India - Agricultural credit projects : a review of recent experience

Report No. 3415 FILE Copy Agricultural Credit Projects: A Review of Recent Experience in India April 8, 1981 Operations Evaluation Department FOR OFFICIAL USE ONLY Document of the World Bank This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank aluthorization. FOR OFFICIAL USE ONLY THE WORLD BANK Washington, DC 20433 U S A. Office of Director-General Operations Evaluation April 8, 1981 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Agricultural Credit Project: A Review of Recent Experience in India Attached, for information, is a copy of a report entitled "Agricul- tural Credit Projects: A Review of Recent Experience in India" prepared by the Operations Evaluation Department. Attachment This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. FOR OFFICIAL USE ONLY AGRICULTURAL CREDIT PROJECTS: A REVIEW OF RECENT EXPERIENCE IN INDIA TABLE OF CONTENTS Page No. Preface Summary ............................................................. I. INTRODUCTION ............................................. II. PROJECT HISTORY .......................................... 3 III. THE ISSUES ............................................... 5 A. The Lending Approach ................................ 5 1. The "Line of Credit" Approach Versus the "State Credit" Approach .................... 5 2. The Role of the Land Development Banks .......... 8 a. Volume of Lending and Trends ............... 9 b. Sources of Financing .................... 10 c. Overdues .................................... 11 d. Integration of Short- and Long-term Cooperative Credit Systems ................ 12 3. The Role of the Commercial Banks ............... 14 4. Instrument for Refinancing ..................... 14 a. Issuance of Debentures by LDBs .............. 14 b. Eligibility Criteria for Refinancing ....... 17 5. Costs and Spreads .............................. 19 6. Revolving IDA Funds ............................ 20 B. The Investments Financed ........ne.....d.2.... ..... 21 1. Groundwater Development ........................ 21 a. Groundwater Development Control ............. 21 b. Incomplete Well Investments ................ 27 c. Well Failure and Remedies ................... 28 2. Farm Mechanization ........ ..............o..... 29 a. The Justification of a Farm Mechanization Component .................. 29 b. GOI's Mechanization Policy .................. 34 c. Procurement ................................ 35 IV. THE HIGH RATES OF RETURN ................................ 37 A. Reasons for High Project Profitability .............. 38 B. Methodological Problems in Estimating the Projects' Benefits ....... ni4............... 40 C. Conclusion ................. ........................ 41 This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without IFC authorization. TABLE OF CONTENTS (Cont'd) -2- Page No. Annex I: Comments on Methodological Approach in Project Benefit Calculation A. Fungibility of Project-provided Inputs ............... 43 B. Limits of Comparing Project Participants and "Control" Group Farmers ............................ 44 C. The Problem of Attribution ........................... 46 D. The Farm Model Approach .............................. 47 Tables 1 - 3 AGRICULTURAL CREDIT PROJECTS: A REVIEW OF RECENT EXPERIENCE IN INDIA PREFACE This report presents an overview of lending for agriculture in India through nine state projects and a subsequent nation-wide credit program. The basis for the report is a total cf 10 Projects Performance Audit Reports (PPARs) issued between October 1976 and July 1980. OED issued an earlier review of agricultural credit programs in 1976 (Report No. 1357). That review covered agricultural programs in five coun- tries and 20 projects. It included a discussion of the credit institutions involved and the benefits achieved under those programs. The overview presented in this report calls attention to issues as- sociated with agricultural credit in India, particularly the lending approach, relevance and benefits of the main investments financed (groundwater develop- ment and mechanization), the institutional achievements, and the methodology for economic analysis of agricultural credit projects in general. In respect of many of these issues the experience reviewed here also has wider implica- tions for Bank-financed agricultural credit operations generally and the conditions that might attend their success. AGRICULTURAL CREDIT PROJECTS: A REVIEW OF RECENT EXPERIENCE IN INDIA SUMMARY i. A 1968 IDA/FAO review of Indian agricultural credit institutions and on-farm investments required resulted in IDA support for ten agricultural credit projects. The projects were approved in 1970-73 and all but one completed in 1975-771/. They were state-based and designed to support lending programs of cooperative and commercial banks to finance investments in, mainly, minor irrigation and farm mechanization. Experience with these projects was, on the whole, highly satisfactory. IDA and the Government of India agreed to move into a nation-wide credit project and in 1975 a first line of credit for a two-year period was approved to support the lending program of the Agricultural Refinance and Development Corporation (ARDC). Experience with this nation-wide credit project was also good, with some excellent institutional and other effects. Three nation-wide agricultural credit projects have been approved since. Project Performance and Audit Reports (PPARs) for nine state-wide projects and the first nation-wide project form the basis for this report. ii. The change to the nation-wide lending approach was made in the belief that poor performance of the credit cooperatives required reforms that could best be met by a line-of-credit through ARDC. Consolidation of finan- cial criteria would result in overall improvement. The change in the lending approach was facilitated by the improved competence of ARDC and the partici- pating banks in scheme appraisal, a shift in participating banks' lending policies from security orientation to development orientation, and the intro- duction of special development debentures as a new way of financing. The nation-wide approach as compared with the state-wide approach has the advan- tages of more flexible lending to states, greater coverage of schemes by type and by location, increased control of a strengthened ARDC over state-by-state lending, and savings of supervisory manpower for IDA. The disadvantages are that lending terms and conditions are standard for all states and that direct IDA leverage at state level is smaller. Nevertheless, in all the states visited, government and banking officials unanimously stated that the nation- wide approach had no negative consequences and states without previous credit had in fact benefitted from it. The change in the lending approach, there- fore, appears to have been correct and timely. iii. Almost all long-term credit was, at the time of appraisal, provided through land development banks (LDBs), which were, therefore, selected as the main channel to on-lend project funds to farmers. At IDA request commercial 1/ Bihar Agricultural Credit Project was completed in 1979 and a Project Completion Report (PCR) is currently under preparation by the Borrower. - 11 - banks (CBs) were included as an alternative channel. While agricultural credit in India expanded overall, between 1974 and 1978, LDB's share declined mainly due to their limited capacity to handle increased volumes of loans, their high overdues causing ineligibility for refinance, and continued lending for investments with relatively slow demand. CB's share, on the other hand, increased due to their recruitment of agricultural staff to process operations in this sector and to Government measures encouraging their lending for agriculture. iv. Overdues have been the most critical problem of participating banks, both LDBs and CBs. Causes are many: deficiencies in lending procedures, effects of inclement weather, inadequate short-term financing available to farmers and political pressure in various forms. States where loan recovery has been good, show strong bank management, state government support of loan recovery efforts, and minimal political interference in banking operations. v. Meeting farmers' overall credit needs has been a matter of concern, with short- and medium-term credit provided by banks other than LDBs. LDBs provide only long-term credit. A gradual merger of the two systems is under consideration of the authorities in India. vi. Since special debentures became the main instrument of transfering IDA funds through ARDC to participating banks, technical control by ARDC of the banks became more effective. These developments have, however, also resulted in IDA funds substituting for local capital which could have been raised by ordinary debentures. The debenture issue and the procedures for raising resources are being reconsidered. vii. Margins between borrowing and re-lending interest rates quoted by participating banks as necessary to cover their operating costs were about 4%. The margin of 3% appears insufficient.1/ The second nation-wide credit project (ARDC II) included a study on interest rate spread, which should help determine the appropriate levels. viii. The main investment financed under the projects reviewed was ground- water development. To ensure optimum use of available resources, control on their development is necessary. In the absence of groundwater legislation, the projects included temporary control measures for maximum well densities. However, since the first borrowers in any area are usually the richer and more agressive, the conditions imposed under the projects may lead to inequities of rural income distribution. Experience under these projects indicates that by deferring establishment of effective controls - which would protect the rights of landowners to use a proportionate share of the sustainable supply of an aquifer below their fields - a definite solution has become more difficult. ix. In one project, one third of the loans made for wells remained undisbursed two years after project completion. It indicated inadequate LDB loan supervision, which has since been strengthened. Estimates of well failures vary, but appear to be declining as a result of groundwater investi- gations. 1/ Interest margins were raised to 3.75-3.85% in October 1980 and are above the 3.5% recommendation of the Committee on Interest Spreads. - iii - x. Farm mechanization (mainly tractorization) components in these projects have been a contested issue. Evidence from the projects concerned suggests that private returns on tractor ownership are strongly positive in high-wage areas such as Punjab and Haryana, or areas where reclamation possi- bilities still exist, as in Gujarat. In southern states, tractor owners interviewed saw as main benefit the ability to manage large (16-20 ha) hold- ings without depending on burdensome permanent labor. Social returns of tractorization vary. Tenant displacement was not evident in Punjab and Haryana where as a result of rapid economic development labor has become increasingly scarce. Tractorization there caused - per cultivated ha - a reduction in the use of family labor, a slight increase in the use of perma- nent and total labor, and a strong increase in the use of casual labor. Combined with the additional off-farm employment created by tractorization, the projects in Punjab and Haryana overall created more employment than they displaced. In Karnataka and the southern states where the stage of economic development anA labor wages are lower while unemployment is higher, tractori- zation had at best no or even a negative effect on employment. In Gujarat, the impact of tractors on employment is not clear. xi. The six projects which included farm mechanization components encountered major problems in procuring the required tractors. The organiza- tional difficulties of procuring, within two or three years, tens of thousands of tractors through dozens of financial institutions and hundreds of dealers, were grossly underestimated. A complicating factor was the differing objec- tives held by the Government of India and IDA. The Government wanted tractor procurement to be consistent with its objective of developing the Indian tractor industry and insisted on commonality of procured tractors with those manufactured in the country. Eventually, the Government dropped this require- ment when IDA agreed to also finance domestically manufactured tractors. As a result of the procurement problems, the mechanization components experienced substantial delays. xii. The tractor components benefitted, on average, medium and large, rather than small farmers. These components were, however, not designed to reach small farmers and Government and IDA policies towards mechanization only changed during project implementation in favor of generating more employment and reaching smaller farmers. The components did not serve many small farmers directly, but performed much better in this regard than expected. It should be noted that there was a significant emphasis on lending to small farmers on the Bank's insistence. Increasing percentages of ARDC lending for small farmers meeting specific income criteria thrust increasingly large proportions of agricultural lending to a segment of the farming population whose ability to raise private resources was limited. xiii. The overall economic returns for this set of projects are high; they can be estimated between 35% and 40%. This is well above economic returns of agricultural projects in other countries. The reason for this success lies in the surrounding environment, with proven technologies and a well-developed credit system already in place; the projects acted as catalysts - Iv - providing the right inputs for the earlier introduced, highly profitable, technical packages. Application of these packages required both water and, where needed, power. Both were provided under the projects. Furthermore, while input prices increased substantially during implementation, prices for most agricultural commodities went up even more and were above appraisal forecasts. xiv. The high economic rates of return as quoted in the completion and audit reports, however, somewhat overestimate the real economic returns. Reasons for this are largely methodological. First, fungibility of money makes it difficult to determine the additionality of project funds to funds already available for financing tractors, wells and pumpsets. Many project beneficiaries, in the absence of IDA credits, might have acquired them from existing funds.!/ Only to the extent that project: funds were additional to existing funds, can project benefits legitimately be attributed to the project. It is, however, impossible to be precise about or to quantify the substitution. Second, comparability between beneficiaries and non-benefi- ciaries is often limited, contributing to the margin of imprecision in calcu- lating incremental project benefits. Third, it is difficult to clearly separate the factors external to the project from benefits attributable to credit operations: a fact that affects, in varying measure, the direct attribution of benefits in some other types of projects also. Finally, Bank methodology of computing economic returns on credit projects neither takes account of incomplete or failed investments by some borrowers, nor of adminis- trative costs of lending. xv. In summary, while the above factors contribute to a built-in opti- mism in evaluating the results of agricultural credit projects, there is sufficient evidence to conclude that this set of projects was very timely, generally received strong institutional support, and achieved an impressive measure of success. They have wider significance for agricultural credit projects in general for their timing, their sector context, the rapid insti- tutional evolution that took place in the course of their implementation, the generally positive and flexible interaction between Bank and Borrower, and even in respect of the important issues which surfaced in the projects. They also establish the special advantages of decentralized, incremental projects built and operated independently within an overall, conjunctive use of surface and groundwater development for irrigation. 1/ Lending by commercial banks to agriculture expanded considerably during the past years. AGRICULTURAL CREDIT PROJECTS: A REVIEW OF RECENT EXPERIENCE IN INDIA I. INTRODUCTION 1.01 In December 1968, an FAO/Bank Cooperative Program reconnaissance mission reviewed agricultural credit institutions and on-farm investment requirements in selected States in India to identify high priority agricul- tural credit projects. Ten agricultural credit projects, each in a different Indian State, were eventually identified: Gujarat, Punjab, Andhra Pradesh, Haryana, Tamil Nadu, Karnataka (then Mysore), Maharashtra, Madhya Pradesh, Uttar Pradesh, and Bihar!. The States were selected on the basis of their agricultural potential and credit requirements, as well as on the degree of the institutional development of their credit and supporting facilities. 1.02 The State-based agricultural credit projects were designed mainly to support three - and four-year lending programs by cooperative and commercial banks to finance investments in minor irrigation (about 80% of the total program), land levelling, and, in six states, farm mechanization. Nearly all projects required State governments concerned to establish or strengthen agencies responsible for monitoring groundwater development. Each project required participating banks to adopt specific loan appraisal procedures based on incremental income calculation, to observe uniform lending terms and conditions, and, usually, to improve their staffing and management methods. 1.03 Approval of the ten credits stretched from May 1970 (Gujarat, Cr. 191-IN) till October 1973 (Bihar, Cr. 440-IN). Total IDA lending amounted to US$287.9 million, in support of investment estimated at appraisal to cost US$515.1 million. The first nine projects were closed between 1975 and 1978; they have been evaluated and their project performance audit reports submitted to the Board2/. The Bihar project was closed in March 1980; its project completion report is currently under preparation. 1.04 Experience gained under the State-based projects was satisfactory and the Bank and the Government of India agreed to larger undertakings. However, instead of continuing with projects to cover the remaining Indian States and starting a series of follow-on projects for States already covered, it was decided to move into a nationwide, all-India credit project, which 1/ An eleventh project (in West Bengal) was identified and financed later. However, this is considered a rural development project, with an agricul- tural credit program administered by ARDC as the main component. Its credit component is somewhat similar to the ten projects reviewed here. However, the project is different enough to have not been included in this Overview. 2/ OED reports Nos. 949, 1303, 1386, 1514, 1640, 1658, 1921, 2100 and 2551. - d- - would take care of the credit demands in all the States. In April 1975, a general line of credit was approved to cover a two year lending program on the general understanding that it would be the first in a series of such credits. Eventually, this was followed by a second and third all-India projects; a fourth is currently under preparation. Credit amounts involved have progres- sively increased: $75 million for the first; $200 million for the second, and $250 million for the third project. 1.05 The first all-India project was closed in December 1977. It was evaluated and its project performance audit report submitted to the Board in 19791/. The second was closed in December 1979 and the project completion report is currently under preparation. 1.06 Although the executing agencies for State-based projects were the Land Development Banks (LDBs) and, in most states, commercial banks (CBs), all schemes-V to be financed had to be eventually approved by the Agricultural Refinance and Development Corporation (ARDC) and all funds had to be chan- neled through it. ARDC grew and performed satisfactorily while implementing these projects and it was a correct decision to make it executing agency for the new line of credit, covering the all-India projects. 1.07 Main objective of this overview is to discuss the major issues which arise from this ten-year lending experience. Some of the discussion is already found in the audit reports; other subjects are new. 1.08 The report starts with a summarized history of developments under the projects reviewed. It then goes on to discuss various issues. The most important are: (i) The overall lending approach: whether it was appro- priate to start on a state-oriented basis and then to switch to an all-India approach; advantages and disadvantages of both approaches, as well as the timing of the change over, are discussed. (ii) As mentioned above, about 80% of the overall credit funds have been allocated to minor irrigation develop- ment, raising questions on groundwater development control and the equity implication of the approach adopted under the projects. (iii) Six of the projects financed tractors and other farm machinery in a country where unem- ployment and underemployment is a major problem. The relevance and benefits of farm mechanization are discussed. Procurement of tractors, which proved to be one of the most contentious issues throughout,, is also commented upon. (iv) The projects' rates of return are in the 35-40% range, well above the average for other agricultural credit projects evaluated by OED, and much above the average of other agricultural development projects. It seems relevant to look for the reasons. 1/ OED report No. 2702. 2/ Schemes are mostly area development programs for a particular type of investment on a large number of farms in a compact area. II. PROJECT HISTORY 2.01 In 1968, an IDA mission with assistance from the FAO/IBRD Coopera- tive Program, reviewed agricultural credit institutions and on-farm investment requirements in several Indian States and identified high priority agricul- tural credit projects. Punjab, together with Gujarat, was given priority for an agricultural credit project because of the States' substantial potential for agricultural development. The Gujarat Agricultural Credit Project was appraised in June and July 1969 and the Punjab Agricultural Credit Project in November and December 1969. The Andhra Pradesh (April 1970), Tamil Nadu (September 1970), Haryana (November 1970), and the Karnataka Agricultural Credit Project (March 1971) followed. Finally, Maharashtra was appraised in April 1971; followed by Madhya Pradesh in March 1972; Uttar Pradesh in Novem- ber 1972, and Bihar in early 1973. 2.02 These projects thus constituted a set of ten similarly designed state-oriented credit projects aimed at increasing agricultural production in India. 2.03 The appraisal estimates of the projects- total costs reached US- $515.1 million. The projects were mainly considered as development programs to assist farmers in minor irrigation improvements (mainly tubewells and pumpsets)!/ on the one hand and, on the other, for farm mechanization (mainly tractors).2. Out of the estimated total investment costs IDA was expected to finance US$287.9 million. Physical completion of these projects ranged from March 1976 to December 1977. 2.04 Funding was to be channeled through the Agricultural Refinance and Development Corporation (ARDC), the State Cooperative Land Development Bank system (LDB) and participating commercial banks (CBs). Procurement for the small and labor intensive credit supported operations was to follow local procedures. Drilling and earth-moving equipment was to be procured through international competitive bidding. The State groundwater agencies were to be responsible for overviewing groundwater development and, in some States, determining the relevant spacing and density requirements. 2.05 By and large, the refinancing and on-lending procedures worked as anticipated, although commercial bank participation was smaller than expected. 2.06 Most of the projects' objectives were broadly achieved. Production increased, though less than anticipated (due to a smaller increase in cropped area in some parts, and to yields lower than appraisal estimates). Farm 1/ In all projects except Punjab. 2/ In Gujarat, Andhra Pradesh, Tamil Nadu, Punjab, Haryana and Karnataka. employment expanded significantly. Investments implemented under the pro- jects were financially and economically viable. Rates of return are similar to or above appraisal estimates; increase in paddy prices contributed to this result. Under the projects, incremental income criteria for sub-loan appraisal were introduced. ARDC expanded and improved its refinancing opera- tions, opening the way to the current all-India line-of-credit approach in IDA funding for agricultural credit in India. The State groundwater agencies established under or supported by the projects developed into large and com- petent groundwater research, development and control organizations. Regula- tion of groundwater development by administrative/lending procedures was only partly successful. 2.07 The Agricultural Refinance and Development Corporation first Credit Project (ARDC I) was the first project to cover the entire country; as such it supplemented the State-oriented credit projects by providing credit in areas not covered by them, and, as funds provided under those projects were exhausted, it took over the financing of unfinished schemes. The project was appraised in February/March 1974. The appraisal estimate of project costs was $168.5 million. The IDA credit amounted to US$75 million and financed about 44% of the project cost. The remaining part was financed by ARDC (36%), participating banks and state governments (10% each). 2.08 The project's primary objectives were similar to those of the State- oriented credit projects, namely; (a) increase production and employment through on-farm investments with particular emphasis on small farmers and less developed states; (b) improve institutional credit channels; and (c) strengthen supporting services, especially groundwater development authori- ties. The objectives were broadly achieved. 2.09 ARDC performance under the project was good and it coped efficiently with the additional responsibility allotted to it. The standard of scheme appraisal and supervision was satisfactory. Studies relating to the integra- tion of cooperative credit systems and training requirements were carried out efficiently. Training targets were largely achieved. Commercial bank (CB) participation exceeded appraisal projections. 2.10 Investments financed under the project were financially and econo- mically viable. Rate of return estimates were largely based on results from field surveys of similar investments which were conducted under the State- oriented credit projects. - 5 - III. THE ISSUES A. The Lending Approach 1. The "Line of Credit" Approach Versus the "State Credit" Approach 3.01 Up to 1975, all IDA agricultural credit lending to India was through individual state projects. The main issue discussed during preparation and appraisal of the ARDC credit project was whether the time was opportune to embark on a "line of credit operation", using ARDC as financial intermediary, in view of the critically poor and deteriorating performance of the credit cooperatives. IDA and GO eventually agreed that an all-India line of credit was appropriate, that the identified needs for major reforms of the agricul- tural credit sector could be better met by such a credit and that a consolida- tion of sector policies, control and financial criteria would result in overall improvement. 3.02 The new system was based on the close relationship which developed between IDA and ARDC during the execution of the state-oriented credit proj- ects (and other projects as well), and on IDA's growing confidence in ARDC's capacity and ability. IDA has delegated to ARDC its former appraisal function and is satisfied with the way ARDC handles the approval of overall lending programs and their subsequent supervision. This approach is similar to the way the Bank handles DFC operations, and it may be advantageous to compare current systems of appraisal and supervision for agricultural credit projects with those for DFCs. 3.03 The main advantages of the all-India projects, as compared with State-oriented projects, are: (a) Greater coverage. They cover the whole country and any worthwhile scheme conforming with IDA-approved criteria can be included. This permits a more comprehensive coverage than in previous times when many schemes too small to justify individual IDA appraisal had to be excluded. Another advantage is that the time from identification to implementation is substantially shortened. (b) Strengthening of ARDC. The great measure of independent decision making allowed to ARDC under the project increased its influence in the various States, and, at the same time, contributed to further strengthening of ARDC. (c) Manpower savings. The "line of credit" type of project has resulted in considerable manpower savings to IDA in regard to project preparation, appraisal, and supervision. (d) Flexible targets. Under the all-India projects, indicative targets were set for each State. However, these were not binding and short- falls in lending activities in certain States could be compensated - 6 - by increased- lending in others. Under the State-oriented projects, such reallocation of funds was impossible. The States were under pressure to complete the envisaged financial programs even if demand and/or capability of lending institutions should be less than originally estimated because otherwise India stood to lose part of approved IDA funds. This could conceivably have led to reduced standards of loan appraisals, or to financing of investments which did not conform with appraisal criteria, as well as to retroactively financing some investments. 3.04 The main disadvantages of the "line of credit" approach are: (a) The terms and conditions of loans are standardized for the country as a whole and possibilities of taking into account special condi- tions and requirements of each State are limited. In particu- lar, this constraint adversely affects the less developed States, drought-prone areas, tribal areas and other areas where special circumstances cannot be adequately addressed by a standardized framework. (b) Smaller leverage at State level. In the State-oriented credit projects, the respective State government and the SLDB concluded separate Credit Agreements with IDA; these incorporated State specific covenants such as establishement of a groundwater board, strengthening of LDB staff, etc. Such State-IDA agreements are not feasible under the line of credit approach; the only possible State-specific action would be that IDA insist on specific tasks to be undertaken in a State; otherwise, this State would be excluded from on-lending. 3.05 However, these two disadvantages are more apparent than real. Actually, possibilities for IDA and GOI taking specific local issues into consideration were not changed by the change from state-wide projects to ARDC credits. Such possibilities had already declined under the State projects. Efforts to standardize terms and conditions of project lending had already been initiated under these and would have most certainly continued had follow- on state projects been carried out. Similarly, control of IDA over factors affecting project implementation has not been affected by the supersession. The only change is procedural: previously IDA had to negotiate with State Governments and had to obtain ARDC's (and sometimes RBI's or GOIs) agreement; now it negotiates with ARDC (and sometimes with RBI or GOI) which is respon- sible for obtaining the State Governments' agreements. 3.06 One additional aspect has to be discussed before reaching an overall assessment: whether the individual Indian States are now in a better or worse position. For this purpose, a distinction needs to be made between (a) states which previously had received an IDA credit, and (b) those which had been without it. - 7 - 3.07 (a) States With Previous Credit. In all the states visited, government and LDB officials unanimously stated that changing to the line of credit projects had had no negative consequences, nor had their states been discrimi- nated against in allocating the all-India project funds. Furthermore, those states with better organized and more aggressive LDBs, with credit discipline, have been able to obtain an increasingly larger share of ARDC refinancing resources. A case in point is Andhra Pradesh: with the second best recovery rate in India, it advanced each year, from eighth place among the states in 1975/76, to fourth in 1976/77, and to second by 1977/781/. Similarly, Tamil Nadu, where loan recoveries and credit discipline seriously deteriorated over the period slipped from ninth place in both 1975/76 and 1976/77, to twelfth in 1977/78 . Such developments are in line with IDA's and ARDC's institu- tion building objectives: offering incentives for good results and penalizing bad performance. But agricultural development in the states where LDBs do not perform well will suffer unless CBs are able to cover fully the deficiency in lending. There is clearly a trade-off between institution building and physical development objectives. Some states may discover that they did lose financial resources in the change towards the all-India projects3/. (b) States Without Previous Credit. These states have benefitted from the change; some of them are small and certainly would not have been considered for an individual state project. 3.08 If the all-India, line-of-credit approach seems to be better than the state-by-state project approach, what was the rationale for the commence- ment of lending for agricultural credit in India on a state-by-state basis? Why did the extension of such credit not begin with ARDC? 1/ Ranking according to refinance drawn from ARDC. 2/ Tamil Nadu also illustrates the flexibility provided by the line-of- credit approach. As its groundwater potential nears full exploitation, less additional funds are required each year. The same applies to Haryana, which, although showing the best recovery rate in the country, dropped from fifth to sixth place and then to tenth in the same years mentioned above. 3/ The change per se cannot carry the full responsibility for this develop- ment; certainly some of the states which receive a smaller amount of funds now would probably not have had a follow-on state credit project at all. Further, if it were realized that some states were not receiving a proper share of the total resources made available to India under the lines of credit, IDA could negotiate with GOI for the establishment of a threshold or a ceiling on lending to some or all states under follow-on projects. - 8 - 3.09 In the early 70s, three conditions which eventually ensured the success of the line-of-credit approach were not yet secured: (i) the on- lending approach, involving the identification and preparation of schemes by the land development and commercial banks to be appraised and approved by ARDC, and the floating of a new kind of financial instrument (special development debentures) as a way of canvassing funds from ARDC to the land development banks, had not yet been developed and tested; (ii) the land development and the commercial banks had yet to change their policies and procedures towards a development-oriented (in contrast to emphasizing secu- rity) approach to lending; further, most needed strengthening and further institutional development. Most states also lacked, or had only very weak, groundwater development agencies; and (iii) although a major agency, ARDC was still far from demonstrating its current appraisal and evaluation skills; it also required strengthening financially. All three elements developed suc- cessfully throughout the implementation period of the state-wide projects. On-lending procedures were eventually established and tested, although some SLDBs resisted the change from ordinary debentures (which were cheaper and move familiar) to the' special debentures (see paras. 3.32 - 3.35). SLDBs and commercial banks were strengthened in part by the sheer volume of incremental lending, and in part by the introduction of new procedures and staff training. ARDC expanded financiallyl/. 3.10 In conclusion, it was correct to start agricultural lending to a country like India using a pilot approach by coverirLg states individually. Once methods, procedures and agencies had been tested, developed and strength- ened, it became a natural progression to use the nation-wide approach, taking advantage of the above-mentioned benefits. The timing of the change was also appropriate. 2. The Role of the Land Development Banks 3.11 At appraisal, the sole source providing substantial long term credit was the land development bank system (LDB). This was a two-tiered system, with a state land development bank (SLDB) at the apex in each state, and either independent primary land development banks (PLDBs) or branches of the SLDB at district or taluk2! level. It was, therefore, natural that LDBs should be selected as the main channel for on-lending project funds to farm- ers. At the Bank's insistence, commercial banks (CBs) which, at the time, did not conduct term lending, were also included as alternative channels; their importance has expanded throughout the period. 1/ Total funds mobilized increased more than six times between 1970-71 and 1977-78 and, although it still experiences some management information and data processing problems, ARDC's appraisal and evaluation capabili- ties have expanded substantially. ARDC is now progressively, though cau- tiously, delegating powers to regional offices. 2/ Subdistrict. - 9- (a) Volume of Lending and Trends 3.12 LDB lending operations can be viewed over three distinct periods. In the three years preceding IDA involvement - 1968/71 - the lending volume (commitments) in both real and nominal terms was higher than in the following period 1971/75, when state-wide projects were financed by IDA. The first ARDC operation, approved in April, 1975, was followed by an improvement (1975/78) in lending volume from the preceding period. Whether this improvement repre- sents a reversal of the downward trend is too early to say. LDB lending operations expanded in nominal terms by 8.5% between 1974/75 and 1975/76 and by 21.4% the following year, and in real terms by 1.4% and 18.6%, respec- tively. The current levels are, however, still substantially below the pre-1971 level (see Table 3)1/. And, in terms of the overall expansion of agricultural credit in India, LDB's share in direct finance for agriculture between 1974 and 1978 - the period of fastest growth in LDB lending - declined from 38% to 28%2. 3.13 The main contributing causes to LDB's reduced lending volume appear to be: (a) limitations in LDBs' physical capacity to handle the substantial volume of loans; (b) the eligibility formula (see paras. 3.37 to 3.44), whereby increases in overdues restrict LDBs' sources of funds; (c) minor irrigation has remained the main lending sector, which shows slower growth in some states, especially in Tamil Nadu3/; (d) farmers find alternative sources of funds more easily than in the past, including those from CBs which expanded 1/ This analysis is based on figures provided in the June 1978, ARDC/GOI Preparation Report for a proposed ARDC III project, and other information provided by ARDC. Comments sent by GOI on a draft of the audit report on ARDC I contain figures which are different from those in the ARDC III preparation report. Since the former were officially provided by GOI to IDA to support the request for IDA funds, and were reproduced in the SAR the Bank staff submitted to IDA's Board of Executive Directors to support the President's recommendation for Credit 947, they have been retained as the formal basis for analyzing LDB lending operations. However, even the figures provided by GOI confirm the decline of LDB lending in real terms (see Table 3, footnote 3/). 2/ The figures for total institutional term credit, a comparator suggested by GOI, are not readily available. Even if they were the comparison would become almost tautological, for LDBs provide most agricultural term credit in India. 3/ There is still a great potential for groundwater irrigation development in many other states. The current rigid spacing and density restric- tions imposed to prevent water overexploitation are likely to be made more flexible and better adapted to local conditions as SGDs acquire more knowledge about the available groundwater resources. - 10 - considerably during this period (see para. 3.30) and traditional sources, and have more funds of their own available for investments!'; (e) little lending for land development has taken place, apparently because of unresolved prob- lems in land consolidation, delays in completing irrigation infrastructure, and underestimation of how much work farmers can carry out themselves without borrowing. Further, many farmers remain unconvinced about benefits of land development; (f) LDB's diversified lending, an alternative to compensate for the slowdown in demand for minor irrigation loans, has not developed signifi- cantly. This may be traced to statutory limitations, and lack of experience and aggressiveness on the part of LDBs. 3.14 For the six-year implementation period of the state-wide and ARDC I projects, 1971/72 to 1977/78, the increase in absolute value of loans out- standing was Rs 5,900 million (see Table 3). This increase is quite high, taking into account the total amount of loans advanced during the period (Rs 13,583 million) and their average duration (about 8 years). Even if the longer maturity of loans extended since 1973 is taken into account, this increase still reflects the declining overall performance of LDBs in terms of loan recoveries, lending and overdues. Prior to the IDA credits, however, there was a sharp decline each year in the ratio of loans advanced to loans outstandingi' from 0.37 to 0.21; the decline continued until 1973/74 when the ratio was as low as 0.16. Thereafter it has fluctuated around 0.20. 3.15 The audit was not able to gather any evidence that technical condi- tions for lending imposed by ARDC on LDBs under IDA projects, a common source of complaint by most of the SLDBs and PLDBs visited, have slowed either demand or approvals. (b) Sources of Financing 3.16 During the period 1971 - 1978, ARDC disbursements to LDBs have grown far more rapidly than loans advanced, showing increasing financial support by ARDC and a shift from financing through ordinary debentures (RBI) to special debentures (ARDC) (see Table 3). The percentage of ARDC and IDA disbursements on loans advanced for ARDC finance rose from less than 20% to an average of almost 45%, and for IDA finance from nil to about 33%. More detailed data on Maharashtra, Tamil Nadu, Karnataka and Andhra Pradesh LDBs show deviations compared with all-India data, but on the whole confirm this trend. 1/ Some studies have shown that a significant amount of investments, par- ticularly minor irrigation investments, have been made without recourse to institutional sources of credit. 2/ Table 3, ratio A/B, which is indicative of the turnover of the loan portfolio. - 11 - 3.17 Since present LDB lending in real terms is below the levels prior to 1971 - the year of first IDA involvement - (see para. 3.13 and Table 3), ARDC and IDA funds have substituted for local resources previously available to LDBsl/ 2/. ARDC is replacing, to a large extent, a system of finance which had worked as a channel of funds in the past. It should be noted that this is done because new developmental objectives could no longer justify continuation of the old system. The new system entails a more technical approach to lending, better leverage, and more responsibilities for ARDC. (c) Overdues 3.18 The most critical problem LDBs face is the high level of overdues. Since bad debts have never been written-off, some refer to old operations. Except for a small improvement between 1975 and 1976, however, the all-India position has worsened in spite of pressure from both IDA and ARDC on the LDBs to increase recovery efforts. Only in three states (Haryana, Punjab and Andhra Pradesh) have overdues been kept at a very low level. Three other states (Kerala, Uttar Pradesh and West Bengal) have a 75% recovery rate, which the Bank has deemed acceptable for a country like India. 3.19 The causes of overdues are multiple: (a) deficiencies in lending policies and procedures: poor appraisal and supervision (solved, more recently to a large extent, through the IDA-supported projects), poor mechanism for recoveries, poor organization and shortage of experienced staff, inappropriate use of certain techniques (such as rescheduling) in dealing with defaults3/. (b) weather vagaries, natural calamities and adverse economic conditions; (c) inadequate short-term financing available to borrowers; (d) inadequate returns to service simultaneously loans and daily living expenses, a feature most apparent in the case of small farmers; (e) farmers- efforts to keep the money on hand, as there is still a considerable credit shortage; (f) campaigns promoting non-payment by farmers (these affect mostly, but not exclusively, debts to Government or Government-sponsored bodies), and f) political inter- ference in varied forms: (i) pressure to advance loans to certain categories of borrowers, and, in some cases, without consideration of the economics of 1/ Ordinary debentures at maturities of 10 to 15 years subscribed by the Life Insurance Corporation of India, RBI, CBs, state government, SLDB, and central and other cooperative banks in the state. 2/ The Region states that India-s system of financial planning is so highly developed and centralized that there is almost complete fungibility of resources including ODA, such as Bank Group funds. ARDC disbursements reached Rs 4.12 billion in 1979/80 an increase of 1300% compared with only 300% for IDA funds during the 1970-80 time period. The audit has no difference with this overall judgement; its concern is limited to the role of LDBs. 3/ Rescheduling is regarded as a common banking practice undertaken on a case-by-case basis. In India, however, rescheduling by LDB, has been done en bloc, covering all borrowers within areas declared as affected, because a case-by-case investigation is difficult, time-consuming and costly. - 12 - the operation; (ii) orders given to LDBs and other cooperatives by officials not to take coercive action or press for repayment; (iii) misleading declara- tions regarding terms and conditions of loans, etc. 3.20 Project analysis usually emphasizes incremental returns and profit- ability, the farmers- repayment capacity and the risks involved being over- looked. Satisfactory financial rates of return on investments are necessary for farmers to repay promptly. They are not, however, always sufficient, especially where small farmers are concerned. Absolute returns on minor investments are bound to be small. LDB officers frequently stated that small farmers use incremental returns first to cover additional living expenses and to repay current debts, and second to loan service to institutional credit agencies. Some studies, however, have shown that small farmers are no worse than others at repaying medium- and long-term debts. 3.21 There are considerable differences between the States in their LDBs' recovery performance. States with good recovery performance (e.g., Haryana, Andhra Pradesh) show strong LDB management, State government support of loan recovery efforts, and minimal political interference in LDB banking operations. 3.22 ARDC faces no risk from LDBs with poor collection records since State governments guarantee the redemption of LDB debentures. Further: (a) SLDBs have never defaulted on either ordinary or special debentures;1 (b) dues on special debentures are not yet significant;- (c) the cushion of SLDB and PLDB resources is not yet exhausted-1; (d) although it is against their statutory rules, SLDBs can delay without sanction investment in sinking funds for the repayment of ordinary debentures (this issue is dis- cussed below; see para. 3.33); and (e) an increasingly large portion of ARDC refinance is channeled to CBs, whose financial situation is stronger than the LDBs'. 3.23 Commercial banks are also facing high overdues, but agricultural activities constitute only a small portion of their portfolio. In the early days, high overdues originated mainly because CBs were unfamiliar with agri- cultural credit operations; in particular they underestimated the risks of lending to farmers. ARDC has set up a Committee on Agricultural Loans to CBs which is expected to improve CBs- recovery performance. (d) Integration of Short- and Long-term Cooperative Credit Systems 3.24 Meeting farmers' overall credit requirement has been a matter of concern for IDA ever since the state-oriented agricultural credit projects were identified. The problem arises with the existence of a dual system of 1/ The Region found that a certain deterioration of this situation has taken place during the months following the audits. 2/ It should be borne in mind, however, that a portion of PLDBs' own re- sources is invested in SLDB share capital. In turn SLDBs (and State Cooperative Banks, SCBs as well) are holding a significant portion of ARDC share capital (for both SLDBs and SCBs it is about 24% as of June 30, 1978). Thus, the system's "own resources" are interwoven; if indi- vidual figures were consolidated they would show that actual, realizable rARntirrP.Q nrp miih lpq. th;an whnt- thpv nnpnr tn hp- - 13 - cooperative agricultural credit, with the cooperative banking system (SCB) providing short- and medium-term credit (together with various banking facili- ties), and the land development banks (LDB) providing long-term credit. 3.25 The disadvantages of this dual system, primarily the inconvenience for borrowers in applying to different institutions, and having to produce the same information twice to satisfy administrative requirements, have been recognized for a long time. When, however, the objectives of the long-term cooperative credit system became development-oriented, instead of being aimed primarily at refinancing old debts and financing land purchase through mort- gage loans, those disadvantages became more apparent. There was no certainty that borrowers who obtained long-term credit were also able to secure the short-term borrowers who obtained long-term credit were also able to secure the short-term loans they needed for seasonal inputs. This would jeopardize a significant part of the benefits that would otherwise accrue from their investments. 3.26 The problems of the two systems have been complicated by their loose relationship. For instance, little or no cooperation exists regarding exchange of information on borrowers and borrower dues, and no attempt has been made to establish common services to save manpower and costs. 3.27 A study of the advisability and feasibility of integrating the two systems was agreed upon at negotiations of ARDC I and undertaken by the Hazari Committee. Its report, published in 1978, recommended a gradual merger of the two systems at all levels but neither GOI nor RBI have endorsed either the committee-s approach or conclusions.I/ 3.28 Since full-scale integration seems to be strongly opposed at pre- sent, 2/ some cooperation and coordination of activities on the part of 1/ Under ARDC III, GOI and RBI would review progress towards integration of the cooperative structure in all States and, by December 31, 1980, make proposals as to the timing of future steps. 2/ As an example, the status and prospects of integration in three states (Andhra Pradesh, Tamil Nadu and Maharashtra) at audit were as follows. There is a division of opinion within GOTN and GOM regarding the advan- tages and disadvantages of integration, and decisions are not expected in the near future. In Andhra Pradesh, the SLDB has a considerable number of non-viable sub-branches. An RBI Study conducted in 1975 estimated that only 54% of SLDB sub-branches were viable, about 30% were potentially so and 16% were non-viable. As the short-term credit system had, and still has, a large proportion of non-viable primary cooperative credit societies (PCCS), the situation of both credit systems could be improved through a partial or full integration. The GOAP is planning an experiment involving the partial merger of four selected PLDBs with their more numerous PCCS counterparts. This experiment, which amounts to a sub-contracting arrangement, aims at using the PCCS' village level facilities together with PLDBs' specialized areas of expertise. A decision on starting such merger is yet to be adopted. - 14 - the two systems should be arranged, particularly in respect of exchange of credit information and for sharing common services at various levels. Alter- natively, LDBs could provide short-term finance for working capital as an integral part of any investment loan. 3. The Role of the Commercial Banksl/ 3.29 Commercial bank (CB) involvement in agricultural lending (and agricultural term loans, in particular) is still comparatively recent and such loans constitute only a modest proportion of their total business. CB lending to the agricultural sector increased from 5% of total operations in 1969 to nearly 11% in 1978; nearly half are seasonal loans. 3.30 CB involvement with ARDC increased sharply in the last five years, from Rs45 million refinanced by ARDC in 1973 (less than 5%), to Rs 1,200 million (over 50%) in 1978. This change is due to a number of factors: the relative decline of LDB lending (see paras. 3.11 to 3.13); the recruitment of agricultural staff in CB branches, and the GOI and RBI policy measures to encourage CB agricultural lending. Further, a number of obsolete legal and procedural restrictions, originally aimed at protecting cooperative banks from competition by CBs, have now been largely or fully removed in most states: exemption of stamp duties, simplification of mortgage registration procedures, and removal of cooperative bank priority over secured assets. 4. Instruments for Refinancing (a) Issuance of Debentures by LDBs 3.31 LDBs obtain funds through issue of ordinary debentures, controlled by RBI, and of special development debentures, controlled by ARDC. The former carry no particular restrictions and LDBs can on-lend freely within statutory limits. In contrast, IDA funds, transferred to LDBs through the special debentures, can be on-lent only for purposes stipulated in the relevant IDA credit agreements. In addition, lending must follow technical procedures required and controlled by ARDC. 3.32 Therefore, from the LDBs- perspective, ordinary debentures are a much better source of funds. In addition to the larger operational freedom, funds obtained by LDBs through floating of ordinary debentures were, until 1/ There is only a limited discussion in the audit reports of the role played by commercial banks in implementing this set of projects. The best can be found in the PCR for ARDC I, but, in quantitative terms, it refers only to that particular project. Therefore, this section relies heavily on the appraisal report of the Third ARDC credit project (Report No. 2404a-IN of June 21, 1979), which repeats the discussion in the PCR on ARDC I, but in a wider perspective. - 15 - quite recently, cheaper and more convenient than IDA funds. When the IDA credit projects were started, the margins allowed for LDBs were 2.5% for special debentures and 2.7% for ordinary debentures; these increased as rates of interest were raised. Over recent years, ordinary debentures would provide LDBs with a spread of about 4.25% on minor irrigation loans and 4.5% on diversified lendings, compared with a uniform spread of 3% under the special debentures. 3.33 Further, repayment terms under IDA-supported projects were intended to match those allowed to the ultimate borrowers, implying annual repayments by SLDBs to ARDC. Repayment of ordinary debentures to RBI, however, is made in a lump sum after 10 or 15 years. To generate such sum most loan collections are allocated to a sinking fund of GOI securities, LDB debentures and other securities. At the beginning of project implementation, sinking fund arrangements provided little financial gain for LDBs. Over the years, however, as returns on investments increased, income from sinking funds has made a significant contribution to LDB revenues. Currently, the amount of investments range from one third to more than half the amounts of loans outstanding for most LDBs, and resulting profits are substantial!/. More- over, contributions to the sinking funds can be treated more flexibly than annual payments2/. Some LDBs were, therefore, hesitant to borrow from ARDC under special debentures. 3.34 ARDC, with implicit IDA support, has been trying to reduce the importance of ordinary debentures as a source of funds for LDBs as well as to equalize the rules and procedures for issuing both kinds of debentures. Initially, some restrictive measures were established - for example, not allowing the refinancing of minor irrigation works through ordinary debentures in those areas where ARDC-financed schemes were operational. In accordance with the credit agreement for ARDC I, an effort was made towards the develop- ment of consistent (and if possible, similar) standards by RBI and ARDC in the floating of debentures by LDBs and their purchase by RBI and ARDC. Similar criteria for the issuance of all debentures were eventually adopted. The 1/ Funds "deposited" in the sinking fund were earning from 6% up to 9% - 9.5% at the time of audit, eventually to redeem debentures floated at 4% to 6% interest rates. Since almost no cost is incurred in handling investments and the risk is minimal, the profit is evident. 2/ Although the cooperative laws and by-laws of the LDBs' require that sink- ing fund requirements are to be met, if necessary, from their own funds, RBI has reported several cases of LDBs not fulfilling their commitments to sinking funds on account of poor recoveries. (Report of the Committee on Integration; see para. 3.28) In some rare cases, LDBs were required to invoke the State government s guarantee to eventually redeem ordinary debentures for which sinking fund provisions proved inadequate. - 16 - debenture eligibility formula (see below paras. 3.38 to 3.46) was extended to ordinary debentures in 1977, and rates of interest: for special debentures became almost similar to those for ordinary debentures. These developments have had diverse effects. On the positive side, a more technical approach to long-term lending has been established through introducing regulations on subproject analysis and supervision. From these, a more substantial impact will result on production, employment and income. Also, disbursements under the all-India projects (ARDC I and II) have moved faster and more easily. Since special debentures became the main source of funds for LDBs (see Table 3), ARDC's technical control on LDB lending operations has become more effec- tive. On the negative side, these developments have resulted in IDA funds substituting for local capital which would have been raised by ordinary debentures, thus reducing the overall additional impact that IDA funds were intended to have on Indian agricultural investments!!. 3.35 The credit agreement for ARDC I stipulated also that a study be made on requirements for the issuance of LDB debentures as well as on the most efficient ways of improving the debenture system. The study was to provide the basis for formulating common requirements for the issuance of all LDB debentures. This obligation was waived by IDA in July 1976 in the light of the developments described above. The problems of the issue of debentures have not yet been fully resolved, however: (a) the rationale of the debenture system remains questionable; as a case in point, ARDC is utilizing loans to transfer IDA funds to the CBs; (b) there are lengthy procedures for estab- lishing the suitability of land for collateral to cover the ordinary as well as the special debentures, which in turn may delay the issue of debentures and call for costly bridging finance; (c) although with the growing share of special debentures in LDB lending operations (see previous paragraph) ordinary debentures have become of less significance, their use carries two other problems: (i) the sinking fund established for their redemption blocks funds which could otherwise be devoted to lending2_/, and (ii) the different timing 1/ There are conflicting data on the extent of additionality under the proj- ects. Total resources mobilized by ARDC increased sharply after dis- bursements under IDA-supported projects started in 1971/72. The figure for total resource mobilization for 1972/73 is 2.4 times larger than that for 1970/71, while that for local resources is 1.4 times larger over the same period. From then on, both total resource and local resource mobilization have been increasing at about the same rate; 2.6 and 2.7 times from 1972/73 to 1977/78, respectively. 2/ Funds other than those invested in ordinary debentures of sister banks or deposits with cooperative banks are lost for agriculture. Cross- subscription of debentures between LDBs, which have been sometimes questioned, results in the spreading of risks and better opportunities to use funds within the system. - 17 - of repayment for ordinary debentures, which are redeemable only at maturity, compared to those of sub-loans, which are recoverable annually, may weaken LDB efforts to recover loan installments each year in their entirety, since funds collected are not going to be utilized immediately. 3.36 The debentures issue has been dealt with by the Hazari Committee as an aspect of integration problems (see para. 3.28). The committee's report is strongly critical of the system and suggests the following changes in the procedures for raising resources: (a) for schemes approved by ARDC, the committee recommends that ARDC provide refinancing on the same basis as for commercial banks and SCB; contributions by the State governments now being made available to special debentures would be granted in the form of term loans on the same conditions; and b) for other lending programs, debentures may be replaced by bond issues with State government guarantee. The whole issue of LDB's raising resources through floating debentures was being con- sidered by RBI; necessary instructions will be issued after consultation with ARDC. (b) Eligibility Criteria for Refinancing 3.37 In order to control the level of overdues in the LDB system some restrictions were put on the eligibility of PLDBs (or SLDBs). Under the State-oriented credit projects, it was stipulated that loan recovery together with any State government contribution to share capital must constitute at least 75% of annual collectibles in order to qualify a PLDB/SLDB branch for participation in the project. There were no ceilings on the amounts of the State government contribution. 3.38 With ARDC I came a formula linking floating of new debentures with recovery performance. Under the formula, a PLDB (or SLDB branch) was not eligible for ARDC finance unless it had a recovery rate of 40% or more. For recovery rates between 40% and 75% the lending program increased progres- sively, linked to the increase in recovery; those with recovery rates above 75% would have an unrestricted lending program. Further, State government contributions were limited to 10% of collectibles. The same formula was adopted by RBI for the issuance of ordinary debentures. 3.39 As the profitability of a PLDB/SLDB branch depends largely on the lending volume, the eligibility formula gave banks a strong incentive to improve recovery rates. Moreover, the limit on State government contributions ensured that improvements were achieved through actual recoveries rather than through financial transfers.1. The debenture eligibility formula was for 1/ For example, under the Maharashtra Agricultural Credit Project, the State government transferred its contribution from SLDB branches having a small lending potential under the project (thus excluding them from a further participation in the project) to branches with a greater lending poten- tial, enabling it to utilize all project funds without any substantial improvement in actual recovery rate. Such transfers were not permitted under ARDC I. - 18 - a limited period (initially until October 30, 1977) during which time LDBs were expected to improve recovery to reach a minimum rate of 75% (65% if State government contributed 10% of the demand). Thereafter, PLDB/SLDB branches would not be eligible for refinancing. 3.40 During the first project year (1975/76), there was an improvement in recoveries and the average rate of overdues for LDBs throughout the country decreased from 39% to 35%. This improvement may have been a result of the application of the above formula, but could have also been affected by the exceptionally good harvest. In the second project year (1976/77), recovery rates declined and the average rate of overdues increased to about 40%. Moreover, the number of PLDB/SLDB branches not eligible for any lending increased from 82 in 1975/76 to 242 in 1976/77 (+195%), while the number of those entitled to unrestricted lending declined during the same period from 1,135 to 756 (-33%). The increase in overdues is probably explained by the effect of political intervention which accompanied the national and State elections. 3.41 Therefore, the effectiveness of the formula as a means for improving recovery rates remains unproven. Besides, its development implications need a careful study. Firstly, the farmers penalized are not those in default but rather those who have not yet taken out any loans. It is even possible that small farmers suffer more from the application of this formula since they generally lag behind the larger and more progressive ones in the adoption of new technologies. 3.42 Secondly, the formula may curtail development in areas prone to natural calamities, because recovery rates in less affected areas are gener- ally better. It has been assumed that rescheduling of loan repayments will generally take place following a natural calamity. However, some LDBs (e.g., Andhra Pradesh, Rajasthan) object to such practice, claiming that it raises farmers' expectations for future rephasing of loan repayments and has a bad effect on credit discipline. Experience in Maharashtra and Tamil Nadu, where rescheduling failed to improve actual loan recovery, supports this argument!/. 3.43 The initial formula has since been amended; the last time in January 19792/. Most amendments have aimed at reducing rigidities introduced with the formula, enabling completion of unfinished investments, and allowing for 1/ Although in those specific cases (a) rescheduling involved delaying payments without increasing the overall maturity period thus increas- ing repayment burden in the years following the rescheduling; and (b) the political climate was such that credit discipline would have broken down with or without rescheduling. The Region states that political considerations were the predominant factor. 2/ Under the third project, ARDC undertook to maintain at all times criteria satisfactory to the Bank with respect to LDB eligibility. - 19 - new lending to small farmers. These amendments have increased the eligibility of PLDB and SLDB branches which would otherwise have been deprived entirely or partly of new refinance. For example, in Gujarat the number of SLDB branches with unrestricted eligibility had been decreasing from 78 (as of June 30, 1976) to 11 (June 30, 1977) and 5 (June 30, 1978); it was raised to 28 as a result of the latest regulations. By contrast, over the same period, the number of branches not eligible for lending, which had climbed from 25 to 79 and to 95, declined to 68. Thus, PLDBs or SLDB branches which had become ineligible for lending because of their poor collection efforts have once again become eligible. Caution thus needs to be exercised so that the new flexibility is not misused to frustrate the basic purpose of the linking formula, namely, to help improve PLDB and SLDB loan recovery performance. 3.44 Continued use of the formula may be questioned on grounds of effec- tiveness, equity and development requirements. Since, however, the problem of overdues becomes more acute and no better alternative has yet been found, it would be a step backward to abrogate the present rules. Investigations should be continued not only for the formulation of rules which would not slow down development in nonprivileged areas, but also regarding practical measures or arrangements to improve recoveries. 3.45 The volume of ARDC refinance to CBs has not been linked to their recovery performance, as in the case of LDBs, for two main reasons: (i) lending for agriculture was still a new field of operations into which CBs were reluctant to expand; they needed as much encouragement as possible; and (ii) term agricultural lending involved such a small part of CB volume and resources that even heavy agricultural overdues could be borne without jeopar- dizing the financial stability of the institutions. While the second argument still holds true, CB agricultural operations, and their share in overall ARDC disbursements (nearly 50% today, see para. 3.30), have now reached a suffici- ent scale and maturity to justify closer performance monitoring, including periodic review of overdues. 5. Costs and Spreads 3.46 Over recent years, more and more PLDBs have incurred losses and LDB profitability has, in general, declined, due to: (i) the banks' inability to continue expanding their lending operations (see para. 3.13) and (ii) the current margin between LDBs' borrowing and re-lending rates, which seems to be insufficient to cover current operating expenditures, to maintain adequate provisions for bad debts and to build up adequate reserves. 3.47 A wide range of margins were quoted by LDBs as being necessary to cover their operating costs - usually 3.5 to 5%; generally around 4%. CBs provided similar figures. No estimates were given for possible losses because of defaults. There are some divergencies of opinion regarding the adequacy of such margins because: (i) different proportions of LDB administrative charges are borne by State governments; (ii) the costs of credit distribution, recovery and defaults are not known with sufficient accuracy, and (iii) the - 20 - volume of lending of an individual PLDB affe ts its cost structure as well as its profitability. A World Bank documentl' suggests that lending adminis- trative expenses incurred by LDBs themselves would reach about 5.7% of loans outstanding on average, or about 8% taking into account the services currently provided by Government. Risks are estimated to be 0.5% of loans outstanding. These figures are well supported and seem more realistic. On these assump- tions, the margin of 3% would fll considerably short of requirements and would result in operating losses..2 3.48 At present, however, SLDB income statements continue to show profit because of their substantial resources, the fact that in many cases no provi- sions for bad debts are made, and also because the return on investments offsets losses on lending operations. Such investments, mostly ordinary debenture redemption funds, can be as high as 50% of loans outstanding (see para. 3.34). These will tend to decline as the special debentures redeemable annually substitute for ordinary debentures redeemable at maturity. There are few recent cases of SLDB having incurred losses (Maharashtra in 1973/74). PLDB profitability is a more serious issue. Besides their smaller size, PLDBs, as primary societies, bear the main burden of credit distribution costs and cannot meet these in spite of a wider margin than that for the SLDBs (usually 1.75% against 1.25% for SLDBs). In addition, in contrast to SLDBs, they have no funds of their own for lending, little income from investments, and receive less assistance from state governments. An increasing number of PLDBs are operating at a loss, even some affiliated to the best performing SLDBs (as in Andhra Pradesh). There are, therefore, grounds for concern regarding the future viability of LDBs. 3.49 The LDB cost situation is worsening. Government policy - with explicit IDA support - requires them to continue to expand their lending to smaller farmers. Thus, costs per operation will increase while income per operation is likely to decrease. Measures to compensate the LDBs for such extra costs will have to be devised to prevent this socially desirable shift in lending patterns from squeezing the LDB's profit ratio too low. 3.50 It was agreed during negotiation of ARDC II that ARDC, in conjunc- tion with RBI, would carry out a study of interest spreads. Under ARDC III, GOI undertook to review it with particular reference to the LDBs. IDA further requested that the review concentrate on the cost of lending to small farmers, but no agreement was reached during negotiations because by then the review was almost complete. 6. Revolving IDA Funds 3.51 A special covenant concerning the reuse of IDA funds recovered from participating lending agencies was included in the project agreements for 1/ C. D. Datey (consultant). - The Financial Cost of Agricultural Credit: A Case Study of Indian Experience, World Bank Staff Working Paper No. 296, October 1978. 2/ Interest margins were raised to 3.75-3.85% in October 1980 and are above the 3.5% recommendations of the Committee on Interest Spreads. - 21 - Punjab, Andhra Pradesh, Karnataka and Maharashtral-. Under this covenant, repayments to ARDC by participating agencies had to be deposited in a special account to be used only for refinancing ARDC-approved agricultural development schemes. In this way, ARDC could count on significant amounts of funds, because the repayment periods under its refinancing arrangements are shorter than GOI's terms of lending. 3.52 Repayments to ARDC by participating banks have not, however, been deposited in a separate account, and this was never reported by IDA supervi- sion missions. It was not appropriate to have such covenants included in the project agreements: although the funds involved are substantial, they are available for a period shorter than the shortest term of any loan to be made under any scheme similar to those financed under the projects - even leaving aside the time required to commit and disburse them. Covenants like these do not appear in more recent projects. B. The Investments Financed 1. Groundwater Development 3.53 To make maximum use of available resources without adversely affect- ing either the resources or the farmers involved, a certain degree of overall control on groundwater development is required. In this regard, control measures provided for under the projects proved insufficient and led to some long-term problems which are far from being solved. Besides the need and measures for control, two further issues are discussed: incomplete well developments, and well failures with possible remedies. (a) Groundwater Development Control (i) The Need for Control 3.54 The need to control groundwater development is suggested on two grounds: (a) to protect the individual investment against subsequent develop- ment. Prima facie, this is a legitimate area of concern for both the inves- tors (farmers) and their banker; and (b) to prevent widespread and long-term over-exploitation of the aquifers. Excessive drawings may have negative economic effects. Pumping costs could increase to economically unacceptable levels; some aquifers may become contaminated or destroyed by salt water intrusion. However, over most of India, i.e., the hardrock areas as well as the alluvial areas in high rainfall (over 1,000 mm) zones, this type of drawdown is practically impossible. In the arid States with substantial alluvial deposits (Gujarat, Haryana and Punjab) groundwater is already fully developed. 1/ But not in those for Tamil Nadu, ARDC I and ARDC II. - 22 - (ii) The Measures for Control 3.55 In the absence of groundwater legislation, appraisal missions adopted a pragmatic approach to prevent overdevelopment by stipulating minimum well spacings and maximum well densities. These criteria were considered only temporary. They were also to apply to all wells financed by institutional credit. Thus, they were expected to be effective, since a large proportion of minor irrigation investments was assumed to be financed through institutional credit. Provisory spacing and density parameters were specified at appraisal. Since background information was scarce, these were on the conservative side. During project implementation, they were modified in the light of additional groundwater information obtained by the SGDs. Rigid spacing criteria have been no longer requested since ARDC III. Investigation of available water resources are now undertaken on an area-to-area basis. 3.56 In Andhra Pradesh, unlike the other projects, it was left to the SGD to determine the spacing and density norms, subject to ARDC and IDA approval. According to SGD calculations, groundwater resources in the hard rock area of Andhra Pradesh are sufficient to irrigate about 50% of the cultivated area. In this respect, the situation in Andhra Pradesh appears to be better than in the hard rock areas of Tamil Nadu and Maharashtra, where groundwater resources are sufficient for roughly 10-15% of the cultivated area (para. 3.64). It should be noted that dug wells in the hard rock areas of Tamil Nadu and Maharashtra are not a new development and large numbers had been installed prior to the Agricultural Credit projects. Project investments in minor irrigation amounted to about Rs. 370 million in Tamil Nadu (70% of total project costs) and Rs. 480 million in Maharashtra (90% of total project costs). Equitable allocation of water in Andhra Pradesh should, therefore, be easier than in those states. 3.57 Recently, some state governments (as GOTN) have extended controls to electrification: only wells which comply with the spacing criteria may be electrified. This measure may discourage uncontrolled investments to some extent, but will not be fully effective since farmers are still free to install diesel engines and many may do so despite their higher operating costs. 3.58 None of these State Governments have yet enacted groundwater legis- lation. Most consider it essential first to prove and quantify the damage; any delay in legislation will, however, make enactment and implementation more difficult (see para. 3.72). 3.59 The draft state groundwater legislation is based on a model pub- lished by GOI in December 1970, which includes principles necessary to the orderly development of a groundwater resource. The difficulties lie in implementation. The main problem is how to control development in areas in which the groundwater resource is sufficient for irrigating only a small fraction of the cultivated land (see example in para. 3.64 below). Statutory arrangements, based on spacing and density criteria, will be particularly unpopular since they will support a basically inequitable allocation of - 23 - groundwater resources. However, this is the only type of control that GOI or any of the state governments has seriously suggested to date. Withdrawal control, which would involve allocation of irrigation water quotas, would be more appropriate, but the mechanisms for enforcing it need to be carefully studied. 3.60 A major constraint on the introduction of groundwater legislation is the difficulty, and danger, of attempting to proceed from the general to the particular in terms of groundwater availability and behavior in any given area. While SGDs have developed considerable overall strength and competence, they have neither the manpower nor the data required to permit precise esti- mates of the potential for, or constraints on, development in a specific village. In addition, statistical shortcomings make it difficult to be precise about how much water is currently being extracted and how much remains available. (iii) Effectiveness of Control 3.61 The effectiveness of groundwater control when restricted to wells financed by institutional credit is questionable. Non-institutional credit accounted for a greater than anticipated proportion of the investment in minor 1/ irrigationl under the projects. This conclusion is strengthened by statis- tics from Tamil Nadu which indicate that, when minor irrigation investments approach the limit allowed by the spacing and density criteria, farmers increasingly finance well investments from non-institutional sources, either to avoid conforming to these criteria or because the criteria make them ineligible for institutional loans. 3.62 Despite this, to date there are not many documented cases of cur- tailed loan repayment due to sinking of a new well nearby. The main reasons for this may include the following: (a) the geological conditions in the Dekkan plateau, where an important proportion of groundwater development efforts under these projects are located, are such that the areas influenced by typical dug wells are not much larger than the fields of those farmers who 1/ In a survey of villages in Maharashtra, it was found that 25% of the wells constructed in the period 1972/73-1975/76 were financed under IDA projects. The SLDB statistics indicate that about 75% of the wells financed by its lending program in the same period were under IDA sup- ported programs. Assuming that this percentage applies also to the districts surveyed, it would appear that in those districts SLDB financed only about one third of the wells. The CBs are known to have financed a relatively small number of wells and, assuming that the above percentages hold true for the state, it can be concluded that a large proportion of the wells in Maharashtra were financed from private sources. In Andhra Pradesh, there are no reliable estimates of the scope of non-institution- ally financed well investments, but it is roughly estimated at 25% of total new wells. The data available are insufficient to permit clear conclusions to be drawn and should be supplemented by additional, care- fully planned investigations. There are no reliable data on private money lenders; however, it is clear that, in this state, farmers pre- ferred institutional credit sources because of their lower interest charges. - 24 - can afford wells with pumpsets; (b) the drought conditions prevailing in several states during much of the implementation period made it difficult to distinguish between the reduction caused by drought and that by sinking a neighboring well; (c) the minimum spacing and maximum density requirements were based on conservative calculations; and (d) it appears reasonable to assume that most farmers will not sink a well too near an existing one since this will put their own investment in jeopardy. If correct, this hypothesis would have important policy implications and should therefore be examined through studies and/or surveys. (iv) Control and Equitability 3.63 The spacing and density criteria as a mechanism for allocating use of a limited and valuable natural resource are logical from a financial and hydrological point of view, but they can have retrogressive economic and social effects. It is clear now that in some areas, as in the Tamil Nadu and Maharashtra plateaus, only a very small fraction (10-15%) of the farmers in any hydrologic unit can have a well permitting withdrawals of 16-24,000 m3 per year. The enforcement of spacing criteria (if effective) implies that when a farmer sinks a well, his close neighbors are automatically precluded from digging wells on their land. As the first borrowers in any area are usually the richer and more aggressive farmers, the conditions imposed by the credit agreements, may lead to inequities of income distribution. No guidelines for allocation of groundwater resources among farmers were prepared and the first to apply have been given the first entitlement to investing in a well. 3.64 Identification of this problem is easier than devising a solution, since to share, for instance, 80,000 m3 of annual recharge in a typical square kilometer in Maharashtra among the 40 farms usually comprising such area would give each only 2,000 m3 per year. Withdrawals at this low level would not support investment in any kind of well. 3.651/ By introducing spacing and density norms, the Bank attempted to impose a legal solution which differs from the present Indian thinking. There is no explicit statement on ownership of and rights to groundwater in India. Groundwater has never been declared to be publicly owned. The right of owpership of groundwater seems to rest with the owner of the overlying land' . The right to use the groundwater is acquired through exploitation and 1/ This paragraph, and some of the following draw material from Veerman, T.S. - "Water Policy and Water Institutions in Northern India: The Case of Groundwater Rights", Natural Resources Journal, Vol. 18, July 1978, pages 569 to 587. 2/ This is not absolutely clear. A case can be made out that the Central and State Governments are the ultimate owners of groundwater because ownership rights on groundwater are derived from ownership rights in land, and the governments are the ultimate owners of the land. - 25 - Indian farmers currently have an unrestricted right to groundwater obtained from a well sunk on their farms. Since groundwater cannot be delineated, it must be accepted that all landowners above an aquifer share (proportionally to their land holdings) in their rights to own and use the groundwater. 3.66 Private rights of ownership of groundwater are indeterminate for any individual user, because other co-owners may also take possession of the water. This situation can lead to well interference from neighboring wells, as well as to depletion of the aquifer over time, if eventually groundwater pumping exceeds recharge. Since this is the case in most of the South Indian plateau, it acts as stimulus for farmers to sink their wells as early as possible. The risk of interference and depletion becomes a motivation for interference and depletion, rather than for cooperation and conservation. 3.67 The Bank identified this problem but the solution adopted amounted to an acknowledgment that full private ownership rights reside with the individual who first drilled a well and takes full advantage of available groundwater, thereby depriving the neighbors (within the area determined by the spacing limit) of their right to exploit the groundwater. Two classes of farmers are thus created: (i) the first to get access to groundwater (i.e., to the credit or other financial resources required to finance the sinking and motorizing of a well), usually the better-off farmers, who acquired rights of usage over all groundwater; and (ii) the owners of the rest of the land, who were left with mere rights of ownership over groundwater but with no effective way to exploit it. 3.68 Staff members who participated in the appraisal missions mentioned that the equity issues were very much in their minds, but they concluded that "a more equitable allocation of groundwater rights" was not really an objective at the time. Appraisal teams pointed out that the projects- design was the correct one to get something going within the realm of the possible. The main objective was to maximize production development and to avoid over-exploitation of the groundwater resource (which has been achiev- ed to some extent). It was assumed that equity considerations would be taken care of by administrative means, pricing increases, encouragement of water sharing and other direct provisions, but these had only limited effect.1/ 3.69 A more equitable solution to the problem - and the problem is indeed serious - would have to recognize and protect the correlative and coequal rights of all overlying landowners to use their proportional share of the total groundwater supply. Such a solution could include one or more of the following elements: (a) water sharing among all farmers, proportionate to either their land holding or their water requirements; (b) water rationing among all those who want to sink a well - or have one already; (c) proper 1/ The projects envisaged substantial water sales by beneficiaries to neighboring farmers. Completion and Audit Reports found no indication of sales except in Uttar Pradesh. - 26 - compensation to farmers who lose or must involuntarily surrender their right to use the groundwater underlying their lands; (d) public or social drilling and operation of wells, with water being proportionately distributed or sold to all interested farmers; etc. 3.70 In each of the first two cases, all those receiving water would have to share the investment and operating costs with the owner/operator of the well. Water sharing or rationing could be carried out under either the appropriation system (under which prospective groundwater users must apply for a Government license to draw a certain amount of water), the adjudication system (under which water for quarreling users is apportioned by the judi- ciary), or traditional systems (under which someone with local authority - local councils, elders, leaders, etc. - iron out discrepancies or impose authoritative decisions, or some traditional consensual mechanisms are put to work). In the third case, a decision would be required on whether compensation should be paid by the neighboring well owner(s) or by the government. A suitable government agency./ or a cooperative/commercial institution should be identified to implement the fourth case. 3.71 Two additional points - both requiring flexibility - must be borne in mind when devising a more equitable solution. The first is to intro- duce enough flexibility into the system to accommodate future demands, most likely from many small farmers (who need to be educated as to their right to groundwater) and from other potential users (nearby villages or more distant cities.V). The second is to consider - where there is such a possibility - conjunctive development and use of surface and groundwater; correlative rights to groundwater, once adjudicated, should be flexible enough to be integrated into a rotational or other distribution system for canal water deliveries, or to be shifted to other potential users when the first ones could be more efficiently served with surface water. 3.72 But experience under these projects, and particularly in Tamil Nadu, indicates that by deferring the establishment of effective controls until most of the groundwater resource has been developed makes a definitive solution more difficult. Understandably, farmers who have not yet invested in wells object to such a prior allocative right, while those who have invested can be expected to object to any reallocation of the resource. The question of the control of groundwater exploitation is becoming an increasingly complex and difficult problem and one which needs to be urgently addressed. 1/ Like the SGDs, which are drilling wells for drinking water in Tamil Nadu and Andhra Pradesh. 2/ For example, there is an area in Northern Tamil Nadu where groundwater development for farming purposes has been restricted or forbidden to secure sources of water for future demands by Madras, the State capital. - 27 - (b) Incomplete Well Investments 3.73 Investments begun under State IDA-assisted projects were eligible for additional refinance from ARDC under the general line-of-credit projects which followed (ARDC I and II). Therefore, earlier performance reviews, which had noted that a substantial number of loans had not been fully drawn at completion, assumed that the vast majority of the incomplete investments would be completed under ARDC I and II. The audit report on the Karnataka Project, prepared more than one year after project completion, raised serious questions on the validity of these assumptions. As others before, the PCR found that almost half of the wells financed (24,000 out of 49,398 dugwells, the largest component in LDB financing throughout the South of India) were incomplete by the credit-s closing date, June 1977. Moreover, some 21,124 project wells still remained incomplete 15 months later; two years after completion, sub- loans for 18,333 wells had not been fully disbursed.-I Therefore, of the well loans made under the project 36% were never fully disbursed. The precise sums involved are not yet known. 3.74 The LDB, which disbursed 95% of Karnataka's minor irrigation pro- gram, has a three-installment (50:40:10) procedure for dugwells, calling for field visits before loan approval and again before release of the second and third installments. The borrower is usually his own contractor and the dugwell loan, as distinct from loans for pumpsets and similar facilities, is disbursed as a cash advance. The final 18,333 "incomplete" figure, therefore, covered all loans for which either the third or both the second and third installments had not been drawn. Of these, ARDC reported that almost 11,000 wells remained recorded as incomplete in LDB books; disbursements on 6,200 other loans had been stopped because of evidence of loan misallocation, and on 1,200 because the wells failed to produce adequate water supplies. 3.75 Information on actual field developments was available only on a small sample of these cases reviewed by ARDC. Of 35 cases, 20 had only drawn the first installment and 15 the second but not the third. Fourteen (40%) had been refused further financing for misallocation of either the first or the second installment. Another fourteen reported they were experiencing various difficulties (salinity, collapse of sidewalls, hard rock at lower depth, delays in electric connections), many of which might have been overcome with further support from the LDBs, had these been aware of the situation. Up to 8 borrowers (20%) had satisfactorily completed their wells without further Bank financing, either because they did not wish to go further into debt or because violation of some minor requirement made them ineligible to draw the final 10%. 3.76 ARDC claimed that a substantial proportion of the wells classified as incomplete may be presumed to have been physically completed without drawing down the final installment. ARDC took the view that as the amount of the third installment is small (only 10% of the loan), many farmers choose not to bother to draw it down. Moreover, since the final installment is released only after the beneficiary has obtained a completion certificate, others may 1/ The Region found in a recent survey that the number of incomplete wells seems to be lower than indicated here. - 28 - have given up because they could not satisfy the strict technical norms laid down by the PLDB. In any event, this issue was being examined by LDB when the projects were audited and a study was expected to be completed by January 1980. 3.77 One important aspect that became clear in the course of discussions is that the LDB loan supervision system is inadequate in that it depends on the borrower making contact with the lending agencies and not vice versa. It is a firm requirement of LDB that each borrower is visited before an install- ment is disbursed but visits took place rarely, if ever, after disbursement until the first repayment is due, and then, only if overdue. Consequently, many borrowers who had drawn 50% of their loan and did not apply for further finance were not followed up at all until repayments fell due, and then only if overdue. 3.78 LDB has since tightened up lending procedures considerably and has, for instance, specified maximum intervals between predisbursement visits to keep a check on long-gestation loans. The difficulty of effectively super- vising slow-disbursing cash loans on a large scale should not be underestim- ated, however. LDBs do not have nearly enough supervisory staff to maintain regular contact with borrowers. The existing staff have virtually no access to private transport, and are already overstretched without attending to the additional registers required by ARDC which are supposed to make loan verifi- cation systematic but are in practice rarely maintained. The on-going study by RBI and ARDC of the LDB organization and management, required in connection with LDB participation in the Third ARDC Credit Project, should include a review of LDB supervision procedures and a reappraisal of the maximum workload and area of operations expected of, and the resources available to, each manager and supervisor. 3.79 In summary, a substantial though unquantifiable part of the invest- ments in wells may not have been completed, raising doubts about the true magnitude of the estimates of the projects' impact on agricultural production (see paras. 4.22 and 4.23). (c) Well Failure and Remedies 3.80 Estimates of the proportion of well failure vary from 1% (percentage of claims for well failure subsidy) to about 13% (based on survey). SGDs claim that as a result of groundwater investigations there has been a reduc- tion in the proportion of well failures, but there are no data to verify this. Approved cases of failed wells are usually entitled to a subsidy which pro- vides remission of interest accrued up to the date of application for subsidy and remission of a part of the outstanding loan. As shown above, the subsidy program may have covered only a small proportion of the well failures and only a part of the investment cost. The main risk was still carried by the farmer and this may have deterred small farmers particularLy from investing in minor irrigation. - 29 - 3.81 It seems timely that insurance schemes should be introduced which would cover these risks. Insurance to cover the loan repayment commitments is feasible since SGDs should be able to provide the data necessary for calculat- ing insurance premiums. The premium could be paid by the financing bank and added to the loan amount. 2. Farm Mechanization 3.82 Whether farm mechanization components should have been included in projects in India has been a hotly contested issue. Further, procuring part of the machinery, particularly tractors, led to conflict between the Government and the Bank. These two main issues are discussed below. In addition, comments on the Indian mechanization policy are presented. (a) The Justification of a Farm Mechanization Component 3.83 The audits on these six projects did not attempt to answer fully the general question on whether mechanization in India is justified. Much re- search has been undertaken; papers supporting and contesting mechanization in the Indian context have been published, debates have occured among Bank staff. Mechanization was eventually deemed necessary in those particular states at the particular time. 3.84 Prior to appraisal, farmers in several states appeared to be experi- encing a shortage of labor and bullock power, particularly during the overlap of kharif harvest and rabi planting, the rabi harvest and kharif planting. Labor and bullock shortages further increased with the rapid introduction of high yielding varieties of wheat, rice, cotton and oil seed crops (HYVs), the expansion in minor irrigation resources and the use of fertilizers. Tractors would enable farmers to convert waste land to arable land, increase crop intensity and yields, and free land formerly used for producing and maintain- ing bullocks. Harvesters were to enable a more rapid harvesting and threshing of the grain crops, minimizing field losses. Early harvesting could allow earlier and more extensive land preparation and planting for cotton after wheat and wheat after paddy, thus enhancing yields and cropping intensity. Imported disks and plow bottoms were needed because India did not manufacture hard steel essential to produce superior attachments. 3.85 Consequently, demand for imported tractors was extremely high and new tractors purchased at official prices could be resold at 35% to 70% over cost. The government had introduced quotas on imports of the more popular kinds and delivery dates stretched to three or four years. The investment cost of bullocks had also increased, appearing about to be the same as for a tractor. The financial benefit-cost relationship of the whole package (tractor-irrigation-HYV) was highly favorable for farmers, particularly in Punjab and Haryana, where the acceptance of new production techniques was the highest. 3.86 There were some uncertainties as to whether the strong demand would continue. Domestic tractor production had increased very rapidly, but dealers - 30 - had built up large inventories because farmers considered most domestic tractors inferior to certain imported types. Regulated prices, a liberal import policy and an overvalued exchange rate artificially distorted demand for farm machinery in general and for imported tractors in particular. 3.87 Another important uncertainty was whether farming with tractors could displace agricultural laborers and enable landlords to evict tenants. This led to the Bank's financing evaluation studies under the Gujarat and Punjab agricultural credit projects, and the government financing similar studies in other states. (i) Benefits from Farm Mechanization 3.88 Two of the six projects including a farm mechanization component were in the Ganges region: Punjab and Haryana; the third one was in Gujarat, while the remaining three were in the Central and Southern States of Andhra Pradesh, Tamil Nadu and Karnataka. Conditions in these three areas are quite different. In general terms, empirical evidence suggests that private returns on tractor ownership are strongly positive in high wage pockets such as Punjab/Haryana, or where reclamation possibilities still exist, such as in Gujarat. Social returns elsewhere are likely to be adversely affected by the labor-saving potential created by the displacement of tenants. 3.89 A review!V of several tractorization studies (the PAU study in- cluded) found that in the Punjab-Haryana area: (i) cropping intensity was 0 - 10% higher on tractor farms as compared with bullock farms, or prior to tractorization; (ii) positive effects on cropping patterns were not a general phenomenon, except the reduced areas under fodder; (iii) yield increases ranged from 0 - 61%, but these were also derived from the use of fertilizers and HYVs and could not be regarded solely as attributable to the tractors, and (iv) tractor farms were slightly superior in timing operations, which re- sulted in higher yields. Tractor farms had a substantially higher total value of crop production per ha than non-tractor farms, which could be achieved in various ways, as mentioned above, but which is difficult to attribute solely to tractors. The review also mentioned a possible advantage in marketing as a result of easier transport for tractor farms. All studies in the review showed medium to high farmer benefits and somewhat lower but still satisfac- tory social benefits. The review suggested that farmer benefits were over- estimated, and that financial returns to tractors from agricultural operations must be close to zero, except in situations where area effects are possible. 3.90 The Bank, from 1971 to 1975, supported research on the impact of mechanization in India through the "Agricultural Mechanization Study: India" which comprised the PAU study mentioned above and a similar study in Gujarat State. Simultaneously the Bank carried out a tractor study for the Pakistan 1/ Binswanger, Hans P. The Economics of Tractors in South Asia, An Analy- tical Review, Hyderabad, India, 1978. - 31 - Punjab!/. Findings from the latter and the Gujarat study show that produc- tion was higher on tractor farms than on bullock farms but that this could not be attributed solely to tractors, but rather to the combination of tractors, irrigation and modern inputs. 3.91 Economic benefits of the tractor components of the Punjab and Haryana projects thus stem from several factors: (i) area effects from cultivating waste and fodder land; (ii) increased cropping intensity; (iii) increased yields including the yield effect from timeliness in soil prepara- tion; (iv) transportation benefits (for marketing and social purposes); and (v) labor savings through lowering labor use per unit of agricultural output. Conventional analysis, however, cannot fully take into account some of the other principal advantages of the tractor to the owner or regular hirer. These are, particularly in the Punjab, such factors as (i) the convenience to, and saving of, family labor, whatever its opportunity cost and particu- larly, (ii) the flexibility and independence afforded by the tractor as against reliance on mobilizing sufficient numbers of bullocks and laborers at the right time and place; and (iii) "commercialization" benefits, a rather unorthodox benefit which was repeatedly pointed out to the audit mission by several farmers: purchase of a tractor alerts the farmer to further invest- ment possibilities, thereby accelerating commercialization of his farm. 3.92 In Gujarat, the impact of tractors on output is unclear. There is evidence from various local surveys that the use of tractors has not increased output when substituted for bullock power. However, there is some evidence that because of the tractor, farmers have been able to make better use of other inputs, thereby increasing the value of their production by about 10%. Having a tractor has also meant that less land would be devoted to fodder and less capital invested in bullocks, and this has contributed to increasing incomes. Contrary to some experience elsewhere, the area cultivated by tractor owners increased only marginally (by about 10%) as a result of their introduction. Some of this area expansion was on land formerly uncultivated owing to drainage problems, heavy soil types or similar reason, and some was an expansion on land previously leased. Tractors have made farming easier for their owners; however, the survey results have not been able to show high financial returns as a result of tractor investments. 3.93 The situation in Karnataka was quite different from that in Pun- jab, Haryana and Gujarati2. It was found at audit that tractors could not 1/ McInerney, John P. and Graham F. Donaldson, The Consequences of Farm Tractors in Pakistan. World Bank Staff Working Paper No. 210, Feb- bruary 1975. 2/ The PCR and audit report on the Andhra Pradesh and Tamil Nadu projects concentrated on discussing minor irrigation issues; they did not discuss the effects of tractors on production, and employment. - 32 - increase cropping intensities or yields significantly because a large part of the cultivated area was dry land. Mechanization only facilitated quick and timely operations by substituting mechanical power for casual labor, the shortage of which had presented a problem in peak seasons. Tractor owners interviewed by Bank missions, and most of ARDCs respondents, were candid as to the main benefits they saw from their investments: the ability to manage substantial (16 to 20 ha), often fragmented holdings, without having to sell or rent a major portion and without depending on burdensome permanent labor. (ii) Drawbacks from Farm Mechanization 3.94 The case against tractors focuses not on their productivity but on the risk of displacement of tenants and/or permanent farm labor in favor of self cultivation with tractors. While the displacement of smallholders and tenants through the expansion of tractor owner-operated farms had been reason- ably well established in the case of at least one area in the continentl/ where farm size increased considerably after tractors were introduced, there is no similar evidence of large scale land consolidation and tenant displace- ment in the Indian Punjab. This is partly a result of land ceiling and tenancy legislation implemented since 1953; even if redistribution objectives have not been fully accomplished, pressure on the land together with legisla- tion have certainly created a barrier to rapid area expansion by large farm- ers. Another reason is that subsidies given to labor-saving investments prior to 1971/72 have been abolished and even reversed through a series of GOI policy measures taken before the IDA project tractors were delivered (see paras. 3.87 and 3.107). 3.95 Displacement of labor is often mentioned as the main disadvantage of tractorization. This argument should, however, be viewed in the context of particular wage and employment situations in each area and does not appear to hold for Punjab and Haryana. These two States form part of the Ganges region, which also includes Uttar Pradesh and Bihar, where weather and soil conditions are nearly identical for all four states. Of these four only Punjab and Haryana have done well. Punjab now has the highest per capita income in India, Haryana the second highest. Agricultural production is also highest in Punjab, followed again by Haryana. Industrial production in Punjab has increased 350% since 1947, compared with an increase of 220% for all India, while industrial production in Haryana, with a smaller population, is even larger than that in Punjab. Labor has become increasingly scarce in the two states and industrialists maintain that, if an embargo were placed on migrant labor from Uttar Pradesh, Bihar and the southern Indian states, their industry would be seriously affected. 1/ McInerney and Donaldson, op. cit. - 33 - 3.96 The case of the Punjab (and Haryana) is special to the extent that agricultural laborers- wages have risen faster than the general price index since the late 60s, i.e., throughout the period associated with the introduction of tractors. Increased seasonal immigration from other regions meets the substantial unskilled labor constraints in these States. Given certain rigidities in the labor market, a case can be made for further labor- saving investments on a more limited scale until this regional imbalance can be removed. The clear danger of this approach, in view of the poor medium term prospects for fully absorbing rural unskilled labor nationwide, is that it creates the potential for further labor-saving investments (e.g., tractor- drawn harvesting machinery) ahead of any real labor scarcity or permanent increase in real wages. 3.97 The study conducted by the Punjab Agricultural University in 1972/ 73 shows that in comparing tractor farms with bullock farms, the decline of family labor (when family members reportedly spent more time on entrepeneu- rial and management activities) is more than offset by increases in use of permanent and casual labor, particularly the latter, resulting in a net increase of on-farm employment due to intensification of farming practices. On the basis of a cultivated hectare, tractor farms appeared to have slightly more total on-farm employment; on the basis of a cropped hectare, slightly less than nonmechanized farms. In most other studies included in Binswanger-s reviewY, tractorization had a neutral overall labor effect. Contrary to the findings of the PAU study, these studies indicate that family labor generally increased, while permanent labor was reduced substantially; daily (casual) labor was found to increase in most cases, which is again in line with PAU findings. Another study by the Government of Punjab showed that use of casual labor per hectare (probably cultivated hectare) was significantly higher on tractor holdings than on bullock holdings and that family labor had been partly replaced. 3.98 In Gujarat, the impact of tractors on employment is not clear. Tractor farmers used similar amounts of labor to bullock farmers (170-180 man-days/ha/year), up to 80% of which was hired labor. This similarity is not due to a greater cropping intensity by tractor farmers, but it appears that while less labor is used for tractor cultivation, the balance is often required for other tasks created in the switch to higher value, labor inten- sive crops. In Karnataka some evidence of labor and tenant displacement was found. On average, a tractor rendered redundant about 140 man-days of casual labor per annum; nearly two-thirds of the tractor owners dispensed with one permanent laborer. Applying these results to 2,912 tractors financed under the project, as many as 400,000 man-days of casual labor and 1,100 laborers would have been affected. These results, however, are based on a very limited sample of 20 projects and 20 control farmers confined to one district. 1/ Binswanger, op. cit. - 34 - 3.99 The pattern that emerges is that in Punjab and Haryana, per culti- vated hectare, tractorization caused a reduction in use of family labor, a slight increase in use of permanent and total labor, and a strong increase in use of casual labor. Combined with the additional off-farm employment created by tractorization (production, marketing and maintenance of tractors and implements together with indirect effects) the projects created more employ- ment than they displaced. Again it must be recognized that tractorization is only one element in a package of improvements. In contrast, in Karnataka, where wages are lower, the unemployment rates high, and the stage of develop- ment lower, tractorization had at best a less positive effect on employment or even a negative effect altogether. (b) GOI's Mechanization Policy 3.100 GOI-s policy regarding farm mechanization changed during project implementation, and particularly thereafter following the change of government in 1977; increasing emphasis has been placed on employment generation in rural areas. GOI is concerned that tractors and combine harvesters will displace labor, particularly the combines of which imports have been halted (they are not domestically produced). Regarding tractors, GOI's policy is not against tractorization as such; the new policy is one of selective mechanization with two basic components: (i) tractor use for land reclamation and development also for custom services, with more emphasis on ownership by small farmers and by groups of farmers; and (ii) promotion of implements and tools for animal and manual labor. This should be seen in the light of the present general power shortage in India's agriculture, on the one hand (the 1971 estimate of power availability for India is about 0.4 hp/ha; Punjab about 0.7, and Japan about 2.5), and GOIs concern for employment creation, on the other. But the experience of the Punjab and Haryana projects shows that, given the state of economic development in the area concerned, tractorization can create addi- tional employment, particularly in non-farm sectors and can as such accelerate economic development. The Planning Commission has now set up a committee on farm mechanization (including harvester combines) with the objective of settling the tractor dispute. 3.101 The tractor components of the projects benefitted, on average, medium and large rather than small farmers. But the projects were not designed to reach small farmers, and GOI and IDA policies only changed during project implementation in favor of generating more employment and reaching smaller farmers. The average holding of a tractor owner was at appraisal expected to be 15 to 20 ha. This compares with average farm sizes, as estimated at appraisal, of 5.7 ha for Punjab, 5.2 ha for Haryana, and the majority below 4.0 ha, which since then declined to about 3 ha, due to subdivision of the traditional family farm. Though the projects did not serve many small farmers directly, they did much better in that regard than expected. Moreover, other beneficial effects accrued to small ex-farmers and farmers: (i) on balance the projects are believed to have created employment; and (ii) custom services were rendered to small farmers by project participants. - 35 - (c) Procurement 3.102 The six projects which included farm mechanization components encountered major problems in procuring the required tractors. As a result, all experienced substantial delays. For example, originally scheduled to be fully disbursed in 28 months, the Punjab project required 82 months; the Haryana project, scheduled for disbursement in 41 months, required 68. 3.103 The main reason for the procurement problems was the differing objectives held by the Government of India and by the Bank. The Bank's main objective and declared concern was to ensure that farmers were efficiently supplied with high quality imported equipment of their choice on competitive terms, with appropriate guarantees for after sales service and parts from reputable dealers. The Government of India fully shared the substance of this objective, but wanted tractor financing under the projects to be consistent with its industrial objectives, namely, the development and strengthening of the Indian tractor industry. Since neither of the two most usual procurement mechanisms (centrally organized ICB and individual "prudent shopping" from dealers) would have achieved these objectives, IDA proposed a compromise combining both bulk procurement and farmer choice. This solution, which was included in the credit agreements, involved obtaining quotations from pre- qualified suppliers; advertising these quotations and requesting potential borrowers to indicate their first and second choice; aggregating these choices by brand of tractor, and finally placing the appropriate order. Tractors would be delivered through dealers designated by the supplier in a contract with ARDC based on the original bid terms. 3.104 All the evidence, however, suggests that even without other compli- cating factors, the organizational difficulties involved were grossly under- estimated. To attempt procurement within 2 or 3 years of tens of thousands of machines in six States through several dozen financial institutions and hundreds of dealers would be difficult enough. However, to require all potential borrowers to be canvassed during the period of bid validity, and their choices to hold good through the whole process of bid adjudication, contracts, and delivery, was impractical. The additional administrative costs, foregone production, and intervening price increases largely cancel out any potential price advantages of ICB. 3.105 But further complications were to arise. The first was a GOI proposal that brands of tractors to be financed by the project be limited to those already being manufactured or to be manufactured in India. IDA proposed to finance only those brands imported from Bank member countries by Indian tractor dealers who maintained acceptable pre- and post-sale services irres- pective of whether they were manufacturing or planning to manufacture tractors in India. In the end, prequalification was agreed to include a major emphasis on commonality (interchangeable parts and processes); only manufacturers of models already locally assembled qualified. The GOI-s requirement of inter- changeability meant that models eligible for prequalification should have 75% of individual parts interchangeable with models licensed for production in - 36 - India. Eventually, 00I dropped this requirement in December 1973, when it was agreed that, as GOI had requested, IDA would also finance domestically manu- factured tractors. Reasons for this request were a much larger than anti- cipated domestic tractor production and a change in relative prices that favored domestically produced tractors over imported tractors. 3.106 Another issue arose with regard to the agency within Punjab and Haryana to be responsible for pooling farmer applications, obtaining bids from suppliers and managing delivery of the tractors. In the Gujarat project, the Land Development Bank had been designated as this agency and the first procurement tranche was being implemented satisfactorily. In Punjab and Haryana, several other banks were involved and to preserve impartiality, IDA proposed ARDC. At negotiations, GOI instead argued in favor of designating the Punjab and Haryana Agro-Industries Corporations (AIC) as the respective procurement agencies. This was agreed to. In view of GOI's commitment to strengthen local enterprises, their request was not surprising, but the agreement contrasted sharply with the original IDA objectives (para. 3.103). 3.107 A new problem then arose in Punjab. The Punjab-s AIC failed to have each applicant indicate first and second choice of tractor brand. 27,000 farmers applied for the 8,000 tractors. PAIC proposed to select the 8,000 project beneficiaries by lottery; choice of tractors was to be limited to the five brands imported by Indian tractor manufacturers, price and after-sales service were to be negotiated without normal tender procedures. PAIC would then offer East European tractors to the 19,000 applicants who failed to get an IDA financed tractor. IDA objected to this approach. Formal procedures were drafted, but IDA again objected to the terms as these had included the entire 8,000 tractors in one tranche, and also specified a power range of 30 to 80 hp while the credit agreement stipulated 30 to 60 hp. It also authorized PAIC to set the price paid by the farmer and commission rates for dealer after-sales service, in contradiction to the provision of the agreement (November and December 1970). Proposed revisions were exchanged between GOI and IDA and an agreement was reached in March 1971. However, before publish- ing the tender, PAIC changed terms and conditions by increasing its handling charges. The differences were overcome and tenders were closed June 21, 1971. 3.108 Meanwhile, other issues had developed. Under the GATTl/, GOI had agreed to free farm tractors from custom duties. In addition, domestic tractor prices were regulated. Further, in 1969 and 1970, the open market price of the Indian rupee was considerably below the official exchange rate; i.e., Rs 10 to 12 = US$1 compared with the official rate of Rs 7.5 = US$1. Therefore, during appraisal of the first four credits which included financing farm tractors, Bank staff questioned subsidized prices for imported tractors and suggested that GOI consider increasing tariff duties and taxes on imported 1/ General Agreement on Tariffs and Trade. - 37 - tractors and tractor components. The Bank believed that a substantial coun- tervailing tariff duty would help equalize the difference in exchange rates, provide protection to domestic tractor manufacturers, and reduce or eliminate the "black market" for used tractors. GOI amended its agreement with the GATT countries and imposed a 30% tariff duty against the c.i.f. value together with a 10% excise duty plus a sales tax on imported tractors. It also imposed a 20% tax on the ex-factory value of domestic tractors and increased regulated prices accordingly. GOI also liberalized licensing procedures to enable more Indian industralists to become manufacturers with or without international collaboration. These changes were announced in May 1971 just after the first Punjab consignment went to tender and before opening the bids. Before the change, a popular imported brand of a 45 hp tractor was priced at Rs 24,400; after the change the price increased to Rs 44,400. In the event, exchange rate fluctuation - largely due to external circumstances rather than delib- erate GOI/IDA planning - effectively removed the problem long before any tractors were procured. 3.109 In the particular case of Punjab, another major factor holding up PAIC sales was that the Indian State Trading Corporation (STC) was able to import and sell identical brands and models of tractors at a lower cost. For three popular brands, STC prices were between 75 and 93% of IDA - PAIC price. 3.110 IDA's experience with tractor procurement in India reveals weak- nesses in the original project design and casts doubts on the validity of the compromises subsequently reached. A policy disagreement appears to have been too hastily reconciled by a procedural arrangement which was subject to different interpretations; this blocked progress on all six projects for several years until a more realistic solution was negotiated. Further, the Bank-s suggestion that GOI impose tariff duties on tractors, in contravention to the GOI-GATT agreement, raises the disturbing question whether the Bank is correct in encouraging - or requiring - its borrowers to amend or abrogate their international agreement. IV. THE HIGH RATES OF RETURN 4.01 This set of projects judged on the basis of economic rates of return is an outstanding success. Economic rates of return for tractor com- ponents in six projects range from 16 to over 50%; in three cases reestimated returns are much higher than the appraisal estimates. In the case of ground- water development through minor irrigation investments, a component which commanded 80% of total funds invested, returns for most projects range from 25 to over 50%; in six of the nine cases, returns estimated at completion are higher than appraisal estimates; in the remaining three they are similar. The overall economic return for the set of projects can be estimated between 35 - 40%. This ranks these projects well above agricultural credit projects - 38 - in other countries, and much above most other agricultural development proj- ects.1/. This image of success is consistent and is perceived and shared by all the parties involved (Bank officers as well as officials in GOI, ARDC, SLDB and PLDBs). Since this is the largest Bank-supported agricultural credit program anywhere, and since the rates of return claimed are so high, it seems worth analyzing the anatomy of success of this set of projects. 4.02 While the projects can without serious reservation be termed highly successful, some figures quoted in the project completion and audit reports may overestimate the real returns attributable to them. The reasons for this are largely methodological, and bear both on intrinsic characteristics of credit projects and on the methodology currently applied by the Bank for estimating economic rates of return. The rest of this chapter is devoted to discussing first the reasons for success and then - briefly - the method- ological approach which may lead to some overestimation of the projects' returns. A. Reasons for High Project Profitability 4.03 Some of the success has to be credited to the favorable technical environment surrounding the projects; the projects acted as catalysts pro- viding some of the inputs for the recently introduced, highly profitable, technological package. 4.04 During the 60s a new technology was introduced embodying high yielding varieties (HYVs), of both wheat and rice. Dwarf wheat varieties developed by CYMMIT in Mexico and rice varieties developed in the Philippines at IRRI proved well suited for Indian conditions, particularly in the higher Gangetic plain. Together with large amounts of fertilizer, adequate and reliable water supplies and enough power, they constitute a powerful tech- nical package which leads to a manifold increase in yields and a substantial increase in cropping intensities2/. This technical package was rapidly adopted by a large proportion of farmers, mainly in Northern India, giving rise to what has been called - in India as well as elsewhere - the Green Revolution. 4.05 Application of the new technology required both water and power. Therefore, the farmers who first and most easily adopted the new technology were those who had already secured power and water supplies. By the late 60s, other farmers, lacking either one or the other or both inputs, were demanding 1/ All agricultural credit projects audited in 1979 show an average economic rate of return of 26%, and other agricultural development projects, around 12%. 2/ There is no doubt that this comprises a package. The new seeds without fertilizers or power or with inadequate water supplies are not neces- sarily superior to traditional varieties. - 39 - wells, pumpsets, and tractors in order to take advantage of the HYVs and this set of projects provided funds for these investments. The credit projects provided the missing links which farmers, working in an environment technolo- gically well endowed, were demanding to continue expanding the Green Revolu- tion. New or improved wells and pumpsets allowed farmers to irrigate fields previously dry-farmed, or to improve water security or water control on previously irrigated land. Contrary to large scale irrigation projects with extended construction periods the on-farm well irrigation investments yield returns immediately. Tractors allowed speedy land preparation required for double cropping, as well as quicker harvesting and facilitated crop transports to the markets. This indeed made the projects extremely profitable, since farmers previously left to the vagaries of the weather were now able to adopt the new HYV technology. 4.06 Consequently economic returns to investments in tractors under the Punjab and Haryana projects, in the heartland of India-s Green Revolution, are associated with large increases in production. In contrast, returns on tractors, although high in absolute terms (16 and 21%, respectively) are much lower in Gujarat and Karnataka. In both cases, completion estimates are well below appraisal estimates. In Gujarat, this is mainly due to tractors increasing output only by about 10% (by value). In Karnataka, tractors did not increase cropping intensities or yields significantly because a large part of the cultivated area was - and remains - dry land. Demand for tractors also reflects the project-s role in supplying key inputs for the Green Revolution. In Punjab, 27,000 applications were received for the estimated 8,000 imported tractors, while total requests for the 2,000 tractors to be allocated to Karnataka reached only 2,800, reflecting the different potential benefits to be derived in either case from the new technology. 4.07 In short, the projects' high returns are largely due to their pro- viding essential inputs in a technologically rich environment, where a highly profitable technical package was already available and well known to the farmers, and where the main limitation for further adoption was the lack of either water, power or both. In view of the high complementarity among the four chief elements: seeds, fertilizer, water and power, it is surprising on first sight that the projects made no attempt to ensure that farmers requesting subloans for tractors had guaranteed water supplies, or that farmers requiring wells or pumpsets had adequate power available. 4.08 The existence of proven technologies, well known to farmers who lacked only one or two crucial inputs to put them fully into effect, is indeed the major contributing factor for the successful performance of this set of projects. There is a second factor, however, which cannot be ignored: the existence of a well-developed credit system, which predated the projects and which was extended and strengthened under them. The relevant feature is the land development bank system (LDB), with thousands of units scattered throughout rural India with a long experience in dealing with farmers - and vice versa. New disbursement procedures had to be established for channeling - 40 - project funds to LDBs, but these closely paralleled the existing mechanisms. Further, the projects introduced new evaluation procedures, which took greater account of the economic profitability and prospects of each bor- rower's investment proposals than of the collaterals for subloans. Additional institution-building developed under these projects. Where state groundwater development agencies existed, they were strengthened; where they did not, they were created. Two minor irrigation corporations to assist small farmers were established in Karnataka. At the same time, a large extension system covered most of the rural territory of India adequately. Therefore, the original institutional capabilities together with project-supported institution- strengthening throughout the decade are also responsible for the success of the projects. 4.09 Furthermore, the projects were implemented at a time when inter- national prices for most agricultural commodities, including the two main project-related commodities: rice and wheat, were particularly high, higher than forecast at appraisal. Since these prices predominated throughout most of the projects' implementation stage and despite considerably increased input prices, their effects in the recalculated economic rates of return are also substantial. B. Methodological Problems in Estimating the Projects' Benefits 4.10 Questions could be asked about the methodology, its strengths and weaknesses in assessing the economic returns of credit projects. There are some methodological constraints and problems in evaluating any agricultural credit project, principally the fungibility of the "input" these projects supply, the limited comparability between project beneficiaries and non- beneficiaries, and the attribution of benefits to credit projects of other benefits generated by two or more external factors. These are discussed in greater detail in Annex I, only summary comment on four issues being made here. These four issues are: (i) Fungibility of project-provided inputs. The question of obtaining sub-loans to substitute for other sources of finance. (ii) Limits of comparing project participants and "control" group farmers. The comparison of project beneficiaries with a group of similar farmers who did not have access to project funds. (iii) Problems of attribution. The technical package is composed of several sub-components and it is impossible to separate individual sub-component benefits. (iv) The farm-model approach. An imaginary farm serves as model for establishing cost/benefit streams. - 41 - 4.11 In analyzing these four issues, it becomes apparent that "fungibil- ity" is difficult if not impossible to determine. Farmers may or may not have obtained credit from other sources, may or may not have had own funds avail- able. Similar difficulties exist in comparing farmers who obtained loans with those who did not. There is usually a "quality" difference between credit takers and non-takers. Investing on the farm entails risk-taking and entre- preneurial drive, a trait usually less developed in farmers not participating in agricultural credit schemes. 4.12 Although water is usually considered the limiting factor to in- creased crop yields or higher cropping intensities, there are other factors of almost equal importance contributing to increased production: improved varieties, fertilizers, pesticides, tractor power, etc. Use of these inputs or application of improved technologies, however, is dependent on successful basic as well as adaptive research and on extension activities to deliver research recommendations. In the case of the Indian credit projects contribu- tions to higher output were accomplished through a combination of the above- mentioned factors. Research supplied among other things the new high yielding wheat and rice varieties. Extension acquainted farmers with fertilizer and pesticide application. The projects provided medium-term credit for water and power while seasonal inputs were financed from different sources. The calcu- lation of returns of the individual "inputs" is impossible. 4.13 Equally problematic appears the economic analysis based on "farm models". The extreme variation of farm sizes, cropping patterns, already existing machinery, equipment, buildings, etc. on the farms makes identifica- tion of a representative, average farm which then serves as the analytical model for thousands of other farms difficult and makes this methodological approach of questionable value. C. Conclusion 4.14 The above comments on the shortcomings of methodology should not be taken to detract from the success of the projects. The comments are only meant to enter a caveat to their unusually high reestimated rates of return.l/ Project staff do not fully share the view that inadequate methodology prevents attribution of project benefits between investments in irrigation and other items in the production package such as seed, fertilizer, research and exten- sion; as they see it, the same applies to other agricultural development projects as well, and rates of return remain appropriate to use for comparison with other projects in which the same methodology is employed. The audit 1/ CPS expresses the view that it is necessary to keep the importance of the ERR in perspective. The Bank needs to retain the cost: benefit analysis as a tool to measure the efficiency of its investments but also needs to explore better methods to measure the financial overall impact of its projects. - 42 - disagrees with this view. Agricultual credit projects present some intrinsic difficulties for evaluation, not common to other projects. The Bank's methodology for computing these rates of return is biased in favor of agri- cultural credit projects by not taking into account cases of failure or incomplete investments, delays and overheads/. 4.15 Nevertheless, this set of projects was consistently successful in achieving its objectives; the rates of return on the investment they financed are high and above the average rate of returns computed for other Bank assisted agricultural credit projects, which benefit or suffer from the same shortcomings of evaluation methodology. They are high enough to remain impressive even if some discounting were done to allow for methodological shortcomings, and investments in tubewells and dug wells are probably the most cost effective means of irrigation in India at the present time. 4.16 The significance of the experience registered by these projects also goes beyond their immediate setting. They illustrate explicitly the impor- tance of a mutually reinforcing package in agricultural credit operations, of a broadly congenial sector context, and of Borrower commitment and a fairly rapid institutional evolution in response to changing needs. The projects also provide a good example of a generally flexible and productive interac- tion between Bank and Borrower. At another level, they underline the impor- tance of the conjunctive use of surface and groundwater in areas where both exist so as to maximize the benefits from irrigation. 1/ The Region states that the administrative costs of lending (2% of invest- ment cost) combined with costs of extension services ($0.70/ha annually) would not reduce the ERR substantially. Beginning with ARDC III admin- istrative costs are accounted for in the economic analyses. - 43 - Annex I COMMENTS ON METHODOLOGICAL APPROACH IN PROJECT BENEFIT CALCULATION A. Fungibility of Project-provided Inputs 1. Since the agricultural credit projects financed farmers acquisition of tractors, wells and pumpsets to expand their production, it would be facile to merely assess what the farmers have done with their project-financed tools and compare them and their situation with farmers who did not receive project-financed tools. However, this is not so. The Bank-supported projects did not provide tractors, etc. but financing, expanding temporarily the farmers' liquidity position and purchasing capability, which they used to buy particular capital goods. Therefore, ex-post evaluation of agricultural credit projects is much more complex: what has to be established is not whether farmers with tractors, wells, etc. perform better and are better off than farmers without such tools, but whether a particular set of farmers with a momentarily enhanced purchasing capability 5 to 10 years ago (the partici- pants or "beneficiaries") are better off today and generate more production than those farmers who did not have their purchasing capabilities enhanced at the same time. 2. The problem is, however, more complex since farmers who applied for, and obtained, Bank-supported sub-loans might instead have obtained credit from other sources, or might have mobilized resources on their own. Thus, obtaining sub-loans may have substituted for other sources of financing. This is not just a theoretical possibility. On the one hand, participating farmers must be creditworthy. On the other, the audits of this particular set of projects contain evidence of alternative sources of funds being available for financing project-related investments (see para. 3.13d and 3.63). 3. Not only the farmers but also the projects' financial intermediaries had access to alternative sources of funds. For example, LDB could raise funds through floating ordinary debentures, which could be subscribed by the Life Insurance Corporation of India, RBI, CBs, State governments, SLDB, and central and other cooperative banks in the State. Due to a change in develop- ment objectives and in agreement with RBI and with explicit Bank support, ARDC progressively curtailed LDB's freedom to utilize such funds raised through ordinary debentures and eventually forbade LDB financing minor irrigation projects through ordinary resources. 4. Therefore, to a probably large but unquantifiable extent, project funds substituted for funds already available for financing tractors, wells and pumpsets. "Benefits" generated by such portion of project funds cannot legitimately be attributed to the projects. A significant part of the project - 44 - beneficiaries would have bought their tractors or pumpsets or sunk their wells by recourse to other sources of financing. The impossibility of quantifying the amount of this substitution makes it impossible to know by how much return estimates for the project should be changed. As stated in the impact evalua- tion report on the Kenya First Smallholder Agricultural Credit Project, "the counterf actual element - what would have happened if there were no project - required in any analysis renders impossible the search for definite conclu- sions. It is simply not possible to know what effects the injection of [US$XX million] of projects funds had on the rural economy"l/. 5. Passage of time introduces another complexity in evaluating credit projects. The evaluator has to assess whether there is a significant dif- ference in production, employment and income between two sets of farmers, one who had their sources of funds temporarily expanded 5 to 10 years ago and another who had not. The impact of this temporarily enlarged purchasing capacity becomes blurred over the passage of time. It may be impossible to differentiate and eventually attribute any such difference to a particular financial intervention, say, ten years ago. Other intervening factors may obscure the project's effects, rendering the project's financial intervention "untraceable" after some years. B. Limits of Comparing Project Participants and "Control" Group Farmers 6. A second general methodological problem in evaluating agricultural credit projects is comparability. For proper evaluation, project benefi- ciaries ought to be compared with a group of similar farmers who did not have access to the project-provided funds ("control" group). In some cases this may be possible. For example, in the Punjab, it is possible to draw a sample from the 8,000 farmers who obtained a project-financed tractor and another from the 19,000 who did not, because both are randomly generated subsets of the 27,000 farmers who applied2/. Thus, it is possible to compare two groups of farmers with essentially similar characteristics3!. I/ OED, Kenya: Impact Evaluation Report - First Smallholder Agricultural Credit Project, Report No. 2968, May 5, 1980, paragraph 17. 2/ Tractor allocations under the project were made by lottery. 3/ It should be stressed again that some of the non-participating farmers would have bought a tractor (or whatever good a project is financing) anyway, utilizing other sources of funds. Thus, there may not be dif- ference in production, employment or income between farmers getting their tractors through IDA/ARDC financing and farmers financing theirs with alternative sources of funds. In strict terms, this would be the real test of the credit project's impact: whatever significant difference is perceived between the two groups would have to be attributed to the project (say, technical assistance accompanying the sub-loans; wider selection of brands; access to goods - usually imported - not available to non-participants; longer terms or grace periods; lower rates of interest, etc.). - 45- 7. In most other agricultural credit projects, however, the farmers who obtain project sub-loans are qualitatively different from the rest of the farmers. Usually, participating farmers are more "entrepreneurial" than their colleagues, are more risk-prone and may have adopted new technologies earlier. Most likely, they have stronger links with financing agencies, as well as with input supply, marketing and technology agencies, selling more to commercial and processing enterprises and buying more modern inputs. Usually they identify - either by themselves or with the help of technical agents or commercial dealers - their need for additional machinery (or whatever the credits are financing), become aquainted with the existence of the project funds, and approach the on-lending banks. Even without any project, many of them might have managed to secure some financing to buy the necessary tractors or pumpsets. For most, the projects may have been just one of several possi- ble sources of funds, which happened to be, at the time, more convenient (on whatever grounds) than the alternatives. Furthermore, most of the farmers who applied for project sub-loans may have taken other loans (for farm, household or personal development) previously and subsequently. Evidence from this particular set of projects is fragmentary, but there are clear indications that some farmers, and particularly the larger farmers had pre- viously borrowed for farm development investments. Some of the medium and large farmers visited by the audit mission had owned tractors before obtaining the project-financed units; many already had a well, which was enlarged or improved under the project, or a pumpset, which was replaced. Some also showed evidence of other investments financed by non-project funds. 8. Therefore, any "without-project situation" established ex-post on the basis of random sampling of project beneficiaries and non-beneficiaries would be biased, for even without the project most who became project benefi- ciaries are likely to have developed faster and would be better off than their "control" colleagues. Evaluations of agricultural credit projects are thus inherently biased in favor of the projects because project participants and nonparticipants are likely to be different types of farmers (or farmers in a different stage of development)!/. The wider the project-s coverage, the more serious the evaluation problem. If project funds were restricted and satisfied only part of the existing demand, many "entrepreneurial" farmers would remain non-participants and could be found within a possible "control" group_l. But if project funds were abundant enough and on-lending took place over sufficient time most farmers with entrepreneurial characteristics 1/ The PPAR on Uttar Pradesh explicitly raises this issue: .control" farmers were most likely selected from less progressive farmers, result- ing in possible underestimate of "without-project" production and over estimate of incremental production and rates of return at completion. 2/ In theory, at least. To identify those more "entrepreneurial", non- participating farmers and design a sample of them may be practically impossible in most cases. - 46 - would have become participants. Farmers remaining as non-participants, and thus available to be sampled as members of the "control" group, would thus be farmers of a different kind. This is the case, for instance, of the farmers who borrowed for minor irrigation development in Tamil Nadu. 9. Further, farmers active enough to have sought a development credit for their farms are also more likely than their neighbors to have been involved in non-farm activities. The term "farmer involvement in non-farm activities" readily calls to mind the image of a large Latin American farmer or rancher with investments in industry, buildings, stock, and so on. But it is also fairly common for many small farmers in, say, South and East Asia, to be variously involved in small local trading, local transporting, lending, handicrafts or cottage industries. Sometimes it is the farmer's spouse (either wife or husband) who is involved in these activities. Therefore, they are more likely to transfer funds between farm and non-farm activities, as well as having a higher and more rapidly increasing income. This blurs further the assessment of the projects- impact on the farmers- productive capabilities and makes the task of constructing the "without project situa- tion", more difficult. C. The Problem of Attribution 10. It is mentioned in the main report (para. 4.03) that one reason for the projects' success is that they provided the precise inputs which were required for the successful introduction of a profitable new technology. How much of the observed benefits must be attributed to the credit project which provided those inputs, and how much to the research, development and extension efforts which developed and made available the Green Revolution-s technical package is difficult to identify. This problem has already been raised and discussed elsewhere!/. Computing the "right" share of benefits to be attri- buted to each of a set of activities would call for methods of economic evaluation more complex than the monodimensional methods currently utilized in the Bank. CPS has issued instructions - and warnings - on the methodologies to be utilized when separate project components are interdependent (CPN No. 2.01 of June 27, 1977; paras. 34 - 38). As mentioned already in the PPAR on 1/ For instance, attribution of benefits associated with the Upper Pampanga irrigation project in the Philippines between the development of high yielding varieties, Bank-supported credit projects (the Rural Credit series), a Government-supported input delivery system, and the irrigation project itself. See PPAR, Philippines Upper Pampanga Irrigation Project, Report No. 3063, June 30, 1980, para. 30 to 33. Similar problems had been discussed previously in the audit reports on other irrigation projects which are quoted in the footnote of paragraph 33. - 47 - Upper Pampanga, it appears worthwhile for CPS to explore further the case of interdependent and mutually reinforcing projects (some of which may not be Bank-supported) and issue appropriate guidance.l! D. The Farm-Model Approach 11. A fourth methodological problem which affects rate of return reesti- mates on agricultural credit projects derives from the way in which the Bank computes these rates. The standard procedure consists of preparing farm models, which show the situation "before" the project and "after" the project, and indicate the probable effects of the project-supported investments on the farm costs and benefits. The financial rate of return is computed for the farm investment. Adjustments are then introduced into the cash flows used to compute the financial rate of return to take into account factors such as shadow prices, taxes, subsidies, interest rates, etc. Through this procedure the investment's economic rate of return is computed. Then, this economic rate of return is implicitly assumed for the whole universe of sub-borrowers. This method may be acceptable for appraisal estimates, when actual "with- project" data is obviously not available, but is clearly inadequate for completion estimates. The main flaw is the implicit assumption that each and every sub-borrower would complete the investments financed under the loan or credit, and that all did, within the same time span, yield similar benefits. This is far from true. Some farmers never complete the investments which were financed under the project. Others complete them only after substantial delays. Others complete the investment but do not utilize them, or they do so over a much longer period of time and, therefore, forego some immediate benefits. It is these immediate benefits which are most important for the computation of rates of return, particularly when the rates are relatively high. Also, the Bank's approach ignores administrative costs; the cost of administering the sub-loans as well as the more general overhead costs devoted to supervision and control. 12. There is substantial evidence in this set of projects of both kinds of problems. On the first, there are a number of failed investments; speci- fically, between 1 and 5% of the wells sunk or enlarged under the projects failed to produce an adequate supply of water. But far more important is that a substantial number of wells were reported as incomplete at the end of the project-s life (see paras. 3.74-76). 13. Information on non-completion of project-supported wells over a longer time span (two years after credit closing) comes from one project only. If this project, however, is generally representative, up to one-third of the wells financed may have never been completed, either because of misallocation 1/ The Bank's Development Research Center has initiated a study "Planning and Designing Investments - Indus Basin" which also deals with attribu- tion aspects. - 48 - of funds or because of technical difficulties. The farm model thus cannot serve the whole project. Completion estimates of economic and financial returns of agricultural credit projects should be based on actual surveys_v. 14. Even in farm models, no account is taken of any delays which might have occurred during project implementation. Substantial delays would have serious effects on other kinds of agricultural projects (e.g., irrigation projects). They are not reflected in the farm models, however, and therefore, are not reflected in the projects' economic rate of return either. 15. Inclusion of overhead costs in cost computation for economic return analysis constitutes a gray area in Bank methodologies. There is no explicit indication in the Operational Manual nor any standard practice at appraisal on whether, when and how to include a proper share of the overall costs of the ministries and agencies involved in a project. Implementation of a project may require that additional positions be established in a ministry or agency, that existing positions be upgraded, or that new agencies or departments be established as a requirement for project implementation. In this particular case, ARDC had to expand its staff and new regional and state offices were established or strengthened. The land development banks were strengthened through an extensive training program which covered about 600 medium and senior staff and about 6,000 junior staff members. Commercial banks had to recruit agricultural technicians to cover what, for them, was a completely new field of lending. 16. In some cases, these additional administrative costs are included at appraisal in the project cost tables; less frequently are they included in the cost flows utilized while computing the appraisal's economic rate of return. Very seldom, if ever, have these strictly project-related costs been computed in retrospect after project completion and taken into consideration when reestimating the rate of return that the project is likely to yield. Further- more, the administrative operational costs of the project are not included, i.e., the cost of physically preparing, reviewing and assessing applications; approving the sub-loans; and, disbursing and supervising sub-project implemen- tation. As mentioned in the report (para. 3.48), the cost of delivering credit to farmers in India may be in the vicinity of 5% of the total funds on-lent. 1/ This does not mean that surveys have not been carried out at completion. However, surveys were carried out only on a small sample just for the purpose of providing updated estimates for the parameters for the farm models. They were not intended to be - and are not - representatives of the project as a whole. Table 1 AGRICULTURAL CREDIT PROJECTS IN INDIA MAIN CHRONOLOGICAL DATA Project Name Credit Number Credit Agreement Date Closing Date PPAR Date Gujarat 191-IN June 3, 1970 June 30, 1975 October 4, 1976 Punjab 203-IN Sept. 4, 1970 June 30, 1977 October 5, 1979 Andhra Pradesh 226-IN Jan. 8, 1971 July 1977 October 5, 1979 Haryana 249-IN June 11, 1971 June 30, 1977 October 5, 1979 Tamil Nadu 250-IN June 11, 1971 Dec. 31, 1977 October 5, 1979 Karnataka 278-IN Jan. 7, 1972 June 30, 1977 April 7, 1980 Maharashtra 293-IN March 29, 1972 June 30, 1976 October 5, 1979 Madhya Pradesh 391-IN June 8, 1973 Dec. 31, 1977 October 25, 1979 Uttar Pradesh 392-IN June 8, 1973 Dec. 31, 1979 July 24, 1980 ARDC I 540-IN April 28, 1975 Dec. 31, 1977 October 25, 1979 Table 2 AGRICULTURAL CREDIT PROJECTS IN INDIA MAIN QUANTITATIVE DATA Total Project Costs Time Under- Economic Credit Credit - At - At Com- run or Over- Rate of Number Amount Appraisal pletion Overrun run Return US$M US$M US$M % $ % Gujarat 191-IN 35.0 0 Punjab 203-IN 25.5 40.0 42.7 25.5 200 over 50 Andhra Pradesh 226-IN 24.4 24.4 24.4 0 37 Haryana 249-IN 25.5 44.5 53.0 25.5 66 over 50 Tamil Nadu 250-IN 35.0 62.0 62.0 24.0 0 about 35 Karnataka 278-IN 40.0 70.4 74.5 40.0 55 21 - over 50 Maharashtra 293-IN 30.0 53.4 57.0 30.0 0 17 - over 50 Madhya Pradesh 391-IN 38.0 60.3 69.9 30.0 0 41 - over 50 Uttar Pradesh 392-IN 38.0 72.5 79.8 38.0 38 25 - 39 ARDC I 540-IN 75.0 168.5 179.0 75.0 0 18 - over 50 Table 3 LAND DEVELOPMENT BANKS: ALL-INDIA LENDING OPERATIONS (Amounts in Ra million) I Increase in Loans I ! Loans Advanced ! Loan Out- i ! Outstanding I ARDC Disbursements Years /a ! During Year ! standing I in % of to LDBs ! In Nominal I In Real I end of I in ! Loans I ! Under IDA Terms 1 Terms /h /c! Period ! A/B ! Amounts I Advanced Total !Projects /d D/A :/A 1 (A) I (B) ! (C) I (0) 1 (E) 1967/68 N/A ! 3,560 1 ! 1968/69 1,900 1 2,145 ? 5,060 0.355 i 1,500 ! 78.9 ! N/A - 1969/70 1 1,900 2,034 6,220 0.305 1,160 ! 61.1 267.5 - 0.148 1470/71 ! 1,940 ! 1,940 ! 7,250 0.264 1,030 53.1 266.6 ! - 0.137 1971/72 ! 1,457.5 ! 1,542 7,670 0.212 420 ! 25.8 ! 283.9 53.7 0.174 0.033 0.218 1972/73 1,714.3 ! 1,499 8,492.1 0.202 822 46.1 ! 861.4 635.8 0.502 0.371 0.773 1973/74 1,467.0 ! 1,122 9,140.5 ! 0.160 648.4 46.4 777.6 529.2 ! 0.530 0.361 0.816 1974/75 1,887.3 1,147 9,925.4 0.191 ! 784.9 41.5 770.6 519.8 ! 0.408 ! 0.275 0.66i 1975/76 2,049.4 1,163 ! 10,687.0 ! 0.192 ! 761.6 ! 37.2 i 990.0 906.9 1 0.483 0.442 1.191 1976/77 2,487 ! 1,379 ! 12,123.9 ! 0.205 ! 1,436.9 ! 57.7 ! 1,267.0 ! 1,005.3 ! 0.509 ! 0.404 0.793 1977/78 2, 1,264 13,150.0 0.178 ! 1,026.1 43.7 1,119.4 858.0 0.477 0.365 0.836 Totals ! 13,583 ! 5,900 ! 5,900 6,070.8 4,508.7 0.447 /e 0.332 /e 0.743 /e 6/30/71 - ! 6/30/78 ! Source: 1967/68 to 1971/72 - Figures provided by ARDC. 1972/73 to 1976/77 - ARDC/COL - Preparation Report for a proposed Third ARDC Project, June 1978 (figure for 1976/77, as amended by IDA in SAR of ARDC III, Annex 5, Table 1). 1977/78 - IDA - SAR uf ARPC IIH; Report No. 2404a-IN; May 16, 19?9; Annex 5, Tabie 1 /a End of Periods: as of June 30. /b Base 1970/71 - 100; Deflator: Implicit Price Deflator of Gross Domestic Capital Formation, as reported in IBRD - Economic Situation and Prospects of India, Report No. 2431-IN, April 9, 1979, Table 6.12. /c Figures provided in GOI comments (Annex 1, page 6) conflict with some of the figures in this table, but they confirm the decline of LDB lending in real terms: Amount in Amount in Year Nominal Terms Real Terms 1969/70 Rs. 1555 M Rs. 1665 M 1970/71 Rs. 1706 M Rs. 1706 M 1977/78 Ra. 2781 m Rs. 1497 M /d Disbursements under ARDC I in the period 1974/75 to 1976/77 slightly overlap disbursements under the late statewide projects. The overlapping refers to the aggregate, all-India figures. There is no overlapping in either individual State; ARDC I funds were utilized only after the state credit project funds had been exhausted. Disbursements of ARDC I and 11 overlap in 1977/78. /e Average for period 6/30/71 to 6/30/78.