial bankers we interviewed, -12 - %,hile admitting the possibility of such practices, denied that they were utilised; one banker responded that there was no need for recourse these measures since the available rates of return were "adequate". Officials at other governmental departments and at other financial institutions re- ported no knowledge of such procedures and even the borrowers we interviewed denied their operation.- 2/ Although interest rates vary markedly with the ty-pe of security their administered nature is immediately apparent. The maximum and minimum interest rates for many types of loans remain unchanged for many years, during which period credit conditions varied considerably. For instance, the treasury bill rate -- which is itself somewhat an administered, rate, especially in recent years since the holders of bills have increasingly been publicly owned enterprises and the provident funds -- shows a coefficient of variation of .54 for the period 1951-1975, while that for maximum interest rate charged on bank advances secured by inventory is .20. 1/ It may be noted, parenthetically, that the Monetary Law Act envisaged the possibility of such means of evasion and armed the Central Bank with, powers to combat it. "For the purpose of preventing evasion of any order" controlling maximum interest rates, "the Monetary Board may from time to time prescribe the maximum rates which may be paid to or charged by commercial banks in the form of commissions, discounts, fees or other payments whatso- ever." Monetary Law Act (1909), #103 (3). 2/ Assuming a similar margin is required it is reasonable to expect interest rates to vary with the type of security, since some forms of collateral have lower risk, market or credit, than others. Thus we should expect loans secur- ed by pledges of goverment securities to carry a lower interest charge than those secured by company shares. It will be noted, however, that this hypothesis is not corroborated with respect to charges on loans secured by inventory and those obtained by pledges of real property. Table COMMERCIAL BAJKS' ADVANCES RATES., 1951-1975 Percent Percent End of Year Blls Purchased .Loans and Overdrafts and Discounted Secured by Unsecured Government Securities Sharea of Joint Stock Stock in Trade Imm~vable Property Others Companies MAx. Min. Max. Min. Max. Min. Max Min. Max. Min. Max. Min. Max. Min. (1) () (3) () (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) G5) 1 7.5 2.5 15.0 3.0 5.0 2.25 12.0 4.0 8.0 3.0 12.0 3.0 12.0 3.0 1952 6.o 2.25 8.0 3.0 5.0 2.25 6.o 4.0 8.0 3.0 7.0 4.o 8.0 3.0 1953 6.5 3.0 -8.0 3.5 5.0 3.5 6.5 4.0 8.0 • 4.0 7.5 4.5 8.0 3.5 1954 6.5 3.0 8.0 3.5 4.5 3.0 6.5 .0 8.0 4.0 7.5 4.5 8.0 3.5 1955 6.5 3.0 8.0 4.0 4.5 3.0 6.5 14.0 8.0 4.0 7.5 14.5 8.0 3.5 1951 6.5 3.0 8.0 4.5 4.5 3.0 6.5 14.0 8.0 4.0 7.5 4.5 8.0 3.5 19571 6.5 3.0 8.0 4.5 5.0 3.0 7.0 4.5 8.0 14.5 7.5 4.5 8.0 3.5 1958 6.5 3.0 8.0 4.5 5.0 3.0 7.0 14.5 8.0 14.5 7.5 4.5 8.0 3.5 195.) 6.5 3.0 8.0 11.5 5.0 3.0 7.0 4.5 8.0 4.5 7.5 4.5 8.0 3.5 1,)" 8.0 4.0 8.0 4.5 7.5 jl.5 8.0 5.5 8.0 4.5 8.0 5.5 9.0 4.0 1961 8.0 4.5 8.0 5.0 7.5 4.5 8.0 5.5 8.0 5.5 8.0 5.5 9.0 4.0 1962 8.0 3.75 8.0 5 0 7.5 3.5 8.0 5.5 8.0 6.0 9.0 6.0 8.0 5.0 1961 8.0 3.75 8.0 5.0 7.5 3.5 9.0 5.0 8.0 6.0 9.0 6.0 8.0 5.0 1965 8.0 4.5 9.0 5.0 7.0 4.5 9.0 6.5 9.0 6.o 9.0 .0 9.0 5.5 1965 8.0 4.5 9.0 7.0 7.0 5.5 9.0 7.0 9.0 6.o 9.5 7.0 10.0 5.0 1966 8.0 Ik.5 9.0 7.0 7.0 5.5 9.0 7.0 9.0 6.o 9.5 7.0 10.0 5.0 19f* 8.0 4.5 9.0 7.0 7.0 5.5 9.0 7.0 9.0 6.o 9.5 6.5 10.0 5.0 1969 9.0 5.5 10.0 7.0 8.5 5.75 9.5 7.0 9.5 7.0 9.5 7.0 11.0 5.5 19709 9.0 5.5 11.0 7.5 8.5 5.75 10.0 7.0 11.0 7.5 10.0 7.0 11.0 6.o 197o 10.0 6.5 12.0 8.5 9.5 6.5 11.0 9.0 12.0 8.5 11.0 8.0 12.0 6.5 1971 10.0 6.5 12.0 8.5 9.5 6.5 11.0 9.0 12.0 8.5 11.0 8.0 12.0 6.5 1912 10.0 6.5 12.0 8.5 9.5 6.5 11.0 9.0 12.0 8.5 11.0 8.0 12.0 6.5 1973 10.o 6.5 12.0 8.5 9.5 6.5 21.0 9.0 12.0 8.5 11.0 8.0 12.0 6.5 19704 10.0 6.5 13.5 8.5 10.0 7.0 12.0 9.0 12.5 8.5 12.0 8.0 12.5 6.5 1975 12.0 8.5 14.0 9.5 11.0 7.5 12.0 9.0 13.0 8.5 12.0 8.5 13.0 6.5 Notes: a) 11o data available for December 1950. But interest rate atructure at April 1951 was the same as December 1951, and it is believed it was unchanged from December 1950. SOUICE: Centra] Bank of Ceylon, Annual Reports. -13- Enquiries at the Central Bank and at the commercial banks indicate that other attributes of the loan, or of the borrower have, at best, marginal effect on the rate of interest charged or even on the margin 1/ of security- required.- "In Ceylon the interest rates are not geared to the 2/ size of the loan but ... to the type of security provided against the loan" concludes an enquiry into the finance of small business. There is no evi- dence to question this conclusion or to indicate that interest rates charged 3/ are any more sensitive to other attributes.- This dominant emphasis upon the quality and quantity of security in the loan making decision reflects an unwillingness to be exposed to risk and a determination to so manage the loan as to reduce, almost to eliminate, the possibility of a default.- The stringent and narrowly defined criteria used for assessing the eligibility of loans, are not adjusted or altered 1/ The major qualification to this were the loans made with the express intention of rediscount with the Central Bank under the privileged access provisions -- in effect this means agricultural credit, both for production and for purchase of paddy under the guaranteed price scheme. 2/ Balasuriya (1972), p. 377. 3/ Bank of Ceylon data on interest rates charged on loans classified by type of security and cross classified by purpose of loan shows very little variation in interest rates,in consequence of changes in purposes. See Bank of Ceylon Commission (1968), Appendix G, Table VIII. There is no reason to question the relevance of this experience to the rest of the system. I/ There are no published data on the bad debt experience of banks, but the Bank Supervision Department of the Central Bank reports that actual bad debts are very few. significantly in a deliberate attempt to clear the market, especially in the short term; they are, rather determined by tradition, the dictates of prudence and the desire to minimize the exposure to risk. In addition, interest rate variability (among different types of loans and borrowers) is small and reflects primarily the form of the security offered. The consequence for bank asset allocation is far reaching. Applicants meeting the specified criteria are almost invariably granted the loan, and these loans constitute the first claim on bank resources, together with the legally required reserves (deposits at the Central Bank and, for significant periods, substantial proportions of vault cash). Vault cash and correspondent accounts constitute the next claim on resources, since both these items facilitate -- and, in minimal quantities are necessary to -- operations, although there is no evidence that significant efforts are expended to economize such items. Residual funds are placed in government securities. This sequential allocation of assets -- with loans and advances being determined by the amount of loan applicants which met the eligibility criteria and the portfolio government securities being determined by the residual funds available -- involves an implicit assumption of no signifi- cant shortage of funds to the system. In fact, the banking system has been characterised by surplus funds over considerable periods of time, as is indicated by the fact that bank reserves have usually exceeded minimal requirements and that there has been little recourse to borrowing from the Central Bank if we exclude funds obtained under the provisions of the special credit schemes. Apparently the eligibility criteria are so severe that there may be a shortage of effective demand and the alternative approach we indicate is appropriate. The structure of the banking system reinforces this tendency. In the past overly strigent eligibility criteria resulted in a shortage of effective loan demand and thus in surplus funds, the foreign owned banks were customarily able to divert such resources to the head office so that they were generally fruitfully utilised; there was little real in- centive to lower eligibility requirements in an effort to develop new loan business. The Ceylonese owned banks had a greater incentive to adapt eligibility criteria. However, prior to 1961 the foreign banks dominated quantitatively and the traditions and customs of British banking were paramount. The Bank of Ceylon, despite its semi-official status, relied heavily upon correspondent banks abroad and was cognisant that it would be judged by the very same criteria, whether or not appropriate, that were applied to expatriate banks. The situation did not change significantly after 1961 when the Peoples Bank was established and the Bank of Ceylon brought under full goverment ownership. This was a period of a massive expansion of resources for Sri Lanka banks since, in 1961, foreign owned banks were not allowed to open new accounts, but the Sri Lanka banks apparently did not respond with different or less stringent eligibility criteria. That custom, risk aversion and the traditional approach to risk management were strong may be only a part of the explanation. As nation- alised enterprises they had little direct incentive to make greater profits, especially if the higher potential profits were available only by under- 1/ Managerial compensation was not directly linked to performance. - 16 - taking greater risk and engaging in what might be considered unorthodox banking practices. Nationalised enterprises are particularly sensitive to comment and criticism by politicians, the press and the publicj. and experience elsewhere suggests that managers of such enterprises are usually reluctant 1/ to expose themselves to the charge that they had taken "undue" risks. It is the way of the world, from which Sri Lanka is not immune, that the aban- donment of orthodoxy is characterised as "bold and imaginative[, if a success and "dangerously rash and imprudent" if a failure. Given a situation where the benefits of risk taking do not accrue directly to the manager, but the potential costs of criticism under failure are very real, it is no wonder that safety is so often sought in orthodoxy and in following precedent. 1/ The Bank of Ceylon Commission was charged by its terms of reference to investigate "whether the Bank has granted any loan, overdraft or other accomodation... without adequate security or adequate safeguards or contrary to accepted principles of banking practice." See Bank of Ceylon Commission (1968) p. ii, and chapter 8. -17- II Following the taxonomy developed by Silber (1973), a variety of selective credit controls can be devised, to operate on borrowers or on lenders, to encourage certain activities or to discourage others. Sri Lanka's experience with such controls have been limited to efforts to en- courage the banks to increase their loans to non-plantation agriculture, and to lengthen the term maturity of their loans. Despite significant differences in the operating details of these schemes -- which, we will argue, have affected the relative efficiency of their operations -- they share a common approach. Each scheme attempts to influence the flaw of bank credit to favoured uses by allowing the banks to rediscount eligible paper, suitably defined, at preferential rates at the Central Bank. Thus they operate on the lenders, to encourage favoured activities, by attempting to increase the rate of return on such operations. a. Agricultural Credit The New Agricultural Credit Scheme (NACS) was established in September 1967 to replace the direct provision of agricultural credits by the Government, originally the Department of Food Production and from 197 the Department of Agrarian Services. Apparently the farmers tended to regard the credit provided by the Government departments as outright grants, and defaults were very high. It was hoped that transferring the credit scheme to the commercial banks, primarily the Peoples Bank, would dispel such a misconception and establish the provision of agricultural credit on a sounder basis. - 18 - The farmer obtains a loan -- subject to a maximum -- from his cooperative society for 240 days at 9 percent interest; if the loan is not repaid on schedule, it carries a penalty of 3 percent above.the original rate. The co-operative borrows from the Peoples Bank for 170 days at 5 percent, it too being subject to a 3 point penalty for over- due loans. The Peoples Bank, in turn, may rediscount such agricultural loans at 1.5 percent at the Central Bank. In addition, for a fee of 0.5 percent, the Central Bank will meet 75 percent of any default loss exper- ienced by the People Bank on such loans. It will be noted that the Central Bank defines eligibility for discount at privileged rates not merely in terms of the purpose for which the loan was made but also dictates the interest rates charged on, and the maturity of, the original loan. The NACS also carries a significant guarantee provision; the guarantee was not extended to 100 percent of the loan in an effort to provide the Peoples Bank with an incentive for collection from the co-operative society. The Central Bank further provides rediscount facilities at preferential rates to finance the purchase of paddy under the Guaranteed Price Scheme (GPS). The farmer wishing to avail himself to the GPS must sell his paddy to the cooperative societies which finance this operation by discounting usance bills with the People Bank -- which, in turn, is able to rediscount these bills at the Central Bank at the special rate of 3 per- cent. Unlike the NACS, however, there is no problem of default since the transfer of the paddy to the Government provides the co-operatives with the funds to repay the commercial bank from which the loan was taken. -19- In Table 2, column 8, we show the ratio of total amount borrowed by the commercial banks from the Central Bank under these schemes to the total amount of advances outstanding to the agricultural sector. The importance of the Central Bank as the ultimate source of the bulk of agricultural credit will be noted. With the exception of two years the commercial banks rediscounted over 50% of their agricultural paper and thus were primarily conduits. Indeed, these data understate the importance of the Central Bank in the provision of funds. Both the GSP and the NACS apply only to non-plantation crops, while our data on bank advances to agriculture (Table 2, col. 2).include the advances made to the plantations. Commencing in 1969 the Central Bank surveys identify bank advances outstand- ing for paddy, livestock and diary (Table 2, column 3). We attribute the remainder of agricultural advances to the plantation sector. We are aware that other crops -- what are referred to as "subsidiary food crops," such as chillis, maize, onions, etc. -- are also included in the NACS but no data are available on advances for these activities. The consequent over attribution to the plantation sector is, however, believed not to be large, since the bulk of NACS loans are made for paddy. It is unlikely that significant bank credit was provided the non-plantation sector prior to 1967 so that the provision of funds to agriculture, as noted from Table 2, column 3, is quite impressive. However, it will be noted, from column 9, that a significant proportion of these loans have been rediscounted so that the Central Bank, and not the commer- cial banks, is the ultimate source of the funds to agriculture. The Table .. Bank Advancew Outet.Ading to Ariculture, 1966-1915 Ila. millIn bank Advancen Outatanding Dua to Central 14on1-Plantation 10) to Agriculture Bank for Refinance (7)/(2) (7)/(3) Agricultural Loane Bank Advancen Outatanding of Agricultural løane Finwiced fram Com- tl Pady Other-mainly Under Inder Total mercial Banke' Own Funda n:1, of Year Itvectock Plantatkon GI'S 1ACG (3) - (7) and Diary Ra. million (1) (2) (3) (4)- (5) (6) (7) (8) (9) (10). (11) 1966 40.4 .A. .A. 25.7 a 25.7 63.6 1967 102.,4 .A. .A. - 59.5 59.5 58.1 1968 10.3 LA. LA. 13.7 62.7 76.4 54.5 1969 176.6 120.3 56.3 8.7 57.6 66.3 37.5 55.1 5h.0 3.6 19ir0 203.0 132.1 70.9 98.0 50.7 148.7 73.3 112.6 1911 226.7 153.1 73.6 93.9 59.7 153.4 67.8 153.6 1972 248,1 1714.0 14.1 76.3 16.2 92.5 37.3 53.2 81.5 3.8 1913 341.2 263.4 77.8 82.0 126.0 208.0 61.o 79.0 55.4 2.5 97)ri4 4o8;6 351.6 57.0 145.7 135.0 280.7 68.7 79.6 70.9 2.2 iv. >, 1'y5 453.5 372.7 80.8 193.0 40.2 233.2 51.11 62.5 139.5 4.1 tes: al The NAt,S wta in5tituted in 1967. STh,e figure derived fram (3) - (7) la negative, which le ion-vlid mInce the banke caniot rediacount more than 100% of their eligible paper. IL le a meamsure of the potential under estmatlon of non-plantation loans in (3) by aur inability to .nolude loani to "oubeidiary food cropm." r hICE: Central Bank reginters. O - 20 - commercial banks have operated largely as conduits, passing onto the co-operatives, and ultimately onto the farmers, the funds that are ob- tained, in large part, from the Central Bank. If the objective of the authorities was to increase the flow of credit to non-plantation agricultural uses it would be valid to conclude that the GPS and the NACS were successful in that credit to these activities increased significantly upon the introduction of the scheme. If the objective, however, was to provide the banks an incentive to re- orient their lending policies towards non-plantation agriculture -- by assuring access to the discount window and by compensating the banks for 75% of the default loss, thus making these loans more attractive b!r in- creasing their liquidity and their safety -- and to utilize a greater pro- protion of their own funds of these activities, the verdict must be mixed. A part of the funds devoted to this sector were obtained from the commercial banks' own resources. Our estimation of this finance, in Table 2, column 10, is baised downwards since we have underestimated the total amount of bank advances outstanding to non-plantation agriculture in consequence of the lack of data on "subsidiary food crops" which are also covered by the NACS. While this figure is very small -- in relation to the total volume of bank advances outstanding (column 11) and especially in relation to the needs of non-plantation agriculture -- it may be noted that it is not trival compared to the quantity of (their awn) funds that the commercial banks devote to the plantation sector. The official schemes have succeeded not only in making credit available to agriculture; they have also had some success, -21- adnittedly partial, in diverting the commercial banks' awn funds to such a use. What factors dete=mine the amount of agricultural paper rediscounted by the commercial banks under these schemes? The absolute amount of rediscount is limited, of course, by the amount of .loans made on the GPS and NACS. But the proportion of available paper that is re- financed varies considerably -- the observed variations are too great to be attributable, reasonably, to errors in the data resulting from the failure to include subsidiary food crops. It will be recalled that the commercial banks make cultivation loans to the cooperatives for 270 days at 5% and are allowed to rediscount such paper with the Central Bank at 1 .5%, obtain- ing default insurance for a further charge of 0.5%. This interest rate differential provides the commercial bank with an incentive to rediscount the paper since such a process would increase the net rate of return to the bank; indeed, since the proportion of eligible paper which can be rediscounted 1/ is not limited by the regulations so that the leverage available is potentially total, the rate of return that the commercial bank can obtain on such a loan 2/ and rediscount process is theoretically limitless.- A profit maximising bank 1/ There may be de facto limits in consequence of moral suasion. Let i - interest rate charged by the commercial bank I = rediscount rate (of eligible loans) at the Central Bank iB = net rate of return earned by the bank on its own resources c = transactions costs (rate) of commercial bank in making loans. p = proportion of loans refinanced, 1 > p 0 E(D) ex-ected value- of default on loans, expressed as a rate d = proportion of defaults which are to be borne by the commercial bank, 1 ..,. d .. 0 i = charge made for such credit insurance, or default com- pensation, expressed as a rate Then -L - UR p + E(D)d + iD + c) 3:B 1-p Therefore, provided iL R( + (D)d +1D + 1. iB> ( L - (E(D)d + iD) - c) ,. if p > 0 2. gi. 0 and becomes limitless as p approaches unity. - 22 - would arbitrage indefinitely, making eligible loans, rediscounting these and using the proceeds to make further loans -- until the supply of loans eligible for rediscount at preferential terms was exhausted or the in- creased supply of funds drove down the interest rate that the bank could earn oo that arbitrage was no longer profitable. In Sri Lanka this latter possibility is not relevant since the interest rate is pegged; the avail- ability of eligible paper is more significant, since there are both 1/ quantitative and qualitative conditions- which restrict the supply of such assets. The possibility of default-assuming that not all of the default risk can be transferred to the insuring agencies -- constitutes a further 2/ constraint, provided we replace the assumption of profit maximisation - with an unconstrained return-risk objective. As the bank rediscounts its loans and uses the proceeds to purchase new agricultural paper, leading to further rediscounting, the risk to the bank rises on two accounts. First, as the absolute volume of loans th8t it makes to this sector increases the absolute exposure to risk rises, provided that a part of the risk must be carried by the bank. Secondly, increasing arbitrage involves increasing 1/ Most significantly, a defaulting co-operative is not eligible for further loans under the NACS, a factor that has been recognised as significant in affecting access to, and utilisation of, the scheme. See, for instance, Central Bank of Ceylon (1972) esp. pp.11-13. 2/ Under profit maxdnmisation default serves merely to lower the rate of return to the bank. Provided the net return to the bank, accounting for expected default costs as well as other transactions costs, were greater than the rate of discount, the bank would find it worthwhile to arbitrage without limit. 23- asset specialisation so that the reduction in the variance of returns associated with asset diversification is denied. Given the shape of the utility function, the degree of this risk aversion, the bank will decide upon an optimal degree of rediscount. Finally, it must be emphasized that our analysis of the profitability of rediscounting makes an implicit assumption that the commercial bank is able to utilise the funds obtained from the rediscount to acquire additional earning assets, thereby increasing the total return on the banks' own resources. If the bank is unwilling or unable to make additional loans with the funds released by the rediscounting, there is no increase in the rate of return to the bank. Earlier we criticised the relevance of this unconstrained risk- return approach in the Sri Lanka context, and suggested an alternative. The banks exercise very rigorous eligibility requirements for granting loans -- there is little difference on this between banks and that there is little opportunity to soften these requirements by paying a higher in- terest rate -- so that the allocation of bank assets among broad portfolio classes is largely determined by the availability of "suitable" loan demand; once (effective) loan demand is satisfied and necessary reserves and vault cash maintained, the residue of the funds are placed in government securities. 1/ Only if the released funds are held idle, as excess reserve, is this valid. The bank can always increase its net rate of return by using the funds to buy treasury bills. However, the lower the rate of return on the use to which the released funds can be directed the lower the benefits from rediscounting. There is, in effect, a hierarchical approach to bank asset allocation, in sharp contrast to the simultaneous determination of intersectoral allocation of traditional risk-return bank theory. In such an,allocative framework the commercial bank would only wish to rediscount -- as contrasted to buying default loss insurance -- to obtain additional funds if there were loan applications of suitable quality, however strictly defined, in excess of available resources (after the loans had been made to agricul- ture). Since the banks wish to meet all applications which satisfy current eligibility criteria -- we could find no real indication of credit ration- ing by varying terms as a means of attempting short term adjustments in the loan market -- they tend to obtain funds to service such loan demands by 1/ recourse to the discount window." The critical factor in determining such recourse is the bank's conception of the alternative use of funds. Given the allocative processes of the banks the only alternative use which would direct, in practice, the banks to discount would be unsatisfied loan demand which met the current eligibility criteria. Traditional analysis which would examine the recourse to the discount window from the perspective of profitability raises to a position of commanding importance a dimension that, other evidence indicates, is not overwhelmingly important in the decision making process of the banks -- and fails to examine dimensions that are of central significance. The bank management policies operative in a particular Assured access to rediscount facilities and at such preferential rates may be a major factor accounting for the absence of liability management -- that is, resource mobilisation by actively manipulating the structure of liabilities -- by Sri Lanka banks. If an institution knows that it can rely upon obtaining funds without limit, provided it has NACS paper, and at very preferential rates, there can be little incentive to attempt to develop other sources of finance. context are of central importance in determining the working of a rediscount scheme- b. Medium and Long Term Find (MLTF) As originally enacted, the Monetary Law Act (1949) #82, restricted the Central Bank to the rediscount or purchase of commercial paper "with maturities of not more than 180 days from the date of their discount;" acquisition of credit instrmients arising from the finance of production was restricted to maturities "of not more than 270 days from the date of their discount." It will be observed that these provisions do not restrict the original maturity of the instruments eligible for discount so that commercial banks could make term loans; provided a balanced portfolio of such loans were maintained, so that some loans were due within six or nine months, the commercial bank would have assured access to the discount window. Nonetheless, in an effort to encourage a lengthening of the maturity of bank loans, these restrictions upon Central Bank operations were eliminated in 1963. In 1964 the Medium and Long Term Fund (MLTF) was established to allow the Central Bank to rediscount, a preferential terms, long term loans made by the banks. The Central Bank has, inter alia,the power to determine 1/ In our interviews we asked bankers -- largely at the Peoples Bank since the overwhelming bulk of NACS loans eligible for privileged rediscount are through this one bank -- the principles which determined their recourse to the discount window. However, the question was phrased the answer was that they rediscounted when they need to, to meet other "legitimate" demands. Questions oriented, whether directly or obliquely, towards the role of the rate of return invariably received negative responses. The risk of agricultural loans was very important and the poor repayment experience was evidently troublesome. But rediscount does not alter, positively, or negatively, their exposure to default: the Central guarantee scheme offers protection here. - 26 - the rate of interest to be charged by the commercial bank and the purposes for which the long term loans may be made to be eligible for preferential discount at the FUnd. The Fund is not a guarantee scheme, even in part, and the institution making the original loan is responsible to the Central Bank for repayment. Examination of changes in the maturity composition of bank advances is difficult since no data are collected_on the original maturity of loans; the available data refer to the time that remains for the loan, at the moment that the data were collected. Thus a loan originally issued for a 10 year maturity in June 1964, that is, a "long term" loan in the data, would be classified as a medium term loan in the 1970 data and as short term in 1974. Further, there may be an implicit under- standing that a de jure short term loan would be renewed, thus making it de facto of longer duration. Finally, access to the MLTF is available only for loans which were made for three or more years, so that the classifi- cation criterion for the Central Bank data on the maturity distribution of advances is not identical with that for the MLTF. Despite these difficulties, which make fine differences in the distribution difficult to interpret, the broad pattern of changes may be observed. 1/ Among the other terms that the Central Bank has specified for eligibility, from time to time, the following may be noted: with limited exceptions, refinance is available for loans granted only to Ceylonese individuals and institutions; refinance is available for a period of not less than three years and of not more than fifteen years "or the final date for repay- ment of the loan granted by the credit institution to its borrower"; the refinanced loan is repayable in instalments but the Central Bank linevertheless retains the right (which is expressly reserved to the Bank) to require repayment of all or any part of refinance at any time on demand'?; any request for refinance shall be made within six months of the date on which the loan was originally granted by the commercial bank; and that the application for refinance must be made before the original loan was dis- bursed by the commercial bank. See Central Bank of Ceylon (1968). - 27 - In absolute terms the quantity of loans granted f or the medium term (1-5 years) and the long term ($+ years) has increased since 1963. However, as a proportion of total bank credit outstanding medium and long term finance was approximately stable between 1963 and 1969, and has declined markedly since 1970, as may be noted from Table 3, columns h and 6. Further, the commercial banks made no use whatsoever of the preferential discount facilities at the MLTF up to 1966, and made only trivial use till 1968; there was a sharp increase in recourse to these facilities up to 1971 and there has been a marked decline since then. Nonetheless, relative to the total quantity of medium and long term credit outstanding the borrowing from the METF is trivially small. We find it difficult to account for the very marked decline in the provision of medium and long term finance after 1969. Admittedly, since 1969 there has been a small shift in the structure of deposits towards demand deposits but this has been so small and the relationship between the maturity profile of liabilities and that of assets so tenuous that it unlikely to afford an adequate explanation. Other structural features, such as the expansion of branch banking and the overwhelming dominance of the system by but two banks would serve to reduce the volatility of deposits and thus, ceteris paribus, permit a lengthening of maturities. No data are available on interest rates on bank loans of different maturities, but discussions at both the Central Bank and the commercial banks suggest that the structure of interest rates on bank loans is relatively invariant with respect to maturity of the loan, so that changes in the term structure are unlikel7 to be the cause of the observed shift towards short term advances. - 28 - TABLE 3 TERM LOAINS, AND T-lEIR REDIS'COUNTING, BY COMMERCIAL BANKS 1963 - 1975 Bank Advances Outstanding Total Debt Outstanding to Medium and Long Term End Medium· (1-5 yrs.) Long (5+ yrs) Fund by Commercial of Total . . (3)2) (5)/(2) Banks Year Rsm. Rsm.. % Rsm. (1) (2) (3) (4) (5) (6) (7) 1963 690.4 247.,9 35.9 175.9 25.5 0.0 1964 752.6 255.9 34.0 185.9 24.7 0.0 1965 762.9 273.5 35.9 202.2 26.5 0.0 1966 822.2 311.8 37.9 203.6 24.8 0.0 1967 976.1 359.6 36.8 232.4 23.8 0.2 1968 1,268.0 435.8- 34.4 344.1 27.1 2.3 1969 1,513.4 566.2 37.4 425.5 28.1 24.0 1970 1;544.1 '444.0 28.8 190.5 12.3 29.6 1971 1,763.2 526.2 29.3 168.3 9.5 34.4 1972 2,168.0 530.7 24.5 153.3 :7.1 32.4 1973 2,240.2 575.5 25.7 151.5 6.8 30.7 1974 3,299.0 644.4 19.6 191.4 5.8 24.2 a) 1975 3,512.9 676.9 19.3 216.2 6.1 21.1 Note: a) As of 30 November Source: Central Bank, Annual ReDorts, and registers. - 29 - However, since the latter part of the.1960s there has been a steady tendency for prices to rise -- in contrast to the earlier period when the price level was volatile, but there was no secular tendency for prices to rise. If the term structure of nominal interest rates does not show a commensurate rise to reflect changing inflationary expectations the term structure of real interest rates flattens and even reverses, thus making long term loans unattractive to the lender. Farther, where there is con- siderable uncertainty about the pace of expected inflation, the lender is able to reduce his exposure to such risk by lending short. Even though we have emphasised ths safety consciousness of the banks, the shift in the structure of loans since 1969, and especially between 1960 and 1970, is so sharp that it appears unlikely it can be attributed entirely to risk aversion by the banks in the face of uncertain inflationary trends. In our interviews, commercial bankers repeatedly asserted that there was a decline in the demand for longer term accomodation; and though they were not aware of - the potentially greater erosion of long term interest rates by rises in prices they did not direct major attention to this factor. Since there is no indication that qualitative terms for loans were drastically tightened during this variod we are not able to account for this decline, or even to determine whether or not such an alleged decline in demand was genuine. Direct comparisons between the borrowings from the MLTF and the total of medium and long term bank advances are not possible since the definition of medium term in the bank data is not the same as that used for access to the Fund. However, both in absolute or in relative terms, total outstandings to the Fund are small and, indeed, declining. Whether the purpose of the Fbnd was to encourage the banks to divert an -30- increasing share of their own resources to term finance, or whether its purpose was to conduit Central Bank funds through the commercial banking system, does not significantly affect our negative judgement on this scheme. The MLTF provides for the rediscount of eligible paper at preferential rates so that, provided the interest rate on the loan exceeds the rediscount rate by more t'a the transactions costs, it is profitable for the commercial bank to make such loans and to rediscount them. There is no incentive to the bank to divert its own resources to term finance, for only by becoming a conduit for official funds will it raise its return and thereby profit from the scheme. By its very design the scheme does nothing to encourage the banks to make term loans and to retain them in its portfolio. However, the evidence is unmistakable that the scheme failed even as a conduit. This failure can be attributed directly and most significantly to the requirement that "without prejudice to the security given by the credit institution to the (Central) Bank, the credit institu- tion will be and at all times will remain liable to the Bank as its debtor for the due repayment of refinance granted to it by the Bank. If the commercial banks had been motivated by simple profit max- imisation they would have made substantial use of the special facilities provided by the MLTF, since such utilisation would have served to increase 1/ Central Bank of Ceylon (1968), Sec. (2) (F). -31 - their profits; indeed, theoretically, they would have an incentive to re- disccunt their entire portfolio of term finance loans, provided they were able to obtain a positive rate of return with the funds so released. The increased risk would not enter their decision processes. Even if the objective function were accurately depicted by the conventional risk- return function we would reasonably expect the banks to be willing to bear the additional risk, at least up to a point, since the net rate of return to the bank would increase markedly with the proportion of its term finance portfolio that it rediscounted. However, neither of these objective functions adequately describe the behavior of the Sri Lanka banks. We have argued that these institu- tions, in the determination of their loan policies evidence so severe an aversion to risk that they take the form of excluding loans that fail to meet sharply defined eligibility criteria-- there is little, if any, willingness to bear a higher risk in an effort to obtain a higher return. The rediscount facilities at the Medium and Long Term Fund were so devised as to increase the rate of return to the banks, while leaving the entire burden of default risk upon them. It failed, accordingly to focus on the dimension that was central in the banks' decision process. It is not surprising therefore, that the commercial banks failed to respond. 1/ It may be noted, parenthetically, that the traditional measure of risk aversion, the ratio of the second derivative of the utility curve to the first derivative, where utility is a function of wealth, is entirely irrelevant in such a situation. - 32 - Another aspect of the operating conditions of the scheme would increase the risk to the discounting bank. For all rediscounts made under the provisions of the MLTF the Central Bank "retains the right (which is expressly- reserved to the Bank) to require repayment of all or part of the refinance at any time on demand". Presumably this provision was instituted to protect the Central Bank from unforeseen abuse of the privileged discount scheme. However justified it may be on other grounds, this provision subjects the discounting bank to the risk that it may be called to repay its loan on demand, thus reducing further the incentive to use the discount facility. - 33 - II It is self evident that a clear conception of the objectives of a policy is necessary both for its effective formulation and for an evaluation of the degree of its success or failure. Unfortunately no clear statement of the objectives of either the New Agricultural Credit Scheme or of the Medium and Long Term Fund is available, or can be imputed from interviews. Was the objective to channel funds towards (non-planta- tion) agriculture and the term finance of commerce and industry, regardless of whether these funds were derived from official sources or from the banks' own resources? That is, would the bpnking system acting as a conduit for official funds be considered acceptable? Or was the central objective of the policy to influence bank lending policies in such a fashion that the banks' own resources were utilised for favoured uses? On either count, the policy to increase the term financing activities by the commercial banks must be judged a failure. The camercial banks did not restructure their portfolios towards term finance, n,o r did they operate efficiently as conduits for Central Bank funds. The design of the policy was fundamentally deficient in so far as it failed, to focus on the central determinant of bank lending policy. Rediscounting ,,under the MTLF is with recourse so that the commercial bank is responsible for the entire default loss, if any, experienced on the loan. In Sri Lanka risk operates primarily by excluding classes of loans and there is little if any willingness on the part of the banks to trade additional risk exposure for extra income. The failure to recognise the centrality of risk avoidance, together with the recourse provision effectively doomed efforts to increase the maturity structure of loans. Other policy attributes, notably the right to compel repayment upon demand of any funds obtained from the MLTF, evidence a similar insensitivity to the role of risik in the decision process_ and serve to reinforce the failure of the policy. The record of the NACS, as a selective credit control, is more mixed. Under this scheme substantial funds were channelled to the agri- cultural sector, the commercial banks acting largely as conduits for official funds. The risks in lending for agricultural production are also high, probably higher than those attending term finance. But the NAGS, as a central feature of its operation, incorporates a credit guarantee scheme whereby the Central Bank, for a small fee, covers up to 75% of the amount of the loan in default. This provision has been widely used by the banks. Defaults on loans have been high bit the commercial banks have been compensated for (the bulk of) their losses and they continue to grant credit. (Moral suasion is also of importance but is not the sole factor.) In this case the success of the NAGS, even as a conduit for official funds, can be directly attributed to a recognition of the role of risk in the portfolio decisions of the banks. This is in sharp contrast to the Medium and Long Term Abnd which assumed, implicitly if not explicitly, either that risk played no role in the decision processes of the banks or that lengthening the term did not affect the risk of the loan. 1/ Whether or not objective risk alters with ten to maturity is unimportant. What determines behaviour is the subjective, or perception of, risk. A scheme which permits the rediscount of a particular type of asset at privileged discount rates increases the profitability of that asset provided it is discounted; it cannot alter the profitability of that asset if it is not discounted but is retained in the portfolio of the bank. Thus, of itself a pure scheme of rediscounting at preferential rates can only be effective, if at all, to convert the banks into conduits for of- ficial funds. It does not alter the parameters of the configuration of the risk and rate of return of the asset unless it is rediscounted, so that it cannot operate to alter the use of the banks' own funds, By its very nature it is inappropriate if the objective is to reorient the banks's own resources towards favoured sectors. A pure credit guarantee scheme, withoug being tied to a preferential rediscount scheme, would reduce the risk exposure of the bank, and thus make the asset more attractive in a portfolio for an institution where risk entered the the objective function. This would not increase the holding of the privileged asset in a case of perfect profit maxinisation; but under the alternatives, of risk-return balance or of the ovezwhelming risk avoidance manifest in Sri Lanka, it would be effective. The (relatively) successful results of the NACS, whether as a conduit for official funds or even that small part which constitutes a diversion of the commercial banks' own resources, can safely be attributed not to the rediscounting mechanism per se but rather to the guarantee component of the scheme. It is particularly important to recognise this since a guarantee scheme can be introduced independently of a preferential rate rediscount scheme, and vice versa. ------ . . - 36 - Although by its very conception the pure rediscount scheme can operate, if at all, only to encourage the conduit of official funds through the banks, it is possible to adapt it to encourage the redirection of the bank's own funds into favoured sectors. For instance, such schemes customarily allow the rediscounting of favoured assets at a preferential discount rate, thus encouraging rediscount. However, if favoured assets were guaranteed access to the discount window (but not at a privileged rate) their liquidity and thus their attractiveness as a portfolio asset would rise to the bank. Directly or covertly, most central banks engage in credit rationing at the discount window so that the assurance that certain assets would always be acceptable for discount without limit would augment the attractiveness of such an asset to the bank and thus encourage the diver- sion of bank funds to its acquisition. Of course, such a policy would only be successful if the prospects of credit rationing by the Central Bank were real and the banks feared the consequences thereof. In the Sri Lanka context, with the dominant banks publicly owned, we do not believe such a fear is real. 1/ It may be noted parenthetically that the operating conditions of the MLTF provide a case study of how not to encourage the banks to utilise their own funds for the favoured use. There is a requirement that an "appli- cation for refinance be made to the Bank before the expiry of six months from which the loan was granted by the credit institution." If there were any possibility that the funds might be needed after six months the bank is not encouraged to hold the loan in its own portfolio and only rediscount it when the need arises; it must rediscount immediately, or suffer the risk of holding an illiquid asset, thereby augmenting the conduit nature of the scheme. - 37 - Alternatively, by restricting the proportion of holdings of assets of favoured sectors which can be rediscounted at preferential 1/ rates- it would be possible, in some sitiiations, to-encourage the banks to increase the allocation of their own funds to favoured uses. The rate of return on the bank's own funds depend directly upon the proportion of the loans discounted, provided the interest rate on the loan exceeds the discount rate charged by more than the transactions costs. Thus, abstracting from risk, and assuming that profitable opportunities for released funds are available, it has an incentive to rediscount all of its loans to the favoured sector. However, if it were restricted to the pro- portion that it could discount, it would still have an increase in its net rate of return -- while at the same time, retaining some of those loans in its portfolio. If it wished to increase its total income by increasing its rediscounting it would, of necessity, have to increase the proportion of its own funds that it was prepared to allocate to the favoured sector. (In this context the policy maker's task is to determine the maximum level of the proportion of the assets which can be discounted, together with the differential in rates. The proportion must be high enough, given the rate differentials and the costs of operation, to yield an attractive rate of return to encourage the bank to acquire the assets; it should be low enough, however, to restrict the conduit operation and to encourage the maximum diversion of banks' own funds to such favoured uses.) Such a scheme 1/ It is not necessary to prohibit the rediscount of the residue. Indeed, in an effort not to increase riskiness (by illiquidiffy)~tE residue should be available for rediscounting, but under the customary pro- visions and rates. A - 38 - of course, would only be effective in a world of profit maximisation, or at least of a trade off between risk and return; in the Sri Lanka context, of portfolio allocation on the basis of strictly enforced eligibility criteria, this is unlikely to be successful. It is not the purpose of this essay to devise "optimal" schemes for the restructuring of credit flows. Indeed, we doubt that it is pos- sible to devise schemes which will be equally successful in all contexts. It has been the burden of our argument that bank behaviour in the deter- mination of its portfolio is the central variable in determining whether or not a particular selective credit control policy can be devised, except by chance, independent of the institutional constraints within which it is expected to operate. We suspect that the behavioural assumptions implicit in the NACS and the MLTF were imported from those of overseas banking systems, particularly those of Britain and the United States. However, appropriate these assumptions may be in the context of their origin, it is clear that they ignore central dimensions of the decision processes in Sri Lanka -- and, probably, in many other developing countries. The crucial importance of examining the objective functions of the institutions to be influenced, and of devising schemes consistent with these functions, cannot be overstressed. LIST OF WORKS CITED Balasuriya, C. Anton (1972): Industrialisation and the Problem of Access to Finance of Small and Medium Sized Firms in Ceylon. Unpublished Ph.D. thesis, Faculty of Economics, UAiversity of Edinburgh. Bank of Ceylon Commission (1968): Report. Sessional Paper No. XXVII, 1968. Colombo, Sri Lanka. Baumol, William J. (1967): Business Behavior, Value and Growth. Revised edition; New York Central Bank of Ceylon (1968): "Operating Instructions, No. BC20/68", dated 4 September, 1968. Edwards, F.R.(1977):"Managerial Objectives in Regulated Industries: Expense Preference Behavior in Banking", Journal of Political Economy (February), pp. 147-162. and A.A.Heggestad (1973):"Uncertainty, Market Structure and Performance: The Galbraith-Caves Hypothesis and Managerial Motives in Banking" Quarterly Journal of Economics (August), pp. 455-473. Ghandhi, J.K.S. (1976): Commercial Bank Behaviour and Selective Credit Controls: A Sri Lanka Case Study. Preliminary Draft Report, Pub- lic and Private Finance Division, Development Economics Department, The World Bank, Washington, D.C. Hester, Donald, and James Pearce (1975): 1ank Management and Portfolio Behavior. New Haven, Conn. Hinderliter, R.H. and H. Rockoff (1976): "Banking Under the Gold Standard: Analysis of Liquidity Management in the Leading Financial Centers", Journal of Economic History (June), pp. 379-396 Hodgman, D.R.(1963): Commercial Bank Loan and Investment Policy. Urbana, Ill. Jayawardena, N.U. (1965): Some Aspects of Financial and Economic Policy, Pt. II: Banking and Finance and their Role in the Private Sector. Unpublished report, Colombo, Sri Lanka. Keynes, J.M. (1913): Indian Currency and Finance. London. Markowitz, H. (1952): "Portfolio Selection", Journal of Finance (March), pp. 77-91. (1959): Portfolio Selection. Cowles Commission Monograph # 16, New York. Penrose, E.R. (1959): Theory of the Growth of the Firm. Oxford. -2- Rao, D.C. and Ira Kaminow (1973): "Selective Credit Controls and the Real Investment Mix: A general Equilibrium Approach", Journal of Finance, pp. 1103-1118. Robinson, R.I. (1962): The Management of Bank Funds. New York, N.Y. Silber, William L. (1973): "Selective Credit Policies: A Survey", Quarterly Review of the Banca Nazionale del Lavoro, pp. 328-353. Tobin, James (1958): "Liquidity Preference as Behaviour Towards Risk", Review of Economic Studies (February), pp. 65-86. Williamson, Oliver (1963): "A Model of Rational Managerial Behavior" in R.M.Cyert and James March, A Behavioral Theory of the Firm. Englewood Cliffs, N.J. (1964): The Economics of Discretionary Behavior. Englewood Cliffs, N.J.