SV\/P672 Evaluation of Financial Policy Credit Allocation in Bangladesh Arvind Virmani WORLD BANK STAFF WORKING PAPERS Number 672 WORLD BANK STAFF WORKING PAPERS Number 672 Evaluation of Financial Policy Credit Allocation in Bangladesh Arvind Virmani The World Bank Washington, D.C., U.S.A. Copyright (© 1984 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing October 1984 This is a working document published informally by the World Bank. To present the results of research with the least possible delay, the typescript has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. The publication is supplied at a token charge to defray part of the cost of manufacture and distribution. The World Bank does not accept responsibility for the views expressed herein, which are those of the author and should not be attributed to the World Bank or to its affiliated organizations. The findings, interpretations, and conclusions are the results of research supported by the Bank; they do not necessarily represent official policy of the Bank. The designations employed, the presentation of material, and any maps used in this document are solely for the convenience of the reader and do not imply the expression of any opinion whatsoever on the part of the World Bank or its affiliates concerning the legal status of any country, territory, city, area, or of its authorities, or concerning the delimitation of its boundaries, or national affiliation. The full range of World Bank publications, both free and for sale, is described in the Catalog of Publications; the continuing research program is outlined in Abstracts of Current Studies. Both booklets are updated annually; the most recent edition of each is available without charge from the Publications Sales Unit, Department T, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A., or from the European Office of the Bank, 66 avenue d'I6na, 75116 Paris, France. Arvind Virmani is an economist in the Development Research Department of the World Bank. Library of Congress Cataloging in Publication Data Virmani, Arvind, 1949- Evaluation of financial policy. (World Bank staff working papers ; no. 672) 1. Credit control--Bangladesh. 2. Finance--Bangladesh. I. Title. II. Series. HG3711.B3V57 1984 332.7'09549'2 84-20985 ISBN 0-8213-0429-1 Abstract The Bangladesh Goverrnment has used a variety of policies to influence the allocation of credit to specific borrower groups like poor farmers, uses such as housing, and sectors such as agriculture. These policies have also been used tc reduce allocation for certain uses such as inventory holding ("speculative purposes"). The policies used have included interest controls credit guarantees, refinance and margine requirements. This paper attempts a systematic analysis of all the policies used by the Bangladesh government in the light of the explicit or implicit objectives of the government. The effect of each of the policies used, alone or in combination, is analysed with a view to evaluating their usefulness and efficiency. The analysis then provides the basis for suggested policy charges in Bangladesh, as well as lessons for financial policy makers in other countries. ABREGE Le Gouvernement du Bangladesh, souhaitant orienter les pratiques du credit bancaire de maniere a favoriser certaines categories d'emprunteurs tels que les agriculteurs defavorises et certains secteurs comme le logement et l'agriculture, ou, au contraire, a d6courager le detention de stocks a des fins speculatives, a mis en oeuvre tout un eventail de mesures. C'est ainsi qu'il agit notamment sur les taux d'int6r&t, les garanties de credit et les criteres regissant les refinancements et les marges ben6ficiaires. Le present document s'efforce de proc6der a une analyse systematique de toutes les mesures appliquees par le Gouvernement, a la lumiere de ses objectifs explicites ou implicities. II examine l'impact de chacune de ces politiques, utilisee isolement ou en association avec d'autres, afin de juger de son utilit6 et de son efficacit6. Sur la base de cette etude critique, il propose des modifications et tire un certain nombre de conclusions dont peuvent slinspirer les responsables de la politique financiere d'autres pays. Extracto El Gobierno de Bangladesh ha utilizado diversas politicas para influir en la asignaci6n del credito a grupos especificos de prestatarios, como los agricultores pobres, iy a ciertos usos y sectores, como la vivienda y la agricultura. Estas polit-icas se han utilizado ademas para reducir la asignaci6n del cr6dito a deterrninados usos, como el mantenimiento de exis- tencias ("fines especulativos"). Entre las politicas aplicadas cabe citar el control de los intereses, garantias de creditos, refinanciamiento y exi- gencias de margenes. En este trabajo se intenta hacer un analisis sistema- tico de todas las politicas ap]Licadas por el Gobierno de Bangladesh a la luz de los objetivos explicitos o implicitos del mismo. Se analiza el efecto de cada una de las politicas util:izadas, por si sola o en combinaci6n con otras, con vistas a evaluar su utilidad y eficacia. El analisis proporciona luego la base para sugerir cambios de politicas en Bangladesh, asi como para extraer ensenianzas de posible utilidad para los responsables de elaborar las politicas financieras en otros paises. CURRENCY EQUIVALENTS The Bangladesh Taka (abreviated TK) is fixed in relation to a basket of reference currencies, with the Pound Sterling serving as intervention currency. On January 11, 1982, the exchange rate was set at TK 38.372 buying and TK 38.472 selling per Pound Sterling. Depending on exchange rate movements between Sterling and the US Dollar, the Taka/Dollar cross-rate is subject to change. In recent months, this rate has fluctuated between TK 19 and TK 20.9 per US$. MEASURES 1 Crore = 10 million 1 Lakh = 100 thousand FISCAL YEAR The Bangladesh Fiscal Year (FY) runs from July 1 to June 30. Chapter Table of Contents Page I., Introduction 1 1.1 Objectives - Explicit 1 1.2 Objectives - Implicit 1 1.3 Types of Policies 2 1.4 Overall Allocation 3 II. Interest Rate Policy 8 2.1 First Major Revision 8 2.2 Second Major Revision 8 2.3 Effect of Credit Ceilings 10 2.4 Guarantees and Collateral 11 2.5 Distributional Effects 15 2.6 Speculative Credit 16 2.7 Interest Ceilings and Interest Subsidy 16 2.8 Penalty Interest Rates 17 2.9 Objectives and EfEfects of Second Revision 19 2.10 Intra-Bank Lending Rates 22 2.11 Real and Nominal Interest Rates 23 III. Credit Allocation Polic 24 3.1 Overview 24 3.2 The Target Approach 25 3.3 Ceilings and Prohibitions 30 3.4 Margin Requirements: Overview and Analysis 32 3.5 Margin Requirements: Differential Impact 37 3.6 Refinance and Counter Finance: Overview 47 Table of Contents (continued) Page 3.7 Refinance Usage, Intentive Effects and Usefulness 52 3.8 Guarantee Programs 62 3.9 Urban-Rural Branch Licensing 64 IV. Issues of Financial Policy 69 4.1 Efficiency or Inefficiency in Credit Markets 69 4.2 Welfare Objectives and Credit Markets 70 4.3 Differential Interest Rates: Availability vs. Cost 72 4.4 Financing of Specialized Banks: Refinance, Debentures 73 or Deposits 4.5 Buffer Stocks: Alternative for Margin Requirements 79 and Refinance 4.6 Term Finance 80 V. Agricultural Credit 81 5.1 Introduction 81 5.2 Expected Marginal Product of Loans 82 5.3 Targets and Disbursements; Differential Performance 85 5.4 The Special Agricultural Program and Commercial 88 Bank Performance 5.5 Introduction of Urban Organized Banks into 91 Agricultural Lending 5.6 Interest Ceilings and BKB Performance 92 5.7 Recovery Performance 94 5.8 Bad Debts vs. Liquidity Problems 101 Footnotes 103 References 103 List of Tables Table Number Page 1. Sectoral Distr.Lbution of Advances by 4 Scheduled Banks 2. GDP in Current Prices by Sectors 6 3. Credit Allocation Index 7 4. Interest Rates on Loans and Advances 9 5. Advances Classified by Size of Accounts 12 6. Advances Classified by Securities 13 7. Loans to Agricuiltural Hunting, Forestry, 14 Fishing 8. Advances Classified by Rates of Interest 18 9. Advances to Small Loans and Small Industry, 27 House Building and Transport 10. Rates of Growth of Prices and Imposition 33 of Margin Requirements 11. Distribution of Advances by Margins for 38 Specified Securities, and Average Loan Size by Margin for the same Securities 12. Distribution of Advances by Rates of Margin 40 for Related Securities 13. Distribution of Advances by Rates of Margin 41 14. Distribution of Advances by Securities 44 15. Distribution of Advances by Economic Purpose 46 16. Use of Refinancie by Bangladesh Krishi Bank 48 (BKB) and Bangladesh Samabya Bank (BSB) 17. Average (weighted) Deposit Rate 53 18. Refinance: Usage as a Percentage of 57 Entitled Limits 19. Advances and Deposits of Scheduled Banks 59 excluding BKB and BSB List of Tables (continued) Page 20. Payment of Government Guarantees to the 63 Banks Under Special Agricultural Credit Program 21. Rural Branch Expansion 65 22. Administrative Costs of Rural Branches 67 23. Sonali Bank Statistics 68 24. Credit-Deposit Ratio by District 78 25. Marginal Product of Loans in Agriculture 83 26. Agricultural Credit 86 27. Special Agricultural Credit Program 89 28. Bangladesh Krishi Bank Borrower Characteristics 93 List of Figures 1. Rationing Equilibrium 36 2. Recovery Profiles by Banks for the Special 96 Agricultural Credit Program 3. Recovery Profiles: BKB, Sonali, Agrani and 97 Janata Banks 4. Recovery Profiles: Janata, Uttara, Rupali 98 and Pubali Banks 5. Recovery Profiles by SACP Programs 100 DEFINITIONS AND ABREVIATIONS Bangladesh Bank (BB): The Central Bank of Bangladesh. Co-operative Banks: Consist of an apex co-operative bank and 62 central co-operative banks. Bangladesh Samabaya Bank Limited (BSBL): The apex co-operative bank; sometimes referred to as the Bangladesh Jatiya Bank (BJSB). Scheduled Banks: Banks permitted to carry out all normal banking operations including the taking of deposits from the public. They are therefore subject to reserve requirements. Nationalized Commercial Banks (NCBs): The six commercial banks (Sonali, Janta, Agrani, Rupali, Pubali, Uttara) owned by the government. Specialized (Scheduled) Banks: Special Purpose Banks which are allowed to take deposits. Consist of: Bangladesh Krishi Bank (BKB): The Agricultural development bank; Bangladesh Shilpa Bank (BSB): Specialized bank for making medium and long term loans to industry. Foreign Commercial Banks': 'Branches of foreign owned banks. Bangladesh Shilpa Rin Sangst'ha (BSRS): Specialized Institution for making long and medium-term loans to heavy-medium industries. Not a sc'heduled bank. House Building Finance Corporation (HBFC): Specialized Institution for making loans for Housing. Integrated Rural Development Program (IRDP): A two-tier co- operative system separate from and independent of the co- operative banks. Financed through the Sonali Bank. PREFACE This paper applies thet analytical framework developed earlier by the author in The Nature of Credit Markets in Developing Countries: A Framework for Policy Analysis, (World Bank Staff Working Paper No.524). The paper was prepared originally as a report for the Bangladesh Financial Sector Review Mission. Many of the issues before this mission and its objectives were similar to those which motivated the author's earlier work. The current paper was therefore conceived from inception as a practical application of this policy framework. A distinction between empirical testing and practical application needs to be borne in mind in reading the paper. The former was not attempted, given the data limitations and time constraints under which the paper was written. It is presented more as a practical guide to others wishing to apply the siame framework in the context of similar limitations. Arvind Virmani I. INTRODUCTION 1.1 Objectives - Explicit The Bangladesh government has outlined three operational objectives of financial policy. These are-I/ (a) Macro-Economic Objective: To control the expansion of bank credit so as to contain inflation while stimulating production and investment. As with most governments no explicit trade-off is considered and the relative emphasis on one or other sub-objective has varied with economic and social conditions. (b) Credit Allocation Objective: To channel credit into priority and essential sectors and away from speculative activities and hoarding. Agriculture, small industry and exports were the earliest sectors to be defined as priority sectors. The emphasis on agriculture as a priority/essential sector has remained throughout. The emphasis given to other sectors has varied however. Small industry expanded to include small businesses and trade, new sectors like residential housing and Transport were added and then detleted from the list. (c) Institutional Objective: To consolidate and strengthen the banking system, to provide new financial institutions and to promote money markets. 1.2 Objectives - Implicit Underlying these explicit or apparent objectives of financial policy are deeper or implicit objectives which guide their application. This is particularly true of the credit allocation and institutional objectives. The implicit objectives are: - 2 - (a) Efficiency Objective: To correct imperfections in the credit market which lead to economically or socially inefficient use of credit. This clearly influences the choice of sectors and borrower categories for both positive and negative credit allocation measures. It also influences the institutional objectives. (b) Welfare Objective: To provide welfare subsidies to target groups, to transfer income between different groups of borrowers, and to change the division of gains in transactions between traditional lenders and borrowers. (c) Subsidy Objective: To correct imperfections or impediments in other markets, or to mitigate these effects, by providing subsidies through the credit system. 1.3 Types of Policies The most important financial policies used to meet the overall objectives can be divided broadly into three groups. This is done on the basis of the sub-set of the objectives they are primarily directed towards. They can and often do have effects on other objectives, but it is useful as a first step to make the categorisation, as follows: (a) Credit Ceilings and Prohibitions: These were directed primarily at controlling the flow of total credit, but often have an allocation component. They include ceilings on total lending by individual banks, lending to private and public sectors, lending to different loan size classes, and lending for certain economic purposes. The allocation component is designed primarily to countervene the likely allocational impact of overall credit control policies. -3 - (b) Allocational Instruments- These included lending targets and guidelines, margin requirements, refinance policies, government guarantees, and linking to new urban and rural branches. Specialized banks when coupled with refinance policy have been another instrument for this purpose. Further the effects of refinance policy have been most strongly influenced by the credit ceiling instrument. (c) Int:erest Rate Policy: Interest rate policies, including the use of penalty rates, of course have the broadest ramifications. Ceillgns on (loan) interest rates have been directed primarily towards the welfare and subsidy objectives. A secondary objective has beenm the stimulation of investment. Interest rate policies of course influence the allocation of credit, and the viability of financial institutions. The use of penalty rates of interest on loan overdues, has been used as an indirect means of couiiteracti g some of these secondary effects. 1.4 Overall Allocation The present paper focuses on the post 1975 period, and on the issue of credit allocation. Clearly however issues can seldom be so neatly compartmentalized. To the extent that Macro-economi and institutional issues interact with credit allocation Issues they will be dealt with. Table 1 gives an overview of the allocation of credit by broad economic sectors. Over the period from June 1975 to December 1975 the proportion of total advances to the agricultural sector rose from 10.8% to 18.6%, while those to manufacturin;v declined from 52.4% to 40.0%. The proportion of advances going to the transport eector also showed a declining trend while those to the construction sector in reased. The proportion of advances going to wholesale and retail trade fluctuated considerably over the Table 1: Sectoral Distribution of Advances by Scheduled Banks (Percentage of Total) Sectors 30-6-75 30-6-76 30-6-77 30-6-78 30-6-79 30-6- 80 Agriculture, Forestry, Fishing 10.76 10.78 11.31 14.13 16.65 18.59 Manufacturing 52.38 48.73 43.81 42.76 40.12 40.02 Construction .76 .69 1.14 1.37 2.67 1.43 Transport and Communications 3.56 4.89 3.93 3.86 3.62 3.17 Transport 4.31 4.87 3.89 3.82 3.50 3.14 Trade and Catering 23.59 27.50 27.64 22.94 26.18 25.82 Trade-Wholesale and retail 23.09 26.65 27.23 22.71 25.98 25.60 Jute goods and Tea + 11.80 6.43 6.29 9.48 9.32 + Data not available. Source: Calculated from Bangladesh Bank Bulletin. - 5 - period. These figures suggest a redirection of credit away from transport and communication and manufacturing, into construction and agriculture, forestry and fishing. Table 2 shows however that the share of the different sectors in total GDP varied considerably over time. Any shifts in the pattern of credit allocation between sectors would therefore be partly due to changes in GDP share and partly to government policy and other factors. It is therefore useful to construct an index which removes this effect, and is presented in Table 3. The doubling of the index for agriculture over this period confirms the profound shift of institutiona)l credit towards the agricultural sector. The index for the transport and communication sector also continues to show a down trend over the whole period. None of the other sectors now show a clear trend over the entire period, with the construction sector replacing the trade and catering sector as the most erratic. The only other information of note in Table 3 is the kink which occurs in the index for Trade and Catering in June 1978. Trade and Catering follow the inverse sawtooth pattern with the rising trend interrupted by a full in 1977-78. It is likely that a general reduction in loan interest ceilings in May 1977 is responsible for this effect which occured between .June 1977 and June 1978. Table 2: GDP in Current Prices, by Sectors (Percentage) GDP in Market Prices 1974-75 1975-76 1977-76 1977-78 1978-79 1979-80 Agriculture, Forestry, Fishing 62.53 53.36 50.94 55.45 54.39 54.10 Manufacturing 6.66 7.61 8.22 7.22 7.12 7.25 Construction 4.51 5.13 5.51 4.72 4.25 5.39 Transport and Communications 4.49 6.17 6.96 6.77 6.59 6.68 Trade & Catering 8.73 10.00 9.86 9.85 10.80 11.05 Soulrce: "Bangladesh: Recent Economic Dlevelopments and Selected Development Issues,"' March 3, 1982. (This is an internal document with restricted circulation.) Table 3: Credit Allocation Index; Ratio of Advance Proportions to GD? Proportions 1974-75 1975-76 1976-77 1977-78 1978-79 1979-80 Sector Agriculture .172 .201 .222 .255 .306 .344 Manufacturing 7.865 6.403 5.330 5.922 5.635 5.520 Construction .168 .135 .207 .290 .628 .265 Transportation .793 .793 .565 .570 .549 .475 Trade 2.702 2.750 2.803 2,329 2.424 2.337 Source: Tables 1 and 2. II. INTEREST RATE POLICY 2.1 First Major Revision Table 4 gives an overview of changes in interest rate ceilings on loans and advances. Over the period there have been two major changes in loan interest ceilings. The one in May 1977 reduced the nominal ceilings on almost all loans by one percentage point; from 13% to 12% on general borrowing and from 11.5 to 10.5 on exports. The exceptions to this were the rates on Jute, Jute goods and Tea and rates charged by the specialised banks and co- operatives, which remained essentially unchanged. The effect of this way to reduce the gap between the interest charged by these institutions and by others, with the exception of the Bangladesh Samabaya Bank (BSB). Interestingly BSB which makes medium to long-term loans retained a higher ceiling rate of 13%. The professed objective of this reduction in ceilings was to provide a stimulus to economic activity, particularly to commerce and trade. Credit ceilings had been in effect since the IMF borrowing in 1974-74. These had been raised periodically. In particular they had been raised some months before the lowering of the interest rate ceilings. Credit ceilings were therefore not on effective constraint in the succeeding period. In 1975-76 and 1976-77 the implicit GDP deflator declined by 23.9% and 3.2% respectively indicating a strong though decelerating deflation. In 1977-78 this process was reversed by an inflation rate of 14.9%. 2.2 Second Major Revision The other major change in interest rate policy occured in October 1980 (also Table 4). The effect of this set of changes was to differentiate loan interest ceilings by economic sectors. Thus intra-sectoral rates were unified while intersectoral ceilings were made different for agriculture and Table 4: Interest Rates on Loans and Advances (Scheduled Banks) Since Since Since July 1974 July 1976 May 1977 1978/79 October 1980 1981 General Borrowers 12.0-13.0% * 11.0-12.0% * 15.5% 16.0 (i.e. all those not separately specified) Export Commodities 12.0-13.0% 11.5% 10.5% * 12.0% * (except Jute, Jute Goods and Tea) Industry 12.0-13.0% * 11.0-12.0% * 14.0% 14.5% 1/ BSB General Lending 12.0-13.0% * 11.5-13.0% * 14.0% Agriculture 12.0-13.0% * 11.0-12.0% * 19n0% * Jute, Jute Goods and Tea 10.5% * * * 12.0% * BKB 11.0-11.5% * * * 12.0% * Co-Operatives 12.0% * * * 12.0% * Loans for Socio-Economic Purposes 12.0-13.0% 12.0% 2/ 11.0% 11.0-11.5% 13.0% * Bangladesh Bank Rate 8.0% 2/ * * * 10.5% (Discount Rate) Concessional Discount 6.0% 3/ * * * * Rate for lending to BKB and BSB X Unchanged from previous period. 1/ Except Long term loans which remain at 14%. 2/ Effective June 1974. 31 Effective September 1976. BKB: Bangladesh Krishi Bank BSB: Bangladesh Samabaya Bank 10 industry. The lowest ceiling was on agricultural advances at 12%, followed by industry at 14% and all others at 15.5%. One exception to this classification were loans for socia-economic objectives which had an overall interesst rate ceiling of 13%. The other apparent exception was export credit (pre-shipment, packing and post-shipment) at 12%. However interest ceilings on export commodities had always been relatively low. Jute, Jute goods and Tea were traditionally the most important export commodities and agricultural exports continue to constitute the majority of exports. Thus the rise in the ceiling rate on export commodities can be seen as consistent with intra-sectoral unification of ceilings. This revision in the interest rate structure was designed primarily as a resource mobilisation measure. It was hoped that the increase in the relative return to an asset competing with speculative stocking activities would shift assets out of the latter, and thus reduce inflationary expectations. However it was also expected to result in more efficient use of credit. Though in the case of both major changes in interest policy the primary objectives were macro-economic, the effects on credit allocations were equally if not more important. It is on these we focus. 2.3 Effect of Credit Ceilings In case of credit for productive purposes, I have shown earlier 2/ that an effective ceiling on loan interest rates would reduce loan amounts and increase collateral requirements. This effect would be stronger for thse who were just meeting credit requirements but now run into a collateral constraint. T1hose who are already constrained by collateral may get no loans at all. The funds thus made surplus reduce the opportunity cost of funds. This reduces the direct impact and also results in loans being provided for the first time to less productive borrowers having sufficient collateral. The - 11 - net effect is to shift credit from those having less collaterable wealth to those with more. In Tables 1 and 3 it was noted that there was no significant change in trends of credit allocation to productive (non-trade) sectors between June 1977 and June 1978. As there is no particular reason to expect the availability of effective collateral to differ significantly across sectors the facts are consistent with our analysis. Table 5 shows that between June 1977 and June 1978 there was a small but significant increase in the proportion of loans going to smaller borrowers. Thus the proportion of loans in the TK 1000 to TK 3000 category increased from 4.44 to 5.79% while proportion of total loans iln size less than TK 3000 increased from 8.82% to 10.31% (see also section 2,5). The previous analysis suggests that this may represent a shift of credit to somewhat less productive small producers who however had sufficient collateral available. This is consistent with the usual observation that small (indiustrial) producers have lower debt-equity ratio. 2.4 Guarantees and Collateral Table 6 gives a distribution of credit by securities. What is surprising in this table is the slharp decreases, from TK 143 crores on June 1977 to T1K 23 crores on June 1978, in the use of machinery and other fixed assets as security. As such assets are most prevalent in manufacturing this would suggest a decline in loans to the manufacturing sector. This conclusion is contradicted by the data in Tables 1 and 3, however. The correct explanation is found in the corresponding, and even sharper, increase from TK 252 crores to TK 509 crores, in use of miscellaneous securities over the same period. This category includes other secured advances, advances secured by guarantees and unsecured advances. In 1976-77 TK 56.5 crores of advances - 12 - Table 5: Advances Classified by Size of Accounts - All Banks (% of all Loans) Loan Size (Taka) 30-6-Z5 30-6-76 30-6-77 30-6-78 30-6-79 30-6-80 < 1,000 4.66 3.63 4.38 4.52 3.16 2.73 1,000 to < 3,000 3.73 3.53 4.44 5.79 6.97 7.64 3,000 to < 10,000 1.45 2.21 2.03 2.09 2.16 2.04 10,000 to < 25,000 1.67* 1.56* 2.50 2.34 2.73 2.36 25,000 to < 50,000 2.49* 2.91* 2.43 2.68 2.15 1.77 50,000 to < 100,000 3.16 3.40 4.78 3.98 4.12 3.13 100,000 to < 10,000,000 14.10 14.00 15.20 17.55 16.74 16.54 > 1,000,000 68.74 68.76 64.24 61.05 61.97 63.79 Cummulative % < 1,000 4.66 3.63 4.38 4.52 3.16 2.74 < 3,000 8.39 7.16 8.82 10.31 10.13 10.37 < 10,000 9.84 9.37 10.85 12.40 12.29 12.41 < 25,000 11.51* 10.93* 13.35 14.74 15.02 14.77 < 50,000 14.00 13.84 13.35 14.74 15.02 14.77 < 100,000 17.16 17.24 20.56 21.40 21.29 19.67 < 1,000,000 31.26 31.24 35.76 38.95 38.03 '36.21 Total 100.00 100.00 100.00 100.00 100.00 100.00 * 10,000 to 20,000 ** 20,000 to 50,000 + < 20,000 Source: Bangladesh Bank Bulletin pp. 24-25. Table 6: Advances Classified by Securities - All Banks 30-6-75 30-6-76 30-6-77 30-6-78 30-6-79 30-6-80 A. Gold etc. 0.47 0.67 0.95 0.95 0.92 B. Stock Exchange Sec. 0.96 1.08 1.02 0.23 1.31 0.68 C. Merchandise 57.86 58.03 53.98 52.90 60.60 62.63 i. Export Commodities 33.26 26.01 20.77 22.55 30.52 31.99 ii. Imported intermediaries 13.69 23.13 23.27 20.94 18.43 16.11 iii. Imported Ind. Machineries 10.41 8.40 9.77 8.54 11.34 14.16 iv. Imported Agric. Products nDe MAchiner- ar.d Other Fixed 14.87 1,.01 12.4, 1.57 2.10% 1.89 E. Real Estate 8.12 6.77 7.25 7.15 7.69 7.69 F. Financial Obligations 2.50 2.65 G. Miscellaneous 16.32 16.75 21.85 35.16 24.82 24.18 Source: Bangladesh Bank Bulletin, pp. 58-59. - 14 - Table 7: Loans to Agriculture, Hunting, Forestry and Fishing: Classified by Size of Accounts Size of Account (Taka) 30-6-76 30-6-77 30-6-78 30-9-78 30-6-79 30-9=80 - Loan amounts (1,000 Takas) < 10OO0 3361.10 4780.19 6083.06 6593.65 5298.45 6592.21 1,000 to < 3,000 2616.34 4268.14 7617.50 7027.70 11629.47 18675.55 3,0O0 to < 10,000 102.82 242.62 809.05 897.46 1648.02 2456.03 10,000 to < 20,000 63.93 40.46 316.29 469.36 1081.76 3096.75 20,000 to < 50,000 49.97 90.32 230.32 345.86 624.73 636.49 50,000 to < 100,000 76.95 216.11 143.01 340.34 395.77 455.01 > 1,000,000D 3838.93 3409.85 5852.91 6822.76 10542.71 18357.15 Total 10112.68 13047.69 21052.14 22497.13 31220.91 50269.19 Proportion of Total Loans (%) <( 1,000 33.24 36.64 28.90 29.31 16.97 13.11 1,000 to < 3,000 25.87 32.71 36.18 31.24 37.25 37.15 3,000 to < 10 000 1.07 1.86 3.84 3.99 5.28 4.89 10,000 to < 20,000 .63 .31 1.50 2.09 3.46 6.16 20,000 to < 50,000 .49 .69 1.09 1.54 2.00 1.27 50,000 to < 100,000 .76 1.66 .68 1.51 1.27 .91 > 100,000 37.96 26.13 27.80 30.33 33.77 36.52 Total 100.00 100.00 100.00 100.00 100.00 100.00 Cumulative Propotio~n(. < 1,000 33.24 36.64 28.90 29.31 16.97 13.11 < 3,000 59.11 69.35 65.08 60.55 54.22 50.26 < 10OO 60.18 71.21 68.92 64.54 59.50 55.15 < 2-OGO- 60.81 71.52 70.42 66.63 62.96 61.31 < 50,000 61.30 72.21 71.51 68.17 64.96 62.58 < 100,000 62.06 73.87 72.19 69.68 66.23 63.49 Total 100.00 100.00 100.00 100.00 100.00 100.00 Source: Bangladesh Bank Bulletins - 15 - to the public sector Jute industry were rescheduled. This was followed in 1977-78 by rescheduling and restructuring of the debt of the public sector Jute and Tea industry. This was done under the condition that the debt would now be backed by government guarantees. The change in the nature of securities therefore primarily represents a reclassification of this debt from fixed capital as collateral to government guarantees as collateral. 2.5 Distributional Effects The distributional effects of credit ceilings, arising from the unequal distribution of wealth and consequently of collateral among households are best demonstrated for the agricultural sector. In this sector there is a high co-relation between the distribution of land and the distribution of wealth. land is of course the best form of implicit or explicit collateral in agriculture. In addition, productivity and credit requirements would also be linked to the value of land. Therefore among the land owners we would expect small landowners to obtain smaller loans, while most of the landless creditors would fall in the smallest category. Table 7 gives the allocation of credit to the agricultural sector, by loan size. Between June 1.977 and June 1978 there was a sharp reduction in the proportion of loans of less than TK 1000 from 36.6% to 28.9%. Some of these were shifted into the next loan size category of TIC 1000 to TK 3000, which increased from 32.7% to 36.2%. However the table of cumulative percentages shows clearly the decrease in the proportion of total loans going to smaller loan categories. As our analysis indicated this would be the direct effect of a decrease in interest ceiling. A word of caution is in order, however: As the loan classes are in monetary terms, a proportional increase in the size of all loans, for example due to inflation, would shift the entire distribution to higher loan sizes. This would result in a - 16 - reduction of the proportion of loans in the lowest and a rise in that of the highest size classification. The proportions in the other loan size groups would remain approximately the same. This is not however the pattern in Table 7. Otherwise a Gini type coefficient is to be preferred. 2.6 Speculative Credit An analysis of pure speculative credit is somewhat different from that of productive credit as applied above. The speculator/trader is assumed to use his own funds as well as any borrowed funds for buying a commodity stock. The return to this stock arises purely from the anticipated increase in its price. In effect we separate out the productive aspects of trade from this analysis. The traders own contribution or margin constitutes the security or collateral as far as the bank is concerned. It can be shown that a reduction in credit ceilings in this case will tend to eliminate credit for the less speculative commodities: That is the commodities for which price rises are expected to be low may not obtain any credit. In tables 1 and 3 and the accompanying discussion it was pointed out that the reduction in credit ceilings in May 1977 led to a downward kink in the rising share of output-adjusted credit going to the trade sector. This is consistent with the above analysis. 2.7 Interest Ceilings and Interest Subsidy It is useful at this point to draw a distinction between interest ceilings and interest subsidies. Both may be motivated by a desire to provide subsidies, as for example the lower than average credit ceilings on Jute, Jute goods and Tea and on other export goods which prevailed before October 1980 (Table 4). However they can have quite different effects on the efficiency of credit use and on the distribution of credit. As already indicated interest ceilings have negative consequences in both these cases. On the other hand, - 17 - that when the banking system is inefficient in allocating credit to specific sub-sectors or sets of borrowers, an interest subsidy to the bank can remove this inefficiency [see Virmani (1982)]. In such a situation an interest subsidy increases loans to the neglected sector and to borrowers with inadequate collateral (wealth). However this policy has not been tried in Bangladesh. 2.8 Penalty Interest Rates In February 1979 the government introduced a system of penalty rates of interest on overdue loans. Scheduled banks were allowed under the system, to charge a penalty rate of interest, which could be incremented by 1% every month, until total rate (normal plus penalty) became 20%. The objective of this penalty was to decrease the disincentive for lending to sectors where repayment of loans has been a problem. Loan overdues can create a liquidity problem in addition to reducing the expected return to the bank. As borrowers who are delaying payments are likely to be the more risky customers it is logical to charge them a higher rate. The penalty rate does this without imposing a cost on good borrowers and is therefore an efficient mechanism for doing so. The application of penalty rates, particularly in agriculture, has been somewhat erratic, with differeni: banks having different rules. Thus Sonali bank charges the full penalty while BKB limits itself to a maximum penalty rate of 2.5% and Pubali to 11'. It is also reported that in other economic sectors, banks often charge penalty rates only to the public sector borrowers and seldom to the private sector. However this is quite consistent with the rationale given above, if the public sector is more erratic and deficient in its repayment performance than the corresponding private sector borrowers. Table 8: Advances Classif ied by Rates of Interest All Banks (Percent of total Loans) 30-6-76 30-6-77 30-6-78 30-6-79 30-6-80 Interest Rate 110,0%: 0.19 0.28 0.15 6.79 0.12 0.11 10.5% 26.41 23.31 16.34 18.93 25.25 19,86 11.0% 1.78 1.68 3.49 6.79 7.40 10.20 11.5- 7.28 5.85 5.87 5.49 8;38 7.26 12.O% 16.10 15.66 41.52 38,39 38.10 37.56 > 12.0 to < 13.0% 2.37 2.53 1,32 1.19 0.69 0,53 H 13.0% 33.38 40.57 17.46 9.04 5.01 3.89 > 13.0% 0 0 0 0.07 3.80 3.29 - 19 - Table 8, giving the disl:ribution of advances by rate of interest, shows that the proportion of loans carrying a rate of interest rate of more than 13% jumped from .07% of total advances in June 1978 to 3.8% in June 1979. They remained at approximately that level in June 1980. A small proportion of these of course constitute high interest loans under the Rural Finance Experimental Project. The rest are due to the use of penalty rates and give an approximate measure of the extent of usage of such rates. As late repayment is a major problem in the agricultural sector, penalty rates will be discussed further under agricultural credit. Such rates can be a very useful instrument of policy, when the repayment period is carefully defined to take account of natural agricultural fluctuations. 2.9 Objectives and Effects of the Second Revision Detailed data of the kind used above is not available for June , so the effects of the October 1980 revision of interest ceilings are not fully visible. However some preliminary conclusions can be drawn. As previously noted this revision involved a change-over to a sectorally oriented system of interest ceilings. This involved the setting up of an implicit hierarchy of sectors; agriculture and exports are on top with a ceiling rate of 12%, followed by manufacturing with a ceiling of 14%, while all other sectors have a ceiling rate of 15.5%. Small loans and loans in less developed areas, which do not fit into the sectoral classiLfication, fall between the ceiling rates for the agricultural and the manufacturing sectors, with a ceiling rate of 13%. There are two possible justifications for this hierarchy. One is based on the supposition that production and investment are sufficiently sensitive to interest rates, and that the economic rate of return has the hierarchical pattern postulated. T'hrough a plausible case can be made for the - 20 - second proposition, the first is highly dubious. However even if it has some merit it must be questioned whether the provision of implicit subsidies to creditors is the appropriate means of providing incentives. Thus it is necessary to look more closely at other policies, like exchange rates and tariffs, which directly affect exports and manufacturing. The other possible justification is to provide welfare subsidies to target groups. Thus the largest numbers of poor are thought to be in the rural areas, and lower interest rates to agricultural creditors may be thought of as providing a subsidy. However even if these justifications are accepted credit ceilings can have unexpected and harmful effects. To understand this it is very important to draw a distinction between interest ceilings and interest subsidies, and between cheaper credit and sufficient credit. Many policy makers and even economic analysts may see the differential interest rate ceilings as a cross-subsidisation scheme from general creditors to farmers, exporters, small loanees and industrial borrowers. This is quite misleading. A true cross subsidisation scheme would involve an interest tax (or a tax per loan) on all bank lending to general borrowers and an appropriate interest subsidy (or subsidy per loan) on the favoured borrowers. This would provide the correct incentives for channeling credit into the desired sectors. The effect of the differential credit ceilings will be to redirect credit away from agriculture and exports and to a lesser extent from industry into general lending. This perverse effect would be accompanied by redirection of credit within each sector to those with sufficient collateral and away from those with little collateral. Thus the poor would tend to get relatively less credit than they would in the absence of effective credit ceilings. Though adverse intersectoral flows may be partly stemmed by forced lending through specification of targets, this has - 21 - its own limitations. Among these are ineffectiveness against adverse distributional effects and the fungability problem (discussed in a later section). Two observed effects of the differential credit ceiling support the above analysis. Firstly these havre a differential impact among the Nationalized Commercial Banks (NCBs). A rise in deposit interest rates also raises the interest costs of all banks, and may have a differential impact across banks depending on the distribution of each bank's deposits. Profitability is more strongly affected, however, by the sectoral distribution of a bank-s existing loan portfolio. Thus the Pubali Bank which was strongly oriented towards small loans, exports and fisheries will now earn 12% or 13% on most of its loans. This has put a profit squeeze and provided a disincentive to expand further into these sectors. Thus, it is being forced to expand into areas of related expertise in industry and trade in which higher interest rates are allowed. A similar reduction in profits is taking place for the specialized banks such as the Bangladesh Krishi Bank (BKB). Though the rise in deposit rates affects only a small fraction of its funds, the relative profitability of agricultural lending vis a vis agriculture related commercial lending has declined. Given that in 1979 approximately 70% of its loans went to agriculture the lower interest rate of 12% on such loans puts a profit squeeze on the BKB (discussed further in agricultural section). As a result the BKB is planning to reduce the proportion of agricultural loans to 60% and to expand faster into sectors like agro-industries, storage, transport and banking to which higher credit ceilings are applicable. - 22 - 2.10 Intra-Bank Lending Rates The rates charged by banks on their internal transfer of funds between branches and between branches and headquarters are of course not directly relevant to final borrowers. They affect the incentive structure of banks, however, and can affect credit flows to final borrowers. For commercial banks, these rates vary from 11% to 14% at the present time, as shown below Bank Headquarters to Branch Branch to Headquarters Sonali 11% 11% Janata and Agrani 12% 12% Pubali 14% 13% Thus Pubali bank branches have clearly no incentive to borrow funds from headquarters for lending to agricultural and export sectors (12%) and for socioeconomic purposes (13%), and little incentives for lending to industry (14%). If the average administrative costs for all lending operations (1 to 2%) are taken account of, even Sonali bank branches have little or no incentive to borrow internally for lending to these sectors. Banks should of course be free to set intra-bank lending rates according to their particular circumstances. Given the current structure of interest rates, however, we can calculate the approximate range in which these rates should lie. The average interest cost of deposits is currently about 7% of total business. The average administrative costs of lending are about 4%, though the administrative costs of rural lending by commercial banks are somewhat higher at about 5%. As the lowest loan rate is 12%, and taking the 4% cost as divided equally between the deposit and lending sides intra-bank lending rates should be approximately 9 to 10%. - 23 - 2.11 Real and Nominal Interest Rates The anlysis presented in this chapter, as in the rest of the paper, is in nominal interest rate terms. Given the established practice of using a rate constructed by subtracting actual or expected inflation from the nominal rate (the 'real' rate) this requires an explanation. The framework on which the analysis of the paper is based, divides the basic financial market into two submarkets: The deposit marlket and the loan market. The former is supplied with funds by depositors and has banks as demanders of funds. In conventional terms this is the savings side of the market. When savings are treated in aggregate clearly a 'real' rate of interest must be used in the analysis. If deposit markets are perfect, changes in the rate of inflation have no effect on aggregate saving. If deposit rates are controlled by the government, however, changes in ithe rate of inflation will influence both the total amount of savings and the amount flowing into deposits. This in turn will affect the opportunity shadow cost of funds to the banks (termed s in the present paper). our analysis focuses on the loan sub-market. The shadow (opportunity) cost of funds (s) is of curse an input into this market. In the loan market, competitive banks will make a loan to a borrower as long as their expected profits are positive; the loan amount, the loan interest rate and other variables are all expressed in nominal terms. The firm similarly maximizes its expected profit evaluated in nominal terms. In deciding whether to take a loan or not it merely compares these net profits with the nominal opportunity cost of own funds. The terms of the loan contract, including the loan interest rate, will differ for firms with different characteristics (e.g. returns). The allocation of loans will therefore be determined by the distribution of these characteristics across firms. If we compare two - 24 - situations; one in which all prices are double that in the other, there is no reason to expect any difference in the distribution of these characteristics across firms. Therefore the allocation of credit should not be affected if other things remain unchanged. If the deposit interest rate is controlled, however, the flow of deposits into the banking system might fall. This in turn would raise the shadow cost of funds to the banks. Our analysis shows that in this case loan amounts must fall and loan interests rates rise for every existing borrower. In addition collateral requirements may rise or fall for each borrower, and this could affect the distribution of loans. Without more detailed information, however, it is not possible to determine the direction of change for any particular borrower. The first order approximation is therefore to assume no effect of inflation on credit allocations; the analysis can be carried out in nominal terms. III. CREDIT ALLOCATION POLICY 3.1 Overview Several different policy instruments have been used for different sectors at different times. Among these are targets, prohibitions and ceilings, margin requirements, refinance or counter finance, guarantees and urban-rural branch approval. One way of organizing these instruments is in terms of positive and negative instruments. The negative instruments are designed to reduce or eliminate the flow of credit to certain individuals or for certain purposes. Ceiling and prohibitions and margin requirements fall into this category. The positive instruments are designed to channel the flow of credit into specified areas. All the other instruments mentioned fall into this category. Logically of course the categories are complementary in the - 25 - sense that if a negative instrunient is successful it must have a positive aggregate effect on the unspecified areas, and similarly for a positive instrument. The banking system (i.e. the financial system excluding the Central Bank and the government) can be conceptually seen as having an inflow of funds on the deposit side and an outflow of funds on the loan side. In other words, instruments can be classified in terms of how far they are from the final borrower. Right on the outflow side are instruments like lending targets, lending ceilings and prohibitions, exemptions from overall credit ceilings, and margin requirements. In the middle are collateral guarantees and urban- rural linked branch expansion. Farthest away, on the inflow side, are instruments like refinance and liquidity guarantees. 3.2 The Target Approach Loans to small businesses, small loans and exports were the earliest to be designated as priority sectors (1973). In February 1975 banks were notified that the small loan category be given priority in any credit allocation following from an increase in deposits. In September 1975 a lending target of TK 1 crore per bank was fixed for nationalised commercial banks' lending to this category. During the same financial year (1975-76) restrictions on housing loans were lifted. In 1976-77, deposit-linked targets were imposed on each nationalised commercial bank. The targets were 1 1/2% of deposits (as on June 1976) to be loaned to the small loan category, 1 1/2% to the transport sector, and 1% of deposits for residential housing. Informal guidelines were also issued for increased lending in less developed areas at a controlled rate one percent below the general interest rate ceilings (i.e. initially at 12%). In 1977-78, the loan target for small loans was raised to 2% and maintained at that level - 26 - for the next three years. In contrast, targets for house building and transport were maintained at the same level in 1977-78 and then eliminated the following year. In 1977-78 loan interest ceilings of 11% on loans for rural housing and 5% on loans for multistoried buildings were imposed. These compared with the general ceiling of 12%. Table 9 shows, for the three nationalised commercial banks the loans made to the three targeted sectors as a percentage of their total loans and as a percentage of target. The data for the three banks suggest that banks were quite successful in meeting the small loan targets, but not those for housing and transport. There is some variation among banks however: Sonali bank was successful in small loans and unsuccessful in the other two; Agrani bank was successful in transport and partially unsuccessful in small loans and housing; and Janta bank was successful in both small loans and transport. This variability in success (as measured against targets) suggests that it is not useful to impose uniform targets on banks. Each bank has its own special orientation and expertise. Given the different sources of information and information links the banks would tend to specialise in special segments. Uniform targets reduce the flexibility of the system. A recognition of this probably lead to the elimination of targets on house building and transport loans after a trail period of two years. Table 9 also gives the allocation of credit to each priority sector as a proportion of the total advances of each bank. In table 1, the share of advances of all scheduled banks going to the transport sector showed a declining trend over the 1978-80 period; it went from 4.3% in 1975 to 3.1%. in 1980. The allocation of credit by the three commercial banks in table 9, is _ 27 - Table 9: Advanees to Small toans rnd Smatl Tndtstry. HouRse Pu1ldfnr and Transport as a proportion of targeted amounts - Three Banks kec. 1975 Dec. 1976 Dec. 1977 Dec. 1978 Dec. 1979 Dec. 1980 1. SONALI BANK (S) Small loan. and small lndustry As a percentage of total advanceti 2.19 2.o9 2.44 2.72 2.43 2.25 As percentage of target: N.A. 701.00 126.56 109.51 97.82 98.80 House Building Percentage of total advances .07 .17 .46 .74 .64 .54 Percentage of target IIA. N.A. 36.46 59.55 N.A. N.A. Trans porlt As a percentage of total advanceo .82 .89 .71 .94 .49 .37 As a percentage of target. NA. N.A. 36.89 50.26 N.A. NA. 2. AGRANI BANK (A) Small loans As a percentage of total advanceis .80 1.91 3.63 2.43 1.61 1.38 As a percentage of target N.A. 291.00 201.18 95.67 62.47 53.15 Rouse Building As percentage of total advances a .07 .91 1.29 1.26 2.13 As percentage of target N.A. N.A. 75.34 101.48 N.A. N.A. Transport is percentage of total advance. * 1.14 1.18 1.27 1.00 1.10 as percentage of target N.A. N.A. 65.54 67.02 N.A. .A. 3. JANATA BANK (J) small Loans As percentage of total advances .0 1.71 1.87 2.59 2.19 * As percentage of target N.A. 608.00 117.91 129.77 112.09 a Rouse Building As percentage of total advances a * * .89 1.56 * As percentage of target N.A. N.A. C 89.03 N.A. N.A. Traaport As percentage of total advances 2.72 2.11 1.89 1.92 1.91 a As percentage of target N.A. N.A. 118.89 128.67 N.A. N.A. 4. ThREE BANKS (S. J, A) Smll Loana (S, J, A) As percentage of total advances .96 1.45 2.43 2.61 2.18 a As percentage of target V.A. 364.00 138.58 114.02 95.71 a House building (S & A only) A percentage of total advances a .13 .63 .93 .83 .98 As percentage of target N.A. N.A. 49.97 74.09 N.A. N.A. Transport (S, J, A) As percentage of total advances a 1.52 1.33 1.42 1.16 a As percentage of target N.A. N.A. 75.84 83.04 N.A. N.A. N.A. a Not applicable a - Data not available Sorce: Annual Reports of the respective banks - 28 - broadly consistent with this trend. It can be concluded therefore that the setting of a formal lending target for the transport sector had little effect on this down-trend. Loans for housebuilding, by Sonali and Agrani banks, rose fairly rapidly from 0.13% of advances in December 1976 to 0.93 of advances in December 1978 (Table 9). This rising trend was maintained by the Agrani Bank during the next year, while that of Sonali Bank was reversed leading to a i small overall decline. Clearly the increase is not solely due to targeting policy, as the removal in 1975-76 of restrictions on lending to this sector, would have led to some re-allocation towards it. The earlier analysis of loan rate ceilings also suggests that differentially lower ceilings on rural housing loans (-1%) and multistoreyed housing (-6%) might have been responsible for breaking the uptrend in housing loans. However rural housing loans were an insignificant part of the institutional loans for housing, before the special scheme was introduced in 1977-78. Similarly multistoreyed housing is a relatively new phenomenon in the housing market and would not have had an immediate impact on total housing credit. This is consistent with the different effect on the three banks. I have shown elsewhere [Virmani (1982)] that there are certain problems of information flows and interpretation which can lead to inefficiency in the form of exclusion of certain types of borrowers and certain types of economic activity. The analysis suggests that there may be a discrepancy between the social and private returns to loans which requires a short term subsidy to the banks. It is also suggested there that the subsidy will (probably) have to be linked with a requirement to lend to the identified borrower categories. Any element of forced lending, must however be only for a short introductory period. This induces banks to collect information and generate links with the - 29 - neglected sector, a kind of forced learning by doing. A policy of targeting/guidelines when accompanied by appropriate incentives can therefore be justified for short periods for new and unfamiliar activities. Thus the initial success in increasing the share of house building loans may have been partly due to the guideline targeting approach. This analysis is also supported by the differential performance of banks in terms of loan share of the small loan category. Thus targeting had little effect in the case of Sonali bank which already had a proportion of its advances going to this sector, and already had information links. On the other hand the Agrani and Janata banks had only .8% and 0% respectively of their loan portfolio going to the small loan category on December 1975. This had increased to 3.63% for Agrani by December 1977 and 2.59% for Janata by December 1978. In both cases it declined after the peak. Three years in one case and four years in the other was probably more than enough time for the initial information links and learning to take place. Normal criteria would eventually begin to apply. Thus the final stable level of lending to this sector represents a relatively efficient level of lending compared both to the low initial levels as well as to the peaks reached under pressure. In 1979-80 banks were asked to raise their credit deposit ratios in less developed areas, including the Chittagong Hill-Tracks to 75%. It is highly unlikely that general development problems can be addressed solely or even primarily by forced lending. Widespread experience, including that of many banks in Bangladesh, indicates that a package of inputs and programme (technology, marketing etc.) is necessary. Moreover with the interest changes in October 1980, interest ceilings for these areas are at 13%, that is 2.5% below the general rate. This provides a disincentive to bank lending. Further as the target is linked to deposits from the same areas, if the target - 30 - is forcefully imposed, this would also provide a disincentive for raising deposits from these areas. A target of 1% of bank deposits for lending for non-traditional exports was introduced in 1979-80 and continued thereafter. Policies directly related to export profitability need to be examined very carefully as suggested in the introduction. If there are also information problems in export lending, guidelines and targets can be useful for a few years if accompanied by adequate incentives. However from October 1980 the interest ceiling for export lending at 12% was 3.5% below the general ceiling. Informal targets were also set for agricultural lending. However as target setting was not the major policy tool for agriculture, credit to this sector will be considered in the section of the primary policy tool used. 3.3 Ceilings and Prohibitions In the pre-1975 period of high inflation, there were many restrictions on the use of various types of financial securities, capital equipment and real estate as collateral. These were gradually lifted. In the absence of interest rate ceilings such restrictions tend to raise interest costs to borrowers (see Virmani 1982). Thus credit would be shifted from those borrowers whose interest rates were previously against the ceiling to those who were below. The latter would now be paying higher rates but getting relatively more credit. Inefficiency is increased and credit tends to shift from unfamiliar, more risky borrowers to familiar less risky borrowes as perceived by the bank. Such restrictions would also tend to shift credit from productive to speculative purposes as the latter are less likely to be based on explicit collateral than on implicit collateral; for example, the margin of stock value constituting the borrowers equity in a commodity stock. Implicit collateral (e.g. the value of accounts receivable) can of course be used even - 31 - on production loans; but this is done only for a bank's best or prime customers. The most widely used ceilings, usually in times of general inflation, are those on credit to large borrowers, or to all borrowers. Thus for example in February 1979 banks were exhorted to limit credit to borrowers and reduce it where possible. Later in the year inventory norms were set and the banks were told to reduce credit gradually in accordance with these norms. The objective was to reduce credit to those with high inventories and large credit. For producers this acts basically as a loan amount ceiling. In December 1979 formal ceilings were imposed on all borrowers. Industrial borrowers were limited to 90% of their maximum outstandings in the previous year. Traders were limited to 85% while export finance, term finance and special programmes were exempt. In August 1980 the cuts were withdrawn. Though credit ceilings are usually used for controlling general credit expansion they have allocational effects which are the primary focus of our paper. Credit ceiling may also be applied selectively; to mitigate the harsh effect of general ceilings on particular sectors or borrowers. To this extent they can be seen directly as an allocational instrument. I have shown elsewhere (Virmani, 1982) that ceilings on loan amounts to producers, when effective, reduce the cost of loans to the borrower, through a lowering of interest rates. Further they are likely to be more effective against the larger, more organised better placed borrowers, mainly because they are easier for a government to police. Thus they are likely to have perverse welfare effects within the set of borrowers. When applied to all existing borrowers they are likely to channel loans away from current borrowers to new less productive borrowers. When applied to a subset of borrowers they will tend to shift credit to those not subject to these restrictions, while lowering - 32 - interest rates to the former. But a better way of rechanelling without adverse distributional effects is to put an interest tax on lending to non- favored borrowers. Overall efficiency in use of credit will deteriorate under both policies, unless all favored creditors were previously getting less than the efficient amount of credit. This is unlikely to be true. Loan ceilings on credit which is going into stocking for speculative purposes has somewhat different effects. In this case it can be shown that an effective ceiling on loan amounts per borrower will increase the interest rate on loans. However if these rates cannot be raised because of interest rate ceilings then no loans will be made to the potential borrowers. Loan ceilings can therefore be useful in this case if they are effective. 3.4 Margin Reguirements: Overview and Analysis In Bangladesh, as in many other countries, government setting of minimum margin requirements have been used primarily as an instrument for controlling "speculation and hoarding", in specific commodities. Thus they were in most active use when prices of particular commodities considered to have large adverse social consequences started to rise very rapidly. Among the specific commodities have been rice and paddy, sugar, oil and oil seeds, jute and jute goods, onions and chillies, salt, cotton yarn and textiles, and synthetic yarn and textiles. As shown in Table 10, at the start of the period under consideration the margin requirement on Jute and Jute goods was implicitly 100% as all lending for this purpose was banned. This was motivated by the high rates of growth of prices, which rose by 63% in 1974-75. As price growth eased to 4.4% in 1975-76, credit was allowed to this sector. However a sharp rise of 40% in the prices of these goods in 1977-78 following on the heels of a rise of 11% the previous year does not seem to have led to an increase in margin Table 10: Rate of Growth of Prices and Imposition of Margin Requirements (In percentages) 1974-75 1975-76 1976-77 1977-78 1978-79 1979-80 1. Jute and Jute Goods Margin 100. 0(?) Rate of price growth 63.00 4.37 11.32 39.99 1.18 2. Salt Margin 50 0 b/ Price growth 751.0 -66.84 20.47-' 5.8&- -17.9 3. Sugar (Alloted by food Dept) t¶argin 0(?) 50 Rate of growth of crude prices 0 -21.10 1.60 12.00 4. Oil & Oil Seeds (Oil Mills-Margins) Local 0(?) 50 Imported 25 0 Finished Product 100(?) 50 Mustard Oil Prices (Rate of Growth) -32.18 -15.00 35.20 -0.20 5. Cotton, Yarn & Textiles and 50 0 50 Synthetic Yarn & Textiles Rates of Growth in; Sari Prices -0.47 -1.45 10.65 26.49 2.81 WPI for Manufactures 49.73 -43.07 -2.63 5.97 7.41 6. General Merchandise Trade Margins 0(?) 50 60 Rate of Growth of COL Index for Middle Income Groups 67.17 -8.40 2.40 12.60 8.30 18.50 COL Index for Non-Food 52.10 11.40 4.40 9.50 9.30 14.60 COL Index for Food -16.00 1.05 15.64 6.80 20.96 a/ There were informal guidelines in margins during this period. b/ Informal norms were prescribed to banks. Sources: Resume of Financial Institutions, Finance Ministry, Various Issues. Bangladesh Bank, Annual Reports. Statistical Absgract of the Bangladesh (-ov,rnmnt. - 34 - requirements. Thus the use of this instrument seems to decline over time, perhaps corresponding to a decline in concern over hoarding. Similarly the margin requirement of 50% on advances against salt was accompanied in 1974-75 by rate of growth of 751% in salt prices. These margin requirements were eliminated the next year as prices fell by 66.8%. Guidelines were issued in 1976-77 as prices again rose by 20.5% and inventory-credit norms were established the following year. The sharp decline of 18% in salt price in 1978-79 led to some positive credit measures. Lending, for sugar alloted by the food department, could be done at 50% margin in 1977-78. As sugar prices were controlled throughout the period, the best available (though imperfect) measure of tightness or rationing in the sugar market was gur prices (jaggery). Thus gur prices, after falling by 21.1% in 1976-77 increased by 1.6% in 1977-78 suggesting a corresponding increase in demand pressure relative to supply in the sugar market. This is confirmed by the still higher price rise of 12% in gur prices in 1978-79. Credit margins on borrowing by oil mills against local oil seeds and oils were put at 50% in 1977-78. Similarly a 25% margin requirement was specified on imported oils and oil seed during the same year. Prices of mustard oil, which constitutes the most important local oil, provide the answer. Price declines of 32% and 15% in 1975-76 and 1976-77 were followed by a price rise of 35% causing alarm to the government. A decline of 0.2% in mustard oil prices in 1978-79 led to a removal of margin requirements on imported oils and seeds and a fall in margin requirement to 50% on finished products of oil mills. Margin requirements seem to have been directed primarily against agricultural commodities, and within that primarily on goods of common consumption. Thus, for example, lending to private rice and paddy traders has - 35 - been banned for most of this period, giving an effective margin requirement of 100%. The other major good in the consumption basket is clothing. As we might expect both cotton yarn and textiles and synthetic yarn and textiles have been subject to margin requirements (Table 10). The early period also witnessed wider use of margin requirements on industrial inputs suggesting use of margin requirement as a more general credit control and inflation fighting device. As recently as 1978-79 margin requirements on general merchandise trade were raised from 50% to 60%, following observation of accelerating inflation in 1977-78 and 1978-79 (Table 10). Before determining the effects of this policy instrument in different contexts, it is useful Ito analyse the effect of margin requirements in a simplified context. Consider a situation in which credit is being used purely for speculative purposes, that is to buy and hold commodity stocks in anticipation of a price rise. Consider a bank and a speculator who are neutral in their attitude to risk and are therefore interested only in the expected returns. To begin with we assume that the speculator has a fixed amount of outside funds or equity available to him for purchase of commodities. Given this equity, we can define a loan supply curve for each borrower-lender pair which is limited at one end by the condition that banks' have positive expected profits. The speculator's demand curve is perfectly elastic; if at any interest rate it is profitable for him to borrow it will be profitable to borrow an infinite amount. Thus a 'rationing equilibrium' can emerge as shown below (r*, L*), if banks act competitively. In such a situation, if government imposes a margin requirement m (m > e which is e + L* e wi is effective, it is equivalent to a loan ceiling - 36 - Figure 1 LS L~~~~~~~~~~L E ~ ~~~~ X , .~~~~~~~~~~~~~~~~~~~~. 1~~ e L = e (--1) (from m = e+ L , e is the value of the stock). As shown in the ceilings exist, and the new interest required is higher than these ceilings no loans will be made. Even when the effect of government imposed margin requirements falls on loans for productive purposes, it still acts as a loan ceiling. The effect of loan ceilings has been shown above (section 12). When commodity stocks are used as collateral by producers, the ratio of the loan value to the collateral is of course referred to by banks as the 'margin'. This ratio, or bank set ,margin requirement' will, in general, be different for different producers, even for the same collateral commodity. This is because the returns from and riskiness of the productive activity in which the loan is used varies across borrowers. All the terms of the loan contract will therefore differ: more productive borrowers will obtain better terms than less productive ones. This will tend to result in lower effective margins for the former than for the latter. Returning to the case of speculator borrowing, the speculator's equity contribution may not be fixed as assumed. If, instead, the speculator has a rising marginal cost of equity funds, similar results are obtained. The - 37 - only difference is that an effective margin requirement raises the marginal return to speculator equity and therefore brings forth additional equity funds into speculation. As a result interest rates do not rise as much as in the fixed equity case, and the number of eliminated borrowers will be smaller. Thus effective margin requirements can reduce the flow of funds for speculative purposes and raise the cost to current borrowers. 3.5 Margin Requirements: Differential Impact In Table 11 and 12 we present data on the distribution of credit by margin requirements for some of the commodities reviewed above for which data is available for the relevant time periods. These statistics relate to Nationalised Commercial banks (NCBs). The results for sugar, and oil and oil seeds are broadly consistent with the above analysis. Thus for sugar the imposition of a 50% margin requirement in April 1978 resulted in a substantial change: Between June 1977 and June 1978, the proportion of credit with no margin requirements, went from 84% to 53%, while that with 50% margin requirement went from 5% to 28%. This was accompanied by an increase in proportion of credit subject to 25% margin. As this credit includes credit going to public sector sugar mills, this represents a fairly effective shift. This shift in the allocation of credit by margins was accompanied by a fall in loan size per borrower from TK 6280 to TK 1040. Total loans against sugar also declined from TK 408 thousand to TK 281 thousand. Both these are consistent with our analysis. By June 1979 however virt:ually no loans were at 50% margin, with about 90% of the loans being at margins of 25% or less. The average loan size had increased back to TK 5820, evert though total loans against sugar were still at a reduced level of TK 245 thousand. - 38 - Table 11: Distribution of Advances by MarpinR for Specified Securities, and Average Loan Size by tlrgl, for saae Seeurities 1. 01 Seeds - Proportion 30-6-76 30-6-77 30-6-78 30-9-78 3-6-79 30-9-80 Percent of Advances ex 88.00S 35.98S 4.512 17.73X 10.32S 202 4.54 8.03 11.05 7.00 251 33.76 9.94 12.81 6.68 30S 15.53 11.22 4O2 50S 3.45 2.65 70.63 63.00 75.23 75.13 > 502 .52 .79 .0 4.68 11.19 Total (All margins) 100.00 io0.0o 100.00 Average Loan Size (in 1000 TK) * OS 66.01 (11) 14.13 .91 (18) .76 1.92 202 1.51 (25) 1.44 11.25 2.55 25Z (17) (39) 3.73 3.37 (2) 11.28 3.72 302 1.86 (3) 6.59 402 502 (9) .94 (10) 1.14 (49) 2.93 (64) 2.63) (67) 1.64 > 502 (1) 1.56 (1) 3.43 0 (54) 2.06 (5) 2.09) (15) 1.09 Total (All margins) (89) 8.81 (125) 3.45 (102) 2.00 (241) .73 (88) 2.54 (100) 1.46 Total Aeount 299.80 431.92 203.49 223.74 146.09 2. 2D111E OIL Percent of Advances 0 95.35% 36.412 21.782 38.612 43.782 10 87.56S 15 20 25 4.04 52.162 30 61.34X 40 50 57.18 > 50 (99.99S) 0 0 52.69S Average loan size (in 1000 TV)* 0 (4) 119.48 (4) 49.82 (5) 2.75 (4) 31.94 (4) 203.46 10 (1) 104.12 20 25 (6) 3.37 (4) 43.15 30 (5) 67.14 40 50 > 50 (99.992) (1) .67 (1) .67 (2) 18.07 (14) 69.96 Total (All mArgins) (14) 35.80 (18) 30.40 (46) 1.37 (38) 3.13 (15) 22 .06 (28) 66.38 Total Advances 501.23 547.26 63.20 118.91 330.93 185.82 3. Sugar 4 GUR Percent of Advances 0 84.332 84.41S 53.04X 56.962 48.842 10 16.93 25 9.842 10.312 18.60S 22.12 24.04 30 so 4.74 27.72 14.15 Averae Loan Size (in TKI 1000)* 0 (20) 21.94 (50) 6.89 (236) .63 (13) 10.65 (14) 8.53 10 (3) 13.80 25 (5) 10.24 (3) 13.02 ".) 52.22 (1) 53.80 (1) 58.78 30 50 (10) 1.93 (30) 2.59 (20) 1.72 > 50 (1) .99 Total (39) 13.34 (65) 6.28 (270) 1.04 (44) 5.53 (42) 5.82 520.38 408.04 280.70 243.19 244.60 * Numbers In brackets are the total nuuber of borrowers taking nans at specifted margins. Source: Scheduled Bank Statistics, variouR lisues. NCR: Nationslited Consercial banks. - 39 - The 50% margin requiremLent on oil and oil seeds was imposed in April 1978. In the case of oil seeds, the proportion of advances with this margin increased from 3% in June L977 to 71% in June 1978; for oils the proportions went from virtually 0 to 57% over the same period. Total loans advanced against oil seed and oils fell from TK 432 thousand to TK 203 thousand and from TK 547 thousand to TK 63 thousand respectively. Average loan size similarly fell from TK 3450 to TK 2000 for oil seeds and from TK 30 thousand to TK 1 thousand for oils. The margin requirement on finished products was reduced to 50% in July 1978 while that on imported oils and oil seeds was eliminated in August 1978. As imports consist largely of oils, both these would affect the loans advanced against oil. By June 1979 the proportion of advances at 50% margin was virtually zero, while total loans had increased to TK 119 thousand and loan size per borrower to TK 3 thousand. In contrast, by June 1979 the proportion of advances against oiLseeds subject to 50% or higher margin had increased from 10% to 80%. This wias accompanied by a marginal increase in total amount loaned and loan amount per borrower. Again these results support the analysis. The common feature of these two cases is that both were agricultural consumer goods considered to be of socio-political importance. As far as urban markets are concerned production trade and consumption are quite clearly separated. Both goods are not very important as intermediate inputs. Therefore, it is relatively easy t:o distinguish between stock holding for speculative from that for productive purpose. In addition, in both cases margin requirements were imposed at the start of a sharp upswing in prices. These features contrast with the case of textile goods. - 40 - Table 12: Distribution of Advances by Rates of Margin for Related Securities Nationalizred commercial Banks (Percent of advances for all margins) 1. Cotton Textiles, Yarn - Export 30-6-76 30-6-77 30-7-78 30-6-79 0 5.87% 6.43% 5.622 * 10/15 2.62 20 50.35 18.35% 23.24 12.13 25 35.54 41.15 38.17 40.22 30 3.19 30.02 9.36 35/40 50 2.31 19.80 28.33 75 Total Advanices (in 1000 TK) 2427.77 587.01 986.82 931.70 Average Loan size* (1432) 1.69 (172) 3.41 (946) 1.04 (358) 2.60 2. Cotton Textiles - Irmport 0 1.07% 53.92% 41.77 75.85% 10/15 20 3.65Z 1.87 4.41 9.11 25 83.52% 38.18 44.33 3.27 30 40 7.96% 1.47 3.20 50 3.13 4.20 7.78 Total Advances (in 1000 Taka) 2394.02 7205.20 6584.11 3746.16 Average Loan size (174) 13.76 (323) 22.30 (236) 27.90 (429) 8.92 3. Cotton & Synthetic Yarn - Imp. 0 19.97% 55.66% 71.47% 51.43Z 5-15 13.56 7.89 20 5.57 2. j 5.69 7.83 9.66 30 27.42% 10.94X 17.64% 40 24.571 2.37 3.35 50 12.82% 5.98 8.68 3.34 75 5.82 Total Advances (in 1000 Taka) 424.91 1321.68 2617.09 2078.27 Average Loan size (191) 2.22 (342) 3.86 (358) 7.31 (374) 5.56 4. other Textiles 0 55.20% 2.47% 16.92% 27.00% 15 18.67% 17.85 15.76 22 5.58 14.24 25 32.86% 38.52 26.05 33.53 ;1, 3.07 5.77 16.06 50 4.252 4.58 . 10.37 Total Advances (in 1000 TK) 381.51 246.83 466.33 419.33 Average Loan slze (62) 6.15 (83) 2.97 (170) 2.74 (126) 3.33 30-6-75 30-6-76 30-6-77 30-6-78 30-6-79 30-6-80 5. Merchandise No. of Accounts 9377 11030 57774 143437 112005 153609 Amount (in 1000 'TY) 46999.00 54422.00 622S4.00 78112.00 111896.00 160603.00 Average Loan Size 5.01 4.93 1.08 .54 1.00 1.05 Average Loan Size (in 1000 Taka) 1. Export Conmoditles 10.78 4.93 1.07 1.82 2.48 2.49 (Jute, Cotton, lextileS 4 Yarn, etc.) iI. Import Commodities 2.77 3.89 3.19 3.24 3.30 3.55 iiI. Other Import Herchandise 3.06 2.91 .27 .11 .27 .34 (wheat, rice, oilseeds, oll, sugar, olives) * No. of borrowers Source: Scheduled Bank Statistics, Various Issues. - 41 - Table 13: Distribution of Advances by Rates of Margin: All Advances of NCBs (In percent) 30-6-78 30-6-77 30-6-78 30-9-78 30-6-79 30-6-80 Margin 50% .2.19% 3.21% 6.30% 6.54% 6.18% 4.98% 60% .07 .07% .05% .51 1.40 1.38 99.99% .24 1.55% .98% .28 .28 0.37 Total Volume (in 1000 Taka) 69239.15 88351.33 115516.37 131177.94 142627.00 210419.38 Average Loan Size (in 1000 Taka) 50% .22 .L8 .16 .16 17.23 .16 60% .06 .04 .02 .20 .29.20 .26 99.99% .10 .69 .14 .05 .03 .05 Total .66 44.87 .15 .16 16.30 .14 NCBs: Nationalized Commercial Barnks. Source: Scheduled Bank Statistics, Bangladesh Bank. - 42 - In September 1975 margin requirements on cotton and synthetic goods (yarn and textiles), was reduced from 50% to 0%. In June of 1976 therefore there were virtually no advances against cotton textiles and yarn (export commodities/Category 1) and cotton textiles (import commodities/Category 2) carrying a margin of 50% or higher. Approximately 86% of advances under each category had margins of 20 or 25%. In the use of cotton and synthetic yarn (import commodities/Category 3), 12.8% of advances were at 50% margin with substantial percentages having margins of 0, 30 and 40%. Similarly other textiles (Category 4) had 87% of advances with margins between 0 and 25%, and 4% of advances at 50% margin. The imposition of a 50% margin requirement on cotton and synthetic goods (yarn and textiles) in March 1977 had a gradual effect on margins applicable to cotton textiles and yarn (export). Proportion of advances carrying a 50% margin increased to 2.3% in June 1977 and 19.8% in June 1978. However the initial fall in total advances from TK 2428 thousand to TK 587 thousand was reversed by June 1978. The immediate effect on average loan size was to double it from TK 1.7 to TK 3.4 thousand. This fell back to TK 1.0 thousand by June 1978. In the other three textile categories there were very little or no effect on proportion of advances having margins of 50% or higher. Thus the proportion of advances having a margin requirement of 50% in June 1976, 1977 and 1978 for category 2 were 0%, 3.1% and 4.2% respectively; for category 3 were 12-8%, 6-0% and 8-7% respectively, and for category 4 were 4.3%, 4.6% and 0% respectively. Similarly total advances and average loan size decreased for category 4 and increased for categories 2 and 3 between June 1976 and June 1977. Part of the reason for these contradictory effects are, the plethora of special policies applicable to imports and exports and the freedom to - 43 - import under the Wage Earners' Schieme since April 1977. The major reason is, however, the difficulty in separating use for consumption from use as an intermediate input, and the holding of stocks for speculation from holding of stocks for production. Thus, borrowing at 0% margin went up between June 1976 and 1977, from 1% to 54% for category 2 and from 20% to 56% to category 3. These were the two categories of goods for which perverse effects were strongest. This suggests borrowing by textile production units, and among them the public sector textile corporations which are major borrowers. Thus the imposition of margin requirements was largely ineffective as credit was diverted from existing trader-borrowers, to new borrowers and to producers. Over a longer time period, new traders can also enter the business if entry (information) conditions are not too difficult. This would further reduce the effectiveness of the policy. General margin requirements on merchandise were increased to 50% in August 1977. The effects of this are shown in Tables 13, 14 and 15. The proportion of all NCB advances having a margin requirement of 50% doubled from 3.2% to 6.3% between June 1977 and June 1978 (Table 13). The increase of 3% points appears small; but the margin requirement was increased only on merchandise and not on other securities. The former constitute approximately 65% of total loans (Table 14). By making a simple adjustment we can say that the re-chanelling was approximately 4.8% (instead of 3%) within the advances which had merchandise as security. Over the same period, the proportion of all loans by NCBs going to trade declined from 35% to 28% (Table 15). This is consistent with our analysis. A similar conclusion emerges for all banks if we look at the proportion of loans going into trade credit (Table 1). Margin requirements on mtsrchandise were increased further to 60% in July 1978. As a result, by June 1979 the proportion of advances with 60% Table 14: Distribution of Advances by Securities Nationalized Commercial Banks FY 75 FY 76 FY 77 FY 78 FY 79 FY 80 A. Gold and Precious Metals .7 .9 1.2 1.2 1.2 1.0 B. Securities 1.3 1.2 1.1 1.4 1.4 1.0 C. Merchandise 74.0 72.6 65.0 62.0 64.2 68.3 (i) Raw Jute 21.9 15.4 6.5 6.1 12.6 11.6 (ii) Jute Manufacture 14.8 11.5 11.3 14.1 11.8 7.8 (iii) Other Export Commodities 6.2 6.4 8.2 7.5 6.4 14.4 (iv) Imports 31.11 39.3 38.9 34.3 33.3 34.4 D. Machinery and Fixed Assets 2.0 2.4 2.1 2.0 2.6 2.1 E. Real Estate 2.6 2.3 2.7 2.9 3.2 2.9 F. Financial Assets 1.9 2.2 3.1 2.4 2.3 1.6 G Miscellaneous 17.5 18.5 24.8 28.1 25.2 23.5 (i) Secured 11.4 12.3 17.6 16.5 19.0 16.9 (ii) Unsecured 6.1 6.1 7.3 11.6 6.47 6.6 Source: Scheduled Bank Statistics, Bangladesh Bank. - 45 - margin increased to 1.4% from .05% in the previous year; an increase of only 1.35% points (Table 13). The proportion of advances going to the trade sector increased to 32% (from 28%, Table 15), along with the use of merchandise as a security (Table 16). The following year the advances having a margin of 50% or over declined to 6.7% of total advances from 7.9% the previous year (Table 13). Thus the results were quite mixed; any effects of the increased margin requirements appear to be quite transitory when applied on such a broad scale. The reason is that both borrowers and equity funds can move much more easily between production and trade, and between different types of securities. Even if immediate movement is precluded, given time such movement seems almost inevitable. Thus the use of margin requirements appears most effective for agricu'ltural consumer goods for wlhich short run price movements have adverse socio-political effects. But even in this case the policy may not be effective for very long, as the possibility of speculative stock holding within the agricultural sector would increase over time. This could also lead to a diversion of agricultural credit from productive to speculative usage. In fact one of the banks gave us an example of diversion of credit from agricultural production to stockinig of chillies in anticipation of a price rise. Table 15: Distribution of Advances by Economic Purposes ' Nationalized Commercial Banks FY 75 FY 76 FY 77 FY 78 FY 79 FY 80 Agriculture, Hunting, Forestry and Fishing 2.2 3.86 4.5 7.98 10.6 11.5 Mining and Quarrying 0.1 .1 1.4 .77 .1 1.5 Manufacturing 50.4 43.42 39.2 40.2 39.2 39.4 Construction 1.0 .9 1.5 1.7 1.7 1.6 Electricity, Gas and Water .1 .1 .1 .32 .7 .5 Wholesale Trade 30.4 36.2 35.0 28.2 32.2 31.4 Finance.and Real Estate 4.1 2.6 2.8 2.6 3.2 2.9 Transport, Storage and Communication 6.1 7.1 6.8 5.7 4.9 4.2 Others 5.5 5.7 8.8 12.6 7.5 6.8 Source: Scheduled Bank Statistics, Bangladesh Bank - 47 - 3.6 Refinance and Counter Finance: Overview Central Bank (Bangladeslh Bank) loans to scheduled banks go under several names, including refinance, counterfinance and back-to-back finance. The percentage of refinance has varied between 30% and 100%. One way of organizing these different programs is to separate the ones which have 100% counterfinance or back-to-back finance from those which have less. In the first case the bank is acting puretly as an intermediary between the central bank and the final borrowers. In the second case a proportion of the lending must be financed out of deposits and therefore the allocation of deposits between refinanced and non-refinanced lending becomes relevant. An alternative way of looking at it, which is more fruitful in the present case is in terms of specialised banks which direct lending to specific sectors or sub sectors, and commercial banks. For the purpose of the refinance issue, the agriculture oriented banks - BKB and BSBL - are of relevance to the first category. All agricultural sector lending of these two banks has been refinanced at 100% by the Bangladesh banks (though very recently the proportion has been reduced to 95% for BSBL). The refinance rate has been 6% since 1974. It has thus been subsidized to the extent of 2% points (below the regular discount rate of the Bangladesh bank) between June 1974 and October 1980 and by 4.5% since the discount rate was raised to 8.5% in 1980. Table 16 shows the sanctions and disbursement of loans from the Bangladesh bank to the agricultural banks (BKB and BSBL). The amounts that these banks are entitled to have refinanced is calculated at 100% of their disbursements. Disbursements as a proportion of sanctions fluctuated considerably over the period but showed a rising trend; going from 78% in 1975-76 to 100% in 1979-80. However in 1979-80 disbursements exceeded Table 16: Use of Refinance by Bangladesh Krishi Bank (BKB) and Bangladesh Samabya Bank (BSBL) Bangladesh Bank Loans to BKB & BSBL 1974-75 1975-76 1976-77 1977-78 1978-79 1979-80 1980-81 1. Sanctioned 2489 3300 4616 8885 8734 12970 15859 2. Disbursed 2075 2589 3922 6347 8617 13015 16739 3. Cumulative Disbursement (from 1974-75) 4664 8586 14933 23550 36565 53304 4. Cumulative Sanctions (from 1974-75) 5789 10405 19290 28024 40994 56889 Agricultural Banks (BKB & BSBL) 5. Entitled Borrowing /1 2605 2768 4301 5434 8189 14600. 21102 (at 100 of disbursement) 6. Cumulative Entitlement 5373 9674 15108 23297 37897 58999 7. Disbursed/Sanctioned (Row 2 - 1) .83 .78 .85 .71 .99 1.00 1.06 8. Disbursed/Entitled (Row 2 5) .80 .94 .91 1.17 1.05 .89 .79 9. Cumulative Disbursement/ Cumulative entitlement (Row 3 *- 6) .87 .89 .91 1.01 .96 .90 10. Average of Rows 7 and 8 .815 .86 .88 .945 1.02 .945 .925 11. Cumulative Disbursement/ Cumulative Sanctions(Row 3 * 4) (.83) .80 .83 .77 .84 .89 .95 1/ It is not entirely clear from the data whether disbursements are for agricultural credit alone or also include trade, storage marketing of agricultural goods. Source: Bangladesh Bank, Annual Reports - 49 - sanctions for the first time and in 1980-81 this reversal become more pronounced. This peculiar situation arises because of delay in disbursement of loans committed in the previous year. The proportion of refinancing actually provided can be measured approximately by the loans disbursed by the Central Bank divided by the entitlement, as measured by the agricultural banks advances for the year. These rise to a peak of 117% in 1977-78 and then decline to 79% in 1980-81. Because of the carry-over of previously committed loans and the partial carry over of previously sanctioned loans, a better (but still approximate; because of the abrupt start at 1974-75) measure is the cumulative disbursement divided by the cumulative entitlement. This shows refinance proportion rising from 87% in 1975-76 to 101% in 1977-78, suggesting that most backlogs were cleared by this time. Refinance remained aLt the fairly high level of 96% in 1979-80 but fell sharply to 90% in 1980-81. This suggests that even though actual refinancing by the Bangladesh bank may have fallen below 100%, due to the application of constraints ona general credit expansion, the goal remained at 100%. In contrast, in 1979-80 and 1980-81 loan sanctions are actually below loan entitlements for the first time, suggesting an implicit movement away from the goal of 100% refinancing of agricultural banks. The commercial banks, and the BKB for its non-agricultural lending, are entitled to refinance on a wide variety of items. The current rates of refinance are as follows: Food and Fertilizers (BADC), Petroleum and Petroleum Corp. 100 Exports Non-Traditional 100% Traditional 50% Small Loans 0% (100% until late ) Sugar 50% Jute 50% SACP 50% - 50 - The refinance rate for the special agricultural programe (SACP), like the agricultural lending of BRB and BSBL, carried a 2% subsidy on the Bangladesh bank rate. All other sectors got refinace at the discount rate which was 8% between June 1974 and October 1980 and 10.5% thereafter (Table 4). The total amount of refinance taken under these catagories rose steadily from 251.53 lakhs in 1979 to 703.88 lakhs in December , a growth of 180%. The annual compound growth rate from June 1979 to June was 67% a year, representing a fairly substantial increase in refinance. Jute was the largest user of refinance throughout the period, but grew by only 23.5% from 119 lakhs in June 1979 to 146.98 lakhs in December . Refinance to the petroleum corporation started in December 1979 and has been the second largest since December 1980. Food and fertilizer have changed places for the next position. Given the diverse items for which refinance is provided it is useful to look at some of the explicit and implicit objectives. Both food and jute are subject to annual fluctuations in production. Refinance is seen as a means of providing credit for holding stocks in a deflationary situation. Our earlier analysis of speculative credit suggests that commercial banks would tend to provide little credit in this situation. But this question really needs to be looked at in terms of the buffer stocking policy of the government. Macroeconomic issues are also important. A properly worked out buffer stocking policy would trade off sectoral effects and macro economic effects, and work out a self supporting buffer-stock policy. Any desired subsidies, should in general go direct to the producer and not to private traders. Sugar production has a strong seasonal pattern. Both lender banks and potential borrowers know that prices will fluctuate in a seasonal pattern - 51 - over the year. The analysis of speculative credit does not indicate that banks will not provide credit in this situation. However a seasonal buffer stocking policy which is self-supporting could be worked out if necessary. If it is thought necessary to provide subsidies to producers or consumers these could be provided directly. Provision of refinance to the petroleum corporation is used essentially to fill the gap between the rise in oil prices and the rise of final product prices. Thus it can be seen as a means of providing a consumption subsidy to oil consumers. Even if such a subsidy is to be given there does not appear to be significant reasons for using the banking system for this purpose. The provision of refinance to BADC for fertilizer has a completely different purpose. Subsidies are provided explicitly through the budget: Delays in receipt of the subsidy or short term liquidity problems are however met by providing 100% refinance to banks for their lending to BADC. This appears to be a more satisfactory method for achieving a similar purpose. In the case of non-traditional exports information problems can effect the flow of credit, as I have shown elsewhere (Virmani 1982). A determination must be made, however, whether this is the primary, or even a major problem. This requires a prior examination of the foreign exchange markets and the policies which impinge directly on exports and imports. In addition one should also look at more general agricultural or industrial policies which impinge on sectors which are potential exporters. Given appropriate policies in these areas, credit policies can be used to remove any inefficiencies in loan markets. These generally relate to new export products with which bankers are unfamiliar. Credit policies when used for an initial period can be useful in removing inefficiences. - 52 - 3.7 Refinance Usage, Incentive Effects and Usefulness The use of refinance policy as an instrument for redirecting credit allocation can be examined in the context of the analysis in Virmani (1982). We start by considering the simplest case of a perfect deposit market, in which a bank acts as a price taker at a deposit interest rate of i (a more realistic case is considered below). Let the refinance proportion be h and the refinance rate p. If the refinance rate is greater than or equal to the deposit rate, no refinance would be availed of and refinance provides no incentive. If the refinance rate is less than the deposit rate, the cost of funds for the refinanced sector becomes i = hp + (1 + h)i which is less than i the deposit rate. I have shown that the direct effect of this is to increase loan amounts and reduce interest rates to borrowers. Loans would also be made for the first time to previously marginal potential borrowers. Both these would shift credit allocation towards the refinanced sectors. If interest ceilings exist, the effect of refinancing is to reverse, partially, the harmful effects of such ceilings. The lowering of the 'free' interest rate on loans means that on the margin borrowers are shifted from having an effective credit ceiling to not having one. The effect of refinance policy on collateral requirements is uncertain, as these may rise or fall. For those borrowers who are still constrained by the loan interest ceilings, the adverse distributional consequences of interest ceilings on the poor (little wealth), may be countered or enhanced by refinance policy. Table 17 gives the average weighted deposit costs over the period. The refinance rate for the agricultural banks has been 6% since 1974. (The refinance rate is of course 8.0/10.5 for the other sectors). This compares with a interest rate of 3.5 to 4.3% till 1980 and 7.0% in respectively. Under such a situation our analysis suggests that these banks would not have Table 17: Average (weighted) Deposit Rate FY74 FY75 FY76 FY77 FY78 FY79 FY80 FY81 A. Weighted Average Deposit Rate 3.0 3.5 4.2 4.3 4.2 4.3 4.3 7.0 B. Deposit Cost Adjusted for Reserve Requirements i- - i - 2.3 1.3 2.4 2.5 3.4 2.5 3.4 3.4 7.0 3 Co A -B 1.7 1.1 1.7 0.9 1.7 0.9 0.9 0 Ln JI - 54 - taken any refinance, at least till October 1980. Because of the existence of reserve requirements, however, we might expect that the effective cost of funds is higher than the average interest rate. It is easily shown that the effective rate i is as follow i - bai bai i g = - - g 1-a 1-a 1-a where a is the total reserve requirement, b is the proportion of reserves required to be put in government or other securities earning a rate rg . It is assumed that a proportion (1-b) of reserves is held as cash which earns no nominal interest rate. For Bangladesh a = 0.25, ab = 0.2 and it = 8.5% (rate on ad hoc treasury bills). Using these amounts we find that the adjusted cost of funds is less than the unadjusted (table 17). This happens because deposits can costlessly earn a net return of (8.5 - i)%. On the other hand we have not adjusted for the transaction cost of dealing in government securities, or for the administrative cost of servicing depositors. The first would reduce the negative adjustment in the above formula. The second means, that the effective cost of deposits is i + t and that adjustment raises both the effective interest cost and the effective administrative cost. A rough calculation suggests an administrative cost of 1.3 to 2% of deposits. Use of these would reduce the gap between the adjusted and unadjusted (effective) cost by 0.5%-0.7%, still leaving the former less than the latter. If we take the unadjusted values as an upper limit, there would have been no refinance taken under this situation till October 1980. This contrasts sharply with what actually happened. Clearly the perfect deposit market approximation is not useful for current purposes. The more appropriate assumption is to assume that deposit - 55 - rates are fixed by the government: and total deposits available are therefore fixed exogenously or given. We also make the approximation that the deposits with each bank are also not subject to their control in any way. In this case it is easily shown that banks will earn rates equal to s - i (positive) times their deposits, where s is the opportunity cost of funds (a shadow-price). The loan decisions of the bank, and the effects of various policies, can effectively be analyzed by using s instead of i. Consider the question of refinance. In this case refinance policy can be effective as long as the refinance rate p is less than the opportunity cost s. The effective cost of funds for the refinanced sectors is therefore p hp + (l-h)s which is less than s. As an aside it should be noted that as the agricultural banks were getting refinace at 6% their opportunity cost of funds (s) must be higher than this. The allocative effect is virtually the same here as in the previous case both with and without interest ceilings. To restate, there would be a shift in credit allocation towards; the refinanced sector, and the full amount of entitled refinance would be used. The experience of the agricultural banks is broadly consistent with this analysis. As shown in the beginning of this section (and Table 16) the banks were taking virtually 100% of this refinance entitlement till 1979-80. In Table 1, and the accompanying discussion it was shown that the allocaton of credit has increased progressively from 11% of total credit to 30.6.75 to 19% in 30.6.80. It was also shown that the index for output adjusted credit to agriculture similarly doubled from .17 to .34 over the same periods. The contrast between the performance of the agricultural banks and the commercial banks (to be analyzed subsequently) suggests that refinance policy combined with the existence of specialized agricultural banks was an important causative factor. - 56 - Table 18 gives the usage of refinance by the scheduled banks (including limited usage by BKB for non-agricultural purposes) by purpose and by bank. Only for rural credit, small loans and petroleum is there anywhere near fuil usage of refinance, with a simple average over the period of 96%, 97% and 97% respectively. The lowest usage is in the food sector with an average of 50%. Usage for all sectors taken together ranges form 69% to 87% with a simple average of 79%. There is a similar variability in bank wide usage. This underusage appears inconsistent with the analysis presented above. The reason for this discrepancy lies in the existence of bank wide ceilings on total advances. Consider a situation prior to the introduction of refinance. There is some division of loanable deposits (total minus required reserves) between sectors which are to be refinanced and those that are not, say LR and LNR respectively. When refinance policy is introduced with refinance proportion h(O < h < 1), hLR of extra funds become available. This is equivalent to an amount hLR Of the loanable deposits of banks becoming free. In the absence of credit ceilings, loans to the refinanced sector would rise as already LR analysed. If they rise to an amount L = (lh) it would mean that all the excess funds have been used up, and s and LNR would remain unchanged. If initial Li is less than LR must fall and LNR rise, and vice versa R ~~~(1-h) s if initial LR is greater than LR/(l-h). In either case, after full RR adjustment, there would tend to be an absolute and relative increase in credit going to the refinanced sector. - 57 - Table 18: Refinance: Usage as a Percentage of Entitled Limits: Nationalized Commercial Banks June 1979 Dec. 1979 June 1980 Dec. 1980 June 1981 Dec. 1981 All Banks Rural Credit 100.00 100.00 98.35 83.07 96.96 96.96 Small Loans 99.93 95.40 87.54 100.00 100.00 97.65 Export Bills 55.24 88.30 57.56 79.41 86.98 93.27 Food 67.70 48.77 74.26 -50.57 39.97 15.88 Jute 79.43 97.14 91.17 96.15 81.98 85.92 Fertilizer 84.86 79.79 53.66 79.25 75.69 99.69 Sugar Mills 100.00 87.48 100.00 35.05 100.00 100.00 Petroleum - 100.00 100.00 100.00 94.35 92.37 Textiles - - - - 75.91 100.00 Other 0 86.98 80.36 74.17 75.60 96.68 Total 68.8 86.45 80.61 82.44 75.63 79.92 All Sectors Sonali 74.7 97.93 96.57 84.81 91.44 82.84 Janata 85.1 88.51 76.71 94.00 78.69 85.79 Agrani 56.3 69.85 45.75 60.44 15.14 33.82 Pubali 50-4 89.02 72.28 49.19 55.71 81.35 Ruaalu 74.70 78.00 92.05 84.94 80.11 76.15 Uttara 75.10 7Z.36 73.80 85.27 84.19 90.00 BKB /_ 36.4 100.00 20.00 - - - - No refinance related lending i/ Non-agricultural refinance. Source: Bangladesh Bank - 58 - If overall credit ceilings exist, which are totally independent of the refinance policy, total loans cannot increase. Existing deposits are enough to finance these loan. This means that any refinance taken goes into free reserves. If the net return from allowed investment out of free reserves is lower than the refinance rate, no refinance would be taken. If the net return is initially greater than the refinance rate full refinance would be taken. Only if the marginal net return falls with amount invested would partial refinance be taken. In the case of NCBs, until October 1980 the refinance rate was 8%, while ad hoc treasury bills had a rate of return of 8.5%. If an adjustment for administrative costs were factored in, this would suggest at most a marginally higher net return on allowed investment. For the refinance rate at 10.5% is clearly greater than the gross return which remains at 8.5%. The peak rate of calculated reserves, which generally occur in June due to seasonal factors, were 26.9% in 1979, 26.6% in 1980 and 27.9% in . Given that reserves in other months can be significantly lower than these values, and that the required minimum ratio is 25%, there seem to be no excess reserves over this period. In the context of the above analysis (credit ceilings operative), the absence of free reserves is consistent with the low profitability of allowed investment. Leaving aside food and export (bills) financing, which is exempted from credit ceilings, this still leaves open the problem of explaining usage of refinance at greater than zero rates (Table 18). The explanation lies in the fact that credit ceilings have not been independent of refinance policy. Another element which also enters is forced lending. Credit ceilings for a forthcoming year start from the previous years ceilings. Increases in credit ceilings, though based primarily on expected increases in deposits and on past Table 19: Advances and Depo;:its of Scheduled Banks excluding BKB & BSB(In 1000 Taka) 1975 June Dec. 1975 June 1976 Dec. 1976 June 1977 Dec. 1977 June 1978 Dec. 1978 June 1979 Dec. 1979 June 1980 Dec. 1980 1. Demand and Time Deposits 93434 194909 107564 128125 133413 156708 160146 194664 211362 244209 248818 .294633 2. Total Credit Advances & Bills 67927 84883 82136 96548 100544 129965 127568 162434 159775 200739 224316 249290 3. Govt., Food credit 443597 195212 1718%6 70339 47341 241127 349689 441905 370256 125757 1290545 4. Credit tMet of Food 6349103 8293088 8041794 9584461 t 12522559 12515673 15893711 15535595 19703644 22305843 23638455 5. BJMC borrowing (Jute) 1856499 2111395 1677290 1987770 ' 22644883 2440136 3382444 1887549 2508468 1698779 2667808 6. Credit Net of Food and Jute 4492604 6181693 6364504 7696691 * 10257676 10075537 12511267 13648046 17195176 20607064 20970647 Textile 7. Sugar 895278 1661799 1277283 1608189 i 2165125 2123377 2277115 2106665 2932503 4485495 4404240 BADC 8. Net Credit 3597326 4519894 5087221 5988502 8092551 7957160 10234152 11541381 14262673 16121569 16566407 9. Petroleum Corp. 0(?) 933 1758 . 426 4275 10. Net Credit 7781360 11498781 15694069 Credit/Deposit Ratio It. (a) Total 0.6795 0.7905 0.7476 0.7481 0.7991 0.7815 0.8164 0.7350 0.8068 0.8964 0.8023 12. (b) Minius Food and BJMC Credit 0.4808 0.5892 0.5917 0.5429 * 0.6546 0.6291 0.6427 0.6457 0.7041 0.8282 0.7176 13. (c) Minus Food, BJMC Credit, Textile Corp. Sugar Corp. BADC 0.3850 0.4308 0.4729 0.4674 * 0.5164 0.4965 0.5257 0.5460 0.5840 0.6479 0.5623 14. (d) Minus (c) sad Petroleum * 0.4859 0.5440 0.6307 * Not avallable. Source; Bangladesh Bank - 60 - performance are also influenced by priority sector lending. Small loans and rural credit are among the more important priority sectors. As already indicated they are the two sectors with the highest (close to 100%) usage (Table 18). This means that at least over the period 1979 to 1980 credit ceilings were sufficiently and consistently adjusted for such lending to make refinance an operational incentive. In the case of petroleum and jute goods government has often virtually forced banks to make loans to meet unexpected situations. In such cases both refinance and credit ceilings have been wholly or partially adjusted as an inducement. In the case of sugar mills, BADC (fertilizer) and food, a similar linking of forced lending, provision of refinance and ceiling adjustment has evolved as a general long run policy. In effect the last two are decided by tacit negotiation. In such a situation refinance policy has a highly erratic incentive effect, and implicit or explicit pressure has to be applied at various points to maintain or expand lending. A very approximate measure of the application of this policy package of forced lending, refinance and credit ceilings can be obtained by getting an estimate of the non-refinance lending or the proportions of free deposits available for refinanced sector lending. In Table 19 this is done by calculating modified credit-deposit ratios; first by subtracting food and jute credit to the public sector jute mills, and then by further subtracting credit to BADC and the textile and sugar mills corporation. This final modified credit-deposit ratio rose from .39 in June 1975 to .56 in December 1980. If we take the average of June-December figures for 1975 and 1980 the rise was from .41 to .60, an increase of 46%. This suggests that the deposits available for refinance related lending declined almost continuously over the period. Further the total credit-deposit ratio rose from a June-December - 61 - average of .74 in 1975 to one of .85 in 1980, a rise of only 15%. Both these facts suggest an increase in forced lending over the period. As export bills currently lie outside the credit ceilings, our analysis suggests that refEinance should be fully used as long as the opportunity cost of funds (s) is larger than the refinance rate (p). Though usage has averaged 77% over the period, it has risen from 55% in June 1979 to 93% in December , showing a rising trend. A possible explanation is that export bills have only recently been exempted from the credit ceilings. In the absence of a wide market for government securities, provision of refinance has been an important means of expanding the monetary base. With deposit interest rates fixed, the opportunity cost of funds (s) is greater than the deposit interest cost (i). Refinance provides a profit to the banks as long as the finance rate (p) ls less than s (assuming no excess funds i.e. no credit ceilings or adjustable ceilings). If the goverment provided general refinance this would merely transfer seigneurage from the government to the banks. In this situation, use of refinance as a means of subsidizing credit to particular sectors where lending is inefficient entails little budgetary costs, and can be an effective alternative to a policy of interest subsidy to banks. The two policies differ only in their effect on collateral requirements. An interest subsidy reduces collateral requirements while refinance may raise or lower them. Even though a rise in collateral will be compensated by a greater decrease in the interest rate borrowers with inadequate collateral will not benefit fully. Thus the poorest might not benefit from the refinance policy. This effect could be mitigated by providing a collateral subsidy for loans to the poorest borrowers. - 62 - 3.8 Guarantee Programs There are two types of guarantee programs in existence. One can be termed a liquidity guarantee and the other a loan guarantee. The liquidity guarantee is applicable to the normal lending programs of the BKB and BSBL. As already noted these two banks have been getting 100% refinace from the Bangladesh Bank for their agricultural lending programs. Moreover, this refinance is only available on their total outstanding loans net of overdues. In other words the refinance loan is made only on fresh advances. It has to be repayed in full to the Bangladesh Bank when the loans become due from final borrowers irrespective of whether they repay their loans to the BKB and BSBL. The guarantee program allows these two banks to borrow from the government, up to 30% of the value of the loans disbursed, to cover overdues and late repayments. The interesting thing is that BKB has never used this program while the BSBL has always used it (Table 20). The reason for not using is ofcourse that it involved a shift from 6% financing to 11% financing, while deposit costs are much lower than the latter. The other reason is that overdues have not been a very serious problem for BKB. In contrast, the BSBL does not take (discussed further in section on agricultural credit). The loan guarantee is applicable to all banks participating in the Special Agricultural Program (SACP) often referred to as the TK 100 crore program. This allows banks to borrow upto 30% of all (SACP) loans defaulted/overdue on the due date. 30% of all amounts recovered after due date have to be repayed. Actual payment of loan guarantees under this program have, however, been less than entitlements, due to budgetary reasons. Table 20: Payment of Givernment Guarantee to the Banks Under Special Agrscultural Credit Programme Name of the 1977-72R iQ77_7R 1978-79 _ Bar.ks Guarantee Guarantee Guarantee Guarantee Guarantee Guarantee Guarantee Guarantee Balance remain- Guarantee Guarantee -_______ claim money Paid claim m R claim money paid claim money paid ing unpaid claim money naid Sonali Bank 184.90 184.90 184.90 184.90 215.72 215.72 388.58 323.11 65.47 566.21 Janata Bank 140.83 140.83 140.83 140=83 160.95 L60.95 193.35 163.65 29.70 260.85 Agrani Bank 127.07 127.07 127.07 127.07 166.00 66.00 210.40 175.35 . 35.05 97.74 Rtupali Bank 105.87 105.87 105.87 105.87 118.15 !.18.15 131 22 1Q9.36 21 .6 115.45 Pubali Bank 111.42 111.42 111.42 111.42 86.34 .86.34 257.76 215.46 42.30 72.39 Uttara Bank 38.06 38.06 38.06 38.06 23.82 23.82 56.63 47.20 9.43 34.80 Krishi Bank 221.10 221.10 221.10 221.10 210.90 :10.90 601.00 500.87 100.13 241.22 Total 929.25 929.25 929.25 929.25 981.88 981.88 1838.94 1535.00 303.94 1388.66 Source: Bangladesb Bank - 64 - In Virmani (1982) I showed that a collateral subsidy (a lump sum or a proportion of effective collateral available from a borrower) has effects similar to an interest rate subsidy. Thus it increases loan size and lowers interest rate and collateral requirements. Therefore if inefficiencies are present, either due to information problems or due to lack of collateral with poor producer-borrowers such a subsidy will improve efficiency in a way that benefits the less wealthy. When banks are risk averse we would expect the collateral subsidy to be more effective than the interest subsidy, though this case was not explicitly analyzed. This analysis can be extended to the loan guarantee as used in Bangladesh. This essentially involves paying to the bank a fraction of the under-repayment by the borrowers. In this case too loan size would rise and loan interest rates fall. However the effect on collateral requirements is ambiguous, and consequently so is the relative benefit to the less wealthy. There is however an extra problem in the way in which the Special Agricultural Program (SACP) is implemented. This is the requirement that no collateral be taken when giving loans under this program. This restriction would tend to lower loan amounts an raise interest rates to borrowers. Many potential borrowers would be excluded. The distributional effects are not totally clear. 3.9 Urban-Rural Branch Licensing The Nationalised Commercial Banks (NCBs) were inducted into agricultural credit by the goverment for the following reasons. (a) To use the existing deposit base of NCBs to redirect credit toward agriculture - 65 - Table 21: Rural Branch Expansion (No. of Branches) 1974 1975 1976 1977 1978 1979 No. of Branches I NCB's Rural June of Year 648 743 949 1505 1822 Urban 762 833 917 1030 1144 Sonali Rural Dec. of Year 193 222 250 367 439 566 Urban 156 178 200 233 270 281 No. of New Branches during past yeat- NCB's Rural 95 206 556 317 Urban 71 84 113 114 Ratio of Rural to Urban 1.34 2.45 4.92 2.78 Sonali Rural 29 28 117 72 127 Urban 22 22 33 37 11 Ratio of Rural to Urban 1.32 1.27 3.55 1.95 11.55 NCBs: Nationalized Commercial Banks Source: Bangladesh Bank Bulletin, :Scheduled Bank Statistics and Sonali Bank Annual reports. - 66 - (b) The use of BKB would have necessiated provision of refinance, which would inevitably have been limited by overall monetary base expansion (c) The NCBs growing branch network in rural areas and their latent capacity for rapid expansion of the branch network in rural areas. In addition to the Special Agricultural Program (SACP) the linkage between the rural and urban branch licensing was a major method used. These licenses for new branches were linked in a theoretical ratio of 1 urban to 2 rural. Table 21 gives the urban and rural branches for all NCBs and for Sonali bank. Between 1976 and the ratio actually fluctuated between 1.3 and 4.9 with an average ratio of 2.7 over the entire period. There was also considerable variation among banks as is shown by the case of Sonali bank where the ratio varied between 1.3 and 11.6. For Sonali the average for the period 1976 to 1980 was 2.9. Thus the expansion was forced by the Bangladesh Bank and the government at an even faster rate than suggested by the professed ratio of 2. As there is an excess demand for urban branch licences, the idea of the linkage is of course to cross subsidize rural banking operations with urban. Table 22 gives an estimate of the administrative costs for rural branches. If the very high initial costs when volume of business is very low, were ignored, the average cost lies between 4.2% and 5.3% of the volume of business (deposits plus loans). Total administrative costs (for Sonali bank) as a percentage of business volume, are between 1.5% and 1.7% (table 23). If this is taken as the upper limit of costs for urban branches, the minimum difference in costs is between 2.7% and 3.6% (The minimum ratio of rural to urban costs is 3). - 67 - Table 22: Administrative Cost of Rural Branches Average Size Business Administrative Cost as a (Taka thousand) Percentage of Business 477 17.3' 1,028 8.4 1,623 5.3 1,945 5.0 2,509 4.8 2,991 4.8 3,733 5.3 4,765 4.2 Note: The national average for the size of business of a Sonali branch was 16,669 (Thousand Taka) in 1979 and 19,164 (Thousand Taka) in 1980; the administrative cost as a percentage of business was about 1.8% for the two years. * Computed from a sample of 10 rural branches that have been in operations since 1978 or earlier. - ** Size of Business is defined as the sum of total deposits and outstanding advances. Table 23: Sonali Bank Statistics(in 1000 Taka or p.9 as applicable) Dec.1976 Dec.1977 Dec.1978 Dec.1979 Dec.1980 Deposits 38441 46615 59950 83879 109666 Advances 24189 36248 48330 67559 96263 Total Business 52620 82863 108280 151438 205929 Interest Paid 1118.96 1479.63 2104.23 2759.46 4742.38 Expenditures 921.45 1230.99 1889.19 2387.90 3485.22 Interest Cost/Advances 4.63% 4.08% 4.35% 4.08% 4.93% Interest Costs/Deposits 2.91% 3.17% 3.51% 3.29% 4.32% Admin. Costs/Advances 3.81% 3.40% 3.91% 3.53% 3.62% 0 Admin. Costs/Total Business 1.75 1.49 1.74 1.58 1.68 Ratio of Rural to Urban Branches 1.25 1.58 1.63 2.01 2.09 Growth in Expenditures 309.54 658.20 498.71 1097.32 Rate of Growth 33.59% 53.47% 26.40% 45.45% Source: Sonali Bank Annual Reports - 69 - To calculate the approximate profitability we take all costs as a proportion of deposits. From Table 22 we can take an average cost of 4.8% which translates approximately as 9.2% of advances. We can similarly use 3.2% (1.6x2) as the corresponding figure for urban branches. Interest costs as a proportion of deposits similarly averaged 4.4% of deposits over the period 1977 to 1980. Interest ceilings were at 12% during this period. Therefore rural loses are approximately 1.6% of advances while urban profits are approximately 4.2% a ratio of 2.6. This is very similar to the ratio of new rural to urban branches for Somali. This is (merely) suggestive of the fact that not too much incentive existed and pressure had to be applied by the government. The declining ratio of total new rural to total new urban branches since 1978 is also consistent with this picture. So is the recent proposal by the NCB's suggesting Et transfer of their unprofitable rural branches to BKB. IV. ISSUES OF FINANCIAL POLICY 4.1 Efficiency or Inefficiency in Credit Markets I have shown elsewhere that when an inefficiency exists in the credit market the expected marginal product of the loan will be different from the marginal cost of funds to the banks (deposit rate i in free market) plus the marginal cost of transferring these funds from depositors to borrows (marginal administrative costs of the banks). In the case of usual concern, when loans are thought to be inadequate, the expected marginal product of loans will be greater than the marginal costs mentioned above. The important thing to note is that the relevant interest rate is not the loan interest rate (r). It: is sometimes thought that r is just the marginal costs of funds or the deposit interest rate i (or i+t) adjusted for - 70 - risk. This is not true. Both the loan rate (r) and collateral (C) take account of the risk of non-repayment, and the loan interest rate r does not bear any definite relationship to the marginal product of loan or capital even in an efficient market Virmani (1982). Given the scope of the study it was not possible to. estimate the marginal product of loans in the various subsectors where inefficiency may be a possible justification of intervention in the credit market. Some attempt was made for the agricultural sector (see section 5.2). We are therefore left to make intuitive judgement. 4.2 Welfare Objectives and Credit Markets There are two welfare issues which are directly related to the way in which credit markets function, and which can therefore legitimately be addressed through credit policy. The more important one relates to the unequal distribution of wealth or land: this can lead to an inefficient distribution of credit for productive purposes and possibly even for consumption purposes. To some extent this is not really a welfare issue but an issue of inefficiency. The less wealthy or landless producers have little collateral and can therefore run into a collateral constraint. This results in less than efficient flow of loans to them at higher interest rates. A collateral subsidy has been analyzed and suggested as the best policy intervention in this case. This is a welfare issue only in the sense that a distributional problem results in credit problems which can in turn worsen income distribution. One can of course argue that we should go to the source of the problem; the unequal distribution of land or total wealth. To the extent that it is the credit factor which adversely effects income flows and subsequent wealth distribution, however, this is the causitive factor. - 71 - The second issue is one of the distribution between lender and borrower of the gains from the credit transaction. This question has most often arisen in terms of monopolLstic money lenders. It can be shown [Virmani (1982)] that when there are no informational inefficiencies present, competitive and monopolistic credlit markets are equally efficient as far as production loans are concerned. However in the latter case the lender obtains the benefits from the credit transaction. In effect most of the productive surplus generated by the use of credit is transferred to the monopolist. In addition when information problemns exist, a monopolist will be less efficient than a competitive market, given the same information problems. In a dynamic context, however, the monopolist has a greater incentive to collect information which will eliminate the inefficiency (this case was not explicitly modeled). Nevertheless, it can be legitimate for government financial policy to attempt to promote competition and eliminate monopoly. In addition to any distributional gains this may also have possitive production effects (again in a dynamic sense) which must be weighed against the costs of intervention. It is necessary at this point to consider the faulty logic and emotionalism that characterises policy related to money lenders. Usually they are thought of as exploitative monopolists who charge very high rates of the order of 100% to 150%. If this is true, tremendous welfare gains would be made if institutional banks could make cr,edit available in these areas even at rates of 30-35%. Absolutely no case can be made in this context for putting loan interest ceilings which are even 'Lower than in competitive institutional credit markets. On the other hand if the money lender's loan rates are in the 30-35% range, these could still be due to monopoly, but can also be due to several - 72 - other factors. The most important [see Virmani (1982)] are information problems which can provide (ineffect) an informational monopoly by moneylenders in small localized regions. The effect of these is to severely fragment the credit market and impair the flow of funds within the rural areas. In such a situation it is worth considering policies which increase the incentive for moneylenders to compete in areas outside their traditional information field. Among the possible policies are increased competition from formal banks, increased supply of funds to moneylenders in deficit areas, and increased opportunities for financial investment/savings deposits in surplus areas and in the surplus (usually post-harvest) periods. The Rural Financial Experiment project in Bangladesh has used (experimental) interest rates ranging from 15 to 36%. The recovery performance has been quite satisfactory at the end of this range. This may of course be due to the tremendous subsidy provided to banks for recovering loans. On the whole, however, the evidence from this project is consistent with the above analysis. 4.3 Differential Interest Rates: Availability vs. Cost In the simplest terms the institution of interest rate ceilings on credit has been seen as a conflict between credit availability and costs. Those who get the loans are seen to have got subsidized credit, while many people end up not getting it. The actual picture is more complicated: The direct effect of (loan) interest rate ceilings will be to reduce credit to all borrowers. Indirect effects can reverse this fall for favored borrowers if ceilings are economy wide. In addition a (direct) rise in collateral requirements will result in some borrowers not getting credit at all (for the first time). - 73 - The simple picture when used for the case in which ceilings are applied only to one sector concludes that borrowers in that sector get subsidized credit. It usually neglects both the reduction in credit to individual borrowers as well as the reduction in the number of borrowers. More importantly, there will also be a shift of credit away from the chosen sector, further reducing credit. Thus instead of getting 'subsidized' credit many borrowers within the sector would get less credit. The differential credit ceilings introduced in Bangladesh in 1981, presumably to 'subsidize' certain sectors like agriculture and exports would have a perverse effect on these sectors. There is already some impressionistic evidence of an unfavorable intersectoral shift. Once this becomes clearer the government will have to choose between maintaining the have a perverse effect on these sectors. There is already some impressionistic evidence of an unfavorable intersectoral shift. Once this becomes clearer the government wiLll have to choose between maintaining the inverted structure and forcing btnks to maintain the desired portion of lending, or evening out the pattern of ceilings. I would recommend the latter solution; to be achived by a rise in the interest rates. If a subsidy is to be provided it should be an interest rate subsidy to banks for lending to favored sectors. This could be financed by a tax on all other lending. Both these would shift credit in the desired direction. 4.4 Financing of Specialized Banks: Refinance, Debentures or Deposits There are many different specialised banks, and many different ways of financing them. The three most important are Refinacing, Reserve Debentures, and Deposits. Refinancing is the primary means of financing the agricultural banks -- BKB and BSBL. A secondary and much smaller source of deposits for BKB, a scheduled bank, is bank deposits. BSBL being an apex co- - 74 - operative bank is allowed to collect deposits from members only. Reserve Debentures eligible for fulfillment of reserve requirements by commercial banks are a major source of funds for the House Building Finance Cooperative (HBFC), and also a source for the Bangladesh Shilpa Rin Sangstha (BSRS) which makes medium/longterm loans to industry. The third major source is deposits. The IRDP Co-operative system is financed indirectly by Sonali Bank deposits, while the Bangladesh Shilpa Bank (BSB) which can directly take deposits, has deposits and foreign financing as its primary source. In economic terms direct provision of refinance to a specialized bank is exactly equivalent to allowing the bank to issue debentures, and increasing the reserve requirements of deposit taking institutions to ensure absorbtion of these debentures (I assume that the public market is limited as it is in Bangladesh). This follows from the assumption that monetary expansion is equal in the two cases. In other words any practical difference arises only if the goverment is led to expand money supply by different amounts in the two cases. All other things can be made equal by an appropriate choice of the interest rate on debentures. Consider first the provision of refinance R to the specialized bank, which is then fully on-lent by the bank. If money supply is to be unchanged, the reserves of the banking system must fall by an amount R ; because the first round effect is -(l-a)R, we must have R' - R= a This can be done by increasing the reserve requirement by a proportion da - R where MB is a (l-a)MB the monetary base. If the same amount of finance is to be provided to the specialized banks through issue of debentures, the value of such debentures is equal to R. To accomodate these into reserves, reserve requirements must change by R (i.e., da - R/D) where D is the amount of deposits. The first round effect is to reduce commercial bank loans by Dda which is equal to R, so - 75 - that money supply is unchanged. Both these policies transfer loanable funds from the commerical system to the specialized banks. The required change in reserve requirements will in general be different (1 - a)MB in the first case and -D in the second. However both policies provide a cross subsidy from the commercial to the specialized bank, and the interest rate on debentures can be set to make the subsidy equivalent. Consider first the simple free deposit market case. We have shown earlier that, if transaction costs; are ignored, the effective cost of deposits to the commercial banks is given by i = i big where i is the nominal cost 1a of deposits, a the reserve requirement, and b the proportion of reserves held in a debentures earning interest ig. It is easy to show that a change in reserves will raise, leave unchanged or lower the effective cost i' as i is greater than, equal to or less than bi . In the present case b = .2 - 4 Thus setting ig around 1.25i would mean that both policies have .25 5' no effect on the commercial banks. As the current average i = 7% and the current rate on government securities is 8.5% we have big = 6.8 which shows approximate neutrality of interest with respect to reserve requirements. A provision for equal taxation of commercial banks in the two policy cases would involve setting the interest rate on debentures (igl - ig2) lower than 8.5 and i - abi 1 i-abig2 making (1-a)MB aD In the more realistic case of government fixed deposit interest rates and exogenously given deposits, the banks earn an oligopoly profit per unit of loan equal to the difference between the shadow/opportunity cost(s) and the interest cost. Loanable funds would be changed by .R in the refinance policy case, and by R in the reserve debenture policy case. Therefore in addition to the factors considered above tax equivalence must - 76 - also adjust for the different amounts of rent transferred under the two policies i.e. [ R (i-abig) (1 + (1 - a)(iabi)(1 + (1-a)2 ga1 -B) - a a) g2J aD In this exogenous deposit case, if deposits could be switched (somehow) from commercial banks to the specialised banks without changing reserve requirements, there would be no monetary affects. Only a transfer of oligopoly rent woud take place. In the Sonali Bank -- IRDP case it is not really a transfer of deposits from a bank to a specialised program, but an induced transfer of loans from one sort of borrower to another. An implicit deposit transfer mechanism is involved in BSRS financing: To an extent it was favored in the grant of licences to open urban branches, and allowed to pay higher interest rates (still less than s) to depositors. Some of the more practical considerations can now be illustrated by evaluating different means of financing the agricultural credit system (BKB, BSBL, IRDP). The main criticism of the use of refinance as a method of supplying funds to BKB and BSBL is that agricultural credit becomes too dependent on global monetary policy and inflation considerations. This is particularly true in times of high government deficits when there is internal and external pressure to reduce inflation. Our analysis shows that this is only valid if reserve requirements are held fixed. If reserve requirements are made flexible, monetary policy can be delinked from the use of refinance or the reserve-debenture policy. When macro policy considerations dictate injection of liquidity into the system refinance policy is a more direct and simple way of providing funds. The reserve debenture policy is more disruptive, requiring injection of reserves into the whole system followed by changes in reserve requirements. Until about 1979/1980 Bangladesh government deficits and public sector losses were such as to provide sufficient scope for injection of - 77 - liquidity through refinance. It was therefore the preferred policy during the period. With the recent reversal of this position pressure has grown on the BKB to depend to a greater extent on its own deposit resources. In such a situation an alternative deposit based policy can be considered. At present both BSBL and the IRDP system are permitted to take deposits only from members. Serious consideration should be given to allowing them to collect deposits from, or offer short-term financial instruments to, the rural public. This would bring them on par with BKB. BKB (and the cooperatives) is not permitted to have urban branches, while the commercial banks can open rural branches and in fact have been forced and/or encouraged to do so. One possibility would be to allow BKB to open urban branches. This would tend, however, to progressively dilute the agriculture-rural orientation of the BKB. An alternative possibility emerges from looking at Table 20 which gives the credit-deposit ratios by district. The credit deposit ratios for the rural areas have grown progressively from 32% in 1976 to 69% in 1980, but still remain below the 75% level possible without any refinancing (at the prevailing reserve ratio). Even though there is considerable variation between districts -- 27% in Chittagong to 160% in Mynemsingh -- the 69% ratio suggests that agricultural lending; could be suppported by rural deposits, without any need for external finance or refinance. Two methods for doing this are available. One method for increasing the flow of deposits to BKB, is to transfer all rural branches of the NCBs to BKB. This alternative is suggested by the expressed desire of NCBs to get rid of their many unprofitable branches. As we shall show below their agricultural lending experience has not been entirely satisfactory. In addition BKB is quite willing to take over Table 24: Credit/Deposit Ratios by District (x) FY 76 FY 77 FY 78 FY 79 FY 80 Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural A. CHITTAGONG 43 25 47 29 53 45 48 31 54 42 Chittagong 48 18 51 63 56 36 53 19 60 27 Chittagong Hill tracts 21 141 21 35 30 48 34 46 24 193 Comilla 27 47 40 63 49 74 43 46 47 83 Noakhali 37 31 53 44 61 63 55 65 65 77 Sylhet 32 14 27 19 34 32 26 23 30 27 B. DACCA 119 45 110 52 112 72 11 77 119 84 Dacca 126 41 115 45 116 55 1l1(?) 50 122 53 Faridpur 33 41 61 56 56 82 50 90 58 107 Mymensingh 31 57 40 79 56 113 55 160(?) 61 160 X Tangail 23 38 40 28 40 66 46 82 48 92 Jamalpur NA NA NA NA NA NA 101(?) 123 103 172 C. KHULNA 57 28 66 56 74 60 77 47 78 98 Barisal 20 30 29 56 36 81 33 54 40 79 Jessore 34 22 99 43 53 58 54 82 57 122 Khulna 85 25 84 60 96 59 100 65 94 94 Kushtia 28 13 45 45 49 52 53 15 74 69 Patuakhali 27 69 53 110 57 103 56 129 69 178 D. RAJSHAHI 27 33 37 50 42 76 46 82 47 92 Bogra 30 32 35 43 47 55 52 59 51 66 Dinajpur 27 58 31 97 32 143 31 128 40 135 Pabna 14 22 65 30 38 58 52 63 38 65 Rajshahi 21 19 33 91 37 52 33 57 34 67 Rangpur 42 51 56 71 58 100 68 127 50 150 Total 86 32 84 43 89 59 86 51 95 69 Source: Bangladesh Bank - 79 - these branches. If rural branches were transferred to BKB the primary responsibility for providing funds; to BSBL and IRDP would be transferred to BKB. An alternative method for linking the flow of funds to agricultural banks with rural deposit mobilization is to make it mandatory for nationalized banks to transfer all excess reserves from their rural operations to BKB, BSBL and IRDP. This transfer would be in the form of loans at rates which take account of interest payments and administrative costs. In other words BKB, BSBL and IRDP would have the right to borrow from the NCBs the difference between the loanable funds generated by rural deposits and the loans made in rural areas. This proposal is likely to be more acceptable to NCBs, but more difficult to implement because of the problem of defining rural areas and rural loans. Some mix of the two maethods may prove to be more feasible. 4.5 Buffer Stocks: Alternative for Margin Requirements and Refinance When analysing margin requirements it was suggested that they may be useful for agricultural commodities. As buffer stocks fulfill the same price stabilizaton function it is useful to briefly consider the two together. As a general principle buffer stock policy is more relevant for commodities whose prices fluctuate regularly on a seasonal and/or annual basis. Margin requirement changes are more appropriate for unexpected and temporary changes in prices. Put somewhat differently buffer stocks are more appropriate when production fluctuates from year to year or when havesting is concentrated at one point in a year, resulting in a strong seasonal fluctuation in prices (i.e., demand is uniform). Margin requirements are more appropriate when unexpected economic events lead to strong speculative demand. Buffer stocking and refinance are not as clearly separable. Buffer stocking would itself give rise to a cyclical, seasonal or annual, pattern of - 80 - demand for credit. In the seasonal case the cycle is exactly the inverse of the demand for credit for agricultural production of the corresponding commodity. Therefore if both sets of borrowers were supplied from an integrated financial system, cyclical demand fluctuation would be eliminated and there would be no need for special refinance. On the other hand annual cycles would still lead to annual fluctuation in credit demand from the stocking agency. If the buffer stocking operation is self supporting (after taking account of explicit budgetary subsidies) the banking system woud generally be willing to supply the required credit. 4.6 Term Finance In the agriculture area there is strong push for providing medium, and long term finance for irrigation improvement. In industry the specialised banks, Bangladesh Shilpa Bank (BSB) and Bangladesh Shilpa Rin Savings (BSRS), are there specially for providing term finance. I have shown elsewhere that term lending creates special problems for banks because of the difficulty of distinguishing between productive and unproductive, and honest and dishonest borrowers. Both problems have arisen in the case of term lending for industry. Industrial lenders can borrow up to 70% of the cost of a new project from BSB. Additional loans of up to 15% can be obtained from ICB and IFC. By inflating the cost of imported machinery (overinvoicing) and of land, and covering the difference from the public (shares) the promoters can end up by not using any of their own funds. Thus they may have little commitment to the success of the project, resulting in poor repayment performance. As I have shown it is necessary in such situations to develop procedures for distinguishing between productive and honest borrowers and the rest, and restricting loans to the former. A World Bank research project on Kenya is attempting to develop such procedures. Similar work has also been - 81 - done in a neighboring country. Use should be made of these and other attempts, to do the same thing in Bangladesh. Other suggestions are made in the financial review of Industrial Sector lending. V. AGRICULTURAL CREDIT 5.1 Introduction The explicit purpose of agricultural credit policy has been to promote the growth of output and the welfare of farmers. The first goal implies either that rural credit markets are inefficient or that credit is an effective means of providing a production/investment subsidy to agricultural. If rural credit markets are inefficient, then effective intervention could promote output growth by improving efficiency. On the other hand a determination of whether a production subsidy should be provided through credit requires a prior exaLmination of the incidence of taxes and subsidies on fertilizer, irrigation water, irrigation equipment and output. Once a neutral policy environment is created, it cannot be ruled out a priori that an explicit credit subsidy (not interest ceilings) may be one component of an integrated subsidy package. The second policy objective of increasing farmer welfare can relate to the entire agricultural sector, or to sub-sets of farmers, usually the poor and landless. An appropriate production subsidy package is one means of increasing farmer welfare if accompanied by measures to market the output and minimize any adverse impact on output prices. The other welfare question relates to the distribution of gains between traditional lenders and farmer borrowers. This has been addressed earlier. Returning to the question of credit market efficiency, there are two opposing views on the matter. One view is that if agricultural productivity - 82 - increases due to introduction of HYVs enough credit will be generated to meet demand. In other words the existing level of technology and non-credit related constraints-risk, availability of water, technical knowledge etc., are significantly more important factors. The other view is that rural credit markets are very fragmented, and excess funds do not flow smoothly from surplus to deficit areas. This causes inefficiencies in credit markets. 5.2 Expected Marginal Product of Loans In an earlier paper I have given a method for measuring the degree of inefficiency, if any, in the market. I have also shown that if there is any inefficiency in rural markets it is probably not due to the physical problem of moving funds. Any inefficiencies arise due to informational problems which constrain traditional lenders, and new potential lenders, from providing credit in unfamiliar areas and to unfamiliar borrowers. I have shown that if the credit market is functioning efficiently, the expected marginal product of loans must be equal to the opportunity cost of loans. The latter is the marginal cost of raising deposits plus the marginal cost of transferring them to borrowers (marginal administrative costs). The average cost of funds in 1979-80 was 4.3% of deposits (Table 17). The rural administrative cost, for commercial banks have ranged around 10.6% of deposits (from a sample of Sonali branches, 2 x 5.3). This yields an approximate average cost of 15%. Though administrative costs for the agricultural banks may be lower, a figure of 15% can be used for illustrative purposes. To obtain the marginal product of loans, we use the fact that fertilizer input and the output resulting from it are separated by the produttion period for crops. Consider a simple case in which fertilizer input is separated from agricultural output by T months. Use of fertilizer therefore requires use of own funds or credit to purchase fertilizer. Any Table 25: Marginal Product of Loans in Agriculture Marginal Product Price Fertilizer 1 2 Time Period Annualized of Fertilizer at (Adjusted) Price p 3 1 of Production Rate or Return Aug. level of use (TK/md) A 3 (months) 5 x 12/T 1 2 3 4 5 6 1. Aman 1979, Local, Broadcast 2.03 125 137.4 .846 10 1.02 HVUW, Transplant 2.46 120.9 137.4 1.165 9 1.55 2. Boro 1979-80, Local 4.53 122.65 180.5 2.078 7 3.56 HYU 2.74 109.51 180.5 .642 7 1.10 3. Aus 1980, H UW, Broadcast 2.01 93.60 182.7 .029 6 .06 4. Aman, 1980, Local Transplanted 3.91 117.38 186.6 1.460 9 1.95 Source: IFDC/BARC, Agricultural Production, Fertilizer Use and Equity Considerations, Results and Analysis of Form Survey Data, 1979/80, Bangladesh 1981. - 84 - credit taken is then repaid T periods later. It is easy to show that the T month rate of return on loans is (p. MPF/pF - 1) where p is the output price, PF the fertilizer price and MPF the marginal product of fertilizer. If fertilizer is not used efficiently because of rationing, the marginal product of loans will be less than this amount. Table 25 presents estimates for the marginal product of loans using estimates of the marginal products of fertilizer at average levels of usage. A reasonable assumption might be to assume that fertilizer usage is efficient in HYV crops, but may not be so in traditional. The second thing, to keep in mind is that high rates of return for short periods do not necessarily translate into the high rates suggested by simple annualization. This is because the opportunities available for the rest of the periods may be quite restricted due to market inefficiencies. The estimated annualized rate of return on loans for HYV crops varies from 6% for 1980 Broadcast Aus to 155% for 1979 Transplant Aman, with 1979 Boro in the middle with 110%. Given a cost of funds of 15% a rate of return of 6% would suggest an excessive use of fertilizer which is very unlikely at average levels of fertilizer usage. Some upward adjustment may also be required for the fact that 1979 and 1980 were bad crop years and what we need is an average of the marginal product over several years. As most HYV are grown in irrigated areas, fluctuations in their output is much less (flooding, can still be a problem). A 9 percentage point difference is hard to explain on this basis given the high returns from other crops in the same period. A possible explanation for the 1980 Aus and 1979 Boro is under use of fertilizer due to shortages and rationing. In such a situation the calculated rates provide an upper limit, and actual returns on loans would be less. However there is no evidence to suggest that there was a shortage - 85 - during this period. Sample testing done by BADC indicates that free market prices were not significantly diffierent from controlled prices. The rates of return calculated from traditional varieties show somewhat less variation ranging from 102% through 195% to 365%. Such high rates are somewhat unbelievable, particularly in bad agricultural years. The caution about annualizing rates is of course very relevant here, but even the unadjusted returns range from 64% t:o 146%. The tentative hypothesis about more efficient use of fertilizer in HYVs than in traditional crops, is supported by the 1979-80 Borro returns but contradicted by the 1979 Aman returns. If one could ignore the 6% return, the rest of the data become fairly consistent, indicating a minimum expected marginal product of loans of 100%. This suggests enormous inefficiency in the agricultural credit market. We cannot just ignore the anomalous observation, however, and any judgment of credit market inefficiency mtust be tempered by caution. 5.3 Targets and Disbursements: Differential Performance Almost the entire gamut of policy instruments has been applied to agricultural credits. This includes targets, refinance, guarantees, branch licensing and special programs. In addition interest rate policies, margin requirements and credit ceilings also impinge on agricultural credit. Among the earliest to be used was the target guideline approach. Table 26 presents the targets, disbursements and achievements of BKB, BSBL and the NCBs for the 1975-81 period. In 1975-76 the target for total disbursements was TK 49.56 crore while achievement was at TK 48.29 crore or 97% of the target. By 1980-81 the targeted disbursements had more than tripled to TK 3438.31 crore while disbursement had increased only to TK 279.6 crore showing a performance of Table 26: Agricultural Credit (in lakh takas) TARGET 1974-75 1975-76 1976-77 1977-78 1978-79 1979-80 1980-81 BKB 1672 2200 3635 4323 6588 10300 17986 BJSB/BSBL 1300 1550 2000 2200 2475 2900 3800 Sub-Total 2972 3750 5635 6523 9063 13200 21786 NCB's 1600 1206 2835 4714 6654 7530 13045 Total 4572 4956 8470 11237 15717 20730 34831 DISBURSEMENT BKB * * 3168 3729 5867 11896 18532 BJSB/BSBL * * 1133 1705 2010 14600 2570 Sub-Total 2605 2768 4301 5434 7877 14600 21102 NCB' 881 2061 2194 3654 5069 5536 6858 Total 3486 4829 6495 9088 12946 20136 27960 ACHIEVEWC'NT: Disb/Target X (%) BKB 87.15 86.25 89.05 115.49 103.04 BJSB 56.65 77.5 81.21 93.24 67.63 Sub-Total 87.65 73.81 76.33 83.31 86.91 110.61 96.86 NCB's 55.06 170.89 77.38 77.51 76.79 73.52 52.57 Total 76.25 97.44 76.68 80.88 82.37 97.13 80.27 Rate of Growth of Disbursement BKB 17.71 57.33 102.76 55.78 BJSB/BSBL 50.48 17.88 34.53 -4.96 Sub-Total 6.25 55.38 26.34 NCB's 133.94 6.45 66.55 38.72 9.21 23.28 Total 38.53 34.50 39.92 42.45 55.75 * Not available BKB: Bangladesh Krishi Bank BJSB/BSBL Bangladesh Jatigo Sumabaya Bank/Bangladesh Samabaya Bank Limited. Source: Annual Reports of the Bangladesh Bank - 87 - 80%, the lowest level since 1976-77. Just as targets were designed to provide an impetus to disbursement the actual disbursement performance influences the setting of targets. Nevertheless the disbursement performance has varied considerably over time and between the banks. Both BKB and BSBL showed improved performance in terms of targeted disbursement over the period 1976-77 to 1979-80, but then performance plummeted quite sharply in 1980-81 (Table 26). In contrast the commercial banks' performance shows a deteriorating trend over the period, going from 77% in 1976-77 to 53% in 1980-81. BKB performed the best in terms of targets, with a peak of 115% in 1979-80, and with 103% even in 1980-81. After the early period BSBL's performance and has remained between these two in terms of targets. Total disbursements rose at a rate of 44% a year between 1976-77 to 1980-81, from TK 64.95% crore to TK 279.6 crore. However much of this growth was due to expansion in lending by BKB which rose from TK 31.68 to TK 185.32 crore an overall compound rate of growth of 56% a year. The growth rate for the commercial banks was 33% a year, and for BSBL only 23% a year. This contrasts with the relative posiltion of the last two when compared on a target-achievement basis. It also suggests that the government and the Central Bank, perhaps, expected tWoo much from the commercial banks. The 33% growth rate includes Sonali Bank lending to the IRDP co-operative system, in which the former acts merely as a conduit. The commercial banks' performance viz a viz direct lending to fina:L borrowers may not match this growth. The BKB's performance sieems by and large to have matched expectations. Thus for example the rise in targets from TK 65.88 crore in 1978-79 to TK 103.0 crore in 1979-80, was accompanied by a rise in disbursement performance from 89% to 115% of target (A 102% growth of - 88 - disbursement). As a consequence BKB's share of total disbursements has risen from 49% in 1976-77 to 66% in 1980-81. A major policy instrument affecting this performance has been the refinance policy coupled with specialization. Refinance policy was analyzed earlier. Being a specialized agricultural bank BKB is excluded from non-agricultural lending, though trade, storage and processing activities, particularly in rural areas, come within its purview. In addition specialization has yielded information economies which resulted in better performance. This is independently indicated by its recovery performance which will be analysed below. An additional factor in past growth performance is that the BKB has not been formally subject to credit ceilings on agricultural lending (nor has the BSBL). 5.4 The Special Agricultural Program and Commercial Bank Performance Table 27 gives the targets and disbursements under the Special Agricultural Credit Program (SACP/TK 100 crore program). Disbursements increased from TK 60.78 crore during 1977 (calender year) to TK 72.68 crore in 1980-81 (financial year), a total growth of only 20%. Achievements as a proportion of targets were equally disappointing, declining progressively from 61% in the 1977 program to 36% in 1980-81 program (Table 27). For the first two years the program was on a calendar year basis, so that the data is not directly comparable to that in Table 26. If we take an average of the total disbursement for 1976-77 and 1977-78 from this table, however, we obtain an approximate figure of TK 77.92 crore for calender year 1977. Of this, disbursement under SACP 1977 which started only in February was TK 60.78 crore or 78% of the total. This gives some idea of the (initially) ambitious scope of the program and the pressure put on commercial banks to increase agricultural lending. - 89 - Table 27: Special Agricultural Program (SACP/100 crore) (In 1000 Taka) SACP Program of: 1977 1978 1979 1979-80 1980-81 Total Under SACP TARGET 10000 10000 5715 10000 20000 DISBURSEMENT 6078 5347 2370 4127 7268 ACHIEVEMENT 60.78 53.47 41.46 41.27 36.34 (Target/Disbursement) As of: Dec.1977 Dec.1978 Dec.1979 Dec.1980 Outstandings Sonali Bank SACP 1000 1659 2737 4250 Normal Program 653 971 1556 3050 IRDP 721 1084 1960 3514 Total 2374 3714 6253 10814 Agrani Bank SACP 745 1002 1302 1620 Normal 43 40 304 Total 7138 1042 1606 * NCB s Total 3938 9216 15063 10570@ * Not available @/ Provisional. - 90 - The push didn't last long however, as by the next year, SACP disbursements at TK 53.47 crore were only 49% of total (again taking 1977-78 and 1978-79 average for total disbursement). By 1980-81 SACP disbursements were only 26% of total disbursements. As one of the major purposes of the SACP was to bring NCBs into agricultural lending, we would expect its greatest impact on their agricultural lending. The data on the outstanding loans of Sonali Bank and Agrani Bank (Table 27) is consistent with this. Though the figure for outstandings includes overdues, the SACP outstandings for December 1977 will include no overdues while those for the normal program do. Leaving aside IRDP lending by Sonali, both banks had a major proportion of their disbursements channeled through the SACP (Table 27). This clearly remains true for the Agrani Banks until 1980-81; but less so for Sonali Bank, if allowance is made for the possibility of higher overdues under SACP than under the normal program. The major increase in the rate of growth of commercial banks disbursements (Table 26), from 6% in 1976-77 to 67% in 1977-78 can be largely attributed to the SACP program. The rate of growth of loan disbursements by NCBs slackened somewhat to 39% in 1978-79 and then plummeted to 9% in 1979-80. We have already shown that disbursements by all banks under SACP as a proportion of total disbursements also declined in absolute terms. Therefore part of the still fairly high growth rate of 39% in NCB loan disbursements must be due to increased lending under the normal program. This is suggested by the gathering momentum of the rural branch expansion program. The ratio of new rural to new urban branches of NCB grew from 1.3 in 1976-7 to 2.5 in 1977-8 and to a peak of 4.9 in 1978-9 (Table 21). The banks' loan disbursements under SACP continued to slacken, however, and reached 68% of first year (1977) - 91 - levels in 1979-80. The growth oil rural branches also registered lower rates in 1978-79 and 1979-80 (Table 21), suggesting a slackening of the rural program. Loan disbursements by 21CBs therefore increased only by 9% in 1979- 80. A major causative factor was the much more rigorous application of credit ceilings on commercial banks in 1.979-80. Though these ceilings had been in operation for sometime, and NCBs' agricultural lending fall within their ambit, aggregate inflationary pressures and IMF related contacts led to increasingly forceful application. 5.5 Introduction of Urban Organised Banks into Agricultural Lending Throughout this experience or experiment of getting the NCBs into agricultural lending there has been a feeling that the distribution of credit has not been satisfactory. In particular it has been felt that tenants and small farmers are not getting enough credit. These effects are predicted by my earlier analysis (competitive monopoly, Virimani (1982)), and I will briefly outline the implications of the amalysis. Given the importance of information and knowledge in the operation of credit markets, urban organized banks are specilized to operate in that environment. Basically they have developed procedures for dealing with organized and established firms. When these banks are forced into the rural- agricultural environment they are confronted with a completely different informational problem. They will therefore seek out and lend to agricultural producers who bear the greatest similarity to urban clients. These are usually the largest, most educatedl and most modern (in terms of irrigation and HYV technology). This is the group which benefits most from the introduction of Urban banks. The medium farmers would tend to get some benefit because of the erosion of monopoly profits of traditional lenders (arising mainly from informational problems). But they would continue to borrow from them. The - 92 - smallest farmers do not benefit at all because the banks have so little knowledge or competence in dealing with them that they cannot even offer any potential competition to traditional lenders. 5.6 Interest Ceilings and BKB Performance Table 28 gives the distribution of BKB loans by land holding and by size of loans. The proportion of loans going to borrowers having 12.5 acres to 33 acres of land rose sharply from 5% in 1976-77 to 11% in 1977-78. Similarly the proportion of loans of amount greater than TK 3000 increased from 12% to 40% of all loans. There was a decrease in the interest ceilings in May 1977, but this was not applicable to the BKB and BSBL which already had lower applicable ceilings. In August 1977 margin requirements of 50% were imposed on all merchandise. We have shown in the analysis on margin requirements, that for traditional borrower/merchants there would be a tendency for loan interest rates to rise and loan amounts to fall. Further some traders would now not get any loans at all. This provides an opportunity for large farmers who do not normally engage in trade or in speculative stocking to do so. Only the largest farmers would tend to have enough own fund to meet margin requirements. They would thus take larger loans using the increased part for speculative stocking. This is consistent with the shift in loans towards the largest land owners from 5% to 11% of total loans. However there is an increase in loan proportion from 28% to 39% in the 2.5 acre to 7.5 acre group. This suggests that some of the medium size land owners with surplus funds may also have engaged in speculation. Different land owners would tend to have different amounts of liquid funds available for providing margin money. Therefore we do not expect a proportional shift in the loan amounts taken. Table 28 shows that the - 93 - Tables 28: Banfgladesh Krishi Bank Borrower Characteristics I. Distribution of Loans According to Size of Holding (Excluding Tea, Cold Storage Tobacco Marketing & Processing); II. According to Size of Loans; III. According to Security Type. 1976-77 1977-78 1978-79 1979-80 1980-81 I. % of all loans for year Land holding of borrowers Landless and up to 2.5 acres 54.70 39.13 42.74 35.72 32.08 2.5 to 7.5 acres 27.58 34.19 30.26 26.05 31.33 7.5 to 12.5 acres 13.10 15.37 13.80 25.36 16.51 12.5 to 33 acres 4.62 11.31 13.20 12.87 20.08 Total 100 100 100 Cumulative % < 2.5 54.70 39.13 42.74 35.72 32.08 2.5 to 7.5 82.28 73.32 73.00 61.77 63.41 7.5 to 12.5 95.38 88.69 86.80 87.13 79.92 12.5 to 33 1001 100 100 100 100 II. % of all loans for year Loan Size < TK 1000 53.86 32.83 34.69 24.67 8.23 1000 to < 3000 34.20 26.50 26.15 18.41 27.75 3000 to < 10,000 4.07 24.41 20.67 14.53 18.92 10,000 to < 20,000 5.38 9.33 11.05 22.06 8.68 20,000 to < 50,000 0.19 2.78 4.81 16.39 13.66 > 50,000 2.30 4.15 2.63 3.94 22.76 100 100 100 100 100 Cumulative % 53.86 32.83 34.69 24.67 8.23 88.06 59.93 60.84 43.08 35.98 92.13 83.74 81.51 57.61 54.90 97.51 93.07 42.56 79.67 63.58 97.70 95.85 97.37 96.06 77.24 100 100 100 100 100 Ill. % of all loans for year Security Real Estate 33.79 53.16 53.89 38.26 45.01 Hypotecation (incl SACE') 61.20 41.86 35.34 55.86 26.04 Stored Crop Pledge 2.38 0.26 3.54 3.46 25.87 Source: BKB annual reports. - 94 - greatest shift was in the TK 3000 to less than TK 10,000 category, from 4% of total loans to 24%. In October 1980 the structure of interest ceilings was changed so that agriculture was subject to a ceiling of 12%. The general ceiling, applicable to storage, transport, trade, etc. was much higher at 15.5%. In the section on interest ceilings it was shown that there would be a tendency to shift out of agricultural lending into general lending (here it could be agriculture related). It was shown earlier that BKB's disbursement performance both in terms of growth rates and in terms of achievement of targets rose progressively from 1976-77 to 1979-80. This trend was sharply reversed in 1980-81; The growth rate declined from 103% to 58%, while performance as percentage of target declined from 115% of target to 103% of target (Table 26). This was accomplished by a shift of the loan distribution away from the smallest loans of TK 1000 and less, to those of between TK 1000 and TK 10,000. The former declined from 25% of total in 1979-80 to only 8% in 1980-81, while the later increased from 33% to 47%. 5.7 Recovery performance Many different measures of recovery performance have been used. When it is not possible to separate out loans by maturity and by initial starting time of loan, the best measure is the following: A ratio of the total repayment during the year to the value of loans plus interest falling due during that year. Another possible measure is the ratio of the repayment of principal to the principal due during the year. The latter measure may be useful when comparing different institutions and programs, with different mandated interest ceilings or other restrictions. This runs into an accounting problem, however. Banks which compound interest, transfer all overdue interest to the principle account, in effect converting it to an - 95 - overdue loan. The problem of recovery performance varying merely because of differences in rates of interest charged, may not be solved. Another problem with both these measures, when used in a comparative analysis, is that recovery performance can differ merely because of different rates of growth of disbursement. ]Non repayment at the due date does not mean that no further repayments will be made in future years. In fact repayments are often stretched out over several years. For example, many banks in Bangladesh are still receiving repayments on loans made in pre-liberation times. For this reason banks with faster growth in disbursements would tend to show poorer recovery performance. A unique set of data is available for Bangladesh which allows us to look at the pattern of loan repayment for a given loan. This relates to the SACP/100 crore program, which was started in 1977. In this program all loans had approximately the same term. Accounts were kept separately for each years' program. These were initially on a calender year basis, but in 1979 were switched to a financial year basis, with the intervening program lasting only 6 months. Data on accumulated recovery of principle is available for each bank and each program at four points of time. A recovery profile can therefore be constructed given total disbursements under each program for all banks. A linear regression of accumulated recovery on time from effective/average due date is used for this purpose. The results for the six nationalized commercial banks and BKB are presented in Figures 1, 2 and 3. The results by program are presented in Figure 4. From Figure 1 it is clear that the recovery profile of the Bangladesh Krisi Bank (BKB) lies entirely above that of others. This superiority is more noteworthy given that the BKB's recovery performance under the SACP is known (even to it) to be inferior to its recovery performance Figure 1: Recovery Profiles by Banks for the Special Agricultural Credit Program SONFLL ----------- MRAWRsN *JRNSRT -. - JAnaT - RUPVLt *---* PU-MLI 1 40.0- - 140. 0 120.0- 120.0 100.0 1 100.0 80.0 - 80.0 60.0- 60.0 40.0- - 40.0 20. 0 - / - 20. 0 0.0 20.0 0.0 10.0 20.0 30.0 40.0 50.0 Figure 2: Recovery Profiles: BKB, Sonali, Agrani and Janata Banks -------- 120.0- - 120.0 1 00/ a 10 e.o 60.0- va0,0 40.0 -u40.0 20.20.0 I o.o 10.0 20.0 30.0 40.0 50. Figure 3: Recovery Profiles: Janata, Uttarg. Rupali and Pubali Banks 1_ _ __ _ _ __0_ __ _ __ _ _ __ _ _ - 2 . UJJ.I .I 0.0 80.0 .- 80.0 60.0 ,' 60.0 0 °I~~~t- 40.0 . ' 40.0 20.0 - / -0.0 0.0 I -0.0 0.0 10.0 20.0 30.0 40.0 50.0 - 99 - under normal programs. From Figures 1 and 2 we can also conclude that Sonali bank's performance is quite close to that of BKB. It thus comes in the second position even though initial recovery is lower than for Uttara and Agrani banks. Given the differences in the intercept terms and slopes it is more difficult to rank the others. Broadly speaking however, figures 1 to 3 show that the performance of the BKB, the Sonali bank and the Uttara bank is better than that of the Janata, Rupali and Pubali banks, with Agrani bank in an intermediate position. Among the commercial banks the better performance of the Sonali Bank is probably due to its larger rural branch network at the start of the program, and its involvement with the IRDP program. The BKB-s better performance is clearly due to its deeper and longer involvement in agricultural lending. Figure 4 compares the recovery performance for the different programs for all banks taken together. The 1977 and 1978 programs are treated separately while the 1979 and 1979-80 program data are aggregated so as to have enough data points. The performance under the 1978 program was definitely worse than that of the L977 program. There have been wide-spread complaints of politicisation of the program in terms of the choice of recipients. If true, this provides an explanation of the deterioration; political pressures would take time! to build up and take effect. The 1979 and 1979-80 programs show significant improvement over the 1978 program. Their performance relative to the 1977 pxogram is more difficult to evaluate: initially lower recovery is made up in subsequent periods because of the higher marginal recovery rate. At the end of 1979-80 banks were given permission, independently and directly to select borrowers for the 1980-81 program. This depoliticisation affects the most recent programs more strongly, but would also effect recovery from older programs. This is a Figure 4: Recovery Profiler by SACP Programs ...... -------- PRO678 120.0-- 120.0 100.0 _ 100.0 80.0 80.0 40.0 40.0 20.0- - 20.0 0.0 10.0 20.0 30.0 40.0 50.0 - 101 - possible explanation for the upward movement in recovery rates for the 1979 and 1979-80 programs. Another explanation for the improvement in the 1979-80 program is the introduction of penalty rates in February 1979. Banks were allowed under this scheme to charge a penalty rate of 1% a month up to a maximum of 20%. Given the normal interest of 12%, this permitted a rise in the interest rate on overdue loans by as much as 8%. As all the banks considered, compound interest, penalty interest rates would tend to raise the average interest payable, in proportion to the delay in repayment. This would tend to increase the slope of the recovery profile line for the 1979 and subsequent programs. The application of penalty rates has however been quite variable, largely because of lack of clarity about its applicability to SACP and other agricultural lending. Only Sonali bank is known to apply the full penalty. The BKB uses a maximum penalty of 2.5%, while Agrani bank uses a penalty of only 1% for rural loans. 5.8 Bad Debts versus Liquidity Problems The existence of such a recovery time profile suggests that bad debts may not be a serious problem in agricultural lending. This is in fact what most bankers involved in agricultural lending in Bangladesh believe. It would be wrong to conclude, as some dlo, that there is no problem of bad debts ,,r non-repayment risk; this aspect is not captured well by the indices used above. The analysis shows, however, that even with the best performance, recovery of loans may normally be stretched over as much as three years. I1lerefore liquidity may become a ser3Lous problem, particularly for 1antitutions which depend on refinancing of loans (or total loans minus wverdues) such as BKB, BSBL and the lRDP system. - 102 - In the discussion about the liquidity guarantee it was noted that BSBL has always used this facility, while BKB has never used it. The previous section's analysis suggests that this may at least partly be due to the latter's superior recovery performance (BSBL is not involved in the SACP). Nevertheless, given the rapid growth of BKB lending in the past five years liquidity could become a serious problem in future. In addition the significantly lower rates of interest will no longer be a problem because BKB can now charge penalty rates on overdues. Therefore the BKB's policy on use of penalty rates may have to be reviewed to allow full use of permisable penalty rates. - 103 - Footnotes 1/ Resume of Financial Institutions, Ministry of Finance, Government of Bangladesh 1973 and 1975. 2/ See Introduction and Section 4.5 of Virmani (1982). R_ferences 1. Virmani, Arvind, The Nature of Credit Markets in Developing Countries: A Framework for Policy Analysis,World Bank Staff Working Paper No. 524, 1982. World Bank An Analysis of Developing NEW Country Adjustment Publicationis Experiences in the 1970s: Low- Compounding and Discounting of Related Income Asia Tables for Project Analysis Christine Wallich (with a Guide to Their Interest Staff Working Paper No. 487. 1981. 43 Applications) pages (including references). Second Edition, Revised aild Stock No. WP 0487. S3. Expanded Adjustment Experience and Aspects of Development Bank J. Price Gittinger Growth Prospects of the Senm- Management Project planners and analvsts will find Industrial Countries William sDiMment . this book a convenient and time-sav- Frederick Jaspersen William Diamond and ing reference for the preparation and StFfederikin Pasperseno.47 913 Raghavan- analysis of development projects. Six- Staff Working Paper No. 477. 1981. 132 Deals exclusiveIv with the manage- decimal tables for 1 percent through 50 pages (mcludtag 3 appendzxes). ment of development banks. The book percent show the compounding factor Stock No. WP 0477. $5. is divided into eight sections, each for I and for I per annum, the sinking dealing w,ith one aspect of manage- fund factor, the discount factor, the Adjustment in Low-Income ment of its problems, and of the var- present worth of an annuitv factor, Africa ious ways of dealing with them. and the capital recoverv factor. The Robert Liebenthal EDI Series in Economic Development. The first edition of this book underwent Staff Working Paper No. 486. 1981. 62 Johns Hopkins University Press, 1982. seven printings in ten vears and was pages (including bibliography). 2nd printing, 1983. 311 pages. translated into Arabic, Chinese, Stock No. WP 0486. $3. LC 81-48174. ISBN 0-8018-2571-7, Stock French, and Spanish. This new edi- No. H 271.$29.5 hrdcver;ISB 0- tion-with narrow-interval compound- Aggregate Demand and 8018-2572-5, Stock No. JH 2572, S12.95 tabes added frhier interest Macroeconomic Imbalances in paperback. rates, updated project examples, a Thailand: Simulations with the Capital Accumulation in lators to perforu n the cocputar ons SIAM 1 Model Eastern and Southern Africa: A discussed, and an annotated bibliog- Wafik Grais Decade cf Setbacks raphv increases the proven usefulness Staff Working Paper No. 448. 1981. 132 Ravi Gulhati and Gautam Datta of its predecessor, both in the class- pages (including 3 appendixes). Analyzes the magnitude of the setback room and at the project site. Stock No. WP 0448. $5. in capital accumulation in eastern and May 1984. About 208 pages. southem Africa. This phenomenon is ISBN 0-8018-2409-5. Stock No. BK 2409. examined in twentv-eight statistical ta- $10.95. NEW bles. The authors sample sixteen coun- Translations of this new edition will be tries and rely on expert observations to available in 1985. Still available are the explore the proximate causes of the following translations of the first editzon: Alternative Mechanisms for setbacks. French: Tables d'interets composes et d'ac- Financing Social Security World Bank Staff Working Paper No. 562. tualistation. Economica, 4th printing, Parthasarathi Shome and Lyn 1983. 74 pages. 1979. Squire ISBN 0-8213-0169-1. Stock No. WP 0562. ISBN 2-7178-0205-3, Stock No. IB 0542, Reviews, clarifies, and evaluates theo- S3. $6. retical literature about the effect of so- cial securitv on capital accumulation Capital Market Imperfections Spanish: Tablas de interes compuesto y de and labor supply. Analyzes empirical and Economic Development descuento para evaluaci6n de proyectos. studies using U.S. data, the impact of Vinayak V. Bhatt and Alan R. Roe Editorial Tecnos, 1973; 4th printing, 1980. pay-as-you-go financed and fully Staff Working Paper No. 338. 1979. 87 ISBN 84-309-0716-5. Stock No. IB 0526. funded social security schemes, and zaages (including footnotes). S6. characteristics of optimal social secu- Stock No. WP 0338. $3. ritv svstems. This studv provides a starting point for evervone interested in the relevance of existing theories for The Changing Nature of Export A Conceptual Approach to the financing social security in developing Finance and Its Implications Analysis of External Debt of countries. for Developing Countries the Developing Countries Staff Working Paper No. 625. 2983. 62 Albert C. Cizauskas Robert Z. Aliber pages. Staff Working Paper No. 409. 1980. 43 Staff Working Paper No. 421. 1980. 25 ISBN 0-8213-0292-2.Stock No. WP 0625. pages (including 3 annexes). pages (including appendix, references). $3. Stock No. Wr 0409. 53. Stock No. WP 0421. $3. - NEW Staff Working Paper No. 632. 1984. 144 Growth and Structural pages. Adjustment in East Asia Development Finance Stock No. WP 0632. $5. Parvez Hasan C.ompanies, State and Privately Staff Working Paper No. 529. 1982. 42 C)wned: A Review NEW pages. David L. Gordon ISBN 0-8213-0102-0. Stock No. WP 0529. An informative guide to the function Economic Liberalization and $3. and design of development finance Stabilization Policies in Interest Rate Management in companies as they are set up in devel- Argentina, Chile, and Developing Countries: Theory oping countries. Case histories high- Uruguay: Applications of the and Simulation Results for light the differences among these com- Monetary Approach to the adSmlto eut o panies-their institutional structure, Blnetof paymets South Korea management stvle, financial perfor- Balance of Payments Sweder van Wijnbergen mance, and other features. Looks at Edited by Nicolas Ardto Barletta,ee ge the problems of resource mobilization Mario I. Blejer, and Luis Landau deposit rates raise output and lower and strategies to overcome them. Twenty-eight leading intemational inflation in the short run, and increase Staff Working Paper No. 578. 1983. 84 economists and regional specialists re- growth through their favorable impact pages. view the salient characteristics of the on savings rates. It concludes that this ISBN 0-8223-0226>4. Stock No. WP 0578. monetary approach to the balance of theory depends heavilv on the as- S3. pavments, examine the vanations in sumption that portfolio shifts into time its application, and evaluate its suc- deposits come out of unproductive as- Development Prospects of cesses and failures. Emphasizes the sets, providing less intermediation Capital Surplus Oil-Exporting empirical evidence and dvnamic as- than the banking svstem. Impact of Countries: Iraq, Kuwait, Libya, pects and costs. Provides an important changes in time deposit rates on infla- Catriesa Arabia, U ibAEr examination of economic policies and tion, capital, capital accumulation and Qzatar, Saudi Arabia, UAE their effects in a region that looms medium term growth are discussed, Rudolf Habliitzel large in current deliberations about in- and empirical relevance is demon- Staff Working Paper No. 483. 1981. 53 ternational indebtedness and finance. strated through simulation runs with a pages (including statistzcal tables). June 1984. About 240 pages. macroeconomic model of South Korea. Stock No. WP 0483. S3. ISBN 0-8213-0305-8. 517.50 paperback. World Bank Staff Working Paper No. 593. Developments in and Prospects Energy Prices, Substitution, ISBN 0-8213-0288-8. Stock No. WP 0593. for the External Debt of the and Optimal Borrowing in the $3 Developing Countries: 1970-80 Short Run: An Analysis of and Beyond Adjustment in Oil-Importing International Adjustment in Nicholas C. Hope Developing Countries the 1980s Staff Working Paper No. 488. 1981. 70 Ricardo Martn and Marcelo Vijay Joshi pages (including 2 annexes, references). Selowsky Staff Working Paper No. 485. 1982. 57 Stock No. WP 0488. S3. Staff Working Paper No. 466. 1981. 77 pages. pages (including footnotes, references). ISBN 0-8213-0062-8. Stock No. 0485. S3. Stock No. WP 0466. S3. NEW NEW Exchange Rate Adjustment Domestic Resource under Generalized Currency Links between Taxes and Mobilization in Pakistan: Floating: Comparative Analysis Economic Growth: Some Selected Issues among Developing Countries Empirical Evidence Nizar Jetha, Shamshad Akhtar, Romeo M. Bautista and M. Govinda Rao Staff Working Paper No. 436. 1980. 99 Keith Marsden Fouses on the relationship between pages (including appendix). Reviews the experience with growth taxation and the three main compo- Stock No. WP 0436. S3. and taxation in twentv developing and nents of savings. Emphasizes tax re- developed countries, spanning a wide formn with a view to raising additional A General Equilibrium spectrum of incomes. Do countries revenues and encouraging household Analysis of Foreign Exchange with lower taxes experience more and business savings. Proposals for tax Shortages in a Developing rapid expansion of investment, pro- reform take account of equityvconsid- Economy ductivity, employment, and govern- erations and the need to keep tax-in- Kemal Dervis, Jaime de Melo, and sheds new light on this and other kev resources to a minimum. Highlights Sherman Robinson questions especiallv relevant to devel- appropriate policies on current ex- Staff Working Paper No. 443. 1981. 32 opment economists. It also examines penditures, subsidies, user charges, pages (including references). the mechanisms bv which fiscal poli- public enterprise pricing, self-financing Stock No. WP 0443. 53. cies mav affect growth rates. of investment bv public enterprises. Staff Working Paper No. 605. 1983. 48 Includes three annexes that examine pages. direct taxes, indirect taxes, and tax Prices subject to change without notice ISBN 0-8213-0215-9. Stock No. WP 0605. changes in Pakistan's 1983/84 budget. and may vary by country. $3. NEW The Policy Experience of Private Bank Lending to Twelve Less Developed Developing Countries Municipal Accounting for Countries, 1973-1978 Richard O'Brien Developing Countries Bela Balassa Staff Working Paper No. 482. 1981. 60 David C. Jones Staff Working Paper No. 449. 1981. 36 pages (including appendix, bibliography). This manual is based on British prac- pages (inc,'uding appendix). Stock No. WP 0482. S3. tices and terminology of municipal ac- Stock No. WP 0449. P3. counting, modified to suit the needs of Private Capital Flows to other countries, especiallv those lack- The Political Structure of the Developing Countries and ing a core of appropriately trained ac- New Protectionism Their Determinations: countants. Provides the basic princi- Doulas R. Nelson Historical Perspective, Recent ples of municipal accounting tor those g * v acn with little or no bookkeeping experi- Staff Working Paper No. 471. 1981. 57 Experience, and Future ence and proceeds through successive pages (including references). Prospects levels of difficulty to some of the most Stock No. WP 0471. 53. Alex Fleming advanced concepts currently in use, Staff Working Paper No. 484. 1981. 41 including the pooling of loans. An im- pages. portant feature is the multitude of practical applications and examples of NEW Stock No. WP 0484. 53. forms and records. Private Direct Foreign A joint publication of the Chartered Price Distortions and Growth Investment in Developing Institute of Public Finance and Ac- in Developing Countries Countries countanc and the World Bank. Ramgopal Agarwala K. Billerbeck and Y. Yasugi June 1984. About 900 pages. Sixteen informative tables trace the Staff Working Paper No. 348. 1979, 101 ISBN 0-8213-0350-3. Stock No. BK 0350. distortion in Prices of foreign exchange pages (including 2 annexes). S30. and other factors affecting the growth Stock No. WP 0348. 55. of developing countries. Based on sta- The Nature of Credit Markets tistics frorn thirtv-one developing in Developing Countries: A countnes. NEW Framework for Policy Analvsis Staff Working Paper No. 575. 1983. 78 Arvind Virmani pages. Savings Mobilization through Staff Working Paper No. 524. 1982. 204 ISBN 0-8213-0242-6. Stock No. WP 0575. Social Securitv: The Case of pages. 53. Chile, 1916-1977 ISBN 0-8213-0019-9. Stock No WP 0524. Christine Wallich S5. Pricing "Policy for Development Describes the savings mobilization po- The Newly Industrializing Manage:ment tential in Chile and in five Asian pro- Developing Countries after the Gerald MI. Meier grams. Some sort of social security Oil Crisis Presupposing no formal training in program functions in almost all devel- economics, it explains the essential oping countries. Programs are often Bela Balassa elements of a price svstem, the func- costly, whether measured in relation Staff Working Paper No. 437. 1980. 57 dons of prices, the various poiicies to GNP, government expenditure, pages (including appendix). that a government might pursue in government revenue, or the wage bill. Stock No. WP 0437. S3. cases of market failure, and the princi- This paper compares the successful ples of public pricing of goods and systems. Notes on the Analysis of services provided by government en- Staff Working Paper No. 553. 1983. 109 Capital Flows to Developing terprises. ;It also provides the would-be pages. Nations and the "Recycling" practitioner with an appreciation of the ISBN 0-8213-0123-3. Stock No. WP 0553. underlying logical structure of cost- Problem benefit project appraisal. To give sub- Ralph C. Bryant stance to the applied and policv di Short-Run Macro-Economic Staff Working Paper No. 476. 1981. 67 mensions, manv of the readings are A P i pages. drawn from the experience of develop- Adjustment Polices m South Stock No. WP 0476. S3. ment praclitioners and relate to such Korea: A Quantitative Analysis important sectors as agriculture, in- Sweder van Wijnbergen Notes on the Mechanics of dustrv, power, urban services, foreign Staff Working Paper No. 510. 1981. 182 Growth and Debt trade, and emplovment. The principles pages (including 3 appendixes). Benjamin B. King outlined are therefore relevant to a ISBN 0-8213-0000-8. Stock No. WP 0510. A Practical model to explore the ay host of development problems. in which cpai- The Johns Hopkins University Press. 1983. in which capital inflow from abroad af- 2772 pages (including bibliography and in- fects economic growth. dex). The Johns Hopkins University Press, 1968. LC 82-716. ISBN 0-8018-2803-1, Stock 69 pages (including 4 annexes). No. TH 2803, 535 hardcover; ISBN 0- LC 68-8701. ISBN 0-8018-0338-1, Stock 8018-2804-X, Stock No. JH 2804, 512.95 Prices subject to change without notice No. IH 0338. 55 paverback. paperback. and may vary by country. State Finances in India A three-volume set of papers that ex- plores a range of issues relating to the nature of intergovernmental fiscal rela- Th p suc f W tions in India. The pnmary source for World DebtTables Vol. 1: Revenue Sharing in India medium- and long-term E , ,' c . Christine Wallich Viol. rI: India-Studies in State Fi- extemal debt of mnany nalnces developing countries. = Christine Wallich Vol. III: The Measurement of Tax Ef- Suhas Ketkar, Asia-Pacific fcrt of State Governments, 1973-1976 Economist and Vice President, Raja J. Chelliah and Narain Sinha Marine Midland Bank, N.A. Staff Working Paper No. 523. 1982. vol. 1, 85 pages, vol. 1I, 186 pages, vol. 11i, 85 pages. ScOften the only reliable ISB1N 0-8213-0013-X. vol. I, Stock No..ir WVP 1523, $3, vol. II, Stock No. WP source of information for 2523, $5, vol. Ili, Stock no. WP 3523, S3. countries for which data is hard to come by... Used quantitatively for macroeconomnic detail as well as qualitatively in Structural Adjustment Policies reports discussing the debt picture. I find the in Developing Economies projected servicing payments a strong feature.09 Bela Balassa Belaf WorkingsPaper No. 464. 1981. 36 Jonathan Kayes, International Staff Working Paper No. 464. 1982. 36 Economist, Republic Pages. National Bank of New York Stock No. WP 0464. 53. Structural Aspects of Turkish World Debt Tables, 1983-84 Edition Inflation: 1950-1979 The World Bank's invaluable reference Also available for the first time M. Ataman Aksov guide to the external debt of develop- S Staff Working Paper No. 540. 1982. 118 ing countries. Essential planning tool ummary Report pages. for economists, bankers, countrv risk Debt and the Developing ISBN 0-8213-0098-9. Stock No. WP 0540. analysts, financial consultants and all World Current Trends 55. thos~e interested in the global system Wol:CretTnd of trade and pavments. Provides data and Prospects on the external debt of 103 developing Includes an overview and summary ta- Thailand: An Analysis of countries augmented bv information, bles from the 1983-84 edition. Structural and Non-Structural where available, on major economic 1984. 64 pages. Adjustments aggregates plus indicators used to ana- Ame Drud, Wafik Grais, and lyze debt and creditworthiness. Shows Stock No. BK 0319, $6.50. IDusan Vujovic statistical tables by country, including Companion computerized data figures for extemnal public debt out- Staff Working Paper No. 513. 1982. 93 standing, commitments, disburse- base pages (including appendix). ments, service payments, and net bor- Includes all debt information given in .'SBN 0-8213-0023-7. Stock No. WP 0513. rowings. Reports on private the unabridged volume, and, where '$3. nonguaranteed debt of 19 countries. available, offers continuous historical Gives aggregate position of 13 major series for 1970-82 and projected debt- borrowers-countries with disbursed service payments for 1983-92. Write for Trends in Rural Savings and and outstanding medium- and long- sample purchase agreement. Private Capital Formation in term total debt in excess of $13.5 bil- (9-track, phase-encoded, recording India lion at the end of 1982. Includes peri- densitv 1600 bpi) Raj Krishna and G.S. odic supplements as fresh data are re- Stock No. IB 0500, $5,000 (service bur- Raychaudhuri eaus for reselling to their clients); Stock World Bank Staff Working Paper No. 382. 1984. 328 pages. No. IB 0667, $2,000 (banks and commer- 1980. 43 pages (including 2 tables, 3 ap- Stock No. BK 0315 $75 (annual subscrip- cial corporations); Stock No. IB 0666, $500 pendixes, references). tion) (universities and libraries). Stock No. WP 0382. $3. E The World Bank Publications Order Form SEND TO: YOUR LOCAL DISTRIBUTOR OR TO WORLD BANK PUBLICATIONS (See the other side of this form.) P.O. BOX 37525 WASHINGTON, D.C. 20013 U.S.A. Date Name Ship to: {Enter if different from purchaser) Title Name Firm Title Address Firm City State Postal Code__ Address Country Telephone ) City State_ Postal Code Purchaser Reference No. _ Country Telephone Check your method of payment. Enclosed is my W Check [3] International Money Order 3 Unesco Coupons D International Postal Coupon. Make payable to World Bank Publications for U. S. dollars unless you are ordering from your local distributor. Charge my O VISA 7 MasterCard 7 American Express G Choice. (Credit cards accepted only for orders addressed to World Bank Publications.) Credit Card Account Number Expiration Date Signature OInvoice me and please reference my Purchase Order No. Please ship me the items listed below. Customer Internal Stock Number Author/ Title Routing Code Quantity Unit Price Total Amount S H _ =~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ _____________ All prices subject to change. Prices may vary by country. Allow 6-8 weeks for delivery. Subtotal Cost $ Thtal copies _ Air mail surcharge if desired ($2.00 each) $ Postage and handling for more than two complimentary items ($2.00 each) S Total S IBRD0053 Thank you for your order. Distributors of World Attn: Mr. Giancarlo Bigazzi Attn: Mr. J.M. Hernandez Via Lamarrnora A; Castello 37 Bank Publications 50121 Madrid ARGENTINA Florence SRI LANKA AND THE MALDIVES Carlos Hirsch, SRL, JAPAN Lake House Bookshop Attn: Ms. Monica Bustos Eastern Book Service Attn: Mr. Victor Walatara Florida 165 4' piso Attn: Mr. Terumasa Hirano 41 Wad Ramanayake Mawatha Galeria Guemes 37-3, Hongo 3-Chome, Bunkyo-ku 113 Colombo 2 Buenos Aires 1307 Tokyo SWEDEN AUSTRALIA, PAPUA NEW GUINEA, KENYA ABCE Fritzes Kungl, Hovbokhandel FIJI, SOLOMON ISLANDS,Africa Book Services (E.A.) Ltd. Attn: Mr. Eide Segerback FIJI, SOLOMON ISLANDS, Attn: Mr. M.B. Dar Regen 1g Ee2, Box 16356 VANUATU P.O. Box 45245 S-13 Stockholm The Australian Financial Review Nairobi SWITZERLAND Information Service (AFRIS) KOREA, REPUBLIC OF Librairie Payot Attn: Mr. David Jamieson Pan Korea Book Corporation Attn: Mr. Henri de Perrot 235-243 Jones Street Attn: Mr. Yoon-Sun Kim 6, rue Grenus Broadway P.O. Box 101, Kwanghwamun 1211 Geneva Sydney, NSW 20001 Seoul TANZANIA BELGIUM MALAYSIA Oxford Universitv Press Publications des Nations Unies University of Malaya Cooperative Attn: Mr. Anthony Theobold Attn: Mr. Jean de Lannoy Bookshop Ltd. Maktaba Road, P.O. Box 5299 av. du Roi 202 Attn: Mr. Mohammed Fahim Htj Dar es Salaam 1060 Brussels Yacob THAILAND CANADA P.O. Box 1127, Jalan Pantai Baru Central Department Store, Head Office Le Diffuseur Kuala Lumpur Attn: Mrs. Ratana Attn: Mrs. Suzanne Vermette MEXICO 306 Silom Road C.P. 85, Boucherville J4B 5E6 INFOTEC Bangkok Quebec Attn: Mr. Jorge Cepeda Thailand Management Association San Lorenzo 15311, Col. del Valle, Attn: Mrs. Sunan LCiObrSeTArieAs Deleg. Benito Juarez 308 Silom Road Attn: Mr. Hugo Chamberlain 031 Mexico, D.F. Bangkok Calle 11-13, Av. Fernandez Guell MIDDLE EAST TUNISIA San Jose Middle East Marketing Research Societe Tunisienne de Diffusion Bureau Attn: Mr. Slaheddine Ben Hamida DENMARK Attn: Mr. George Vassilou 5 Avenue de Carthage Attn: Mr. Wilfried Roloff Makarios l Avenue Tunis Rosendems Alle 11 Nacosia TURKEY DK-1970 Copenhagen V. Cvprus Haset Kitapevi A.S. EGYPT, Arab Republic of (granch offices in Bahrain, Greece, Attn: Mr. Izzet Izerel Al Ahram Morocco, Kuwait, United Arab 469, Istiklal Caddesi Al Galaa Street Emirates. Jordan) Bevoglu-Istanbul Cairo NETHERLANDS UNITED KINGDOM AND MBE BV NORTHERN IRELAND FINLAND Attn: Mr. Gerhard van Bussell Microinfo Ltd. Akateeminen Kir;akauppa Noorderwal 38, Attn: Mr. Roy'Selwyn Attn: Mr. Kari Litmanen 7241 BL Lochem Newman Lane, P.O. Box 3 Keskuskatu 1, SF-00100 NORWAY Alton, Hampshire GU34 2PG Helsinki 10 Johan Grundt Tanum A.S. England FRANCE Attn: Ms. Randi Mikkelborg UNITED STATES World Bank Publications P.O. Box 1177 Sentrum The World Bank Book Store 66, avenue d'Ina Oslo 1 600 19th Street, N.W. 75116 Paris PANAMA Washington, D.C. 20433 GERMANY, Federal Republic of Ediciones Libreria Cultural Panamena (Postal address: P.O. Box 37525 UNO-Verlag Attn: Mr. Luis Fernandez Fraguela R. Washington, D.C. 20013, U.S.A.) Attn: Mr. Joachim Krause Av. 7, Espana 16 Baker and Taylor Company Simrockstrasse 23 Panama Zone 1 501 South Gladiola Avenue D-5300 Bonn I PHILIPPINES Momence, Illinois, 60954 HONG KONG, MACAU National Book Store 380 Edison Way Asia 2000 Ltd. Attn: Mrs. Socorro C. Ramos Reno, Nevada, 89564 Attn: Ms. Gretchen Wearing Smith 701 Rizal Avenue 50 Kirby Avenue 6 Fl., 146 Prince Edward Road Manila Somerville, New Jersey, 08876 Kowloon PORTUGAL Commerce, Georgia 30599 INDIA Livraria Portugal UBS Publishers' Distributors Ltd. Rua Atonio Alves Martins 9730-E George Palmer H'ifhway Attn: Mr. D.P. Veer 120Lanham, Maryland, 2076 New Delhi 110002 Lisbon Blackwell North America, Inc. (Branch offices in Bombav, Bangalore, SAUDI ARABIA 1001 Fries Mill Road Kanpur, Calcutta, and NMadras) Jarir Book Store Blackwood, New Jersey 08012 INDONESIA Attn: Mr. Akram Al-Agil Sidnev Kramer Books Pt. Indira Limited' P.O. Box 3196 1722 P Street, N.W. Attn: Mr. Bambang Wahvudi Riyadh Washington, D.C. 20006 )I, Dr. Sam Ratulanigi No~. 37 SINGAPORE, TAIWAN, BURMA United Nations Bookshop Jakarta Pusat Information Publications Private, Ltd. United Nations Plaza IRELAND Attn: Ms. Janet David New York, N.Y. 10017 TDC Publishers 02-06 1st Floor, Pei-Fu Industrial VENEZUELA Attn: Mr. James Booth Building 12 North Frederick Street 24 New Industrial Road Libreria del Este Dublin 1 Singapore Attn. Mr. Juan Pericas Avda Francisco de Miranda, no. 52 ITALY SPAIN Edificio Galipan, Aptdo. 60.337 Licosa Commissionaria Sansoni SPA Mundi-Prensa Libros. S.A. Caracas 1060-A The World Bank Headquarters European Office Tokyo Office U I818 H Street, N.W. 66, avenue d'lena Kokusai Building Washington, D.C. 20433, U.S.A. 75116 ]'aris, France 1-1 Marunouchi 3-chome Telephone: (202) 477-1234 Telephone: (1) 723-54.21 Chiyoda-ku, Tokyo TOO, Japan Telex: WUI 64145 WORLDBANK Telex: 842-620628 Telephone: (03) 214-5001 RCA 248423 WORLDBK Telex: 781-26838 Cable Address: INTBAFRAD WASHINGTONDC ISSN 0253-2115/ISBN 0-8213-0429-1