India - First and Second Agricultural Extension Projects

Docmncnt of The World Bank FOR OFFCIAL USE ONLY RePort No. 13789 PERFORMANCE AUDIT REPORT INDIA MADHYA PRADESH AGRICULTURAL EXTENSION PROJECT - PHASE II (CREDIT 1138-IN) ANDHRA PRADESH AGRICULTURAL EXTENSION PROJECT (CREDIT 1219-IN) DECEMBER 13, 1994 Operations Evaluation Department of This document has a restricted distribution and may be used by recipients only in the performance their official duties. Its contents may not otherwise be disclosed without World Bank authorization.  CURRENCY EQUIVALENTS Name of Currency: Rupee (RPs) Rate of Fxchange: End of Year Appraisal Year (1980-81): USD 1.0 Rs 8.00 Intervening Years (average): = Rs 11.81 Completion Year (1989): = Rs 16.50 The USD credit amounts in this report should be understood to mean USD equivalents. ABBREVIATIONS AND ACRONYMS ADA Assistant Director of Agriculture ADP - Agricultural Development Project AO - Agricultural Officer AEO - Agricultural Extension Officer AP - Andhra Pradesh API - Andhra Pradesh Agricultural Extension Project APAU - Andhra Pradesh Agricultural University BDO - Block Development Officer CAD - Command Area Development CF - Contact Farner DDA - Deputy Director of Agriculture DOA - Department of Agriculture (MP and AP) DOE - Directorate of Fxtension (GOI) DRD - Department of Rural Development (MP and AP) ERR - Economic Rate of Return FAO/CP - Cooperative Program of the Food and Agriculture Organization GCES - General Crop Estimates Survey GOAP - Government of Andhra Pradesh GOI - Government of India GOMP - Government of Madhra Pradesh IAAP - Intensive Agricultural Area Program LADP - Intensive Agricultural District Program ICAR - Indian Council of Agricultural Research ICRISAT - International Crop Research Institute for the Semi-Arid Tropics IRDP - Integrated Rural Development Program IRRI - International Rice Research Institute JDA - Joint Director of Agriculture JNKVV Jawaharlal Nehru Krishi Vishwa Vidyalaya (SAU-MP) KP - Karshak Parashatel (AP mandel farmers committee) LRD Land Records Department (AP) M&E - Monitoring and Evaluation MDO - Mandel Development Officer MP - Madhya Pradesh MPI - Madhya Pradesh Agricultural Extension and Research Project MPII - Madhya Pradesh Agricultural Extension Project NAEPI - National Agricultural Extension Project - Phase I OED - Operations Evaluation Department OF - Other Farmer PAR - Project Performance Audit Report PCR - Project Completion Report SAR - Staff Appraisal Report SAU - State Agricultural University SMS - Subject Matter Spec:alist T&V - Training and Visit Extension System VDO - Village Development Officer (AP's VLW) VEO Village Extension Officer (AP's VEW) VEW - Village Extension Worker VLW - Village Level Worker FISCAL YEAR Government of India: April I - March 31 Governments of MP and AP: April I - March 31 WEIGHTS AND MEASURES Metric System BORROWFR Government of India  FOR OFFICIALUSE ONLY THEWORLD BANK Washington, D.C.20433 Office of Director-General U.S.A. Operations Evaluation December 13, 1994 MEMORANDUMTO THE EXECUTIVE DIRECTORSAND THE PRESIDENT SUBJECT: Performance Audit Report on INDIA MadhyaPradesh AgriculturalExtensionProject-Phase Andhra Pradesh AgriculturalExtensionProject II (Credit 1138-IN) (Credit 1219-IN) Attached is the PerformanceAudit ExtensionProject-Phase II (Credit 1138-IN) Report on INDIA-Madhya PradeshAgricultural and Andhra Pradesh AgriculturalExtension Project (Credit 1219-IN) prepared by the OperationsEvaluationDepartment. The first project is rated satisfactory explainedby the weakercommitmentof the Andhra and the second unsatisfactory,the difference institutionaland staffingplans essential to the Pradesh State Governmentto implementingthe training and visit extensionsystem. Sustainability project results is also rated more likely in Madhya of carrying Pradesh. The PAR assesses the problems of out effectiveextensioncommonto both operations in India and the world. and to T&V extensionprojectselsewhere Attachment This document has a restricteddistributionand maybe used by reciplents their officialdutles. Its contentsmay not otherwlse only In the performance of be disclosedwithoutWorldBankauthorIzation.  FOR OFFICIALUSE ONLY PERFORMANCE AUDITREPORT INDIA MADHYAPRADESHAGRICULTURALEXTENSIONPROJECT- PHASE II (Credit 1138-IN) ANDHRAPRADESH AGRICULTURALEXTENSIONPROJECT(Credit 1219-IN) TABLEOF CONTENTS Page No. Preface ............................................................ i Basic Data Sheets ............... ..................................... iii Evaluation Summary .................................................. vii I. BACKGROUND A. Introduction .................................................. 1 B. Project Preparation ............................................. 2 C. Project Objectives and Description ................................. 5 II. IMPLEMENTATION EXPERIENCE A. Madhya Pradesh ............................................... 7 B. Andhra Pradesh ............................................... 9 III. PROJECT OUTCOME A. General Comments ............................................. 12 B. Madhya Pradesh ............................................... 14 C. Andhra Pradesh ............................................... 15 IV. MONITORING AND EVALUATION ................................ 19 V. FINDINGS AND LESSONS A. Findings . .................................................... 26 B. Lessons . ...................................................... 35 Annex Government Comments on the Draft PAR .............................. 37 Tables 1. Project Costs and Bank Commitments ............................... 6 2. Project Parameters .............................................. 6 3. M&E Staff Strength ............................................. 20 Boxes 1. Seeds in Sehore, Madhya Pradesh .................................. 16 2. Rice in Rangareddy, Andhra Pradesh ................................ 17 3. Cotton in Guntur, Andhra Pradesh ................................. 18 IBRD Maps 15427R, 16103R This report was prepared by Edward B. Rice (Task Manager), who audited the projects in October 1993. Silvana Valle provided the administrativesupport. Thisdocument hasa restricted distribution and maybe usedbyrecpients onlyintheperformance oftheir duties.Its contents official maynototherwise bedisclosed World without Bankauthorization.  PERFORMANCEAUDIT REPORT INDIA MADHYAPRADESHAGRICULTURALEXTENSIONPROJECT- PHASE II (Credit 1138-IN) ANDHRAPRADESH AGRICULTURALEXTENSION PROJECT (Credit 1219-IN) PREFACE This is a Performance Audit Report (PAR) on the India Madhya Pradesh and Andhra Pradesh Agricultural Extension Projects, involvingIDA Credits in the amounts of USD 36.9 million and USD 6.0 million respectively. The objectives were identical - to achieve early and sustained increases in agricultural production through the reorganization and strengthening of the state extension service. The Credits were approved on April 23, 1981 and March 30, 1982. They were closed, fully disbursed, on June 30, 1989 and March 31, 1989, after two (MP) and one (AP) extensions of the Closing Date. The PAR is based on the Project Completion Reports (PCR) prepared by the Resident Staff in India of the South Asia Regional Office and submitted to the Board on December 14, 1990 and September 24, 1990, the Staff Appraisal Reports, the President's Reports, the Credit documents, a study of project files, and discussion with Bank staff. An OED mission visited India in October 1993 and discussed the effectiveness of the Bank's assistance and project execution with the Government of India, the two state Departments of Agriculture (DOA) and other relevant agencies. The DOAs' kind cooperation and valuable assistance in the preparation of this report is gratefully acknowledged. The PCRs provide an adequate account of the operation of the two projects and the performance of the Bank and project executing authorities. The PAR summarizes these sections. The PAR departs from the PCRs in its treatment of the effects and opportunity costs of the projects, and of the model of the extension delivery system that defined the project experience. Following standard OED procedures, copies of the draft were sent to Government for comment. The response from the Directorate of Extension was taken into account in the text and presented in the Annex.  - iii - PERFORMANCE AUDIT REPORT INDIA MADHYA PRADESH AGRICULTURAL EXTENSION PROJECT - PHASE II (Credit 1138-IN) BASIC DATA SHEET KEYPROJECT DATA Appraisal Actual or Actualas % of* Expectation CurrentEstimate AppraisalEstimate Total ProjectCosts (USSm) 60.0 52.7 88 LoanAmount (US$m) 37.0 36.9 100 EconomicRateof Retum (%)* >50 >50 InstitutionalPerformance Partial * Thresholdvalues CUMULATIVE ESTIMATED AND ACTUALDISBURSEMENTS 81/82 82/83 83/84 84/85 85/86 86/87 87/88 88/89 89/90 Total AppraisalEstimate(USSm) 0.5 3.5 9.5 19.5 32.5 37.0 37.0 Actual (US$m) 0.0 0.9 1.9 7.1 12.5 20.4 27.0 33.0 36.9 36.9 Actualas % of Appraisal(%) 0 26 20 36 38 55 73 89 100 100 Dateof Final Disbursement: January18, 1990 PROJECTDATES Appraisal Actual Identiflcation/preparation - 1980 Appraisal 10/80 Board Date - 04/23/81 Signing Date - 05/07/81 Effectiveness 08/05/81 07/22/81 Completion 06/30/86 06/30/89 ClosingDate 06/30/87 06/30/89  - iv - STAFF INPUTS (staff weeks) FY81 FY82 FY83 FY84 FY85 FY86 FY87 FY88 FY89 FY90 Total Pre-Appraisal 5.6 5.6 Appraisal 11.2 1.0 12.2 Negotiation 2.5 0.7 3.2 Supervision 2.4 9.5 11.8 11.6 6.9 4.4 6.8 4.1 3.0 60.5 Completion 0.9 0.5 1.4 Totals 19.3 4.1 9.5 11.8 12.5 6.9 4.4 6.8 4.1 3.5 82.9 MISSION DATA Month/ By No. of Days Specializatlon Performance Type of Year Persons in Represented Rating Problem Field Preparation GOMP/RSI Appraisal IDA Supervision 1 07/82 1 5 Ext 2 M,F 2 02/83 4 6 Ext/Ext/Res/M&E 2 M 3 08/83 2 5 Agr/Ext 2 M,F 4 03/84 2 5 Agr/Ext 2 M,F 5 08/84 2 5 Ext/Res 2 M,F 6 01/85 2 9 Econ/Ext 2 M,F 7 06/85 2 7 Agr/Agr 3 M,F 8 12/85 1 5 Agr Implementation Assistance 9 01/86 2 5 Agr/Agr 2 M,F 10 09,10/86 2 5 Agr/Agr 2 F 11 03,04/86 2 7 Agr/Agr 2 F,M 12 10,11/87 3 9 Agr/Agr/Econ 2 F,M 13 04/88 3 5 Agr/Agr/Ext 2 F,M 14 10/88 3 5 Agr/Res/Agr 2 F,M 15 05/89 3 6 Agr/Res/Agr 2 F,M 16 10/89 3 5 Agr/Res/Soc 2 F Reporting Completion IDA Codes Used Ratinq Types of Problems Ext = Extension 2 = Moderate problems M = Managementincluding Agr = Agriculture 3 = Major problems which are supervision and M&E = Monitoring and Evaluation addressed adequately. coordination. Econ = Economist F = Financial Including Res = Research budgetary support and reporting. OTHER PROJECT DATA Executing Agency State of Madhya Pradesh Follow-On Project NAEP-1  PERFORMANCE AUDIT REPORT INDIA ANDHRA PRADESH AGRICULTURAL EXTENSION PROJECT (Credit 1219-IN) BASIC DATA SHEET DATA KEYPROJECT Appraisal Actualor Actualas % of Expectation Current Estimate AppraisalEstimate Total ProjectCosts (US$m) 9.7 10.2 117 LoanAmount (US$m) 6.0 6.4 106* EconomicRateof Return(%)** >50 >50 Performance Institutional Partial * Dueto variationin $/SDRexchangerate. ** Threshold values ESTIMATED CUMULATIVE AND ACTUALDISBURSEMENTS 81/82 82/83 83/84 84/85 85/86 86/87 87/88 88/89 89/90 Total AppraisalEstimate(US$m) 0.2 1.0 2.1 3.6 5.1 6.0 6.0 Actual(US$ m) 0.0 0.4 0.7 1.0 1.7 2.5 3.9 4.5 6.4 6.4 Actualas % of Appraisal(%) 0 40 33 28 33 42 65 75 107 107 Dateof Final Disbursement: October3, 1989 DATES PROJECT Appraisal Actual Identificationfpreparation - 1980/81 Appraisal - 05/81 Board Date - 03/30/82 Signing Date - 05/05/82 Effectiveness 08/04/82 07/27/82 Completion 03/31/87 03/31/89 Closing Date 03/31/88 03/31/89  - vi - STAFFINPUTS (staff weeks) FY81 FY82 FY83 FY84 FY85 FY86 FY87 FY88 FY89 FY90 Total Pre-Appralsal 1.2 0.5 1.7 Appraisal 0.7 11.1 11.8 Negotiation 4.2 4.2 Supervislon 1.5 11.6 13.8 11.4 20.4 13.3 13.7 3.0 88.7 Completion 5.7 0.2 1.3 0.7 4.5 12.4 Totals 1.9 23.0 11.8 13.8 12.7 20.4 13.3 13.7 3.7 4.5 118.8 MISSIONDATA Month/ By No. of Daysin Speclalizations Performance Type of Year Persons Field Represented Rating Problem Preparation GOAP/RSI 01/81 IDA 2 3 Ext 02/81 1 1 Econ Appraisal 05/81 2 2 Ext/Econ Board to 08/81 2 4 Ext Supervision 02/82 1 2 Ext 06/82 2 5 Ext Supervislon1 09/82 1 4 Ext 1 2 03/83 3 5 ExtVAgr 2 M 3 09/83 3 4 Ext/Agr/M&E 2 P/M 4 02/84 3 4 Ext/Agr/FA 3 P/M 5 07/84 3 5 Ext 3 P/M 6 12/84 2 5 Ext 3 P/M 7 07/85 2 5 Econ/Ext 3 P/M 8 01/86 4 12 Econ/Ext/Agr 3 M 9 06/86 2 5 EcorVExt 3 M 10 01/87 3 6 Econ/Ext 3 M/F 11 05/87 3 5 Ext/Econ 3 M 12 12/87 3 5 Ext 3 M 13 05/88 2 5 Ext 3 M 14 11/88 2 6 Ext/Res 2 M Completion IDA Codes Used Ratina Tvye of Problem Ext = Extension 1 = Problemfree or minor problem M = Management Agr = Agriculture 2 = Moderateproblems F - Financial M&E = Monitorlngand Evaluation 3 = Majorproblems P = Political FA = FinancialAnalyst Econ = Economist Res = Researrh OTHERPROJECTDATA ExecutingAgency State of Andhra Pradesh Follow-OnProject None  - vii - PERFORMANCEAUDIT REPORT INDIA MADHYAPRADESH AGRICULTURALEXTENSION PROJECT - PHASE II (Credit 1138-IN) ANDHRA PRADESH AGRICULTURALEXTENSION PROJECT (Credit 1219-IN) EVALUATIONSUMMARY Introduction versial. AP was one of the few states to raise serious objections to importing T&V. The 1. Bank support for development of the Department of Agriculture (DOA) argued that training and visit(T&V) agricultural extension by limiting new staff recruitment, costs could system in India began in 1975 as a component be kept lowwithout injury to the T&V system. of irrigation projects in three states, including The Bank was concerned that at the low Madhya Pradesh (MP) and Andhra Pradesh incremental levels proposed by the AP (AP). The Bank then switchedto self-standing government it would be unable to support an state-wide extension projects, and later, to effective system. The final project cost of multi-state ("national") extension projects, USD 9.7 million, and the supporting Bank bringing in the rest of the 17 major Indian Credit of USD 6.0 million, were the lowest in states. This audit deals with the last two the Indian T&V program. These figures for single-stateT&V extension projects, approved AP can be compared with those for MPII: in 1981for Madhya Pradesh II and in 1982for project costs of USD 60.0 million and the IDA Andhra Pradesh. The two projects were Credit of USD 37.0 million for a targeted chosen for audit as part of a work program for farming population 70 percent the size of that an OED study focussing on monitoring and in AP. evaluation (M&E). The Indian T&V program has alwayshad high M&E content. Project Objectivesand Description 2. The T&V project format was easily 3. The objectives in the two states were blueprinted, and the preparation exercise for identical: "to achieve early and sustained both states followed a pattern common to a increases in agricultural production (in all group of seven states that were moving districts) through the reorganization and together in the period 1978-80 toward strengthening of the extension service of appraisal. The delays that accompanied the DOA". preparation and appraisal concerned the basic building blocks of T&V itself. Staffing - 4. The projects would establish cadres of especiallythe means to acquire the large mass Village Extension Workers (VEW), of dedicated field workers that constitutes the responsible for eight groups of about 100 farm core of the T&V system - was the most families each. The rest of the system design important issue. The question was whether followed the blueprint -fortnightly messages, and when to transfer into the "reorganized" ten contact farmers selected from each group, extension system a substantial number of the daily visit schedules, intensive supervision, Village Level Workers (VLW) from the backstopping by subject matter specialists Departments of Rural Development. (SMS), and a single line of command through Processingof the AP T&V project was contro- to extension headquarters. There was no  -viii - discussion in the project description in the However there are frequent references in the Staff Appraisal Report (SAR) about the files to severe constraints on staffing at characteristics of the technical programs that supervisory levels, in the size and quality of would be carried to the target populations. It the cadre of subject matter specialists, and in was taken for granted that some new physical support from the research service. technologies were everywhere available and Supervision also complained that advances in others would be generated by the closely- improved technologies were not keeping up linked research services. MPII would cover with the spread of the T&V field force - that the 30 districts left over from MPI - out of 45 "new technologies were not being developed in the state. API would cover all 22 districts fast enough'. Officers from both the research in that state. As just shown, incremental and extension establishments told OED that resources approved for T&V in MP were six the linkage looked better in the reports than times greater than those approved for AP. on the ground: researchers said that attendance by the supervisors at training 5. M&E cells were to be established in sessions was declining and extensionists said the DOAs in both states, and operate along their source for new, monthly lessons was lines described in a manual on M&E drying up. Bank Staff were supervising methodology issued in 1981 by the Bank for research as well as extension projects and were Indian T&V. The MP staffing pattern for the more critical of the first. cell included four evaluation/analystpositions, the AP staffing pattern included none. The 8. The research connection worked better difference proved immaterial, because MP in AP. It is centered at the Andhra Pradesh never filled the evaluation posts. Agricultural University (APAU), located outside Hyderabad. APAU and its regional ImplementationExperience research stations constitutes the largest state research network in India, and was responsible 6. Project implementation was difficultin for developing the training program for T&V both states. But, whereas 15 of 16 Bank staff. But a reading of the AP supervision supervision missions to MP rated that project reports and PCR show that there were no an overall 2 (moderate problems), 10 of 14 other consistentlybright dimensionsto project missions to AP rated that project 3 (major implementation. Overriding all issues was the problems). The main problem in MP was state government's failure to complete the budget shortfalls throughout the first five agreed staffingpattern. The text discussesfive years. The main problem in AP was political, separate factors that undermined the project and lack of government commitment to plan, including the failure to transfer the support the project. By late 1987the situation VLWs and the state government's decision to had improved in both states. create another layer of administration (the mandel) and subject T&V staff to its authority. 7. In addition to the poor funding The cumulative effect of all five factors situation, the supervision reports for MP show overwhelmed the administrationof the project. most concern for the technical capability of The low budget in AP compounded these the T&V program. In those districts where difficulties. At the time of the audit mission, the limited funds were deployed at levels although the staffing complement had been sufficient to support adequate staffing of the retained at the same reduced strength as when VEW posts, supervision usually reported that the project wascompleted in 1989,the amount farmer contacts had been established and of funds available for expenditure headings useful technologies were being transferred. other than salaries and allowanceswas too low to support effective extension of any type, let of the SMS, the competence of the VEWs, alone one that is based on high mobility, and the stream of appropriate technology. intensive supervision and frequent training. The other important consideration is that in 11. For Madhya Pradesh, the PCR the early 1980s the DOA's existing force of mentions only one specific example of field assistants and sub-assistantswere in their extension impact, but that is a big one: the majority undertrained and unprepared to increase in area planted to soybean on black provide a service which was responsive to soils previously fallowed in the wet season, farmers' needs. Finally, and just prior to the along with the subsequent introduction of audit mission, the DOA eliminated all SMS short duration soya and pigeon pea allowing positions at the subdistrict level for the for sequential cropping and the intensification purpose of restoring a measure of supervision. of land use. The majority opinion is that the The audit missionwas told that the last action extension service, and most particularly the had effectively ended T&V in western AP. T&V farmer contacts dedicated to technical transfer, accelerated a process already Results underway. There are no comparable success stories discussed in Bank files about other 9. The projects were successful in greatly farming regions of the state, although many extending the process - initiated by pilot technical breakthroughs of smaller dimension operations in both states in the late 1970s- of can be sited. Since it takes only small creating a public service, at the village level, incremental improvements in crop incomes to dedicated to the extension of agricultural reach the SAR's rate of return threshold technologiesthrough farmer training. Previous targets (as described in the text), it is communitydevelopment programs carried out conceivable that the shared success in the by VLW and DOA field staff in these and central west of the state could by itself justify other states had mixed technical transfer with the expenditures under the project throughout the supply of subsidized inputs and non- the state. Incremental impacts elsewhere agricultural activities such as rural roads and would add to that effect. Given this child-care. The Indian Government recognizes production result, the commitment of the state the technological thrust to be the single most government to the T&V program, and the important contribution of the overall T&V apparent stabilityof the field force to date, the program, and MP and AP are part of that audit rates this project satisfactory. success story. 12. Andhra Pradesh is divided in two main 10. The audit examined the scant and regions, the western drylands and the wetter subjective evidence of technical progress at east, includingthe large, irrigated paddy deltas. farm level attributable to extension delivery Despite a concentration of research activityin systems. The M&E survey reports are not of the drylands,state and Bank reports admit that any help here. The PCRs do not add much. there has been very little progress with Their claims about "the efficiency of the improved technologies: no significant, widely technology delivery competence" and "no accepted, new packages were available to give doubt that the project has had very significant a boost to the extension activities. The story impact" contradict other, disturbing in the rice-lands of the east is different, commentary in the files and the PCRs although here too the relative importance or themselves about the continued weakness in impact of the extension service has been supervision by T&V management, the quality questioned. There have been rice varietal releases, and associated pesticide treatments,  - x - which have spread rapidly through the deltas. exercise was seen by DOA as an obligation to It is certain that extension must have played the Bank, and secondarilyto the Ministry, and some role. But in this case - unlike the not as a source of lessons for Lhe state soybean story in Madhya Pradesh - there is a . program itself. strong sense even among the T&V practitioners that the farmers know as much as Sustainability or more about the new packages, and especially about fitting them to local 14. Direct Bank support for T&V ended in conditions, than the average extension agent, 1989 in Andhra Pradesh and in 1992 in and that the impact of most of the VEWs' has Madhya Pradesh. Since the T&V system was been marginal. The audit sees too little imported to India largely in response to Bank evidence to support the satisfactory rating for urgings, the program could have decayed with the AP project - established at the time of the the end of direct Bank involvement. The audit Project Completion Report - on either missioninspezted budget and expenditure data institutional or production impact. The in both states, and found that the states have commitment factor appears to have been one maintained support for the DOA extension of the decisive missing elements. This force at budget and expenditure levels only conclusion refers to the AP state government, marginally below where they were at project and holds despite the remarkable efforts of completion. This is more impressive for AP, APAU to associate itself with the T&V where Bank support ended five years ago. For extension system and support its front line MP, where a recently completed third phase field staff. operation followed the audited project, the Bank withdrawal is too recent to provide a 13. Lack of government commitment to clear test of sustainability,although FAO/CP, M&E in both states helps explains the poor in its PCR on the first of the national progress of this component, though the Bank extension projects (NAEPI), found early signs shares some of the blame. M&E cells were of budget pressure. Nevertheless, the audit established within the extension divisions of mission also confirmed that talk about the DOAs. The M&E survey program overhauling the programs meant just that, and consisted of two surveys for each of the two did not suggest they would be dismantled. In main cropping seasons and was common to all that respect, the projects can be said to have 17 states. MP and AP rank below average in created a sustainable institutionai infra- all respects: maintaining the survey reporting structure. It appears the state governments schedule, analyzing the results, and carrying are convinced that a public sector extension out special studies. The raw data was not service based on farm visits is necessary for being used as the beginning of an analytical their small farmers. The fact that the T&V exercise, but as the final output. The T&V structure fits the hierarchical structure of the and M&E management teams in MP and AP DOA and civil service is another advantage. show themselves by their inattention to have Within that panorama, the decline of the T&V been indifferent to the possible uses of the structure in western AP must be seen as a surveys,or of their cells' capacity to carry out special case. special evaluations. These dedicated and hard working district M&E staffs are wasting their Findings and Lessons time. A resource committed to M&E and capable of bringingsome needed feedback to 15. The T&V program provided an the program has been largely ignored. The alternative to the community development underlying problem was that the M&E program. It forced agricultural technologies  -xl- back onto the agenda, without challenging the been corrected were not and aggravated the communitydevelopment activities which were situation. Despite the positive impact on carried on as before. It also substantially production, particularly in Madhya Pradesh, expanded the number of farm familiescovered the Bank should have been concerned about by these public services. There is now a large the opportunity costs of the staff and budget state-financed field staff dedicated to resources commitment to these projects. The transferring information about improved report mentions four major adjustments to the technical packages to small and medium scale T&V methodology that should have been farmers. The central and state governments - considered at appraisal to get full force from including Madhya Pradesh and Andhra an enhanced extension operation (referring to: Pradesh - see these as welcome developments. better use of the VLW who remained in place, the extension outreach programs of the 16. A recently completely OED study of agricultural universities, the connection with the worldwide T&V experience ("Agricultural input supply, and the use of targeted Ertension: Lessons from Completed Projects" campaigns). None of these options fit the see para 1.4) discusses the favorable and classic T&V design, and major unfavorable factors. The correspondence with accommodationswould have had to be made. the MP and AP cases is remarkably high, The rigidity and ritualism of T&V was evidence that one is dealing in this audit with apparent in MP and AP, and reduced the common T&V experiences and problems. The extension impact substantially. text discusseseight points taken from Lessons, comparing the global experience with the two 18. The Bank moved in India to help states. Among the common problems are: the correct a real deficit in government's services insufficiencyin technology relevant to farmer for small and medium scale farmers. But the circumstances, particularly for resource-poor Bank should have shown broader vision in environments;the gaps in quantity and quality supporting this reform. The audit is not of the field staff, and of the supervisors and recommending any one or combination of subject matter specialists who were supposed these options. But it is clear that some to support them; the difficulty in meeting consortium of the best of the T&V principles recurrent cost expenditures; the poor and these other systems could have been performance of the M&E units; and the lack launched with a high likelihood of superior of commitment by senior officials. results than what was obtained. In answer to the criticismthat this is unfair second-guessing, 17. Some recent Bank reports on T&V the audit maintains that the alternatives were extension programs mention these problems, not even considered - they were simplyswept but conclude that most have been corrected aside by the dramatic promises of T&V and that the T&V principles are still valid and supporters. Of course training,and visitswith the programs are still on track. The Region farmers, are essential to any enterprise argues that the introduction of the T&V extending new technologies. But that was not system in the two states in the early 1980swas enough to guarantee results, as the case "an appropriate response to the prevailing studies of MP and AP demonstrate. circumstances at that time." The audit challenges those common perceptions. It 19. GOI and the states are presently suggests that the assumptions about the asking themselves how to get more out of the availability of new technologies and T&V apparatus that has now been installed. appropriate staff proved to be invalid,and the They recognize that T&V as originally budget and M&E shortfalls that could have presented is too costly for the states to sustain  -di - much longer. Substantial modifications are 20. Among the conclusions reached in the under review,some already pioneered by the report are that the Bank has to be more adept leading states. Bank supervision staff have at "projectizing"useful development tools such played a leading role in pushing these reforms. as T&V; that the Bank cannot accept low-level They include the abandonment of any pretense tests of project economic impact - as the T&V at maintaining the farmer/agent ratio at 800; SARs proposed - because the opportunity the shift from contact farmers to group costs of those resources are far higher; that meetings; the formal recognition of T&V's the T&V system in these two states has to be necessary role in ensuring and, when opened up, to the point where the T&V necessary, helping to more inputs to the structure may no longer be recognizable; that farmlands; upgrading minimum requirements the relationship of the VEW and VLW field for VEW appointments to get at least forces has to be better managed; that the university graduates; and, since 1989, the T&V/M&E system in these two states has to "broadbasing"of T&V, to incorporate in the be overhauled; and, finally, that the Bank is extension menu the horticulture, animal correct in considering continued support to husbandry, soil conservation and other extension work in both states, provided these specialtieswhich DOA offers through divisions investments can be justified at the sector level. parallel to the extension service but not at the Bank involvement should not be allowed to village level. Simultaneously,the Directorate terminate for the wrong reasons: before of Extension, GOI, is bringing reforms to the the sustainability of the reformed extension T&V/M&E component. The M&E system system - and the good ideas behind it - have had become as ritualistic as the T&V system it been secured. was supposed to monitor.  AUDIT REPORT PERFORMANCE IN)DIA MADHYA EXTENSION PROJECT - PHASE II (Credit 1138-IN) PRADESHAGRICULTURAL ANDHRAPRADESHAGRICULTURALEXTENSIONPROJECT (Credit 1219-IN) I. BACKGROUND A. Introduction 1.1 Bank support for development of the training and visit (T&V) agricultural extension system in India began in 1975. It was first incorporated into the agricultural components of two, contiguous irrigation improvement projects along the Chambal River in Rajasthan and Madhya Pradesh.Y It spread rapidly in 1977: as a minor component of a project extending irrigation schemes along three rivers in Andhra Pradesh; as a major component of two multipurpose "agricultural development" projects in the states of Orissa and Assam; and linked with agricultural research in four, single-state "extension and research" projects (including Phase I of the Madhya Pradesh extension program). When the Bank began a series of stand-alone, nationwide, agricultural research projects in 1978, extension was also reformatted to stand alone. Seven more states were brought into the T&V system between 1979 and 1982. The Bank switched to multi-state ("national") extension projects in 1985, after which previous recipient states were resupplied with funds and another five were brought in - bringing the total to all 17 of the major Indian states. The third in the series of national extension projects was approved in 1987. The pipeline of new Bank lending operations specificallytargeting T&V extension then terminated. 1.2 This audit deals with the last two single-state T&V extension projects, approved in 1981 for Madhya Pradesh 1I and in 1982 for Andhra Pradesh. A cluster audit covering five previous single- state T&V projects was issued by OED in 1 991 Z'. Project completion reports have already been prepared by FAO/CP on the first two of the national projectsy. The third national project should close in early 1995,and a completion report mission is scheduled later that year. Thus, within a five year period in the early 1990sthe most recent phases of all 17 state T&V programs except Andhra Pradesh will have been reviewed in completion reports and audits (other than the present report). The ex-post coverage includes a 1992Bank Technical Paper, by a member of the Bank's staff resident in the New Delhi Office, devoted to an analysisof the whole set of Indi*T&V projects.3' The broad coverage provided by these other reports allowsthe present report to skip some of the details of the T&V program itself and concentrate on the important themes. .1 502and562,approved Credits InJuly1974andJune1975 respectively. component A smallextension designed described alonglineslater in theWest asT&VwasalsoIncluded Bengal Development Agricultural approved Project InApril1975(Cr541). No.8808.PPAR: Y Report andBihar India- theWestBengal Agricultural Extension andResearch theKerala, Projects; andTamilNaduAgricultural Maharastra ExtensionProjects; andthe National Agricultural Project Research (Credits and855).June29, 1990. 690,761,1028,1135,1137 India- National 3 PCRs: Extension Agricultural Project (Cr1523) andNational Extension Agricultural II(Cr1596). Project June1994. i Agricultural in India.by Michael Extension Number190. Paper Mackiln.WorldBankTechnical  - 2- 1.3 The two projects were chosen for audit as part of a work program for an OED study focussingon monitoring and evaluation (M&E). The Indian T&V program has alwayshad high M&E content, and was one of seven case studies selected world-wide to support OED's M&E study. Madhya Pradesh and Andhra Pradesh T&V projects were selected not because they were exceptional examples of either T&V or M&E, but because they were the only completed Indian T&V operations as yet unaudited. Bank veterans of the Indian T&V program expressed a concern that the experiences of the two states were below average for the 17 states, and undistinguishedwith respect to either T&V or M&E. Given the M&E bias of the audit exercise, they suggested that other states with demonstrably "good"and "bad" T&V/M&E be identified to make the case study more robust. For that reason the audit mission visited Karnataka as well, to get exposure to an acclaimedleader among the T&V/M&E programs. But otherwise OED kept Madhya Pradesh (MP) and Andhra Pradesh (AP) as a cluster case study - within the audit framework. The results have justified that decision, as discussed below. For one thing, these two states - among the most populous in the country - are also witness to among the oldest T&V activities in India, a point noted in the previous paragraphs. If the strength of the T&V program is related to the years invested in setting it up, then the failure of MT and AP to perform above the average should offer lessons. 1.4 The composition of this Performance Audit Report (PAR) reflects also the findings and lessons of a study recently issued by OED on the global T&V experience (AgriculturalExtension: Lessons from CompletedProjects' referred to below as Lessons). This was the second OED global report on T&V - the first was issued in 1983k- but it was the first to challenge the practicality of some of the major assumptions underlying the T&V philosophy7'. Its findings relate to the world- wide experience, in which India plays a major role. Thus the audit provides an opportunity to test some of the assertions of Lessons against two Indian state experiences where the strengths and weaknesses of T&V (and M&E) were on conspicuous display. The last chapter of this report targets specific issues brought up in Lessons, and assess their applicability to these two states. The performance of the M&E operation is an important part of the story. B. Project Preparation 1.5 T&V was brought to India by Daniel Benor-vas consultant to the appraisal missionsfor the Rajasthan (appraisal 1973) and Madhya Pradesh (1974) Chambal Command Area Development (CAD) Projects. The t T&V was not used in those Staff Appraisal Reports (SAR). Though the agricultural servicescowbnent of both projects were important to reaching the principal objectives of the irrigation program, which was to make better use of the water already under command, the PCRs and PARs for those projects paid practically no attention to the extension components. Y ReportNo.13000, April 29, 1994 FStrenqthenlna Acricultural Research andExtension: Bank theWorld Experlence. No.4684, Report September 1,1983. Z'OED's 1990 clusterauditraisedsome ofthe sameconcerns. Y OED's policy Isnotto refer to consultarnts bynameexceptInextraordinary clrcumstances. Mr. Benorscontrlbution to and successwiththeworldwide expansion of Bankand government extension activity, throughthemedium ofthe T&V system,is Justthat- extraordinary. ThespreadofT&V in indiaInpartlcular is closely ldentified withhim, and the absenceof any reference wouldbe bothlabored and disrespectful.  -3- Nevertheless, the structure of Indian T&V took shape in the Chambal area during 1974-1975,and provided a basis for Benor and the Bank to propose similar adjustments to existingextension activity in other states. The Andhra Pradesh Irrigation and CAD Composite Project was under appraisal at the time, and the T&V system was brought into that operation as well. Discussions were soon underway with the MP and AP Governments, as well as with other states, to persuade them to expand the new extension system from these small beginningson irrigation schemes to rainfed areas in the rest of the state. 1.6 The Madhya Pradesh AgriculturalExtension and Research Project (Cr 712-IN)was appraised in 1976 and became effective in 1977. The original plan was to extend T&V to all 45 districts of MP. However, the project as agreed limited the initial operation to 15 districts, dispersed over the state in order to pick up districts representative of its many agricultural conditions (MP is the largest state in India). The Bank and the state government (GOMP) accepted at appraisal that a Phase II project would incorporate the other 30 districts. 1.7 Preparation of Phase II wascarried out by GOMP with the assistance of the Bank's Resident Mission in New Delhi. Because the T&V project format was easily blueprinted, the preparation exercise followed a pattern common to a group of seven states that were moving together in the period 1978-80 toward appraisal. This was after the decision to break research out for an independent line of finance. Three states were grouped in a single extension project, the first one approved exclusivelyfor extension activities(the Composite Agricultural Extension Project, approved in 1979, supporting T&V in Gujarat, Karnataka and Haryana). The next four states, including Madhya Pradesh 11,were processed as separate projects. The Bank had concluded that dealing with multiple state governments in a single operation overly complicated negotiations with each on issues peculiar to that state. The Bank returned to the multi-state format in 1985. 1.8 There were no serious conceptual disputes during the preparation, appraisal and negotiation of MPII, or of any of the other three state projects then movingin tandem. There was no challenge to the overall T&V philosophy. Nor were substantial objections voiced in the Bank or India to the probable and improbable consequencesof T&V - for example the predictable institutional shock from strengthening the agricultural extension wing vis a vis other line agencies, or the glowingprojections put forward for farm-level impact.!' MPII was part of one of those few, substantial, blueprinted lending programs in the Bank which gathered strong support within the Bank and were effectively translated into operations broadly acceptable also to governments. 1.9 The disputes and delays that accompanied the preparation and appraisal of MPH concerned the basic building blocks of T&V itself. Staffing - especially the means to acquire the large mass of dedicated field workers that constitutes the core of the T&V system - was the most important issue for MPH. The question was whether and when to transfer into the 'reorganized"N extension system- of the Department of Agriculture (DOA) under the Agricultural Production Commissioner - a Y TheBank'sAgricultural Department senta memorandum, In response to the Region's suggestion thatthe Yellow CoverReviewof theAPAppraisal Report bewaived, complainingthattherewasnot enough timeto satisfy itself that technical packageswereavailable to extend.Thefilesshowno further correspondence on thatissue. LYGOMPusually refers to the 'reorganized" system, rather thanT&V. substantial number of the Village Level Workers (VLW) from the Department of Rural Development (DRD)WY under the Secretary of Rural Development. GOMP agreed it would transfer them but not before negotiations, as some Bank staff advised. However GOMP did comply within the first year of effectiveness, presenting a striking contrast to the evolution of events regarding DRD staff in Andhra Pradesh (paras 2.11ff). 1.10 Processing of the first (and only) AP T&V project was much more controversial. AP was one of the few states to resist T&V. Some of the states to enter the T&V program later fall in that category. Punjab was the most outspoken. But, whereas the Punjabi Govermnent took issue with the basic proposition of T&V, and with the assumption that that state's already advanced farmers could benefit from what it regarded as an overly simplisticextension model, GOAP's objection was less principled and more the result of personal prejudice and conflict within government. The Bank's files show that it attributed govemment's reluctance to expand T&V statewide (from the irrigation schemes) to the Secretary of Agriculture, a man who held a senior position in agriculture throughout the period from Benor's first visit to Hyderabad (capital of Andhra Pradesh) in 1975 till after the Credit became effective seven years later.!' OED's audit mission was told that the footdragging and attempts to reduce costs were more a result of the Secretary's dispute with the Principal Secretary to Govemment for Finance. It was the latter who blocked attempts by Benor, the Bank and the DOA to set up a project at a scale allowing T&V to function fully, and requiring Bank finance commensurate with its contributions in the states that had entered the T&V program earlier. 1.11 Part II of the PCR on the audited project, the Andhra Pradesh Agricultural Extension Project (API), describes a series of visits and discussionsbetween the Bank and GOAP from 1979 to 1982 debating the totals and compositions of project costs and the Bank Credit. The Bank was concerned that at the low level proposed by the government it would be unable to support an effective T&V system. The DOA, under pressure from Finance, argued that by limiting new staff recruitment the costs could be kept low without injury to the program. GOAP intended to transfer enough VLW, and upgrade enough junior DOA field staff, to reach recommended staffing targets without new recruitment to the cadre of front line field extensionists. Housing was another component that was substantiallyreduced, compared with other states. The Bank argued, as it had elsewhere, that it was necessary to build houses in remote locations to guarantee the T&V field staff would remain at station. The Bank wanted enough houses to be built with project funds to accommodate 30 percent of the field staff. GOAP responded that it would accept finance for new construction for only 3 percent of the field staff - the rest could be accommodated if necessary in rental quarters. The final project cost of Rs. 7.83 crores, and the supporting Bank Credit of USD 6.0 million,were the lowest in the Indian T&V program. These figures generated doubts among senior staff in the DOA's field offices whether GOAP was serious about making the program work, and also left some Bank staff feeling the project was badly underfunded. 1't The DRD is referredto in India,in English,as the Pancahyat RaJDepartment. gives a verycynical attitudeto theT&V proposalwas describedin one Bank memo: 'that Secretary ThisSecretary's L21 Impression,seeminglyregarding touchesbut predicts T&V as mainlyBank-GOIgimmick (sic)which he reluctantly will vanish' (memoto files, 11/5/80).  -5- C. Project Objectivesand Description 1.12 The objectives in the two states were identical and expressed in few lines: the five-year projects sought "to achieve early and sustained increases in agricultural production (in all districts) through the reorganization and strengthening of the extension service of DOA". 1.13 The projects were designed to the standard T&V model, and there was not much variation between the two state proposals (except with respect to incremental costs). The projects would establish cadres of villageextension workers (VEW), each responsible for eight groups, of about 100 farm families, per group, formed within a "circle"of three or four villages. The VEW would make the rounds of his entire circleonce in two weeks, deliveringmessages (="recommendations' ="impact points") to the ten contact farmers selected from each group and supposedly waiting for him on the appointed time marked in his "dailyvisit schedule". Each contact farmer (CF) was expected to pass the messages to his nine neighbors - the so-called "other farmers" (OF). Each VEW would thereby pass information directly or indirectly to about 800 families. The VEWs would be supervised by agricultural officers (AO), who in turn were supervised by agricultural extension officers (AEO) located at the 'block" level (the focus of governmental services within districts, throughout India, since the decentralization movement of the 1950s). The line of command, reflecting one of the fundamental features of the T&V model, was to be unambiguous and undiluted, passing from the AEOs to a subdistrict officer, then to the district and possiblya multi-district layer of administration to Departmental Headquarters at Bhopal or Hyderabad. Every state T&V system - its hierarchical structure, titles and grades - varied around this model. For example, in AP: the VEWs were called Village Extension Officers (VEO - however this report willcontinue to use the generic form VEW); in later years the AOs at the next higher rung were sited at the Mandel level, a subdivision of the Block created in 1987; the block-levelT&V postings were eliminated; the DOA put a Joint Director (JDA) in charge of a district, whereas MP put only a Deputy Director (DDA) there; etc. But the basic model was recognizable throughout the country. The "project description" section of the SAR was devoted exclusivelyto reviewing the extension methodology and detailing the organizational structure, staffingrequirements, training program, monitoring and evaluation service, and the houses, vehicles, and other incremental resources required. There is no discussionin this section of the SAR of the characteristicsof the technical programs that would be carried to the target populations. Apart from the changes in input numbers and nomenclature, the "project description" sections are interchangeable. 1.14 Reflecting the different positions taken by the two states with respect to incremental project costs, the proposed outreach of the MP program was smaller but more intense than in AP. MPH would cover the 30 districts left over from MPIL' and a farming population of 3.8 million families. API would cover all 22 districts in that state and 5.5 million families (AP's district populations are larger than MP's). Nevertheless, incremental resources approved for T&V in MP were six times greater than those approved for AP - USD 60.0 millionvs USD 9.7 million total costs. The Credit sizes were USD 36.9 million vs USD 6.0 million. In consequence, the proposed farmer/VEW ratio 1-3 Funds to continuethe programin the original15 districtsof MPI (including the two ChambalCAD districts),after MPIwas exhausted,would be deliveredby a subsequentproject. Four years later, In 1985, the Bank approvedthe NatlonalAgriculturalExtenslonProject,which as originally conceivedwas supposed to cover the 15 but In fact was written to cover all 45 districtsas and when needed.  -6- was significantlylower (more favorable) in MP - 494 per VEW in MP vs 812 per VEW in AP, and the housing budget in AP was nearly eliminated - 195villageresidences in AP vs 3,000 in MP. These and other differences between the two state programs are shown in Tables 1 and 2. The first shows project costs and Bank commitments (of IDA funds) by major category. Table 2 compares physical features of the T&V design. Table 1: Project Costs and Bank Commitments, as Appraised MP H AP I Project IDA Project IDA Costs Commitments Costs Commitments (USD in millions) Incremental Staff Cost 19.4 17.9 2.0 1.8 Civil Works 12.8 11.9 1.2 0.8 Equipment and Vehicles 4.0 3.8 2.3 2.2 Training 1.4 1.4 0.7 0.9 Incremental Operating Costs 7.9 - 1.5 Total Base Cost 45.5 - 7.7 Contingencies 14.5 - 2.0 Total Project Cost 60.0 - 9.7 Unallocated 2.0 0.4 Total IDA Commitment 37.0 6.0 Table 2: Project Parameters MP II AP I Districts (total in state) 30(45) 22(22) Farm families (million) 3.8 5.5 Staff: VEW - Existing - DOA - 4,355 . Existing - VLW transfer 3540 2,421 New 4150 AEO/AO - Existing 317 853 -New 797 SMS - Existing 32 193 -New 345 74 Houses: VEW 3000 195 AEO/AO 500 Vehicles: 4 Wheel Drive Sedan T&V 122 80 M&E 2 1 1.15 The agreements at appraisal on how to assemble the cadres of VEWs and subject-matter- specialists (SMS) were the most difficult. As discussed below, one source was to be the VLWs then reporting to the Departments of Rural Development in both states. Of the 7,690 VEWs it required,  -7- MP agreed to transfer 3,540 VLWs (46 percent) from DRD, and recruit the rest from outside the civilservice, payingsalaries of this last group with project funds. Of the 6,776 VEOs it required, AP agreed to transfer 3,000 VLWs, and upgrade the rest of those needed from their current positions as field assistants and sub-assistantsalready on the DOA roster, including those presently under the administrative control of the CAD Administration on the irrigation schemes. VLWs were better trained than the typicalDOA field assistants (they had been given two years of special training at the post-high school level to prepare them for work within the rural development program). The Bank was aware of the distinction, and it tried to reach agreements that guaranteed at least a minimum level of expertise. With respect to the SMS, which was another vital element in T&V design, AP again insisted it could get along mostly with existing staff. MP agreed to include 377 SMS at the district and subdistrict levels combined, givinga VEW/SMSratio of about 20. AP agreed to include 267 SMS at the two levels, for a ratio of about 25. These ratios are close. But MP proposed to recruit 345 qualified persons, 92 percent of the total needed for those positions, while AP proposed to recruit only 74, or 28 percent, and make up the rest from its existing roster. As discussed below, these distinctions and expectations proved important in determining the outcome of the projects (AP never transferred the VLWs and never recruited SMS, throwing the DOA back on its own staff resources), although not as important as the Bank believed at appraisal. 1.16 M&E cells were to be established in the DOAs in both states, and operate along lines described in a manual on M&E methodology issued in 1981 by the Bank for Indian T&V. The MP staffing pattem for the cell included four evaluation/analyst positions - including three sociologists- along with statisticians. The AP staffing pattern included no evaluation positions. The difference proved immaterial, because MP never filled the evaluation posts. ll. IMPLEMENTATION EXPERIENCE 2.1 Project implementation was difficult in both states. But, whereas 15 of 16 Bank supervision missionsto MP rated that project an overall 2 (moderate problems), 10 of 14 missionsto AP rated that project 3 (major problems). The main problem in Ml was budget shortfalls throughout the first five years, reducing MP's initial advantage over AP in agreed funding levels. The main problem in AP was political. For several periods during implementation the Bank considered cancellingthe AP Credit. The Resident Mission, responsible for supervision throughout this period, was hopeful that the inhospitable political situation in AP would eventually clear up. By late 1987 that was the case, and MP's budget crisis had also passed. The two stories must be separated. A. Madhya Pradesh 2.2 The project became effective in July 1981, only two and a half months after it was signed. But it ran into trouble immediatelybecause of a gathering fiscal crisis at state level. Expenditures on the project were depressed for over five years due to poor budgetary support. The supervision rating dropped to a 3 for the first and only time in 1985, because of the counterpart shortfall, and the project was put briefly on the Problem Projects list. The crisis undermined implementation of all Bank projects underway in MP in the early and mid 1980s - four of them in agriculture. Implementation of the follow-on project, the MP component of the National Agricultural Extension Project (NAEPI) effective in early 1985, was also held back for the next three years. In October 1986Bank supervision reported that only 30 percent of the state's budget commitments to MPII since  -8 - start-up had been released. Thereafter, activities accelerated, recruitment picked up, and by Credit closure in June 1989 (two years late) 90 percent of all extension positions had been filled. 2.3 The PCR says that the early financial problems could have been avoided if the project had been delayed until MPI had been completed - that the state showed itself incapable of absorbing two extension projects simultaneously.This is an odd conclusion,given that Phase II had been anticipated since appraisal of MPI in 1976 and the intention was not to overlap activities under MPI but to extend them to the rest of the districts. There is little justification for arguing that MPI should have been allowed to finish its work in the 15 pilot districts before letting MPH carry the T&V program to the other 30 districts. The DOA could have managed T&V throughout the state, if the funds had been provided. If the Bank wanted to soften the effects of the state's fiscal predicament on the Bank's projects, it should have deferred commitments across the entire portfolio. 2.4 In addition to the funding situation, the supervision reports and other Bank correspondence record the eight-year history of project implementation - two longer than planned. There is continuous criticismof DOA's failure to fill senior staff positions in the M&E cell, of delayed M&E reporting, and of the lack of any analytic inquiry into the issues raised by the data from the M&E field surveys. There were similar problems on the auditing side, with late submissions and, in response, occasional Bank suspensions of disbursements. 2.5 The criticism of most concern related to the technical capability of the T&V program. In those districts where the limited funds were deployed at levels sufficient to support adequate staffing of the VEW posts, supervision usuallyreported that farmer contacts had been established and useful technologies were being transferred. As discussed in Chapter III, the development of the cropping pattern in one large area in the middle of the state was said to be partly the result of this effective village T&V work. However there are frequent references in the files to severe constraints on the full power of the T&V system. These included a shortage in staffing at the AO and AEO levels, in the size and quality of the SMS cadre, and in physical support from the research service. These components backing-up the village extension workers were fundamental parts of the T&V system, and weaknesses here, though not fatal, were seen to be holding the program baclk 2.6 Apart from these staffingconstraints, supervision also complained that advances in improved technologies were not keeping up with the spread of the T&V field force - that "new technologies were not being developed fast enough". One of the positive effects attributed to the Indian T&V program was that it put pressure on the research services to deliver viable packages. Whether the assumptions about the availabilityof new technologies had been over-optimistic is an issue left till later (para 5.7). But it seems clear that the research/extension linkage was not as effective as expected, at least in MP. One of the reasons was that the Bank, by separating the research from the extension components in projects in the late 1970s,had allowed a gap to reopen that had been partly closed by the earlier series of research/extension projects, including MPI. In that project, for example,a research agronomist had been posted at each of the 15 district extension offices. In MPH, after research had been moved to a separate project, that explicit linkage decayed. Though the agreements on MPII called for similar positions in each of the other 30 districts, most of them were never filled (until the seventh year of the project). And most of the "district agronomists' already in place shifted their office back to the local, zonal research station. Arguably that was a better site for them than at the district extension office. But given the budget difficulties of the agricultural research establishment, centered at the state agricultural university at Jabalpur (Jawaharlal Nehru  -9- KrishiVishwa Vidyalaya- JNKVV), this physicalrelocation was accompanied by a deficit in transport facilities and the separation was greater than had been intended. 2.7 The point about enforced separation cannot be pushed too far, because the intent of the Bank's two consecutive national research projects was to work in exactlythe opposite direction. And JNKVV's involvement in backstopping the extension program, in particular the monthly district training programs where the 'lesson plans" (with the annotated list of messages) were prepared for relayingon to the AOs at the fortnightly block training sessions,was graduallyexpanding throughout the 1980s. Whether the linkage would have been even stronger if a research component had been preserved within the project is an unanswerable question. The audit mission met officers from both the research and extension establishmentswho said the linkage looked better in the reports than on the ground: research said that attendance by the AEOs was decliningl' and extension said their source for new, monthly lessonswas drying up. Bank staff insist the AP research contribution to the extension program was weaker than it should have been, and the quality of its work, if not the quantity, declined during the project period. 2.8 The Credit was fullydisbursed and closed on June 30, 1989after two one-year extensions of the Closing Date. Salaries and allowances consumed 67 percent of project costs and 70 percent of Bank disbursements, compared with appraisal projections of 43 percent and 51 percent. The 'facilitating"categories of equipment and vehicles, training, and operating costs took 12 percent of actual project costs, compared with 29 percent planned. B. Andhra Pradesh 2.9 This section starts where the last left off, with the research connection. The development of the linkage between the research and T&V programs in AP has been spectacular, giving the T&V program professional and intellectual support superior to those in most if not all other states. The work is centered at the Andhra Pradesh Agricultural University(APAU), located outside Hyderabad, but the effective interaction has been based at the seven regional research stations and other research facilitiesin the APAU network scattered throughout the state (the physicalestablishment of APAU is larger than in any other state agricultural research system). The regional centers manage the regular training and demonstration programs for extension staff in their agronomic zones. The Vice Chancellor's foreword to APAU's report Extension Highlights 1991-1992 suggests the depth of commitment at APAU to forging the link: 'in the present Indian context, research in agriculture without application/extensionis only a hobby and on the same count, extension without research is folly". 2.10 A reading of the supervision reports, files and the PCR on AP show that there were no other consistently bright dimensions to project implementation. L1YFigures on attendance at monthlymeetingssometimes clearlysupport the research objection. At one research stationvisited by the audit mission,the ratio of extensionto research staff In attendance at monthlymeetingshad declined from 70%to 30%. The researchersclaimed they began to recognizethe same DOA faces returning to allthe metings, on whateverspecialty,and concludedthat some SMSstaffhad been appointedsimplyto 'attend meetings with research', lettingthe other SMSget on about their businessl  - 10 - 2.11 Overriding all issues was GOAP's failure to complete the agreed staffing pattern. The situation deteriorated during the project period, mostly due to political influences. Within months of the project becoming effective, the elections of late 1982 brought to power as Chief Minister of AP a newcomer to the political scene who was to establish a reputation over the next five years as one of the least predictable heads of government in the country. There are five separate factors that undermined the staffing plan of the project. The first, the VLW transfer issue, and the second, the limitations on inter-district transfers, predated the 1982 elections. The other three can all be attributed to the political machinationsof this Chief Minister. 2.12 GOAP had obliged itself to transfer '3,000" VLWs (AP uses the title Village Development Officer - VDO; the exact figure recorded in the SAR was 2,421) from DRD (the Rural Development Department, commonlyaddressed as the Panchayati Rai Department) to DON The order was issued by the state government before negotiations. It was never put into effect state-wide (although in two districts a few hundred VLWs did shift although some of them later returned to DRD). There was resistance to the transfer throughout the cadre of VLWs, who correctly anticipated a loss in salary, perks, job satisfaction, and promotional opportunities. The action was eventually stopped by legal action. In 1985 the Bank agreed with DOA to a staffing plan that was based on 4,000 rather than 6,776 village-levelstaff positions, minus the VLW. 2.13 The second factor was the legal prohibition on the state government's powers to transfer staff in the lowest service categories from the district where they were recruited. The DOA was forced - by the agreement to abandon rights to the 3,000 VLWs, and by its inabilityto recruit - to rely entirely on its junior field staff. It had to respect the spatial distribution of this cadre from before the project. Most of them were concentrated in the higher potential areas of the state, including the irrigation schemes and most particularly the districts of the rice basin of eastern AP. They had been recruited there to carry out the national and state food production schemes and other campaigns of earlier decades. Without a supplementary corp of field extension workers to assign, DOA was unable to bring the other districts, particularly the western, dryland areas, up to strength. The VLW would have helped fill that gap, but they were unavailable. This imbalance, with some good districts at full strength and the other, disadvantageddistricts suffering from huge gaps in the T&V staffing pattern, persists to this day. 2.14 The third factor was the new government's quick decision to retire all civil servants over the age of 55. Normal retirement age in the Indian civil service was 58, so the nominal change was small. And it affected all government departments. But it took away 130 of DOA's older staff, including most of the senior officers who had been involved in preparing the T&V project, or who had been preparing themselves to manage its implementation. 2.15 The fourth factor was also not aimed directly at the project. But it had a profound effect on project structure and was seen by the Bank to threaten the integrity of the T&V system. The new government established a new layer in between the block and the villages. This action had admitted political intentions, since the block administration in AP (as in most of India) was thought to be dominated by the Congress Party. The new layer was called the Mandel Panchayat, the old blocks were broken up into about four mandels, and the blocks disappeared as an administrativeunit in the agriculture and communitydevelopment hierarchies. The post of the Mandel Development Officer (MDO) replaced the post of the Block Development Officer as the key DRD decision-makingcenter. Farmer committees were set up at the mandel level (Karshak Parashatel - KP), and the state-wide  - 11 - organization of these committees was put under the control of the Chief Minister's brother. The mandel affected the project directly because all of the T&V AOs and VEOs were subjected to the authority of the MDO (and, indirectly, the KP). This wasjust what T&V was designed to prevent - the dilution of authority. Although the relationship between the MDO and the DOA staff at the of mandel level was neither uniformlynor clearly established (the DOA retained technical authority its staff, unless they were otherwise needed), there was no question but that the activities of the VEOs thereafter came under the influence of DRD and the farmer committees. The mandel authorities would interrupt the T&V visit schedule whenever they saw it was in their, or the communities, interest to do so.)' 2.16 The fifth factor is a follow-upto the fourth. In June 1987 700 of the T&V AOs (the VEW's supervisor) were transferred from the DOA to the Mandel Panchayat and the direct authority of the of MDO. The previous reform had left the T&V field force in DOA, but fractured the line command. This new action stripped DOA of half of its first-rung officer corps, and left thousands of its VEWs unsupported. The DOA (and the Bank) struggled over the next year to have the action reversed, and ultimatelysucceeded. Nevertheless, although temporary, the shift is said to have further disrupted efforts to keep the project moving. 2.17 The cumulative effect of all five factors overwhelmed the administration of the project. Even a substantiallylarger budget would have been unable to offset these shocks, although it would have improved performance. The low budget compounded these difficulties. At the time of the audit mission, although the staffing complement had been retained at the same reduced strength as when the project wascompleted in 1989,the amount of funds available for expenditure headings other than salaries and allowances was too low to support effective extension of any type, let alone one that is based on high mobility, intensive supervision and frequent training. The other important consideration is that the DOA's existing force of field assistants and sub-assistants were in their majorityundertrained and unprepared to carry out the T&V role they were put into. In spite of this initial deficit, DOA accelerated their entry into the ranks of T&V agents by shortening the agreed training program. As one of the Resident Mission staff said, quoting an official from the AP government, the project expected a donkey to carry the weight of an elephant. 2.18 The next chapter discussesthe difficulties in developing improved technologies appropriate for the dryer, western areas of the state. The DOA, partly in response to that constraint, and under pressure also from the tight budget and the administrativefactors just mentioned, allowed the staffing complement in those districts to fall to levels well below the minimum required for effective T&V. The number of VEWs was so low in the mandels of some districts that the ratio of farmers per VEW 1/ Thedebate Insists at lengthin this text. TheT&V blueprint is notdiscussed overT&V's'singlelineof command' on It. Supportersof the movement in Indiatowardsdecentralized administrative controls,with greaterpower assumed by representatives of grass root constituencies, endorsethe transferof controlto block and mandelelectivebodies - either direct or throughtheir influence on Blockand MandelDevelopment Officers(BDOsand MDOs). The Bankfought long and hard In AP to thwart GOAP's efforts to subjugate the T&V staff to these local DRD and electivebodles. of the T&V staff intothe local administrativestructurecrippled theT&V performance Is uncertain. WhethertheIntegration backstopping,a role and One could argue itgave the T&V staff,alreadycrippled by poor budgets and weaktechnical that the line of command had already been utilityIt badly needed. In the files on the AP projectthe Bank recognized and Karnataka), without apparent damage. The 1992Bank report fracturedin threeother states (Gujarat,Maharastra of this debate. Agricultural Extension in India (see footnote 4) recognizes the strengths on both sides  - 12 - exceeded 4,000, the AOs had to help cover the circles otherwise without support, the subdistrict officers (ADAs) had to share the dilapidated fleet of vehicles to the point where one ADA could only expect mobility one week a month, and supervision effectively ceased. On top of these problems,and just prior to the audit mission, the DOA eliminated all SMS positions at the subdistrict level, converting those SMSs into ADAs and expanding the number of subdistrict agencies. This action was intended to restore a measure of supervision of the AOs and VEWs. And it recognized the fact that the SMS were not trained wellenough to perform the intended role of the SMS in T&V and might as well be put to better use as supervisors. The mission spoke with one "regular"ADA - one of those who was an ADA before the reorganization - who said that that last change had effectively ended T&V in her domain. Similar frustration was evident in all discussionsin this half of the state. 2.19 The Credit was fully disbursed and closed on March 31, 1989 after two extensions. Incremental staff costs consumed 57 percent of projects costs, compared with an appraisal projection of 26 percent. III. PROJECT OUTCOME A. GeneralComments 3.1 The projects were successful in greatly extending the process - initiated in both states in the late 1970s - of creating a public service, at the village by pilot operations level, dedicated to the extension of agricultural technologies through farmer training. Previous community development programs carried out by VLW and DOA field staff in these and other states had mixed technical transfer with the supply of inputs and credit. In fact, in the decade preceding the introduction of T&V, the purely "extension" functions of the VLWs had been superseded by subsidized programs intended to spread the use of already known cropping packages and agricultural programs such as rural roads, child-care, and other activities satisfying equipment, and by non-agricultural needs of the rural households. T&V put new technologiesfor small farm agriculture, specificallythe grain, pulse and legume rainfed farms, back on center stage. In so doing, it pressed the research services into this line of service as well. The Indian Government recognizes this to be the single most important contribution of the overall T&V program, and MP and AP are part of that success story. The importance of preserving a mechanismfor technical transfer to these farmers is no longer questioned. 3.2 Direct Bank support for T&V ended in 1989 in Andhra Pradesh and in 1992 in Madhya Pradesh. GOAP proposed a follow-on operation to the Bank in 1988, but the plan was not accepted. The new structure of Bank support to state agriculture is through Agricultural Development Projects (ADPs), in which extension is only one - and a relatively minor one - among a long list of possible component pieces. For extension, they emphasize training. They are unlikely to include Bank finance for salaries and other operating expenses. Both MP and AP are currently discussingADPs with the Bank. The objective of the ADPs is to support state government agricultural programs as policy reforms are put into place. 3.3 Since the T&V system was imported to India largely in response to Bank urgings, and at the promptings of the charismatic consultant who offered T&V to the world, the program could have decayed with the end of direct Bank involvement. The audit mission inspected budget and  - 13 - expenditure data in both states, and found that the states have maintained support for the DOA extension force at budget and expenditure levels only marginally below where they were at project completion. This is more impressive for AP, where Bank supported ended five years ago. For MP, the Bank withdrawal is too recent to provide a clear test of sustainability,although FAO/CP, in its PCR on NAEPI, finds early signs of budget pressure. Since the bulk of expenditures were always commnitted to salaries and allowances(paras 2.8, 2.19), since state governments cannot readily dispose of large civil service cadres, and since the erosion of funding has been with operating expenses, vehicle repair and other cost headings that make the field staff effective, the continued existence of the T&V programs in both states is not proof positive that they are a permanent part of the institutional and policy fabric. Nevertheless the audit mission also confirmed that talk about overhauling the programs meant just that, and did not suggest they would be dismantled. In that, partly qualified, respect, the projects can be said to have created a sustainable institutional infrastructure. It appears the state governments are convinced that a public sector extension service based on farm visits is necessary for their small farmers. T&V provides an institutional framework which also fits well into the government bureaucracy. Within that panorama, the decline of the T&V structure in western AP must be seen as a special (but instructive) case. 3.4 The current discussionsin India, and these two states, about how to adjust the T&V programs to a low-budget scenario are mentioned below in the chapter on findings. 3.5 The rest of this chapter deals with the evidence of technical progress at farm level, and whether it can be associated with extension deliverysystems. The PCRs are unhelpful here. The MP PCR mentions the revolution in kharif cropping and changes in the cropping pattern in the black soils area of central-west MP, and claimsT&V has had a major role in bringing that about. The AP PCR does not cite any site-specific process of technical transformation, though, like the other PCR, it makes frequent allegations about the favorable on-farm impact at least in those districts where DOA was reasonably well staffed. Claimssuch as "the efficiencyof the technology delivery competence of the extension service has vastly improved" and "there is no doubt that the project has had a very significantimpact in providingvillage-basedprofessional agricultural advisoryservices to most, if not all, villages"appear frequently in both documents. They contradict other, disturbing commentary in the files and the PCRs themselves about the continued weakness at the AO and AEO levels of supervision, in the quality of the SMS, in the competence of the VEWs, and in the stream of technology. Thus one must set the PCRs aside and start afresh in trying to assess impact. The evidence is subjectiveand unquantitative, but worth reviewing. The two states are treated separately. 3.6 Before passing to the state experiences, a comment is appropriate on the appraisal practice of projecting Economic Rates of Return (ERR) for these and other T&V projects. In the two SARs, project benefits are not directly estimated. Rather, the analysis indicates by how much yields on all land planted to the primary crops must increase in response to the project to provide incremental net returns over project costs sufficient to give a 50 percent ERR. The required yield increment is minuscule: for Andhra Pradesh paddy and sorghum yields need to increase by less than a third of 1 percent (0.3 percent) at full development. That translates for paddy to a 6kg increase over average yieldsof 1,670kg(in the late 1970s),and for sorghum to a 2kg increase over average yieldsof 580kg. The MP SAR uses paddy, wheat and "other crops", and the fractions are slightlyhigher: the overall yield increment must reach at least 1.8 percent. The PCRs show that actual increases in yields have exceeded those fractions. For AP, the PCR presents a table showing actual average yield improvements surpassingthose thresholds - and substantiallyfor paddy. For MP, a similar PCR table  - 14 - shows incremental yields surpassing the thresholds substantially for all recorded crops. The line of argument in both cases is to admit that there is no wayto attribute any of the overall yield increments to T&V - or to any other individual or mix of factors - but that available evidence from observations is strong enough to support the assertion that T&V at least contributed enoughon-farm to pass the threshold. At 0.3 percent and 1.8 percent, those thresholds are not at all forbidding and the assumption seems acceptable. The PAR returns to discuss this procedure, and what it implies, in a subsequent section. B. MadhyaPradesh 3.7 The PCR mentions only one example of extension impact, but that is a big one: "The increase in area planted to soybean and the consequent reduction in Kharif fallows in black soil regions of the state is seen as a significantachievement under the project. The soybean area has gone up from 0.3 million ha in 1981-82to 1.3 million ha in 1987-88. The introduction of the short duration pigeon pea and a mixed cropping of long duration pigeon pea with soybean in the Kharif season are other important achievements" (page 5). 3.8 By the time of the audit in 1992,the soybean area had jumped to 2.2 million ha. The change in cropping pattem in the black soil regions consequent to the introduction of soybean - allowing the land to be cropped twice and doubling the intensity of farming - is a major adjustment in farm practice. These heavy,stickysoils had been used occasionallyfor other crops during the wet, difficult, kharif season. But the farmers who planted a kharif crop could not get it out in time to plant a rabi crop, except in some areas with late-sownwheat. It was the high prices paid for the soya crop by new oil seed processing plants that turned the economics of kharif farming around and persuaded farmers to plant rather than fallow. The soya was planted sole, or intercropped - as mentioned in the PCR - with long duration pigeon pea and other crops. The subsequent introduction of short duration soybean and pigeon pea allowed these rainfed farmers to plant mustard, pulses and wheat in the rabi season as well, as they had before, taking advantage of the residual moisture in the soil and the occasional October rain. These and other adjustments in the cropping pattern have supported the further intensification of farming outside as well as inside the black soils belt. 3.9 How much extension had to do with this process is unclear. The soybean story is part of a larger process of expanding oilseed production in the dryland areas of this and neighboring states, and the early years of the oilseed, and soya, revolution predated the arrival of T&V. Some persons outside the extension service but familiarwith the soya story in MP say that extension agents before and after the introduction of soya played at best a minor role in what was going on - that it was driven by the oil seed processors and the very favorable prices paid. But the majority opinion is that extension, and most particularly the T&V farmer contacts dedicated to technical transfer, accelerated a process that was already movingwell. A look at the fortnightly lesson plans that guided the VEWs in the days in between their training sessionsdoes suggest that they were prepared to talk about most technical aspects of the new cropping patterns, including the pests that came with them. Box 1 looks at the seeds, the farmers and the extension agents in one area visited by the audit mission. 3.10 There are no comparable success stories discussedin Bank files about other farming regions of the state. In the wheat belt in the center of the state, modest improvements in farm practices can  - 15 - be attributed in part to the DOA's extension service. But this was a region of strong activity by JNKVV's own extension operation (the State Agricultural University's (SAU) main campus at Jabalpur is in the middle of this region), and any successes would have been shared. Further east, in the rice regions centered around Raipur, there have been improvements in rice yields recorded over the last two decades. Here again, however, T&V extension cannot claim to have been the primary agent of transfer. In this case it was the pre-T&V Intensive Agricultural District Program (IADP) that took a lead role. DOA old-timers who worked under both the Raipur IADP and the new T&V program feel that the LADP program was a more effective (and more expensive) vehicle than T&V for pushing new technologies. Unlike T&V, LADPlinked new technologies with delivery of subsidizedinputs, it could target on the better, more responsive farmers (instead of on all farmers, through uniform farmer/staff ratios and contact farmers), and IADP project staff were involved in marketing as well. Nevertheless, the soya story overshadowed some other respectable technical transfer operation -broadcast to line sowing,rice in Rapur, new horticulture crops, etc. - that deserve mention in a compendium of T&V impact. 3.11 Since it takes only small incremental improvements in crop incomes to reach the ERR thresholds (as stated above), it is conceivable that the shared success in the central west of the state could by itself justify the expenditures under the project throughout the state. Incremental impacts elsewhere would add to that effect. Given this production result, the commitment of GOMP to the T&V program, and the apparent stability of the field force to date, the audit rates this project satisfactory. C. AndhraPradesh 3.12 The state is divided in two main regions, the western drylands,which include districts ranked among those with the lowest rainfall in the country, and the wetter east, including the large, irrigated delta of the Krishna River. There is a remarkable concentration of research activity in the western region - in fact in the vicinityof Hyderabad. The main campuses and experimental fieldsof ICRISAT (the International Crop Research Institute for Semi-Arid Tropics), the India Dry Land Research Program, within the ICAR network (the Indian Council of Agricultural Research, of the Government of India), as well as the Andhra Pradesh Agricultural University (APAU), are located there. 3.13 Despite the concentration of research activity,state and Bank reports admit that there has been very little progress with improved technologies in the western region. The audit missionvisited Rangareddy District near Hyderabad, and later drove through the driest area of the state to Ananthapur in the far southwest. The comment heard most frequently, at the T&V offices as well as the research centers, was that no significant, widely accepted, new technical packages were availableto give a boost to the extension activities. That does not mean there is no businessfor the VEWs. As in MP, the fortnightly lesson plans list the various improved practices to be transferred to the farmers, and they deal with several field crops (and some horticulture crops) and the pesticides and other inputs that comprise the improved package. Advances are mentioned in getting farmers to use chemical fertilizer mixes, instead of just nitrogen, in inserting hybrid sorghums in the rotation, in planting sunflower and protecting it from birds, etc. Box 2 looks at one such success story, of extension support for bringing irrigation to two dryland farms. But examples such as this and other positivecomments are generallyfollowedin interviewsbyexpressions of disappointment at the overall slow progress. The sense of the conversations is that there is not a significant advantage for the farmers who adopt, and that most farmers in any case do not.  - 16 - Box 1: Seds In Shore, MadhyaPradesh TheStateAgricultural University has a campusartdresearch stationin SehoreDistrict, just west of Bhopal.There the level of farmerawareness of improvements In cras and $eiesi srelatively high. The tradtrnal rotationIs based on a mixtureof sorghumand plgeon pea In the kharifseason,followed by wheat and gram In Fe rabi season provided residual soil moisture is adequateand there'ar at:lea%t a few lght rains In Octobr00nd November.Sobean was Introduced in the lt ten year, and has:spread Widely. It Is plantedIn khari,fusual In mure wth sorghumand/or pigea pea In some flelds shortsea.son soya is the doinkantcrop, wth the mixed crops in occasionalrows or scattered. in thicase, whenthe soya is harvestedth othercrops arm, removedsimultanusly to permitpiardin of the rabi crop. The sorthum in this part Of th stat iS Usedfor fodder,the peas for food. The mixedcrops in a dominat yaroti are consiered of secondty valu and whateveris not readywill be pghed under. In otherfeldssoysais thescondy crop. It is haedale, and the good stands of longerseasonsorghum and/orpigeanpea are leftin the field untrady. For th* pea that is severalmonthsaway. The farmersmakea deliberate decision In this case to forego a rabi crop. These arethetwo basicrotations. Therearemanyvariations, including othe crops. Farmersare as keen about the qualityof seed they plant for all these crops as they are:about the rotation. As the researchestablishment releasesnewvareltis especially the popularIshoterseasonvaretieswhih mlain yields and otherpropertiescomparableto the traditional, lone seasonvareties,thefarmersare quickto adopt, The farmersgrade their own seed, normallyretaining half Oftheir reqi0ret, for " subsequentseasonand purchasing the restas certMed seed. Grading is ta skill,and:nrmally donebythe women.The tension- sevice, includingtheT&V program, concentrates on th cOtactfmOrs which inthis rego ae t a men. The information is passedalong to thewives. Thi pressure fot improvedseedhas-nogenderditilnctkrio both menand womenare saidto apprecie the relative advag ofthe differentrotationsand the Impotance of good seedto makingany of the modemrotations wok. HowMIuch ension has to do wlththese prcse is uncertain- introducing oya, or troducing t st seaso ghus and the newer,hybridsoghumst are now enteringthe market,or bringingthe certified sed to the:villages..The medium sal fanrer alonp the mainroad irterviewedby the audit nission -ten milesfromSeho4e town- who had harvested his oya andWas, ploughingunder the sorghumand pigeonpga to ge the land readyfor gram, said he hWdneverheard of The and contact farmers. But he was visted by extension. pople from the unvesity resarch team, and they occasionally provided himfreeseed aspar Oftheir eriment 3.14 The Ananthapur visit was a good example. This is primarily a groundnut zone - 80 percent of the cultivated land is planted to this crop in the wet (monsoon) season, and 20 percent of the (much smaller area of) cultivated land in the dry season. The groundnut is one of ICRISAT's primary crops, and releases of improved varieties by both that international agency, ICAR, and APAU have aimed at Ananthapur as much as at any other dryland district. Accompanying the varietal releases, the research and extension services have been promoting the adaption of what is called the broadbed and furrow land preparation profile - planting on the wide ridges and allowing the narrow furrows to carry away excess water. Yet despite these and other components of what might be described as new technical packages, groundnut yields have been almost stagnant. The TMV-2 variety introduced 60 years ago still outperforms all newcomersin most fields. The primary objectiveof research now wouldseem to be to stabilize,rather than increase, groundnut yieldsaround the 800 kg/ha level. ICRISAT's deep plough has not been accepted, partly because it needs more draft strength than the weak animals of the area can muster. Pest control is one part of this picture where the VEWs appear to have had some impact, as the bugs and diseases return in their seasons, or change with the changing varieties, and the controls have to keep up. The overall sense of stagnation has many other exceptions, which the VEWs and their superiors talk about when making  - 17 - the farm visitswith the audit mission. It is during the discussionsback in the offices and hotels where the overall sense of frustration is communicated.L' ~ ~ '-;~ ~~~~o ' : 'r' . :r: .y.::n.:r. 'ic .,'e.' ... ...-.. ,~~~~~~~~~O ... r...VX. ........ . . . ............ b ft ............. i=: peilwtanzone Th: atrdi w~here rIZtlSlf halS. IW twiabt.d. there vi..tedhI .g hasbeen a substamiat dit. c d .rIStd ............. Increase ln-- west.B.n.:p r the md.h. itt h'tlgated .hyderSba e.... . 'hrv truckfar,ingri* too them- I In the las slo :. f' a'y: r :. d m :~~~~~~~.... tlec-h er. . 4taswellr thSrsts.hi r. own 0: w d g cth.ces of hi fo wore paddy and vnewas-Y .in vegeta .thmon (concentrating t ar lest yr, hewed awlzsor.t tI pleated l.n& Thend farmerliadas recntly a ited........ (.....e4 we.l. h. invariably Het*o tiedliveacres,of whichti. noe planted and teoh four withma Th e t T&V o .. -arte. r, ............. .d ot. bu ris r ei xe o r t........e.h... to n use5 etc. thiseriseands oldt phesticodey, e seric the the (fe at, withohre too irputact oft extension service is not considered significant. There have been rice varietal releases, and associated pesticide treatments, which have spread rapidly through the deltas. The transformation dates back to the 1960s,and in particular to the Intensive Agricultural Area Program (IAAP) which was carried out in two delta districts. LAAP was a sequel to LADP, and attempted to spread that type of campaignto a larger number of farmers. APAU, supported by the international research team of the 'nternational Rice Research Institute in the Philippines (IRRI), has played a major role in increasing rice yields and carrying these programmatic initiativesforward over the last two decades. It has been instrumental also in fitting rice and pulses into a sequential cropping pattern that makes best use of residual water in those irrigated zones of the delta where water releationed and do not pnermnita second paddy crop. Farmer awareness of varietal introductions, and the farmer's abilityto descrbe to the audit mission the progression of new rice and pulse releases, the advantages of each, and the pests they attracted, was remarkable. 3.16 The T&V extension system takes credit in its reports for contributing to these advances. And it is certain that extension must have played some role. But in this case - unlike the soybean story in Madhya Pradesh - there is a strong sense even among the T&V practitioners that the farmers the know more about the new packages, and especially about fitting them to local conditions, than 6 ' ofincenties Thelack forthe DOA effecton performance. feld forcehas an incalculable The situationofone ofthe AOswhokindly attendedthe audit missionin Ananthapur brngs out this sideof the story. Hehas served25yearsin and couldnotexpectmorethanone promotion that category, beforeretirement. He spoke of his brother,whohad enteredprivatepracticeas an engineerand had alreadyhad threepromotions. The meager salarIes explainwhy pronotionsareeagerly soughtwherever theylead- up into, through or outo T&V. He gets a saiary lncrementeach worthabout7 US dollars.Hewasfriendly, outspoken, forlorn,bter, and a gentleman throughoutthe year,currently encounter.  - 18 - average extension agent, and that the VEWs' impact has been marginal. One likelypiece of evidence of this was found at the meetings with farmers called by the VEWs and AOs to present the audit mnission.The previous contact farmers, now substituted by others, 2 ' did not attend. We were told at several stops in the delta that very few of these former CFs attend any meetings with the VEWs, because they have "other pressing work". The failure of T&V to keep up was the reason expressed to the audit mission by several people in a position to have informed judgements on the matter. These include senior APAU research/extension staff and the JDAs (DOA district chiefs) in both of the delta districts visitedby the mission. The JDAs were not saying T&V was a failure, only that the way it was implemented half-heartedly in Andhra Pradesh condemned it to fail. Nevertheless, this is an area of the country where people talk about the popularity and effectiveness of radio (which transmits to farmers one hour a day) in rapid dissemination of new agricultural practices. The character of the VEW field cadre is another factor here - the low level of technical training. The agents are described as having difficultycatching up with the farmers. Thus what is in question here is whether even with adequate support T&V would have been able to take the lead. Box 3 provides an example of this frustration, from a cotton area outside the reach of the irrigation command system of the delta. Box .i Cottoninbuntu Ihra h.ratesh- X ad mion IsThe- visiedmostlythe irrigatedpy Sreas in Gut* - t 0 areas -ithedistrict,andthe mission isite on of theminhthe traditl iotn grow.. zoMe. The fdatlhdrielthedeclinedofpofaiit imortat -~~~~~~~~~~~~~~~~~~~~~pn tQttnoe evn er, ept h ~~fhyrd atl* ....... Thisyears pric clase wascatastroophi - -.byhfal Iffitom: on sn c P.stmlpln aot theneV... Nevefleesh arejhv few itboWhnsandcontinue to plat The i thextelnsionai- tsev ou.e.a the aseofthefrstwoofthe thre rmersmet,thesi the conom wer agai I .... i m u could notofferanything better.The tiprferr to staywith" c b leas I gav it t a-ont Jncone for threemorthsof theyear. The 8a that . then a1aentddnottse obelet t the OutoftroUble. The thirdi farmer wasawomane hs husband wold h on headaadndcto twoseasns earlier, andwas tryingto Md anlalternative p g.n t therains hadbeenf i excessive, soakcing . i ;. field .'the if anrd` i. . 5 ~~~~~~~~~~~~-. runin tIh,-,: crp .... ~.. -.`.i, Sh :-i -skdthe P. . i . xenlolt' g iiO -. .--.i.R lim, we "Wer d--i i. Ei wit to hepherfindanothercrop.Hesaid hewould relay. thequsin oasbjo``ttrselait ,I~k dubious, and apparentlhadasked the samee question before, 5 3.17 The PCR skips over all of these arguments. It discusses"project results' as if one need only demonstrate that the project put the extension system in place in order to claim success. The followingquote is taken from that section of the AP PCR: 1 7 Thepractice Z of replacingCF varies evenwithinthetwo states of MPandAP. Uncooperative or poorlyperforming CF will probablybe substituted. Butsomedistrictshavea deliberatepolicy of replacement - sayevery threeor four years-to giveotherfarmers a chance. There doesnotappear to beanystatewide policy. ThisIs onefeatureof T&V thathasa sinisterside. Somemission contacts saidthat many CF wereIn thatpositionin orderto takeadvantage the minikits of thatwentwiththe privilege.No minikits,no Interest. Where thesemotives operated, replacement policy wouldobviously correspond to the localDOA'splanfor distributing thesefreebenefits. 8 1 VIn Rs comments on the draftPAR,the Directorate of Extension, GOI,suggests the PARcouldhavebeenmore generous:AP's T&Vprogram"has definitely helpedin popularizing the new ricevarieties released by the State Agriculture University." Seethe Annex.  - 19 - "Overallthe project was successful in meeting its principal objectives of reorganizing and strengthening the extension services of DOA It enabled establishment of a single line extension service with functional linkageswith the research system,periodic training to field staff, placement of subject matter specialists at field level, regular village visits and close contact with the farmers (wherever VEWs were in position) and an M&E unit to assess project progress and impact. Consequently, there were significantagricultural production gains" (page 4, emphasis added). 3.18 This reads like the T&V manual, not an ex-post report. And it is contradicted by the PCR itself where it discusses the many weaknesses of the T&V system that was being installed, by the supervision reports which complain that the technical bases of T&V were not being upgraded rapidly, and by the commentators in the state, who dismiss this assumption that T&V leads automatically to impact. Later, the PCR says that rice yields have improved substantially and sorghum yields hardly at all: "the developmental changes are prominent in irrigated farming situations; rainfed agriculture is yet to transform"1-. Nevertheless, the gains in both regions are assumed to be enough to pass the ERR threshold. 3.19 The PCR concludes the project was successful not only in achieving institutional objectives, but in farm impact. However, the audit sees too little evidence to give a satisfactory rating for the AP project on either institutional or production impact. It is impossible to read through the supervision reports and other files, and visit extension field offices in AP, and come to any other conclusion than that in AP the T&V project performance was unsatisfactory. And this conclusion holds despite the remarkable efforts of APAU to associate itself with the T&V extension system and support its front line field staff. It holds also despite the fact that the Bank wielded substantial leverage in this T&V operation, committing only USD 6 million but affecting the shape and activity of a large public service. Bank experts who had worked on the project do not argue now that it is because the project was underfunded that the results were poor. The commitment factor was the decisive element. One veteran officer at APAU told the audit mission that he had been in this business for 13 years, that he had seen 13 Directors of Agriculture and 14 Secretaries of Agriculture, that only one of the directors had come from the agricultural profession, that all the other directors and secretaries were from the Indian Administrative Service and, at least in this case, that they were largely indifferent to the fortunes of T&V.M' IV. MONITORING AND EVALUATION 4.1 Table 3 shows the staff strength sanctioned at project start-up for the M&E cells and actually in place at the time of the audit. For MP it includes positions for all 45 districts. These cells were '9' The expansion of rice helps explainthe stagnation- Indeed the marked decline- in the sorghum crop. GOAP supports a socialjusticescheme that includes subsidies on the consumer was to makecheap priceof rice. Theintent rice availableto all familiesin the state. Rice has been substituting for sorghum and other dryland grains even in westerndistrictswhererice was previouslynot eaten. Drylandfarmers have been forced to move out of sorghum, APAU and DOAare trying to introduceother crops for sorghum Inthe rotation,for examplecotton and sunflower.No extensionservicecan work against strong economic signals. 2' streams of service. By comparison,he said that In Karnatakaall the directors came out of agricultural  - 20 - composed mostly of the district teams: a statistical officer and a few investigators (called Assistant Statistical Officers in MP) at district headquarters, the other investigators scattered elsewhere. The field investigators did the farm interviews and the initial recording of data. Table 3: M&E Staff Strength: Sanctioned (1982) and Actual (1993) MP AP 1982 1993 1982 1993 Headquarters Staff Joint Director 1 1 1 1 DDA - Economics 1 - - - DDA - Sociologist 3 - - - ADA - Statistics 2 1 1 1 ADA - Farm Management 2 - - - Research Assistant 8 - Assist. Stat. Officer 13 9 - - Statistical Assistant - - 2 1 Computer Operator 10 1 4 1 Divisional Staff ADA- Statistics 10 7 Computer Operator 10 10 District Staff Research Assistant 48 29 - - Assist. Stat. Officer 263 219 Statistical Assistant - - 22 20 Junior Investigator - - 53 34 4.2 The MP staffing plan was more ambitious than the AP plan, and included a large analytical cohort at Headquarters, a layer of supervisors at divisional level (about three districts per division, an administrative layer established for revenue purposes that is not staffed by DOA in AP), and a substantiallyhigher ratio of staff per district (45 districts in MP, 22 districts in AP). The Table shows the gaps in the staffing pattern in terms of staff sanctioned at appraisal (1982) and actually in post (1993). MP's proposed large Headquarters team, including the analytical positions, was never assembled. In fact one of the continuing battles between Bank supervisors and GOMP was over its failure to appoint a qualified person permanently to the post of Joint Director in charge of the M&E operation. 4.3 The M&E survey program was common to all 17 states. It was based upon guidelines (a "manual"),prepared by Bank staff in 1981IL, which had themselves been abstracted from a 1977 Staff The MonHtoring 21/ and Evaluationof Trainingand Visit Extensionin India: A Manualof Instructions.Slade,R.H. and Feder,G. October 1981.  - 21 - survey program were discussed Working Paper.A The philosophy of T&V, M&E and the proposed period 1981-1982by the Bank's at a series of regional seminars planned and organized in India in the Ministry of Agriculture. A Regional M&E advisor and the Directorate of Extension (DOE) of the and tried to secure state Monitoring and Evaluation Subgroup of DOE oversaw the survey program in submitting the reports on the guidelines and cooperation in carryingout the program according to time. seasons - kharif and 4.4 The program consisted of two surveysfor each of the two main cropping a two-stage stratified sample, rabi. The first survey, pre-harvest, was the Monitoring Survey. It used numbers of contact farmers (CF) and the first selecting the group and the second providingequal Survey. other farmers (OF). The second survey, post-harvest, was the Monitoring-cum-Evaluation and added the crop of farmers, The second survey repeated the questions of the first to a new sample for the crop had been selected cuts. In fact the second sample was drawn from farms whose fields The size of the samples was cut. These were carried out by various agencies (see below). For the Monitoring determined by the rule that the standard errors be kept below 5 percent.n the district level. For at Survey, the objective was a sample large enough to provide that accuracy the sample was large enough the Monitoring-cum-EvaluationSurvey,the objectivewas to ensure that The established, all-India, to meet the 5 percent standard error test at the state level (for all data). (GCES) - was to be the main instrument crop cut sample survey - the General Crop Estimates Survey estimates at the state for the T&VIM&E field survey. GCES was designed to provide significant by the T&V/M&E demanded rather than the district level. But with the level of crop specificity by the M&E cell.] Survey results were compiled design, the GCES usuallyhad to be "augmented" direct to Headquarters (AP). at the district level and sent through the divisional office (MP) or level and "verified" at state Separately, the GCES crop cuts were "scrutinized" at the district was consolidated and sent to headquarters of the statistical agency. At T&V headquarters, all data operation, and by the late the Ministry and the Bank. In theory, each T&V state that had an M&E two months of the end of within 1980s all of them did, should have prepared four reports per year each of the surveys. each survey and 4.5 The Bank-proposed M&E program called for the states to analyze data in the final survey report. incorporate the findings and recommendations along with the tabular - of problems exposed by the The plan also called for the M&E cells to carry out special studies T&V program. All such reports surveysand of other subjects important to evaluating progress of the and expected to be discussed were aimed in the first instance at management of the extension service, Extension Aaricultural Cernea, Prolects. M.M., and Tepping, B.J.SWP No. 2 A System for Monhorina andEvaluating Thetitledoesnot revealthatthe system wastargeted at the Indian T&Vsystem. 272. December 1977. testfor ERR forthe conventional T&Vproject 23/ Notethatthe5%accuracy testwasovertwicethesizeof thethreshold (seepara3.6). the answers whether fromthefarmsin the augmented" sample actuallyentered 2V Themission couldnot determine - LRD in AP andthe datareported by the M&Estaff. Theagency responsiblefor state-wide statistics the district-wise IADP program areaIn MP). The Bureau in MP- didnot usethem(withthe exception of the data fromtheold Raipur wasusedmostly to crosscheckthe main, GCES sample. Other Informants mission wastoldtheaugmented sample saidthat was not so.  - 22 - in district and block level meetings with extension field staff. But they were to be in English, and distributed as well to the Ministry, the other states and the Bank. 4.6 Most state M&E cells had difficultyin maintaining the schedule for submitting data from the surveys,though eventually, and sometimes two years late, the four annualthe tabulated reports were sent. Most state M&E cells also performed poorly in identifying patterns in the surveys - favorable or unfavorable - or extracting lessons from them. Some states prepared many special states prepared almost none. MP and AP rank below average in all three respects: studies, some maintaining the survey reporting schedule, analyzingthe results, carryingout special studies. In fact MP is considered one of the worst M&E performers in the country. For its global study of M&E, OED recruited an Indian consultant firm to review the experience of a selection of states with T&V/M&E, to try to draw lessons about why a few states did better than others. The firm selected MEP, among eight studied, as an example of poorly performing M&E. AP was not on the firm's list, but its reputation for M&E is not much better than MP's. 4.7 There were two main problems. One was embedded in the questionnaires, and can be traced back to the simple prescription for M&E embodied in the 1977and 1981 Bank proposals. They refer to three zones of concentration, or priority areas of survey inquiry, which were seen to be "the irreducible minimum if monitoring and evaluation is to be efficient and effective."5' These were the visits,standing as a proxyfor "the entire endeavor of institution building undertaken by the project", the adoptionof recommendedpractices, and yields. The visits were measured by farmer recall of the name of the VEW and the number of contacts. Adoption was measured by farmer statements about whether he knew of the practice, and whether he had used it. Yields would come from crop cuts. The problem arose because the visit indicators were unable to measure the quality of the extension contact, so that apparent positive correlations of contact with adoption and yield were unconvincing and were not paid much attention. Also the adoption questions had to be handled in the interview ,ithsophistication, because casual responses about what farmers knew and used could be misleading. The field investigators who did the interviewswere in the majority not up to this standard, and the value of the response wasjeopardized. The matter of the quality of the crop cut is discussed below. Finally, in this set of issues concerning survey design, there could be no purposeful selection of contact farmers for the Monitoring-cum-EvaluationSurvey, even though these farmers were the principal point of entry of the recommended practices (the methodology for augmenting the sample allowed some room to redress the imbalance). Thus, the tabled crop yield data was largely at the mercy of the luck of the GCES sample as to how many contact farmers were included among those interviewed. 4.8 The second main problem was that the raw data was not being used as the beginning of an analyticalexercise, but as the end of it. From the initial Bank supervision of the two projects, there were recurring complaints that the data was being handled mechanistically. Neither the interesting observations that jumped out of the tables - all of which followed a prescribed format - nor troubling inconsistenciesthat were also thrown up by the data, were pursued. For example, the in the MP survey reports the tables occasionally show that other farmers surpassed contact farmers in number of extension contacts, that other farmers with fewer contacts had yields superior to farmers with many contacts, or that contact farmers with many visits from the VEWs had lower yields than 25J Slade and Feder (page 4), basedon Cernea and Tepping. See footnotes21 and 22.  - 23 - those with one visit. An officer in one MP district explained this last result by saying the irregularity of the frequent visits caused the transfer process to be worse than no visits at all! these occasional inconsistencies was that nobody paid them any attention. The problem with The evidence was counterintuitive, but the M&E cell reported it without comment. There are many anomalies in all the reports which should have called at least for an explanation, if not for a study, and did not. 4.9 An equally important feature of the formats for the four yearly survey provided no structure for showing year-wise results, so that progress reports is that they over time could be assessed. Perhaps other states improvised, but MP and AP did not. The ADA-Statistics handedly trying to keep the survey operation going had, himself carried in AP who is single- figures, which he provided to the audit mission. But they were not called worksheets with year-wise for by the Bank's proposed M&E survey format, and so they were not routinely tabulated. In fact that informants said: that if the Bank had demanded a series of annual figures, is exactly what several they would have been incorporatedz'. There are some exceptions, even in MP and AP, but few. One occurred in 1988, when Kerala issued a special study called An EmpiricalAnalysis of T&V in Kerala, providing trend data (without much analysis). Seven other states including MP and AP repeated Kerala's study, in a once-off exercise. 4.10 Despite that 1988 activity,the record of these two states in carrying out special studies of any kind is terrible. MP has done a total of four studies, three of them in 1981 before the audited project began.A' AP has done two studies. These totals include the 1988 Empirical studies. They can be compared with the most active state M&E cells. Tamil Nadu, which is complimented in the OED consultant's study as the best M&E performer among those reviewed, has produced 48 studies. Kamataka, visited by the audit mission,has produced 28 studies. Haryana, another good exampleof T&V/M&E, has produced 26. Some of the subjects dealt with by the Kamataka M&E cell are exactly the type one would expect an aggressiveM&E unit would want to tackle: of productivity trends of key crops, of yield potentials and yield gaps in drylandcrops, of low productivitysubdistricts, of the impact of T&V on adaption rates by agro-economic region. 4.11 The T&V and M&E management teams in MP and AP show themselves by their inattention to have been indifferent to the possible uses of the surveys,or of their cells' capacity to carry out special evaluations. The ex-head of the M&E cell in AP, who had been responsible for establishing that unit and guiding it through its first six years (and who has a good reputation in T&V/M&E circles in India), told the mission, when asked why he did not do more studies, that if he delivered the four primary survey reports on time he felt he had "fulfilled his responsibility".2 The senior officialof MP's T&V operation was more imaginative- in three statements. He said that the cell had done other studies, but not in English and not available in translation. He said that M&E was not & TheBank's1981 Manualsuggested thattablesshowing year-to-year trendsbe constructed, sayingit wouldbe a trivialmatter' to designthem(page51 of the Manual - seefootnote21). But It did not includerepresentative tables. trend 2/ Theauditmission wastoldthata thirdspecial study- of extension impact on womenfarmers - was underway. 28 There is onereference, in Bank filesof 1985, to thegooduseswhich regular thissamepersonmadeofthe findingsof the surveys.Thissuggests,assuming thecomment wasaccurate, thatperformance dueto the repetitive mayhavedeteriorated, perhaps character of the surveysandtheirresuRts.  - 24 - really essential in MP, since the extent of T&V's impacton crop production wasso large and obvious. And he said that the three sociologistpositions in the M&E staffing pattern - that had never been filled - had been unnecessaryin the first place: "since we are all sociologists".& 4.12 These two cells produce only an illusion of M&E. The district M&E staff, and the ADA responsible for visiting the districts to keep the teams moving on schedule and getting the samples and crop cuts done, are dedicated, hard working, poorly compensated, and wasting their time. A resource committed to M&E and capable of bringingsome needed feedback to the program is largely ignored. Repeated Bank supervision requests to T&V management to get the cells to use the data properly met no apparent response. 4.13 The M&E exercise in these two states - and supposedly in some of the other states - was seen by T&V management and the DOA itself as a commitment to the Bank, and secondarily to the Ministry, and not as a source of lessons for the state program itself In fact the senior M&E operator in both states complained that the cell was being used by T&V management to support the positions it wanted to present, and to suppress at times disagreeable material. This is a darker corner of the T&V/M&E experience which the audit did not explore.3N 4.14 The preceding comments do not address the problem of data corruption through error or abuse. The audit mission also did not investigate the quality of the crop cuts, and crop cut reports, or the plausibilityand recording of all the other questions and answers in the four annual surveys. There is no evidence in the files that the Bank, either Headquarters or the Resident Mission, ever investigated this issue either. There is no shortage of statistical talent in Indian states (the crop cuts go back to the Moghul period) or in the offices of the Land Records Department (MP) or Bureau of Economics and Statistics (AP) that are responsible for administering GCES in the two states. Nevertheless, crop cutting and proper validation and recording of data require special skills. The mission found evidence suggestingthat the system set up to ensure the data was accurate has been compromised. For example, travel funds for both carrying out and supervising the crop cut were insufficient. The LRD office in each district, under instructionsfrom its headquarters, would allocate the cuts authorized for each of the selected priority crops to its own staff, to the T&V staff, and to other field agencies. In MP the T&V and/or M&E staffs were responsible for supervising the local LRD or revenue ("patwari") officers who performed the crop cuts, especially of the augmented sample. They should sign off on the three forms givingthe specificationsof the field and the weights of the measured crops when green and after drying. In AP, the VEWs, and not the M&E field investigators (who were nevertheless responsible for the interview surveys),did most of the cuts on behalf of the extension service. However more senior T&V or M&E officers were obliged to observe the cuts and sign off as well. This is where slack in the system threatened the integrity of the results. had been in chargeof M&E ' This officer foronlythree months at the timeof the auditmission,althoughhe had been in DOAmuch longer. Hisviews presumably are not representative,though they callto question his appointment. 3' The missionalsodidnotlookintothe issue of whetherthe M&Ecellshould havebeen locatedelsewherethan where it was. Inboth states the cellwas placedunder the JDA- Extension.The Bankhad arguedthat the M&E cells in India should reportdirectlyto the Secretaryof Agricultureor at least to the to Director, ensure Independence.  - 25 - Given the immobilizationof the VEWs and their supervisors, as well as the M&E district staff; the way was open for errors to pass unnoticed. The mission was told in one district that only about 10 percent of the crop cuts was supervised by anybody. 4.'5 The Indians take crop cuts seriously, and the audit mission was told there is little danger of the investigators faking cuts and of outright falsification of results.l' But supervision is an essential part of the protocol. Audit questions about the M&E team's capability to properly carry out its survey responsibilitieswere usually met with embarrassed glances at superior officers, making it clear that compromiseswere indeed being made. This could be a recent phenomenon, associated with the decline in the operational budget. But it needs to be investigated. As mentioned above, there is no sign that the Bank has pursued the matter. The M&E Subgroup of the DOE in New Delhi has not done it either - or got the capacity to do it. They appear to share a common assumption that the system works well enough anyway - in the absence of full supervision. The audit considers that assumption dangerous. 4.16 In recent years, the Directorate of Extension has moved to correct some of the deficiencies in the M&E program. Partly from Bank/New Delhi urgings3', it commissioned a group of studies in 1990 and 1991, and was about to undertake several more at the time of the audit mission in October 1993,to look in depth at the issues of most importance to the proper functioning in India of T&V and T&V/M&E. Subjects of the earlier batch of studies included Research Support to Extension, Feedback Provided by Extension Functionaries and its Impact on Research, and Input Consumption and Production Trends Before and After Introduction of T&V. One of the two newer studies was to be the Quality of SMS Support to AgricultureExtension. The audit mission reviewed the first batch and found the reports of good quality. Another report, by the consultant firm CARDS, dealt exclusively with the performance of the M&E program: Functioning of State Monitoiing and Evaluation Units.3yThis is another good report. It calls for the movement of the survey routine away from descriptive to diagnostic reporting, and it recommended changes in the survey format. The revised format was piloted in one district in each state in 1991,and was supposed to be implemented fully starting with the kharif season of 1993. The descriptive and analytical 3*/The audit,is uneasy about this assertion,though the missionfoundno solid evidenceto challengei. The mission heard tNoreasons whythe official yieldstatisticsmay in fact be influencedby 'external,pressures. One has to do with the formulasfordistributing GOIsubsidies to the states, whichfavorthose states demonstratingshortfalls fromnormal or projectedcrop productionlevels. Theother has to do withthe traditlonal method forpresentingyields,whichuses an indicatorthat expresses yields as a percentage of some Nmaximum" yield,set in earlierdays, that does not allow a figureto be entered in excess of that maximum.Thus actualyields havean artificial, finiteupperlimit. Whetherthese factors are present, and indeed Influencecrop statisticstabled In the M&E survey reports, Is not known. 32 AKthough the Bank has provided no technicalassistance to the T&VIM&E program since the departure of the RegionalM&Eadvisor,the Resident Misslonhas kept steady pressure on the state T&V authoriies to carry out and makeuse of the fieldsurveys and to expand the numberof special studies. Italso pressured GOIto strengthenthe Directorate ot Extension, and workedwith DOEto developa series of country-wide seminars startingin 1988and, as mentionedIn the text,the consultant reports carriedout in the early 1990s. V3 Studies. NewDelhi,1991. and RuralDevelopment Centrefor Agricultural  - 26 - material in the CARDS M&E report included MhP and AP data. It was useful to this audit as well as OED's M&E study (para 1.3).23 4.17 In fact, OED contracted the same consultant firm to do a repeat study of M&E in the Indian T&V system, to try to discern the differences that explain why some states did relatively better than others in organizing, implementingand using M&E. That new report, available in OED as a Working Paper to OED's M&E study,3' mentions the importance to success of high-level support within the state govemment. Support seemed to be based more on the personalities involved, and could not be predicted easily. That analysiswould fit the profiles of MP and AP. V. FINDINGS AND LESSONS A. Findings 5.1 The T&V program provided an alternative to the community development program. The latter had taken over public services for the rural population, incorporating the national extension service established in the mid-1950s. But it had put the emphasis not on extension per se but on the promotion of well-known packages of production inputs - usually associated with the OGreen Revolution" and invariablysubsidized. It had also moved on from agriculture to other social services. The extension program of the 1950s had been created at a time when there were few new technologiesof any sort to promote other than the irrigated high yieldingvarieties. By the time some newer technologies began to appear, the VLWs had turned much of their attention away from agriculture. T&V forced agricultural technologies back onto the agenda, without challenging the community development activities which carried on as before. In fact with the advent of the Integrated Rural Development Program (IRDP) program in 1982, there was supposed to be enough work to keep both T&V staff and the VLWs busy with agricultural operations. There are now large state-financed field staffs dedicated to transferring information about improved technical packages to small and medium scale farmers. The central and state governments - including Madhya Pradesh and Andhra Pradesh - see this as a welcome development. 5.2 The T&V experience in these two states is only a part of the national program, and the lessons are not fully transferrable.3' Nevertheless, since they are among the biggest states in India, and since they are among the three earliest states to adopt the T&V methods, their experience ay The consultant report also proposes that six new Progress Reports be introduced into the M&E routine at statelevel, one each for the VEW, AEO, SDAO (subdistrict), DDA/Extension, SMS, and Master Trainer (the research link). The PR are to be prepared fortnightly or monthly, and were put into effect in late 1993. These new forms are additlonal to the enormous reporting requirements already imposed on the field staff. The audit did not review systematically the coverage, overlap, relevance, use, and quality of this gigantic flow of reports - already overwhelming before the new forms were put into play. 35/'India: Performance of State Monitoring and Evaluation Units under the Training and Visit System of Agricultural Extension". CARDS (op.cit.). OED Working Paper. 1994. 3' For a good description of T&V throughout India, its strengths and weaknesses, and recent changes in the blueprint, the reader is referred to the 1992 report by Macklin (footnote 4)  - 27 - deserves attention. At the least it can help illuminate the advantages and disadvantages of this popular extension model. The recently completely OED study of the worldwide T&V experience ("Lessons' - see para 1.4) lists and discusses the favorable and unfavorable factors. The correspondence with the MP and AP cases is remarkably high, further evidence that one is dealing in this audit with common T&V experiences and problems. 5.3 The followingeight points are drawn in their original order from the list of ten performance findings in Lessons (pages 30ff). The global findings are put in quotes, and compared with the MP/AP experience in follow-up statements. The discussion in Lessons was based on a review of extension programs of all kinds - not just T&V - since the OED study dealt with the universe of Bank-supported extension activities. But 90 percent of the free-standing projects in the Lessons reviewfollowedthe T&V model, and therefore the analysistracks the good and bad outcomes of that special system. The use of Lessons as a reference in this audit report is therefore appropriate. 5.4 * 'The Bank's support for extension in free-standing and component projects has substantially increased the attention given to agriculturalservices, especially in the smallholderpopulation". As just stated, this is the most important result of the MP and AP programs. 5.5 * "Thefocus of most investmentson smallholdersand improvementsin the organization and programmingof serviceshas enabled Increased coverage of smallholdersby public sector extension services". All informants in the two states would agree with this. The relationship between the VEWs and the contact farmers has been strengthened by the practice of the former providing the latter with free "minikits"of seeds and pesticides distributed under the national and state crop and input campaigns. This resembles the practice under the community development program of distributing subsidized assets and inputs to beneficiaries of its activities. But T&V was not uniquely tied to these free minikits, and thus its outreach could be, and was, broader. Since the area of overlap of beneficiaries between the two programs is small3', T&V has greatly extended the numbers of families benefiting one way or another from these public services. The contact farmer approach has been abandoned in some states, in favor of group contacts, because the relay of messages by contact farmers to other farmers was not working as well as expected and the strategy was seen to be limiting even greater increases in coverage.A' MP and AP are also discussiig more frequent use of groups, though the contact farmer model still dominates. 37 T&V contact farmersand their closestfarmer friendswere usuallytoo well off to qualifyfor the subsidizedinputs under the communitydevelopmentprograms. 3U Intheory,theVEWshouldvisit thefields of each of his CF,entertaining questionsalso fromotherfarmerswhoJoined the walkinglesson, and later In the day hold a group meetingin some common spot. In fact, the contact pointhas usualtybeenthefarm visit,withthe singleownerand perhapsa couple of friends. Thegroup aspectsalreadybulftInto the originalT&V design have not been respected- or effective- in practice.  -28 - 5.6 * "The majority of borrowershad serious difficulty in meeting the mrcurrent cost exuenditures of national extension services, to the extent that in many cases the sustainabilityof the instituted systemswas in doubt". This is already a debilitating problem for AP. It is less serious in MP, although the PCR on NAEPI calls attention to the fact that GOMP had not yet transferred the MPH program costs from the 'plan" to the 'non-plan" budget, where it would join other programs fully absorbed into the normal state budget. The two states have maintained the T&V staffing cadre close to levels called for during implementation. But gaps in postings are increasing, especially in AP. Of greater immediate concern is the gradual reduction in budgetary allocations for transport and other operational expenses, emasculating the field staff. Signs of this problem are appearing in MP, though it is far less serious there than in AP. 5.7 * "An insuMciency in technology which was relevant to farmer circurstances was a major constraint in many services, and was particularlyprevalent for resource-poor environments". The comparisons between central-west MP (soybean), eastern AP (paddy), and western AP (groundnuts, etc.) illustrate this point well. T&V, even supported by three powerful research establishments, was relatively helpless in making headway in the AP drylands. Macklin, in his comprehensive report on T&V in India, says the same thing: "the generation of improved relevant, risk-tolerant technologies for rainfed agriculture has been given far less emphasis".&The other two cases are just as instructive. In eastern AP, where releases of new paddy varieties, sequential pulse varieties and associated pesticide applications are continuous and quickly spread, the undertrained village extension workers and subject matter specialists have a hard time keeping up with the farmers. In MP, however, T&V extension does appear to have played a useful role, accelerating a process already activated by other catalysts (in this case, high prices and aggressive buyers from oil seed processors). Lessons hints at a typology of conditions appropriate for T&V: it is good for spreading new crops and cropping systems (including irrigation practices) that offer a quantum leap in technology or practice, especially if they have not yet been fully embraced by commercial dealers; T&V is largely redundant in the spread of profitable new seed varieties; and T&V is ineffective in the drylands in the absence of sparkling new packages. The MP and AP stories fit that typology. 5.8 * "Apoor understandingby researchers and extension staff of the circumstances of farm households (ie a lack of farming system perspective) and poor linkages between extension staff and farmers". researchers, Neither of the two states fully fit this description. The most important cases of technical progress observed during the audit - in the soybean zone in MP and the paddy command areas of eastern AP - are testimony to the general strength of the research contribution in India, to its links with T&V extension (especially in AP), and to the concern for farming systems where these were relevant to technical progress. Thus, the conversion of kharif fallow to soya cropping, the development of quick maturing varieties to permit the rabi planting in sequence with soya, and other adjustments in the cropping rotation, were correctly aligned with the possibilities for improvement apparent in the Page13. -WOp.Cit.  - 29 - traditional farmingsystems. The possibilitiesshift dramaticallywithin short distances in MP - mustard, sorghums, pigeon pea and soya itself coming in and out of the optimal pattern as the soils and rainfall patterns vary. The research establishment has been trying to fit new, more profitable patterns to every niche (though, as mentioned above, Bank supervision has not been impressed with state-wide results). In eastem AP, however, with a dominant paddy culture in the wet season and a second paddy crop in the dry season where irrigation permits, research (and extension) have concentrated on improvements in paddy varieties to fit different areas and rainfall regimes. That focus - with attention also to the rotation crop and less work on unirrigated systems - does not seem inappropriate. 5.9 In one respect this concern of Lessons does apply: the apparent lack of awareness by extension staff of the principles of farm economics, as sometimes revealed in their technical messages. The continued promotion of practices with limited benefit/cost ratios in western AP is indicative of this deficiency. 5.10 *!Y"Staff gualitv was a major constraintin nearlyall of the free-standingprojectsand in others which requiredvery largenumbers of staff Most of the free-standing pnjects provided for continuous technical train ng for extension agents, ;v.hichwas especially important due to their usual relativelylow level of education ard basic agricultural training. However, serious difficulties occurred in having insufficient subject matter specialists who had adequate technicaland practicalexperienceto participateeffectively in regularstaff training in the state-wideprograms, so that training was often overly theoreticalor too generar. These problems are clearly reflected in the MP and AP experiences and may be the most damaging. AP suffered especiallyin the ranks of the VEWs, since it was denied the better educated VLWs from the communitydevelopment program, denied also new recruits, and had to fall back on its existing junior field staff. The SMS problem was severe in both states. It is not clear whether the Bank should have anticipated that the T&V standards for the SMS position would be rejected in both states because of civilserviceprocedures. Instead of new recruiting, or selecting promisingcandidates from within the civil service for further training, the DOAs had to offer the newly established SMS positions - and the promotion and salary advance they carried - to aspiring AEOs of any professional background. There is no way at present to retain a staff member in a particular stream such as agronomy or plant protection and still guarantee him equal speed in moving up through the ranks. Promotion depends on seniority, not achievement, and usually takes the officer to another post.1' The SMS position was corrupted, and that removed one of the fundamental pieces of T&V architecture. Given that the AEO ranks were also substantiallyundermanned in both states, relative to the VEWs, the whole support system for the front-line extension force was compromised. The supervision reports for both projects recognized this weakness and asked without success that the gaps be filled. The optimism of the PCRs stands in striking contrast to the supervision reports and the reality on the ground. 4' Twootherissuesraised aboutparicpation in Lessons, fromthissequence. areomitted culture, and top-down! 4tl Seefootnote16.  - 30 - 5.11 * "It was apparent that unless there was solid commitment by senior officials of central and implementing agencies to implement the improved service, operational problems could be expected in not only funding support, but also in approvals for recruitment, procurement and establishing satisfactory terms of service". This was a problem more for AP than MP. There is good evidence that the lack of ownership in AP can be traced to the Agriculture leadership as well as Finance. Also, the Chief Minister who assumed power in the first year of project implementation then subjected all services to the arbitrary demands of his political enterprise. In MY there were no equivalent enemies or shocks. In fact MP supported T&V through four, consecutive Bank lending operations. However, the officials in Agriculture interviewed in both states in 1993 seemed genuine in their respect for the T&V model, although they wanted more flexibility and commiserated with extension staff over the lack of budgetary resources. 5.12 * Finally, among the points discussed by Lessons, "in general, monitoring was weakly developed in free-standing projects". Chapter IV discusses the M&E component of T&V in the two states, one of the most depressing parts of the projects. The main fault was that it was never turned into an instrument of use to DOA and T&V management. It was seen primarily as a service aimed at the Bank and GOI. 5.13 Some recent Bank reports on the T&V extension program have mentioned many of these problems and recommended that if they had not already been corrected they ought to be. The conclusion of most of these analyses however was that the rigidity of the T&V model had been exaggerated and most of the faults have been recognized and are being removed. The T&V principles were still valid and the program was still on track. The Region, in its comments on the draft PAR, also argues that the introduction of the T&V extension system in the two states in the early 1980s was 'an appropriate response to the prevailing circumstances at that time.9 Lessons appears to accept T&V as an appropriate model for regular face-to-face extension where justified, provided modifications are made to make it more responsive, although if one were to sum the proposed adjustments it would be hard to recognize the modelly. 5.14 It is worth revisiting these common perceptions. The MP and AP T&V experiences resemble those in other parts of the world and demonstrate the same miscalculations in the T&V approach. The two most important are (1) the assumption that profitable new technologies and husbandry practices are everywhere present in rechargeable numbers large enough to legitimize the creation of a ubiquitous, multipurpose, staff intensive, extension service, and (2) the assumption that the service will be staffed by properly trained persons at three critical levels: the VEW; their supervisors, the AOs and AEOs; and their technical support cadre, the SMS. Both of these assumptions proved nonviable in the majority of areas serviced by T&V in the two states. Other assumptions could be added here, for example that the states would maintain the budgets for operating costs, and that an M&E operation would be well staffed and steer the program away from error. These two wVMemo, G. Mukanl to G. Donaldson, August 24, 1994. 3y Lemonsalso pointsout that fiscal,humanand technology resourceconstraints do not alwaysIndicate that a 'permanent' face-to-face service Is warranted In all areas throughout a state.  - 31 - assumptions were also in jeopardy. But in these cases the constraints were "manageable" and the results could have been better, whereas with the other two factors the problems would seem to be inherent in the system and irreducible. 5.15 The Bank moved in India to help correct a real deficit in government's services for small and medium scale farmers. Increased farm production and farmer incomes implied improved farming systems. The communitydevelopment program as it was implemented in the mid-1970spaid too little attention to the technical requirements for extending the "green revolution" outside of the HYV packageswhere it started. There was a need to force the institutional establishment to recognize the deficit, and the T&V intervention served that purpose exactly. 5.16 But the Bank should have been more adept in supporting this reform. It hit hard with a single model that was overly rigid and mis-specified,at least for Indian conditions. Some of the personnel problems - for example with VLW transfer in AP and SMS postings in both states - should have been better anticipated, but were bypassed to protect the theory. There were strategic alternatives to the T&V model. Four can be cited. 5.17 One would have been to work within the structure of the then existing community development program to turn its attention back to agricultural technologies. In fact most of the VLWs were working in agriculture, and were better trained to do so than the VEWs in the DOA extension systems. The VLWs still do work in agriculture, for example in distributing pumps and pipes, sprayers and bicycles,and draft animals and ploughingequipment packages under IRDP. Audit meetings that brought together VEWs and VDOs were strange events, because both cadres talked as if only the first was handling agriculture, while the other was doing something else. What they meant was that the VEWs were doing the only bona-fide extension work, as defined by the T&V model. 5.18 This distinction must be discussed in more depth. The VEWs are associated with the minikits of seeds and pesticides from the central and state crop improvement campaignsWwhich they distribute in the first instance to the contact farmers; the VLWs are identified with the packages of capital assets which they distribute to farmers selected for the "subsidy"programs.4' It would appear in most districts that T&V staff do get involved to some degree with IRDP and the other "subsidies" - helping make the farm plans, distributing the equipment, and inspecting the assets once in place. For the T&V supervisorystaff this is a normal requirement of their positions in the hierarchy of DOA, which loads on them assignmentsoutside of T&V but well inside the responsibilitiesof DOA-.t' But the reverse is not common: the VLWs do not normallyinvolve themselves with the minikits, or with any technical extension work. In the western region of AP this distinction has led to a remarkable situation where, for example, an AO who was appointed to the position of Mandel Development 44Forexample,among the'Centrally Sponsored Schemes",theIntensive Programfor RiceDevelopment, the National Programfor PulseDevelopment,the Oilseeds Production Program andthe Spices Development Program. 4' Forexample,amongthe GOIprograms, agricufural inputs andservices delivered withinIRDP, theScheduled Caste ActionPlan,the Backward CasteActionPlan,andtheScheduledTribeActionPlan. 4eThereportdoesnot discuss theseextraassignments. Theyinclude, for example, dutieswelloutside bothT&Vand IRDP, suchas inspection of licensedpesticide, fertilizer and machinery dealers.  - 32 - Officer while retaining the AO post in DOA uses his few VEWs (T&V) to help with IRDP bLt does not use his more numerous VLWs (DRD) to help with the minikits. The VLWs are deliberately excluded from the technical diffusion process, even though they are better trained than the VEWs. on This bifurcation of responsibilitiesis the outfall from the introduction of T&V, with its insistence to the purity of the diffusion process. The best trained field staff (VLWs) are now confined distributing assets to marginal farmers, while the undertrained VEWs are responsible for technical transfer to the farmers with better resources, in particular the contact farmers. One could argue that those roles should be reversed. The potential for this confusion could have been foreseen, had the appraisal team looked behind the T&V blueprint. The mix-up is most apparent in western AP, but of its elements are everywhere present. There is clearly a need to redimension the responsibilities been a way to make better use the whole of the field operation serving agriculture. There must have of extension techniques within the existingcommunitydevelopment structure, although that of course would have required crafting new relationships between DRD and DOA 5.19 The second option would have been to build upon the extension structure already developed at the state agricultural universities. India had rejected the full-blownAmerican land grant university system, where teaching, research and extension were all assembled under one authority. GOI had accepted that a strong research capability could usefully be incorporated in the SAUs. GOI also and encouraged the SAUs to develop their own extension wing, with their own, qualified SMS extension agents. But this activity is confined to extension of the technology releases by the research wing. Although the SAU extension work is in principle competitivexviththe DOA extension service, in India, and in particular in MP and AP, it has not created any outright conflict (thanks to common sense in both agencies). Constructive institutional arrangements could have been developed to bring of the SAUs more directly into the recruitment, training and backstopping of the expanding cadres both SMS and extension agents that were to be created under T&V. Benor and the Bank were willingto consider that the SAUs be responsible for the SMS cadre, an idea that GOI as well as the in DOAs rejected. But the Bank could have pressed anyway for better integration of the SAUs strengthening the SMS role. This would have avoided one of the defects of the T&V model cited by Lessons and mentioned above: sending extension agents out everywhereregardless of whether they had strong profitable packages to deliver. 5.20 The third option would have been to allow extension to retain its rights to promote and help organize, if not actual carry out, the distribution of inputs as well as technical advice. This is anathema to the T&V school, which wants to protect the extension agents from any obligations that interfere with straight informationdiffusion. But it would have given greater legitimacyto the agents in the areas where there was not much new technology to extend, and it would have enabled agents throughout the states to back up their messageswith reliable promises of available purchased inputs. If there was one common complaint about the T&V model the audit mission heard throughout its interviews - in New Delhi as well as the states - it was that the input function should not have been disassociated from the extension function. The 1990 study commissioned by DOE on Feedback Providedby Exvension(para 4.16) found that the most recurring problem reported by extension agents was the non-availabilityof inputs at the opportune time (page vii). Actually the states - including MP and AP - maintained within the DOAs a small, parallel cadre of input delivery agents, to support the T&V field staff. The argument from the states was that this separation was nevertheless superficial and inefficient.  - 33 - 5.21 The fourth option is an enhanced version of the third. In this case, the extension service and inputs are uniquely associated with specific campaigns, which feature a proven technology and well- targeted farmers and farming systems. The extension agents in this regime are not distributed uniformlyby district, forever, but gathered as needed in squadrons according to the physical and temporal needs of the campaign and only when there is something worth campaigning for. Campaigns can be strengthened by the rigor of a T&V protocol. And T&V can take over and execute campaigns after the technologies are confirmed. But in principle campaigns and T&V do not fit together. Campaigns target inputs or technical packages at farmers and farming systems in favorable situations. T&V offers whatever it can to all farmers47'.In India, T&V is operating in a countryside awash with campaigns, and has to accommodate them. Some T&V supporters say the campaign mentality is by nature top-down, and that campaigns in practice do not allow for two way communicationwith the farmers - two of the faults of traditional extension which T&V is designed to eliminate. Properly designed campaigns can overcome both these constraints - the first, by basing themselveson thorough farmingsystems analysis;the second by planning for high levels of beneficiary participation. That T&V itself escapes those criticismsis also untrue, as the OED report Lessons makes clear-&. 5.22 The audit is not recommending any one or combination of these options. But it is clear that there were alternatives to the basic T&V model, and some combination of the best of the T&V principlesand these other systemscould have been launched with a high likelihoodof superior results than what wasobtained. In answer to the criticismthat this is unfair second-guessing,the audit would maintain that the alternatives were not even considered - were simplyswept aside by the dramatic promises of T&V supporters. Of course training, and visits with farmers, are essential to any enterprise extending new technologies. But that was not enough to guarantee results, as the case studies of MP and AP demonstrate. The fixation on the simple model incurred costs and distracted attention from other essential elements for over a decade. 5.23 The notion of T&V "incurringcosts" is worth examining. The ERR calculations assume that T&V can at least claim the equivalent of an increase of about 1 percent at full development in crop yields across farming regions, and that is enough to give 50 percent rates of return on project expenditures. The calculation is self-serving. All it saysis that there are feasible adjustments to make to traditional farmingsystemsthat can have substantial payoffs,some only requiring low-cost changes in current husbandry practices (seed selection, seed spacing, planting depths, etc.), and others involvingmore expensivepackages of new technologies requiring cash inputs. The T&V extension program claimed just enough of those economic benefits to justify itself 1 ' But one can imagine a host of other interventions that could have had at least as good and maybe better impact. Radio 47Atthetimeof the auditmission, a newformulation of GOIpolicywasunderconsideration whichwouldconcentrate T&Vserviceson campaigns targetedonlyat marginal farmers.Thatlastconditionmayweakenthe contributions of boththe campaign andtheT&Vapproaches. 4' A variation of thisoptionwouldbe to havean improved T&VsystemIn locallties whereconditionsareappropriate (includinggoodprospects forcontinuous supplyof relevant technology), butto relyoncampaigns in otherareas where few profitable Innovations canbe expected. 4 Apparently T&Vextension wasnotthefirstBankprogramto usethisapproach to theERR.It hadbeenusedearlier for someof the Indianagricultural creditprojects.  - 34 - campaigns, mobile video shows, farmer training centers, minimum price support programs strictly applied, credit to processing factories for workingcapital to purchase raw materials from farmers, are but a few of the options that could have excited the farmers to do some things differently or to take on new tools. The soybean story shows where the real incentives for adoptions of new farming systems lie (para 3.9). 5.24 So the issue is not whether T&V can claim a 6kg increase in paddy yields in the deltas of Andhra Pradesh, but what else could that state have done with even the few resources it did commit to T&V. The Bank has to be concerned with the opportunitycosts to the T&V program, and from this perspective the audit suspects that the costs of dedicating one of the largest civil service cadres in the states to this stylized, simple extension model were not worth the results. The two questions that should be asked are: (1) whether, if a combination of well trained VEWs, AEOs and SMS could be mustered, if a good extension/research linkage could be secured, and if adequate operating costs could be guaranteed, was the unalloyed T&V technology-transfer system the best way to make use of these resources; and (2), if none of these conditions could be met, did the T&V system nevertheless make any sense. Those questions should have been asked before embarking with this model, and were not. 5.25 GOI and the states are presently asking themselves how to get more out of the T&V apparatus that has now been installed. The most common criticism at senior levels of government - in GOMP, GOAP, and GOI - is over the rigidity and ritualismof T&V. The hierarchical structure of T&V fits well with the hierarchical structure of the departments and civil service, but these other two features were alien. The governments recognize that T&V as originally presented is too costly for the states to sustain indefinitely. Substantial modifications are under review, some already pioneered by the leading states. These adaptations include the abandonment of any pretense at maintaining the farmer/agent ratio at 800; the shift from contact farmers to group meetings; the for-malrecognition of T&V's necessary role in ensuring that inputs move to the farmlands; and upgrading minimum requirements for VEW appointments to get at least university graduates. 5.26 Another major reform has been underway since 1989,when GOI and DOE started the charge toward "broadbasing"T&V. Until that point, T&V had generally been concentrated on the grain, pulse and oilseed operations on participating farms. What that focus neglected were the many other technical activitiessupported by the DOAs - the horticulture, animal husbandry,veterinary medicine, soil conservation, fishery and other serviceswhich DOA was not staffed to offer to individualvillages. The outreach of DOA units supporting these services usuallyended at the block level. They did not have their own extension cadre. Broadbasingmeant that the VEWs would be trained to assume these jobs as well. That would make T&V more visible, more relevant to the whole farming system, and more attractive to the state departments of finance.' 5.27 As mentioned, DOE is also bringing reforms to the T&V/M&E component. The M&E system had become as ritualistic as the T&V system it was supposed to monitor. The MP and AP operations under audit were two of the worst offenders. At the time of the audit mission the reforms had been accepted in those states mechanistically,but there still seemed to be little appreciation of what M&E could offer T&V management. 50 Theworrisome sideof 'broadbasing' Is thatit puts another elephant on top of thedonkey(para2.17).  - 35 - 5.28 The criticismsabove of the T&V and T&V/M&E systemsdo not give adequate recognition to the majority among the staff who are hard-working and intent on making their part of the operation function well. The best of the VEWs can compete with the best extension agents anywhere. Where they have been properly equipped, they can make T&V or any extension program flourish. The audit mission met many of that type of agent, including those who remain agents and those who have been promoted. B. Lessons 5.29 The Bank has to be more adept at "projectizing" useful development tools. The T&V blueprint made sense as a structural element of an extension-based project, but it was not enough to call a project by itself, and it covered up some important facts of bureaucratic life. There were maniyother elements - already part of the institutional landscape - to be incorporated in the design. The single-mindednessof the T&V model as applied in MP and AP has invited the criticism that T&V is characterized by excessive rigor and ritualism, faults which could have been corrected if the initial conditions had been more accurately described. 5.30 The Bank also has not to be satisfiedwith passing low-leveltests of project economic impact. The MP project passes that test; the AP project does not. But it is unlikely the either project would pass a test based on the real opportunity costs of tyingup these intensive staff and budget resources. 5.31 The T&V system in these two states needs to be opened up, to the point where the T&V structure may no longer even be recognizable. Training and visits will remain key operational concepts, but how and by whom they are handled should be open to all options. 5.32 In particular, the extension services in the two states should support the campaigns pressed by state and central governments, provided the essential qualities of the T&V philosophy are preserved: its ultimate goal of a universal embrace, commitment to diffusion of technologies,research linkage and disciplinedapproach to the work week. Those conditions nevertheless allow much room for maneuver, for example in ensuring the inputs arrive on time and assembling squadrons for campaigns. At the same time the rationale for campaigns could be expanded to permit their use not only for input supply but for highly targeted, cost-less changes in husbandry practices. 5.33 The relationship of the VEW and VLW field forces should be reconsidered. The fact that the best of the two services is now excluded from technical transfer, and has seen its role shifted to addressingthe production needs of more marginalfarmers, is an unanticipated and unwelcome effect of the T&V intervention. 5.34 The T&V/M&E system in these two states also has to be overhauled. The stimulus has to come from senior management in the Secretariat and DOA as well as in the Extension Division. The M&E staff has to include an analyticalwing, that is looked to by those senior officers for diagnostic work, to confirm project impacts, to identify weaknesses in the program, and to point the way to improvedperformance. The surveycontent should be re-examined to be more relevant, even beyond the survey reforms instituted in 1993. 5.35 With reforms of the T&V and M&E systems underway, the Bank is correct in considering continued support to extension work in both states. Justification would depend on the priority for  - 36 - extension among other investment opportunities at the sector level. It would be inappropriate for Bank pursuit of this particular developmental model to pull up short - for the wrong reasons - before the reformed system and the good ideas behind it have been secured.  37 ANNEX Government of India Ministry of Agriculture Dept. of Agriculture and Cooperation Directorate of Extension Krishi Vistar Bhavan Pusa, New Delhi 110012 Dated September 28, 1994 TO: The Chief Agriculture & Human Development Division Operations Evaluation Department The World Bank 1818 H Street N.W. Washington, D.C. 20433 U.S.A SUB: Madhya Pradesh Agricultural Extension Project - II (Credit 1138-IN)and Andhra Pradesh Agricultural Extension Project (Credit 1219-IN) Draft Performance Audit Project Sir, I am to refer to your D.O. letter dated the 16th August, 1994 addressed to Mr. J.C. Pant, Secretary (Agriculture), Government of India on the subject cited above. Our comments on the Performance Audit Report (PAR) in respect of Agricultural Extension Projects pertaining to Andhra Pradesh and Madhya Pradesh are enclosed for necessary action at your end. Yours faithfully, (V.V. Sadamate) Director (Extension Management) Copy to: Chief, Agriculture Unit The World Bank 69-70, Lodhi Estate, New Delhi 110003  38 Comments on the Draft PerformanceAudit Report Relating to MadhyaPradesh,AEP-Phase-U(Credit 1138-IN) and AndhraPradesh,AEP (Credit 1219-IN) The Operations Evaluation Department (OED), an independent department of the World Bank has forwarded Draft Performance Audit Report (PAR) of Madhya Pradesh Agricultural Extension Project -Phase II (Credit 1138-IN) and Andhra Pradesh Agricultural Extension Project (Credit 1219-IN). The OED has sought our comments on the Draft Report to be included, if any, in the final report before the same is submitted to the Board of Executive Directors of the World Bank. Our comments on the draft Performance Audit Report are given as below: I. Andhra Pradesh The Andhra Pradesh Agricultural Extension Project (Credit 1219-IN)commenced during 1981-82and the project was completed in 1989-90. The State Government could utilise the total allotted project cost although there were starting problems under the project in respect of allotment of funds. Although the Govemment of Andhra Pradesh made adequate provision under the project but expenditure on civil works, vehicle/equipment and incremental operating cost has been below both at desired and projected levels. The Directorate of Extension is of the opinion that the management of project activities especially at field level was weak in the State mainly due to the absence of field level supervisory staff. This adverselyaffected scheduled visits, transfer of technology and solving of problems of the farmers. The Government of Andhra Pradesh based on the visits of the World Bank Mission and Government of India's Inter- DisciplinaryTeams also pointed out the above weak areas and emphasised time and again for the regular active involvement of all supervisers in their responsibilitiesin the field and identifyingtraining needs of the extension functionaries. Although no significant widelyaccepted new package of practices were available to give a boost to the extension activities under the project, it has definitely helped in popularising the new rice varieties released by the State Agricultural University and also advocated need based pesticides used for rapid spread. Moreover, the project has helped in solving the problems of the farmers experienced by them in the field. It may also be mentioned that the APAU provided adequate support to the T&V Extension System in the State. As indicated in the PAR, the M&E Unit did not perform its job as conceived in the project. As per the assessment of the Directorate of Extension as also the World Bank, the performance of M&E Unit in Andhra Pradesh was below the average in all the respects. The M&E Survey Report brought some feedback on the field problems of the weak areas, but the same were not highlighted to the appropriate forums like monthly workshops and fortnightly training sessions.  39 II. MadhyaPradesh The implementation status of Agricultural Extension Project in Madhya Pradesh was rated as moderate both by the World Bank and Directorate of Extension. It has been rightly pointed out in the Performance Audit Report of OED that there has been inadequate budget support under the project throughout the project implementation period. However, the State Government made adequate provision in the budget during the extended period. The main problem in the project was large vacant positions of VEWs, AEOs, SMSs and M&E staff. The progress under civil works was also poor as a result large number of quarters remained incomplete even after the completion of the project. Like Andhra Pradesh, the State of Madhya Pradesh also ranked below the average among the T&V States in respect of M&E activities. M&E unit have been submitting M&E survey reports to the State Government, but the feedback was not brought out for the use of extension staff for necessary corrections.  IBRD15427R ir 6r IV BHIND rr MORENA N TA 115 -, P R V, HIVPUR c yj y 61 H: R F WA A CHHATARPUR" > MANDSAUR -/ I/ 'L GUNA SATNA -y P ANNA IDH'' -7 SAGAR Y 21-- k, DAMOH -7 9ATtA) '--,SHAAPURI RAJGARH,, 1, ........ 17. .... O .. .... . . . . .. ... . .... .. . 0 S N A H C)Ot Lik f, -'JABALP UR --------- - SEHORE -------NARSIMAHAPUR 0 M A N D LA JHABUA DEWAS . .... ... .. HOSHANGABAD ::':'-'D:HAR S E0 N I .. ........... 1, I I I B I LA S P U R BETUL KHARGON BALAGHAT 0 ---------- EAST "N C, NIMAIR It, R A t PU R 5 H T R A DUR6 INDIA 0 MADHYA PRADESH EXTENSION PROJECT r -2 PHASE U 20-- Le -M2 R81 PM-1993 A:k:::::: 3-1954 1981"4-1985 5-1986 h, 0 Olk., T.- D-,-l B-d- x O w K ?O x !O B.-6- OCTOBER 1990  7. 77 1 n < / i n INDIA ANDHRA PRADESH AGRICULTURAL EXTENSION PROJECT 9- ;7AO AJAGRICULTURAL RESEARCH FACILITIES ANDTRAININGCENTERS Ia.~~~~~~~~~ '$ MAt[JHfA / 5 < .t cfv,n bod F A tE FH PR 5J H c / 0 -' > D' KA RI M N AGA i A 0El -A A K AE . NIZAMABAD A~~~~~~~~~~~~~~~~~~ r . 7 S 0K i tA A I - 9 ) 2 .ve ( ' w>onooor&n2 , 7 S M E DA WA RA NGAL APAT NA A W AeRAN A Ll 5 A Hyderobod ohedra'ogai KH A M M A M . U. (7 HYDERABAD NALGONDA ~ 17- ~ ~ s / / r WE STT 2 3 t\ °~~~~~ ~ ~~~Moaro O O 0 !9ioect Components 7 ./ 2 M . AH8U8NAGAR wKRISHNA cl A Seed Forms \Y:ofovod. f De-elopment o-d Demo-st-rtoon F-rms C *-000 G U N T UR A-OO CE 0 Research Stton- U Tro-ing Center- AAOO :. . 0 M.chiiliptnam. Uniessty Comp.ses 4 ~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 Crae * Sitat Capitol K U R N OO L - / * High.oty MoIor RvR AAO.E Conals J} j- ->g- - State Roandories AN G ANTHAPUR p } NE L L RE AAA 0 CUDDAPAH \ ~~~~ ~~~~~ ~~~~~~~~~~0 AQO 20 d0 60 80 100 120 lAO 160 P80 200 220 0 20 40 60 80 100 120 140 i't~~~ t MadrasTrptl M O*o 2 83-Md