Document of The World Bank FOR OFFICIAL USE ONLY Report No. 16848-IN THE IMPACT OF DAIRYING DEVELOPMENT IN INDIA: THE BANK'S CONTRIBUTION THE IMPACT OF: KARNATAKA DAIRY DEVELOPMENT PROJECT (CREDIT 482-IN) RAJASTHAN DAIRY DEVELOPMENT PROJECT (CREDIT 521-IN) MADHYA PRADESH DAIRY PROJECT (CREDIT 522-IN) NATIONAL DAIRY PROJECT (CREDIT 824-IN) NATIONAL DAIRY H PROJECT (CREDIT 1859-IN/LOAN 2893-IN) June 30, 1997 Operation Evaluation Department 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. Currency Equivalents (annual averages per US$1) Currency Unit = Rupees (Rs) Year Officiala Marketb Reaf 1975 8.38 9.75 17.00 1976 8.96 10.10 16.95 1977 8.74 9.60 16.61 1978 8.19 9.65 15.66 1979 8.12 9.10 16.63 1980 7.86 16.99 1981 8.66 17.01 1982 9.46 17.18 1983 10.10 13.40 17.94 1984 11.36 15.00 18.46 1985 12.37 13.85 19.13 1986 12.61 14.15 19.84 1987 12.96 14.55 20.89 1988 13.92 17.00 21.73 1989 16.26 19.00 22.50 1990 17.50 19.80 23.95 1991 22.74 28.00 26.19 1992 25.92 32.30 28.10 1993 30.49 31.45 29.90 1994 31.37 31.60 31.61 1995 34.95 1996 34.95 a. "International Financial Statistics," IMF, Vol. XLVIII, 1995. b. Pick's Currency yearbook, World Currency Yearbook and Currency Alert, various issues. c. (1994 Exchange Rate X GDP Deflator India)/GDP Deflator US Units and Conventions Dates = dd/mm/yy 100 liters of milk = 102 kilograms Director-General, Operations Evaluation : Mr. Robert Picciotto Director, Operations Evaluation Department : Mr. Roger Slade for Ms. Elizabeth McAllister Division Chief, Agriculture and Human Development Division Mr. Roger Slade Task Manager : Mr. Wilfred V. Candler Abbreviations and Acronyms FOR OFFICIAL USE ONLY Al artificial insemination AMUL Anand Milk Union Limited ARC Agricultural Refinance Corporation BCT Bhagavatula Charitable Trust DAHD Department of Animal Husbandry and Dairying DANIDA Danish Agency for International Development DCS Dairy Cooperative Society (village level cooperative) EC European Community EPC Effective protection coefficient ERR Economic rate of return FRAC Food and Research Analysis Center GDP Gross Domestic Product IAS Indian Administrative Service ICDP Intensive Cattle Development Project ICR Implementation Completion Report IDA International Development Association IDC Indian Dairy Corporation IER Impact Evaluation Report IFPRI International Food Policy Research Institute IMF International Monetary Fund IRMA Institute for Rural Management, Anand KDDC Karnataka Dairy Development Corporation MIS Management Information System MMPO Milk and Milk Products Order MPDDC Madhya Pradesh Dairy Development Corporation MPU Milk Producer's Union (Regional Cooperative Society) MUV Manufactured Goods Unit Value Index NDDB National Dairy Development Board NDP National Dairy Project NDPII Second National Dairy Project NEP New Economic Policy NGO Nongovernmental organization NPC Net protection coefficient ODA Overseas Development Corporation of the United Kingdom OED Operations Evaluation Department (of the World Bank) OF Operation Flood OFI Operation Flood I OFII Operation Flood II OFIII Operation Flood III PCR Project Completion Report PAR Performance Audit Report RDDC Rajasthan Dairy Development Corporation RERR Reestimated Internal Economic Rate of Return Rs Rupee SAR Staff Appraisal Report 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. SDR Special Drawing Rights SEWA Self-employed Women's Association SFDA Small Farmer Development Agencies SMP skim milk powder SNF solids-not-fat TMDD Technology Mission for Dairy Development US United States of America WDCS Women's dairy cooperative society WFP World Food Program WID Women in Development The World Bank Washington, D C. 20433 U.S.A Office of the Director-General Operations Evaluation June 30, 1997 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: The Impact of Dairying Development in India: The Bank's Contribution Karnataka Dairy Development Project (Credit 482-IN) Rajasthan Dairy Development Project (Credit 521-IN) Madhya Pradesh Dairy Project (Credit 522-IN) National Dairy Project (Credit 824-IN) Second National Dairy Project (Credit 1859-IN/Loan 2893-IN) Sector Impact Evaluation Attached is an impact evaluation report on the five dairy projects in India named above. The first credit, for the state of Karnataka, was approved in FY74 for US$30 million; two further state projects followed in FY75 for US$27.7 and US$16.4 million. The first national credit was approved in FY78 for US$150 million, followed by a second national project in FY88, supported by a credit of SDR 121.2 million (US$150 million, equivalent) and a loan for US$200 million. The evaluation is based upon three visits to India and field work in five Indian states by Operations Evaluation Department (OED) staff, two participatory surveys and a study of the protection of the Indian dairy industry, an examination of Bank files and documents, discussions with the responsible Bank staff, government representatives, academics, and the staff of the National Dairy Development Board (NDDB) in India. A draft of the study was discussed at a two-day workshop at the Institute for Rural Management, Anand, Gujarat state in March 1997. The generous assistance of the Swiss Development Corporation in financing the workshop is gratefully acknowledged. Operation Flood is an Indian program to foster farmer-controlled cooperatives capable of paying an attractive price for very small deliveries (1 to 2 liters) of milk. It owes its genesis to a visit in 1964 by the then Prime Minister of India, Lal Bahadur Shastri, to Anand in Gujarat state. He was so impressed with the operation of the cooperative Anand Milk Union Limited (AMUL) that he decided to support replication of the AMUL model nationwide. This expansion (Operation Flood 1) was supported directly by the World Food Program and indirectly by the three Bank-funded state projects. A subsequent visit to Anand by the Bank's then President, Mr. McNamara, resulted in an offer to support a further expansion of the program (Operation Flood II) with IDA funding through the National Dairy Project. This expansion was also supported by extensive food aid from the European Community (EC). National Dairy II, again with support from parallel EC food aid, funded a third phase (Operation Flood III). Inauguration of Operation Flood was accompanied by an extraordinary jump in the growth rate of total Indian dairy production from 0.7 percent cumulative to 4.7 percent. This is a growth rate for the dairy industry quite comparable with the growth for wheat and rice production under the Green Revolution. Technological change underlying this growth rate included increased by-products from the Green Revolution crops and a switch to crossbred cows based on semen from European dairy breeds. It 2 also reflected government withdrawal from direct collection and processing of milk and assurance that food aid would not be used to depress dairy prices. Operation Flood only procures about 6.3 percent of national production so that it is the policy changes which applied to the whole industry, as well as rising incomes and demand for milk and the technological changes mentioned above which explain the faster growth rate. The higher growth rate in production underpinned an increase in milk consumption from 107 to 193 grams/head/day between 1970 and 1994. Operation Flood itself occupied a leadership position within the industry. By 1996, Operation Flood was procuring milk from 6.3 million farmers organized into 53,000 farmer-controlled village-level Dairy Cooperative Societies, processing it through 170 Milk Producers Unions grouped into 22 State Dairy Federations and supplying over 100 urban centers through a national milk grid. Although itself a narrowly focused marketing intervention (with some support for farm-level production), Operation Flood had wide ranging beneficial impacts on milk consumption, farm incomes, employment generation, education, and women. It was an innovative intervention in that it threw the Bank's support behind a policy change which was already under way and to an existing Indian organization, the NDDB. This contrasts with many Bank projects, particularly in Africa, which call for the creation of an institution which is also expected to get the project up and running. The major difficulty encountered by Operation Flood has been the reluctance of state governments to cede full control of the industry to the farmer cooperatives. State governments, which in some cases own the processing plants used by the cooperatives, interfere by setting producer or consumer prices, making key managerial appointments, and mandating continued overmanning of dairy plants. Progress has been made in reducing this interference, which varies greatly from state to state, but it is a continuing problem. Bank plans for dairy development in India no longer include special assistance to NDDB and Operation Flood. Contents Preface ...... ................................................... v Executive Summary ......................................... ....... I 1. Bank Dairy Projects....................................... ........ 9 Design Features of Bank Dairy Projects.. .................................. 9 Project Context ........................................... ....... 10 Literature Review ......................................... ........ 13 Report Structure .................................................. 18 2. Project Design......................................................19 Karnataka......................................................19 Rajasthan......................................................20 Madhya Pradesh.................................................21 The National Dairy Project..........................................21 The Second National Dairy Project....................................23 Changed Paradigm................................................ 24 3. Impact of the First Four Projects at Completion...........................25 Targets................................................... 25 Organization................................................ 25 Impact.................................................... 26 Milk Production................................................. 27 Milk Consumption .......................................... ...... 28 Importance of Milk Income .........................................29 Profitability of Dairying........................................................................ 30 Benefits .......................................................................................... 31 Summary......................................................32 4. Developments Under National Dairy H ......................................................... 35 Expansion and Nonfunctional Cooperatives..............................35 5. Village-Level Impact ................................................................................ 51 Karnataka ........................................................................................ 51 Madhya Pradesh ................................................................................ 53 6. Project Impact ......................................................57 Milk Production..................................................57 Supply or Demand Shift?....................................................................... 61 Economic Performance ......................................................................... 66 Exploration of the Free Trade Option...................................69 Industry Structure............................................................................... 70 Farmer Incomes ..................................................73 Credit......................................................... 73 Women........................................................ 73 Employment Generation............................................ 75 Improved Technology............................................. 75 ii Adulteration ....................................................76 Education ...................................................... 77 Cooperative as NGO ......................................... .....78 7. Findings and Recommendations .................................... .....79 Findings.......................................................79 Recommendations................................................ 85 Lessons ........................................................ 79 References ........................................................... 89 Annex A: Estimation of Economic Impact............ ................. .....99 Annex B: Estimation of the Cost of Operation Flood ........................... 103 Annex C: Estimation of Producer and Consumer Surplus ................. ...... 129 Annex D: Estimation of the Impact of Free Trade in Dairy Products .......... ......... 135 Annex E: Borrower Comments ............................................... 141 Tables 1.1: Bank-supported Loans and Credits with "Dairy" in the Title.................9 3.1: DCSs and MPUs Formed........................................... 25 3.2: Karnataka: Daily Household Milk Production.................. ......... 27 3.3: Madhya Pradesh: Average Milk Production and Disposal ................... 28 3.4: Madhya Pradesh: Average Milk Prices, Landless and Large Farmers...................... 28 3.5: Karnataka: Home Consumption of Liquid Milk........................................................ 29 3.6: Karnataka: Ratio of Milk Sales to Total Expenditure............................................... 30 3.7: Madhya Pradesh: Milk as a Proportion of Average Annual Income. ....................... 30 3.8: Madhya Pradesh: Gross Margins in Milk Production .......... ............ 31 3.9: Madhya Pradesh: Opinions on Impact of DCSs on Beneficiaries............................ 31 3.10: Madhya Pradesh: Farmers' Opinions of the DCSs and Their Reason for Joining ..... 32 4.1: Formation of Dairy Cooperative Societies ........................ ..... 35 4.2: Milk Procurement, Production, Cooperative Membership........................................ 38 4.3: Water in 200 Milk Samples Submitted for Testing ........ .............................. 40 4.4: Loan Repayments on Bank Projects to NDDB .......................... 45 4.5: Profitability of the Cooperative Sector.......................................... 46 4.6: Revised Effective Protection of the Indian Dairy Industry........................................ 47 4.7: Indian and World Market Prices for Selected Dairy Products, 1991-95 .................. 48 5.1: Survey Results: Need for Veterinary Services . .......................................... 55 5.2: Survey Results: Sources of Veterinary Services ........................ 55 6.1: Milk: National Production and Operation Flood Procurement and Sales............. 60 6.2: Nominal and Real Processor Prices for Cow Milk.......................... 65 6.3: Price and Production Changes in Wheat, Rice, and Milk .................. 65 6.4: Estimates of Economic Benefits of Bank-Assisted Projects ....... .......... 66 6.5: Impact Estimates of Cost of Operation Flood............... ............... 67 6.6: Alternative Estimates of Capital Costs of Operation Flood per Beneficiary in Constant 1996 US$ ....................... ..... 68 6.7: Alternative Estimates of Annual Gross Benefits Under Operation Flood and Policy Change, 1996............. .......... 69 6.8: Costs and Benefits of Alternative Policies for Urban Milk Supply, 1996 .............................................. 70 iii 6.9: National Milk Procurement by Operation Flood, Other Organized, and Informal Sectors. .....................................71 6.10: Punjab: Milk Samples Drawn and Found Adulterated ....................77 6.11: Progress Towards Anand Principles............................. 78 7.1: Projected Source of Expenditures for Operation Flood III, 1992-93 ................... 85 Boxes 1.1: Early Bank Interest in Operation Flood........................ 12 3.1: Establishment of the Anand-Pattern Structure ........................ 26 4. 1: Milk Assembly at its Worst .............................. .....41 4.2: Artificial Milk ........................................... 42 6.1: A Level Playing Field? .......................... 72 Figures 4.1: India: Operation Flood Milk Procurement, 1970-95 ............................................ 39 6.1: India: M ilk Production, 1951-94 (Linear Fit)...................................................... 58 6.2: India: Milk Production, 1951-94 (Exponential Fit).............................................. 59 6.3: Time Series of Total Milk Production and Real Price of Milk . ............... 63 6.4: Demand and Supply Shifts .................................... 64 7.1: India: Milk Production, 1951-94, and Operation Flood Milk Procurement, 1980-94...................... 84  V Preface Operation Flood, which was supported by the five projects now evaluated, is a huge undertaking. In 1996, it involved 9.3 million farmer-members supplying an average of 10,900 metric tons of milk per day through 55,042 functional village cooperative societies to 170 milk producers unions (MPUs) who sold it as liquid milk and processed products. This vast organization grew out of a single small cooperative society in the state of Gujarat established in 1946. This growth has been supported by US$0.69 billion (1996 dollars, 43 percent of project cost), disbursed by the Bank and, and US$1.1 billion (1996 dollars) of food aid from the World Food Program and the European Community. Even so, Operation Flood represents only 6.3 percent of total milk production and 22 percent of marketed milk in India. Total milk production has also grown rapidly since the inception of Operation Flood. In India's dairy industry, the Bank has followed a simple, consistent, and transparent development strategy: to support the expansion of dairy production by small producers through a successful indigenous development program. This program, despite substantial government assistance, enabling legislation, and government control of some apex organizations, is now dominated by fanner-controlled village-level dairy cooperative societies (DCSs) and some regional MPUs. An initial IDA credit of US$30.0 million was approved in March 1974, and two further IDA credits of US$27.7 and US$16.4 million were approved in December 1974, all supporting projects in various states. A fourth IDA credit of US$150.0 million approved in June 1978 supported a first national project, and a further IDA credit of SDR 121.2 million and loan of US$200.0 million approved in December 1987 supported a second national project. The first four projects were audited by the Operations Evaluation Department (OED) in 1987, and studies were sponsored by the Bank, the Danish International Development Agency (DANIDA), and the International Food Policy Research Institute (IFPRI) on the impact of the Bank-supported projects in Karnataka (Alderman 1987) and Madhya Pradesh (Mergos and Slade 1987). The final project, which closed on April 30, 1996, was audited in November 1996. This report is based on the above-mentioned audits and studies, the associated Project Completion Reports and Implementation Completion Reports, examination of other Bank documents, and interviews with Bank staff. OED Missions also visited India in January and November 1996. As part of the study, a participatory farmer survey was undertaken in Karnataka, and a resurvey of villages, originally studied in 1983, was organized in Madhya Pradesh. A study of the level of protection of the Indian dairy industry was also undertaken. A workshop at the Institute of Rural Management, Anand, Gujarat state, with 43 Indian participants was held to discuss a draft of this report in March 1997. The generous assistance of the Swiss Development Corporation in financing this workshop is gratefully acknowledged. The valuable assistance provided to OED staff in the preparation of this report by the government of India and the state governments of Karnataka, Rajasthan, Gujarat, Bihar, and the Punjab is gratefully acknowledged. The hospitality and generous commitment of time by the staff of Operation Flood, the National Dairy Development Board (NDDB), and other participants Vi in the Indian dairy industry too numerous to mention also contributed greatly to the success of the evaluation. A draft version of this impact evaluation report was sent to the government of India, the governments of Karnataka, Rajasthan, Gujarat, Bihar, and the Punjab, and the NDDB for comments, and these are attached in Annex E. Where appropriate, the text has been changed to reflect these comments. 1 Executive Summary Operation Flood is not an all purpose poverty removal program. It cannot be, because it focuses clearly on a single productive activity, dairying, while the ranks of the rural poor include many different categories of the disenfranchised: the old, the infirm, the tribals, the landless, the small farmers, the artisans and so forth. (Similarly, OF) is not an all purpose development program, aimed at removing economic and social inequalities existing in rural India for centuries at one stroke. Dr. V. Kurien (quoted in Doornbos and Nair 1990) 1. The first of the five Indian dairy projects included in this study was approved by the Board in March 1974. At that time India was following a self-reliant macroeconomic development policy with heavy emphasis on import substitution (a policy which remained essentially unchanged until 1991). In the dairy sector, a policy decision had been made to utilize farmer-controlled cooperatives to develop the dairy industry and to make the import substitution as efficient as possible. The government also decided to withdraw from its own direct efforts to develop the dairy industry (indirect efforts via extension, research, artificial breeding, etc., continued). Dairy commodity food aid was to be sold at commercial prices with the proceeds earmarked for support of this Operation Flood. 2. The inauguration of Operation Flood corresponded to a turnaround in per capita incomes which had been falling to a period of sustained, if modest, growth. Rising per capita incomes, together with rising population and a high income elasticity of demand for milk, resulted in a rapid growth in the demand for milk (technically, a rapid shift in the demand function for milk). In the absence of supply side adjustment, this would have led to a rapid escalation in the price of milk or the need for extensive imports. In fact, the shift in the supply function from the policy changes introduced with Operation Flood resulted in an even larger shift in the supply (function) and declining real prices. 3. Operation Flood itself is an Indian program to foster a farmer-controlled cooperative dairy industry capable of paying an attractive price for very small deliveries (1 or 2 liters) of milk. The program owes its genesis to a visit in October 1964 by the then-Prime Minister, Lal Bahadur Shastri, to Anand in Gujarat state. The Prime Minister was so impressed with the operations of the cooperative Anand Milk Union Limited (AMUL), and so disenchanted with the performance of various state-sponsored schemes to promote the dairy industry, that he asked Dr. V. Kurien, the Managing Director of AMUL, to replicate the AMUL model throughout India. The result was Operation Flood I (OFI), a program supported directly by the World Food Program (WFP) and indirectly by three IDA-funded projects in the states of Karnataka (Cr. 482), Rajasthan (Cr. 521), and Madhya Pradesh (Cr. 522). 4. A subsequent visit to Anand in 1978 by the Bank's then-President, Mr. McNamara, resulted in an offer to support an extension of the program with further IDA funding. At the same time, the European Community (EC) decided to use a major portion of its dairy surplus to support Operation Flood directly rather than indirectly through the WFP. The result was Operation Flood II (OFII), funded by the National Dairy Development Board (NDDB) of India, a parastatal; the Bank (through the National Dairy Project (NDP), Cr. 824); and the EC through food aid. The earlier Bank-supported state projects continued as an integral part of OFII. On 2 completion of OFI, the project involved 22 state cooperative federations working through 164 milk producers unions (MPUs). Subsequently, Operation Flood III (OFIII), supported by the Second National Dairy project (Cr. 1859/Ln. 2893), was designed to increase the number and coverage of MPUs and DCSs with the policy objective of ensuring full farmer control at the village, union, and federation levels. Farmer control was to include the right to set producer and consumer prices, hire and fire staff, and go bankrupt. 5. Very substantial financial assistance has been provided to NDDB and Operation Flood through food aid provided on concessional terms but sold by Operation Flood at market prices. This has been used to establish processing infrastructure (including a 30 percent grant element offered by NDDB in connection with loans for processing plants and equipment), for developmental expenditures, and to provide public goods (and some social expenditures). Though larger than normally offered in the private sector, this financial assistance does not differ in principle from the tax holidays and infrastructural investments typically offered to private companies in developed and developing countries to influence their locational decisions. Operation Flood itself has studiously avoided subsidizing recurrent costs in order not to undermine the independence of the cooperatives. However, where states have retained some control over their cooperatives, quite extensive subsidies (or reimbursements) have been provided by state governments for losses incurred. IDA funds were on-lent by the central government to the NDDB at an interest rate of 7.5 percent and from the NDDB to the MPUs at 10 percent (approximate commercial terms). Inflation has substantially eased the burden of these loans. 6. From Independence until about 1971, dairy production grew at about 0.7 percent a year. Since then, it has grown at more than 4.7 percent a year. As shown in Figures 6.1 and 6.2, there is a clear break in the trend line for dairy production. Continuing the earlier rate of growth would have resulted in the production of only 24.2 million metric tons of milk in 1995, compared with the estimated actual production at the higher growth rate of 66.3 million metric tons. This is an increase in the production of dairying as a whole. Operation Flood too has grown fast: from 1970-71 to 1985-86, procurement rose from 0.14 to 2.96 million tons per year. Since 1985, its procurement has increased at about 4.3 percent per year compounded. Nevertheless, Operation Flood still procures only about 6.3 percent of total milk production, or about 22 percent of marketed milk. Some of the expansion of Operation Flood does not represent new milk production but rather a switch from the home manufacture of ghee and other products or from milk previously sold to informal milk buyers (dudhiyas). Total production has increased much faster than can be explained by the additional production generated and procured by Operation Flood (Figure 7.1). 7. The overall expansion of the dairy industry from the early 1970s has been comparable to the more widely recognized Green Revolution crops of wheat and rice. Two key policy changes which accompanied the decision to provide direct financial assistance to the cooperatives to develop the dairy industry were a cessation of direct efforts by the public sector to promote dairy production and the decision to sell dairy food aid at commercial prices within India. Both of these changes reduced the price risks for farmers, small-scale traders (dudhiyas), and private processors. Crossbreeding of local cows with specialized dairy breeds provided the technology for rapid increases in milk production, and Operation Flood provided the example of large-scale modem milk processing fed by a well-organized milkshed procuring milk from a large number of producers in very small amounts. As shown in Figure 7.1, the actual volume of milk handled by Operation Flood remains a small fraction of the increased supply. It is thus not possible to 3 attribute increased production simply to the dairy processing, marketing infrastructure, and technical support provided by Operation Flood directly (important though these contributions have been). Rather, one has to look to the changed dairy policy environment which accompanied the decision supporting Operation Flood. This was not "policy based lending" in the sense of an operation rich in policy conditionality. But, it was an operation where the Bank saw an ongoing policy that it felt was promising and could usefully be supported. The Bank saw the policy as promising both because of its likely production impact, but more importantly because of its capacity to reach the poor. 8. Without the policy change, there would have been no Operation Flood; and without the attractive alternative of an expansion of the cooperative sector, it is highly unlikely that the public sector would have ceased its efforts to promote production while holding down prices. On the issue of the independence and farmer control of cooperatives (a key objective of Operation Flood and the Bank), progress has been mixed (Table 6.11). Under India's constitution, agriculture is a state responsibility. This explains the differences in the treatment of dairy cooperatives among states. 9. A byproduct impact of Operation Flood and accompanying dairy expansion has been the establishment of an indigenous dairy equipment manufacturing industry (only 7 percent of dairy equipment is now imported) and an impressive body of indigenous expertise that includes animal nutrition, animal health, artificial insemination (AI), management information systems (MIS), dairy engineering, food technology, and the like. The Bank's insistence on International Competitive Bidding means that, as far as Operation Flood goes, India's dairy manufacturing industry has been established in competition with international suppliers. However, supplies of Indian equipment to non-Flood dairy processors may have benefited from the artificial stimulus of "self-reliance" which ruled in many industries prior to 1991 and raised the cost of those processors. This indigenous infrastructure explains in large part why milk processing and marketing costs of Operation Flood have not exploded in the face of having to procure and account for minute quantities of milk. 10. Operation Flood recognized that farmers' profits can be increased by lowering their costs (that is, increasing their efficiency) just as well as by raising producer prices and that access to lower-cost producers enhanced Operation Flood's competitive position. Accordingly, Operation Flood supplied Al services (including crossbreeding), vaccination and veterinary services, cattle feed, improved forage seeds, and training in the urea treatment of dry forage, milk testing, recording, and accounting. For Al, milk-testing, and village society secretarial services, Operation Flood trained villagers to a high degree of competence and thus controlled costs. It capitalized on rural India's ample supply of very low-cost labor. In addition, Operation Flood provided access for its villages to poverty-oriented programs such as credit (and subsidies) to purchase cattle, subsidized biogas construction, and improved water supply. 11. Unusually, the project is a single-commodity "integrated development project." There was an obvious production impact (40 million tons more of milk were produced in 1995 than would have been produced if the pre-1971 growth rate had continued) of the policy changes associated with Operation Flood. It is also a nutrition project. Under the new policy, per capita milk consumption has risen from 107 grams/head/day in 1970 to 193 grams/head/day in 1994. Operation Flood has also had clear benefits for the poor and women. For those landless people who own or have been able to purchase a milch animal (a cow or buffalo), it has been a boon. Of course, those who are both cowless and landless have only been slightly affected by some of the 4 social programs supported by the village dairy cooperative societies (DCSs) out of their profits and by the increased prosperity of the village economies. Operation Flood, in collaboration with NGOs, has established 6,000 women's dairy cooperative societies (WDCSs) and higher incomes for some women. Indirectly, it has expanded the number of children attending school because of the high income elasticity of demand for education in Indian villages. This is a project that has produced diverse "results on the ground." The multidimensional benefits from this single- commodity production and marketing project raise a question as to the relative efficacy of production projects that aim to remove or reduce poverty directly by raising incomes compared with health, nutrition, and education projects designed to remove or reduce poverty indirectly, or compared with more complex area development projects. 12. Project sustainability is a complex issue. The first point is that where states continue to interfere with pricing decisions or make key appointments, MPUs and state federations can be expected to make losses; however, state governments can be expected to reimburse (or "subsidize") them for their losses. Thus, ironically, it may be the state-controlled, often loss- making, MPUs which are the most "sustainable." Their future is not in doubt since they are not allowed to go bankrupt. It is the full Anand-pattern, independent cooperatives which have the right to go bankrupt and are thus fully at risk. Although the cooperative sector as a whole makes a profit, Tables 4.4 and 4.5 show that there are substantial arrears (27 percent) in the payments due to NDDB from MPUs and state federations under the Bank projects and that some 46 MPUs and state federations made losses in 1994-95. However, over the last two years, 27 previously loss-making MPUs have been restored to profitability. NDDB's equity is sufficient to repay all loans even if there were no further repayments from the states. Thus, even though some MPUs may eventually go bankrupt (through mismanagement or overinvestment), this does not threaten the sustainability of the cooperative sector as a whole, even in the face of substantial state assistance to the corporate sector which is not currently available to cooperatives. Operation Flood continues to expand and consolidate the cooperative sector. 13. Viewed as lending to sustain and help implement a new policy, Bank investments need to account for only 0.6 percent of the observed acceleration in the growth of dairy production to yield an estimated economic rate of return (ERR) of 10 percent. If 2 percent of the observed increase was due to the policy change, then this would return 10 percent on the entire Operation Flood investment by the WFP, EC, government of India, and the Bank. 14. It has already been remarked that Operation Flood is an Indian program; the basic institutional structure was put in place during the first phase, when the Bank was supporting parallel dairy development in three states. Thus, when the program was massively expanded under Operation Flood II, there was an already existing indigenous institution ready to implement the project. Bank institutional support thus involved genuine institution building, as different from the institution creation which is characteristic of many Bank projects. The great difficulty of simultaneous institution creation and project implementation has been widely demonstrated in Bank projects, particularly in Africa. 15. Total project cost was US$1.6 billion-US$0.9 billion as grant and US$0.7 as loan (1996 dollars). These resources were channeled predominantly through NDDB. Most of the money was passed on to MPUs for construction of milk processing plants with a 30 percent grant element. It also paid for development of the national milk grid, establishment of DCSs, and some of the costs of the development of NDDB's support capacity. Currently, NDDB receives no operational subsidies. 5 16. Initially, the dairy industry was highly protected (Table 4.6) and until 1984-85, skim milk powder (SMP) received more protection than butter-oil (ghee). Subsequently, SMP protection has fallen relative to ghee, to the point that though ghee is still substantially protected, SMP is essentially unprotected. This reflects in large measure the rise in world prices for SMP relative to ghee, a change that is not yet fully reflected in the Indian market. For recombined milk, the protection is about 18 percent. Effective protection coefficients range from 0.88 for SMP on an import basis to 2.34 for ghee on an export basis. Since SMP and ghee are joint products, the effective protection for milk, about 18 percent, provides the best overall measure of current protection. As the industry has grown, the level of protection has fallen significantly as world prices have risen in response to reduced export subsidies in the European Union and United States. Under its New Economic Policy and its adherence to the World Trade Organization, India is removing all quantitative controls on imports on dairy products. In 1993- 94, record procurement by Operation Flood led to record production of milk powder and ghee. This led to depressed prices in India (as low as Rs 36,500 or US$1,160 a ton for SMP, versus US$1,569 on the world market), and most MPUs recorded large losses, the majority of which they subsequently made up. It is probably fair to say that the Indian dairy industry is at least as competitive as the North American and European industries. 17. The perishability of milk (it will spoil if not cooled within four hours of milking) and the impossibility for consumers of knowing whether a given sample of milk is a health hazard dictates a public interest in milk marketing. It is a simple matter to adulterate milk by adding water (and thus perhaps fecal matter and associated health hazards); by correcting for spoilage by the addition of caustic soda; by replacing easily cleaned stainless steel equipment with plastic and mild steel; or indeed by bypassing the cow entirely, relying on a mixture of vegetable oil, soap, and urea to produce "artificial milk." Since these devices can be used to reduce both marketing margins and milk quality, there is a Gresham's Law of Dairying that states "poor milk drives out good." To rely on the beneficial results of unregulated competition in such circumstances is not in the public interest. 18. India does not yet have the capacity, despite ample and appropriate regulations, to ensure that milk is not contaminated. Small traders, in a field interview, listed bribes to the health inspector to avoid having their milk tested as a routine cost of doing business. When detected by corporate milk processors, poor quality milk simply commands a lower price: it is not usually rejected. Thus, for all practical purposes, the Indian milk market is very largely unregulated. Effective regulation is a major public health need. Without effective public health controls, there is an evident tendency for ethical processors (those who observe the law) to be driven out of business by unscrupulous processors. 19. In the villages, there are many milk producers and few buyers. This is inherent in the very small volumes of milk produced by any one farmer. Without an adjacent urban center to ensure that depressed prices are countered by the entry of new buyers or a cooperative society that ensures a benchmark milk price, the market structure involves many weak sellers with only a few discretionary buyers. Farmers are the weak sellers, since their cows produce highly perishable milk twice a day. An individual trader can easily choose not to buy on a particular day. This situation is a recipe for market failure. Traders, in maximizing their own profits, can be expected to act oligopolistically. They will maximize their short-term profits, even as they drive the suppliers down their long-run supply curves and thus jeopardize long-term profits. At a distance from urban centers and in the absence of a cooperative, there is no reason to expect milk 6 traders to act competitively. Policies predicated on competitive behavior are thus unlikely to yield expected benefits. 20. The Anand principles call for a three-tier system of cooperatives owned and controlled by farmers, with professional management. They are to be free of government or political interference and to have the right to set purchase and selling prices, to hire and fire staff, and go bankrupt. A major rationale of the Bank-supported Second National Dairy project (NDPII) was the conversion of cooperatives to the Anand pattern. All states that benefited under this follow- on project signed undertakings to convert their cooperatives to the Anand pattern. Actual compliance has been mixed, with many states still appointing civil servants as managing directors of the federations or even MPUs and government nomination of at least some of the directors to the federations and unions. By and large, farmers control the village DCSs. Factionalism can be a problem but is not often serious. Even so, arbitrary and unhelpful rulings by the local representative of the Registrar of Cooperatives occur. At the MPU and federation level, the effects of government appointments are reflected in poor pricing decisions that sometimes lead to losses. Thus, there are MPUs (notably, but not exclusively, in Rajasthan and Madhya Pradesh) that are experiencing financial difficulty. This is recognized as a problem by NDDB, and strong remedial steps are being taken, including the closing of three MPUs that have no prospect of becoming profitable again. These problems are not inherent in the cooperative structure but rather reflect the failure of the projects to achieve the full Anand structure for cooperatives in the states concerned. These problems are exacerbated by the requirement that cooperatives register under an archaic and paternalistic Cooperatives Act. 21. The Bank's recent Livestock Sector Review for India concludes that Bank policy should be to level the playing field between cooperatives and private companies in India's dairy sector. The Bank report takes this to include eliminating all state intervention in cooperatives, repealing the Milk and Milk Products Order (MMPO), establishing a mechanism to monitor the milk market to ensure fair competition, and strengthening public monitoring and enforcement of hygiene standards. This omits reference to the need to remove tax concessions and other incentives provided to private investors. 22. Both "ensuring fair competition" and "enforcing hygiene standards" are easier said than done. Effective mechanisms for achieving these desirable objectives would very largely remove the need for the MMPO. Private companies and full Anand-pattern cooperatives can be expected to behave quite differently in a monopsony situation with cooperatives treating farmers (their owners) more generously than private companies would. Thus, some (slight) tilt toward cooperatives may be the best operational policy recommendation for encouraging (if not ensuring) fair competition. For those who see no difference in the treatment of producers by cooperatives and corporations, the recommendation of any tilt towards cooperatives will appear unjustified. 23. The problem of ineffective enforcement of hygiene standards is serious. The legislative and administrative basis for enforcing hygiene standards is already in place. The problem is the lack of enforcement of the Prevention of Food Adulteration Act of 1954. Pending proper enforcement of the act, retention of the MMPO provides another avenue for enforcing hygiene standards. Unfortunately, it is not clear that the MMPO is being implemented any more conscientiously than the Prevention of Food Adulteration Act. 7 24. Similarly, pending the removal of incentives for companies to invest in dairy plants, the MMPO provides a slight restraint on the waste of public resources from the overexpansion of private sector processing capacity, which absorbs large amounts of public direct investment and tax concessions. In the Punjab alone, under these incentive programs, corporations have installed or applied to install an extra 3.8 million liters/day of capacity to cope with an expected production increase of 0.56 million liters/day. Up to half of this excess investment will be provided from state funds. Recommendations 25. The policy recommendations that follow from the analysis in this impact study differ slightly in content, and certainly in sequencing, from current Bank proposals: * Elimination of all state intervention in Operation Flood cooperatives. Where this involves the transfer of state-owned assets to the cooperatives, the cooperatives could issue bonds to the state to be retired over a suitable period of time. * Elimination of all government incentives for investment in the dairy industry. In the absence of such a change, and as a distinct second best solution, any incentives should be equally applicable to cooperative or corporate investments. * A renewed effort to enforce the provisions of the Prevention of Food Adulteration Act of 1954. This should include inspection and approval of quality control/milk testing procedures of processors, inspection of processing facilities, and testing dudhiya milk supplies for adulteration. This would require increased state expenditures. However, the sums involved are dwarfed by the savings which would follow from the withdrawal of government incentives. The livestock sector review suggests that states could experiment with paying private sector organizations to carry out the inspection and enforcement function. 26. Pending successful implementation of the above three conditions, repeal of the MMIPO would be premature. This is an issue of sequencing of reforms. The MMPO provides a weak bulwark against overinvestment, substandard equipment, and unhygienic operating conditions. 27. Continued support for the Indian dairy industry can be justified by the intention to bring about the policies recommended in para. 25 and perhaps afterward, depending on whether a tilt towards cooperatives is felt to be appropriate. 28. At the very least, there should be no further lending to the dairy sector in states which have not yet adopted the full Anand pattern for Operation Flood cooperatives or do not treat these cooperatives equally with private corporations. Lessons 29. As has been hinted at earlier, the very success of the expansion of the Indian dairy industry poses in a stark fashion the question of attribution. To what extent was this a reaction to increased demand, itself a function of rising incomes and population increase? To what extent can it be attributed to technology, such as increased straw from the Green Revolution and the adoption of crossbred cattle? What part did the removal of the risk of price declines based on 8 massive imports of milk products affect the willingness of farmers and private processors to invest? What part was played by Operation Flood's demonstration to other investors of the potential viability of well-organized and modem processing of milk collected from a large number of small producers? What role did individual and prescient personalities play? (Neither the visit of the prime minister of India nor the visit of the president of the Bank was an entirely unorchestrated event.) 30. Despite these attribution problems, a first lesson stands out: Lesson 1: A well conceived investment project in support of an already adopted and appropriate policy change can be associated with results out of all proportion to the investment involved. This occurs in part because it solves the "ownership" problem, in part because it avoids having to create an institution at the same time that the new institution is being expected to implement the project, and in part perhaps for a number of reasons not yet well understood. This lesson intentionally finesses the exact extent of causation involved. 31. A second lesson concerns the multidimensional impact of the project: Lesson 2: By raising incomes, an apparently simple single-commodity project can have multiple beneficial effects, including nutrition, education (especially of girls), and job-creation. This poses a question as to the relative efficiency of production projects that aim to reduce poverty directly, as compared to health, nutrition, and education projects designed to remove poverty indirectly or area development projects designed to intervene in many ways simultaneously. 32. Though not explicitly a "targeted" project, nevertheless 60 percent of the beneficiaries were marginal or small farmers and landless. Lesson 3: By focusing a project on a predominant activity of the poor, "self- selection" is likely to result in a major portion of the beneficiaries being poor. This provides an alternative to "targeted" projects for reaching the poor. 9 1. Introduction Design Features of Bank Dairy Projects 1.1 Bank dairy projects have adopted two distinct strategies (Table 1.1). The first focuses on remedying a production shortage by providing new and genetically superior purebred dairy stock, often imported. The second (the Operation Flood model) involves provision of a reliable market for existing producers. There is overlap in these designs in that both provide for milk processing. The first model includes milk collection, but primarily from its "own" farmers, and the second model includes access to inputs, veterinary services, and crossbred cattle. Despite this overlap, the models are clearly distinct. Table 1.1: Bank-supported Loans and Credits with "Dairy" in the Title Amount/Loan Credit Title Country Approved (million US$) Cr. 0234 Integrated Dairy-Beef Development Korea 1/71 7.0 Cr. 0236 Intensive Dairy Production Turkey 2/71 4.5 Cr. 0269 Addis Ababa Dairy Development Ethiopia 7/71 4.4 Cr. 0482 Karnataka Dairy Developmenta India 6/74 30.0 Cr. 0504 Dairy Development Sri Lanka 7/74 9.0 Cr. 0521 Rajasthan Dairy Developmenta India 12/74 27.7 Cr. 0522 Madhya Pradesh Dairy Developmenta India 12/74 164 Cr. 0580 Dairy Development Tanzania 7/75 10.0 Ln. 1193 Second it. Dairy Development Korea 10/75 15.0 Cr. 0824 National Dairya India 6/78 150.0 Cr. 1196 Smallholder Dairy Development Zambia 12/81 7.5 Ln. 2576 Second Dairy Development Sri Lanka 6/85 38.0 Cr. 1859 Second National Dairy' India 12/87 121.2b Ln. 2893 Second National Dairy' India 12/87 200.0 a. Designed as a marketing project. b. SDR, million. 1.2 Only in the case of Turkey was the production strategy unambiguously successful (with a reestimated economic rate of return (ERR) of 19 percent). The Zambian project is of particular interest, since it was designed as a "production" project but implemented as a "marketing" project. The project was appraised on the assumption that crossbred cattle could be freely imported from Zimbabwe. However, the 1980-81 drought led to a ban on cattle exports from Zimbabwe, and transport costs made it uneconomic to purchase cattle from further afield. This led to a redesign and sharp contraction of the milk production component. Nevertheless, milk collections exceeded the original projections, about half of them from existing small producers. At audit, the five projects that are the subject of this impact evaluation were considered satisfactory. Howc ver, before it is concluded that "marketing is the way to go," it should be noted that an attempt to redesign the first Sri Lankan dairy project from a production project (which was proving unsatisfactory) to a cooperative marketing (Anand-type) project was not successful, and the dairy component of the Pakistan Livestock project (Ln. 1366), which was modeled on Operation Flood, was also judged unsatisfactory. 10 Project Context 1.3 The first of the five Indian dairy projects included in this study was approved by the Board in March 1974. At that time, India was following a self-reliant macroeconomic development policy with heavy emphasis on import substitution (a policy which remained essentially unchanged until 1991). In the dairy sector, a policy decision had been made to utilize farmer-controlled cooperatives to develop the dairy industry and to make the import substitution as efficient as possible. The government also decided to withdraw from its own direct efforts to develop the dairy industry (indirect efforts via extension, research, artificial breeding, etc., continued). Dairy commodity food aid was to be sold at commercial prices with the proceeds earmarked for support of this Operation Flood. 1.4 The inauguration of Operation Flood corresponded to a turnaround in per capita incomes, which had been falling, to a period of sustained, if modest, growth. Rising per capita incomes, together with rising population and a high income elasticity of demand for milk, resulted in a rapid growth in the demand for milk (technically, a rapid shift in the demand function for milk). In the absence of supply-side adjustment, this would have led to a rapid escalation in the price of milk or the need for extensive imports. In fact, the shift in the supply function from the policy changes introduced with Operation Flood resulted in an even larger shift in the supply (function) and declining real prices towards international prices, thus reducing the implicit taxation of consumers. 1.5 Operation Flood is an Indian program to foster a farmer-controlled cooperative dairy industry capable of paying an attractive price for very small quantities (1 or 2 liters) of milk without operational subsidies. Operation Flood was developed in the context of less successful policies which have been summarized: In 1959, the government of India set up the Delhi Milk Scheme, starting a new pattern for dairy development. This pattern was based on the government milk plants collecting milk at its chilling centers through traditional middlemen, pasteurizing it and marketing it in glass bottles through specially set up milk booths all over the city. This pattern was followed until the 1960s in almost 100 cities of the country where similar "Milk Schemes" were set up under state development programs. During this period, the government also started Intensive Cattle Development Projects (ICDPs) which focused on providing artificial insemination and veterinary services to the milk producers. Such programs were further supported with credit to the farmers through Small Farmer Development Agencies (SFDAs). The government milk schemes ran into serious problems. When a milk scheme was started, it paid a reasonable price to the farmer, charged its overheads to the consumers and the consumer was provided with good quality milk at reasonable prices. However, during the summer months, when supplies dwindle and prices go up, the milk schemes were unable to raise the producer prices and their supplies dwindled. The milk schemes then resorted to the import of milk powder to augment their meager supplies. Since the imported powder was rather cheap, the milk schemes were able to maintain low prices in the cities. This removed the incentive of the urban markets for rurally- produced milk, and India's milk production remained stagnant, and its dependency on imports gradually increased. 1. More accurately "stagnation." 11 Providing credit for the purchase of milch2 animals also did not work as it did not contribute to bringing in any additional milch animals to the production system and was merely a transfer of resources-many times only on paper. This, however, brought a bad name to the dairy cooperative movement in many parts of the country as the loans were provided through cooperatives who were supposed to recover the loans through milk money. Many of those who took loans did not bring their milk to the cooperatives and instead sold it directly to the traditional traders. Large amount of loans had, therefore, to be written off. The mid-60s was a trying period in the history of the Indian dairying as the dairy development schemes of the government did not seem to work, milk supply in the cities had to be rationed, imports of as much as 60,000 tons per years had to be resorted to, to feed the milk supply schemes which were handling less than 1 million liters of milk per day. Above all, the dumping of cheap imported milk powder was working against the interest of the indigenous milk producers (Aneja 1994, pp. 13 and 14). In October 1964, on the occasion of the inauguration of AMUL's cattle feed plant, the then Prime Minister of India, Lal Bahadur Shastri, spent the night as the guest of a village milk cooperative society near Anand. Impressed by the socioeconomic changes brought about by the milk cooperatives, he expressed the desire for a national-level organization to replicate Anand Model dairy cooperatives throughout the country ... .Thus, in 1965, NDDB (the National Dairy Development Board) was registered (Banerjee 1994, p. 11). 1.6 Ten years of staff work from the Delhi office culminated in a visit to Anand in 1978 by the Bank's then-President, Robert McNamara (Box 1.1). This visit resulted in an offer to support an extension of the program with funding from the International Development Association (IDA). At the same time, the European Community (EC) decided to use a significant portion of its dairy surplus to support Operation Flood directly, rather than indirectly through the WFP. The result was Operation Flood II (OFII) funded by the NDDB, the government of India, the World Bank (through National Dairy I, Cr. 824), the EC (through food aid), and the farmer-owners of the village dairy cooperative societies (DCSs). The earlier Bank-supported state projects continued to be supported, but as an integral part of OFII. 1.7 The National Dairy project (NDP, or OFII) significantly exceeded the target number of milk producers unions (MPUs) to be formed (125 new MPUs versus a target of 50) and DCSs (29,400 new DCSs versus a target of 20,000). Despite the more rapid growth rate, the DCSs formed conformed to the Anand pattern of farmer control. The same could not be said for the federations and MPUs. Though the details varied from state to state, farmer control was diluted or usurped by appointment of civil servants and politicians and NDDB or federation cadres to serve on the boards of the federation and MPUs. Most MPUs had elected farmer representation, and in many, farmer representatives were in the majority, but the Anand principles of farmer control, including the right to set prices, hire and fire, and go bankrupt, had been undermined.4 2. That is, a cow or buffalo. 3. Among the key principles of the Anand model were village dairy cooperative societies (DCSs) organized into milk producers unions (MPUs) at the district/regional level, which in turn were members of a statewide dairy cooperative federation. This three-tier structure, which was farmer-owned and employed professional managers, formed a single- purpose dairy cooperative that provided a guaranteed market for milk (paid for in cash at approximately weekly intervals) and provided help with modern inputs such as artificial insemination (AI), veterinary services, and concentrates. Farmer ownership meant that these cooperatives, like other private sector organizations, could go bankrupt, since there were no state guarantees. 4. This failure to get adoption of the full Anand principles underlies much of the academic literature critical of Operation Flood. 12 Box 1.1: Early Bank Interest in Operation Flood "Amul's milk collection system is efficient and businesslike. Villagers line up twice a day at village collection centers where they receive a uniform price. The contribution of each farmer is tested for fat content on the spot. Quality tests are also conducted upon arrival at the factory. Cash payment for the morning milk is paid in the evening and payment for the evening milk is paid the following morning. Several private trucking firms (selected by competitive tender) carry the milk from the village to Anand. Collection and delivery follow a rigorous schedule. Bulk transport of pasteurized milk from Bombay, a 270 milk haul, is done by refrigerated rail car" (memorandum to files, October 17, 1968). "When we last talked of the possible staff contribution of the Agricultural Division to the Fourth Plan Review.. .I suggested that livestock development with special emphasis on 'dairy' might be one of the subsectors deserving additional study by the Bank ... .Besides reviewing research results and Government and cooperative dairy schemes and plans, a dairy man would need to look into the activities of private firms such as Hindustan Lever, Nestle, Polsons and Glaxo, also active in the dairy field" (letter, October 18, 1968). "Anand is, statistically, all that was reported and Kurien said at Reading. What you have to experience to believe is the enthusiasm of a highly skilled team of technologists and marketeers working in an atmosphere that is modem and delightful. K takes a great interest in the architecture of his factories and the 'surrounds' are like a Garden City. ". . the future lies with the hybrid cow-in the beginning a first cross of European milk breed (mainly Jersey) on the native cow which in one generation will double production and about halve costs of production. It's the most exciting cattle development prospect in the world at the present time. It has possible repercussions beyond India along the lines, for Southeast Asia, I discussed with you.. .last year. "... In my view the prospect is the animal equivalent of the rice and wheat story with the possibility of even greater initial gains and impact upon costs of production which, with the buffalo, can only remain high" (letter, March 3, 1969). "One key organizational issue concerns the relationship between state livestock departments (largely veterinary-oriented) and the National Dairy Development Board. .. .The technical side (of any project proposal) will also call for detailed review so as to bring a rational nutritional and breeding policy to bear on any project put to the Bank for support. Dairy Engineering, on the other hand, is a field in which India has built up considerable expertise. The bottlenecks mostly on the farm" (memorandum, January 13, 1970). "I have now been to Anand .... .I do not think we have a project here in the near future but I would consider it a possibility for FYI 974. In the mean time I think it will be worthwhile to take more than a passing interest in the way Operation Flood develops. . ." (letter, June 14, 1971). 1.8 This set the stage for the second National Dairy project (NDPII) or Operation Flood III (OFIII), where the first nominated project objective was: (a) Adjustment in cooperative institutional structures towards characteristics of the OF model (elected boards for DCSs, MPUs or Federations, model bylaws insuring democratic and accountable operations, professional management answerable to the cooperative boards, management and ownership by the cooperatives of dairy infrastructure, autonomy for each cooperative in pricing, marketing, appointment of key personnel, and employment of labor). (SAR, National Dairy II, p. 12) 13 1.9 This project also involved major investments in dairy processes and transport equipment, help in the formation of additional MPUs (in the event unused) and DCSs, staff training, and improved dairy technology, but the key project objective was to restore the full Anand pattern of farmer control, which had been eroded under the "too-fast" growth that had occurred under the NDP. 1.10 As an aside, it may be noted that the NDP could have followed either of two growth strategies. A slow growth strategy would have extended the project only to those states willing to adopt the full Anand model as a precondition for Bank-supported investments. The alternative fast growth strategy involved setting up federations and MPUs even in states that were not willing to relinquish all control to boards elected by farmers. The expectation in this latter case was that Operation Flood could "work from within" and persuade state politicians of the value of full farmer control and the Anand principles. In the event, the fast growth strategy was adopted; and even with hindsight, it is not possible to fault this decision. It was the hope of correcting the deviations from Anand principles that had occurred during the implementation of the NDP that provided the rationale for the NDPII. 1.11 The NDDB, does not fit easily into any of the pigeonholes that constitute the Bank's normal institutional paradigm. Described by Liana Gertsch (1990) as "corporatism," the NDDB can perhaps best be described in the Bank lexicon as "a parastatal operating in the producer's interests," almost paradoxical in Bank experience. Formally a parastatal (with all its board members appointed by the government), it has "sworn an affidavit before the Bombay High Court that it is not a government agency .... Perhaps a better indication that it leaned more towards the private sector is that in order to gain access to funds generated by the sale of commodities, it had to create a wholly public institution (the Indian Dairy Corporation (IDC)) with legal right to engage in commercial activity on behalf of government" (Gertsch 1990). In practice, the NDDB established its financial independence (and hence managerial autonomy) through the sale of concessionally-priced dairy imports. Not beholden to the government for funding, the NDDB acts as the technocratic (and policy) apex organization for the dairy cooperative movement. Literature Review 1.12 There has been a flood of literature on Operation Flood and the dairy sector in India (Alderman, Mergos, and Slade 1987). An important and critical strand of this literature has been generated by the Indo-Dutch Program on Alternatives in Development, sponsored by Dutch Aid. The work of these Indo-Dutch authors is summarized in two books that, while critical of Operation Flood, offer few alternative policy prescriptions (Doornbos and Nair 1990; Doornbos and others 1990). Recently, the Bank has also criticized the cooperative sector (World Bank Reports 1996). Ranged on the other side are indigenous authors such as Somjee and Somjee (1989), Mascarenhas (1993, 1988), and T. Shah (1991). 1.13 Topics discussed at length in the literature are: * Production Impact. It is generally agreed that the rate of growth in Indian milk production has increased significantly during the lifetime of Operation Flood (Fulton 5. It has also taken on, or been assigned, a number of other cooperative tasks not directly related to dairying. 14 and Bhargava 1994; Kumar and Singh 1993). But even this has been challenged (Mishra and Sharma 1990). However, the critics emphasize, quite correctly, that this increase in milk production should not be attributed exclusively to Operation Flood (Doornbos and others 1990; Nair 1985; Aneja 1994). * Socioeconomic Impact. The apologists emphasize that some landless, tribal, and marginal farmers have benefited (nearly 60 percent of milk producers are landless or marginal farmers) (Fulton and Bhargava 1994). The critics argue that much of the increased milk income went to large farmers and that many landless and otherwise disadvantaged were unable to benefit from Operation Flood.6 * Women. Apologists point to the efforts made to include women in DCSs, even to the point of establishing women-only DCSs. Critics emphasize the number of mixed (that is, male-dominated) DCSs. Somewhat confused messages emerge from the debate, since there is no consensus as to whether more or less work for women is to be welcomed. Certainly some critics cite the withdrawal of women from the domestic manufacture of ghee in order to sell whole milk as a loss for the "subordinate gender" (George 1991). * Implementation. Critics have pointed to a contradiction between the development of the initial Anand village cooperatives, which were formed by farmers on their own initiative, and Operation Flood, which is seen as a "top-down" program dedicated to persuading villagers to form DCSs. These critics also doubt the realism of attempting to replicate the Anand pattern nationwide, given the diversity of states in India (Joshi 1990; Shekhawat 1990). Apologists acknowledge that progress toward the Anand pattern differs markedly among states and cite various modifications to the structure of DCSs and the services provided to show that Operation Flood has been implemented adaptively. Also, though "spearhead" teams attempt to interest villages in the formation of DCSs, the final decision rests with the villages, not the spearhead team. * Sustainability. Critics concerned about the level of assistance provided to Operation Flood in the period of its most rapid expansion doubted whether it would be able to survive without such assistance (Doornbos and others 1990). Other writers have pointed to the rapid decline and phasing out of this assistance as evidence that this problem has not eventuated (Chaterjee and Acharaya 1992). 1.14 Total project cost was US$1.6 billion-US$0.9 billion as grant and US$0.7 as loan (1996 dollars). These resources were channeled predominantly through NDDB. Most of the money was passed on to MPUs for construction of milk processing plants with a 30 percent grant element. It also paid for development of the national milk grid, establishment of DCSs, and some of the costs of the development of NDDB's support capacity. Currently, NDDB receives no 6. A number of schemes have tried to give the poor access to Operation Flood by providing milch animals (often on credit). In fact, the more binding constraint for the landless is often a lack of free or cheap fodder, a problem that cannot be remedied by a single capital injection. 7. More precisely, both the village and the MPU have a veto. Both must agree on the desirability of a DCS, since the MPU will collect milk only from a DCS with sufficient volume (or the promise of sufficient volume) to cover the variable costs of milk collection. 15 operational subsidies. Even though Operation Flood handles only 6.3 percent of Indian milk production, it is a huge undertaking. Operation Flood operates in 22 states, but the bulk of its procurement (46 percent) is obtained from two states-Gujarat and Maharashtra (Table 4.2). In 1996, it had 9.3 million members and NDDB estimates 6.3 million "pouring" members,8 who supply 10,900 metric tons of milk per day through 55,042 village-level DCSs to 170 MPUs for processing and sale. Some of the expansion of Operation Flood does not represent new milk production but rather a switch from home manufacture of ghee and other products or from milk previously sold to informal milk buyers (dudhiyas). Fortunately, there is a rich Indian literature on Operation Flood. This has been examined, together with Bank internal reports and files. The evaluator visited a total of 40 villages in five states to discuss with farmer members and managing committees their experience of Operation Flood. This included women-only DCSs, DCSs which were thriving, DCSs which had closed and reopened, and even a DCS which had closed and stayed closed. Discussions were held with academics, formal sector milk processors, and the small-scale, informal dudhiya milk collectors who provide keen competition for the cooperatives in peri-urban areas. A two-day workshop was held at the Institute of Rural Management at Anand, Gujarat state, in March 1997 to discuss the draft report with academics and managers in the dairy industry.9 The project's impact on the intended beneficiaries was studied more formally through participatory evaluations in nine villages in Karnataka and a repeat survey in Madhya Pradesh to capitalize on "semi-baseline" data available from 1983. 1.15 The evaluation intended to focus on five questions: * What have other commentators, both national and foreign, said about the program? In responding to this question, the evaluation has been fortunate in having only to extend a literature review published in 1987 (Alderman, Mergos, and Slade 1987; Kumar 1997) (paras. 1.12-1.13). * How has the expansion of Operation Flood compared with the growth of the rest of the Indian dairy industry? Did the rate of growth of the dairy industry change perceptibly with the introduction of Operation Flood? What role, if any, can be attributed to the Bank's assistance in the growth of Operation Flood? (paras. 6.3- 6.11 and Figure 7.1). * How competitive is the Indian dairy industry? Is this a case where an "infant industry" that initially had to be supported with grant and concessional funds has finally reached the point where support is no longer required? If not, how significant is the support required? (paras. 4.32-4.34). * What was the impact on the poor? (paras. 1.16, 5.9, and 6.34-6.35). 8. Originally, there was only one member per family, and only families supplying (or "pouring") milk could be members. Thus, members and "pouring members" were synonymous. Over time, in part to give greater access to women, some DCSs have come to allow two members per family, some members who have ceased to pour have been allowed to retain their membership, and members in DCSs which have closed have not been removed from the books. NDDB estimates that about 68 percent of members (that is, 6.3 million people) supply milk to a DCS at some time during the year. 9. This workshop was dominated by representatives of the cooperative sector. However, the list of invitees included donor organizations, the private sector, and government policymakers. Representation was diverse enough to encompass a wide range of views on the industry and the draft report. The assistance of the Swiss Development Corporation in providing funds for this workshop is gratefully acknowledged. 16 Can the impact be improved? (paras. 7.27-7.34). In the event, as described in Chapter 6, a wide range of other unforeseen benefits were associated with Operation Flood. These included impact on each of the Bank's four overarching objectives of poverty alleviation, women in development, governance, and the environment. Findings with respect to these four objectives are summarized in the next sections. 1.16 Poverty. Operation Flood was not designed as a production or poverty project. However, by virtue of providing a market to very small milk producers, the majority of its beneficiaries were marginal or small farmers and landless. This did not exclude the "better off" (many of whom are poor in any absolute sense), but it did include a majority of poor beneficiaries (albeit poor who happen to have, or are able to obtain, at least one milch animal). As the 1987 Performance Audit Report (PAR) said: Not only are the level and nature of the benefits flowing to members of the established DCSs impressive overall, but the extent to which such benefits are reaching the extremely poor and needy (destitute, widows, landless and near-landless) in certain 'spearhead' villages is unusually noteworthy; and also indicative of the AMUL (Anand Milk Union Ltd.) model's potential to benefit a target group which is widely espoused by all donors, including especially the Bank, but which has proved elusive to reach in practice (para. 2.4). Operation Flood turned out to be a major income- (paras. 6.34-6.35) and employment- (6.47) generating project. 1.17 Women. "Women's issues" are touched on in the above quotation by the mention of "widows." However, neither the Staff Appraisal Reports (SARs) nor the PAR make explicit mention of women. Anecdotal evidence suggests that not only widows but also wives have benefited disproportionately from Operation Flood. Not only do women often look after the dairy animal(s), but also the sums of money received are typically used for running the household (cattle feed, food, clothes, and educational requirements) (paras. 5.7, 6.34, 6.38, and 6.42). Operation Flood has cooperated with NGOs, notably the Self-employed Women's Association (SEWA) and Bhagavatula Charitable Trust (BCT), to establish, in some cases with Ford Foundation support, 6,000 women-only dairy cooperative societies. These are generally thought to operate more smoothly than the male-dominated DCSs.10 10. A much less sanguine view has been expressed in the Bank's study of gender in India (World Bank 1991): 4.15 All the constraints to women's access to public programs and commercial transactions with the 'outside' which have been outlined earlier are operative in the dairy sector. The common pattern is for women to handle most of the production aspects and for the men to assume the cooperative membership and control the cash income. There are a number of problems associated with this traditional 'inside/outside' division of labor. Some affect the overall efficiency of dairy production, others affect the welfare returns to the family and to the woman producer herself. Women, for example, do not usually gain access to training in modern livestock management and dairying techniques which is available to men through the co-op structure. Instead, they must learn second- hand through the men or continue with traditional practices, both of which lower their efficiency and reduce returns to investments in training. Non-member producers also miss the chance to be trained in the responsibilities and rights associated with cooperative membership and to benefit from the exposure provided by Farmers Induction Training and other member activities. 4.16 For women from poor households the greatest disadvantage is, however, that they have no control over dairy income which is collected by the male household member. In cases where the women used to deal with traditional milk traders who came to the household compound, they lose what small degree of economic autonomy they had when marketing arrangements are formalized through the cooperative structure. The fact that 17 1.18 Privatization and Governance. The projects were not consciously designed to be innovative with respect to privatization and governance. Nevertheless, their emphasis on village- level DCSs and farmer control of MPUs and state cooperative dairy federations meant that they broke new ground in supporting membership-based nongovernmental organizations (NGOs) (in this case cooperatives) (World Bank Reports 1996), development, and project administration. Because DCSs are, in principle at least, open to all, they represent a major addition to village social capital." The difficulty in getting true independence for the cooperative movement is one of the themes/issues explored in this evaluation (paras. 4.17-4.21). 1.19 Both cooperatives and companies can exist in either the public or private sector. Public sector companies are generally referred to as parastatals. Private sector cooperatives are referred to in India as farmer-controlled or (full) Anand-pattern cooperatives. In most states the dairy cooperatives are to some degree mixed, with the state imposing decisions on pricing or staffing, often with some state ownership of productive assets. It is thus a false dichotomy to talk of public sector cooperatives versus private sector companies. 1.20 As between private sector cooperatives and companies, there is no difference in the managerial objectives; in both cases, management aims to maximize the return to the owners. The difference is in the owners. In a private company, the owners are the suppliers of capital, while in a producer cooperative the owners are the input suppliers (milk suppliers in the case of Operation Flood). This has important behavioral implications with respect to the exercise of monopsony power. A private single buyer will seek to minimize procurement price, since this will maximize the return to the owners. Cooperatives have no such incentive to lower prices excessively, since any profit from a lower price will only need to be returned to the producers, who are the owners. 1.21 In the villages, there are many milk producers and few buyers. (This is inherent in the very small volumes of milk produced by any one farmer.) Without an adjacent urban center to ensure that depressed prices are countered by the entry of new buyers or a cooperative society that ensures a benchmark milk price, the market structure involves many weak sellers with only milk payments to cooperative members in many villages are, for efficiency reasons, now made less frequently and, hence, in larger sums, has apparently increased the likelihood that at least some of the money is diverted by the men before essential household expenses are met. In short, for non-member women producers, Operation Flood has too often meant more dairy work but no increase, and sometimes even a decrease, in their access to dairy income. See also paras. 4.18-4.22 of the gender study. Since India is a big and diverse country, both views may be right. The evaluation met many women in the field who had been greatly advantaged by Operation Flood. 11. An internal memorandum of November 14, 1996, states: The Bank's Social Development Task Force has described social capital as follows: Social capital is not just the sum of institutions which underpin society, it is also the glue that holds them together. It is the shared values and rules for social conduct expressed in personal relationships, trusts, and a common sense of 'civic' responsibility that makes society more than a collection of individuals. Without a degree of common identification with forms of governance, cultural norms, and social rules, it is impossible to imagine a functioning society. The promotion of DCSs, which are male-dominated, is viewed by some as a missed opportunity. They argue that given the dominant role of women in dairy production, the DCSs (and extension services) should have been designed to give women a dominant role. Where they have emerged, WDCSs have vitally changed the role of women within the village and thus added significantly to social capital. 18 a few discretionary buyers. Farmers are the weak sellers, since their cows produce highly perishable milk twice a day.12 An individual trader can easily choose not to buy on a particular day. This situation is a recipe for market failure. Traders, in maximizing their own profits, can be expected to act oligopolistically. They will maximize their short-term profits, even as they drive the suppliers down their long-run supply functions and jeopardize long-term profits. Absent a cooperative and at a distance from urban centers, there is no reason to expect milk traders to act competitively. Policies predicated on competitive behavior are thus unlikely to yield expected benefits. 1.22 The Environment. Dairy plant effluent, if not treated, has a high biological oxygen demand and can be a serious pollutant of surface water and aquifers. Effluent treatment plants were an integral part of Bank-financed processing plants. However, the high cost of operating these plants tempted some MPUs to bypass or underutilize their plants. This was identified as a problem in project supervision in the early- and mid-project stages. Current operation is believed to be satisfactory, but because of high cost, proper effluent treatment needs to be monitored from an independent authority. Report Structure 1.23 The next chapter reviews project design, and the third chapter discusses the impact of the first four projects. Chapter 4 describes developments under the most recently completed project. Chapter 5 reports the major findings from participatory village studies commissioned in Karnataka and Madhya Pradesh. The sixth and longest chapter focuses on current impact. A seventh and final chapter contains findings and recommendations. 12. Milk can be processed into ghee (butter oil), chenna (homemade cottage cheese), or mithai (traditional sweets). All of these byproducts are highly fuel- and labor- (women's labor) intensive. These technical possibilities do not substantially strengthen the farmers' bargaining position. 19 2. Project Design Karnataka 2.1 The first of the three state projects was in Karnataka, where dairying was already fairly advanced. There were 400 existing dairy cooperatives at the village level, the Bangalore Dairy Milk Producers Cooperative Society Union (serving Bangalore), and many multipurpose cooperative societies. A key thrust of the project was to incorporate the existing dairy cooperatives into an Anand-type three-tier cooperative structure. The project also planned to open about 500 dairy "wings" within the multipurpose cooperatives, which would be converted to single-purpose Anand-type cooperatives when volume and experience justified it. New DCSs were to be established in villages without cooperatives to bring the total number of project- sponsored DCSs to 1,800. The DCSs would be organized into four MPUs, which would own the Karnataka Dairy Development Corporation (KDDC). The DCSs were projected to serve a total of 450,000 farm families. Production would be stimulated by an artificial insemination (AI) program of crossbreeding native village cattle with high-producing exotic breeds, help with fodder production, veterinary services supported by two regional diagnostic laboratories, consultant services to assist the KDDC with demonstration farms, applied research trials on pastures, and milk marketing studies. Hardware was to include the renovation of two processing plants, the construction of two new processing plants and four cattle feed mills, and the establishment of milk collection routes and centers. Provision was also made for the importation of 500 in-calf heifers to stock three KDDC farms to produce purebred exotic breeding stock. 2.2 The KDDC was set up as a parastatal. (The central government was to contribute Rs 2.5 million and the state government Rs 0.8 million of equity plus Rs 3.2 million of working capital to be shared, in unspecified proportions, between the governments.) As part of this transaction, the state government was to transfer its dairy plants and other assets to KDDC. The governments undertook to gradually sell their equity to the MPUs at cost, starting in the sixth project year, with the expectation that by the tenth year the MPUs would be majority shareholders in KDDC. In the event, the governments did not sell their equity to the MPUs. Under NDP, the name of KDDC was changed to the Karnataka Milk Federation, but its parastatal structure remained unchanged. In Karnataka, significant progress has been made towards professionalizing the management of the dairy industry, but ownership of state dairy plants has still not been legally transferred to the federation. Thus, the state still has residual powers flowing from its initial investment in KDDC and its failure to transfer title as expected. With hindsight and recognition of the debilitating effects of government interference in management, it would probably have been better to insist on a fully cooperative structure from the start. Similar design difficulties have affected the other two state projects. 2.3 The project was to be guided by the NDDB, which would provide "spearhead" teams to assist with the establishment of both MPUs and the village-level DCSs that would supply them. DCS membership was to be limited to cattle owners, 10 percent of whom were expected to be landless.13 A further 50 percent were to be marginal farmers with less than 2 ha. of land. 13. In later years the prestige of DCS membership, and the desire to allow both husband and wife to belong, has led to a breakdown of this criterion in some DCSs. A survey in Karnataka (Sampark 1997) found, in addition to double 20 2.4 A very rapid expansion of milk production was projected, mainly based on the replacement of local dairy and buffalo breeds with crossbred dairy cows. The semen used in the Al program was to come mainly from Jerseys and Fresians. As a result of this breed change, which would be supported by readily available veterinary services and improved concentrate supply, milk production in a typical DCS was expected to rise from 60,000 liters/year (1/year) to 290,000 1/year in the fifth year of the project. 2.5 The bulk of International Development Association (IDA) funds (US$26.4 million) were to be channeled through the Agricultural Refinance Corporation (ARC), a parastatal bank, to the KDDC and MPUs. Loans for dairy plant construction were to be supported by a 20 percent equity participation from the central and state governments in order to retain a 4:1 debt-to-equity ratio in the MPU. This financing arrangement applied also to the two other state projects. Parallel arrangements were made for local banks to provide credit for the purchase of cattle and feed. The ARC would stand ready to refinance these loans, giving the banks a 2 percent margin under their lending rate. 2.6 The major foreign exchange costs of the project were the expansion of the Bangalore and Mysore milk processing plants, the construction of new processing plants at Hassan and Tumkur and four new feed mills, and investments in KDDC farms and technical services. Rajasthan 2.7 The Rajasthan credit was for US$27.7 million. Rajasthan had a government milk scheme that subsidized urban milk prices but had failed to stimulate the emergence of a cooperative sector. This failure was recognized in 1972, and the Rajasthan Animal Husbandry Department commenced the organization of dairy cooperative societies based on the Anand pattern. At appraisal, there were some 100 such societies organized into two unions, which were to form part of the project. 2.8 The project called for the creation of a new apex organization, the Rajasthan Dairy Development Corporation (RDDC), to foster the development of five MPUs, which would process and market milk from a total of 1,800 DCSs. In turn, the DCSs would collect milk from a total of 240,000 families, most having less than 2 ha. of land or being entirely landless. 2.9 Provision was also made for the importation of 100 exotic in-calf heifers and 33 bulls, dairy plants and feed mills for the MPUs, a regional diagnostic laboratory and production of vaccines, consulting services, and a training center for each MPU. There was no provision for direct farm-level credit, since the ARC already operated a credit scheme for wells and minor irrigation. 2.10 The NDDB was again to be relied upon to provide key staff and spearhead teams to help establish the MPUs and DCSs. All procurement of Bank-funded items would be by the RDDC; and Bank funds would flow to the MPUs and RDDC through the government of India and ARC. (husband and wife) membership, members who did not supply milk to the DCS and some who did not even have milch animals. 21 Madhya Pradesh 2.11 The Madhya Pradesh credit was for US$16.4 million. The dairy industry, as well as the cooperative movement, was underdeveloped in Madhya Pradesh. Started in 1960 to supply the government of Madhya Pradesh's milk plants, the cooperative movement suffered from low official milk prices, and at appraisal only 100 dairy cooperative societies were active. 2.12 At project inception, the government milk scheme was in disarray. It provided about 10 percent of urban demand at Rs 1.70 per liter (compared with Rs 1.50-2.50 per liter in the private market). Producer prices were not competitive. The government-operated milk plants had a capacity of 20,000 liters/day (1/day) in Bhopal and Indore, and 2,000 1/day in Ratlam, Ujjain, and Hoshangabad. The project called for a complete revamping of supply with the construction of three 100,000 1/day plants, each capable of expansion to 200,000 1/day. These plants would each be the center of three MPUs under the auspices of a new umbrella organization, the Madhya Pradesh Dairy Development Corporation (MPDDC). Each MPU would have 400 newly-formed DCSs with memberships of about 130 farm families (predominantly small farmers and the landless) for a project total of 160,000 families. Equipment for the milk-processing plants and associated feed mills would take about 80 percent of the foreign exchange costs of the project. 2.13 Other important project components were the importation of 100 exotic cattle for crossbreeding, the provision of consulting services, and the construction of a training center for each MPU. The NDDB was to be contracted to supply at least some of the consulting services and spearhead teams to facilitate the establishment of the MPUs and to take the lead in the formation of village-level DCSs. In accord with the Anand principles, the DCSs, MPUs, and MPDDC would be farmer-controlled. Initial working and start-up capital, to be provided by the governments of India and Madhya Pradesh, would be repaid out of profits, so that eventually both ownership and control would be in the hands of farmers. The MPUs and MPDDC would employ professional managers and technicians, again in accord with the Anand principles; they would also be free to set retail and farm-gate milk prices. 2.14 Funds would flow from the Bank to the government of India, which would channel US$2.75 million to the state government for training and extension services and the balance through the ARC to participating local banks and hence to the MPDDC and MPUs. The MPDDC was to be the sole procurement agent for all items financed by the Bank. The National Dairy Project 2.15 This fourth dairy project was an integral part of OFII, which was to be jointly funded by the Bank (US$150 million from the IDA), the EC (US$100 million in the form of food aid), and the government of India (Rs 94 million, or US$10.9 million as a grant, and a further Rs 670 14 million ($77.7 million) as a loan); farmers were to contribute Rs 48 million (US$5.6 million). This relatively large project followed a visit in 1978 by then-President of the Bank (Mr. McNamara) to Anand and his offer to Dr. Kurien to provide major funding for a successor project to OFI. 14. In the event, the government of India's projected contribution was not needed. (Annex B, Tables B7 and B8 show projected, SAR, estimated, and actual financing.) 22 2.16 Conceptually, the NDP was a follow-on project aiming to extend the Anand model to additional states, bring the three states assisted earlier by the Bank frilly into the Operation Flood system, and provide further resources for expansion of those states that had benefited from OFI and the earlier state projects. An important design difference was that the specific investments to be made in OFII were not spelled out in the SAR, nor indeed were they appraised by the Bank. Rather, a line of credit was to be provided to the IDC for on-lending to state dairy cooperative federations and MPUs.15 The NDDB was expected to appraise investment proposals on behalf of the IDC; if satisfactory, they would be funded. The first five appraisals were to be vetted and approved by the Bank, but subsequent appraisals would merely be sent to the Bank for information. 2.17 As part of a process of bringing the state projects into line with the Anand model, the state dairy development corporations were to be wound up and replaced with state cooperative milk producers federations, to be owned in due course by the MPUs, which were to be owned in due course by their supplying DCSs. 2.18 Though the intention of the project was clearly to replace state-owned apex organizations with state cooperative federations owned by their constituent MPUs, and though the SAR (Schedule A) included a model agreement between IDC and a state government to introduce the three-tier Anand model, adoption and implementation of these model agreements were not included in project conditionality. The consequence, at project completion, of this apparent oversight was well described in the PCR (Box 3.1) and unfortunately continues to plague Operation Flood. Progress is being made in the assertion of farmer control, but state domination of the apex "cooperatives" is still widespread. 2.19 The Anand model clearly started from the ground up. Though attempting to emulate this model, Operation Flood and the Bank projects were "top down," at least to the extent that target rates of growth were set in the Bank projects. In the event, the apex federations and MPUs were less independent of government than the Anand model stipulates. This is not to say that DCSs were ever organized without proper grassroots support. The spearhead teams served to mobilize farmer support, not to substitute for it. Some villages rejected the organizing efforts of the spearhead teams. This mobilization effort was the key to the rapid growth in number of DCSs. But there was a certain tension im licit in setting targets for how many villages would voluntarily decide to form a DCS. 6 15. By the time the NDP was appraised, the earlier projects had found that channeling funds from the ARC through participating banks to the MPUs was too cumbersome and had replaced the procedure with direct lending from the Indian Dairy Corporation (IDC) to the MPUs (with loans guaranteed by both the federations and the state governments). Originally, the IDC and NDDB were set up as twin institutions with a common chairman and board of directors. The IDC was responsible for the financing of cooperative dairy processing plants and acted as project unit for implementation, including letting tenders and bid evaluation. NDDB was a purely technical organization responsible for plant design, consultancy services, staffing of "spearhead" teams, and research. Under NDPII, the IDC was folded into NDDB with the objective of reducing overheads and streamlining project implementation. 16. Thus one investigator reported: "In the Bhopal region . .. the village farmers have had almost no hand in the planning or implementation of the new dairy scheme. Inquiries in three (sampled) DCS villages ... established that the DCSs were organized as follows: some officials from Bhopal came to each village, offering to show a film about the benefits of cooperative dairying and then urged village leaders to sign up members of a DCS. The selection of which villages to include in this process, and when to approach them, was made entirely in the dairy offices in Bhopal. Moreover, the Bhopal Dairy Union (not to mention the Madhya Pradesh Cooperative Dairy Federation) was not formed by a process of unifying leaders and initiatives from among the village DCSs. Instead, it worked the other way 23 The Second National Dairy Project 2.20 The managements of Operation Flood and the Bank were acutely aware that the rapid expansion achieved under the NDP had, to a significant extent, sacrificed at the federation and MPU level the Anand principles of farmer control (Box 3.1). The follow-on project, NDPII, was proposed to restore the Anand principles, serve additional DCSs, and allow for further expansion of milk processing capacity. 2.21 External assistance of US$510 million was to be provided (US$360 million from the Bank and US$150 million in the form of food aid from the EC). Domestic resources equivalent to US$164 million were to be provided by NDDB. Hardware of one sort or another to facilitate milk collection, transportation, testing, processing, and retailing dominated project costs (70 percent). Another 24 percent of project cost was for the development of DCSs and productivity enhancements at the farm and village level. Institutional strengthening was to take 6 percent. The bulk of these funds were to flow through the NDDB to MPUs and federations, but this would be subject to tight conditions on progress toward adopting the full Anand principles, thus remedying the defect of the previous project design noted in para. 2.18. The first condition agreed was: Operation Flood (OF) model criteria to be adopted by a cooperative, or to be in meaningful progress towards adoption, as a condition ofNDDB funding under the project: (i) Adoption by Federations, MPUs, and DCSs of the Operation Flood cooperative model bylaws to ensure democratic and accountable operations; (ii) Elected boards of producer members for DCSs, MPUs, and Federations; (iii) Autonomy for the Federations and MPUs in the appointment of professional management, employment of staff, pricing, and marketing; (iv) Ownership and management of plants (dairies, chilling centers, feed plants, etc.) by the MPUs, or in areas where the plant serves a central purpose (city dairy or central feed plant) by the Federations; and (v) Proper internal control procedures and accounting and auditing of the Federation, MPUs, and DCSs involved, including a system whereby this information is regularly reviewed by the respective Boards and General Body Meetings (World Bank Reports 1987a, p. 35). 2.22 As the PAR observed: "With hindsight, the project design was unrealistic .... It is now evident that the amount of money provided could not be productively invested in states willing to implement the Anand principles in full" (Memo to Executive Directors). Entrenched political and bureaucratic interests were extremely reluctant to relinquish their influence over the industry. In practice, in order to achieve target levels of disbursement, it has proved necessary to waive, in most cases, the conditions in para. 2.21. around: the DCSs were formed at the behest of the Union and the Federation. Consequently, in the Union's annual reports, members of the Board of Directors, consisting of representatives from the component DCSs, are not even mentioned by name. They are probably irrelevant to the decision making process within the Union ... ." (Attwood 1985). 24 Changed Paradigm 2.23 With the preparation of the Bank's Livestock Sector Review in 1996, the Bank's paradigm for the development of the Indian dairy industry changed. Whereas the SAR for the NDPII reported that the Indian dairy industry was competitive (para. 4.23) and provided resources to be on-lent to the cooperative sector, the sector review reported that the industry was not competitive and recommended that funding of loans to Operation Flood be phased out (World Bank Reports 1996, para. 6.10). In large measure, this change seems to have been triggered by a belief that the private sector might be more efficient than the cooperative sector and a serious misreading of the cost of Operation Flood. The Livestock Review (World Bank Reports 1996, Table 4.2) estimated this cost as US$4.26 billion in 1990 dollars (US$5.06 billion in 1996 dollars) versus a reestimate by this study of US$2.98 billion in 1996 dollars (Annex B, Tables B3 and B 15).1 Whatever the reason, the Bank ceased to believe in the project design to the point that, in recommending that the Bank reverse itself on an extension of the project, an internal memorandum (October 4, 1995) could say: "However, the key argument [for no extension of the project beyond December 1995] is the project design is not consistent with the present GOI [Government of India] and Bank strategy of leveling the playing field for agroindustrial development between cooperative and private entrepreneurs, thus putting the project at odds with the recommendations of our macroeconomic work as well as the recent livestock sector review." 17. The impact estimate includes an allowance for state government assistance and central government assistance to the dairy industry as a whole. If these factors are subtracted, then the impact estimate (for costs included in the Livestock Review) is US$1.87 billion (1996 dollars) versus US$5.06 billion. 25 3. Impact of the First Four Projects at Completion 3.1 The project completion reports (PCR), implementation completion report (ICR), and PARs are silent as to any benefits to the beneficiaries under these four projects. Rather, they report on progress in establishing DCSs, expanding milk procurement, installing milk processing and feed production capacity, introducing crossbred cattle, and the like. As to how and whom these changes benefited, there is little to be gleaned in the reports themselves. However, two joint studies between the Bank and the International Food Policy Research Institute (IFPRI) (Alderman, Mergos, and Slade 1987; Alderman 1987) and a study sponsored by the NDDB (Singh and Acharya 1986) undertaken about the time of project completion throw some light on the projects' immediate impact. Targets 3.2 After some delay, the four projects came close to their intended targets in terms of DCSs and MPUs formed (Table 3.1), dairy processing and feed plants constructed, and Al. However, with respect to milk procurement and importation of dairy stock, they fell far short. Table 3.1: DCSs and MPUs Formed DCSs MPUs Project Target Actual Target Actual Karnataka 1,800 1,803 4 4 Rajasthan 1,084 3,148 5 5 Madhya Pradesh 1,200 849 3 12 NDP 33,300 42,700 89 164 NDPII 70,000 69,675 190 170 Source: World Bank Reports 1987c, World Bank Reports 1997b, and study estimates. Organization 3.3 The three state projects provided for parastatal dairy development corporations to be created as apex organizations. The boards of directors were to be a mix of farmer-elected, ex officio, and government-appointed members, and the same pattern applied to MPUs. OFII provided for the apex organizations to be reconstituted as cooperative milk federations. However, this turned out to be simply a change in corporate name. The constitution of the boards of directors and the operating style were unaffected by the change. This was a problem at completion and continues to be a problem to this day (paras. 4.17-4.21). 3.4 The transfer or lease by states of government-owned milk plants and farms to the new dairy development corporations was to be a condition of project funding at the state level (World Bank Reports 1978, Schedule A, p. 1). This condition was waived, however, and had not been implemented at project completion, nor has it been finalized to this day. Ironically, ten years after project closing, and despite a further follow-on project focused exactly on this issue (para. 2.20), not all the agreed conditions for lending to the states have been met, although there has been significant movement. Great progress was made in creating a cadre of dairy professionals 26 and organizing them into spearhead teams to mobilize milk producers and to run the resulting MPUs and the support services of the NDDB. Box 3.1: Establishment of the Anand-Pattern Structure A three tier Anand-pattern structure has generally been established throughout India. As established, however, this structure needs to be further rationalized in a number of directions if the objectives of establishing an Anand-pattern structure owned and controlled by the farmers are to be achieved. In Gujarat, the federation was essentially established as a result of the felt need of the individual union to establish a common brand name and a common marketing platform for unions and to enable unions to coordinate various activities. In most other states, however, the federation was established as an instrument for implementing Operation Flood II in order to establish unions and DCSs in an effort to create a farmer-owned and -controlled organization. In many states, particularly the earlier IDA-aided projects in the states of Rajasthan, Karnataka, and Madhya Pradesh, existing government dairy development corporations were converted into federations. Whereas the unions in Gujarat were essentially strong, viable, and owned and controlled by farmers through elected boards, the federations and unions formed in many parts of India were established with government-nominated boards and controlled by government. In many cases, competent professionals employed by the federations ensured that they were building a system which would ultimately lead to the emergence of an Anand-pattern structure. In certain cases, however, the systems developed were essentially aimed at providing benefits to the consumer, sacrificing the milk producer. In many cases, the feeder balancing dairies at the union level have remained the property of the federation and not of the union, resulting in a strong federation and a weak union with the farmer losing control of the system. Nowhere, except in Gujarat, Kerala, and Pondicherry, is the federation actually controlled by an elected board. During the initial years as Operation Flood was being introduced, it was necessary to establish an organization (i.e., federation) responsible for implementing the program in the states. As milk collection and sale have expanded and DCS and unions have been established, it is now necessary to effectively transfer ownership control and effective power to the elected boards at the union level. In an effort to strengthen unions, Operation Flood III now aims at transferring feeder dairies and balancing plants to the concerned unions and ensuring the establishment of elected boards at the union and federation level. Initial difficulties are expected as bureaucratic structures already established at the federation level may impede the transfer of power to the unions and their elected leadership. Source: World Bank Reports 1987b, p. 266. Impact 3.5 The NDDB-sponsored management information system (MIS) provides comprehensive feedback to all levels of Operation Flood on the business and procurement aspects of the program. No such comprehensive monitoring system is in place to measure other dimensions of the projects' impacts on the intended beneficiaries. This being the case, even well-designed, but geographically constrained, statistical samples are nevertheless in the nature of "anecdotal" information, given the huge size, vast spread (and hence diverse climatic and social conditions) of Operation Flood as a whole. 27 Milk Production Karnataka 3.6 A Bank-sponsored study of the impact at completion of the Bank's project in Karnataka compared conditions in Operation Flood villages with those in non-Operation Flood villages (Alderman 1987). The study involved 42 villages with cooperatives and 20 without, divided equally into two climatic zones. Thirteen households were interviewed in each village in five successive rounds of interviews. This study showed that milk production per family was from one and a half times to twice as high in Operation Flood as in non-Operation Flood villages (Table 3.2). The higher production in Operation Flood villages was due primarily to herd composition, not numbers of animals and not yield per animal. Operation Flood villages had a higher proportion of buffalo and crossbred milch animals and fewer local cows. A multiple regression that controlled for (among other things) literacy, household size, farm size, and 1/household/day estimated 2.56 more 1/household/day from Operation Flood villages than non- Operation Flood villages (Alderman 1987, p. 18). A multiple regression analysis showed that the nominal price of milk was 3.2 percent higher in Operation Flood villages; deflated by a village cereal price index, it was 1.9 percent higher (Alderman 1987, Table 9 and p. 29). This raises some questions: is the elasticity of supply very high? or does the introduction of Operation Flood shift the supply function? or were DCSs established mainly in villages with high levels of dairy activity? Perhaps some of all three, but probably more of the latter. Table 3.2: Karnataka: Daily Household Milk Production (1/household/day) Village Round (Date) Flood Non-Flood Ratio 1 (January-April 1983) 3.61 1.80 2.00 2 (May-July 1983) 2.56 1.57 1.63 3 (August-October 1983) 2.95 1.57 1.88 4 (November-January 1984) 3.07 1.93 1.60 5 (March-April 1984) 2.91 1.77 1.64 Source: Alderman 1987, Table 1. Madhya Pradesh 3.7 By contrast, a Bank-sponsored study in Madhya Pradesh found no such increase1 in milk production, as summarized in Table 3.3. Indeed, in rabi DCS villages averaged 14 percent less milk production (a "decrease") than non-DCS villages. This is confirmed by Mergos and Slade (1987, Table 6.1) who reported an average milk production of 2.91 liters per day in DCS villages against 2.83 liters per day in non-Operation Flood villages. However, after controlling for area of land, percent irrigated, number of milk animals, value of milk animals, and several inputs, Mergos and Slade (1987, Table 6.3) found a 13.4 percent higher production in DCS villages. 18. Strictly, not "increase" but "difference between Operation Flood and non-Operation Flood villages." There is no "before project" baseline data, so that relative changes cannot be measured. We cannot tell whether the difference is due to Operation Flood or to a sampling error. This applies equally to the Karnataka and Madhya Pradesh studies; the "increase" reported in Karnataka could reflect sampling error. 28 Table 3.3: Madhya Pradesh: Average Milk Production and Disposal (1/household/day) Season Produced Sold Consumed Converted Rabi DCS villages 3.59 1.95 1.53 0.11 Non-DCS villages 4.16 2.40 1.54 0.22 Kharif DCS villages 4.31 2.68 1.45 0.18 Non-DCS villages 4.34 2.54 1.53 0.27 Source: Singh and Acharya 1986, Tables 5.8 and 5.9. 3.8 Another concern is the distributive effect of Operation Flood. Some inconclusive results from the Madhya Pradesh sample are reported in Table 3.4. The rows of Table 3.4 refer to the season and whether milk was sold to a DCS or other buyer (usually a dudhiya). The DCS column refers to whether there was a DCS in the village. These results are difficult to interpret. Presumably, the difference for landless and large farmers supplying the DCS are due to large farmers providing higher-testing milk, since the DCSs use a test-based formula to determine payout. There is no reason to suppose that this quality differential also characterizes milk supplied to non-DCS purchasers, whether in a DCS or non-DCS village. Indeed, as Singh and Acharya (1986) note: One must note here that the formal price figures, taken as such without other considerations in a non-DCS context, are deceptive in the sense that they are purely nominal, The private vendors invariably cheat the producers not only in the measurement of milk but also in the determination of unit price level, which is supposed to be based on 'Mawa' (total solids) content (93-94).19 Table 3.4: Madhya Pradesh: Average Milk Prices, Landless and Large Farmers (Rs/1) Milk Buyer DCS in Village Landless Large Ratio Rabi DCS Yes 2.41 2.66 1.10 Other Yes 2.42 3.04 1.26 Other No 2.57 2.78 1.08 Kharif DCS Yes 2.37 2.47 1.04 Other Yes 2.46 2.36 0.96 Other No 2.43 2.46 1.01 Source: Singh and Acharya 1986, Table 5.10. Milk Consumption Karnataka 3.9 The study of project impact in Karnataka addressed the crucial issue of the impact of Operation Flood on milk consumption (Alderman 1987). Multiple regression analysis was used 19. This quote is, of course, merely an assertion, though a very widely believed one (para. 4.11). Jaipur dairy now provides a service by which consumers who suspect water has been added can have milk samples checked (Table 4.3). Though not a random sample, this service has clearly demonstrated that some vendors certainly sell watered milk. 29 to explain the budget shares of milk, ghee, and other dairy products in the consumption of Operation Flood and non-Operation Flood villages. A key conclusion was that "the average budget share of milk was 9 percent higher in the control (that is, non-Operation Flood) villages than in the cooperative (Operation Flood) villages, given equal income and production" (Alderman 1987). Unfortunately, this is difficult to interpret since production (and incomes) were higher in Operation Flood villages. Taken at face value, the above quote would seem to support the substantial (but not well documented) literature that claims Operation Flood has led to the sale of milk which producers used to consume themselves. However, the raw consumption data in Table 3.5 are more reassuring. They suggest a slightly higher average consumption in Operation Flood than in non-Operation Flood villages, despite the fact that a much lower proportion of milk is being retained in Operation Flood villages because milk production there is substantially higher. But the real issue, as Alderman (1987) notes, is the net effect on nutritional status. A rise in the price of any (noninferior) good should lead to lower consumption of the good and higher consumption of substitutes. In the study sample, it would appear that the income effect just about compensated for the substitution effect, leaving total consumption of milk per household substantially unaffected. Table 3.5: Karnataka: Home Consumption of Liquid Milk (1/household/day) Vdlage Proportion (%) Round (Date) Flood Non-Flood Flood Non-Flood I (January-April 1983) 0.81 0.83 22.4 46.1 2 (May-July 1983) 0.64 0.60 24.9 38.6 3 (August-October 1983) 0.81 0.62 27.3 39.6 4 (November-January 1984) 0.79 0.72 25.7 37.2 5 (March-April 1984) 0.71 0.56 24.3 31.5 Source: Alderman 1987, Tables 1 and 10. Madhya Pradesh 3.10 Levels of milk consumption in the Madhya Pradesh study have already been reported in Table 3.3. Even though milk consumption was 5 percent less in DCS villages in kharifthan in non-DCS villages, overall one would have to conclude, as for Karnataka, that on the basis of these samples, there was little reason to be concerned about a wholesale diversion of milk from consumption to the market. Rather, the concern in Madhya Pradesh should be that dairy income in DCS villages may not have increased significantly. Multiplying average price by average output gives a 6.25 percent higher dairy income in non-DCS villages. Importance of Milk Income Karnataka 3.11 The ratio of the value of milk production to total expenditures in the Karnataka sample is given in Table 3.6 for a range of farm sizes. All landholder classes had higher ratios of milk sales to total expenditure in Operation Flood villages. In both Operation Flood and non-Operation Flood villages, the larger farmers got a larger proportion of their income from milk sales than smaller farmers. However, the spread was less in Operation Flood villages (6.8 percent instead of 8.4 percent). 30 Table 3.6: Karnataka: Ratio of Milk Sales to Total Expenditure (Percent) Village Landholder Class Flood Non-Flood Landless 16.4 10.7 Marginal 16.0 12.0 Small 21.8 12.5 Other (large) 22.8 19.1 Source: Alderman 1987, Table 19. Madhya Pradesh 3.12 The Madhya Pradesh sample confirms this relationship. As shown in Table 3.7, the ratio of milk sales to other income is higher in Operation Flood villages for all classes except the landless, and slightly higher on average. When dairy income is regressed on other income and a dummy variable for a DCS village, the presence of the DCS is estimated to raise dairy income by Rs 527. The t-statistic of 1.43 is significant at the 10 percent level. On the basis of these two studies, one could say that with the important exception of the landless in Madhya Pradesh, Operation Flood appears to have raised the proportion of income gained from dairying. However, the increase has been quite modest. Table 3.7: Madhya Pradesh: Milk as a Proportion of Average Annual Income Land Holding DCS Village Non-DCS Village Descriptor Size Milk Total Ratio Milk Total Ratio Bighas4- <-Rs/household-> % <-Rs/household-> % Landless 0 1,015 4,151 24.5 1,469 4,702 31.3 Marginal 0-6 1,433 5,184 27.7 999 5,787 17.3 Small 6-11 1,785 6,061 29.5 1,086 5,164 21.0 Low Middle 11-21 2,131 6,612 34.3 3,022 9,500 31.8 High Middle 21-31 2,290 10,273 28.4 1,747 11,161 15.6 Large 31+ 3,829 13,558 28.2 3,680 14,573 25.3 All Classes 2,135 8,144 26.2 2,193 9,407 23.3 a. I Bighas = 0.23 hectares. Source: Singh and Acharya 1986, Table 7.1. Profitability of Dairying 3.13 Only the Madhya Pradesh study addressed profitability (Singh and Acharya 1986). This study provides impressive evidence of the profitability of dairying. After all costs have been accounted for, the profit percentages are as given in Table 3.8. The gross margins reported are so high that they raise some question as to whether all variable costs have indeed been taken into account.20 Only in the case of the landless in rabi is it reported that milk production fails to cover "variable costs" (which include the opportunity cost of all nonpurchased inputs and services used in milk production), and then the loss is only 5 percent, suggesting that even in this case it may be a useful way of using spare labor, albeit at a slight discount. 20. "The variable cost includes the actual cost of all purchased inputs and services such as fodder, concentrate feed, animal health care, breeding, hired labor, etc. and the opportunity cost of all non-purchased inputs and services used in milk production" (Singh and Acharya 1986, p. 94). 31 Table 3.8: Madhya Pradesh: Gross Margins in Milk Production (Percent) Season Operation Flood Village Non-Operation Flood Village Rabi 30.0 42.0 Kharif 83.6 72.0 Source: Singh and Acharya 1986, Tables 5.11 and 5.12. Benefits 3.14 Only the Madhya Pradesh study asked directly about benefits (Singh and Acharya 1986). The reaction was remarkably positive, as shown in Table 3.9. This table provides evidence of substitution of grains for milk and milk products in consumption (more than would seem reasonable on the basis of the village comparisons discussed earlier). Remarkably, the second and third biggest increases are attributed to increased education of girls and boys, respectively. If this claimed increase is indeed causal, this would be a major positive impact of Operation Flood. Table 3.9: Madhya Pradesh: Opinions on Impact of DCSs on Beneficiaries (Percent) Query Increase No Change Decrease For household: Food consumption 43.2 52.5 4.3 Coarse grain consumption 30.1 59.1 10.6 Wheat and rice consumption 38.3 53.1 8.6 Milk consumption 28.6 36.2 35.0 Curd consumption 7.2 46.0 46.0 Buttermilk consumption 6.6 45.7 47.1 Ghee consumption 11.8 51.4 35.9 Home milk processing 8.3 39.2 50.9 Family income 36.7 42.7 19.7 Intensity of labor in dairy work: Family 36.1 37.4 26.5 Adult male 37.9 39.0 23.0 Adult female 34.8 43.6 21.5 Family labor in agricultural work 32.3 61.6 5.9 Clothing for family 47.8 48.5 3.7 School attendance of: Boys 62.8 29.1 8.1 Girls 67.4 29.7 2.9 Use of fertilizer 52.1 42.0 5.6 Indebtedness 58.7 28.9 11.9 Availability and use of: Veterinary services 68.1 26.5 4.5 Fodder supplies 58.1 31.6 9.9 Household status 17.0 80.8 2.0 Health Awareness 43.7 50.5 2.0 Source: Singh and Acharya 1986, Table 7.19. 3.15 The reported increase in indebtedness is ambiguous. If it represents increased credit- worthiness that has been utilized profitably, then it should be counted as a benefit. If it has been 32 used unwisely, or is simply indebtedness without augmented creditworthiness, then it should be counted as a cost. 3.16 The companion study (Mergos and Slade 1987), for which results are given in Table 3.10, was less specific about the nature of benefits received from the cooperative system. Of those with an opinion, 86 percent believed that the operation of the DCSs was good or satisfactory. The primary reason for membership in the DCSs was to sell milk and take advantage of the year-end bonus. Despite this focus on milk price and collection, over 70 percent of those with an opinion thought that the DCSs provided additional benefits. Perhaps the best testimony to the good reputation of the DCSs is that 63 percent of those interviewed in the non-DCS villages said that they would join a DCS if it was formed in the village. Table 3.10: Madhya Pradesh: Farmers' Opinions of the DCSs and Their Reason for Joininga b) Additional Benefits Received by a) Performance of DCSs (0%) Membersb %) Member Non. All Member Non. All Good 43 28 34 Yes 67 52 58 Satisfactory 43 34 38 No 29 19 23 Bad 12 10 11 Don't know 4 29 19 c) Reason for Joining d) Reason for Not Joining Members (%) Nonmembers (%) Sell milk 76 Could not buy shares 17 Get bonus 9 No spare time 13 Obtain vet. services 7 DCS "not useful" 19 Other 8 No specific reason 51 Total 100 Total 100 a. Households with milch animals in DCS villages. b. In addition to milk price and milk collection. Source: Mergos and Slade 1987, Table 2.7. Summary 3.17 At completion, the growth of production of the industry as a whole had received a jolt and had started to grow significantly faster. At the same time, the rationing of milk in the major cities had been dropped, and the formal sector market for milk had also been put onto a new fast- growth path. Literally millions of small farmers were being drawn into the cooperative system, and vast new programs in veterinary services, Al, improved animal nutrition, crossbreeding, and vaccination were being introduced. Yet the three village-level studies for which results are readily available show relatively little difference between Operation Flood and non-Operation Flood villages. Certainly the studies do not show the degree of difference one would expect if these village-level differences were to be used to explain the national-level phenomena observed. The easiest escape from this apparent paradox is the likelihood that dairying was growing vigorously in both Operation Flood and non-Operation Flood villages. At the same time, demand was growing vigorously due to rising per capita incomes, population growth, and the high income elasticity of milk (1.1 urban, 1.6 rural). That this increase in demand did not result in 33 21 rapidly escalating real prices or heavy imports of milk solids is best explained by the removal (or at the very least the relaxation) of the dead hand of the public sector from dairy development (a removal that was part and parcel of Operation Flood, the program being supported by the Bank projects) and the continuation of the import-substitution strategy which provided high protection levels to dairying and thus an incentive for domestic production to grow. 21. A reviewer has emphasized the "self-reliant" development model being followed at the time (para. 1.3). As shown in Table 4.6, effective protection was very high so that, in the absence of these macroeconomic policies, it is likely that imports would have been substantial.  35 4. Developments Under National Dairy II Expansion and Nonfunctional Cooperatives 4.1 The NDP, which supported OFII, was followed by the NDPII, which supported OFIII. OFII had supported the establishment of 25 apex organizations, 125 new MPUs, and over 32,000 additional DCSs.22 OFIII, the follow-on project, was intended to support the establishment of only 20 new MPUs, but 27,000 new DCSs. 9 Thus, while OFII was primarily an expansion project, it has been followed by a period of active consolidation, as shown in Table 4.1. Table 4.1: Formation of Dairy Cooperative Societies Cumulative Cumulative Functional Year Organized Disbanded Balanceb Disbanded (%) 1980-81 13,284 1,639 11,645 12.3 1985-86 42,692 6,973 35,719 16.3 1988-89 58,883 12,941 45,942 22.0 1991-92 64,057 13,674 50,383 21.3 1992-93 65,469 14,622 50,847 22.3 1993-94 67,247 15,048 52,199 22.3 1994-95 69,771 16,795 53,066 24.1 1995-96 72,744 17,702 55,042 24.3 1996-97 (Prov.) 73,931 18,268 55,663 24.7 a. Includes both DCSs for which the Registrar of Cooperatives had officially acknowledged the winding up of the coop, and coops which may pour at other times of year. b. DCSs that "poured in June." Source: NDDB (Anand) and study data. 4.2 The NDP closed in November 1985, when about 42,000 DCSs had been organized and 36,000 were functional in 164 MPUs. OFIII has seen a further 30,000 DCSs organized (functional DCSs have risen by 19,000) in 170 MPUs. (MPUs peaked at 173 in 1988-89). There are no separate statistics either on DCSs formally wound-up by the state Registrar of Cooperatives, DCSs waiting to be wound up, and DCSs that did not supply in June but are active at other times of the year. This means that the "disbanded" figures in Table 4.1 are something of an overestimate. To further complicate matters, there are examples where mixed (that is, male- dominated) DCSs were disbanded in a village, only to be replaced by a women's dairy cooperative society (WDCS) and even cases where a mixed DCS and WDCS coexist in the same (large) village. 4.3 The bottom line is that Operation Flood has continued to expand. There are now 21,000 more active DCSs "pouring in June" than there were at the completion of OFIL. Average procurement has also risen, but less quickly. Average daily milk procurement per functional DCS has fallen from 220 I/day to 194 1/day, and average daily milk procurement per member has 22. Sixteen state-level federations and nine "apex MPUs" in small states where only one MPU was needed. 23. In fact, only eight new MPUs were formed. 36 dropped from 2.1 I/day to 1.5 1/day.24 Like DCSs themselves, some of the 9.3 million farmer members of DCSs have become "nonfunctional" either because the DCS is nonfunctional or because they have ceased to "pour" to the DCS. Allowing for nonfunctional DCSs and multiple memberships, NDDB estimates that there are about 6.3 million "pouring members" who supply milk to a DCS at some time during the year (Annex B, Table B 16). 4.4 The reasons why a quarter of the DCSs are classified as "nonfunctional" are not well documented. Many will simply be "in recess" in June with milk being poured at different times of year. Others have been officially disbanded by the Registrar of Cooperatives, and others are awaiting this action. Five major causes seem to explain the closure of DCSs: * Active price competition, especially in peri-urban villages. With only a small distance to transport milk and negative milk processing costs, small traders can offer prices higher than the cooperative. As urban areas expand, the area served by small traders expands. To the extent that farmers get more for their milk than they would from the cooperative, this expansion is to be welcomed. The question as to whether the consumer is better served by informal traders is an open one, since there is evidence that watering of milk is widespread (Table 4.3). * Changed needs of milk producers. Where villages become more decentralized, the single DCS pouring site may cease to properly meet farmers' needs. Itinerant buyers who will come to the farm might then provide a service the DCS does not match. (In some large villages, DCSs have established one or more satellite pouring points to reduce waiting lines and travel times.) * Village politics. Though the Anand principles call for the DCSs to be open to all without discrimination, this may not be honored, leading to tensions within the village that can eventually undermine the DCS. * Dishonesty. Despite a sophisticated system of checks and balances and "double- entry" checks on physical quantities and payments, embezzlement does occur, often followed by the collapse of the DCS. * Higher income. In a Madhya Pradesh study (Hiremath, Singh, and Mergos 1997), several DCSs were found to have closed due to the introduction of soybeans. These provide a higher income than dairying but fewer byproducts for cattle feed than the cotton and groundnuts they replace (para. 5.11). 4.5 If the volume of milk supplied by a DCS falls, for whatever reason, below an economic level, then the MPU will cease collecting. A DCS may thus become "nonfunctional" not because of explicit dissension or a wish of the membership to disband, but because the volume of milk supplied falls to a level that makes collection uneconomic. 24. Calculated from NDDB data. This is an ambiguous statistic, since farmers who have joined a DCS are often reluctant to drop their membership even if they cease to own a milch animal. Some DCSs allow two members per family, and some even allow membership from families who do not own a milch animal. There are no national statistics on pouring members in earlier years. Case studies provide examples of villages where only half the membership actually pours (Sampark 1997). 25. With many buyers (since it is a short distance to town) as well as many sellers, there is a better chance of atruly competitive market for milk than in remote villages where market failures are to be expected in the absence of a DCS. 37 4.6 With almost a quarter of DCSs that have been formed nonfunctional in June, this is a serious problem for Operation Flood. MPU staff were well informed about the range of circumstances that can lead to the collapse of a DCS, but the causes of collapse are not routinely documented. Under OFIII, the NDDB initiated a Cooperative Development Program to help deal with this problem. This is a multidimensional program aimed at in-service training of all participants in the cooperative system from farmers to boards of directors and professional staff, with particular emphasis on women. The training includes village-level discussions of cooperative principles and rights and responsibilities, facilitated discussions among management committees of neighboring DCS on problems and procedures, visits by DCS chairmen (and chairwomen) to Anand to see the support infrastructure that has been built up and to be further indoctrinated in their rights and responsibilities, and technical training for secretaries on AI and accounting procedures. In principle, this program funded all the expenses for one five-member team in each MPU and the nonsalary expenses of another team. The program is active in 89 MPUs, some of which have financed a total of five teams. The impact of this educational program has been twofold: first, a greater sense of belonging and ownership of DCSs by their 26 members, and second, better performance, particularly at the DCS level. The result has been an increase in milk poured in villages under the program compared with other villages. Since higher procurement helps spread overhead costs, this increased milk production suggests that the program has a one- to two-year payback period. While the Cooperative Development Program is intended to be preventative, the Technology Mission for Dairy Development (jointly funded by the central government, state governments, and NDDB) has been active in reviving nonfunctional DCSs. 4.7 It is Operation Flood policy not to receive milk from small traders or dudhiyas. This is very largely driven by quality concerns (i) that the milk may have been watered and (ii) that elapsed time since milking may have allowed souring to commence. In addition, cooperative ideology puts a high value on a direct link to the producer. This said, it may be noted that where a dudhiya is collecting from producers not otherwise served by the DCS (say from a village several miles distant) or by direct pick-up from farmers, there might be a case for collaboration. This would be true especially if the cans supplied by the dudhiya were kept separate and individually tested on receipt at the factory or cooling center. 4.8 As reported in the ICR and Table 4.2, the investments under NDPII turned out to be fairly evenly distributed with the exception of Gujarat, which received 40 percent of funds released by NDDB. However, because of the high level of milk procurement in Gujarat, the investment per kilogram of milk procured turns out to be only 36 percent above the average for the project as a whole. More extreme is the underinvestment in Maharashtra, which received only 24 percent of the average investment per kilogram procured. That the investment in Gujarat was not excessive relative to milk procurement is suggested by the ICR (para. 18) when it observes, "The data on rated capacities and actual utilization in the Operation Flood dairies 26. Though not by any means widespread, abuses that have been identified and corrected as farmers have better understood the system and their rights include (i) the failure to actually test milk (that is, an arbitrary assignment of fat content by the DCS Secretary); (ii) acceptance of milk from a milk trader; (iii) misconduct by the Managing Committee or Chairman, who has been forced to resign; and (iv) refusal of membership on the basis of caste or "politics." By contrast, a DCS chairman who had been on the Cooperative Development program, responding to a question posed during a group interview, said spontaneously: "The village cooperative has to follow the Anand principles. Firstly, just as we treat all milk the same in the society, we should treat all people the same. Secondly, there should be no politics within the society. The society should be like a Hindu Temple. You take your shoes off before entering and no politics within the Temple." 38 showed considerable inter-and intra-state variation. Gujarat and Maharashtra do better than most." It appears that on efficiency grounds Gujarat was not overbuilt. However, the high proportion of milk production procured by the cooperative system in Gujarat relative to other states means that investment per kilogram of production (not procurement) was four times as high in Gujarat as in almost any other state. Similarly, the investment per member was higher in Gujarat than elsewhere. Table 4.2: Milk Procurement, Production, Cooperative Membership Milk Milk Production Procurement Release of in 1990 Producer (volume in Funds by (volume in Members of OOOs) NDDB Rs/kg 000s)" Rs/kg Cooperatives (kg/day) (Rs million) Procurement (kg/day) Production (000S)b Rs/Member Gujarat 3,313 3,843 1.16 9,657.5 0.40 1,847 2.08 Uttar Pradesh 819 806 0.98 26,553.4 0.03 478 1.69 Karnataka 1,098 860 0.78 6,545.2 0.13 1,294 0.66 Tamil Nadu 1,203 602 0.50 9,246.6 0.07 1,884 0.32 Maharashtra 1,908 391 0.20 10,235.6 0.04 1,069 0.37 Andhra Pradesh 848 628 0.74 8,246.6 0.08 701 0.90 Other States 2,091 2,472 1.18 77,290.4 0.03 1,719 1.44 All India 11,280 9,602 0.85 147,775.3 0.06 8,992 1.07 a. Basic data from Dairy India 1997, p. 160. b. Basic data from Dairy India 1997, p. 185. Source: ICR, p. 13. 4.9 The history of milk procurement by Operation Flood is summarized in Figure 4.1. There was an initial period (1970-1980) of rapid growth (15.2 percent per annum) from a small base under OFI. Five years (1980-1985) of extraordinary growth of 22.5 percent per annum followed under OFII, and subsequently more modest growth of about 4.2 percent per annum (from a much larger base) under OFIII. Under OFII, procurement by Operation Flood has grown slightly less fast than production. Two factors may help explain this. In the early period, there was an 18- month to two-year hiatus in implementation due to the reorganization of NDDB and delays in state governments' willingness to adopt the Anand principles. In the later years, competition from the private sector has slowed deliveries to the cooperative sector. 4.10 Retail Sales. Retail sales by Operation Flood have risen from an average of 1.0 million I/day in 1970 to 5.0 million 1/day in 1985-1986 to 9.40 million I/day now. 4.11 Quality and Competition. Competition both for milk procurement and sales is lively. There is no restriction on informal traders purchasing milk, and the demise of some DCSs testifies to the vigor of this competition. Tales of "unfair competition" from these traders abound, including tying suppliers by either a loan from traders to farmer or paying 30 or 60 days in arrears so that if suppliers leave, they will lose one or two months of milk income. Failure to test milk, falsification of test results, refusal to take milk (or setting a giveaway price) during the flush season, and long measures are repeatedly mentioned. Urban milk demand is met by cooperatives (60 percent), informal urban producers (30 percent), and informal milk traders and new private suppliers (25 percent). Estimated market shares add to over 100 percent, reflecting different market shares between markets. 5 g I, I ' I a s I II 04f y= 1.4027e0us R2 = 0.8375 00,,0 ** y.= .08060 2252X =00, R 2. 1 4 Period Rate of Procurement Growth per annum 1970-1980 15.2% 1980-1985 22.5% 1985-1995 4.2% 10 y = 0.1831eo 1s24X 10 ***** R2.m 0.7691 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 I II I lI I I Il Figure 4.1 India: Operation Flood Milk Procurement 1970-95 (exponential fit to 1970-80, 1980-85, 1985-1995 subsamples) 40 4.12 There is also frequent reference to the watering of milk by informal traders and by farmers. Some farmers interviewed mentioned the ability to add water to the milk as one of the advantages of selling to informal traders. Data on the watering of milk are difficult to obtain. An exception is in Jaipur, where for a number of years Jaipur Dairy has had a free service that tests milk for added water at consumers' request. (Actually the test is for solid-non-fat (SNF), and it is assumed that any SNF level below 8.5 percent is due to the addition of water.) This is a "self- selected sample," and as such is representative of milk that consumers suspect may have been watered (not the informal milk supply as a whole). The relevant data is given in Table 4.3. With low overheads and a rather simple approach to milk processing, informal traders provide vigorous competition. Table 4.3: Water in 200 Milk Samples Submitted for Testing Percent No adulteration 17 0.1-20.0% water added 40 20.1-40.0% water added 35 40.1% and up water added 8 Source: Jaipur Dairy for year 1995. 4.13 Private companies do not generally procure their milk directly from farmers. Rather they buy from contractors at the factory. Contractors usually employ informal traders to buy from farmers, thus perpetuating the unfair trading practices referred to in para. 4.11. By using contractors, private companies minimize costs but have no control over quality between the farm gate and the receiving platform (Box 4.1). This extreme example should not be taken as typical; rather, it warns of the existence of a problem. 4.14 Problems with effluent treatment have already been alluded to in para. 1.22, and a review of Bank supervision reports shows intermittent concern with nonfunctioning control equipment, and poor maintenance in some factories, at some points in time. NDDB itself operates a Quality Assurance Group which is designed to help MPUs remedy weaknesses in equipment or standard procedures. This is a continuing challenge, even as factories are being prepared for certification by the International Standards Organization for export quality production. 4.15 Increasingly, both the corporate and cooperative sectors have been plagued by contractors and transporters adding "artificial milk" to their deliveries (Box 4.2). Since this "milk" satisfies the normal butterfat and SNF tests, improved testing procedures are needed for bulk milk delivered to the factory. Although these comments are not based on extensive empirical investigation, they were not seriously challenged by leading Indian dairy academics and others when presented at a Workshop on a draft of the study held in Anand, Gujarat state, in March 1997. Since we want to know village-level impact, it is inherent that evidence has to be drawn from disparate studies, and comprehensive evidence is unlikely to become available.27 27. A reviewer has put this caution more strongly: "The data on private sector quality is weak. Data from one state-- Punjab (with one table, Table 4.3, subject to self-selection problems)-and one visit to a single milk plant (Box 4.1) are insufficient to draw general conclusions about the whole industry and country. Assertions of knowledgeable observers are not a good substitute. Several references are made throughout the text about unfair trade practice by the private sector (paras. 4.11, 6.33, etc.). However, no evidence is provided: Instead, this is based solely on references to 'tales,' conclusions made by observers (paras. 3.8, 4.4, 4.11, 5.13, etc.), and Box 4.2 on artificial milk. With thin 41 Box 4.1: Milk Assembly at Its Worst The following account of a private sector milk assembly center is not intended to reflect the hygiene standards of the subsector as a whole, but it does suggest that enforcement of the Adulteration of Food Act of 1961 is a major challenge, at least in the Punjab (Table 6.10 shows that 16 percent of milk samples drawn and tested by the Punjab Ministry of Health in 1994 were judged to be adulterated and 24 percent in 1995.) The center was used to transfer milk from cans to bulk storage in order to cool it, correct for excess acidity, and transfer it to a small bulk tanker for transport to a manufacturing plant. The building used was a former ice plant that opened directly onto a main road, with the exception of a dirt parking area 100 feet wide. The building had a high roof and a row of windows high up, but these were so dirty that the building appeared quite dark inside. The darkness was accentuated by dirty dark walls and an apparently dirt floor, which may have overlain a concrete floor at some depth. Milk delivered in cans was first poured into large (6 feet in diameter) steel (not stainless steel) drums. There was an open 100 kg. sack of caustic soda on the floor, which was evidently used to correct the pH of any milk that had got too acidic or sour. After this adulteration (if needed) of the milk, the liquid was pumped through plastic piping (again no stainless steel, no cleaning in place) to rectangular steel containers. These looked to be holdovers from the ice plant, since many were severely rusted and stained; some were rusted right through at the top. These containers, at 4 feet high, would have been extremely difficult to clean, and there was no evidence that this was even attempted. The milk was cooled in these rusted steel containers until a small tanker truck was available, at which stage it was pumped into the tanker for onward transport to a noncooperative processing plant. New mild steel containers were in the process of construction. Apart from the electric motor, pump and plastic piping, it was a Dickensian scene. Source: Field Visit. 4.16 Working Capital. 1993-94 was a trying year for Operation Flood. High procurement levels coincided with low world prices for dairy products. This led to depressed prices in India (as low as Rs 36,500 or US$1,160 a ton for SMP versus US$1,569 on the world market) and precluded export except at a substantial loss. Many MPUs ran out of working capital as inventories of dairy products (which could only be disposed of at depressed prices) locked up working capital. Many unions fell behind in making payments to farmers and had to declare "milk holidays" on which they would not receive milk (see also para. 4.24). NDDB provided substantial interest-free loans to help with this crisis. This facility was withdrawn when the crisis abated, and NDDB found that some MPUs were using the credit facility to avoid painful, but needed, decisions. Some of these loans are yet to be repaid. evidence, one cannot generalize as to the extent of unfair trade practices in the whole dairy industry and country." For readers unfamiliar with the Indian dairy market, the text has cited enough examples to alert them that quality and competition can be problems in India. Those familiar with the industry will be able to make their own judgments. 28. Gulati and Bhide (1997) report SMP quotations in 1993-94 ranging from Rs 36,500 per ton to an inferred price of Rs 71,400 per ton (Annex Table 5.1). This latter estimate involves assuming 920 grams per liter of milk versus the more usually accepted estimate of 1,020 grams per liter. Correcting for this error, the high price drops to Rs 64,600 per ton. In any case, the high price is an inferred price, from consumer milk prices, which may not have adjusted downwards fully to reflect the lower cost of raw materials. The average price used for estimation of protection coefficients in 1993-94 was Rs 56,000 per ton. Had the lowest price been used, the protection estimate would have been reduced by about a third. 42 Box 4.2: Artificial Milk Problems with the supply of artificial milk have been reported in both the corporate and cooperative sectors. An article in the New Delhi Statesman (July 8, 1995) reported on the seizure from a contractor of "artificial milk" and goes on to explain that "the ingredients used in the manufacture of synthetic milk are caustic soda, water, refined oil, common salt, sugar and urea." This artificial milk was destined for delivery to a large corporate processor.a Similarly, an article in the Asian Age (March 12, 1997) says: Delhites Panic Over Synthetic Milk .. . The large-scale supply of synthetic milk in Delhi and adjoining districts in Haryana and Uttar Pradesh has become the cause of serious concern for consumers in the region. Spurred to action by a deluge of complaints, the Food and Research Analysis Centre (FRAC), a voluntary organization, collected samples of milk from various outlets in and around New Delhi last year. Of the 33 milk samples collected by FRAC, almost 60 percent of the samples were found to stray from specified standards laid down by the Prevention of Food Adulteration Act. "The samples were not adulterated but merely sub-standard," says Mr. S.P. Virmani, founder- president of FRAC, admitting however that some samples .. . had been found to contain synthetic adulterants ... In January this year the Gurgaon administration busted a large milk adulteration racket that involved supplies from the Rajasthan Government Milk Cooperation Saras. According to the Gurgaon police, at least three quintals of milk were siphoned off from the tankers every day, and replaced with adulterated supplies. Saras (an MPU) tankers from Bhilwara, Ajmer, Jogpur, and other areas of Rajasthan would collect at a destination in Binaulla village in Gurgaon district, where the fraud was perpetrated. Local administration sources reveal that a large number of private as well as government were involved in the racket. a. In a letter to the editor (August 4, 1995), the Manager of Corporate Communications understandably took exception to the article, mentioning "25 different tests and analysis .... These checks rule out any possibility of adulteration of liquid milk going undetected." The letter stopped short of denying that the culprit was a supplier to the company or that the material had been seized. It seems unlikely that the contractor was caught by officials on his very first use of the recipe. 4.17 Cooperative Structure. A continuing problem for Operation Flood is the rejection by politicians and bureaucrats of the full Anand principles of farmer control. Even in Gujarat, where the existing system is farmer-controlled at all three levels, politicians continue to harass the cooperative system in the hope of forcing it to accept government assistance and hence political appointees on boards of directors. This is a continuing war of attrition, with the public sector giving ground reluctantly to increased control by the NDDB and farmers when the cost of supporting governmental-dominated cooperatives becomes too high, only to attempt to regain control once financial equilibrium is restored. 4.18 The issues of government interventions, cost control, and subsidies are confusingly intertwined. The question then arises as to when a payment by government to cover excess operating costs by the cooperative system is a subsidy. A clear distinction needs to be made between a government subsidy to an Anand-pattern cooperative in which all tiers are farmer- owned and -controlled and payments to cooperative with "mixed ownership and control." Typically, mixed-control cooperatives have higher costs or lower revenues owing to government interventions, including the inability to dispense with redundant staff or inappropriate appointments, inadequate salaries to retain key staff, and prices that are fixed too high for 43 farmers or too low for consumers. There is a more subtle and corrosive effect of (state) government intervention. If management knows that at the end of the day the government will bail out the organization, the incentive to efficient management is fundamentally undermined. (This is the key reason that insistence on the independence and the Anand principles is so important.) Nothing focuses the mind of a manager or a board director like the prospect of the organization going bankrupt. A clear conceptual distinction needs to be made between payments needed to compensate for governmental managerial interventions and a genuine subsidy to an Anand-pattern cooperative that has full discretion to manage its affairs. Operation Flood officials were adamant that no subsidies, in this narrower sense, had been received by (fully) Anand- pattern cooperatives. There may have been a subsidy element in the first two phases of Operation Flood, but in OFIII (and NDPII) all funding was provided by loans from the Bank, profits on concessional dairy imports, and the NDDB. There was no contribution from the central government and hence no subsidy from the central government directly to Operation Flood. Both the central and state governments provide support to the dairy industry generally (including Operation Flood areas), such as veterinary and extension services and Al. 4.19 Much of the reluctance of state governments to relinquish their influence over cooperatives is because some of the physical plants used still belong to the state and loans to the dairy sector are guaranteed by the state (as required by the Bank). State intervention is then rationalized as a legitimate concern to assure the value of the state's assets, even when the net result of an intervention is to reduce system efficiency, raise cooperative costs, and hence undermine the value of the state's assets. 4.20 The continued appointment of civil servants as managing directors of milk federations and MPUs is one symptom of the above conflict.30 Only in Gujarat, Karnataka, and Rajasthan has it been possible to appoint dairying professionals to managing directorships. 4.21 Even in Karnataka and Rajasthan, the professional managers are not direct employees of the MPU but are seconded to the MPU from a cadre of professionals employed by the milk federation or NDDB. The result is that appointments tend to be for a few years without the expectation of having to make a career in the service of the MPU. In Bihar the Anand principles have been accepted to the point that professional staff are employed directly by the MPUs. 4.22 Cooperatives have to operate in accordance with the cooperatives act of the relevant state. In almost all cases, these acts are antiquated holdovers from pre-Independence days with a markedly paternalistic bent. The result is a bias that allows the registrars of cooperatives much more leeway in cooperative affairs than is open to the registrars of companies. In addition, the 29. Doornbos and others (1990) identified both input and investment subsidization and expressed the fear that NDDB would be unsustainable if these subsidies ceased. A close reading of this data shows that the input subsidy (farmers being paid more for their milk than was recovered from consumer) is better described as a consumer subsidy and that this was financed by profits from concessional imports of dairy products, not from the central government. In the event, the dairy cooperative system as a whole is currently profitable, though individual MPUs and even federations are still unprofitable where state governments control milk prices or limit the ability of cooperatives to correct overstaffing. 30. It is incorrect to refer to a government payment to an organization with a government-appointed managing director as a subsidy, just as the concept of a cooperative with a government-appointed managing director is a seeming contradiction. 44 acts usually require cooperatives to have their accounts audited by the auditor general, a requirement that may well involve a delay of several years.3' 4.23 A fundamental problem for cooperatives is the conflict between efficiency and equity. Mention has already been made of the greater efficiency of milk collection from large farmers. The same issue applies to the collection of milk from small DCSs and the servicing of remote villages.32 Economic efficiency would dictate either not collecting from these more expensive sites or at the very least paying a lower price to reflect the higher cost of providing service. Typically, cooperatives prefer to provide service equally to all suppliers at the same cost on grounds of equity (and to some extent simplicity), but are always subject to receiving a sufficient volume of milk to cover variable costs. Cooperatives typically provide for some cross- subsidization from the better off to the poor.3 Given the Bank's overarching objective of reducing poverty, careful thought needs to be given before the Bank advocates an increase in efficiency if efficiency is to be obtained by paying lower prices to the poor or withdrawing service altogether. In the final analysis, MPUs will decide how much cross-subsidization they wish to provide. Their decision will, of course, be constrained by the need to operate profitably. Where the volume of milk and existing route pattern do not allow variable costs to be covered, MPUs withdraw service to DCSs as is witnessed by the number of DCSs which have become nonfunctional. 4.24 As argued earlier (paras. 1.20-1.21), cooperative and private trader are likely to behave differently in a monoposonistic relationship with milk producers. A similar difference in behavior can be expected between cooperative and private companies at the corporate level. Private companies, in attempting to assure their profitability and survival, are unlikely to distribute profits to the point that working capital is seriously eroded. Cooperatives may be tempted to return too much to producers to the point that too little money is retained as working capital. Political considerations may lead boards of directors to authorize overly generous payments to producers. In India the asymmetry between the tax treatment of farmers, who do not have to pay income tax, and companies (including cooperatives) that have to pay income tax at the 35 percent rate on profits is an additional powerful argument for maximum producer payout to reduce tax liability. 31. Cooperatives can get their accounts audited privately, but this does not exempt them from the requirement to also have their accounts audited by an auditor general. This option results in a double dose of audit fees. Understandably, it is seldom used. Again, Bihar provides an exception. It has recently been permitted to substitute private auditors for the services of its Auditor General. 32. It is interesting to see how this is described in the literature critical of Operation Flood: OF authorities have shown a preference for establishing cooperatives in regions which were far away from the main metropolitan centers thereby avoiding tough price competition with the many small traders near cities ... . The infrastructural investment for organizing this type of procurement is, however, sizable, especially when milk has to be marketed in metropolitan cities. (Doombos and others 1990, p. 117). The reader is left with an impression of a lack of competitiveness of the cooperative system and excessive investment, as if it would somehow have been better to leave remote villages unserved in order to compete successfully with the small peri-urban traders who are already providing an economic and low-cost service. 33. The extent of the subsidy, if any, is open to debate. It is standard practice worldwide for cooperatives and private milk processors to pay all suppliers to the plant the same amount, regardless of transport costs. In part, this is a matter of simplicity and goodwill; even more importantly, though the marginal transport cost of additional milk collected from remote villages exceeds average transport costs, the marginal processing cost for the additional milk falls below average processing cost, leaving net impact on profitability moot. 45 4.25 The above-mentioned temptation for cooperatives to pay too much to their producers is reflected in somewhat slow repayment of project loans by MPUs as shown in Table 4.4. Ironically, two of the three states (out of 20) that were helped by specific Bank projects have double the rate of arrears compared to OFI and OFII. The most significant arrears are from Rajasthan, where the state government has asked the NDDB to take over management of the cooperative sector until financial profitability can be restored. The NDDB is confident of resolving the financial problems in Rajasthan within five years. Table 4.4: Loan Repayments on Bank Projects to NDDB (Rs Milion)a Outstanding Project Lent Dueb Received Amount Percent ofDue Karnataka 244.4 321.6 312.6 9.0 3 Madhya Pradesh 142.4 257.5 142.5 114.9 44 Rajasthan 169.2 254.6 85.1 169.5 67 National Dairy I and II 8,054.1 2,781.2 2,098.3 682.9 25 Total 8,610.1 3,614.9 2,638.6 976.3 Percent 100 73 27 a. To March 31, 1996. b. Includes interest. c. Loan repayment data relate to IDA funds in the case of Karnataka, Rajasthan, Madhya Pradesh and relate to World Bank, EC and NDDB/IDC funds in the case of National Dairy I and II. Source: NDDB. 4.26 Some cooperatives have developed the device of part payment in equity. In this case, a part of the final (bonus) payment for milk is in the form of equity in the DCS, which in turn is reflected in DCS equity in the MPU.34 The result is that the MPU limits its tax payment, and the cooperative members increase their equity in the MPU, thereby increasing working capital or allowing the governments loans or equity to be reduced. 4.27 Financial Sustainability. Financial sustainability is an extremely complex topic in the context of Operation Flood. Just as the Bank insists that the central government guarantee funds made available to the NDDB, the NDDB requires that state governments guarantee loans to the federations and MPUs. Thus, nominally at least, the NDDB's financial security depends on the creditworthiness of the state governments rather than on individual projects. This reliance on the states' creditworthiness is particularly important in states that continue to interfere with the pricing and staffing decisions of federations and MPUs (Table 6.11). The Bank-supported projects aimed to free the cooperative sector from state interference. Had they succeeded, direct lending without a state guarantee would have made sense. But the project was unable to eliminate state interference altogether. Many states take the position that as long as they are guaranteeing loans to federations and MPUs and as long as they own part of the assets, they have the right to appoint directors and managers and to have a say in pricing and staffing decisions. 34. Strictly, the final payment has to be described as an adjustment to the price of milk paid during the year, since a bonus would need to be paid out of profits which would first have to be declared and taxes paid. 46 4.28 The profitability of federations and MPUs improved markedly in the three years before the NDPII closed. As Table 4.535 shows, loss-making MPUs and state federations were reduced from 73 to 46 (13 of them in Karnataka and Andhra Pradesh) over two years, and the increase in profits and reduction in losses has been spectacular: overall losses of Rs 584 million in 1992-93 have been converted into profits of Rs 285 million in 1994-95. Even so, accumulated losses over this period were Rs 627 million or about US$18 million. Table 4.5: Profitability of the Cooperative Sector (132 state federations and MPUs) 1992-93 1993-94 1994-95 No. making Profit or No. making Profit or No. making Profit or Net Net loss Net Net loss Net Net loss State profit loss (million Rs)" profit loss (million Rs)a profit loss (million Rs)a Gujarat 15 - 57 13 2 28 12 3 36 Uttar Pradesh 9 14 (33) 11 12 (49) 12 11 (26) Karnataka 2 11 (131) 3 10 (125) 10 3 115 Tamil Nadu 5 6 (171) 6 4 53 6 5 (3) Maharashtra 9 2 (7) 10 1 28 10 1 112 Andhra Pradesh 2 9 (168) 2 9 (121) 6 4 (58) Other States 17 31 (131) 26 22 (142) 29 19 119 All Indiab 59 73 (584) 71c 60 (328) 85c 46 285 (= a loss. -= not available. a. After tax. b. Includes unaudited accounts. Losses were Rs 581 and 254 million in FY93 and FY94, respectively, converting to an after-tax profit of Rs 285 million in FY95. c. Totals to 131. Source: ICR and mission data. 4.29 While payments to the NDDB are still often in arrears, the rate of increase in these arrears has slowed substantially. The NDDB is taking steps to correct the problem, which is not currently a threat to the survival of the dairy cooperative system. Total overdue loan installments increased 27.5 percent (from Rs 796 million to 1,020 million) from March 1995 to March 1996 and represented about one-third of repayments due at the end of March 1996. 36 From 1990-91 to 1994-95, these arrears grew at an annual rate of 180 percent from a very modest base of Rs 82 million. 4.30 To put the numbers in perspective, the NDDB's surplus (excess of income over expenditure) was Rs 466 million in FY95. Its total assets were Rs 27.8 billion, and its equity was Rs 14.1 billion. Thus, total arrears were equal to around three years' surplus, and total loans outstanding (Rs 10.7 billion)37 could in extreme circumstances be written off and still leave the NDDB with about 30 percent of its equity. 35. India has 170 Milk Producer's Unions and 20 federations. For those not covered in Table 4.5 (58 organizations), either audited accounts were not available (some states require a state or cooperative auditor to do the auditing, often resulting in delays of up to several years), or there was an amalgamation of units. 36. The Implementation Completion Report (ICR) for NDPII, para. 50, puts arrears at 1 percent of NDDB's loan portfolio, but this is presumably a misprint. Total overdue loans (Rs 1,020 million) exceed total overdue on Bank projects (Rs 976.3 million) since some loans were advanced using NDDB's own funds. 37. Rs 7.3 billion committed to subborrowers and 2.8 billion of accrued interest as of March 1995, plus 564 million of capital and interest outstanding under the Working Capital facility in March 1996. 47 4.31 The above figures are subject to various interpretations. Some will fix on the first derivative and emphasize that Rs 627 million accumulated losses over three years, arrears are continuing to increase, and 46 MPUs and federations continued to make losses in 1994-95. Others will look at the second derivative and note that the rate of increase of arrears has slowed, the number of MPUs and federations making losses had been reduced, and, for the first time in three years, 1994-95 saw MPUs and federations, as a whole, make a profit. This looks more like a "turn-around" situation than the threat of a pending financial crisis; however, others may read the data less optimistically, and for the 46 loss-making MPUs, it is to be expected that loan recovery will be difficult. 4.32 International Competitiveness of the Indian Dairy Industry. Empirical research undertaken as part of the impact study shows that the protection of the Indian dairy industry has declined remarkably over the last 20 years (Table 4.6). The initially high protection has now been significantly reduced. Indeed, for skim milk powder (SMP) in 1994-95 and 1995-96, protection was negative, as the effective protection coefficients (EPCs) are less than 1.00. In large measure, this reflects a recovery of world prices, though the stable environment (and Table 4.6: Revised Effective Protection of the Indian Dairy Industry (at Shadow Exchange Rates) Real World Prices (FOB Netherlands) US$/Tonne Net Protection Effective Protection Ratios Coefficients Coefficients Butter Real SMP/ Year SMP Ghee Milk SMP Ghee Mdk SMP Oil/Ghee Real Butter 1975-76 2.03 1.67 2.42 2.37 1.89 3.27 1834.63 3617.57 0.507 1976-77 4.27 2.20 4.16 7.27 2.67 9.29 1035.29 3117.65 0.332 1977-78 5.16 1.69 3.13 10.96 1.93 4.76 736.20 3144.17 0.234 1978-79 3.92 1.47 2.30 6.22 1.61 2.93 830.32 3294.22 0.252 1979-80 3.15 1.41 2.46 4.35 1.54 3.20 906.10 300659 0.301 1980-81 1.50 1.14 1.73 1.64 1.21 1.97 1741 80 3493.85 0.499 1981-82 1.68 1.13 1.49 1.87 1.20 1.65 1764.58 4425.00 0.399 1982-83 1.93 1.25 1.83 2.21 1.34 2.12 1588.59 4064.11 0.391 1983-84 2.02 1.76 2.47 2.34 2.03 3.14 1326.22 3275.26 0.405 1984-85 2.83 2.19 2.87 3.70 2.69 3.95 1165.80 2722.37 0.428 1985-86 2.20 4.42 5.81 2.62 5.42 7.80 1103.05 1903.37 0.580 1986-87 2.05 2.75 2.98 2.40 3.67 4.05 1021.67 1600.00 0.639 1987-88 1.67 3.27 2.95 1.87 4.79 4.01 1321.52 1539.18 0.859 1988-89 1.12 1.97 1.70 1.17 2.32 1.93 2263.77 2158.95 1.049 1989-90 1.15 1.62 1.47 1.21 1.82 1.61 2154.92 2477.05 0.870 1990-91 1.11 2.05 1.86 1.15 2.45 2.19 1693.22 2022.02 0.837 1991-92 1.12 1.92 1.65 1.17 2.25 1.89 1694.44 2027.78 0.836 1992-93 1.02 1.71 1.33 1.05 1.94 1.44 2068.00 1995.54 1.036 1993-94 1.13 2.12 1.55 1.19 2.58 1.72 1747.31 1825.27 0.957 1994-95 0.86 1.92 1.28 0.88 2.24 1.38 1844.78 1972.74 0.935 1995-96 0.93 1.52 1.01 0.95 1.69 1.04 2195.00 2525.00 0.869 Notes: 1. This table provides a summary of information from the report sponsored by OED (Gulati and Bhide 1997). 2. Annex 5.1 of Gulati and Bhide (1997) uses a conversion for milk of 910 grams per liter rather than 1,020 per liter. Data has been corrected for this error by multiplying by 0.9. 3. Net Protection Coefficients (NPC) for SMP for each year are calculated from Tables 5.1 and 5.5 taking the SER values for that year for the different regions under the Importable and Exportable Hypothesis and then averaging them. The NPC for ghee and milk are similarly calculated from Tables 5.3, 5.7 and 5.4, 5.8, respectively. 4. The Effective Protective Coefficients (EPC) for SMP are similarly calculated from Tables 5.9 and 5.13. For ghee and milk, similar calculations are made on the data from Tables 5.11, 5.15 and 5.12, 5.16, respectively. 5. Real World Prices for SMP are calculated by deflating the World Prices from Table A 5.3 by the MUV series for 1996. Real World Prices for ghee are similarly calculated from Table A 5.7. 48 assistance with investment in processing plants) the Indian industry has enjoyed undoubtedly played a role. Table 4.7: Indian and World Market Prices for Selected Dairy Products, 1991-95 (US$/metric ton) Item 1991 1992 1993 1994 1995 1996 Milk: Producer Price India: Cows (4% fat) 154 n.a. n.a. 153 191 India: Buffalo (7% fat) 242 n.a. n.a. 223 255 Germany 360 393 n.a. n a. n.a United Kingdom 329 344 n a. n.a n.a. United States 270 289 n.a. 285 275 Milk: Wholesale Price India 336 360 n.a. n.a. n.a. United Kingdom 438 435 n.a. n.a. n.a. Butter: Wholesale Price India 1,832 3,788 n.a. n.a. 2,868 World 1,413 1,363 1,350 1,294 2,215 1,546 Whole Milk Powder: Wholesale Price India 3,089 n.a. n.a. n.a. 2,186 World 1,388 1,638 1,488 1,520 2,075 Skim Milk Powder: Wholesale Price India 1,620 1,907 1,782 1,712 2,045 World 1,425 1,681 1,569 1,563 2,075 1,795b a. Converted to dollars from Gulati and Bhide (1997), Annex Table 5. lAB, using market exchange rate. b. USDA, 1996-97 (North Europe Low). Source: World Bank Reports 1996, Table 5.5. 4.33 It is worth pausing for a moment to consider the data in Table 4.6 in more detail, especially since the results for the final year 1995-96 are atypical. Looking just at milk, had the time series stopped in 1994-95, it would have been appropriate to conclude that the Indian dairy industry has been very successful in reducing initially very high levels of protection, but protection was still about 30 percent. The last row of Table 4.6 is somewhat misleading since it is a peak in world prices of SMP and butter which explains the low protection in 1995-96. The real world price for SMP was the second highest on record (the highest price occurred in 1988- 89), and butter was at its highest real price since 1989-90. As shown in Table 4.7, the latest estimate of world prices for SMP and butter suggest a 10 percent fall for SMP and perhaps 30 percent for butter, implying roughly an 18 percent level of protection for India on the basis of the latest world prices. 4.34 The protection figures in Table 4.6 refer to processed milk products. Since both milk itself and collection, transport, and processing are "non-tradables," it is not possible to infer how protection is allocated among these non-tradables. What is clear is that the Indian system of getting milk from the farmer's pail to the point where it has been cooled and can be handled in bulk is both organizationally sophisticated and technically inefficient. Indian producers get about 63 percent of the total cost of wholesale urban milk (Gulati and Bhide 1997, Table A3. 10) for supplying warm milk in small quantities. The share of wholesale cost going to chilling centers which supply bulk chilled milk is 81 percent (this includes the 63 percent to producers as one of its costs). Processing margins, starting with chilled bulk milk at rural centers is thus 19 percent. It is this figure which needs to be compared with developed-country processing margins, since they pick up bulk chilled milk from the farm. 49 4.35 A byproduct impact of Operation Flood and accompanying dairy expansion has been the establishment of an indigenous dairy equipment manufacturing industry (only 7 percent of dairy equipment is now imported) and an impressive body of indigenous expertise that includes animal nutrition, animal health, Al, management information systems, dairy engineering, food technology, and the like. The Bank's insistence on International Competitive Bidding means that, as far as Operation Flood goes, India's dairy manufacturing industry has been established in competition with international suppliers. However, supplies of Indian equipment to non-Flood dairy processors may have benefited from the artificial stimulus of "self-reliance" which ruled in many industries prior to 1991 and raised the cost of those processors. This indigenous infrastructure explains in large part why milk processing and marketing costs of Operation Flood have not exploded in the face of having to procure and account for minute quantities of milk.  51 5. Village-Level Impact 5.1 Two village-level participatory studies were undertaken as a part of the evaluation. The first of these was in nine villages in the Mysore district of Karnataka state (Sampark 1997) and the second in 12 villages in three milksheds in Madhya Pradesh (Hiremath, Singh, and Mergos 1997). The villages in this latter study had been surveyed in 1983, and the in-depth village studies were supported by a survey of secondary data on two hundred villages in the same milksheds. 5.2 Both studies covered villages with and without a DCS. In the Madhya Pradesh sample, there were villages that had had a DCS in 1983 and still had one in 1996, villages in which the DCS had ceased to operate, a village where a DCS had been open since 1983, and a village that had never had a DCS. Both studies administered questionnaires to individual families and used a participatory approach. In Madhya Pradesh, this took the form of focus group discussions; in Karnataka, social mapping and group validation of survey findings were used. 5.3 Despite the similarity of approach, quite different pictures emerged from the two studies. The Mysore MPU in Karnataka appears to have substantial operational independence.38 Certainly it is free to set producer prices and, with the support of an integrated business plan, it lowers producer prices in the flush and raises them in the lean period. The MPU appears to have satisfactory relations with the supplying DCSs and has not been substantially adversely affected by the entry of private companies under the liberalization policy. The three MPUs studied in Madhya Pradesh appear to suffer from greater governmental interference that prevents them from raising producer prices to be fully competitive with small traders and private companies. This, and the withdrawal of MPU-subsidized animal health services, has led to a decline in relations between the MPUs and their supplying DCSs. In addition, the introduction of soybeans in Madhya Pradesh has revolutionized agriculture with large declines in the area under cotton and groundnuts. Since soybeans provide less feed residue than cotton and groundnuts, this has adversely affected dairying. In addition, the higher incomes from soybeans have boosted prosperity to the point that wage labor provides incomes competitive with those from dairying. The Mysore district has not enjoyed a similar technological change. Karnataka 5.4 The nine villages included in the Karnataka survey comprised five with operating DCSs (the DCS villages), two where a DCS had closed (the "defunct Flood" villages) and two that had 38. A reviewer has commented that this may be indicative of the Mysore MPU being "one of the better ones," and while the results will be representative for Mysore Union they may be unrepresentative for India as a whole. This comment applies to all village studies, since a sample of even 40 villages in one state or milkshed, however carefully selected and however representative of the area being sampled, is bound to be unrepresentative of the 50,000 plus villages with DCSs. However, an evaluation interested in impact is compelled to look for evidence at the village level. 39. The village-level focus of these studies meant that the operations of the MPUs were not studied in detail. Comments thus primarily reflect the "view from the village" and may appear differently at the regional level. 52 never had a DCS (the "never Flood" villages). In addition to the participatory activities, the survey interviewed a stratified random sample of 451 people. 5.5 Depending on how benefits are measured, quite different conclusions can emerge as to who benefits most from the cooperative system. Thus DCS membership substantially exceeds the number of people "pouring" to the DCS. In an effort to open up the DCS to women, membership rules have been revised in several villages to allow two members per family. In addition, some families that have ceased to own cattle have retained their membership. Membership may thus not be a good measure of "beneficiaries" as usually understood. Small and marginal farmers dominate the people pouring (74 percent of pourers contributing 74 percent of milk to DCSs).41 Landless42 contributed another 15 percent of pourers (pours 11 percent of milk). Large farmers were 11 percent of pourers (pours 15 percent of milk). 5.6 Important though dairying is to the landless, not all of them can participate because of the cost of cattle and the need to purchase feed. Though the direction of causation may be in doubt (are people wealthier because they own milch animals, or can only the wealthier afford to own milch animals?) dairying families earn more money in DCS villages than dairying people in defunct- and never-Flood villages. In DCS villages dairy families have higher incomes than nondairy families, and in all villages the landless with dairy cattle are better off than those without. 5.7 There was a general preference for selling to the DCS rather than to a private vendor. Vendor prices tend to be higher than DCS prices close to an urban center and lower (if a vendor is even active) in more remote villages. Women who have no access and control over the family's total income do have some control in the case of milk money. Milk money is small when it comes in daily or every week, and the expenditure can be on small items. Leaving a small amount of money to their wives is not seen as a problem by some men. 5.8 The closure of the DCSs in the two defunct-Flood villages followed embezzlement by the secretaries, irregular committee meetings, and interference from state-level authorities. This resulted in poor cash returns to producers, falling milk procurement, and eventually a decision by the MPU that the DCSs were no longer viable (and hence the MPU would cease to pick up milk from the villages). Closure affected the landless and scheduled castes and tribes most adversely. Without the DCSs, there was no alternative market, and cattle often had to be sold. The villagers were not in a position to reactivate the DCS. 5.9 If the impact is to be evaluated from the viewpoint of equity, the study (Sampark 1997) concludes that there is definitely a positive impact on those who are at the lower end of the economic ladder, both in terms of landholding and caste to which they belong. However, those who do not have access to resources to own cattle are left out of the fold of dairy cooperative and have not been able to access this opportunity for augmenting incomes. 40. Stratified by landless and farm size and to include dairying and non-dairying families and, in DCS villages, dairy producers who poured and did not pour to the DCS. 41. Sampark 1997, Tables VI.B.1 and VI.B.2. 42. The landless in non-DCS villages had no dairy income (and presumably no cows). Actual consumption in DCS villages was 0.8 1/day for landless, 0.9 I/day for marginal farmers, 1.3 I/day for small farmers, and 1.59 I/day for large farmers. 53 Madhya Pradesh 5.10 As noted above, the 12 villages surveyed were initially surveyed in 1983 in connection with the completion of the NDP (Mergos and Slade 1987). In 1983 nine of the villages had DCSs, and in four of these (the defunct-Flood villages) the DCSs had closed. Three villages had no DCSs in 1983, although one had formed a DCS in the meantime. Thus, in 1996 there were five DCS villages, four defunct-Flood villages, and two never-Flood villages. Although the data was available from the earlier study, the names of the respondents could not be recovered. Accordingly, a new stratified random sample of 30 families from each village (360 interviews) and a supplementary survey of village-level data was conducted in 200 villages to check the generality of the village-level observations in the 12 resurveyed villages. 5.11 In general, Madhya Pradesh emerged as having a much more fluid dairying situation than Karnataka. Price controls prevented the cooperative sector from raising prices to compete fully with private companies, and the rapid expansion of the area under soybeans reflects a more profitable farming system that provides fewer crop residues for ruminants. This, together with a fairly widespread rash of dishonesty amongst DCS secretaries, had led to the collapse of a significant number of DCSs. For the MPU as a whole, the proportion of functional DCSs has dropped from 77 percent of those organized in 1983 to 55 percent in 1996. Hiremath, Singh, and Mergos (1997, p. 91) note, "Our survey of 200 DCS villages indicated that DCSs were functioning in 184 villages and in the remaining 16 villages the DCSs had become dysfunctional. The reasons for DCS closure include declining milk production, poor leadership in the village, bad choice of secretary and complaint against the DCS's operations. Nearly 31 percent of DCS's closure was attributed to competition from the private sector as a result of the new economic policies. Some 38 percent of the DCS closure was mainly the results of declining milk production in villages that may be due to the impact of changing cropping patterns. Among the remaining dysfunctional DCSs, 25 percent had closed due to both private competition and internal reasons and 6 percent due to internal reasons alone."43 5.12 The high proportion of internal reasons for DCS collapse is disturbing. A key assumption of the Anand system is that villages can manage their own affairs themselves. In part, the recent Bank-sponsored withdrawal from the provision of preventative animal health care seems to have resulted in less frequent contact between villages and representatives of the MPU with a growing feeling of estrangement on the part of individual producers and less opportunity for MPU staff to monitor emerging difficulties in the operation of DCSs. This is probably the most serious concern as to the sustainability of some MPUs. 5.13 Part of the loss of suppliers to small-scale dudhiyas around urban centers is to be welcomed. With short distances from farm to consumer, the dudhiyas can be economically efficient. In practice, part of their competitive edge may be due to the routine adulteration of milk by farmers or in transit. Dudhiyas also provide credit and, in doing so, provide an additional service not available from DCSs. Competition with private companies represents a different issue. To the extent that companies can operate more efficiently because they are not subject to price fixing and overstaffing mandated by the state, the solution is in the hands of the state and 43. Poor leadership in the village and bad choice of secretary. 44. NDDB supports a Cooperative Development Program to improve farmers' understanding of the cooperative system and their rights and responsibilities (para. 4.6). This recognizes that tensions within the village can undermine the operations of the DCS. See also Attwood 1985, p. 37. 54 the cooperative sector.45 If excessive debt in the cooperative sector is the problem, then some financial restructuring may be required. To the extent that private companies are able to pay more to convenient villages with low transport costs, the cooperatives may eventually be forced to reconsider their panterritorial pricing policy. 5.14 The loss of DCSs due to declining milk production is to be welcomed if it represents the realization of new opportunities for farmers to use their resources more profitably or rising incomes that will ultimately lead to higher home consumption.46 In the Bhopal milkshed, it appears that much of the decline in milk availability is the result of the introduction of soybeans. 5.15 As for Karnataka, the surveyed relationships between wealth and dairying need to be interpreted cautiously. A striking change for all villages from 1983 to 1996 has been the increase in the years of education for the head of household which had doubled and in some cases almost tripled. Within this general trend, the households without milch animals (with or without a DCS) have increased their educational levels faster than the dairying households. This may reflect the reliance of nondairy households on nonfarm jobs that require education at higher levels. 5.16 In 1983, incomes were higher in non-DCS villages than in DCS villages, but by 1996 this relationship had been reversed. In the period between the two surveys, real income rose at 6.5 percent per annum in the DCS villages but at only 3.5 percent in the non-DCS villages. Despite this, the defunct-Flood villages were reported to be the most developed in terms of educational and income levels, agricultural land owned and irrigated, numbers of milch animals owned, and milk sold. This suggests that these villages were close to town and thus able to capitalize on their locational advantage to get high milk prices from dudhiyas even in the absence of a DCS. 5.17 In those villages with a DCS in 1996, the DCS's performance was rated highly with respect to milk price, provision of a reliable market for milk, and regularity of payment. Respondents were much less satisfied with the provision of veterinary services and fodder. In large measure, this more critical view of veterinary services was the correct reflection of a new "unbundling" policy introduced at the Bank's urging by OFIII. Under OFI and OFII, both regular health camps and emergency veterinary services had been provided free or at highly subsidized rates by the MPU. In part, this represented a view that "prevention is better than cure" and that farmers were better off paying a compulsory insurance premium through a slightly lower milk check than remaining uninsured. The Bank argued that this undermined the market for the private supply of veterinary services and by implication that the risk element could be ignored. It was agreed that the health camps would be discontinued and even emergency service moved toward full cost recovery. The results of this policy can be seen in Table 5.1 and Table 5.2. The need for veterinar services increased substantially, and the proportion of cases where the need was satisfied fell. This was particularly marked for members of a DCS, but also applied to 45. To the extent that corporate competition is based upon government equity and tax holidays not available to the cooperative sector, the solution is the withdrawal of this assistance to private companies (Box 6.1). See also para. 6.31. 46. If reduced supply is due to price controls holding prices at levels below what the market could bear, it is not to be welcomed but is evidence of the need to remove state interference from pricing decisions. 47. A reviewer reports: "Vets in Anand interviewed during a visit said that increased cost recovery reduced emergency calls from farmers for nonemergency matters. Since the visits were practically free, farmers tended to seek more help than necessary since the cost of vet visits was undervalued." The reviewer also criticized the survey reported in Table 5.1 as not adequately answering the question of whether farmers are better or worse off. Perhaps a question that should have been asked was: "Do you prefer the present system where you pay for veterinary services directly or the old 55 nonmembers and non-DCS villages. There was a slight rise in the proportion of cases where needs were met in non-DCS villagcs. Table 5.1: Survey Results: Need for Veterinary Services (Percent) DCS Village Members Nonmembers All Non-DCS Village Change Change Change Change 1983 1996 (%) 1983 1996 (%) 1983 1996 (%) 1983 1996 (%) Need for veterinary service 15 55.36 269.1 27 30.88 14.4 20 46.11 130.6 13 44.44 241 8 Need was satisfied 96 43.55 (54.6) 88 71.43 (18.8) 93 50.60 (45.6) 55 56.25 2.3 Source Hiremath, Singh, and Mergos 1997, Table 6.5.1. 5.18 As shown in Table 5.2, there was a switch by DCS members from DCS-supplied services to government and private sources, with a fall in the ability of nonmembers and non- DCS villages to access these alternative services. Table 5.2: Survey Results: Sources of Veterinary Services DCS Village Members Nonmembers All Non-DCS Village Change Change Change Change 1983 1996 (%) 1983 1996 (%) 1983 1996 (%) 1983 1996 (%) DCS 64 11.30 (82.3) 45 0.00 100.0 60 8.43 (85.9) Private veterinary doctors 7 19.35 176.4 10 4.41 (55.9) 6 18.07 201.2 25 16.25 (35.0) Government veterinary hospital 29 69.35 139.1 45 26.47 (41.2) 34 73.50 116.7 75 73.75 (1.7) Indigenous treatment - 0.00 - - 0.00 - - 0.00 - - 4.44 - -Not applicable. .. Not applicable. Source: Hiremath, Singh, and Mergos 1997, Table 6.5.1. 5.19 An unappreciated benefit of the MPU's provision of veterinary services was the resulting increased contact between MPU staff and the village. The loss of this contact probably explains some of the deterioration in the performance of DCS secretaries and lethargic management committees. 5.20 Since farmers attitudes may well differ from state to state or even MPU to MPU, the decision on bundling or unbundling services is probably best left to the management of individual MPUs. system where these services were 'subsidized,' but at the cost of receiving less money for every liter of milk?" Unfortunately, we cannot go back and rerun the survey. In fact, both questions are needed. The indications from the questions in Table 5.1 are that farmers would probably be willing to pay something to revert to the old system. The reviewer is right to question whether they would be willing to have their milk money docked enough to pay the full cost.  57 6. Project Impact 6.1 Caveat #1. As argued earlier, "project," and hence project impact, is here interpreted as relating to the fundamental policy change support which both largely terminated direct new public sector support for dairy development and launched Operation Flood. Indirect assistance to the dairy industry as a whole through animal health, Al, research, extension, and miscellaneous credit programs continued unabated. 6.2 Caveat #2. The impact of the Bank's assistance is evaluated relative to the preexisting low-growth policies with direct government involvement in milk procurement, pricing, and distribution. The evaluation takes India's highly protectionist self-reliant policy as given. This imposed high costs on consumers as indicated by the high Net Protection Coefficient of about 3 in 1975 (Table 4.6). In the event, this "second best" policy has brought India close to a "first best" result. Currently the industry is subject to only modest effective protection, say about 18 percent, in large part due to rising world prices and a real devaluation of the rupee. Milk Production 6.3 Milk production and Operation Flood milk procurement and sales are shown in Table 6.1 and Figures 6.148 and 6.2. These show a change in national milk production from a growth rate of 0.7 percent per annum until 196949 and 4.7 percent per annum cumulative since. Though this change occurred prior to the approval of the Bank's first project (March 1974), it can best be attributed to the policy changes accompanying the launching of Operation Flood in 1970. This change in government dairy policy abandoned government schemes that underpaid producers and tried to compensate by providing improved technology and credit and replaced it with a dairy cooperative scheme that was allowed to capture the rents generated by sale of milk powder and butter-oil supplied as food aid. 6.4 Had the first dairy credit been approved in 1994, rather than 1974, it might have been recognized as a new-style sector adjustment credit, albeit in support of policies already in process of implementation. The projects being evaluated represent a series of five investment credits predicated upon a major policy change made, "owned," and persisted with by government. 48. Table 6.1 corresponds almost exactly with Figure 2.1 "Measurement of Project Impact," in Mergos and Slade (1987). This figure is used by the authors as the basis for a methodological discussion of the measurement of project impact. As they say: "We assume that there is an underlying upward trend.. . . We are concerned with the additional change, attributable to the project, over and above the underlying secular trend" (p. 23). 49. Linear regression suggests the change occurred almost simultaneously with project approval. The exponential model suggests 1971, five years after the decision to rely on the cooperative sector for dairy development. Period Annual Product Increase Million Metric Tons 1951-1969 0.14 1971-1994 1.87 70 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - y = 1.8688x - 20.284 m 60 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -F -R2 0.98W2-f - .2 40 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - . - - - - . - . - - - - E 0 30.-.--.--.---y.1366x+ 18.076-. -..................................................... a. 120 - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -- 3 0. years Figure 6.1 lndia: Milk Prodluction, 1951-1994 (linear fit to 1951-69, 1971-94 subsamples) Period Annual Growth Rate 70 --------- -- ---- --- - --- ------- ---------- - - -- --- -- --- --- 1951-1969 0.7% 1971-1994 4.7% y R=099356 60 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -2 - - -9 c 50 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -- - 0 .2 50 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - .2 c 0 30 O.5897 20 - - - - - - - - - - - - - - - - - - - - - - - - - - - - 10 -- - - - - - - - - - --- --- - - - - --- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 0 i i i i i i i b i i i [i i i i i i a i I i i | i U) t- r r ) v) tn rz ; years Figure 6.2 India: MiIk Production 1951-1994 (exponentialfit to 1951-69, 1971-94 subsamples) 60 6.5 Given the extensive time lag from policy decision to project approval, one might well ask how Bank projects could be credited with any of the benefits from the policy change. The projects supported the policy change long before they were approved by the Board. This support should be dated from the first time the Bank indicated its willingness to consider appraising the first project (Box 1.1). Not only was this indication of Bank support helpful in endorsing the wisdom of the government's policy change, but it helped validate the EC's and WFP's decision to provide generous amounts of dairy products to India to support Operation Flood. As the credits took shape, they became a potent argument against policy reversal in the event that the bureaucratic beneficiaries of the old policy were ever in a danger of mounting a counterrevolution. Moreover, the Bank projects allowed the government to support the major investments needed for the new policy without the government having to bear the immediate pain of finding all the resources. Table 6.1: Milk: National Production and Operation Flood Procurement and Sales Milk Production Operation Flood Operation Flood (Million Metric Tons) Milk Sales Milk Procurement Percentage 1951 17.41 1956 19.72 1961 20.38 1966 19.37 1969/70 20.74 1970/71 0.37 0.14 1971/72 22.50 0.39 0.41 1.8 1975/76 0.58 0.43 1977/78 28.40 0.81 1980/81 1.05 0.96 1981/82 34.30 1.22 1985/86 44.00 2.23 2.96 6.7 1988/89 2.65 2.99 1990/91 54.90 3.02 1991/92 3.16 3.52 1992/93 3.16 3.98 1993/94 60.80 3.23 4.17 6.9 1994/95 63.31a 3.52 4.05 6.4 1995/96 66.31 a. Dairy India 1997, p. 160, "anticipated." b. Dairy India 1997, p. 160, "target." Notes- 1. The production estimates are total production and are based on a sample survey. Purchases and sales are from NDDB. 2. For some reason the production survey does not include the amount marketed, so there are no official estimates of total marketings. 3. To some extent this compares different things, since the fat content of milk can vary from say 3.0 percent in a poor Fresian to 12 percent in a good buffalo. A proper accounting would include data on butterfat and solids-not-fat (SNF) produced. Source: Aneja 1994; NDDB undated; and Gupta 1997, Table 108, and study data. 6.6 Given the policy decision that had been made and the dynamic nature of Operation Flood management, the most likely "counter factual" if the Bank had failed to support Operation Flood is that one or more other donors would have stepped into the breach. However, without any donor support, it is reasonable to suppose that the expansion involved in Operation Flood would have had to be implemented more slowly. This slower expansion would have forced increased reliance on imported milk powder, urban milk shortages, or significantly higher milk 61 prices. Any of these alternatives would have invited a return to more interventionist and managed (more strictly, mismanaged) dairy policy. Such a policy reversal would then have undermined the expansion of the whole dairy industry, which was the real payoff from the policy change. 6.7 It takes nothing from the achievements of Operation Flood to suggest that its biggest contribution may have been to ensure that the non-Operation Flood portion of the industry was allowed to grow. This was excellent policy-based lending-not to achieve a change in policy but rather in support of a policy change that was already being implemented.o 6.8 Two key policy changes were involved with this series of projects. The first was to stop direct (especially central) government support for intervention to provide low cost milk by the importation of skim milk powder and butterfat and direct milk collection and processing. This policy change had already been adopted when the first Bank project was approved, and Bank lending can thus be seen as supportive of this changed policy. The second policy change was to get state governments to stop implementing dairy policy through the cooperatives by interfering with pricing or staffing. Table 6.11 shows that progress was made in this direction, but it was incomplete. NDPII was predicated on attaining independence of dairy cooperatives; this goal was not fully achieved. Supply or Demand Shift? 6.9 An increase in production (or consumption) over time can be brought about in two ways either by a movement along the supply (or demand) curve or by a shift in the relevant curve. These have very different developmental implications and are at the core of the attribution of benefits in this impact study. An increase in production due to a shift along the supply curve implies that higher prices are being paid for the product (due to the demand curve having shifted to the right). No technical progress has been achieved and should prices decline to their former level, then production would also decline to its original value. If this happened, the observed increase in production would be brought at the cost of higher prices and be caused primarily by changes in the economy generally. A (rightwards) shift of the supply function itself, however, represents a willingness by producers to supply a larger quantity at the old price (or the same quantity at a lower price). This is "economic progress" or "development" and represents improved technology, costs reductions, and accumulated investments in production (or just possibly a price decline in alternative products). To the extent that the demand curve has shifted to the right, this represents a stimulus to the dairy industry due to macroeconomic factors such as population growth, higher incomes, and reduced transaction costs in a liberal economy. To the extent that the supply curve has shifted to the right, this represents technical progress and development. It is important in interpreting an observed increase in production to know to what extent it is the result of macroeconomic factors at work in the economy and to what extent it represents development on the supply side. 50. The rapid (4.7 percent) growth of milk production has been challenged as "biologically impossible" (Mishra and Sharma 1990). They estimate milk yield would have to increase "7.6 percent a year during 1971-72 to 1977-78" and in 1985-86 by 18 percent. They make no allowance for the crossbreeding of local cows (which triples the yield). When this is factored in, even mating no more than 3 percent of local cows to exotic breeds leads to a cumulative improvement in yield of 2.2 percent per annum which, while high, is not biologically impossible. 62 6.10 Over a twenty-five year period, it is to be expected that both demand and supply functions will have shifted. Demand will have shifted under the influence of macroeconomic variables such as population and income increase and lowered transactions costs, while the supply function will have shifted due to changed technology (such as crossbred cows) and accumulated investment in the industry. These variables cannot (or certainly have not) been observed directly. As shown in Table 6.2, the real price of producer milk has fallen from Rs 7.77 per liter in 1975-76 to Rs 5.29 per liter in 1993-94 (both in 1990-91 Rs). Real urban processed milk prices are graphed against total milk production over the life of Operation Flood in Figure 6.3. This clearly shows that the real price of milk has fallen even as production has increased. This fact argues that there has been a larger shift in the supply function. The first and last year's observations from Figure 6.3 have been transposed to Figure 6.4 in order to consider the relative magnitude of the shifts. Point Al in Figure 6.4 represents 1977-78, and A2 represents 1993-94. The price elasticity of demand for milk in India has been estimated at -1.077 (Dairy India 1997, p. 59). As shown in Figure 6.4, this implies a shift from demand function DI to D2. This is a shift in quantity demanded from QI to Q2. If short-run supply were perfectly inelastic, the shift in supply would be from QI to Q2. In the less extreme case where supply elasticity51 is taken to be 1, represented by supply function SI, the shift in the supply function is from Q l" (the amount which would have been supplied at the lower price in 1993-94 using the original SI supply timeline) to Q2 (the amount which actually was supplied). In either case, the shift in supply exceeds the shift in demand. Both have shifted-supply more than demand. 6.11 As shown in Figure 6.4, there appear to have been both supply and demand shifts. The annual rate of production increase (i.e., shift of the supply function at constant prices) for Operation Flood (4.7 percent) exceeds that for rice and wheat from the commencement of the green revolution (3.4 percent). For wheat alone, the rate of increase (5.0 percent) has been slightly higher than for milk, while for rice, the rate (2.8 percent) has been significantly lower. Thus, the supply function shift for milk under Operation Flood and associated policy changes52 has been quite comparable (even greater than) the supply shift for the green revolution grains in India. Some comparisons between the growth rates, price declines, and economic significance of wheat, rice, and milk are given in Table 6.3. 51. It has not been possible to identify an estimate in the literature of the supply elasticity for milk in India. 52. Government production support to the dairy industry (animal health, Al, extension, and credit schemes for the purchase of cows) has been provided both before and after the period of accelerated growth in both Operation Flood and non-Flood areas. 8 A, 7.5 --1977-78 10 '1981-82 7 6.5 - "1990-91 1985-86 & 55 0 A2 A 1993-94 "i 5 45-- 45 20 25 30 35 40 45 50 55 60 65 Total Milk Production (million metric tons) Fig 6.3: Time Series of Total Milk Production and Real Price of Milk (Note shift of origin) D1 SI D2 8 - . .- - - .- - - . . . . . . . A,(1977-78) 5 I I 6o S AL 3(19394 0D1 D2' a. ,. 4 0I III II 0 10 230 40Q 50 60 Q2 Total Milk Production (million metric tons) Figure 6.4: Demand and Supply Shifts (1977-78 and 1993-1994) 65 Table 6.2: Nominal and Real Processor Prices for Cow Milk Nominal Price Year of Cow Milk GDP Deflator Real Pricea 1975-76 2.395 30.8 7.775974 1976-77 2.500 32.7 7.645260 1977-78 2.500 34.2 7.309942 1978-79 2.230 34.8 6.408046 1979-80 2.870 40.2 7.139303 1980-81 3.330 44.9 7.416481 1981-82 3.820 49.5 7.717172 1982-83 4.000 53.1 7.532957 1983-84 4.000 57.7 6.932409 1984-85 4.000 62.0 6.451613 1985-86 5.000 66.6 7.507508 1986-87 5.000 70.9 7.052186 1987-88 5.580 77.0 7.246753 1988-89 6.000 83.2 7.211538 1989-90 6.000 90.1 6.659267 1990-91 6.000 100.0 6.000000 1991-92 6.380 114.5 5.572052 1992-93 8.000 124.9 6.405124 1993-94 8.000 151.2 5.291005 a. 1990-91 Rs. Source: Gulati and Bhide 1997, Table A5.36. Table 6.3: Price and Production Changes in Wheat, Rice, and Milk Growth Rate Rate of Price Change Farm Gate Value of (%) (%) Production (1991) Wheat 5.6 (2.3)a Rs 172 billion Rice 2.7 (1.1) Rs 356 billion Milk 4.7 (1.4) Rs 242 billion a. Parentheses indicate a decline, i.e., real wheat prices declined at 2.3 percent per annum. Notes: 1. The growth rate for rice and wheat is calculated for the period 1961-94. The data is from the World Bank database. 2. The growth rate for milk is for the period 1971-94 (see Figure 6.2). 3. The rate of price change for wheat and rice is calculated for the period 1966-91 (for producer price in constant 1990 Rs). The producer price data is from the World Bank database. The GDP deflator for India is from the International Financial Statistics Yearbook (1995). 4. The rate of price change for milk is calculated over the period 1975-96 (for cow milk at processor level in constant 1990 Rs). The data for cow milk is from Gulati and Bhide (1997), Table A5.36. The GDP deflator for India is from the International Financial Statistics Yearbook (1995). 5. The farm gate value of production for milk is calculated using producer price of milk from Gulati and Bhide (1997), Table 3.1. 66 Economic Performance 6.12 Total Bank investment in the five projects was $500 million nominal, which converts after compounding at 10 percent and allowance for inflation to $1.9 billion in 1996 dollars (Annex A). By 1996, the higher growth rate associated with the Operation Flood period (especially the associated policy changes, para. 6.3) was resulting in an extra 43 million metric tons of milk per annum. Since the start of the faster growth trend, and using a 10 percent rate to compound its value to 1996, the total increment was 1,086 million metric tons. Each ton would require about $310 of imported ingredients if it were to be replaced with recombined milk. If even 0.56 percent of the observed increase in milk production were due to Bank investments, the projects would return an economic rate of return (ERR) of 10 percent; if all investments (from the WFP, EC, and government of India) are taken into account, at most 2 percent of the increased production would need to be attributable to the project to yield an ERR of 10 percent. It seems highly likely that at least these modest increases can be attributed to the projects. Two cost headings not included in these calculations are (i) state reimbursement to state federations and MPUs for losses incurred on the instructions of government or under state-appointed managers and (ii) general assistance to the dairy industry since these expenses would have been incurred even without a change in policy. 6.13 Using a narrower definition of project impact (i.e., direct nonpolicy impact), the PARs for the first four projects and the ICR of the final project provide the estimates of economic benefits listed in Table 6.4. Table 6.4: Estimates of Economic Benefits of Bank-Assisted Projects Estimator Result Karnataka (Cr. 482) ERR 22% Rajasthan (Cr. 521) ERR 28% Madhya Pradesh (Cr. 522) ERR small or negative National Dairy I (Cr. 824) ERR 17-37% National Dairy II (Cr. 1859) ERR 21% Source: World Bank Reports 1987c, p. 54, and World Bank Reports 1997a, Table 9. 6.14 The cost of Operation Flood depends on what items are included. Table 6.5 provides the impact's estimates of these costs (see Annex B for the derivation of this table and how these estimates differ from earlier estimates). One item which is missing from Table 6.5 is direct assistance from state governments to MPUs and state federations. It has not been possible to obtain estimates of these transfers. Typically they are made to compensate for losses made by MPUs or federations. However, such losses occur in two quite different ways. The first is where a fully independent, farmer-controlled MPU or federation makes losses in the normal course of its business. Government transfers to make up for such losses can clearly be described as "subsidies." A second situation is where the losses are directly due to government dictated pricing decisions or losses made under government nominated management or due to government mandated overstaffing. Government transfers in such circumstances are better described as "reimbursement" to the cooperatives, even though they may in the case of pricing directives be subsidies to consumers or producers. Since it has not been possible to derive these costs, it has not been necessary to try to partition them into subsidies and reimbursements. 67 Table 6.5: Impact Estimates of Cost of Operation Flood (1996 US$ billion) Item Amount Real Direct Subsidies Central Government 0.033 State Government 0.054 State Governments (Processing Plant Transfers)a 0.799 Commercial Banks 0.006 Farmers 0.001 Food Aid (Through Consolidated Revenue) 0.962 Food Aid (Direct to IDC) 0.184 Total Real Direct Subsidies 2.039 Bank Loans and Credit 0.694 Total Loans and Subsidies for Operation Flood 2.733 Other Central Government Dairy Expenditures 0.250 Total Government Assistance to Dairy Industryb 2.983 a. Budget-based figure. See Annex B, para. 17. b. Includes food aid and Bank support but is net of some state government assistance. Source: Annex B, Table B15. 6.15 A further cost which has not been included is the general support that state governments extend to the dairy industry in general through extension, research, animal health, AI, and regulatory services. Again, it was not possible to assemble the necessary cost data. Though provided to both Flood and non-Flood areas, these costs still enter the overall envelope of financial assistance to the dairy industry as a whole. 6.16 A cost which is included, but is not at all well documented (Annex B, para. 17), is the value of government dairy plants (and sometimes associated debts) transferred to dairy cooperatives or leased to them without charge or for a peppercorn rent. This has been set at US$0.8 billion in 1996 dollars. 6.17 The costs in Table 6.5 refer to the total value of assistance over the period 1971-72 to 1996-97, expressed in 1996 US dollars. It is important to note that for the costs covered, these cost elements cover both capital and recurrent costs. Annual central government expenditures on animal health, for example, have been converted to 1996 US dollars and summed. 6.18 The most general level of assistance is total central government assistance to the dairy industry in general, including Operation Flood. Much of this assistance (especially for Operation Flood) was provided from food aid and Bank loans and credits. Most of this assistance flowed through the government's consolidated accounts, appearing first as consolidated revenues and then as consolidated expenditures. The total cost of central government, including transfer of state assets to cooperatives, is estimated to be US$2.98 billion. 6.19 Of this, US$250 million was for identifiably non-Flood central government support to the industry in general, leaving US$2.73 billion as the assistance to Operation Flood (strictly assistance to NDDB/IDC and Operation Flood, see para. 6.21). The sources of this funding are shown in Table 6.5. 68 6.20 This leaves two possible sources of overestimation. First, there is the very shaky estimate of US$0.8 billion for the value of plant transfers. If this is deleted, the estimated expenditure falls to US$1.93 billion (now probably an underestimate). 6.21 Another possible source of overestimation is the inclusion of all payments to NDDB/IDC from the central government as costs of Operation Flood. Starting in 1979-80, the consolidated accounts itemize expenditures on NDDB/IDC. How much of this was for Operation Flood? Earlier they did not make this break-out. It was felt that some of the payments to NDDB/IDC might be for closely related dairy activities which were highly supportive of Operation Flood. Accordingly, the above estimates have included all central government payments to NDDB/IDC. 6.22 Finally, we need to know how many beneficiaries. Since many farmers have only one or two cows, they will be able to supply milk at some times of year, while not at others. NDDB has provided estimates (Annex B, Table B 16) that 6.3 million members supply milk at some time in the year, but only 3.5 million poured milk in December 1996. Accordingly, costs per beneficiary are also provided on the basis of the 6.3 million members who supplied at some time in the year and the 3.5 million farmers who supplied in December, 1996. 6.23 These alternative cost and beneficiary estimates have been consolidated in Table 6.6 into a range of estimates of per beneficiary capital costs. Even though some state assistance to the dairy industry is excluded, the first row of Table 6.6 is clearly an overestimate, since total assistance to the dairy industry has been divided by the producers of only 6.3 percent of total production. The second and last rows probably provide good upper and lower bounds on the cost of Operation Flood. Remembering that this covers capital and recurrent costs (para. 6.17), the annual cost per member (who pours at some time in the year) for capital and recurrent expenditures is between US$10 and US$17 .s Table 6.6: Alternative Estimates of Capital Costs of Operation Flood per Beneficiary in Constant 1996 US$ Per Members Per PourerP Total Assistance to Industry 473 851 Assistance to Operation Flood 433 780 Assistance Net of Plant Transfers 306 551 a. 6.3 million members. b. 3.5 million members who supplied in December 1996. Source: Annex B, Table Bl7. 6.24 Another way of looking at economic performance is to consider either how the same outcome (consumption and price level) could have been achieved by alternative means (trade, in this case) or the change in producer and consumer surplus induced by the project. The appropriate calculations for one year (1996) are provided in Annex C. The results are given in Table 6.7. That the dominant gains are in consumer surplus confirms the argument advanced elsewhere (paras. 6.9-6.11) that the dominant impact has been a shift in the supply function. As 53. If one wanted to express this for "an average year of the project," this could be doubled since there has been a build-up in membership. On the other hand, the industry still has substantial assets; not all these resources have been used up. 69 calculated in Table 6.7, the gains in just one year would have paid for the whole of Operation Flood and associated policy changes, thus underlining the key importance of getting policies right. Table 6.7: Alternative Estimates of Annual Gross Benefits Under Operation Flood and Policy Change, 1996 Consumer Producer Total Total < - (Rs billion- )- > (US$ billiona) Import Replacement n.a. n.a. 288.9 9.2 Short-run Supply Assumption Inelastic 435.0 0.0 435.0 13.8 Unit Elasticity 269.0 17.0 252.0 8.0 a. Exchange rate= 31.37. Exploration of the Free Trade Option 6.25 The political realities in India, prior to the late 1980s, ruled out a reversal of the highly protectionist policy of economic self-reliance or, more modestly, an exemption for the dairy industry. Nevertheless, it is appropriate to examine the magnitude of benefits forgone by refusing to follow a "first best" policy of free trade, at least for the dairy industry. 6.26 An attempt to model general free trade in a formerly protected economy runs into the intractable problem that the new price regime would change the distribution of income and wealth within the economy, which would lead to shifts in demand functions, which in turn determine the new price structure and hence incomes and wealth. There is no satisfactory way to model these simultaneous and recursive changes. 6.27 This section finesses this problem by asking what would have been the impact of an exemption of self-reliance for the dairy industry, leaving the rest of the economy protected? In the interests of clarity, the assumptions are simple and minimal. The first assumption is that only urban consumption of milk would have been affected (i.e., about 30 percent of production). This implies no effect on village-level consumption and, more seriously, no impact on processed products. The second assumption is that, under free trade or self-sufficiency, without Operation Flood the supply function would have shifted half as much as it did under Operation Flood. That is, technical advances would have been adopted, but not as quickly as under the stimulus of Operation Flood. The results for 1993-94 (the last year for which production data are available) are given in Table 6.8. 70 Table 6.8: Costs and Benefits of Alternative Policies for Urban Milk Supply, 1996 (US$ Billion) Operation Flood Free Trade Consumer Surplus 0.19 0.52 Supply Elasticity = 0 Producer Surplus (0.57) (3.28) Foreign Exchangea 0.00 (0.73) Supply Elasticity = 1 Producer Surplus (0.29) (1.63) Foreign Exchangea 0.00 (1.77) a. Raw material only. 6.28 The argument for free trade is that it leads to a maximization of consumer surplus and indeed Table 6.8 suggests that consumer surplus would have been increased by free trade, as compared to Operation Flood. However, the extra foreign exchange cost of imported raw materials exceeds the gain in consumer surplus (and this foreign exchange represents other consumption forgone in the rest of the economy).54 To the extent that weight is put on the loss of producer surplus (much of it going to the very poor), then the advantages of the free trade option are reduced. Industrial Structure 6.29 The industrial structure which emerged as a result of Operation Flood involved an organized sector composed of Operation Flood cooperatives, private companies, and "other cooperatives and government schemes," and an informal sector composed of small traders (dudhiyas), urban milk producers, and intravillage trade directly from producer to consumer. Both the small traders and urban milk producers sell directly to urban consumers without pasteurization or other factory treatment. 6.30 Although Operation Flood cooperatives comprise the largest single dairy organization, it is engaged in active competition with small traders in peri-urban villages and with some private companies in urban milk sales. Relative milk procurement is given in Table 6.9. 6.31 Prior to the New Economic Policy (NEP) announced in the 1992-93 budget, competition within the dairy industry was controlled in two ways. Firstly, NDDB had a monopoly on the importation of dairy products. Secondly, construction of new processing capacity was subject to industrial licensing. Repeal of industrial licensing under the NEP resulted in a boom in investment in new processing capacity, since this was obviously a growth industry. This investment also enjoyed a large number of incentive payments and tax holidays as described in Box 6.1. The well-known case for the economic efficiency of private investment and competition rests on the assumption of an absence of significant subsidies and taxes. Clearly, this condition is not currently met in the Indian dairy sector. As a result of the NEP (liberalization accompanied by subsidies), significant excess processing capacity has been created. Thus, prior to liberalization, the Operation Flood cooperatives had about 58 percent of the processing capacity 54. Of course, with a full macroeconomic analysis, the exchange rate would shift too, thus changing consumption patterns and incomes throughout the economy and leaving the position of the demand function itself moot. 71 in the Punjab with the balance in the rest of the organized sector. Since liberalization, the supply of milk to the organized sector is expected to rise by 0.56 million 1/day. The Operation Flood cooperatives have raised their processing capacity by 0.28 million I/day (roughly in line with a 50 to 60 percent market share). The balance of the organized sector has raised installed capacity by 1.1 million 1/day and applied for a capacity expansion of 2.71 million 1/day.55 Thus, the relatively orderly, low cost, and competitive expansion of the dairy industry which characterized the Operation Flood period has been succeeded by a rapid and excessive expansion of processing capacity under the NEP, driven by subsidies far in excess of anything which was offered in the Flood period. Table 6.9: National Milk Procurement by Operation Flood, Other Organized, and Informal Sectors (MMT and percent) 1961 1972 1980/81 1988/89 1992/93 1994/95 1995/96 1996/97 1. Total production of milk MMT 20.4 23.0 31.6 53.7 59.0 64.1 66.1 70.1 2. Organized sector 0.8 1.3 3.1 5 8 n.a 6.4 n.a. n.a. 3.68 5.55% 9.93% 10.88% n.a. 10.10% n.a. n.a. 3. Operation Flood cooperatives n.a. 0.24 1.01 3.58 3.85 3.75 4.00 4.40a 1 03% 3.21% 6.67% 6 53% 5.85% 6.05% 6.28% 4. Non-Flood organized sector 1.04 2.12 246 2.63 4 52% 6.72% 4.20% 4 16% 5. Handled by Operation Flood 18 56% 32.33% 61 33% 58.77% cooperatives as percentage of organized sector 6. Handled by the informal sector 19.65 21 72 28 46 47.87 56.72 96.30 94.44% 9007% 89.12% 89.88% a. Provisional. Notes: 1. Data for rows 1, 2, and 3 is from Shah and others, undated draft, Table 3.1. 2. Rows 4, 5, and 6 have been calculated from the data in rows 1, 2, and 3. 3 Data for rows 1, 2, and 3 for the years 1994-95 to 1996-97 is supplied by the NDDB, Anand. 6.32 A Milk and Milk Products Order (MMPO) was introduced eight months after the 1992- 93 budget to dampen the "irrational exuberance" of private sector dairy processing investors. This again requires licensing of large plants (over 50,000 1/day) but has not been very effective. 6.33 Most private processors rely on contractors and informal traders to procure milk from farmers, only taking possession of the milk at the factory receiving dock. They thus have no 55. "Fears about over-capitalization were further reinforced because many state governments were reckless in announcing all manner of concessions and subsidies in order to attract private capital to their state. In Haryana, for example, some 40 new plants were proposed to raise processing capacity to a total of 40 lakh liters per day (lId) from the existing 15 plants which had a capacity of 10 lid. This looked ridiculous because a majority of the existing plants were hardly viable; the Haryana Cooperative Federation, which has been operating for nearly two decades, had never been able to collect more than 2 lid in flush and 50,000 liters in lean seasons. There seemed no way the new plant could become viable. Vij has suggested that the private businessmen were interested more in the tax breaks and subsidies which they could claim under the present regime. In his estimate, the 11 new plants proposed would earn, by way of tax exemptions alone, nearly Rs 100 crores by making disproportionate investment in land and machinery expressly for claiming these concessions; there would be additional subsidies and concessions from both the state as well as the central governments, most of which were not available to co-operatives" (Tushaar Shar and others, Institute for Rural Management (IRMA), Undated, "Institutional Structures for Dairy Development: India's Post-Independence Experience," Draft, p. 47.) 72 direct contact with producers and no control over the treatment of milk before it reaches the factory (Box 4.1) or the prices paid to farmers. Milk is, of course, tested on receipt. Box 6.1: A Level Playing Field? India is a very diverse, federal republic, meaning that much industrial policy is constitutionally determined by the state governments. Both before and after the NEP, the most rapid growth of private sector dairying has been in the Punjab. Even before the NEP, one multinational had a plant with 0.8 million I/day processing capacity, compared to a total cooperative capacity in the state of 1.5 million I/day. Since the NEP, the private sector has added about 1.1 million I/day nominal capacity. It is nominal since reporting requirements are lax, and investors have an incentive to overstate intended processing capacity in order to attract larger state equity investments. Incentives available to attract new industrial investment (including private corporatea dairy processing) in the Punjab include: * A 49 percent state equity in large companies as ajoint venture, thus halving the capital an investor must put up. State equity is contributed in cash, whereas the entrepreneur's contribution can include assets and good will, possibly valued quite arbitrarily. * Reduction for seven years of 8.8 percent state sales tax to 4.4 percent. * Exemption from a 20 percent surcharge on power consumption. * An income tax holiday for seven years. In addition, any milk described as originating from outside the state is exempt from a 4 percent tax on milk for processing. Since the cooperative system is organized on an intrastate basis, this results in a defacto concession to b the private sector. The sales tax concession alone far exceeds a reasonable profit margin for milk processing. Forced to register under the Registrar of Cooperatives, cooperatives have to operate in a much more paternalistic and interference-prone environment than companies that answer to the Registrar of Companies. Though perhaps partially self-imposed, cooperatives relate their salary scales to those within the public service. Managerial and professional staffs can typically at least double their salaries on joining a private company.' It is possible to argue about the level of assistance provided to the cooperative sector in the past,d but as of today, the "playing field," in Punjab at least, is heavily tilted in favor of the private corporate sector, which can only lead to serious misallocation of resources. In the milkshed for the Ropar District Cooperative Milk Producers' Union, 0.9 million 1/day (nominal) capacity was added by private investors in the eight-month period following the announcement of the NEP and before the introduction of the MMPO. The Ropar MPU has a capacity of 0.1 million 1/day. Though the five plants that resulted draw from a wider area than the Ropar MPU, the probability that they could all survive was low ex ante, and in fact one has already closed, and others may do so. Capital investment cost is about Rs 1,000 (US$27) per liter. The investment has been about US$25 million, some of it now of little more than scrap value. a. But not cooperative investment, since this falls under the Registrar of Cooperatives. b. Similarly, D.M. Attwood and B.S. Baviskar (1987) remark in the context of sugar". . . although the [sugar] cooperatives receive some helpful subsidies from the government, much larger subsidies are directed to the private factories at the expense of the more efficient cooperatives." c. Indeed, the managerial and professional staffs may well prove to be the main beneficiaries of the liberalization in the dairy industry, as cooperatives are forced to go some way toward meeting private sector salary scales for key staff. d. The activities of the NDDB have been almost entirely self-financed, albeit including substantial profits generated by the sale of concessionally priced food imports. Bank assistance was advanced in the form of loans, which are being repaid on agreed terms. The government accepted, as it should, the foreign exchange risk of borrowing externally, and inflation has substantially eroded the real cost of the NDDB's debt, as it has for all debtors. It is clear that any debate as to subsidy levels is unlikely to be fruitful, unless it starts with a conceptually clear definition of the subsidy element to be estimated. 73 Farmer Incomes 6.34 We have already seen that milk production has grown at over 4 percent annually since the inception of Operation Flood. In constant (1995) rupees, the annual payment by the cooperative system dairy farmers has risen from Rs 2.1 billion in 1972 to Rs 34.0 billion in 1995 ." Village interviews during the evaluation mission brought out two key points. First, dairying is shifting, in a few cases, from a sideline activity to a serious economic enterprise and, in some cases, even becoming the main source of farm income. Second, dairying was particularly valued because of the reliability and regularity of payments. In many families, dairying is relied upon to pay recurrent household expenditures, leaving the crop income to finance investments and major social events. A cost of production study carried out by the Punjab Agricultural University in two milksheds in the Punjab (Gill and others 1995) reported an average gross revenue of 9.30 Rs/l in 1994-95, with a gross margin of 3.61 Rs/1 and a net return of 3.06 Rs/l, or 33 percent. The study correctly included imputed costs of labor, capital, and land. Rates of return and payback periods were correspondingly high (45 percent) and low (2.5 years), respectively. 6.35 This high return to dairying is in part borne out by reports that milk vendors continue to be able to purchase milk from some non-DCS villages at discounts of as much as 30 percent off the cooperative price. Credit 6.36 The impact of a DCS on credit arrangements in a village is very complex. In the first place, neither a DCS nor an MPU will extend credit officially. However, some DCS secretaries may informally provide advance payments, though this is discouraged due to the possibility of embezzlement and discrimination in favor of an elite. Some MPUs visited have acted as brokers, putting DCSs or individuals in touch with private banks or official credit schemes. 6.37 Membership in a DCS often improves a person's creditworthiness with retailers and others (Table 3.9). However, the willingness of dudhiyas to extend credit quietly, quickly, and in confidence in times of need is highly valued and in part explains their persistence even in villages with DCSs (George 1991, p. 284). The high real rate of interest charged is still preferable to borrowing from a money lender, since future milk income can be used as security. Women 6.38 Women in most Indian villages have a strongly socially-constructed role.58 They run the household and are employed outside the home only in very poor families where the husband is 56. Dairy India (1997, p. 18a) gives the price of milk as 9 Rs/kg. Volume is from Table 6.9. 57. Dairy India (1997, p. 5). This agrees closely with the estimate of Rs 30.0 billion given in the ICR, Appendix D, p. 6. 58. India is, of course, a highly diverse country. The generalizations expressed in this section undoubtedly have many exceptions. However, they are useful in emphasizing the revolutionary impact women-dominated DCSs can have on women, at least in some villages. 74 unable to earn even a minimum survival income.59 They are traditionally dependent on their husbands for even quite modest and essential expenditures. They play virtually no visible role in village politics or social organizations and are understandably shy. Their household tasks include the care and feeding (and dung utilization) of dair animals. In a male-dominated DCS, the husband may or may not collect the milk income. Who collects milk money is less important than for what it is used. Provided it is used for current household expenditures, food, and education, the woman's practical needs are met. Empirical data on this point is weak. 6.39 The social (and production) impact of women's DCSs (WDCSs) is dramatic. In such DCSs only women members are accepted, the management committees are composed exclusively of women, and the secretary and Al worker can be women. Women thus find themselves empowered, since they are "authorized" to meet collectively outside the home and explicitly make their own decisions. Latent leadership talents emerge, and self-confidence is built and reinforced, even among the followers. 6.40 The Bank's study of Gender and Poverty in India (World Bank 1991, para. 4.18-4.31) emphasizes the lead role played by NGOs (notably, the Self-employed Women's Association and Bhagavatula Charitable Trust), in some cases with Ford Foundation support, in establishing WDCSs. Increasingly, WDCSs are being mainstreamed in Operation Flood's expansion. There is, of course, no compulsion for NGO-organized village women's dairy societies to join Operation Flood. For the most part, they have chosen to do so and have been welcomed. 6.41 Because dairying is their primary, and usually only, commercial activity, women, when empowered, prove more adept at the utilization of improved husbandry methods than their husbands. At home all day, they are better able to identify when an animal comes into heat, for instance, and thus to get it to the AI worker in a timely fashion. 6.42 When women get the milk income (by far the dominant pattern in a WDCS), it enables them to make most household expenditures without having to ask their husbands for money and even to save small amounts for emergencies. Women report much reduced domestic tension from having an independent, if modest, income. 6.43 Discussion with WDCS members reported some initial opposition from the village men in some cases, but especially when the women demonstrated their ability to increase production and a reduced need for income from the husband, opposition changes to support. The man who had recently been ashamed to have his wife be seen taking an active role outside the house switched almost unconsciously to taking pride in her leadership qualities as demonstrated within the WDCS. As for the women, they are excited by their achievements. 59. The Bank's study of Gender and Poverty in India (World Bank 1991, p. viii) emphasizes the inside/outside dichotomy with most women's tasks being confined within the household, resulting in conscious or unconscious female seclusion. This dichotomy is one which both boys and girls are led to expect almost from birth-hence, the reference to a socially-constructed role. 60. The women's literature emphasizes the collection of milk money by men, even though the money may be given to the wife for safekeeping (Rangnekar, Vasini, and Rangnekar 1994 and Sinha undated). 61. NDDB notes that the absolute number of women members of DCSs increased from approximately 66,000 in 1984- 85 to approximately 1.70 million in 1995-96. 75 6.44 Visits to the dairy factory and feed plant and interstate travel to Anand (for WDCS chairwomen) gets women out of the village as a group and widens their horizons. Moreover, the thorough understanding of the cow's reproductive cycle, which is part and parcel of the effective use of Al, provides a compulsory tutorial on the possibilities for family planning. 6.45 No doubt Operation Flood will be criticized by some for not promoting a "gender neutral" approach to cooperative development.62 As their name suggests, WDCSs do imply a gender-differentiated role for women. Because it provides social endorsement for women to meet outside the home, to develop leadership skills, and to take business responsibility, it changes significantly the socially-constructed pattern by which women expect to live their lives. It is a vastly improved pattern, and that should be enough. 6.46 Nominally gender neutral, the original project design resulted in male-dominated management committees for the DCSs and male articulation of needs to the MPU and Operation Flood. The dominant productive role of women was not reflected in the management of the DCS, nor was technical information extended to them directly. The Cooperative Development program, especially in the context of WDCS, has increasingly emphasized the role of women and has organized visits to the dairy factory and to discuss the rights and responsibilities of (women) members as well as Al, animal feeding, and improved husbandry. With something like 6,000 WDCSs (and more to come) or approximately 300,000 active members of WDCSs, Operation Flood provides an example of large-scale program redesign to address gender issues. Employment Generation 6.47 As mentioned above, one of Operation Flood's key impacts was to allow women to give up coolie (laboring) work outside the home. In addition to releasing children to go to school, this also translated into "employment creation." With 3.5 million pourers, it is reasonable to assume that 5 percent represented women who were able to stay home rather than go out to work. This withdrawal of women from the labor force will have opened up an additional 175,000 laboring jobs (admittedly ill-paid jobs), predominantly for the very poor. Operation Flood has been a major employment generating project. Improved Technology 6.48 Operation Flood has been assiduous in promoting improved technologies, including Al; crossbred cows; vaccinations; improved feeds and forage seeds; urea treatment of straw, fan, and sprinkler cooling of cows in summer; and biogas plants. In particular, the program has provided training to villagers in Al, elementary animal health, and first aid. Far more animals are inseminated, at higher conception rates, by these Operation Flood technicians than by the professional veterinarians stationed at government expense at Al centers, even when the density of the government system reaches as high as one center to six villages. The changes promoted by 62. The normal DCSs are gender neutral according to their bylaws. Membership is open equally to men and women (but only one person per family, a rule relaxed in some DCSs). The only gender-specific requirement is that at least one woman be on the management committee. But in practice, such societies are usually run by the men of the village, with the token woman making no significant contribution to decisions made by the committee. 63. Biogas certainly beats handmade cow-dung patties as a cooking medium, especially for the women, who traditionally not only have to cook but also make the dung patties. Many programs promote biogas plants. Indeed, the key way Operation Flood promotes them is to broker farmer access to such programs. 76 Operation Flood have been widely adopted by other farmers and villages. Anecdotal information suggests that maximum yield per animal is rising quite fast, but none of the available statistics are in a form that allows for checking yields within the cooperative system. And indeed, village level time-series data only hint at the trend in procurement per farmer that might have been expected to follow from the higher yields reported in interviews. Adulteration 6.49 Adulteration is a long-standing problem in India, as it is elsewhere in the developing world. As incomes increase, consumers can be expected to become more discriminatory, and hygiene regulations are likely to be enforced more strictly. In their 1986-87 study, Kaur and Gill (undated) make the throwaway comment that "the share of creamery in the consumer's rupee was 15.18 percent which was apparently low but they earned more by way of adulterating milk." In a 1993 study, Gill and Kaur (1993) report that 70 percent of the 90 consumers interviewed mentioned problems with adulterated milk, while 68.8 percent mentioned unhygienic milk and problems of taste. Middlemen (dudhiyas) in 32 percent of cases mentioned problems with adulteration of milk by farmers. However, the dudhiyas 'most frequent complaint was that they were "harassed by the staff of Health Department who took bribes by threat of taking milk samples" (Gill and Kaur 1993). Box 4.1 describes one unhygienic noncooperative milk cooling center; Box 4.2 documents problems with "artificial milk," and Table 6.10 reports that 16 percent of the milk samples drawn by the Health Department in the Punjab showed adulteration" in 1994 and 24 percent in 1995. More significant is the very small number ofsamples taken. In 1994, the number was less than 600, and in 1995 it will be less than 500 if the few nonreporting villages are sampled at the same rate as in 1994. Two large DCSs (of the 5,100 DCSs in the Punjab) would take this many samples in a single day. The conclusion from Table 6.10 is that in the last two years little has been done to prevent adulteration of milk in the Punjab, and quite likely the same conclusion applies to other states. As reported in Box 4.2, when consumers in Delhi became dissatisfied with the quality of the milk on offer, they turned to an NGO (FRAC) rather than the Health Department. An Indian reviewer has commented that this study's concern with adulteration of milk may be misplaced: My experience suggests that consumers generally pay for what they buy. In the Indian context, normally each dudhiya supplies not one kind of milk but 3 or 4 kinds. The consumers are aware which kind they are buying-namely they pay a lower price for adulterated milk, still lower for much more adulterated milk, and a fairly high one for 'pure' milk. Also, contamination (which arises because of dilution with impure water) is taken care of in the Indian household by simply boiling the milk. This relaxed view overlooks the energy cost of having to boil milk and the plight of less well- informed consumers. Also, where there is chemical adulteration and the milk is used for processing, there may be a low-level but widespread public health hazard, even before identifiable symptoms occur. 64. Actual adulteration was probably even higher, since milk can be declared as "cow milk" even though it originated with a buffalo. This results in more lenient standards for judging milk to be adulterated. 77 Table 6.10: Punjab: Milk Samples Drawn and Found Adulterated 1994 1995 Drawn Adulterated Percentage Drawn Adulterated Percentage Jalandhar 10 3 30 17 3 18 Hoshiarpur 16 1 6 51 20 31 Gurdaspur 16 11 69 n.a. n.a. n.a. Amritsar 17 3 18 60 31 52 Ferozepur 47 7 15 24 6 25 Faridkot 100 2 2 12 1 8 Bathinda 64 9 14 61 10 16 Sangrur 79 15 19 n.a. n.a. n.a. Patiala 51 0 0 27 10 37 Ropar 43 23 3 36 9 25 Ludhiana 52 6 12 n.a. n.a. n.a. Kapurthala 33 10 30 9 3 33 Mansa 28 3 11 n.a. n.a. n.a. Artihgarhsahib 9 0 0 n.a. n.a. n.a. Notes: None of the samples drawn from tr-e cooperative sector were found to be adulterated. n.a. = District data not yet reported for 1995. Source: Punjab Ministry of Health, Study Data. 6.50 Improved hygiene was never an explicit objective of the Bank-supported projects. It never appeared in project conditionality, presumably because the projects confined themselves to supporting Operation Flood where cooperative control of milk from producer to factory stage, together with milk cooling stations and routine pasteurization of all milk, ensured high standards of hygiene. In principle, the cooperative system has a "double-entry" accounting system for milk solids, whereby the individual producer's milk is tested for SNF and butterfat. This allows estimation of the total SNF and butterfat in each delivery of milk by the DCS, which is independently checked by the MPU. Recently, the problem of artificial milk has reared its head within the cooperative system (Box 4.2), as some transporters have replaced the milk in their bulk tankers with artificial milk prior to delivery at the dairy. This will require much more sophisticated testing of milk receipts by dairies than has been customary in the past. 6.51 Most noncooperative milk processors hold down costs by "out-sourcing" milk procurement. This means that they have no direct control over milk adulteration between the producer and the factory stage. Weak operation of public milk testing services thus affects most adversely the quality of milk processed in the corporate sector. Education 6.52 Education was a high priority in all villages visited. For the poorer villages, milk income meant the difference between going to primary school and not being able to attend. For slightly wealthier villages, it helped children stay in high school or allowed those who would have dropped out to be able to afford to stay in school. In still wealthier villages, where school attendance was universal, some of the earnings of the cooperative had been set aside to help the local schools and, in one case, a local private university. 6.53 The above field observations are reinforced by the survey results reported in Table 3.9. The survey shows that increased school attendance for girls is the second most frequently mentioned impact of a DCS and increased school attendance for boys the third. Significantly, the 78 impact on education for girls is listed more frequently. This is confirmed by OED's recent study in Karnataka (Sampark 1997, p. 67): Women and Education of Children: Among the landless and marginal farmers, women go for coolie work. The main reason for girls dropping out of school is household chores along with care of siblings. Dairying enables women to stay home and supplement their income. They shoulder the responsibilities, thus enabling their young children (8-15 age group) to attend school. 6.54 In some families several children were in school who would not otherwise have been, and in others children were staying in school. Unfortunately, there is no direct quantitative data on extra school attendance. However, the likely scale of the impact can be gauged by noting that if even one family in three had an extra child in school, with 3.5 million pouring members this would be 1.1 million extra children. Even one family in ten having an extra child in school yields 350,000 extra children in school. Including the indirect effect of higher dairy incomes in general, the impact is even bigger. Of course, this extra enrollment does not come free. Extra teachers and extra school rooms have had to be provided, but given the high income elasticity of demand for education in India, Operation Flood can be seen as a major "demand side" educational project. Cooperative as NGO 6.55 Cooperatives are included in the Bank's definition of nongoverment organizations (NGOs) as membership NGOs (World Bank 1995, p. 14). As we have seen , the degree of autonomy for the MPUs and state federations is mixed. Significant progress has been made (Table 6.11) towards organizational autonomy and the full Anand cooperative principles. 6.56 At the village level, the DCS clearly is farmer-controlled and autonomous. The project thus supports 53,000 NGOs and is probably the largest program involving membership NGOs in program delivery supported by the Bank. Table 6.11: Progress Towards Anand Principles Compliance 1991a 1995b States SFs MPUs No. % % % Free to set consumer prices 8 36 80 80 Free to set producer prices 7 32 87 87 Elected boards 14 64 75 87 Autonomy in Staff Recruitment -- -- 93 76 Notes: a. By State, from NDDB, April-September 1991, "Progress Report on Operation Flood." b. Data on MPUs and federations data from Supervision Letter, November 28, 1995. 79 7. Findings and Recommendations Findings 7.1 Operation Flood can be viewed as a twenty-year experiment confirming the Rural Development Vision (World Bank Reports 1997c). It does this in several dimensions. It demonstrates: * The role of agriculture in poverty reduction * Rural development involves more than agricultural production * The value of national "ownership" in development * The beneficial effects of higher incomes in relieving the worst aspects of poverty * The capacity of agriculture to create jobs * The capacity of agriculture to benefit the poor at low cost * The importance of a commercial approach to development * The capacity of single-commodity projects to have multidimensional impacts * The importance of getting government out of commercial enterprises * The importance of market failure in agriculture * The power and problems of participatory organizations * The importance of policy But it warns: * This is not easy, and * The Bank has difficulty in recognizing success These facets of Operation Flood are discussed in greater detail below. 7.2 The Role ofAgriculture in Poverty Reduction. We have seen that not all of Operation Flood's procurement represents new production; some of it represents switching from other purchasers or uses to Operation Flood. By the same token, not all the payments made by Operation Flood represent new income to villagers. Nevertheless, the income flows from Operation Flood to villages are now massive. We have already seen (para. 6.34) that annual payments by the cooperative system to farmers now average about Rs 34.5 billion (or US$1.0 billion per annum). Allowing for both cash costs of milk production (especially cattle feed) and switching, perhaps a quarter of this, say US$250 million per annum, is net increased income. 80 Faster growth of incomes in Operation Flood villages and higher levels of milk production have been identified in village studies (see paras. 3.6-3.765 and 5.16). 7.3 For milk production as a whole, the switching problem is irrelevant since this is, by definition, extra production. About 10 percent of the increased production is handled by the formal sector, and the rest is either consumed within the village or handled by the informal sector (Table 6.9). This is extra value created regardless of how it is marketed. Since Operation Flood only handles about 6.3 percent of production, the extra cashflow to villagers from the higher dairy growth rate approximates about US$16 billion per year. Investment in dairy stock and recurrent cash costs are involved in the generation of this gross revenue at the village level, but clearly, cash flows of this magnitude can be expected to generate real poverty alleviation and linked benefits. Studies have shown that as many as 70 percent of suppliers are from the marginal and small farmers, together with some landless, so a large part of these benefits are reaching the very poor. 7.4 More than Agricultural Production. As pointed out in the introduction (paras. 1.1-1.2), Operation Flood is only incidentally a production project. Primarily, it is a marketing project. It has also been demonstrated in Zambia (para. 1.2) that provision of a market may be enough to stimulate production. Theorists, in illustrating the nonagricultural dimension of rural development, are inclined to cite the value of rural road building in reducing transport costs and, hence, opening villages to the cash economy. In fact, what Operation Flood provided was a reliable buyer in the village. In many cases, this demonstrated need for a better road resulted in either road construction by villagers or political agitation to get a road built by the panchayat. This raises an interesting question (para. 7.8) as to whether development is more effectively promoted by supply- or demand-side interventions. Is it better to build a road in the expectation that this will attract trade or promote trade in the hope that it will result in the construction of needed infrastructure? 7.5 National Ownership. As described earlier (paras. 1.3, 1.5, and 6.3-6.7), Operation Flood was an Indian project long before the first Bank-supported project was approved by the Board. Operation Flood reflected the political/developmental insight in 1964 of Lal Bahadur Shastri, a farm boy who rose to be Prime Minister. The Bank was most constructive when it simply assisted the program. 7.6 Higher Incomes Reduce the Worst Aspects ofPoverty. Unfortunately, only anecdotal information is available on the impact of Operation Flood on education (paras. 6.52-6.54 and Table 3.9). However, in every one of the approximately 40 villages visited, improved education for children was cited as among the most important impacts of higher dairy income. One of the key reasons very poor women prefer to stay home to look after a cow, rather than go out for wage labor (apart from it being a lot easier), is that this allows a child (usually a daughter) to go to school rather than having to stay home to look after younger siblings. For the very poor, education seems to be nearly as important as being able to buy enough to eat. 65. A reviewer cautions against a too facile acceptance of these sections since (i) the three-village studies show relatively little difference between Operation Flood and non-Operation Flood villages (para. 3.17); (ii) possible sampling error and the absence of baseline studies make the conclusions based on the Madhya Pradesh and Karnataka village studies questionable (footnote 39); (iii) available evidence indicates that the proportion of income from Operation Flood is modest (para. 3.12); (iv) a resurvey in 1996 of the villages in Karnataka found that defunct Flood villages were reported to be the most developed in terms of educational and income levels, number of milk animals owned, and milk sold (para. 5.16). 81 7.7 Health services are more of a public good than education. Parents know how much education they are able to provide their children. With health, the issue is the availability of service whether or not it is used by an individual respondent. Accordingly, the health impacts of improved prosperity were less salient than the educational impact. However, a recurrent theme in DCSs visited was that they had used profits from the DCS to support some improvement in the health services available to the village. 7.8 This finding poses a major question: Are the "worst symptoms of poverty" best reduced by the provision of "supply-driven," single-purpose health, education, and nutrition projects or by "demand-driven," income-generating projects? The new growth economics would assure us that both demand and supply have their own separate effects (and that there is an interaction term) in determining both the level of education, health, and nutrition enjoyed and the further growth in incomes. However, this generalized advice is not very helpful without quantification. It is probably a faiir comment that neither OED nor development researchers, more generally, nor indeed hardheaded field workers engaged in implementing projects are in a position to quantify these effects. 7.9 Capacity ofAgriculture to Create Jobs. Much of the milk collected by Operation Flood would otherwise have been processed in the village, but some comes from one or more additional cows being kept by farmers (or, more accurately, their wives) or as a result of better husbandry. However, a proportion of the extra production came from women who were enabled to stay home to look after a newly purchased cow and thus avoid the necessity to go out for work. Each such change opened up a job for someone else needing work. If even 5 percent of producers made this switch (unfortunately, we lack data to properly quantify this point), this would represent the creation of 175,000 jobs. 7.10 Benefiting the Poor at Low Cost. Taking the NDDB estimate of 6.3 million direct beneficiaries and all-up identifiable costs of Operation Flood (Annex B), Table 6.5 and Table 6.6 show that cost per beneficiary over the life of the project was from US$306 (1996 dollars) to US$433. If, however, beneficiaries were taken to include the families of these pourers, the number would be about 32 million. 7.11 Some of the benefits were undoubtedly enjoyed by larger farmers (still in most cases poor by any absolute standard), but there is strong evidence that the majority of beneficiaries were small and marginal farmers and were from "other backward castes."67 Since the Bank does not routinely calculate the unit costs of reaching the poor, it is not possible to say whether the above costs are low in any comparative sense. 7.12 A Commercial Approach. Operation Flood has had a remarkably hardheaded and commercial attitude to the supplying DCSs. If a DCS cannot supply sufficient milk to cover the variable costs of collection, then after appropriate discussion, the MPU withdraws its collection service. Of 72,700 DCSs formed, MPUs were collecting from only 55,000 in June 1996. This is a very substantial attrition, driven by commercial realities. NDDB has also been active in 66. Those Keynes so unkindly described as "the slaves of long dead philosophers." 67. In all DCS villages, marginal farmers make up the largest group of the pourers followed by small farmers. Marginal and small farmers together constitute more than 70 percent of total pourers, and there is a gradual increase in the representation of these groups (Sampark 1997). Fulton and Bhargava (1994) estimate that 60 percent of members are landless or marginal farmers (para. 7.14). 82 supplying managerial staff, if requested, to MPUs that are in financial difficulties. Three of 173 MPUs have been closed or merged with other MPUs when it became evident that they were not independently financially sustainable. 7.13 Multidimensional Impacts. The multidimensional impact of this single-commodity project has been illustrated above. These effects include, education, employment, income, and contributions to village-level social amenities. Thus, although a project may be quite narrow in its interventions, it may still have a wide range of impacts. 7.14 Getting Government Out of Commerce. Operation Flood had two major thrusts. The first was to expand the geographical range of villages from which milk could be collected and processed. The second was to provide these services in the context of Anand-pattern cooperatives. As shown in Table 6.10, considerable progress has been made in this regard. Those states which have not yet adopted the Anand pattern have Cooperative Federations with the most severe financial problems. 7.15 Market Failure. As discussed above (para. 1.21), the dairy industry is particularly vulnerable to market failure which in part explains the prevalence of cooperatives in the dairy 68 industry worldwide. By maximizing the return to producers rather than the suppliers of capital, cooperatives avoid the tendency to market failure in dairy processing. 7.16 Participatory Organizations. The DCSs are farmer-controlled, village-level organizations. By and large they have functioned well. However, as mentioned in para. 4.4, there have been problems or organizational breakdown based upon caste rivalries within the village, reflecting state-level political conflicts, malfeasance by DCS secretaries (often in apparent complicity with the board of the DCS), and simple incompetence of the secretary. NDDB has supported an aggressive program of Cooperative Development (para. 4.6) designed to inform members of their rights and responsibilities, to instill a sense of pride and "privilege to serve" in DCS board members, and to train secretaries and milk-testers. The observation that villagers have a great capacity to organize themselves does not mean they do not need any help with this activity. 7.17 It is significant that most villages prefer to select, rather than elect, their board members. Obviously, this leaves room for power to be seized by the village elite. However, it also reflects a desire to avoid the divisive effect in the village of electoral campaigns and a desire to avoid the cost of an election. The process of selection (consensus among DCS members) does not mean that less thought has gone into who will be the board members than would occur with a multicandidate election. 7.18 The Importance of Policy. The study has focused on Operation Flood; however, Operation Flood has grown fast but not significantly faster than the dairy industry as a whole. The magic of Lal Bahadur Shastri's decision to endorse Operation Flood was that this same policy protected the whole dairy industry from dumped imports of dairy products and arrested the proliferation of government schemes designed to develop the industry (paras. 6.3-6.7). 68. Even where there is no technical basis for market failure, monopoly marketing organizations are an almost ubiquitous developmental tool with very adverse effects on small farmers and agricultural development generally (Bauer and Paish 1952). 83 Figure 7.1 illustrates the quite modest share of increased dairy production which was procured by Operation Flood. 7.19 This is Not Easy. Four characteristics of Operation Flood mentioned above warn of some difficulties in going from Vision to Action (let alone Results). The first is that NDDB is a financially independent parastatal. Second, this was an indigenous program enjoying top-level political support before the Bank ever indicated its interest. Third, Operation Flood has been very economical in the use of foreign technical assistance, and fourth, this is a project which has been steadily supported for over two decades. 7.20 Financial independence, together with an assertive and talented leadership, meant that NDDB could operate on a wide canvas more decisively and independently than normal parastatals. This is not a combination which can easily be found in the public sector (nor indeed among NGOs which are predominantly small-scale operations). 7.21 Such projects require identification by donors of indigenous initiatives which are producing results, albeit perhaps on a small scale. It argues also for being very careful not to "drown" such local initiative in either excessive resources or excessive technical assistance and for being willing to patiently "grow the project" over a significant time span. It may also argue for more encouragement of south-south transfers of technology and managerial practices. 7.22 Recognition of Success. One of the most curious features of the present impact study is the extent to which the above findings seem to run counter to current views in the Bank. Thus, in recommending against a further extension of the final project, an internal memo said "the key argument [for no extension beyond December 1995] is the project design is not consistent with GOI [government of India] and Bank strategy of leveling the playing field for agroindustrial development between cooperatives and the private entrepreneurs, thus putting the project at odds with the recommendations of our macroeconomic work as well as the recent livestock sector review" (World Bank Reports 1997, Box 3.2).69 7.23 In part, the Bank's new paradigm, enunciated in the livestock sector review, was based on the observation of the restrictions imposed by the MMPO which had been introduced to compensate for NEP's repeal of the Industry Licensing Act. It seems to have been assumed that repeal of MMPO would immediately transfer the Indian dairy industry into a "first-best world" of private investment, easy entry, and competition. The use of private resources was expected to guarantee against overinvestment, and easy entry and competition were expected to guarantee that monopoly positions would be quickly eroded. The present study, on the other hand, interprets the MMPO as an attempt to escape a third-best world of heavy state subsidies to the private (but not cooperative) sector for new processing investments and a moribund public testing and inspection system for milk quality and hygienic operating conditions which does not protect milk quality in the corporate sector (paras. 6.49-6.51). 7.24 The overestimation of project cost and particularly the central government's contribution of its own resources (Annex B, Tables B3 and B 15) seems to have confirmed the view that 69. In fact, no sooner had the Bank conveyed to the government the decision not to extend the project than the Secretary of the Treasury wrote assuring the Bank that project extension was highly consistent with the government of India's strategy for development of the dairy industry. 70 y 1.868x - 20.284 N 60- R2 . 0.9802 50- C ag 40 - .9 Total Flood Procurement y =0.1366x + 18 076 SR2= 0.5898 0 ~20- 10-- 0 - 0)L -10 -20 years Figure 7.1 India: Milk Production, 1951-1994 (linear fit to 1951-1969 and 1971-1994 subsamples) and OF milk procurement 1980-94 85 Operation Flood was primarily a public sector activity and an extremely expensive one at that. Thus the livestock sector review says An analysis of the 1992-93 expenditures indicates that: * Activities which could be taken over by the private sector, such as poultry development, cattle breeding and dairying (especially Operation Flood), receive about 60 percent of total agriculture funds. Funding of such activities should be phased out (para. 6.10). Operation Flood is quoted as having actual government of India expenditures of Rs 805 million in 1992-93 (51 percent of total expenditures on Animal Husbandry and Dairying). This agrees exactly with budget estimate figure for 1992-93 given in Annual Plan 1993-94 for the Department of Animal Husbandry and Dairying (page 28).70 The Annual Plan document also gives the projected source of these expenditures (Table 7.1). Table 7.1: Projected Source of Expenditures for Operation Flood III, 1992-93 (Rs million) Loan to NDDB 620 Assistance from EC 165 Funds-in-Trust 20 Total 805 Source: Annual Plan 1993-94, Statement 11.2. 7.25 Since the loan to NDDB represents on-lending from the Bank, and the EEC assistance is a grant, there is at most 20 (about US$750,000 at the official exchange rate) of the Rs 805 million which is a direct charge on Indian resources.71 7.26 As has been brought out earlier, this study concludes that even though Operation Flood is apparently an "old-style" project dedicated to improving the marketing channels for a single commodity through the nexus of increased incomes, the projects have had a big impact on employment; support of membership NGOs; increased leadership opportunities for women; education, nutrition, and poverty reduction, sometimes in excess of newer-style projects dedicated to just one of these objectives. The study concludes that the impact of the Bank's dairy assistance to India has been large and positive. Operations felt, even after an extensive review of the livestock sector in India, that the project was at odds with Bank strategy. It is unusual for an OED report to feel that the Bank has an unrecognized success story. Recommendations 7.27 The Bank's livestock sector review concluded that Bank policy should be to "level the playing field" between the cooperative and companies72 by (i) eliminating all state intervention in cooperatives, (ii) repealing the MMPO, (iii) establishing a mechanism to monitor the milk 70. Real Expenditure is given as Rs 850 million. 71. Annex B, Table B15 estimates total central government resources over the period 1971-72 to 1996-97 as US$33 million, in 1996 dollars, or Rs 1.15 billion. 72. World Bank Reports 1996, para. 4.52. 86 market to ensure fair competition, and (iv) strengthening public monitoring and enforcement of hygiene standards. This omits reference to the need to remove tax concessions and other incentives provided to private investors. 7.28 Both "ensuring fair competition" and "enforcing hygiene standards" are easier said than done. Effective mechanisms for achieving these desirable objectives would very largely remove the need for the MMPO. As argued earlier (paras. 1.19-1.20), private companies and full Anand- pattern cooperatives can be expected to behave quite differently in a monopsony situation, with cooperatives treating farmers (their owners) more generously than would private companies. Thus, cooperatives should be treated no worse than private companies. 7.29 The problem of ineffective enforcement of hygiene standards has already been discussed (para. 6.49 and Boxes 4.1 and 4.2). The legislative and administrative basis for enforcing hygiene standards is already in place. The problem is the political and bureaucratic will to enforce the Prevention of Food Adulteration Act of 1954. Pending proper enforcement of the act, retention of the MMPO provides another avenue for enforcement of hygiene standards. 7.30 Similarly, pending removal of incentives for dairy plant investment by companies, the retention of the MMPO provides some protection against the waste of public resources from the overexpansion of processing capacity by the private sector (utilizing large amounts of public direct investment and tax concessions). 7.31 The policy recommendations that follow from the analysis in this impact study differ slightly in content, and certainly in sequencing, from current Bank proposals: (i) Elimination of all state intervention in Operation Flood cooperatives. Where this involves the transfer of state-owned assets to the cooperatives, the cooperatives could issue bonds to the state to be retired over a suitable time period. (ii) Elimination of all government incentives for investment in the dairy industry. In the absence of such a change, and as a distinct second best solution, any incentives should be equally applicable to cooperative or corporate investments. (iii) A renewed effort to enforce the provisions of the Prevention of Food Adulteration Act of 1954. This would include inspection and approval of quality control/milk testing procedures of processors, inspection of processing facilities, and testing dudhiya milk supplies for adulteration. This would require increased state expenditures. However, the sums involved are dwarfed by the savings which would follow from the withdrawal of government incentives (Box 6.1). The livestock sector review has suggested that states could experiment with paying private sector organizations to carry out the inspection and enforcement function. 7.32 Pending successful implementation of the above three conditions, repeal of the MMPO would be premature. This is an issue of sequencing of reforms. The MMPO provides a weak bulwark against overinvestment, substandard equipment, and unhygienic operating conditions. 7.33 Continued support for the Indian dairy industry can be justified by the intention to bring about the policies recommended in para. 7.31 and perhaps afterward, depending on whether a tilt towards the cooperatives is felt to be appropriate. 87 7.34 At the very least, there should be no further lending to the dairy sector in states which have not yet adopted the full Anand-pattern for Operation Flood cooperatives or do not treat these cooperatives equally with private corporations. Implementation should be required, ex ante, since NDP and NDPII have shown that formal adoption of the Anand principles is not necessarily followed by their implementation. Lessons 7.35 As has been hinted at earlier, the very success of the expansion of the Indian dairy industry poses in a stark fashion the question of attribution. To what extent was this a reaction to increased demand, itself a function of rising incomes and population increase? To what extent can it be attributed to technology, such as increased straw from the Green Revolution and the adoption of crossbred cattle? What part did the removal of the risk of price declines based on massive imports of milk products affect the willingness of farmers and private processors to invest? What part was played by Operation Flood's demonstration to other investors of the potential viability of well-organized and modem processing of milk collected from a large number of small producers? What role did individual and prescient personalities play? (Neither the visit of the prime minister of India nor visit of the president of the Bank was an entirely unorchestrated event.) 7.36 Despite these attribution problems, a first lesson stands out: Lesson 1: A well conceived investment project in support of an already adopted and appropriate policy change can be associated with results out of all proportion to the investment involved. This occurs in part because it solves the "ownership" problem, in part because it avoids having to create an institution at the same time that the new institution is being expected to implement the project, and in part perhaps for a number of reasons not yet well understood. This lesson intentionally finesses the exact extent of causation involved. 7.37 A second lesson concerns the multidimensional impact of the project: Lesson 2: By raising incomes, an apparently simple single-commodity project can have multiple beneficial effects, including nutrition, education (especially of girls), and job-creation. This poses a question as to the relative efficiency of production projects that aim to reduce poverty directly, as compared to health, nutrition, and education projects designed to remove poverty indirectly or area development projects designed to intervene in many ways simultaneously. 7.38 Though not explicitly a "targeted" project, yet 60 percent of the beneficiaries were marginal or small farmers and landless. Lesson 3: By focusing a project on a predominant activity of the poor, "self- selection" is likely to result in a major portion of the beneficiaries being poor. This provides an alternative to "targeted" projects for reaching the poor. 73. This requirement should also be incorporated in the up-front conditionality of statewide adjustment lending.  89 References Alderman, Harold. 1987. "Cooperative Dairy Development in Karnataka, India: An Assessment." IFPRI Research Report. No. 64. Washington, D.C. Alderman, Harold, George Mergos, and Roger Slade. 1987. "Cooperatives and the Commercialization of Milk Production in India: A Literature Review." IFPRI Working Papers on Commercialization of Agriculture and Nutrition. No. 2. Washington, D.C. Aneja, R.P. 1994. "Dairying in India: A Success Story." Asia-Pacific Association of Agricultural Research Institutions. FAO Regional Office for Asia and the Pacific. Bangkok. Attwood, D.W. 1985. "Sociological Aspect of Dairy Development Projects: Report of a Field Study in Madhya Pradesh, India." Montreal: McGill University. Attwood, D.M., and B.S. Baviskar. 1987. " Why do Some Cooperatives Work but not Others? A Comparative Analysis of Sugar Co-operatives in India." Economic and Political Weekly 22(26):A38-54. Banerjee, A. 1994. "Dairying Systems in India" FAO World Animal Review 79. Bauer, P.T., and F.W. Paish. 1952. "The Reduction of Fluctuations in the Incomes of Primary Producers" The Economic Journal 52:750-80. Doornbos, Martin, and Liana Gertsch. 1994. "Sustainability, Technology and Corporate Interest: Resource Strategies in India's Modem Dairy Sector." Journal ofDevelopment Studies 30:3. Doombos, Martin, and K.N. Nair. 1990. "The State of Indian Dairying: An Overview." In Martin Doornbos and K.N. Nair, eds., Resources, Institutions and Strategies Operation Flood and Indian Dairying. Dutch Studies on Development Alternatives. New Delhi/Newbury Park/London: Sage Publications. Doombos, Martin, Frank van Dorsten, Manoshi Mitra, and Piet Terhal. 1990. Dairy Aid and Development: India's Operation Flood Indo-Dutch Studies on Development Alternatives 3. New Delhi/Newbury Park/London: Sage Publications. George, Shanti. 1991. Development Perspectives for the 1990s. Basingstoke, United Kingdom: Macmillan. Gertsch, Liana. 1990. "The National Dairy Development Board of India and Corporatism: The Politics of Public Policy Making." Working Paper Series No. 71. The Hague: Institute of Social Studies. Gill, G.S., and M. Kaur. 1993. "Study of Marketing Problems in Milk Trade in Punjab." SARAS Journal ofLivestock and Poultry Production 9:93-96. 90 Gill, G.S., P.S. Khatra, N. Singh, and S. Singh. 1995. "Pilot Study on Cost of Production and Marketing of Milk in Two Milksheds on the Punjab State." Department of Economics and Sociology, Punjab Agricultural University. Government of India. Ministry of Agriculture. Department of Animal Husbandry and Dairying. Undated. Annual Plan 1993-94. Government of India. Ministry of Agriculture. Department of Animal Husbandry and Dairying. Undated. Annual Report 1993-94. Gulati, Ashok, and Shashanka Bhide. 1997. "Indian Dairy Policy and Protection, 1975-95." National Council of Applied Economic Research New Delhi. Draft Working Paper No. 2 prepared in conjunction with OED Impact Evaluation Report, The Impact ofDairying Development in India: The Bank's Contribution. Operations Evaluation Department, World Bank, Washington, D.C. Gupta, P.R. 1997. Dairy India. Fifth Edition. New Delhi. Hiremath, B.N., Katar Singh, and George Mergos. 1997. " Operation Flood: A Resurvey of Selected Villages in Madhya Pradesh." Institute of Rural Management Anand. Draft Working Paper No. 4 prepared in conjunction with OED Impact Evaluation Report, The Impact of Dairying Development in India: The Bank's Contribution. Operations Evaluation Department, World Bank, Washington, D.C. Kaur, M., and G.S. Gill. Undated. "An Economic Analysis of Marketing of Milk in Ludhiana District (Punjab)." Source unidentified. Kumar, Nalini. 1997. "Operation Flood: Literature Review and Reconciliation." Draft Working Paper No. I prepared in conjunction with OED Impact Evaluation Report, The Impact of Dairying Development in India: The Bank's Contribution. Operations Evaluation Department, World Bank, Washington, D.C. Mergos, George, and Roger Slade. 1987. "Dairy Development and Milk Cooperatives: The Effect of a Dairy Project in India." World Bank Discussion Paper 15. Washington, D.C. Mishra, S. N., and Rishi K. Sharma. 1990. Livestock Development in India: An Appraisal. Institute of Economic Growth, Studies in Economic Development and Planning, No. 54. New Delhi: Vikas Publishing House Pvt Ltd. National Dairy Development Board. Undated. "Operation Flood: A Progress Report." Anand, India. Sampark. 1997. "Impact Study of Milk Cooperatives at Village Level (Operation Flood III)." Draft Working Paper 3 prepared in conjunction with OED Impact Evaluation Report, The Impact ofDairying Development in India: The Bank's Contribution. Operations Evaluation Department, World Bank, Washington, D.C. Sahni, G.S. 1993. "Development of Dairy Cooperatives in India." Indian Dairyman: Journal of the Indian Dairy Science Association 45(11): 505-508. 91 Shah, Tushaar, Vishwa Ballabh, Pratima B, and Jayesh Talati. Undated. "Institutional Structures for Dairy Development: India's Post-Independence Experience." Draft. Anand, India: Institute of Rural Management. Singh, Katar, and J.C. Acharya. 1986. "The Impact of the Madhya Pradesh Dairy Development Project." Anand, India: Institute of Rural Management. World Bank. 1991. "Gender and Poverty in India: A World Bank Country Study." Washington, D.C. 1995. "Working With NGOs: A Practical Guide to Operational Collaboration between the World Bank and Nongovernmental Organizations." Operations Policy Department. Washington, D.C. 1996a. "The World Bank Participation Sourcebook." Environmental Department Papers 019. Washington D.C. 1996b. "Monitoring Environmental Progress: Expanding the Measure of Wealth." Environment and Sustainable Development Publication. Washington D.C. World Bank Reports 1974a. "Appraisal of Karnataka Dairy Development Project, India." Report No. 43 1a-IN. Washington, D.C. 1974b. "Appraisal of Madhya Pradesh Dairy Development Project in India." Report No. 522a- IN. South Asia Projects Department, General Agricultural Division. Washington, D.C. 1974c. "Appraisal of Rajasthan Dairy Development Project in India." Report No. 523a-IN. South Asia Projects Department, General Agricultural Division. Washington, D.C. 1975. "India Livestock Sector Review." Report No. 785-IN. South Asia Projects Department. Washington, D.C. 1985. "Tanzania: Dairy Development Project (Cr.-580)" OED Report 5747. Washington, D.C. 1978. "India: National Dairy Project: Staff Appraisal Report." Report No. 1964-IN. Washington, D.C. 1987a. "Staff Appraisal Report National Dairy Project II." Report No. 6897-IN. Country Department IV, Asia Region. Washington, D.C. 1987b. "Project Completion Report National Dairy (Cr. 824-IN)." South Asia Projects Department, General Agriculture II Division. Washington, D.C. 1987c. " Project Performance Audit Report: India: Karnataka Dairy Development Project (Credit 482-IN), Rajasthan Dairy Development Project (Credit 521-IN), Madhya Pradesh Dairy Development Project (Credit 522-IN), National Dairy Project (Credit 824-IN)." Operations Evaluation Department Report 6857. Washington, D.C. 92 1996. "India Livestock Sector Review: Enhancing Growth and Development." Report No. 14522-IN. Agriculture and Water Operations Division Country Department II. Washington, D.C. 1997a. "Implementation Completion Report: India Second National Dairy Project (Cr. 1859- IN/Ln. 2893-IN). Report No. 16218. Agriculture and Water Operations Division, Country Department II, South Asia Region. Washington, D.C. 1997b. "Project Performance Audit Report: India: Second National Dairy Development Project (Cr. 1859-IN and Ln 2893-IN)." Forthcoming. Washington, D.C. 1997c. "From Vision to Action in the Rural Sector." Report No. 15560. Washington, D.C. Further Reading Alvares, Claude. 1985. Another Revolution Fails. Delhi: Ajanta Publications. Aneja, R. P. 1991. "Delicensing May Result in Greater Freedom." Indian Dairyman: Journal of the Indian Dairy Science Association 43(10):477-479. Athreya, Geeta. 1986. "CHAD Women's Dairy Cooperative: A Case Study from Tamilnadu." In Marty Chen, Manoshi Mitra., Geeta Athreya, Anila Dholakia, Preeta Law, and Aruna Rao, eds., Indian Women: A Study of their Role in the Dairy Movement. Ghaziabad, Uttar Pradesh, India: Vikas Publishing House Pvt Ltd. Apte, D. P. 1988. "The Role of Co-operative Dairy Schemes in Rural Development in India." In D.W. Attwood and B. S. Baviskar, eds., Who Shares? Co-operatives and Rural Development. Bombay/Calcutta/Madras: Delhi Oxford University Press. Aravindan, M. 1992. "Dairy Science Education." In P.R. Gupta, ed. Dairy India 1992. 4th Annual Edition. Delhi. India. Azad, Nandini. 1985. "Improving Working Conditions for Rural Women Through Creation of Alternative Employment Options: The Case of the Working Women's Forum-Dindugal Dairy Women's Project and Adiram Pattinam Fisherwomen's Project." In Shimwaayi Muntemba, ed., Rural Development and Women: Lessons from the Field, Vol. II. Geneva: International Labour Organization. Baviskar, B.S. 1988. "Dairy Co-operatives and Rural Development in Gujarat." In D.W. Attwood and B. S. Baviskar, eds., Who Shares? Co-operatives and Rural Development. Bombay/Calcutta/Madras: Delhi Oxford University Press. Belavadi, N.V., and S.N. Singh. 1994. "Coop Law vis-1-vis Cooperative Principles." The Cooperator 31(7). Bennett, Lynn. 1992. "Women, Poverty, and Productivity." Economic Development Institute. World Bank: Washington, D.C. 93 Berger, Marguerite. 1987. "Agricultural Extension for Women Farmers in Developing Countries: Institutional Constraints." Quarterly Journal of International Agriculture 26(1):28-45. 1989. "Giving Women Credit: The Strengths and Limitations of Credit as a Tool for Alleviating Poverty" World Development. 17(7):1017-1032. Chatterjee, A.K., and R.M. Acharya. 1992. "Heading for 21st Century." In P.R. Gupta, ed., Dairy India 1992 4th Annual Edition. Delhi. Chatterjee, Meera. 1990. "Indian Women: Their Health and Economic Productivity." World Bank Discussion Papers 109. Washington, D.C. Chen, Marty, and Geeta Athreya. 1986. "BCT Women's Dairy Program: A Case Study from Andhra Pradesh." In Marty Chen, Manoshi Mitra., Geeta Athreya, Anila Dholakia, Preeta Law, and Aruna Rao, eds., Indian Women: A Study of their Role in the Dairy Movement. Ghaziabad, Uttar Pradesh, India: Vikas Publishing house Pvt Ltd. Chen, Marty, and Anila Dholakia. 1986. "SEWA's Women Dairy Cooperatives: A Case Study from Gujarat." In Marty Chen, Manoshi Mitra., Geeta Athreya, Anila Dholakia, Preeta Law, and Aruna Rao, eds., Indian Women: A Study of their Role in the Dairy Movement. Ghaziabad, Uttar Pradesh, India: Vikas Publishing House Pvt Ltd. Chen, Marty, Manoshi Mitra, Preeta Law, Geeta Athreya, Anila Dholakia, and Aruna Rao. 1986. "Recommendations for Policy Planning for women in Dairy Production." In Marty Chen, Manoshi Mitra., Geeta Athreya, Anila Dholakia, Preeta Law, and Aruna Rao, eds., Indian Women: A Study of their Role in the Dairy Movement. Ghaziabad, Uttar Pradesh, India: Vikas Publishing House Pvt Ltd. Das, S.K. 1994. "Mother Dairy Delhi." Indian Dairyman: Journal of the Indian Dairy Science Association 46(1):3-7. Dieckmann, N. 1994. "The Integration of Social and Gender Issues in Smallholder Dairy Production." FAO World Animal Review 79:23-33. Doornbos, Martin, Liana Gertsch, and Piet Terhal. 1991. "Dairy Aid and Development: Current Trends and Long-Term Implications of the Indian Case." In Edward Clay and 0. Stokke, eds., Food Aid Reconsidered: Assessing the Impact of Third World Countries. EADI Book Series, 11. London: Frank Cass. Published in collaboration with The European Association of Development Research and Training Institutes (EADI), Geneva. Doornbos, Martin, Pietervan Stuijvenberg, and Piet Terhal. 1987 "Operation Flood: Impact & Issues." Food Policy 12(4):376-3 83. Dubey, Ajay. 1994. "Cooperative Companies in Liberalized Economies: International Experience and Rationale in Dairy Sector." Working Paper 65. Anand, India: Institute of Rural Management. Franco, F., and P.G. Vijaya, Sherry Chand. 1991. Operation Flood and the Voluntary Sector: A Study of Tribal Dairy Cooperatives in South Gujarat. New Delhi: Indian Social Institute. 94 Fulton, Joan, and Mukesh Bhargava. 1994. "The Results of a Marketing Intervention: Dairy Cooperatives in India." Journal of International Food and Agribusiness Marketing 6(1):33-58. George, J. 1988. "Dairy Development Policy Instruments and Implications for India." Oxford Agrarian Studies. 27:1-28. George, Shanti. 1992. "Generalization in Rural Development: 11 Villages in South Gujarat, India." Journal ofInternational Development 4(4):437-462. 1990. "Operation Flood and Centralized Dairy development." In Martin Doornbos and K.N. Nair, eds., Resources, Institutions and Strategies Operation Flood and Indian Dairying. Dutch Studies on Development Alternatives. New Delhi/Newbury Park/London: Sage Publications. 1988. "Co-operatives and Indian Dairy Policy: More Anand than Pattern.". D.W. Attwood and B.S. Baviskar, eds., Who Shares? Co-operatives and Rural Development. Bombay/Calcutta/Madras: Delhi Oxford University Press. 1987. "Stemming Operation Flood-Towards an Alternative Dairy Policy for India." Economic & Political Weekly September. Government of India. Ministry of Agriculture. Department of Animal Husbandry and Dairying. Annual Report 1992-93. International Cooperative Alliance. 1992. "Cooperative Development Program of the NDDB." Review ofInternational Co-operation 85(1):65-74. Jamal, Shagufta. 1994. Women in Dairy Development. New Delhi: Concept Publishing Company. Jain, M.M. 1986. Growth Pattern of the Dairy Sub-Sector in Rajasthan. Bombay/Nagpur/Delhi: Himalaya Publishing House. Joshi, V.H. 1990. "Operation Flood: Constraints and Potentialities in Saurashtra." In Martin Doornbos and K.N. Nair, eds., Resources, Institutions and Strategies Operation Flood and Indian Dairying. Dutch Studies on Development Alternatives. New Delhi/Newbury Park/London: Sage Publications. Koshy, Abraham, and A.A. Gopalakrishnan. 1991. "Managing Dairy Development-Insights from a Study of Dairy Farmers in Kerala." Indian Dairyman: Journal of the Indian Dairy Science Association 43(10):452-461. Khanna, R.S. 1994. "Indian Dairying: Introspection and Opportunities." Indian Dairyman: Journal of the Indian Dairy Science Association 46(4). Kumar, Ashok, and Kulwant Singh. 1993. "Impact of Milk Cooperatives on the Rural Economy of Rajasthan." Indian Cooperative Review 30(3):228-237. 95 Kurien, V. 1992. "Food Aid for Dairy Development." In P.R. Gupta, ed., Dairy India 1992. 4th Annual Edition. Delhi. Kurup, M.P.G. 1995. "Livestock Sector in India." Draft. An Analysis and Overview, 1994. A compendium of the livestock sector situation analysis reports. A Project of the Government of India in Collaboration with the Swiss Development Cooperation. Law, Preeta. 1986. "Overview." In Marty Chen, Manoshi Mitra., Geeta Athreya, Anila Dholakia, Preeta Law, and Aruna Rao, eds., Indian Women: A Study of their Role in the Dairy Movement. Ghaziabad, Uttar Pradesh, India: Vikas Publishing House Pvt Ltd. Madan Mohan, T.R., and N.P. Singh. 1994. "Buyer-Supplier Relationships in Hierarchical Organizations: A Case Study of Indian Dairy Cooperatives." Working Paper 59. Anand, India: Institute of Rural Management. Mascarenhas, R.C. 1993. "Explaining Success in South Asian Rural Development: The Importance of Routine." Public Administration and Development 13.475-487. 1988. A Strategy for Rural Development: Dairy Cooperatives in India. New Delhi/Newbury Park/London: Sage Publications. Mattigatti, Raveendra, H.G. Shankara Murthy, G.K. Hiremath, S. Mallikarjunappa. 1993. "Resource Productivity in Cow Milk Production-An Impact of Operation Flood Program." Indian Cooperative Review 30(2):145-152. Mayoux, Linda. 1995. "Alternative Vision or Utopian Fantasy?: Cooperation, Empowerment and Women's Cooperative Development in India." Journal of International Development 7211-228. Mitra, Manoshi. 1990. "Anand in Bihar: Replication of Caste and Class Patterns." Martin Doornbos and K.N. Nair, eds., Resources, Institutions and Strategies Operation Flood and Indian Dairying. Dutch Studies on Development Alternatives. New Delhi/Newbury Park/London: Sage Publications. 1986a. "Women and Work in the Livestock Economy: An Introduction." In Marty Chen, Manoshi Mitra., Geeta Athreya, Anila Dholakia, Preeta Law, and Aruna Rao, eds., Indian Women: A Study of their Role in the Dairy Movement. Ghaziabad, Uttar Pradesh, India: Vikas Publishing House Pvt Ltd. 1986b. "The Andhra Pradesh Dairy's Women's Program: A Case Study from Andhra Pradesh." In Marty Chen, Manoshi Mitra., Geeta Athreya, Anila Dholakia, Preeta Law, and Aruna Rao, eds., Indian Women: A Study of their Role in the Dairy Movement. Ghaziabad, Uttar Pradesh, India: Vikas Publishing house Pvt Ltd. Munshi, Kaivan D., and Kirit S. Parikh. 1994. "Milk Supply Behavior in India: Data Integration, Estimation and Implications for Dairy Development. " Journal ofDevelopment Economics 45:201-223. Nair, K.N. 1985. "White Revolution in India: Facts and Issues." Economic and Political Weekly 20:2-29. 96 Parthasarthy, G. 1991. "White Revolution, Dairy Co-operatives and Weaker Sections." Economic and Political Weekly 26(52):al 77-a 183. Patel, Amrita. 1992. "Twenty-five Years of NDDB." In P.R. Gupta, ed. Dairy India 1992. 4th Annual Edition. Delhi. Patel, A.S. 1988. "Co-operative Dairying and Rural Development: A Case Study of Amul." In D.W. Attwood and B.S. Baviskar, eds., Who Shares? Co-operatives and Rural Development. Bombay/Calcutta/Madras: Delhi Oxford University Press. Patel, R.K. 1993. "Present Status and Promise of Dairying in India." Indian Journal of Agricultural Economics 48(1):1-33. Patel, P.R., and H.A. Ghasia. 1993. "SUMUL Swings Balance." Indian Dairyman: Journal of the Indian Dairy Science Association 45(5):185-194. Patel, S. 1990. "The Anand Pattern: A Socio-Historical Analysis of its Origin and Growth" In Martin Doombos and K.N. Nair, eds., Resources, Institutions and Strategies Operation Flood and Indian Dairying. Dutch Studies on Development Alternatives. New Delhi/Newbury Park/London: Sage Publications. Quarterly Economic Report of the Indian Institute of Public Opinion. 1995. "Dairying in India: An Assessment." Volume 151. Rajagopalan, R., ed., 1996. "Rediscovering Co-operation." Three Volumes. Institute of Rural Management Anand, Gujarat, India. Rangnekar, S., P. Vasiani, and D.V. Rangnekar. 1994. "A Study on Women in Dairy Production" World Animal Review. 79(2). Ross, John, and Elizabeth Frankenberg. 1993. "Findings from Two Decades of Family Planning Research." The Population Council, New York. Satish, S., and John Farrington. 1990. "A Research Based NGO in India: The Bharatiya Agro Industries Foundation's Cross Bred Dairy Program." Network Paper 18. Agricultural Administration Research and Extension Network. Overseas Development Institute. Regent's College, Inner Circle, Regent's Park, London. Sen, D., and Rani G. Jhansi. 1990. "Women in Dairying: A Case Study." Journal of Rural Development 9(5):809-83 1. Sharma, Miriam, and Urmila Vanjani. 1993. "When More Means Less: Assessing the Impact of Dairy Development on the Lives and Health of Women in Rural Rajasthan (India)." Social Science Medicine 37(11):1377-1389. Shah, Dilip R. 1992. Dairy Cooperativization An Instrument ofSocial Change. Jaipur/New Delhi: Rawat Publications. 1993. Review Article "Replicability, Dependency and Equity Issues in Operation Flood Projects." Journal ofIndian School ofPolitical Economy 5(2). 97 Shah, Tushaar. 1991. "Management of a Village Dairy Cooperative Fieldnotes: from Laxmipura (Sabarkantha District)." Case Study. Institute of Rural Management Anand, India. 1987. "Structural Effects of India's Dairy Co-operatives and the Technological Transformation of a Traditional Dairy System." Administration and Extension 25(4):199-214. Shekhawat, P.S. 1990. "Operation Flood in Rajasthan in Resources, Institutions and Strategies Operation Flood and Indian Dairying" In Martin Doornbos and K.N. Nair, eds., Resources, Institutions and Strategies Operation Flood and Indian Dairying. Dutch Studies on Development Alternatives. New Delhi/Newbury Park/London: Sage Publications. Sidhu, J.S., and R.S. Sidhu. 1990. "Case Studies of Successful and Unsuccessful Primary Co- operative Service and Milk producers' Co-operative Society in Punjab." Indian Journal ofAgricultural Economics 45(3):367-373. Singh, Katar, and Rakesh Saxena. 1994. "Some Macro-Economic Aspects of India's Livestock Sector: A Situation Analysis." Draft. Institute of Rural Management Anand. Singh, Mahendra. 1994. "Impact of Liberalization on Milk Producers" Indian Dairyman: Journal of the Indian Dairy Science Association 46(4):168-173. Singh, R.K.P. 1996. "Dairy Cooperatives: Organizations of the Poor in Bihar. Rediscovering Co- operation. Co-operatives in the Emerging Context" In R. Rajagopalan, ed., Rediscovering Co-operation, Volume III Institute of Rural Management Anand, India. Sinha, Frances. Undated. "Women and Dairy Cooperatives in UP: The Inside-Outside Dichotomy." Draft. Economic Development Associates contribution to National Commission on Rural Labor Study of Women in Dairying. Somjee, Geeta. 1989. Narrowing the Gender Gap. New York: St. Martin's Press. Somjee, Geeta, and A.H. Somjee. 1989. Reaching out to the Poor: The Unfinished Rural Revolution. London: Macmillan. Srinivasan, Viji. 1993. Indian Women: A Study of their Role in the Handicrafts and Dairying Sectors. New Delhi: Har-Anand Publications. Tandle, M.K., B.R. Athani, and A.G. Kumbar. 1993. "Cattle Cross Breeding in India: An Overview." Asian Livestock 18(12):153-157. Thirunavukkarasu, M., C. Ramasamy, and R. Prabaharan. 1994. "Effects of Operation Flood on Poverty Alleviation." Indian Journal ofDairy Science 47(1):46-50. Tripathi, M.K, K.K. Singhal, R.K. Mishra. 1994. "Livestock Population and Production Growth in India." Indian Dairyman: Journal of the Indian Dairy Science Association 46(6):337- 339. 98 United States Department of Agriculture. 1994. "Estimates of Producer and Consumer Subsidy Equivalents. Government Intervention in Agriculture, 1982-92." Economic Research Service. Statistical Bulletin No. 913. Washington, D.C. United States Department of Agriculture. 1995. "Dairy Yearbook." Economic Research Service Report. Statistical Bulletin Number 924. Washington, D.C. van Dorsten, F. 1990. "The Impact of Amul on the Milk Economy of the Kheda District." In Martin Doornbos and K.N. Nair, eds., Resources, Institutions and Strategies Operation Flood and Indian Dairying. Dutch Studies on Development Alternatives. New Delhi/Newbury Park/London: Sage Publications. Verhagen, M. 1990. "Operation Flood and the Rural Poor." In Martin Doornbos and K.N. Nair, eds., Resources, Institutions and Strategies Operation Flood and Indian Dairying. Dutch Studies on Development Alternatives. New Delhi/Newbury Park/London: Sage Publications. Vijayalakshmi, S., J. Sitaramaswamy, and John De Boer. 1995. "Rationalization of Milk Procurement, Processing and Marketing in Southern India." Agricultural Systems 48(3):297-314. World Trade Organization 1995. "The World Market for Dairy Products." International Dairy Agreement. First Annual Report. Geneva. Annex A 99 Estimation of Economic Impact 1. As shown in Table Al, the nominal value of Bank assistance to Operation Flood under the five projects was US$500 million. When this is compounded at 10 percent, it becomes the equivalent of US$1.9 billion in 1996 dollars. 2. The increased milk production attributable to the faster growth of the Indian dairy industry from 1975 to 1996 is shown in Table A2. This table also shows the total compounded to 1996 at a 10 percent rate of interest. 3. A million metric tons of milk requires about 85,000 tons of skim milk powder (SMP) and 50,000 tons of ghee (butter-oil), or 60,000 tons of butter. In 1996 world prices were about US$2,195 per metric ton for SMP and about US$2,525 per metric ton for ghee. Thus, the cost would be US$310 per ton of milk for raw materials alone. For 1,086 million tons of milk, raw materials would cost US$336.7 billion, indicating that the Bank's assistance to Operation Flood would have had to account only for 0.56 percent of the observed increase in production to break even at a 10 percent rate of interest. (And this takes no account of benefits likely to be experienced after 1996.) 4. Support for Operation Flood was also provided by the World Food Program, European Community, and the government of India. Allowing for the Bank's contribution of at least 20 percent of the total, all contributions would need to account for around 3 percent of the observed increase in milk production to have an economic rate of return (ERR) of 10 percent. Capitalizing the current annual increment of 40 million metric tons (that is, assuming growth reverted to the pre-Operation Flood rate) would add 43 percent to benefits, or allow only 2.1 percent of these augmented benefits, for the investment to yield an ERR of 10 percent. 100 Annex A Table Al: Disbursement Flows and Present Value Calculations for the Five Dairy Projects in India Disbursed (USS) Major Dairy Dairy Madhya Dairy Agriculture National National To/al With Trans Project Pradesh Rajasthan Credit Dairy I Dairy II Dairy ! MulV MUV Disbursement 1996 as 0 Period C0824 C0522 C0521 C0482 C1859 L2893 Total 1990=100 1996=100 Adjusted Values 20 1976 5,46792 175,129 13 180,59705 45 84 0387 466936.35 3,141,314.28 19 1977 825,565.59 439,341 10 452,05366 1,716,960.35 50.35 0.425 4041591.67 24,718,007.06 18 1978 2,473,112.06 2,249,043.31 5,908,422 18 10,630,577.55 5794 0.489 2174553074 120,903,352.90 17 1979 3,982,098 27 5,740,036 58 7,924,522 18 2,086,499.47 19,733,156 50 65.62 0.554 35641172.03 180,147,244.90 16 1980 9,179,626.67 1,183,037.74 2,553,38199 1,988,522 59 14,904,568 99 71.98 0.607 2454139367 112,767,041.00 15 1981 8,329,055 19 1,753,338 86 2,086,986 80 3,801,315.43 15,970,696 28 72.27 0.610 26191323.14 109,407,656.60 14 1982 26,437,505.28 3,303,663.52 2,487,810 48 2,404,187 29 34,633,166 57 71.16 0600 57683008.74 219,0531,12970 13 1983 33,210,676 14 1,115,777 73 4,687,084 19 1,794,778 It 40,808,316.17 69.54 0.587 6955136084 240,110,160.90 12 1984 32,554,624.48 0.00 711,717.77 3,412,302 39 36,678,644.64 68.05 0.574 63881748.17 200,488,291 20 11 1985 30,033,15140 000 000 7,364,757.78 37,397,909 18 68.61 0.579 6460283043 184,319,414.80 10 1986 6,273,262.57 000 000 119,05082 6,392,313.39 80.89 0.683 936601537 24,293,031.74 9 1987 000 000 0.00 000 000 8884 0750 000 0.00 8 1988 000 000 000 000 46,943,174 83 46,943,174 83 9531 0804 5837483035 125,131,633.10 7 1989 000 000 000 000 6,570,546 91 6,570,546 91 94.66 0799 8226718.99 16,031,547.98 6 1990 000 000 000 000 18,079,191 00 000 18,079,191 00 10000 0.844 21427457 17 37,960,047.46 5 1991 0.00 000 000 000 34,331,465 00 000 34,331,465.00 102.23 0863 39802066.24 64,101,625.70 4 1992 000 000 000 000 38,289,047.00 000 38,289,047 00 10664 0900 42554555 99 62,304,125.42 3 1993 000 000 000 000 20,144,029 00 26,855,062 00 46,999,09100 10633 0.897 52387212 13 69,727,379.34 2 1994 000 000 000 000 000 40,879,844 00 40,879,844 00 1021 0.930 439622458 53,194,317.43 I 1995 000 000 (005) (028) 002 21,887,180 00 21,887,179 69 15 18 0972 22521866 10 24,774,052.70 0 1996 0.00 000 000 000 000 27,500,515 00 27,500,515 00 18.52 1000 2750051500 27.500,515 00 150,000,00000 16,400,000.00 23,315,01690 29,331,88944 164,357,45376 17,122,60100 500,526,961.10 1,900,071,889.00 Note- Values are in 1996 US dollars. Source. Disbursement data from OIS. Annex A 101 Table A2: Calculated Milk Production Increase Due to Faster Growth of the Indian Dairy Industry between 1975 and 1996 Differences in Million Year Number of Years Metric Tons of Milk Compounded 1975 21 1.58 11.71 1976 20 3.55 23.90 1977 19 5.52 33.78 1978 18 7.49 41.66 1979 17 9.46 47.83 1980 16 11.43 52.53 1981 15 13.40 55.98 1982 14 15.37 58.37 1983 13 17.34 59.87 1984 12 19.31 60.61 1985 11 21.28 60.72 1986 10 23.25 60.31 1987 9 25.22 59.47 1988 8 27.19 58.29 1989 7 29.16 56.82 1990 6 31.13 55.15 1991 5 33.10 53.31 1992 4 35.07 51.35 1993 3 37.04 49.30 1994 2 39.01 47.20 1995 1 40.98 45.08 1996 0 42.95 42.95 489.85 1,086.17a a. Total compounded amount.  Annex B 103 Estimation of the Cost of Operation Flood Summary This Annex provides a detailed examination of cost components of Operation Flood and the Bank-supported projects and compares the estimates contained in the Bank's Livestock sector review and the impact study. For those not wishing to go into further details, the comparison between the projected and actual cost of the five Bank projects, in nominal and 1996 US dollars, are: Nominal Cost of World Bank Projects (US$ million) Project Projected Actual Karnataka (Credit 482) 72.7 51.6 Rajasthan (Cr. 521) 51.8 41.8 Madhya Pradesh (Cr. 522) 31.2 25.0 NDP (Cr. 824) 293.8 290.7 NDP II (Cr. 1859/ Ln 2893) 674.0 502.0 Total 1,123.5 911.1 Real Cost of World Bank Project (US$ million, 1996 dollars) Project Projected Actual Karnataka (Credit 482) 129.7 92.2 Rajasthan (Cr. 521) 91.9 74.1 Madhya Pradesh (Cr. 522) 56.2 45.0 NDP (Cr. 824) 498.1 493.0 NDP II (Cr. 1859/ Ln 2893) 758.2 584.8 Total 1,534.1 1,289.1 1. World Bank Reports 1996. 104 Annex B The estimates of the cost of Operation Flood, which include the Bank projects but goes well beyond them, in billions of constant 1996 dollars from the Livestock Review and the impact are: Item Review Impact Real Direct Subsidies Central Government 2.36 0.033 State Government (in cash) 0.054 State Government (in kind) 0.799 Commercial Banks 0.006 Farmers 0.001 Food Aid (through consolidated revenue) 0.962 Food Aid (direct to IDC) 1.36 0.184 Total Real Direct Subsidies 3.72 2.039 Bank Loans and Credits 1.34 0.694 Total Loans and Subsidiesa 5.06 2.733 a. Net of some state government assistance. Source: Tables B3, 310, and B15. Earlier Estimates 1. Estimation of the cost of Operation Flood is not easy. The first expenditures on Operation Flood that the impact has identified (food aid in 1971) occurred over 25 years ago, three years prior to the first Bank project (approved in 1974). Before 1976-77 Operation Flood expenditures were not identified as a line item in the budget documents, and indeed it appears that in this early period food aid may have gone directly to IDC, thus bypassing the government's consolidated revenue and expenditure accounts (para. 15). There have been significant changes over these twenty-five years in the presentation of budget data by the Ministry of Finance. This annex lays out the impact's estimates in some detail, both because others may wish to refine or correct these estimates and because the revised estimates are radically lower than the Bank's Livestock Sector Review (World Bank Reports 1996).2 Before 1976-77, payments to IDC and NDDB were not shown as line items in the budget, and it was not until 1979-80 that expenditures on Operation Flood were separated from other payments to IDC and NDDB. For purposes of this exercise we have treated all payments to IDC and NDDB as Operation Flood expenditures. This inflates slightly the estimated cost of the operation, but avoids arbitrary inference. 2. The estimates in the Review provide a good starting point, since they identify the major cost elements, and refinements of these estimates can then be made step-by-step. Estimates of the value of food aid received by NDDB are presented in Table B 1. The Review's estimate of Government Spending and World Bank Loans for the Dairy Sector are given in Table B2. These estimates are then consolidated in Table B3 to give the estimated cost of Operation Flood both in 1990 dollars (US$4.26 billion) as originally reported and in 1996 dollars (US$5.06 billion) for comparison with other estimates. 2. Hereafter, the Review. Annex B 105 Deflation 3. The first adjustment is in the deflator used to go from "current" to "real" dollars. Columns 2-5 of Table B4 have the same headings as Table B 1, but they give the value of food aid received in current dollars (i.e., the cost of the food aid in the year it was received). Column 6 of Table B4 then gives total current dollar value of food aid received year-by-year. Column 7 gives the Indian Gross Domestic Product (GDP) deflator (with 1990 = 100). This measures the rate of inflation in India. The deflator of 20.5 for 1971 means that it is estimated that to buy as much skim milk powder in 1990 as was bought for US$3.74 million in 1971 would require (3.74*100)/20.50 = US$18.23 million in 1990. These are the mechanics of using the Indian GDP deflator, and it can be readily checked that the total value of food aid reported in Column 6 of Table B I is highly correlated with the estimates derived in Column 8 of Table B4. 4. Next is the question of which deflator to use to relate dollar values year-by-year to dollar values in 1990. Movements in dollar prices are measured by the US GDP deflator, or for the dollar price of internationally traded goods, by the Manufactured Goods Unit Value Index (MUV). If we want to know how many dollars we need in 1990 to buy as much skim milk powder as a given number of dollars in 1971, we need to use a deflator of changes in dollar prices. Column 9 of Table B4 gives the MUV deflator (1990 = 100), and Column 10 gives the real (1990 dollar) value of food aid using the MUV deflator. The total value of food aid, expressed in US 1990 dollars, drops from $1.145 billion to $0.89 billion. 5. The same problem occurred with respect to the value of Bank projects. Column 3 of Table B2 gives the value of the projects deflated by the Indian GDP (i.e., as reported in the Review), and Column 4 gives the values in 1990 dollars as recalculated using the MUV index. Using the latter index results in the total value of Bank assistance in 1990 dollars being reduced from $1.13 billion to $0.85 billion. Table B2 also gives those costs in 1996 dollars. Timing 6. A second problem focuses on the timing of disbursements to Operation Flood. In Table B2, the Review has assumed that the plan amounts were disbursed in the mid-year of the plan and similarly that Bank assistance was assumed to be disbursed in the year of approval, as different from the actual disbursement profile over the life of the project. Table B5 reproduces Table B2, with the differences that (i) plan amounts are assumed to be disbursed evenly over the life of the plan (a very minor change), and (ii) the actual disbursement profile is used for Bank projects, and the MUV index is used to get constant dollars. This table also reflects that some amounts of Bank assistance were canceled and never disbursed. With these two changes, the estimate of plan expenditures drops from $1.72 billion3 (Table B2) to $1.58 billion (Table B5), both in 1996 dollars, and the value of Bank loans from $1.34 billion to $0.69 billion. 3. US$1.46 billion in 1990 dollars (Table B2, Column 3) converts to US$1.72 billion in 1996 dollars using the MUV deflator. 4. US$1.134 billion in 1990 dollars (Table B2, Column 3) converts to US$1.34 billion in 1996 dollars using the MUV. 106 Annex B Double Counting 7. Table B3 reflects a fundamental misunderstanding of the Indian planning and budgeting process. In India, as in most other countries, Government planning and budgeting are based on consolidated revenues and expenditures. Consolidated revenue reflects total receipts whether from taxation, parastatal earnings, or foreign grants and loans. Government expenditures are then consolidated expenditures whether financed from domestic or foreign resources. Thus, food aid and Bank assistance are already included in Government expenditures (see, however, para. 15 for the probability that prior to 1976 food aid bypassed this standard budgetary process). Moreover, the government grant listed in Table B3 is derived not from the government accounts, but rather from a side calculation that all Bank loans are supported by a 30 percent grant from government: "Since the loans ($1.13 billion) cover only 70 percent of program costs, the NDDB received an additional grant from the government of about $480 million" (the Review, para. 4.32). While there was some contribution of domestic resources to the earlier projects, the bulk of the additional funding came from food aid and NDDB's own retained resources. Since there is no clear basis for the estimated grant (nor evidence of it as a separate item in the consolidated expenditures), it is dropped from Table B6. 8. The total loans and subsidies (US$1.58 billion) in Table B6 is derived from the estimate of Total Government of India Dairy Expenditures in Table B5. The changes in Table B6 are the result of eliminating double counting. The negative entry for NDDB's contribution in Table B6 may appear counterintuitive at first glance; and indeed, given the rather tenuous data underlying these calculations, it could be due to rounding. However, there is no inherent reason why the combination of disbursements by the Bank and the value of food aid grants should not exceed the value of government-budgeted expenditures. What this means in real terms depends on where the negative is "captured." It might be a residual in consolidated revenue, but is more likely to have been passed on to NDDB as food aid which NDDB sold, retaining the excess as undisbursed EC funds to be utilized for special program approved by the EC. In fact, NDDB's equity in 1995 was about US$553 million (1996 dollars). Some of this represents gains on other activities, but a part of it may be profits made on food aid and not yet disbursed for the benefit of the dairy industry. There is thus some evidence that it is likely that the sum of Bank loans and food aid grants exceeded the cost of the project, since it has not only financed the project, but also resulted in a significant build-up in the equity of NDDB. These initial corrections drop the estimated cost of Operation Flood from US$5.06 billion (1996 dollars) in the Review to US$1.58 billion in Table B6. Some further refinements lead to an "all in" estimate of almost US$3 billion. 9. Table B6 still includes an element of overestimation of costs since the government's dairy expenditures include not only expenditures on Operation Flood, but also expenditures on the rest of the industry (see para. 19 below). (There is also some question as to how food aid was valued for budgetary purposes, which affects the estimate of government expenditures.) Under Counting 10. In addition to the upward bias in the Bank estimates discussed above, there are also elements of undercounting. Table B I shows that from 1971-79 the value of butter-oil is described as "not available," and the total value of food aid does not include butter-oil. When butter-oil in the earlier years is valued at the average real cost for the years when a price is available (US$2,513 per metric ton, 1990 dollars), this adds US$90 million (1996 dollars) to the value of food aid. Annex B 107 Omitted Costs 11. As appraised, the state governments were expected to make substantial contributions to the three state projects, and the central government was to contribute significantly to all projects except NDPII. The projected contributions in nominal dollars (rupees for NDP) taken from the SARs are given in Table B7. 12. Actual project costs, as reported in the PCRs and ICR, deviated substantially from projected financing, as shown in Table B8. Bank contribution is somewhat understated in the PCRs for the first four projects due to using the exchange rate for the first project year. The Revised World Bank contribution uses the exchange rate appropriate to the year of disbursement. 13. There was no mention of the use of food aid to fund the first three projects. However, Table B9 shows that food aid received over the period of the state projects (Rs 4.19 billion) greatly exceeds the total cost of the three state projects (Rs 1.09 billion). An examination of central government revenue accounts shows that from 1976 food aid was included in consolidated revenue so that the government contribution recorded in consolidated expenditure could easily have been financed from food aid. The same cannot be said of state government expenditures and contributions from banks and farmers, which presumably represent additional resources not included in the estimates provided in the Review. Since significant inflation characterized the period of implementation of the three state projects, Table B9 has been restated in constant 1994 rupees in Table BIO. Since we are working here in rupees, the Indian GDP deflator is the appropriate deflator. Unfortunately, it is not available after 1994. Column 2 of Table B 10 is the nominal rupees (given in Table B9) deflated by the Indian GDP deflator to give real 1994 rupees. Column 3 then converted this to 1994 dollars at the official exchange rate of 1:31.37. Column 4 is similar to Column 3, but the rupee expenditures were distributed in accordance with project disbursement profiles and then converted to dollars, year-by-year, at the official exchange rate and deflated by the MUV to give real 1994 dollars. The striking difference between Columns 3 and 4 indicated the extent of real devaluation of the rupee which has taken place over this period. Column 5 has been obtained from Column 4 by using the MUV deflator to give values in 1996 dollars to be comparable with most other estimates. The higher estimate has been used (i) to be consistent with other rupee:dollar conversions and (ii) to avoid the appearance that the impact might be trying to minimize its estimate of the cost of Operation Flood. To see these costs in perspective, Table B7 (SAR estimates) and Table B8 (PCR/ICR actuals) have been recalculated in constant 1996 dollars as Table B 11 and Table B 12, respectively. Farmer and commercial bank contributions totaled $7 million, less than 1 percent of total actual expenditure.5 Data Correction 14. Table B 13 gives the physical quantities and prices of food aid received by NDDB. This table is copied from Annex Table 4.3 of the Review (itself based on information provided by the staff of NDDB) and was used in the construction of Table 4.1 in the Review. For the impact study, NDDB has clarified and corrected some of this data. NDDB explained that the data refers to financial (April-March) years rather than calendar years. In addition, they note that whole milk powder was never received from either the World Food Program (WFP) or the European 5. Note that estimated actual expenditure on the three state projects in Table B12 in 1996 US dollars is US$211.3 as compared to US$211.5 shown in the last column of Table B10. 108 Annex B Community (EC). Quantities of whole milk powder reported for the years 1986 and 1987 were actually for skimmed milk powder (SMP). NDDB further notes that 3,000 metric tons of SMP was received during 1993-94. No food aid or cash assistance from EC was received after 1993- 94. They point out that the volume of butter-oil received in 1975 was 8,210 metric tons instead of 3,210 metric tons. Moreover, no butter-oil was received after 1988. Table B14 reproduces Table B13, with the exception that the above corrections have been made together with the corrections suggested in para. 10 above. Approximations 15. The Review estimate used all dairy expenditures under the fifth, sixth, and seventh plans. In fact, Operation Flood also received funding under the fourth and eighth plans and the two annual plans for 1978-79 and 1979-80. The first year when transfers to NDDB/IDC appear as a line item in the budget is 1976-77; in earlier years Operation Flood expenditures were included with all other dairy expenditures. All other dairy expenditures in the period 1971-1975 comes to Rs 26 million; the value of food aid over this period was Rs 463 million (US$184 million, 1996 dollars). What appears to have happened over this period is that food aid bypassed the standard budget process being delivered directly to IDC. When Operation Flood expenditures are separated from the annual budgets, it appears that total government assistance to Operation Flood (including revenues contributed from food aid and Bank loans) was about US$1.87 billion (1996 dollars).6 This is in general agreement with the US$1.58 billion estimated (using the MU'V deflator and correcting for year of disbursement) by the Review, using expenditures for the dairy industry as a whole but from only three plans. Missing Data 16. Table B8 and Table B12 give the total cost of the five Bank-funded projects. Two types of contributions to Operation Flood are missing from these tables and from the Review. The first is contributions in kind (such as processing plants transferred to cooperative entities, sometimes accompanied by debts) and facilities made available to the cooperatives at no charge or for a peppercorn rent. The second set of missing data referred to is payments made by state governments to federations or MPUs for losses incurred. Unfortunately, it has not proved possible to generate reliable estimates of the magnitude of either of these costs. 17. A hint as to the possible value of transfers of state plants to state federations and MPUs is provided by the PCR of NDP which presents a budget for the profitability of a representative MPU. This budget provides for the transfer of a government plant to the MPU in the first year (1979) (PCR, Table 18.2). The plant is valued at Rs 24 million. This is specified as the book value of the plant. At the 1979 market exchange rate, this would have been US$2.6 million; corrected to 1996 dollars, this would be US$4.7 million. As a very general order of magnitude, if this had applied to all 170 MPUs, this contribution could have been valued at as much as US$0.8 billion in 1996 dollars. Actual value could have exceeded book value, but only selected MPUs will have had plants transferred to them (sometimes with debts attached).7 These are very rough figures, but may serve to help put external assistance into perspective. 6. US$184 million as food aid prior to 1976 plus budget totals US$1.69 billion from 1973-74 on. 7. There is also the question of the value to attach to a loss-making operation. Annex B 109 18. With respect to the second set of costs, not only is the data missing, but even if it were present, there would be a problem of its correct interpretation. If a fully independent MPU loses money and then goes to the state to make up the loss, there is no doubt that that is a subsidy to the MPU. If the state directs the MPU to sell its milk below its break-even price, making up the loss is a reimbursement to the MPU and a subsidy to consumers. As mentioned in the main text, government interference in the cooperative sector takes many forms, including appointment of key staff, insistence on overstaffing, and price setting. This interference is not ubiquitous and varies greatly from state to state. In the event, it has not been possible to generate data on state transfers, and hence, it has not been necessary to attempt to partition these transfers into subsidies and reimbursement for services provided. Had the states not been willing to make these subventions, then either the affected MPUs would have had to close, or the states would have had to allow them to set break-even prices and reduce costs by laying off staff and closing unprofitable milk routes. Cost of General Support 19. In addition to the above costs which apply directly to Operation Flood, governments also provided support to the dairy industry generally, particularly with respect to animal health and Al, but also including research, extension, regulatory and inspection services, and policy development and implementation. It has not been possible to derive the state expenditures for this purpose (this too is missing data), but the value of central government assistance to the dairy industry over the period 1971-72 to 1996-97, expressed in 1996 dollars, is US$1.69 billion' (adding in food aid of US$184 million prior to 1976-77, we get US$1.87 billion). This estimate is obtained from the respective budget documents, converted at the official exchange rate, and 8. This is the sum, in 1996 US dollars, of all budgeted amounts under consolidated revenues and expenditures. As discussed in para. 15, as much as US$184 million of food aid may also have gone directly to IDC. Under later accounting conventions, this too would have been reflected in the consolidated accounts. 110 Annex B deflated by the MUV index. These central government expenditures can by analyzed as follows (in 1996 US$ billion): Total Assistance to Dairy Industry 1.69 of which was Paid to NDDB/IDC 1.44 of which was Ear-marked for Operation Flood 1.26 20. Central government assistance to the dairy industry, net of direct assistance to NDDB/IDC, was thus about US$250 million (1996 dollars) over the period 1971-72 to 1996-97. This assistance would have been directed to the dairy industry countrywide, regardless of whether Operation Flood was active in the area being served. Impact Estimate 1996 21. All of the above corrections lead to revised estimates in terms of 1996 dollars as shown in Table B 15. Note that the difference in the total cost estimates (both in 1996 dollars) between Table B12 (US$1.29 billion) and Table B15 (US$2.73 billion) is that the former refers to Bank projects (and thus excludes the first phase of Operation Flood), while Table B 15 refers to Operation Flood as a whole and includes allowance for US$0.8 billion of state government assistance not included in the estimated Bank-project financing. Total Loans and Subsidies comes from the consolidated expenditure budgets of the central government (US$1.69 billion), plus food aid delivered directly to IDC (US$0.184 billion), plus state government expenditures (reported in the PCRs) of US$0.054 billion, plus inferred value of transferred plants US$0.799 billion, plus contributions from commercial banks US$6 million and from farmers US$1 million, to give US$2.733 billion. To this can be added US$0.25 billion of central government expenditures on the dairy industry not channeled to NDDB/IDC, to give total government assistance to the dairy industry of US$2.983 billion. If we subtract Bank Loans and Credits (US$0.694 billion) from Total Loans and Subsidies (US$2.733 billion), we get Total Direct Subsidies (US$2.039 billion). Subtracting food aid and identified state government assistance (in cash and kind) and commercial bank and farmer contributions, we get a central government contribution of 2.039 - 0.184 - 0.962 - 0.799 - 0.054 - 0.006 - 0.001 = 0.033. This matches almost exactly with the US$0.03 billion reported by the PCRs as a further net central government expenditure (i.e., not financed from food aid). Number of Beneficiaries 22. At project closing (March 1996), there were 9.3 million members of DCSs. Some 72,744 DCSs had been organized, and 55,042 were functional (Table 4.1). NDDB has provided the estimates (Table B16) of direct beneficiaries (i.e., those who supply milk to a DCS at some time during the year) of Operation Flood in 1996. 23. It is important to note that small producers with only one animal are only able to supply milk for about six months per year, and even with two cows there will be significant dry periods when neither animal is in milk. This means that there is a major difference between the number of people pouring in any given day and the number who pour at some time during the year. It is this latter number which is used as an estimate of beneficiaries. 9. The number of people who have benefited from Operation Flood at some point in time would, of course, be larger. Annex B Ill 24. Note also that Table B 16 provides both estimated and actual pourers. Actual pourers refers to data collected in December 1996. Since December is in the flush and estimated pourers exceed actual pourers, it is likely that some of the estimated corrections are slightly conservative. Cost per Beneficiary 25. The necessary data has now been assembled to calculate cost per beneficiary. This is a simple calculation; all that is required is to decide who will be counted as a beneficiary and which cost estimate to use. Table B 17 presents per beneficiary costs on the basis of members who pour at some time of year (6.3 million) or pourers in December 1996 (3.5 million). Three costs are used: total assistance to the industry, assistance to Operation Flood,10 and assistance to Operation Flood less value of imputed plant transfers. 10. As discussed in para. 19, central government funds paid to NDDB/IDC included about US$180 million (1996 dollars) not earmarked for Operation Flood. The high total payments to NDDB/IDC figure has been used as central government assistance to Operation Flood. 112 Annex B Table BI: Food Aid Received by NDDB (1990 US$ million) Skim Milk Whole Milk Year Powder Powder Butter-Oil Butter Total (1) (2) (3) (4) (5) (6) 1971 18.27 0.00 n.a. 0.00 18.27 1972 26.75 0.00 n.a. 0.00 26.75 1973 31.05 0.00 n.a. 0.00 31.05 1974 27.72 0.00 n.a. 0.00 27.72 1975 42.58 0.00 n.a. 0.00 42.58 1976 90.36 0.00 n.a. 0.00 90.36 1977 75.43 0.00 n.a. 0.00 75.43 1978 15.89 0.00 n.a. 0.00 15.89 1979 21.56 0.00 n.a. 0.00 21.56 1980 67.67 0.00 51.36 4.36 123.39 1981 39.36 0.00 45.47 3.65 88.49 1982 136.65 0.00 62.15 16.03 214.82 1983 48.55 0.00 31.96 10.62 91.13 1984 8.23 0.00 1.53 1.32 11.08 1985 49.00 0.00 30.55 9.76 89.32 1986 0.00 12.48 4.76 4.92 22.17 1987 0.00 7.33 0.54 1.04 8.90 1988 42.00 0.00 5.40 9.11 56.51 1989 34.85 0.00 3.69 14.37 52.90 1990 21.36 0.00 0.00 0.00 21.36 1991 0.00 0.00 1.47 0.00 1.47 1992 16.09 0.00 0.00 0.00 16.09 Total 813.37 19.81 238.88 75.18 1,147.23 Source: World Bank Reports 1996, Table 4.1. Annex B 113 Table B2: Government Spending and World Bank Loans for Dairy Sector Development (using Indian GDP Deflator) Millions ofReal Millions ofReal US Dollars US Dollars US Dollars Millions of US (1990) (1990) (1996) Dollars Using Indian GDP Using MUV Using MUV (1) (2) (3) (4) (5) Government of India Dairy Expenditures 5th Plan, 1974-78 60.0 184 130.89 155.68 6th Plan, 1980-85 432.0 749 621.23 738.97 7th Plan, 1985-90 434.0 521 455.36 561.61 Total 926.0 1,455 1207.47 1436.27 World Bank Projects Rajasthan Dairy 27.7 88 68 81.04 Madhya Pradesh Dairy 16.4 52 40 47.99 Karnataka Dairy 30.0 95 74 87.77 National Dairy I 150.0 431 259 307.94 National Dairy II 360.0 468 405 481.99 Total 584.1 1,134 846 1006.73 Source: Columns 1-3 are from World Bank Reports 1996, Table 4.2; Columns 4 and 5 are Column 2 deflated by MUV index. 114 Annex B Table B3: Livestock Sector Review Estimates of Cost of Operation Flood Item 1990 US$ billion 1996 US$ billion Direct Subsidies Government Dairy Expenditures 1.50 1.79 Government Grant to NDDB 0.48 0.57 Food Aid 1.15 1.36 Total Real Direct Subsidies 3.13 3.72 Bank Loans and Credit 1.13 1.34 Total Loans and Subsidies 4.26 5.06 Source: World Bank Reports 1996, para. 4.32. Annex B 115 Table B4: Food Aid Received by NDDB Total Value Total Value Total Real of Skimmed of Whole Total Value Total Value Total Value India GDP Dollars Using MUV Total Real Milk Powder Milk Powder ofButter-Oil of Butter of Food Aid Deflator India GDP Deflator Dollars Year (Current $) (Current $) (Current $) (Current $) (Current $) (1990=100) Deflator (1990=100) Using MUV (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) 1971 3.74 0.00 n.a. 0.00 3.74 20.50 18.23 26.44 14.13 1972 6.09 0.00 n.a. 0.00 6.09 22.70 26.82 28.81 21.13 1973 8.29 0.00 n.a. 0.00 8.29 26.70 31.05 33.37 24.84 1974 8.73 0.00 n.a. 0.00 8.73 31.50 27.72 40.66 21.48 1975 13.13 0.00 n.a. 0.00 13.13 30.80 42.62 45.21 29.03 1976 29.60 0.00 n.a. 0.00 29.60 32.70 90.52 45.84 64.57 1977 25.81 0.00 n.a. 0.00 25.81 34.20 75.47 50.35 51.26 1978 5.54 0.00 n.a. 0.00 5.54 34.80 15.91 57.94 9.56 1979 8.65 0.00 n.a. 0.00 8.65 40.20 21.51 65.62 13.18 1980 30.37 0.00 23.04 1.53 54.94 44.90 122.36 71.98 76.33 1981 19.47 0.00 22.49 2.13 44.08 49.50 89.05 72.27 61.00 1982 72.64 0.00 33.04 8.58 114.26 53.10 215.18 71.16 160.57 1983 27.99 0.00 18.43 5.31 51.73 57.70 89.65 69.54 74.39 1984 5.11 0.00 0.95 1.36 7.41 62.00 11.96 68.05 10.89 1985 32.61 0.00 20.35 6.08 59.04 66.60 88.65 68.61 86.05 1986 0.00 8.85 3.37 3.00 15.23 70.90 21.47 80.89 18.82 1987 0.00 5.64 0.41 1.00 7.05 77.00 9.15 88.84 7.93 1988 34.94 0.00 4.49 7.50 46.92 83.20 56.40 95.31 49.23 1989 31.48 0.00 3.34 12.43 47.25 90.10 52.44 94.66 49.91 1990 21.36 0.00 0.00 0.00 21.36 100.00 21.36 100.00 21.36 1991 0.00 0.00 1.67 0.00 1.67 114.50 1.46 102.23 1.63 1992 20.16 0.00 0.00 0.00 20.16 124.90 16.14 106.64 18.90 Total 405.69 14.49 131.57 48.91 600.66 1,145.11 886.20 Source: Columns 1-5 are from World Bank Reports 1996, Annex Table 4.3, corrected for calculation errors; Columns 6-10 are by calculation; Column 8 gives Total Real Dollars using India GDP Deflator and is highly correlated with Column 6, Table B 1; Column 10 gives Total Real Dollars using MUV Deflator. 116 Annex B Table B5: Government Spending and World Bank Loans for Dairy Sector Development (US$ million) Nominal US Dollars Real 1996 US Dollarsb (1) (2) (3) Government of India Dairy Expenditures' 5th Plan, 1974-78 63.90 160.63 6th Plan, 1980-85 467.20 785.46 7th Plan, 1985-90 447.30 636.00 Total 978.40 1582.09 World Bank Projects Rajasthan Dairy 23.30 41.39 Madhya Pradesh Dairy 16.40 29.60 Karnataka Dairy 29.30 52.82 National Dairy I 150.00 254.32 National Dairy II 281.50 316.70 Total 500.50 694.00 a. Corrected for disbursement profile. b. Deflated by MUV index. c. Includes provision for Technology Mission on Dairy Development. Source: See text, para. 6. Annex B 117 Table B6: Revised Estimates of Cost of Operation Flood (1996 US$ billion) Item Amount Contribution from NDDB (0.16) Real Direct Subsidies Government Dairy Expenditures 0.00 Government Grant to NDDB 0.00 Food Aid 1.05 Total Real Direct Subsidies 0.89 Bank Loans and Credit 0.69 Total Loans and Subsidies 1.58 Source: See text, paras. 7 to 9. 118 Annex B Table B7: Proposed Financing of Bank Projects' Karnataka Rajasthan Madhya Pradesh NDP NDPII (Cr. 482) (Cr. 521) (Cr. 522) (Cr 824) (Cr. 1859/Ln. 2893) Financier (US million) (US$ million) (US$ million) (Rs billion) (USS million) Central Government 8.0 8.4 4.6 1.6 State Government 12.0 9.8 6.3 Banks 8.3 3.5 2.1 Agricultural Refinance Corporation 2.1 1.6 Farmers 5.4 0.3 0.2 World Bank 39.0 27.7 16.4 1.3 360.0 IDC/NDDB 0.2 314.0 Total 72.7 51.8 31.2 3.1 674.0 a. This table is restated in constant 1996 dollars in Table B 1. Notes: 1. Central government contribution for NDP includes Rs 0.9 billion in food aid to be provided by EC. 2. The IDC/NDDB contribution for NDPII includes funds of US$150 million to be generated from commodity aid. Sources: SAR for Cr. 482-IN, p. 14; SAR for Cr. 521-IN, p. 15; SAR for Cr. 522-IN, p. 14; SAR for Cr. 824-IN, p. 19; SAR for Cr. 1859-IN/Ln. 2893-IN, Table 4.2. Annex B 119 Table B8: Financing of Bank Projects According to Bank Completion Reportsa Karnataka Rajasthan Madhya Pradesh NDP NDPII (Cr. 482) (Cr 521) (Cr. 522) (Cr 824) (Cr 1859/Ln 2893) Financier (Rs million) (Rs million) (Rs million) (US$ million) (US$ million) Central Government 65.5 59.9 14.2 State Government 148.5 101.7 329 Banks 3.5 25.8 Farmers 3.4 1.0 1.0 World Bank 244.4 180.7 131.2 154.3c 282.0 Revised World Bankb 290.1 205.2 141.1 150.0 281.0 IDC/NDDB 20.2 221.0 EC (Food Aid)d 120.5 Total 465 1 343.2 205.3 295.0 503 0 Revised Total 510.9 367.9 215.0 290.7 502 0 a. This table is restated in constant 1996 dollars in Table B12. b Revised data on Bank disbursements. c Actual rupees converted at exchange rate of US$1 = Rs 9.88 (PCR, p. 260) d As shown in Table B9, Rs 4.19 billion of food aid was received during the life of the state projects. Note- The IDC/NDDB component of NDPII includes funds (about US$110 million) generated from commodity aid Sources: PCR for Cr. 482-IN, Table 3.1, PCR for Cr. 521-IN, Table 3.3; PCR for Cr 522-IN, Table 3.4; PCR for Cr. 824-IN, Table 5; ICR for Cr 1859-IN/Ln. 2893-IN, Table 8B. 120 Annex B Table B9: Consolidated Actual Financing of State Projects and Dairy Food Aid Received Over the Same Period Financier Rs million Central Government 139.6 State Government 283.1 Banks 29.3 Farmers 5.4 World Banka 636.5 IDC/NDDB Total (for the three state projects)b 1,093.9 Food Aid Received (1976-86) 4,192.3 a. Revised data from Bank disbursements. b. From Table B8. Annex B 121 Table B10: Consolidated Actual Financing of State Projects and Dairy Food Aid Received Over the Same Period in Constant 1994 Rs and Constant 1994 and 1996 US$ 1994 1994 Year-by-Year Year-by-Year Financier Rs million US$ million 1994 US$ million 1996 US$ million (1) (2) (3) (4) (5) Central Government 441.8 14.1 24.9 26.9 State Government 891.3 28.4 50.3 54.2 Banks 100.1 3.2 5.6 6.1 Farmers 16.9 0.5 1.0 1.0 World Banka 2030.3 64.7 114.4 123.3 IDC/NDDB Totalb 3480.3 110.9 196.2 211.5 Food Aid Received (1976-86) 12,752.5 406.5 731.0 787.4 a. Revised data from Bank disbursements. b. For the three state projects. Source. Table B9 in constant 1990 rupees. 122 Annex B Table B11: Proposed Financing of Bank Projects in Constant 1996 US$ millions NDPII Karnataka Rajasthan Madhya Pradesh NDP (Cr 1859/ Financier (Cr. 482) (Cr. 521) (Cr. 522) (Cr. 824) Ln 2893) Total Central Government 14.3 14.9 8.3 257.0 294.5 State Government 21.4 17.4 11.4 50.2 Banks 14.8 6.2 3.8 24.8 Agricultural Refinance Corp. 3 7 2 9 6.6 Farmers 9.6 0.5 04 10.5 World Bank 69.6 49.1 29.5 208.9 405.0 762.1 IDC/NDDB 32.1 353.1 385.3 Total 129.7 91.9 56.2 498.1 758.2 1,534.0 Notes. 1. Central Government contribution for NDP includes US$122 million in food aid. 2. The IDC/NDDB contribution for NDPII includes funds of US$142 million to be generated from commodity aid Source: Table B7 updated to constant 1996 US$ million. Annex B 123 Table B12: Financing of Bank Projects According to Bank Completion Reports in Constant 1996 US$ millions NDPII Karnataka Rajasthan Madhya Pradesh NDP (Cr 1859/ Financier (Cr. 482) (Cr. 521) (Cr 522) (Cr 824) Ln 2893) Total Central Government 11.8 12.1 3.0 26.9 State Government 26.8 20.5 6.9 54.2 Banks 0.6 5.5 61 Farmers 0.6 0.2 02 1.0 World Banka 52.4 41.4 29.5 254.3 316.7 694.3 IDC/NDDB 34.3 117.0 151.3 EC (Food Aid)b 204.4 151.1 355 5 Total 92.2 74.1 45.0 493.0 584.8 1289.1 a. Revised data on Bank disbursements. b. As shown in Table B9, Rs 4.19 billion of food was received during the life of the state projects. Notes: 1 In NDPII, EC food aid calculated on basis of data supplied by NDDB. IDC/NDDB contribution by difference. 2. Total EC food aid received from 1976-96 was $807.5 million (1990 US$) equivalent. Source- Table B8 updated to constant 1996 US$ million. 124 Annex B Table B13: Physical Quantities of Food Aid, Prices, and Total Value as Used by the Livestock Sector Review Skimmed Milk Powder Whole Milk Powder Butter-Oil Butter Total Value of Food Vol Value Vol Value Vol Value Vol Value Aid in Millions of Year 000 mt Price/mt $000 000 mt Price/mt $000 000 mt Price/mt $000 000 mt Price/mt $000 Current Dollars (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) 1971 6.28 595 3,736.60 0.00 0.00 2.29 n.a. n.a. 0 0 3.74 1972 9.98 610 6,087.80 0.00 0.00 2.45 n.a n.a. 0 0 6.09 1973 11.98 692 8,290.16 0.00 0.00 3.71 n a. n.a 0 0 8.29 1974 9.05 965 8,733.25 0.00 0.00 4.28 n.a. n.a. 0 0 8.73 1975 10.56 1,243 13,126.08 0.00 0.00 3.21 n.a. n.a 0 0 13.13 1976 26.86 1,102 29,599.72 0.00 0.00 7.17 n.a. n.a. 0 0 29.60 1977 17.63 1,464 25,810.32 0.00 0.00 1.78 n.a. n.a. 0 0 25.81 1978 7.82 708 5,536.56 0.00 0.00 4.68 n.a. n.a. 0 0 5.54 1979 22.40 386 8,646.40 0.00 0.00 6.04 n.a. n.a. 0 0 8.65 1980 31.15 975 30,371.25 0.00 0.00 12.29 1,875 23,043.75 1 1,525 1,525 54.94 1981 18.81 1,035 19,468.35 0.00 0.00 9.37 2,400 22,488.00 1 2,125 2,125 44.08 1982 77.44 938 72,638.72 0.00 0.00 14.04 2,353 33,036.12 4 2,146 8,584 114.26 1983 37.57 745 27,989.65 0.00 0.00 9.33 1,975 18,426.75 3 1,771 5,313 51.73 1984 7.70 663 5,105.10 0.00 0.00 0.60 1,580 948.00 1 1,360 1,360 7.41 1985 48.97 666 32,614.02 0.00 0.00 15.86 1,283 20,348.38 6 1,013 6,078 59.04 1986 0.00 738 0.00 9.52 930.0 8,853.60 2.81 1,200 3,372.00 3 1,000 3,000 15.23 1987 0.00 840 0.00 5.86 962.5 5,640.25 0.34 1,200 408.00 1 1,000 1,000 7.05 1988 22.00 1,588 34,936.00 0.00 0.00 3.03 1,481 4,487.43 6 1,250 7,500 46.92 1989 17.99 1,750 31,48250 0.00 0.00 1.66 2,013 3,341.58 7 1,775 12,425 47.25 1990 14.99 1,425 21,360.75 0.00 0.00 0.00 1,625 0.00 0 1,375 0 21.36 1991 0.00 1,425 000 0.00 0.00 1.00 1,669 1,669.00 0 1,413 0 1.67 1992 11.99 1,681 20,155.19 0.00 0.00 0.00 1,644 0.00 0 1,363 0 20.16 600.66 Source. Columns 1-13, World Bank Reports 1996, Annex Table 4.3 (see text para. 12); Column 14 calculated from other columns. Annex B 125 Table B14: Physical Quantities of Food Aid, Prices, and Total Value as Revised Skimmed Milk Powder Whole Milk Powder Butter-Oil Butter Total Value of Food Vol Value Vol Value Vol Value Vol Value Aid in Millions of Year 000 mt Price/mt $000 000 mt Price/mt $000 000 mt Price/mt $000 000 mt Price/mt $000 Current Dollars (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) 1971 6.28 595 3,736.60 0 0 2.29 561 1,284 0 0 5.02 1972 9.98 610 6,087.80 0 0 2.45 611 1,497 0 0 7.59 1973 11.98 692 8,290.16 0 0 3.71 708 2,626 0 0 10.92 1974 9.05 965 8,733.25 0 0 4.28 863 3,692 0 0 12.43 1975 10.56 1,243 13,126.10 0 0 8.21 959 7,874 0 0 21.00 1976 26.86 1,102 29,599.70 0 0 7.17 972 6,972 0 0 36.57 1977 17.63 1,464 25,810.30 0 0 1.78 1,068 1,901 0 0 27.71 1978 7.82 708 5,536.56 0 0 4.68 1,229 5,752 0 0 11.29 1979 22.40 386 8,646.40 0 0 6.04 1,392 8,408 0 0 17.05 1980 30.00 975 29,250.00 0 0 12.30 1,875 23,044 1 1,525 1,525 53.82 1981 18.81 1,035 19,468.40 0 0 9.37 2,400 22,488 1 2,125 2,125 44.08 1982 77.44 938 72,638.70 0 0 14.00 2,353 33,036 4 2,146 8,584 114.26 1983 37.57 745 27,989.70 0 0 9.33 1,975 18,427 3 1,771 5,313 51.73 1984 7.70 663 5,105.10 0 0 0 60 1,580 948 1 1,360 1,360 7.41 1985 48.97 666 32,614.00 0 0 15 90 1,283 20,348 6 1,013 6,078 59.04 1986 9 52 738 7,025.76 0 0 2.81 1,200 3,372 3 1,000 3,000 13.40 1987 5 86 840 4,922.40 0 0 0.34 1,200 408 1 1,000 1,000 6.33 1988 2200 1,588 34,936.00 0 0 3.03 1,481 4,487 6 1,250 7,500 46.92 1989 17.99 1,750 31,482.50 0 0 0.00 2,013 0 7 1,775 12,425 43.91 1990 14.99 1,425 21,360.80 0 0 0.00 1,625 0 0 1,375 0 21.36 1991 0.00 1,425 0.00 0 0 0.00 1,669 0 0 1,413 0 0.00 1992 11.99 1,681 20,155.20 0 0 0.00 1,644 0 0 1,363 0 2016 1993 3.00 1,625a 4,875.00 0 0 0.00 0 0 0 4.88 636.87 a. 1993 price for SMP is taken from Gulati and Bhide (1997), Table A5.17. Source See text, para. 11. 126 Annex B Table B15: Impact Estimates of Cost of Operation Flood (1996 US$ billion) Item Amount Real Direct Subsidies Central Government 0.033 State Government 0.054 State Governments (Processing Plant Transfers)a 0.799 Commercial Banks 0.006 Farmers 0.001 Food Aid (Through Consolidated Revenue) 0.962 Food Aid (Direct to IDC) 0.184 Total Real Direct Subsidies 2.039 Bank Loans and Credit 0.694 Total Loans and Subsidies for Operation Flood 2.733 Other Central Government Dairy Expenditures 0.250 Total Government Assistance to Dairy Industryb 2.983 a. Budget-based figure. See para 17. b. Includes food aid and Bank support but is net of some state government assistance. Source See text. Annex B 127 Table B16: Estimates of Beneficiaries (1996) Classification People (million) Total Household Membership 9.3 Less defunct DCSSa 1.4 less duplicated membershipsb 1.6 Total Effective Membership 6.3 (Effective Membership/Total = 68%) Less 50% of 1 animal ownersc 1.3 Less 25% of 2+ animal ownersd Potential Pourers 4.1 Actual Pourers (December 1996)e 3.5 (Actual Pourer/Potential Pourers = 85%) a 17,700 defunct DCSs at, say, 80 members each b. At 20 percent of non-defunct membership, based on the "Society Village Enumeration" (NDDB 1987-88). c. 40 percent of total effective members have one milch animal (NCAER study 1996). Only 50 percent of such producers have an animal in milk at any one time. d. Similarly, 60 percent of effective members have two or more animals, 75 percent of whom can pour milk on any given day. e. Number of people pouring on any given day. Source- See text. 128 Annex B Table B17: Alternative Estimates of Capital Costs of Operation Flood per Beneficiary in Constant 1996 US$ Per Member Per Pourer Total Assistance to Industry 473 851 Assistance to Operation Flood 433 780 Net of Plant Transfers 306 551 Source: See text. Annex C 129 Estimation of Producer and Consumer Surplus 1. The following estimates are offered principally for illustrative purposes. However, they do give general orders of magnitude of producer and consumer surplus in 1996 arising from Operation Flood and associated policy changes on the basis of one interpretation of the data. No pretense is made that this interpretation is unique or even best. The most that is claimed is that it is clear. 2. In Figure 1, Point A represents the quantity of milk produced in 1995 (66.3 million tons) and an average urban wholesale price of Rs 8 per liter. The 1995 constant elasticity demand curve DD' passes through this point with an elasticity of -1.077 (Dairy India 1997, p. 59). Point B represents the quantity of milk (25 million tons, Figure 6.2) which would have been produced in 1995 had the prepolicy change trend continued and the price (Rs 7.13 per liter) at which reconstituted milk, based on imported ingredients, could have been sold. (No endogenous long- term supply function is assumed for either the with or without policy change situation, although, obviously, when these are estimated, they could be inserted without substantial effect on the analysis.) One estimate of the total value of the supply shift associated with the policy change is the dotted area BA'q2ql. Numerically, this is Rs 289.9 billion and corresponds to "without project" scenario that milk consumption would have been kept at the "with project" level by importation of the required ingredients. (The double shaded area BB'AA' represents a government tax forgone of Rs 35.3 billion in this "without project" scenario.) 3. Points A and B' in Figure 2 correspond to Figure 1, as does the demand curve DD'. Point C represents the price (Rs 19.4 per liter) which would have occurred on the assumption of a perfectly inelastic short-run supply curve and no importation. The shaded area (corresponding to Rs 435 billion) is the consumer surplus of the policy change under the "perfectly inelastic short- run supply curve" assumption. 4. Figure 3 assumes a unit elasticity supply function. Sl' is the supply function without Operation Flood, and S2' is the supply function with Operation Flood and associated policy changes. In both cases, we assume self-sufficiency. The quantity Q- is obtained on the basis of the pre-Flood growth rate, and quantity Q+ is the observed production in 1995. Point E is the equilibrium without Operation Flood, and Point A (corresponding also to A in Figure 2) is the equilibrium with Operation Flood. The gain in consumer surplus as a result of going to Operation Flood is the dotted area (about Rs 269 billion). The producer surplus in the without case is Rs 243 billion and is Rs 260 in the with case. There is, thus, a slight gain in producer surplus (Rs 260 - 243 = 17 billion) in going from the without to with Operation Flood cases. These results are summarized in Table Cl. 1. Gulati and Bhide 1997, Table A5.9. 130 Annex C Table Cl: Alternative Estimates of Annual Gross Benefits Under Operation Flood and Policy Change, 1996 Consumer Producer Total Total (Rs billion) (US$ billiona) Import Replacement n.a. n.a. 288.9 9.2 Inelastic Short-run Supply 435.0 0.0 435.0 13.8 Unit Elasticity Short-run 269.0 17.0 252.0 8.0 a. Exchange rate = 31.37. D 10 B1 A B : A' 9B 7 s 1 *' 4 a oe 1 a 2 6- % I % *% No A. E 4 . a 1 3 - %% 3~f I ' - 1 . No I 0 1 1 1 0 10 20 q1 30 40 50 60 q2 70 80 quantity (million metric tons) Figure 1 (Annex C): "With/Without" Benefits of Operaton Flood D 21 195 C 18 16.5 15 13.5 12 10.5 CL 7.5 B3' AsD 4.5- :3 1.5 - 0 I I 0 10 20 q1 30 40 50 60 q2 70 80 quantity (million metric tons) Figure 2 (Annex C): Consumer Surplus with Perfectly Inelastic Short Run Supply Function SI, 13 ' without' producer surplus D 12- E F T1.1 . - - - - - - - -- -- - --- consumer surplus 1 - 0 0 0 10 - o 0 9 00 9 0 8. 5 4 ,o . . . . . . . w.to"'"pr ou e -c nsu ersur lu 3 2S 0 --------- --I - -- ------- --I - 0 10 20 30 40 50 60 70 80 Figure 3 (Annex C): Consumer and Producer Surplus with Unit elasticity supply function  Annex D 135 Estimation of the Impact of Free Trade in Dairy Products 1. This annex considers the impact of a partial exemption from the Indian policy of self- reliance. The exemption covers only urban milk consumption, which is thought to be approximately 30 percent of total milk production. 2. Figure 1 presents data from 1977-78. The left axis gives urban retail milk prices in 1977-78 rupees, and the right axis gives the same prices in constant 1993-94 rupees. Point A is the observed national production and urban retail milk price. Point U is the inferred urban consumption (30 percent of production), and DuDu' is the urban demand function with elasticity -0.535 (Dairy India 1997, p. 58). Qu is the inferred urban consumption under the self-reliant policy. Pw is urban retail price of milk based on imported and reconstituted ingredients. Point F is the intersection of the urban demand function DuDu' and world price Pw. Qf is the corresponding free trade urban consumption, and Qg is the quantity of milk produced at the world price in the unit elasticity of supply case. 3. The gain in urban consumer surplus as a result of going from self-sufficiency to free trade is the area FPwPiU which is approximately Rs 16.33 billion in 1977-78 (Rs 59.64 billion in constant 1993-94 rupees). The loss of producer surplus depends on the assumption with respect to the elasticity of supply. For a completely inelastic supply function, the loss is the rectangle U'PwPiU, which is approximately Rs 13.284 billion in 1977-78 (Rs 48.50 billion in constant 1993-94 rupees); for a unit elasticity supply function, the loss is the trapezoid GPwPiU, which is approximately Rs 9.44 billion in 1977-78 (Rs 34.5 billion in constant 1993-94 rupees). 4. The direct foreign exchange cost for the limited free trade option also depends on the elasticity of supply. In the perfectly inelastic case, the expenditure on imported and recombined milk is the rectangle QuU'FQf. In the unit elasticity case, it is the rectangle QgGFQf. Raw material imports constitute about 89 percent of the cost of urban milk. The direct foreign exchange costs of the increased urban milk consumption would have thus been 0.89 times the above rectangles. Table Dl records these changes in 1993-94 rupees and 1996 dollars. Table D1: Annual Costs and Benefits of Free Trade for Urban Milk, 1977-78 1993-94 Rupees 1996 US$ Cost/Benefit (billion) (billion) Consumer Surplus 59.64 2.10 Supply Elasticity = 0 Producer Surplus (48.50) (1.71) Foreign Exchangea (18.44) (0.65) Supply Elasticity = 1 Producer Surplus (34.50) (1.22) Foreign Exchangea (36.59) (1.30) a. Raw materials only. 5. A similar analysis for 1993-94 is depicted in Figure D2. The left axis gives urban retail prices in 1993-94 rupees. The observed urban price and total production is again Point A. This defines the Indian urban price Pi (Rs 8.80 per liter). Point U is the urban demand at this price, 136 Annex D corresponding to quantity Qu, and DuDu' is the urban demand function (elasticity -0.535) through this point. Pw (Rs 7.66 per liter) is the urban price of milk based on reconstituted imported ingredients, and F is the price and quantity (Qf) for urban consumption which would occur under free trade. 6. Supply is a little more complex. OS+ is the (unit elasticity) supply function which characterized the with-protection and Operation Flood scenario. OS is the supply function shown in Figure DI, which ruled in 1977-78. The question, then, is: By how much (if at all) should we assume that the supply function would shift under free trade? If we assume that technology explains perhaps half the observed shift with the balance explained by market access and other assurances provided by the self-reliance policy and that the shift due to technology would have occurred in any case, then OS- is the new shifted supply function under the free trade option and without Operation Flood. 7. Three scenarios are illustrated in Figure D2 with respect to urban milk pricing and supply: (i) Simple self-reliance with supply function S- and price-quantity equilibrium2 at D; (ii) Operation Flood with supply function S+ and price-quantity equilibrium at U; and (iii) Free trade with supply function S- and price-quantity equilibrium at F. 8. Point G is where the self-sufficiency/free trade supply function S- intersects world price. The corresponding quantity is Qg. 9. Table D2 expresses the consumer and producer surpluses relative to the self-reliance option in 1993-94 rupees. The same data is reported in Table D3 in terms of 1996 dollars. Table D2: Costs and Benefits of Alternative Policies for Urban Milk Supply, 1993-94 (Rs Billion) Operation Flood Free Trade Consumer Surplus 5.66 14.76 Supply Elasticity = 0 Producer Surplus (16.27) (92.99) Foreign Exchangea 0.00 (20.50) Supply Elasticity = 1 Producer Surplus (8.14) (46.30) Foreign Exchangea 0.00 (49.84) a. Raw materials only. 1. A generous assumption, since technology adoption is directly related to the profitability of the technology. A more realistic assessment might be to give a third of the shift to technology alone, a third to protection and Operation Flood alone, and a third to the synergy if both were undertaken together. 2. This, too, is a conservative assumption, since it is to be assumed that the even higher prices than under Operation Flood would have induced further technological adoption. For the skeptics, and to get orders of magnitude, the Operation Flood scenario can also be interpreted as a self-reliance scenario where all the supply shift is due to higher prices, and Operation Flood has no impact. Annex D 137 Table D3: Costs and Benefits of Alternative Policies for Urban Milk Supply, 1996 (US$ Billion) Operation Flood Free Trade Consumer Surplus 0.19 0.52 Supply Elasticity = 0 Producer Surplus (0.57) (3.28) Foreign Exchangea 0.00 (0.73) Supply Elasticity = 1 Producer Surplus (0.29) (1.63) Foreign Exchangea 0.00 (1.77) a. Raw materials only. s 4 Du 3.5 00 o 3 A .9Pi 10.04 7 0> 2.5 0 2 1.5 F Pw 424 -G - - E 1U, u LU Du' 05 o- 0 Q9 5 1 15 20 25 30 Quantity (million metric tons) Figure DI: Estimation of Impact of Free Trade in Dairy Products (1977-78) S Du 16 S+ 14 ... 12 10 U Pu A o G F - Pw -- -- - - - - - - -- - - - - - - - - - - - - 6 Du' E u 4 2 0 - 4u Q 0 10P '20 30 40 SO 60 70 Quantity (million metric tons) Figure D2: Estimation of Impact of Altemnative Policies (1993-94)  Annex E 141 Borrower Comments National Dairy Development Board CHAIRMAN S OFFICE CDB:542/7/73 July 22. 1997 Mr Roger Slade Division Chief Agnculture & Human Development Division Operations Evaluation Department The World Bank 1818 H Street NW Washungton D C 20433 USA Dear Mr Slade. Thank you for your letter dated June 30. 1997 enclosing a copy of the draft report on "The Impact of Dairying Development in India : The Bank's Contribution Impact Evaluation Report" and inviting our comments on it Your letter was received in Anand on July 16th. hence. we have on1% been able to go through it hurriedly in sun. I find it a heartening confirmation of what we feel we ha% e achieved and of our approach which. as you know. has been to place farmers in control of the resources the create. and thus. in control of their own destmes I would agree with the report s conclusion that a maior impediment to fully achievmng all of or -. -as been the reluctance of the state governments to endorse and support Anand Pattern - that is fi, irmer-owned and controlled -- cooperatives I was pleased with the recognition that talk of a level playing field is meaningless unless one takes full account of the subsidies and concessions offered to. and evasions resorted to. by corporate players and traders The report quite effectively argues for retention and enforcement of the MMPO until such time as there is a genumely level playing field I was paruc:s:- v struck. but not surprised bv. the author s comment that in is unusual for him to encounter a succe -rv that the World Bank does not recomse which. he suggests is the case with Operation Flood I co hope that the Bank's Executive Directors wall give due consideration to the report's findings and recommendations 142 Annex E While my colleagues may have some observations on some of the data/figures indicated in the report which will be provided to you separately. I confirm that we have no objection to the publication of this report by the Bank With land regards. Yours smncerely, (V Kurien) 00' cc to Shn Yogesh Chandra Secretary (AH&D) Department of Animal Husbandry & Dairying Mmstrv of Agriculture Krshi Bhavan New Dellu 110 001 Annex E 143 N K Chawla Tf t %qwi 4 Exeutive "eftor MatnlO Dary Development Board ED:NKC:WB:72 C2 July 2, 1997 Fax No.(202) 522-3123 Dr Wilfred Candler Principal Evaluation Officer Agriculture & Human Development Division Operations Evaluation Department The World Bank 1818 H Street, N.W. Washington DC 20433 USA Dear Dr Candler, Many thanks for your E-mail dated 1st July, 1997. I wish to clarify that no EEC cash aid was received after 1993-94. With kind regards, Yours sincerely, N K Chawla MeR-sP, F W . - oM 4 4 t . * 2eri .wo 0.26o240161o 6 4ow ANAND 388001 INDIA * PHONE 40146 AN 44014- * EMAIL rootfeanana.nduoemet * 'L. r26W4nIr-7 AMD 401 144 Annex E ALL- IN-1 NOTE DATE: 01-Aug-1997 08:10am EDT TO: rslade ( rslade@worldbank.org@INTERNET FROM: dnraju, ( dnraju@finance.delhi.nic.in@INTERNET EXT.: SUBJECT: Comments on "The impact of Dairying Development in India: The Bank's Contribution" Dear Mr.Slade, Thank you for your Fax message informing that the final report will be revised in the light of comments to be furnished by us in about a week's time. We are agreeable to your suggestion to publish the gray cover version of the report in the meantime. Regards, D.N.Narasimha Raju. August 1,1997.  0 CD ro cad C-