Mexico - Ninth Agricultural Credit Project

Document of The World Bank FOR OFFICIAL USE ONLY Report No. 13840-ME PERFORMANCE AUDIT REPORT MEXICO NINTH AGRICULTURAL CREDIT PROJECT (LOAN 2837-ME) DECEMBER 30, 1994 Operations Evaluation Department This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. CURRENCY EQUIVALENTS Currency Unit - New Mexican Peso (Mex$) (year end) 1987: US$1.00 = Mex$ 2.21 1988: US$1.00 = Mex$ 2.28 1989: US$1.00 = Mex$ 2.64 1990: US$1.00 = Mex$ 2.95 1991: US$1.00 = MeX$ 3.07 1992: US$1.00 = Mex$ 3.12 The US$ loan amounts in this report should be understood to mean US$ equivalent. ACRONYMS AND ABBREVIATIONS ACF - Average Cost of Funds AGROASEMEX - Aseguradora de M6xico para la Agricultura (Agricultural Insurance Company of Mexico) ANAGSA - Aseguradora Nacional Agicola (National Agricultural Insurance Company) AGSAL - Agricultural Sector Adjustment Loan BANRURAL - Banco Nacional de Crdito Rural (National Rural Credit Bank) BANXICO - Banco de Mxico (Central Bank of Mexico) CETES - Cerificados de Tesorerla (Mexican Treasury bills) CONASUPO - Compafia Nacional de Subsistencias Populares (National Commission for Distribution of Basic Foods) FEFA - Fondo Especial par Financiamientos Agropecuarios (Special Fund for Agricultural Financing) FEGA - Fondo de Garantia y Asistencia Tdnica (Fund for Loan Guarantees and Technical Assistance) FEPS Final Executive Project Summary FICART - Fideicomiso para Crdito en Areas de Riego y de Temporal (Trust Fund for Credit in Irrigated and Rainfed Areas) FIRA - Fideicomisos Instituidos en Relaci6n con a Agricultura (Agricultural Trust Funds) FIRCAVEN - Fideicomiso para a Restructuraci6n de la Carters Vencida (Trust Fund for Restructuring Portfolios in Arre-ars) FONDO - Fondo de Garantia y Fomento pars Ia Agricultura, Ganadera y Avicultura (Guarantee and Development Fund for Agriculture, Livestock and Poultry) FOPESCA - Fondo de Garantia y Fomento pars las Actividades Pesqueras (Guarantee and Development Fund for Fisheries) FSAL - Financial Sector Adjustment Loan GATT General Agreement on Tariffs and Trade GDP - Gross Domestic Product GIRA - General Interest Rate Agreement GOM - Government of Mexico IDB Inter-American Development Bank LIPs - Low Income Producers M&E - Monitoring & Evaluation MIS Management Information System NAFTA - North American Free Trade Agreement NPC Nominal Protection Coefficient OED - Operations Evluation Department OPs - Other Producers (as opposed to LIPs) PAR - Performance Audit Report PBs Participating Banks PCR - Project Completion Report PRONASOL - Program Nacional de Solidaridad (National Solidarity Programme) SAR - Staff Appraisal Report SARH - Secretaia de Agricultura y Recursos Hydraulicos (Secretariat of Agriculture) SDI Subsidy Dependence Index SERC - Sistem de Evaluaci6n de los Resultados del Cr6dito (Credit Impact Evaluation System) SHCP - Secretarfa de Hacienda y Crdito Publico (Secretariat of Finance) SIM - Sistema de Inspecci6n por Muestre (System of Monitoring by Sampling) SIPA - Sistema d Indices de Precios Agrfcolas (System of Agricultural Price Indices) SUECO - Sistema Unico de Elaboraci6n de Costos de Cultivo (integral System for Determining the Costs of Cultivation) TA - Technical Assistance TORs Terms of Reference WEIGHTS AND MEASURES Metric System FISCAL YEAR January 1 BnDecember 31 FOR OFFICIAL USE ONLY THE WORLD BANK Washington, D C. 20433 U.S.A. Office of Director-General Operations Evaluation December 30, 1994 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Performance Audit Report on Mexico Ninth A.ricultural Credit Project (Loan 2837-ME) Attached is the Performance Audit Report on Mexico-Ninth Agricultural Credit Project (Loan 2837-ME) prepared by the Operations Evaluation Department. The project outcome is rated as satisfactory. Subsidies on loans to agricultural producers were reduced by at least 7 percent of agricultural GDP, exceeding appraisal targets; interest rates to farmers are now positive in real terms, although on average they remain below market rates. The investment and incremental short-term credit provided under the project supported financially viable investments in irrigated areas, but the performance of investments in rainfed areas was mixed. The project strengthened the productive support systems of the Government's Agricultural Trust Funds (FIRA) by providing civil works, training and technical assistance. Institutional development is rated as substantial. In particular, the Bank insisted on a thorough analysis of the rural financial sub-sector, both in the context of a project-financed Agricultural Financial Sub-sector Study and of preparation of a follow-on operation. Although the study was weak, these efforts yielded data and findings that catalyzed key reforms, such as rationalization of the National Rural Credit Bank (BANRURAL) and elimination of the National Agricultural Insurance Company (ANAGSA). Preparation of the proposed follow-on operation was suspended in January 1993, due to disagreements on the pace of interest rate reforms. However, Government appears to be committed to the policies supported under the project, the reform process went beyond the project's stated objectives, Government transfers to FIRA and BANRURAL have fallen sharply, and the value of their equity has increased in real terms in spite of the reduced support. Overall, the sustainability of project benefits is rated as likely. Attachment This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.  FOR OFFICIAL USE ONLY PERFORMANCE AUDIT REPORT MEXICO NINTH AGRICULTURAL CREDIT PROJECT (Loan 2837-ME) TABLE OF CONTENTS Page No. Preface .......................................................... Basic Data Sheet .....................................................iii Evaluation Summary .................................................. v I. BACKGROUND ................................................ 1 II. DESIGN AND IMPLEMENTATION ................................. 4 A. Project Design ............................................... 4 B. Project Implementation ......................................... 8 III. PROJECT OUTCOME ........................................... 14 A. Farm-Level Impact ............................................ 14 B. Institutional Development ....................................... 19 C. Impact on Financial Intermediaries ................................ 20 D. Impact on Sectoral Reforms ..................................... 26 IV. FINDINGS AND ISSUES ......................................... 30 A. Overall Assessment of the Project ................................. 30 B. Key Findings Regarding FIRA/FICART 9 ........................... 31 C. Subsequent Developments and Outstanding Issues .................... 34 Figures 1. Agricultural Lending by FIRA, BANRURAL and Commercial Banks . 18 2. Ratio of Arrears to Loan Portfolio for Commercial Banks on Loans to Agriculture and All Sectors ............................ 26 3. Interest Subsidies on Loans via FIRA and BANRURAL as a Percentage of GDP ......................................... 27 4. Area Insured for Selected Crops .................................... 33 This repoi t was p, cpared by Edward B. Rice (Task Manager), and McDonald Benjamin (cnsultant), who audited the project. Silvana Valle provided the administrative support. This document has a restricted distribution and may be used by recipients only in the performance of their official duties. its contents may not otherwise be disclosed without World Bank authorization. Page No. Tables 1. Appraisal and Actual Project Costs .................................. 9 2. Appraisal and Actual Project Financing ................................ 9 3. Increments in Assets and Equity during Project Years for a Stratified Sample of FIRA Borrowers ........................... 16 4. Subsidy Dependence Indices for FIRA ............................... 21 5. Example of the Tasa Mezcla's Effect on Subsidies and Interest Rates for a Hypothetical Short Term Loan to a "Basic OP" .................... 24 Annex 1: Statistical Overview of the Mexican Economy and the Rural Financial Sector 1. The M acroeconomy .............................................. 39 2. A griculture .................................................... 39 3. T rade ......................................................... 39 4. Finance ....................................................... 39 5. The Banking System's Non-Agricultural Portfolio ........................ . 40 6. The Banking System's Agricultural Portfolio ........................... . 40 7. Agricultural Interest Rates ......................................... 41 8. Agricultural Insurance ............................................ 41 9. Subsidies to Agricultural Lenders .................................... 41 10. BANRURAL's Key Financial Ratios ................................. 42 11. FIRA Interest Subsidy Passed to Commercial Banks Intended for Farmers............................................ 42 12. BANRURAL's Interest Subsidy to Farmers ............................ . 42 13. Returns to Farmers on Selected Crops ................................ 43 14. Sources of D ata ................................................. 43 Map IBRD 26569 1 PERFORMANCE AUDIT REPORT MEXICO NINTH AGRICULTURAL CREDIT PROJECT (Loan 2837-ME) PREFACE This is a Performance Audit Report (PAR) on the Ninth Agricultural Credit Project in Mexico, involving an IBRD loan in the amount of US$400 million. The objectives of the project were to (i) increase agricultural productivity and thus agricultural exports, real farm income and rural employment; (ii) help the Government improve agricultural sector policies by further reducing interest subsidies, and (iii) help to maintain the financial integrity of participating financial intermediaries. The loan was approved on June 16, 1987, became effective on August 28, 1987, was fully disbursed on April 2, 1992, and was closed on June 30, 1992, one year later than the original Closing Date of June 30, 1991. The PAR is based on the Project Completion Report (PCR) prepared by the Natural Resources and Rural Poverty Operations Division in Country Department II of the Latin America and Caribbean Regional Office (LA2NR)* and submitted to the Board on December 21, 1993, the Staff Appraisal Report (SAR), the President's Report, the loan documents, the transcripts of the Executive Directors' meeting at which the project was considered, on a study of project files, and on discussions with Bank staff. An OED mission visited Mexico in February/March 1994, and discussed the effectiveness of the World Bank's assistance with the Agricultural Trust Funds of the Central Bank (FIRA), the Secretariats of Finance and Agriculture, the Central Bank, the National Rural Credit Bank (BANRURAL), other relevant Government agencies, commercial banks and agricultural producers. Their kind cooperation and valuable assistance in the preparation of this report is gratefully acknowledged. The PCR provides a thorough account and assessment of the project experience, and discusses the performance of the Bank and the project executing agencies with a particular focus on sectoral reforms. The PAR elaborates on particular aspects of the overall lending period (1987-1992), including the allocation of resources at the farm level, the problem of arrears, subsidisation of farm beneficiaries and the institutional viability of FIRA. Following standard OED procedures, copies of the draft were sent to Government, FIRA and BANRURAL for comments in November 1994. The Borrower informed OED in December 1994 that it had no comments regarding the Audit Report. * Formerly known as the Agricultural Operations Division in Country Department 11 of the Latin America and Caribbean Regional Office (LA2AG).  Ill PERFORMANCE AUDIT REPORT MEXICO NINTH AGRICULTURAL CREDIT PROJECT (LOAN 2837-ME) BASIC DATA SHEET Key Project Data Item Appraisal Actual or Actual as Percent of Expectation Current Estimate Appraisal Estimate Total Project Costs (US$ million) 1,000.0 920.0 92 Loan Amount US$ million 400.0 400.0 100 Number of Beneficiaries 550,000 868,400 158 Project Rating Satisfactory institutional Performance Substantial Sustainability Likely Cumulative Estimated and Actual Disbursements FY88 FY89 FY90 FY91 FY92 Appraisal Estimate (US$ million) 150.0 250.0 360.0 400.0 400.0 Actual (US$ million) 173.9 371.4 389.9 395.4 400.0 Actual as Percent of Appraisal 116% 149% 108% 99% 100% Date of Final Disbursement: April 2, 1992 Project Dates Original Plan Revisions Actual Appraisal Mission 03/85 11/85 11/25/85 Post-Appraisal Update Mission 05/04/87 Negotiations 11/86 11/86 & 05/87 Board Approval 07/85 06/87 06/16/87 Loan Signature 07/31/87 07/31/87 Loan Effectiveness 09/30/87 10/29/87 08/28/87 Project Completion 12/31/90 12/31/91 12/31/91 Project Closing 06/30/91 06/30/92 06/30/92 Iv Staff Inputs (staff weeks) FY85 FY86 FY87 FY88 FY89 FY90 FY91 FY92 FY93 FY94 TOTAL Pre-appraisal 4.3 7.7 12.0 Appraisal 32.7 5.3 38.0 Negotiation 26.6 26.6 Supervision 1.4 20.6 20.4 6.7 0.8 15.6 4.8 1.6 71.9 Other 2.6 1.3 9.1 13.0 Total 6.9 41.7 42.4 20.6 20.4 6.7 0.8 15.6 4.8 1.6 161.5 Mission Data Mission Dates No.of Staff- Specializations Overall Project Development (mo.yr) Persons Weeks Represented Status b MgMt. b impact b in Field Preparation 03/85 3 2 F, E, AE Pre-appraisal 10/85 1 1 F Appraisal 11/85 5 3 F, A, AE, C, 0 Full Supervision 1 08/87 2 4.5 TM, AE n.a. n.a. n.a. Full Supervision 2 10/87 2 6 TM, F 1 1 1 Partial Supervision 3 11/87 2 1 TM, C Partial Supervision 4 05/88 * 1 1 TM Full Supervision 5 09/88 1 1.5 TM 2 2 1 Partial Supervision 6 01/89 1 1.0 AR Partial Supervision 7 02/89 1 0.3 LW Full Supervision 8 03/89 2 2.5 TM, C 1 2 1 Partial Supervision 9 05/89 1 1 C Partial Supervision 10 06/89 2 1.3 TM, C Partial Supervision 11 07/91 1 0.5 AR Partial Supervision 12 08/91 3 6 F, E, AE PCR Preparation 10/92 2 2.5 PA, AR TOTAL 35.1 Other Project Data Borrower: Nacional Financiera (NAFIN) Executing Agencies: Fidelcomisos Instituidos en Relaci6n con la Agricultura (FIRA) Fideicomiso para Cr6dito en Areas de Riego y Temporal (FICART) Proposed Follow-On Prolect (Suspended) Name: Proposed Rural Financial Sector Project Proposed Loan Amount: US$300 million Processing Suspended on: January 11, 1993 a. F = financial analyst; E = economist; AE = agricultural economist; A = agriculturalist; C = credit specialist; 0 = operations assistant; TM = task manager; AR = architect; LW = lawyer; PA = project assistant. b. 1 = Problem-free or minor problems; 2 = moderate problems; 3 = major problems. V PERFORMANCE AUDIT REPORT MEXICO NINTH AGRICULTURAL CREDIT PROJECT (Loan 2837-ME) EVALUATION SUMMARY Introduction 3. The project was to be implemented by two apex institutions: FIRA, the Agricultural 1. Mexico is the World Bank's second Trust Funds of the Central Bank, and largest borrower for agricultural credit. Since FICART, a trust fund of the Government- 1965, the Bank has approved ten loans totalling owned National Rural Credit Bank, US$1,805 million for agricultural credit projects BANRURAL. The Borrower (Nacional and earmarked an additional US$334 million Financiera, or NAFIN) was to on-lend the for credit components in six rural development World Bank loan of US$400 million equivalent projects. The initial thrust of Bank policy was to FIRA (US$300 million) and FICART to promote agricultural production by large (US$100 million) for production credit (9 commercial farmers; however, by the fourth percent), investment credit (79 percent) and project (approved in 1973), the Bank began to institutional development (2 percent). Ten emphasize access to credit for low income percent of the loan was unallocated. During producers (LIPs). The Bank also attempted to negotiations in May, 1987, support was agreed place greater emphasis on institutional for the construction of FIRAKs new development and on sectoral reforms. The size headquarters. of agricultural interest subsidies was a particular concern for the Bank and was addressed under 4. The Bank loan was to cover the foreign a General Interest Rate Agreement (GIRA) exchange costs of the project, or 40 percent of that has governed agricultural credit since 1984. the total project cost of US$1,000 million. The This Audit reviews the Ninth Agricultural remaining 60 percent of costs were to be Credit Project (FIRA/FICART 9), the last in financed by FIRA and FICART (32 percent), the series of ten agricultural credit projects. commercial banks (14 percent) and sub- borrowers (14 percent). The total project cost Project Design and Implementation of US$1,000 million represented about 15 percent of FIRA's and FICART's projected 2. The objectives of FIRA/FICART 9 were investment programme of US$6,911 million for to (i) increase agricultural productivity and thus the years 1987-1991. agricultural exports, real farm income and rural employment; (ii) help the Government improve 5. When the project was identified in 1985, agricultural sector policies by further reducing the Bank informed the Government of Mexico interest subsidies, and (iii) help to maintain the (GOM) that continued disbursement at the financial integrity of participating financial rapid rates experienced under earlier intermediaries. An important but unstated operations could only be justified if further objective was to provide the Mexican progress was achieved on sectoral issues. The Government with rapid access to foreign Bank initially called for a 50 percent reduction exchange. in interest subsidies to LIPs, the preparation of vi a study of the agricultural financial sub-sector 8. Project implementation was complicated (hereafter the Study), and financial self- by delays in GOM adherence to the GIRA, sustainability for FIRA and FICART within 5 problems with the construction of FIRA's new years. headquarters and disagreements with GOM regarding the Study. Government delays in 6. Negotiations broke down in November, raising interest rates to farmers as mandated by 1986 over the issue of financial self- the GIRA led to temporary suspensions of sustainability for FIRA and FICART (para. disbursements in 1988. Nevertheless, the loan 2.6). In January, 1987, GOM failed to increase disbursed well ahead of schedule, except for the interest rates to LIPs as agreed under the FIRA headquarters component, which GIRA. Bank management was informed in experienced major delays (para. 2.19). The April that the proposed loan would slip to project was extended by one year to allow for FY88, "unless Bank management sees a way in the completion of the civil works, but further which we could proceed more quickly". Senior delays led to reallocation of loan funds from management and regional staff agreed that civil works to on-lending and the project was rapid movement with some accommodation was closed on June 30, 1992. The Study was required. The conditionality on interest rates executed in two phases. Te phase I report was weakened and the requirement of financial was completed in May, 1988. GOM initially self-sustainability for FIRA and FICART was withheld most of the report from the Bank, dropped. Instead, GOM agreed to provide arguing that the Guarantee Agreement only budgetary support to FIRA and FICART so as required it to "exchange views with the Bank to prevent a decline in the real value of their on the conclusions and recommendations of equity below the levels prevailing in January, such study" (paras. 2.20 ff). A complete copy 1986. Processing of the loan proceeded quickly of the phase I report was later given to the after that: a post-Appraisal update mission Bank, but was found to be unsatisfactory. visited Mexico early in May, a second round of Phase 11 of the Study was begun in March, negotiations that month resolved outstanding 1989 and completed in July, 1990. The Bank issues and the project was approved by the was also dissatisfied with the phase 11 Board on June 16, signed on July 31, and document, but did not insist on improvements became effective on August 28, 1987. as it had begun to obtain information for its own analysis. 7. The project was designed to disburse rapidly (paras. 2.9 ff). There was no up-front 9. The project had relatively few covenants conditionality, no linkage of disbursements to and compliance was acceptable, albeit actions such as preparation of the Study, and incomplete (paras. 2.27 ff). Bank supervision no cap on the disbursement rate. Moreover, was on the whole thorough and timely, and the the appraisal report (SAR) and the staff's Bank's policy dialogue with GOM flourished Board Presentation were both characterised by when the new Administration adopted a more important lapses in the accuracy and laissez-faire approach ater the 1988 elections. completeness of information (paras. 2.12, 2.13 and 2.16). Lending pressure was clearly a Project Outcome factor: Board Approval of this project was needed to meet the FY87 lending programme 10. According to FIRA/FICART estimates, target for Mexico of US$2,000 million. the number of beneficiaries (870,000) was Besides, disbursement of the third tranche of a higher than appraisal projections (550,000). commercial bank package for Mexico was LOW income producers (i.e. those with annual linked to Effectiveness of this loan, family incomes less than Mn times the vii regional minimum daily rural wage) received a funds that appears to have been important is larger proportion of the Bank loan than commercial banks' provision of FIRA- expected. However, lending to LIPs fell rediscounted sub-loans to wealthier farmers sharply as a share of FIRA's total lending, from who could have qualified for short-term credit 49 percent in 1987 to 19 percent in 1993. at market rates using commercial banks' own FIRA correctly points out that the minimum funds (para. 3.13). rural wage declined significantly in real terms during the project, so that certain low income 13. The project was to support institutional farmers could have experienced declines in real development (ID) at FIRA by providing income and still have graduated involuntarily to equipment for FIRA's demonstration centres, the other producers (OP) category. The Bank training for FIRA staff and funding for the stood by the increasingly strict definition of construction of FIRA's new headquarters. The LIPs. Since interest rates to OPs were higher project had little impact on the demonstration than rates to LIPs, graduation of loanees to the centres (less than 30 percent of Bank funds OP category brought interest rates on their allocated for equipment and vehicles were loans closer to market rates. disbursed) but the centres had a visible impact on the cultivation practices of neighbouring 11. While the impact of the project on farms (para. 3.15). Nonetheless, 3 of the 9 agricultural production in Mexico cannot be FIRA-operated centres were shut down in the quantified, it is feasible to consider whether or past two years by the State Governments that not investments that the World Bank helped to owned them. Over 13,800 participants received support were financially viable (paras. 3.5 ff). training in more than 850 training events for The Audit concludes that investments in farmers and FIRA staff under the project. The irrigated crops were financially viable long term benefits of the training have not throughout 1986-1990, but that the picture is been evaluated. The new FIRA headquarters less encouraging for rainfed crops. were completed more than two years after Furthermore, not all financially viable Project Closing and FIRA staff moved into the investments were necessarily economically new facility in October, 1994. sound, since the policy environment was highly distorted. The main sources of inefficiency 14. The ID component for FICART was were trade, exchange rate and input/output implemented with funding from non-project price factors, rather than cheap credit. sources. No information is available on the component, since FICART has been disbanded. 12. Diversion of sub-loans was rampant Significant institutional development of among BANRURAL clients, primarily due to BANRURAL began during the project period, lack of monitoring and little pressure to repay, (although not with project funding), notably in rather than to below-market rates of interest. the form of training programmes and upgrading Diversion of sub-loans was much rarer among of computerised systems. There was also commercial bank clients, due to more intensive substantial institutional development at supervision. While substitution by sub- commercial banks, which benefitted from borrowers of loan funds for other resources FIRA's TA and training (para. 3.18). cannot be ruled out, the Audit found no evidence of widespread capital flight using 15. GOM fulfilled the covenant requiring FIRA/commercial bank loan funds during FIRA's equity to be maintained at least at FIRA/FICART 9. Capital flight was allegedly January, 1986 levels in real terms. FIRAs a serious problem during earlier operations. strong financial position at the end of the One type of substitution (or crowding out) of project could not have been achieved without viii GOM's support. A subsidy dependence index programme of restructuring of agricultural (SDI) was calculated to assess the extent of loans. Commercial banks' contaminated FIRA's reliance on subsidies (excluding the agricultural portfolio amounts to Mex$5,000 guarantee fund, FEGA). The Audit found that million (US$1,540 million), of which about half during 1988-93 FIRA would have had to charge is overdue. Most overdues are on commercial interest rates 85 to 121 percent higher than its banks'own funds, rather than FIRA (or World actual average on-lending rates (e.g. 44 percent Bank) funds, since commercial banks advanced instead of 24 percent in 1990) in order to cover bridging loans to agricultural producers at the opportunity cost of its liabilities and equity unrealistically high real rates of interest in the and operate without GOM grants (paras. 3.19 late 1980s. Since 1990, commercial banks have ff). The required increases would have been been more selective, and agricultural credit has lower (74 to 110 percent) if FIRA's costly TA declined as a share of their total lending. activities had been excluded. Moreover, FIRA became a more efficient institution in the 18. Commercial banks have also called for course of the project. The spread (over the increased spreads on rediscounted agricultural market rate) required for self-sufficiency loans. This request was partly satisfied by the declined from 6 percent in 1988 to 2.4 percent introduction of the tasa mezcla or blended in 1992. Excluding FIRA's TA activities, the interest rates scheme introduced in 1989 (paras. respective figures were 3.6 percent and 1.2 3.27 ff). The scheme applies only to short term percent. This was achieved by halving FIRA's loans to OPs. It allows participating banks ratio of administrative costs to risk assets. On (PBs, i.e. commercial banks and BANRURAL) the other hand, FIRA's guarantee and TA to charge unrestricted interest rates on any fund, FEGA, saw a five-fold real increase in funds they contribute to FIRA-rediscounted net guarantee payments (after subtracting sub-loans above the minimum 20 percent premium income and recoveries on guarantees) contribution. The blend of controlled and during the project, due to rising arrears on unrestricted rates has resulted in market rates guaranteed loans and to sharp reductions in the to OPs on these loans. area indemnified by agricultural insurers. 19. The project adopted a two-pronged 16. FICART"s equity rose 9 percent in real approach to reduce interest subsidies and terms over the life of the project, but the encourage further reforms in the rural financial increment was a mere 10 percent of total GOM sector. First, it mandated increases in interest transfers to FICART during the project period. rates to farmers in accordance with the GIRA, FICART was shut down in 1992. FICART's and second, it called for sectoral reforms based trustee, BANRURAL, made significant losses on the recommendations of the Study. The every year, due to low interest rates and an limited objectives with regard to reducing extremely serious problem with loan recoveries. interest subsidies were achieved (para. 3.34). GOM has allowed BANRURAL to shrink Subsidies also became more transparent during considerably and bear the losses it makes on the project period. GOM's agricultural bad loans. As a result, BANRURAL's recovery insurance agency, ANAGSA, which had made rate on current dues appears to have doubled huge losses during the 1980s, was shut down. to 80 percent. BANRURAL's deposits barely increased in real terms, but rose significantly as a share of 17. Arrears have increased dramatically on liabilities. The closure of ANAGSA and agricultural loans and indeed on all loans by shrinking of BANRURAL are partly commercial banks (paras. 3.31 ff). In March, attributable to the Study, which otherwise had 1994, GOM and the Mexican Banking little impact on sectoral reforms (paras. 3.38 ff). Association (AMB) negotiated a massive agricultura potoi-mutst e$,0 ix Findings and Issues 0 Commercial bank lending to wealthier farmers rose sharply following the introduction 20. The project was largely implemented as of the lasa mezcla scheme, suggesting that planned at appraisal and is rated satisfactory. FIRA had crowded out short term lending to While there is a problem of attribution of these OPs. results, institutional development was substantial during the project period. GOM E Interest subsidies fell by at least 0.6 percent was reluctant to undertake certain reform of GDP during the project period, (or more measures proposed by the World Bank and the than 7 percent of agricultural GDP). Greater Inter-American Development Bank (IDB) as transparency and the use of the SDI would conditions for a follow-on operation, but help GOM to evaluate remaining subsidies for appears to be committed to the policies FIRA's and BANRURAL's development supported under the project. Transfers to activities. FIRA and BANRURAL have fallen sharply, and several market-friendly reforms have been U The replacement of ANAGSA with introduced in the broader agricultural and AGROASEMEX, a more commercially financial markets. Overall, the project's oriented agricultural insurer, was an important benefits are considered sustainable. success. Both AGROASEMEX's insurance premia and FEGA's loan guarantee premia 21. The Audit underscores the following need to be raised towards actuarially fair rates project findings (paras. 4.4-4.12): in order to reduce the cost to GOM, allow FEGA and AGROASEMEX to expand their a The unstated objective of providing GOM coverage and encourage entry into the with rapid access to foreign exchange was agricultural insurance market by private achieved. However, this objective was not companies. entirely consistent with the project's long term development objectives. U Preparation of a follow-on operation was suspended early in 1993, but significant policy n The farm-level impact of the project cannot measures were nonetheless adopted. The be quantified. While FIRA's M&E findings promise of funding and Bank/IDB coordination point to large increases in asset and equity were essential for the constructive dialogue. holdings of FIRA clients, there was no control group. This difficulty should be addressed, 22. The Audit also highlights the following even if only by second-best measures. outstanding issues (paras. 4.13-4.19): f Lack of pressure to repay (rather than low a FIRA's future as an institution is being interest rates) led to significant diversion of debated as the Central Bank prepares to funds by BANRURAL clients. Diversion was relinquish trusteeship of all its trust funds. much rarer for FIRAFcommercial bank loanees, Establishment of FIRA as a second-tier due to more intensive supervision, development bank or a transfer of FIRA's trusteeship to NAFIN should be explored. tn 7e problem of OPs masquerading as LIPs was rarely confirmed by FIRA's monitoring t FIRA must assume greater risks in order to staff. The incentive for such masquerading has fulfill its mission of expanding the formal declined as LIP rates have approached OP financial frontier to new, viable and potentially rates, and penalties for PBs are severe if viable clients. This may require expanding the irregularities are detected. coverage of FEGA's loan guarantees. x Crowding out of commercial bank lending to borrowers. More generally, Bank supervision established clients should be avoided. of any future financial intermediary loans in Meico should include an annual review of the a BANRURAL should mobilise more rural financial performance of all participating banks. savings, lend on the basis of stringent criteria and gradually diversify its portfolio into non- E The bold measures taken by GOM since agricultural rural lending. Part of its 1987 have reduced inefficiency in rural financial transaction costs could be covered by a markets. However further measures are transparent, time-bound subsidy that would also needed, especially liberalisation of agdcultural be made available to other participating banks interest rates, in order to increase farmers' for loans to new, potentially viable clients. access to credit. The Bank should seek to conduct a more intensive policy dialogue with a The restructuring of arrears on commercial GOM, tailoring any future financial support for banks' agricultural loans should be monitored the sector to GOM's willingness to consider closely to assess implications for FIRA and further necessary reforms. PBs, and repayment incentives for sub- PERFORMANCE AUDIT REPORT MEXICO NINTH AGRICULTURAL CREDIT PROJECT (Loan 2837-ME) I. BACKGROUND 1.1 The Macroeconomy: Until the debt crisis of the early 1980s, Mexico pursued an inward- looking industrialisation strategy that yielded high rates of growth but led to structural imbalances and growing inefficiency. The discovery of major oil reserves in 1977 allowed the Government of Mexico (GOM) to increase external borrowing dramatically and to postpone urgent economic reforms. In 1982, declining oil prices, rising real interest rates and the strength of the US Dollar precipitated a debt crisis that obliged GOM to break with earlier policies and adopt an IMF-supported stabilisation programme. This programme reduced fiscal and current account deficits, however rising inflation and the collapse of oil prices in 1986 entailed the need for a more fundamental restructuring of the economy. Consequently, the Government cut public spending and the fiscal deficit, reduced tariffs and quantitative restrictions, and allowed the Peso to depreciate significantly in real terms. In 1986 Mexico acceded to the GATT. A commercial bank financing package was arranged to support the government's 1986/87 structural adjustment programme and release of the third tranche of the package was made conditional on the Effectiveness of this loan (FIRA/FICART 9, Loan 2837-ME). 1.2 The Agricultural Sector: About 17 percent of Mexico's surface area is dedicated to agriculture, 24 percent to woodlands and rainforests and 51 percent to grass and pasture lands. Landholdings are highly fragmented, averaging three hectares per farmer. Nearly two-thirds of the tracts are semi-communal ejido lands.' Around eighty percent of agricultural producers are considered low income producers (LIPs), that is, their annual net family incomes amount to less than one thousand times the regional minimum daily rural wage. In 1985, the agricultural sector accounted for about 9 percent of GDP and 35 percent of employment. Growth in agricultural production averaged 2.5 percent annually in the early 1980s but was virtually stagnant in the late 1980s, as the agricultural sector took time to recover from changes in the macroeconomy, the trade regime and sectoral policies. 1.3 Agricultural Policy: Government policy towards agriculture has aimed to combine low prices for urban consumers with sufficient incentives for agricultural producers. This policy is implemented through a combination of subsidies and price controls. The National Commission for Distribution of Basic Foods (CONASUPO) purchases staple crops from producers at guaranteed prices and administers food subsidies to consumers. Fertilisers, irrigation water and farm credit were heavily subsidised until the mid-1980s, when GOM began to reduce input subsidies. During the project period (1987-91), import restrictions on agricultural products were reduced or removed, and 1. These lands belong to ejidal communities established after the Mexican revolution in the 1920s. The ejidatarios were to cultivate the lands communally and were not to sell, rent or pledge their parcels. Usufruct rights were to be passed from one generation of eOidatarios to the next. In 1992, restrictions on the exchange of ejidal lands were eased slightly with the revision of Article 27 of the Constitution, however their value as collateral for bank loans remains limited. 2 guarantee prices were eliminated in 1990 for all crops except maize and beans. The World Bank supported these and other reforms with sectoral adjustment operations, including two in agriculture. 1.4 Agricultural Credit Institutions: In order to stimulate lending to the agricultural sector and induce greater participation by private banks, GOM established three trust funds, collectively known as FIRA (the Agricultural Trust Funds), within the central bank. The first of these, FONDO, was established in 1955 as a promotion and guarantee fund for agriculture. It rediscounts primarily short- term commercial bank loans to agriculture. FEFA, the special fund for agricultural financing, was established in 1965 as a channel for foreign loans to agriculture. It rediscounts medium- and long- term loans to the sector by public and commercial banks. The guarantee and technical assistance fund, FEGA, was created in 1973 to guarantee a portion of commercial bank loans to agriculture and subsidise transaction costs for commercial bank loans to low income producers. 1.5 FIRA's network consists of 10 regional offices, 38 state offices and 90 local agencies. A new headquarters is currently being completed in Morelia, MichoacAn. FIRA also administers six demonstration farms, of which three are owned by state governments. FIRA's highly qualified staff provide TA to farmers and commercial bank staff, supervise sub-borrowers' projects, monitor participating banks (PBs) and assist PBs with the evaluation of larger loans. 1.6 Mexico's commercial banks were nationalised in 1982 and denationalised in 1991. As their new private sector owners are attempting to recover their investments rapidly, banks are reducing the share of agriculture in their portfolios and paring the number of agricultural loan staff, which had increased significantly with FIRA assistance during the 1980s. The banks are also charging high spreads on their activities. The Government has attempted to inject greater competitiveness into the sector by admitting new banks, including affiliates of US banks, although it has balked at admitting large European banks into the Mexican market. 1.7 In order to reach (primarily low income) agricultural clients that commercial banks are reluctant to lend to, GOM complemented FIRA's second-tier activities with direct lending to the agricultural sector via the publicly owned National Bank for Rural Credit, (BNCR). BANRURAL, as it is also called, was formed in 1975 as the result of a merger of three agricultural development banks, two of which were financially weak institutions. Although BANRURAL was the largest bank in Mexico during the 1980s with over 25,000 staff, 500 branches and about 40 percent of all financing for agriculture, it suffered from significant loan delinquencies that proved to be unsustainable. Over the past five years, BANRURAL's staff and scale of operations have shrunk dramatically and in 1991 a large part of its portfolio was written over to a trust fund for the collection of arrears (FIRCAVEN).3 1.8 FICART (the trust fund for credit in irrigated and rainfed areas) was established in 1976 as a second-tier institution just to rediscount sub-loans made by BANRURAL, its trustee. FICART provided a conduit for loans from the Inter-American Development Bank (IDB) that enabled the IDB to avoid having to deal with BANRURAL directly. Although FICART had helped to execute 2. Mexico - Agricultural Sector Adjustment Loan (2918-ME), US$300 million, approved on 22nd Februarv, 1988. and Mexico - Second Agricultural Sector Adjustment Loan (3357-ME), US$400 million, approved 25th June, 1991. 3. FIRCAVEN has in turn given way to a trust fund known as FI)ELIQ, for the liquidation of bankrupt companies. FIRCAVEN's recoveries on Banrural loans in arrears had been negligible. 3 credit components under four earlier Bank-supported projects, FIRA/FICART 9 was the first one in which it participated as a direct beneficiary. It has since been disbanded, due to a weak rationale for its existence, reforms at BANRURAL and high operating costs. BANRURAL assumed FICART"s assets and liabilities in 1993. 1.9 During the 1980s, almost 50 percent of BANRURAL's recoveries were direct transfers from the now defunct national agricultural insurance company, ANAGSA. ANAGSA's coverage was relatively limited in the 1960s and 1970s, but in 1981 a change in the law made insurance with ANAGSA mandatory for BANRURAL loanees. ANAGSA thus crowded out private insurers and, with policies involving no deductible and full coverage, moral hazard problems set in, including significant corruption of ANAGSA staff. The losses became unsustainable and in 1990, ANAGSA was replaced with an agro-insurance company (AGROASEMEX) that is run on a more commercial basis. 1.10 In 1988, the Salinas Administration introduced a national Solidarity program (PRONASOL), with a view to channelling funds to municipalities for development activities. One channel of resources provides financing for small, interest-free loans, averaging Mex$400. These loans are approved by municipal monitoring and evaluation committees and are repaid to a municipal fund for further investment at the village level. This is intended to encourage communal pressure to repay. Loans are unsecured and are based on credito a la palabra, i.e. on the word of the loanee. Recovery rates are around 55 percent, compared to an estimated 20 percent under BANRURAL. A large segment of BANRURAL's less bankable clients have been redirected to PRONASOL, which therefore serves the most risky and least viable tier of agricultural producers. Although the World Bank is supporting PRONASOL with two rural development loans (3310-ME and 3790-ME), it has explicitly ruled out direct funding of PRONASOL's credit activities. 1.11 World Bank Support for Agricultural Credit in Mexico: Mexico is the Bank's second largest borrower for agricultural credit. Since 1965, the World Bank has financed ten agricultural credit projects totalling US$1,805 million and six agricultural projects with credit components, for which US$334 million were earmarked for on-lending.4 The initial thrust of Bank policy was to promote agricultural production by large commercial farmers; however, by the fourth project (approved in 1973), the World Bank began emphasizing access to credit for low income producers. In order to reach them, BANRURAL was made eligible for Bank-supported FIRA rediscounts from 1977 onwards under the fifth and subsequent projects. Interest subsidies became the subject of increasingly frank exchanges between the World Bank and the Borrower, and a General Interest Rate Agreement (GIRA) was eventually reached in 1984. The GIRA called for scheduled increases in agricultural interest rates towards a market reference rate; however, the onset of inflation and rising market rates led to frequent breaches in the agreement and repeated interruptions in disbursements under the seventh and subsequent loans. The slow pace of interest reform had important fiscal and sectoral implications, and all five agricultural credit projects immediately preceding this one were rated unsatisfactory.' Furthermore, disagreement on the pace of interest rate adjustments led the World Bank and the Borrower to suspend preparation of a follow-on to this project and interrupted an constructive dialogue on reforming rural financial markets. 4. Throughout the report, US dollars means US dollars equivalent (in current terms, unless otherwise indicated). 5. OED: Project Performance Audit Report for Mexico, Fifth Agricultural and Livestock Credit Project (Loan 1217-ME), Sixth Agricultural Credit Project (Loan 1569-ME), Seventh Agricultural Credit Project (Loan 1891-ME), Eighth Agricultural Credit Project (Loan 2454-ME), Agricultural Credit Project (Loan 2610-ME), Report No. 8860, June 29, 1990. 4 II. DESIGN AND IMPLEMENTATION A. Project Design 2.1 The Context. The Eighth Agricultural Credit Project (Loan 2454-ME), approved in June, 1984, became effective in December, 1984, and was expected to close in September, 1988. However, by March, 1985, two-thirds of the US$300m loan had already been disbursed. In order to allow sufficient preparation time for the Ninth Agricultural Credit project, while maintaining the continuity of World Bank flows to the agricultural sector, an interim loan of US$180m (Agricultural Credit Project, or FIRA 8A, Loan 2610-ME) was approved in July, 1985. Over half the loan was disbursed by June, 1986, within six months of effectiveness, but in principle adequate time had been bought for a thorough preparation of the Ninth Agricultural Credit Project.' 2.2 Project Objectives and Description: The objectives of FIRA/FICART 9 were to (i) increase agricultural productivity and thus agricultural exports, real farm income and rural employment; (ii) help the Government improve agricultural sector policies by further reducing interest subsidies, and (iii) help to maintain the financial integrity of participating financial intermediaries. An important but unstated objective was to provide the Mexican Government with rapid access to foreign exchange: Mexico's debt-service ratio in 1987 was 55 percent and release of the third tranche of a commercial bank package supporting the Government's structural policy reforms was tied to Effectiveness of this loan. 2.3 The World Bank loan of US$400 million equivalent was to be on-lent by the Borrower (Nacional Financiera, or NAFIN) to FIRA (US$300 million) and FICART (US$100 million) for production credit (9 percent), investment credit (79 percent) and institutional development (2 percent). Ten percent of the loan was unallocated. The loan was to cover the foreign exchange costs of the project, or 40 percent of the total project cost of US$1,000 million. The remaining 60 percent of costs were to be financed by FIRA and FICART (32 percent), commercial banks (14 percent) and sub-borrowers (14 percent). The total project cost represented about 15 percent of FIRA's and FICARTs projected investment programme of US$6,911 million for the years 1987-1991. 2.4 Project Genesis: An initiating brief, dated February 5, 1985, indicated GOM's intention to request a follow-on to FIRA 8 in the amount of US$300-400. The Bank's initial position was that continued disbursement at high rates could only be justified if further progress was achieved on sectoral issues, primarily reductions in the interest subsidies to LIPs and improvements in BANRURAL's operations. The Appraisal mission that visited Mexico in November, 1985, recommended a 50 percent reduction in LIP subsidies during the project period, over and above GIRA requirements as a condition for negotiations. The GIRA was to end in January, 1987 with an increase in interest rates to LIPs from 70 percent to 80 percent of the market reference rate, namely 6. In practice, the preparation was far from thorough. Lending pressure entailed a high degree of tolerance for careless if not misleading analysis that cannot be attributed to a shortage of preparation time (see the section on the quality of appraisal (para. 2.12) and the section on disbursements (para. 2.9)). Thus the Audit does not concur with the PCR's assessment that the project could have benefitted from additional preparation (PCR para. 12). One exception, however, is the FIRA Headquarters sub-component, which was hastily appraised and included at the last minute. 5 the weighted average cost of funds (ACF) to commercial banks.' Thus project staff regarded this condition as having been met in August, 1986, when a letter of amendment to the GIRA set a target rate of interest of 85 percent of ACF for LIPs by July 1, 1987 and 90 percent by January 1, 1988.' The Appraisal mission also recommended inclusion of a component for FICART, which it found to be "a reasonably good institution, both in its administration and in its delivery of credit", provided that BANRURAL arrears did not exceed 15 percent of outstanding balances discounted by FIRA or FICART. A decision meeting in January, 1986 strengthened the requirements for FICART's participation in the project by calling for a study of the rural finance sector that would quantify subsidies to financial institutions. The aim of the FICART component and its conditionality was to acquire previously unobtainable information and thus to be in a position to influence developments in the entire rural financial sector. 2.5 In September, 1986, the loan committee introduced new conditionality for the project, requiring that FIRA and FICART become financially self-sustaining within five years. The same meeting decided to link disbursement of FIRA/FICART 9 to that of the 1986-87 commercial bank financing package for Mexico, in order to ensure that the project would be implemented within an acceptable macroeconomic environment. The commercial bank package was part of Mexico's Economic Program of Growth-Oriented Adjustment and Structural Reform, launched in October, 1986, and as mentioned earlier, release of the third tranche of the package was in turn tied to the Effectiveness of the FIRA/FICART 9 loan. 2.6 A Breakdown in Negotiations: Negotiations were held in November, 1986, but broke down over the issue of financial self-sustainability for FIRA and FICART."o The Government's position was that the trust funds were established with specific socio-economic goals that did not include financial viability. In January, 1987, GOM failed to increase the interest rate to LIPs to 80 percent of ACF, as agreed under the GIRA. With progress stalled on FIRA/FICART 9, an internal briefing note to World Bank management in April, 1987, reminded them that the commercial bank package was linked to loan Effectiveness, that another project would have to be brought forward to meet the US$2 billion commitment target for FY87, and that FIRA/FICART 9 would inevitably slip to FY88, "unless Bank management sees a way in which we could proceed more quickly". 7. This indicator is tracked by the Central Bank and reported on a monthly basis. It measures commercial banks' weighted average cost of term deposits, discounted notes and other deposits except savings and sight deposits. The cost of bank acceptances and bank-endorsed commercial paper have also been included in the ACF since November, 1988. 8. Since on-lending rates to LIPs would have to be well above 100 percent of ACF for commercial banks to have an adequate margin to cover administrative costs and doubtful loans, it was misleading to regard an increase in rates from 80 to 90 percent of ACF as a 50 percent cut in subsidies. 9. Appraisal Mission Issues Paper, dated January 10, 1986. 10. The Bank's position was that FIRA and FICART should earn a net operating profit that would be sufficient to maintain the real value of their equity and provide them with a return of at least 10 percent on their beginning-of-year equity. Their net operating profits would be adjusted as necessary to ensure that their average cost of funds would be equal to at least the Reference Rate, as defined in the GIRA. Furthermore, all grants and subventions received by FIRA and FICART would be related to corresponding development expenses and all extraordinary income and expenditure items would be excluded. Note that this definition of self-sufficiency would still have entailed subsidies to FIRA and FICART: the clause on grants and subventions was unclear, and the average cost of funds to commercial banks did not include costs associated with the reserve requirement (from which FIRA and FICART were exempted) or administrative costs associated with deposit mobilisation (which FIRA and FICART would not have had to incur). 6 2.7 Weakened Conditionality and Rapid Approval: At a meeting on April 27, 1987, chaired by the Senior Vice President for Operations, it was agreed that the ultimate goal was to promote loans at market rates for all borrowers, with up-front cash subsidies where required for social purposes. In the interim, rapid movement with some accommodation was required for FIRA/FICART 9. Satisfactory conditions for Board presentation would be: (i) GOM's agreement to increase LIP interest rates to 75 percent by May 1, 1987 and to 85 percent by December 1, 1987 (instead of 80 percent by January 1, 1987, 85 percent by July 1 and 90 percent by January 1, 1988, as called for by the amended GIRA); (ii) signature of a Memorandum of Understanding for an umbrella agreement on budgeting and controlling credit subsidies that would replace the GIRA; (iii) an increase in LIP interest rates to 90 percent of ACF during the project, pending agreement on the GIRA replacement; (iv) agreed terms of reference (TORs) for a study of financial intermediation and subsidies in the agricultural sector, and (v) a satisfactory policy statement by GOM on the protection of FIRA's and FICART's financial integrity. 2.8 Processing of the loan proceeded rapidly thereafter. A Memorandum of Understanding was signed in late April, the increase in LIP interest rates to 75 percent of ACF took place on May 1, a post-Appraisal update mission visited Mexico on May 4, a second round of negotiations was held on May 11-15 and agreement was reached on TORs for the rural financial sector study, as well as on maintaining the real value of FIRA's and FICART's equity at least at levels prevailing on January 1, 1986. The project was approved by the Board on June 16, signed on July 31, and became effective on August 28, 1987. 2.9 Rapid Disbursement by Design: The initial Project Brief, dated August 7, 1985, noted the rapid pace of disbursements under FIRA 8 and FIRA 8A, arguing that "[w]hile these two loans provide much needed resources to the agricultural sector, their quick disbursement limits the World Bank's ability to influence the development process". It therefore suggested that "to ensure that the Bank is allowed to fulfill its development role, it would be necessary that disbursements be spread over a reasonable period. As a result, it is proposed that Bank disbursements under the project would be limited to no more than US$125 million per year. This would be achieved by either imposing the above annual limit or by limiting Bank disbursements to an appropriate percentage of annual total lending." This ceiling was subsequently dropped, without any comment in the project files. 2.10 The Project Brief also proposed that "short-term credit should be limited to incremental credit in real terms", noting that FIRA 8 had financed short-term credit that was incremental in nominal terms only. This proposal was upheld during project preparation, and it was made clear to the Mexican delegation during the November, 1986 negotiations, that the World Bank wanted to finance "real" incremental short-term lending only. However, a memorandum to project files dated December 12, 1986, states that "[a]ctually, the way the condition in the Project Agreements...is spelled out, it allows some freedom--the word "real" is not specifically mentioned. The reason for this is that should Mexico need faster disbursements, this could be permitted at the discretion of the Bank by not imposing incremental in real terms". 2.11 At the second round of negotiations, the Bank agreed to retroactive financing of 10 percent of the loan amount (US$40 million) for discounts made after January 1, 1987, although they violated the GIRA, and agreed to a doubling of the Special Account to US$40 million "to facilitate loan 7 disbursements.."" As the PCR notes, there was no up-front conditionality, nor was there any linkage between disbursements and the preparation and delivery to the Bank of the Agricultural Financial Sub-sector Study (hereafter the Study), or to implementation of its recommendations. No justification is provided in the SAR for presuming that FIRA/FICART 9 would be disbursed over a four-year period (1987-91). Since the project amounted to only 15 percent of the FIRA/FICART investment programme for 1987-91 and no criteria were established to distinguish between project investments and non-project investments that met GIRA conditions, there was nothing in the appraisal or loan documents to preclude disbursement of the entire loan within one year. 2.12 The Quality of Appraisal: The Staff Appraisal Report (SAR) reveals several shortcomings that are attributable to poor analysis or to a deliberate attempt to remove any obstacles to Board Approval.12 For example, a memorandum by project staff dated June 1, 1987, called for the SAR to add a paragraph that drew attention to the poor collection performance of commercial banks. It noted that the reported 4 to 5 percent arrears rate (as a share of outstanding loans) did not reflect significant rollovers of dubious loans, and that the true arrears rate was closer to 30 to 40 percent. However, since loan conditionality limited World Bank rediscounts to participating banks with arrears under 15 percent of the loan portfolio and Board Approval might have been complicated by these data, Bank management decided to remove the paragraph from the final SAR. Similarly, the SAR for FIRA 8 reported loan losses at BANRURAL of around 20 percent of the portfolio. The SAR for FIRA/FICART 9 made no reference to the write-offs, reporting instead that BANRURAL's "overall arrears position of 9.5% in 1980 has been progressively reduced to 3%". 2.13 A third example is drawn from paragraph 2.13 of the SAR, which affirms that the decapitalisation of FIRA's equity in 1982/83 due to high inflation was reversed during 1984-86, when FIRA's equity increased by an annual average of about 42 percent. This was impossible, since, as the SAR failed to mention, inflation ranged from 59 percent to 105 percent during 1984-86. Finally, Annex 3 includes the following quixotic assertion: "...profits would be insufficient to offset the decapitalization of its loan portfolio caused by...remaining sub-loans at fixed and negative real interest rates. However, as those sub-loans are repaid (within the next five to six years), FIRA would become self-sustaining, earning sufficient profits to avoid decapitalization and further Government support." With FIRA's on-lending rate below the rate charged to LIPs (to provide commercial banks with a spread) and rates to LIPs below the ACF and the ACF below the inflation rate, this was also simply impossible. 2.14 Board Presentation: The need to avoid obstacles to Board Approval led to further lapses in the accuracy and completeness of information during Board Presentation. Staff affirmed that preferential rates were within a narrow band around the market cost of funds, failing to note that the rate to LIPs was 25 percent lower than the ACF and that the ACF itself was well below the rate of inflation. When asked directly whether very large farmers received any measure of subsidy, staff said no. There was no mention of the fact that at ACF plus one percentage point, the on-lending rate to large farmers was insufficient to cover the cost of funds, administrative costs and a provision for risk, that the on-lending rate was insufficient to maintain the real value of the loan, or that loans to 11. See the Summary of Negotiations, dated May 22, 1987. 12. See OED's "Review of World Bank Agricultural Credit Operations in Medco", dated July 21, 1988, which was prepared as a working paper for "A Review of Bank Lending for Agricultural Credit and Rural Finance (1948-1992)", OED Report No. 12143, dated June 29, 1993. Also, see para. 2.16 in this Audit Report. 8 producers in other sectors were at higher rates of interest. One Executive Director remarked on the decapitalisation of Mexico's agricultural credit system, drains on the central Government's budget and the misallocation of scarce resources. The loan was then approved. B. Project Implementation 2.15 General Performance: Difficulties arose during project implementation as a result of delays in adherence to the GIRA, disagreements with GOM regarding the Study and problems with the construction of FIRA's new headquarters. The project was extended by one year to allow for the completion of the civil works, eventually closing on June 30, 1992. Nevertheless, the three full supervision missions that visited Mexico in October, 1987, September, 1988, and March, 1989, gave the project respective ratings of 1, 2 and 1. Disbursements proceeded well ahead of schedule, while the objective of reducing interest subsidies was greatly assisted by a sharp decline in inflation. Indeed, interest rates to agriculture have been positive in real terms since 1988. Furthermore, the new Administration that took office in 1988 introduced a series of reforms that increased efficiency in the broader agricultural and financial sectors (see para. 3.41). These reforms benefitted the project, although few if any of them are attributable to it. 2.16 Project Costs and Financing: Project costs at completion were US$920 million, or 8 percent below appraisal estimates (see Table 1). The shortfall was more pronounced for medium and long term investments (9 percent) than for seasonal investments (1 percent), due to a sharp increase in real interest rates to agriculture. Expenditures on institutional development were almost one-third lower than projected, as a result of new restrictions on the purchase of vehicles and limited use of technical assistance. Panel 2 of Table 1 shows the distribution of project costs and World Bank lending for sub-loans by type of beneficiary. It suggests that LIPs received a considerably larger share of funds than was expected at appraisal. However, detailed comparisons with appraisal figures are vitiated by errors and internal inconsistencies in the appraisal data." 2.17 Overall, the World Bank financed 43 percent of project costs, as opposed to the 40 percent estimated at appraisal. Commercial bank loanees also contributed a larger share than projected, whereas FIRA/FICART, the commercial banks and in particular BANRURAL contributed less than originally planned (see Table 2). Although BANRURAL's shortfall was partly compensated by FICART, both BANRURAL's participation and that of its borrowers were lower than required under the BANRURAL Project Agreement (see PCR para. 42 and Part III, Table 5.2). 13. The SAR simply divides sub-loan costs equally between LIPs and Other Producers (OPs), even though the Loan Agreement allocates less funds for LIPs than for Other Producers and includes the same disbursement percentage for each category (which implies a lower project cost for LIPs than for OPs). There are several other difficulties with the Appraisal data. For example, the project cost table in Annex 1 of the SAR shows that participating banks would contribute US$148 million and sub-borrowers would contribute a further US$148 million. The financing plan on the same page includes contributions of $141 million by each group. It is furthermore clear, from a comparison of the project cost tables in Schedule A of the President's Report and Annex I of the SAR, that the latter erroneously allocates the entire PB and borrower contributions to long term investments and none to short term investments. As noted earlier, no attempt is made to reconcile the original appraisal cost allocation of US$16 million for institutional development with the total of US$22.5 million after negotiations, which includes funding for the FIRA Headquarters. Finally, both the project cost panel and the financing plan panel in Annex 1, Table 1 of the SAR estimate total project costs at USS6,911 million. However, Annex 1, Table 2 indicates total project costs of US$9,865 million. This major discrepancy is not the result of a typographical error, but rather of double counting of PB and borrower contributions. 9 Table 1: Appraisal and Actual Project Costs (US$ Millions) Appraisal Actual Panel 1: Total Project Cost Total Cost I Bank Loan Total Cost J Bank Loan Medium & Long Term Investments 785.0 314.0 714.9 317.2 Short Term Investments 192.5 37.0 189.8 77.0 Institutional Development 22.5 9.0 15.41 5.8 1 Total 1,000.0 400.0 920.o 400.0 Panel 2: Cost of Investments' Low Income Producers (LIPs) 492.0 169.5 487.2 243.2 Other Producers (OPs) 492.0 181.5 417.4 151.0 Unallocated -- 40.0 Total 984.01 391.0 6 394.2 a. The cost of the LIP and OP investments estimated at appraisal (US$984 million) does not equal the appraisal estimate of short plus medium and long term investments (US$977.5 million), because of an inconsistency in the SAR: the institutional development (ID) was originally costed at US$16 million, but with the addition of the new FIRA headquarters at the last moment, the ID component rose to US$22.5 million. The SAR reduced the size of the investment components in the cost table by type of investment (short term versus medium/long term), but not in the cost table by type of borrower (LIP versus OP). Table 2: Appraisal and Actual Project Financing (US$ Millions) Appraisal Actual Amount Share Amount Share The World Bank 400.0 40.0% 400.0 43.4% FIRA 205.5 22.3% > 318.0 31.8% FICART 65.9 7.2% Participating Banks 86.4 9.4% BANRURAL > 141.0 14.1% 1.8 0.2% Sub-borrowers' Contribution 141.0 14.1% 160.4 17.4% Total 1,000.0 100.0% 920.0 100.0% 10 2.18 Disbursements: Disbursements began rapidly (reaching 30 percent of the loan amount within four months of Effectiveness), but were held up briefly in the first quarter of 1988 because GOM failed to introduce the interest rate increases mandated by the GIRA. Indeed, the Bank had agreed to a temporary 5 percent reduction in long term rates for LIPs (relative to the ACF), together with a 2 percent increase in short term rates, effective November and December, 1987, with the understanding that rates would return to GIRA levels as of January 1, 1988. The Government adjusted agricultural interest rates in June and again in August, 1988, bringing them to levels exceeding those required under the GIRA. This opened the way for disbursement against loans discounted after April, 1988, even though they had not met GIRA conditions. By June, 1989, i.e. within two years of Board Approval, 93 percent of the loan had been disbursed. Slow progress on the FIRA headquarters led to a one-year postponement of the Closing Date, so that FIRA could use up World Bank funds for this component. Due to further difficulties with the contractors, the funds were reallocated for on-lending and the project was closed on June 30, 1992. 2.19 FIRA Headquarters: Funding for the construction of FIRA's new headquarters in Morelia, Michoacin, was added during the second round of loan negotiations in May, 1987. Draft bidding documents were sent to the World Bank one year later. The Bank questioned the sharp escalation in estimated costs and suggested revisions in the bidding documents. In January, 1989, the Bank warned FIRA to improve the consistency and completeness of its tendering documents "to protect FIRA from likely high-cost overruns and delays in the time during the construction period", and sent down a consultant with detailed comments to review the tendering documents. The Bank finally cleared the fourth version of the bidding documents in June, 1989. Although project staff at first objected to FIRA's criteria for the selection of a construction agency, they eventually agreed to FIRA's choice and the contract was signed in November, 1989. The contractor experienced liquidity problems from the outset, supplied less materials to the building site than might have been purchased with their 20 percent advance payment, and delayed construction throughout 1990. In March 1991, the World Bank was notified by NAFIN that costs had gone up from Mex$11,850 million (US$4.54 million) to Mex$27,297 million (US$9.65 million). Although FIRA supervisors were aware of the difficulties, feedback to FIRA management regarding the problems was inadequate and FIRA tolerated the delays until April, 1991, when the decision was taken to cancel the contract. The legal wranglings with the contractor that followed the cancellation remain unresolved. The Bank approved FIRA's decision to elicit bids for a new contract and a second contractor was hired early in 1992. Notwithstanding the extension of the project by one year, FIRA headquarters were unfinished at Project Closing. They were still incomplete in March, 1994, when the Audit mission visited Mexico, but by October, 1994 landscaping and interior arrangements had been completed and FIRA staff had moved in.'4 2.20 The Agricultural Financial Sub-Sector Study: Through the Study, the Bank hoped to acquire previously unavailable information and thus to be able to conduct a more informed policy dialogue with GOM regarding the rural financial sector. The Government was concerned that this policy dialogue would lead to conditionality in future operations that would be at odds with GOM's socio- political objectives for the agricultural sector. The Government agreed to the Study because the World Bank made it a condition for including the FICART component, but was averse to any Bank 14. FIRA staff attributed the delays in construction partly to the refusal of the State Government of MichoacAn (which ceded the 28- hectare site to FIRA) to allow the use of explosives for excavation of the foundations. Since the ground is hard rock, labour costs and construction time were considerably higher than projected. 11 involvement in the Study from the outset. The difference in perspective between the World Bank and GOM, which simply reflected a broader disagreement on the need for reforms in the sector, was not reconciled prior to Board Approval. Furthermore, provisions in the Guarantee Agreement regarding the Study were weak, possibly to avoid further delays in approval and disbursement of project funds. The inadequate preparation and weak conditionality for this key component greatly complicated implementation. 2.21 The Bank sent three missions in the first three months following Effectiveness to monitor progress on the Study and ensure that the January, 1988 deadline would be met. The deadline was optimistic. The Secretariat of Finance (SHCP), which was responsible for the Study, initially refused to provide details on the work programme or on the staff involved. However, draft plans for the Study were eventually forwarded to the Bank in November, 1987. In May, 1988, the Bank requested a copy of the first phase of the Study from GOM (field work had been postponed to a second phase). The Government reminded the Bank that Section 3.02 of the Guarantee Agreement only required GOM "..by not later than January 31, 1988, [to] carry out the study included in Part E of the project, under terms of reference as agreed between the Guarantor and the Bank, and thereafter exchange views with the Bank on the conclusions and recommendations of such study" [italics added]. A supervision mission in May/June, 1988 was duly given a copy of Chapter IV of the Study, entitled "Conclusions and Recommendations". The Bank argued that it was in no position to comment on the methodology, assumptions and data included in the Study without receiving a copy of the full document. After repeatedly requesting a copy of the Study from GOM and linking reallocations of funds between loan categories to receipt of the full document, the Bank was provided with a copy. The Study fell short of its stated objectives and the Bank warned that a follow-on operation would be contingent on a more detailed analysis of subsidisation, transaction costs and arrears. 2.22 Phase II of the study began in March, 1989, and the Bank was initially optimistic about its execution, but a February, 1990 supervision reversed this assessment, advising the Bank to obtain the raw data for its own analysis. GOM declined to provide the data, but sent a revised document in July, 1990. The analysis was still regarded as limited, but the Bank began to obtain data, and conducted its own analysis in the context of the preparation of a follow-on operation. 2.23 Monitoring and Evaluation (M&E): Towards the end of the project period, FIRA independently upgraded its capacity for monitoring rediscounts, known as the Sistema de Inspecciones por Muestreo (SIM). Currently FIRA monitors about 13 percent of sub-loans. Eighty percent of this sample is randomly selected, whereas 20 percent are chosen on the basis of prior information regarding possible irregularities. Both participating banks and sub-borrowers are monitored, and loans are recalled if off-farm diversion is uncovered. In the case of irregularities by the PBs, such as deviations from agreed loan terms, a penalty interest rate is charged. The penalty was increased in 1992 from 1/2 to 3 times the ACF. FIRA staff maintained that the incidence of diversions and other irregularities was quite low and that it had not varied significantly over time, although they could not present corroborating data to support the latter judgment (see para. 3.11). 2.24 FIRA's ex-post evaluation capacity was established at the World Bank's insistence in 1976. The SAR for FIRA/FICART 9 also noted (para. 3.18) that "[t]he project would assist FIRA to maintain itself as an efficient institution, strengthening those areas that show weakness, such as monitoring and evaluation procedures". Nevertheless, during the project period, FIRA's M&E unit received little attention from Bank supervision missions and the Sistema de Evaluaci6n de los Resultados del Crdito (SERC), as it is known, is not even mentioned in the PCR. There has also 12 been little support for ex-post evaluation within FIRA itself and in 1991 FIRA's management nearly abolished the SERC. The evaluation unit nevertheless tracked returns, equity and assets for a stratified sample of about 700 farms across the country during the project period, focussing on particular lines of crop and livestock production. It issued information bulletins covering the 1983-87 and the 1988-90 periods respectively that suggest that the farm investments supported by FIRA were financially viable (see para. 3.7). 2.25 At present, FIRA is attempting to integrate the SERC and the SIM with three other information systems, one for costs of production (SUECO), one for agricultural price indices (SIPA) and one for the ex-ante analysis of the profitability of various lines of production. The last of these was introduced in 1991 and has been well received by FIRA staff. The integration of these five sources of data should yield a unified information system of greater relevance to FIRA's needs. 2.26 Reporting and Auditing: The PCR reports inconsistencies in the financial records maintained by the Borrower (NAFIN) and the implementing agencies (FIRA and FICART, see PCR para. 81). Indeed, in 1991 the World Bank reallocated funds among disbursement categories since some were overdrawn, possibly as a result of inadequacies in record-keeping. Delays in reporting contributed to a slow response by FIRA management and by the World Bank to problems with the contractors for FIRA's new headquarters. However, compliance with the project's reporting and auditing requirements was generally good. On the World Bank's side, important weaknesses in appraisal documentation were identified earlier in this Report (see paras. 2.12 and 2.16). 2.27 Compliance with Covenants: The project had relatively few covenants and compliance was acceptable, albeit incomplete." The objective of improving agricultural sector policies by reducing interest subsidies was supported with a covenant requiring adherence to the GIRA. The Government made the required adjustments to interest rates with some delay, but the 1988 increases in interest rates for LIPs went beyond those required under the GIRA. The objective of maintaining the financial integrity of participating financial intermediaries was supported by a covenant in the Project Agreement with the Central Bank (BANXICO), requiring GOM to "take all necessary action to ensure that adequate funds are budgeted annually to FIRA and FICART so as to prevent any material capital erosion, in real terms, which FIRA or FICART may have incurred during the immediately preceding year on account of all their lending operations after January 1, 1986". This was adhered to for both FIRA and FICART by the end of the project period, although in December, 1987, FICART's equity was found to be 40 percent lower in real terms than in January, 1986, and FICART was dissolved after Project Closing. No such covenant was included in the Project Agreement for BANRURAL, which absorbed FICART's assets and liabilities in 1993. 2.28 Another covenant that was meant to support the objective of improving agricultural sector policies was the requirement that an Agricultural Financial Sub-sector Study be prepared and that its recommendations be implemented according to an agreed schedule (see para. 2.20 ff. and Section 3.02 of the Guarantee Agreement). The covenant was too weak to achieve its objectives. Not only did the covenant fail to guarantee World Bank access to the full Study, it also predictably failed to ensure that the Study would be of acceptable quality. Indeed, since the Bank was dissatisfied with the Study's analysis, it did not insist on implementation of the Study's recommendations. 15. See the PCR, Part III, Table 9, for a list of covenants and their status. 13 2.29 There were violations in the following covenants: first, in March, 1989, a World Bank supervision mission found that about $13.5m had been claimed by FIRA for disbursements at LIP rates to ejidatarios who earned more than 1000 times the minimum daily rural wage in their region. This problem arose because, until 1990, GOM defined all ejidatarios as LIPs, irrespective of their incomes. FIRA replaced these rediscounts with eligible sub-loans at the World Bank's insistence. Second, participating banks, notably BANRURAL, did not make the minimum contributions to sub- loans that were required under the Loan Agreement. The World Bank did not enforce this covenant. Finally, the Project Agreements required that access to rediscounts be suspended for participating banks whose arrears exceeded 15 percent of the aggregate outstanding amount of loans rediscounted under the project. However, due to the lack of a clear definition of what constituted arrears and to participating banks' ability to restructure loans just before the reporting dates, this covenant was hard to enforce, and indeed never was enforced. It appears to have been violated, insofar as 40 percent of BANRURAL's portfolio was written off in 1991 for collection by other agencies, suggesting that arrears exceeded 15 percent by any acceptable definition. 2.30 World Bank Supervision: The Bank was on the whole thorough and timely in its monitoring and support of project implementation, providing 10 missions and over 40 staff-weeks of supervision through June, 1989. One possible exception is the Bank's delayed response (in May, 1988) to GOM's failure to increase rates as required by the GIRA in January, 1988. The intensity of supervision dropped sharply after June, 1989, when most of the credit component had been disbursed. Indeed, there was a two-year hiatus in missions between June, 1989 and July, 1991, when a World Bank mission visited Mexico to address problems with the construction of FIRA's headquarters. The PCR (para. 75) rightly states that the project would probably have benefitted from more intensive supervision of the productive support component. However, World Bank staff took advantage of the project and of preparation of a follow-on operation to conduct a fruitful dialogue on rural financial sector reforms (see para. 3.10, 4.2). 2.31 Bank-Borrower Relations: As noted in the PCR (para. 78), the Bank-Borrower relationship was generally satisfactory. It can be divided into three distinct phases. The first phase, lasting from project preparation until the Summer of 1988, was marked by relatively paternalistic GOM policies towards agriculture and by sensitivity at what was regarded to be undue World Bank influence in the rural financial sector. Loan negotiations were broken off in 1986 over the issue of self-sufficiency for FIRA and FICART, the Government refused to allow the World Bank to participate in its review of BANRURAL's operations and, once the project was under way, GOM was reluctant to share the findings and data of the Study with the Bank. A change of Administration in 1988 accelerated the pace of reform in the Mexican economy, and the Bank-Borrower policy dialogue on rural finance flourished as GOM adopted a more laissez-faire approach to trade and finance. Most of this policy dialogue was conducted in the context of the preparation of a follow-on operation, to be financed by the World Bank and the IDB. This phase of the relationship ended late in 1992 with a breakdown in the dialogue that coincided with the promulgation of OD8.30 in the Bank, the adoption of a more doctrinaire approach to rural finance by the IDB, and importantly, electoral considerations in Mexico as the Sexenio (or six-year administration of the President) began to draw to a close. The Government recognised the economic merits of most of the World Bank/IDB policies but argued that the policies were inconsistent with GOM's socio-political objectives in the agricultural sector. There has been little serious exchange of ideas on formal agricultural credit since Project Closing. On the other hand, GOM and the World Bank have jointly undertaken research on informal financial markets during the past two years. It is expected that the Bank and GOM will conduct a broader and more intensive dialogue now that the new Administration has taken office. 14 III. PROJECT OUTCOME A. Farm-level Impact 3.1 Number and Type of Beneficiaries: According to FIRA/FICART estimates, the project exceeded appraisal expectations with regard to the number of beneficiaries. A total of 190,436 families (about 870,000 individuals) were reached, compared to the appraisal target of 100,000 families (about 550,000 individuals)."' It is also likely that the project helped to create more jobs than estimated by the SAR, although this cannot be confirmed because data on job creation are not available for FICART. About 72 percent of the appraisal target of 240,000 jobs was achieved through FIRA's project-related activities. 3.2 Although the reallocation of loan funds resulted in LIPs receiving 61.7 percent of World Bank rediscounts, compared to 50 percent at appraisal, the overall share of LIPs in the FIRA portfolio has declined dramatically since 1987, from 49 percent to about 19 percent of all rediscounts. FIRA's recent SIM data suggest that incorrect classification of wealthier borrowers as low income producers is negligible, but data were not available for the project period. Furthermore, for FIRA loans that were not discounted by the World Bank, LIPs were defined according to the Government's definition, which, until 1990, included all ejidatarios as low income producers. This made non-project FIRA resources available to an undetermined number of OPs at the subsidised rates intended for low income producers. Moreover, after Project Closing, FIRA began to use a cutoff of 3,000, rather than 1,000, times the regional minimum daily wage to distinguish between LIPs and other producers for the purpose of technical assistance grants." 3.3 The definition of Low Income Producers was a bone of contention between the World Bank and GOM during this and earlier projects. FIRA correctly pointed out that the real value of the minimum regional daily wage had fallen dramatically (see Annex 1.2), so that farmers could have experienced a real decline in incomes and still have graduated from the LIP category to become OPs. The World Bank stood by the increasingly strict definition, arguing that it had evidence from Mexican census data that 80 percent of farmers still fell below the 1,000-times-the-daily-wage cutoff. In practice, World Bank staff wanted to address the issue of targeting subsidies to poorer farmers more effectively in the context of the follow-on project; besides, the tightening of the definition was a convenient way of raising agricultural interest rates towards market rates for a broader segment of the market. 3.4 Throughout the 1980s, the World Bank suspected that masquerading of OPs as low income producers was far more widespread than FIRA indicated and that suspect definitions of producers' income were being used, but it could not muster sufficient evidence. The Audit team has not seen FIRA's SIM data for the relevant period and cannot comment on these issues, except to note that 16. The PCR states that 'The credit component of the project has benefitted an estimated 868,400 individuals and about 190,436 families.." [emphasis added], which suggests that the number of families has been double-counted. 17. At the end of 1989, (half-way through the project), the lowest regional minimum daily wage was about US$3.40 per day. A farm family with an annual income equal to 1,000 times this wage, i.e. (US$3,400) would have earned 35 percent more than the average Mexican (based on a GNP/caput of US$ 2,440). A farm income of 3,000 times the minimum daily wage would have equalled more than four times the national average. Since the minimum daily wage was more than US$3.40 in certain regions, farmers with slightly higher incomes could have fallen below the cut-offs in those regions. 15 the incentive for cheating has declined significantly, because the difference between LIP and OP rates is far smaller. Nevertheless, the incentive is still there and continues to entail unnecessary monitoring costs. 3.5 Returns on Investments: Mexican agriculture went through a difficult adjustment in 1988 and 1989, as a result of the removal of trade barriers following Mexico's accession to the GATT, the reduction of support prices for key crops, a sharp increase in real interest rates and the significant declines in input subsidies, including credit subsidies. However, agricultural production rebounded in 1990 and 1991, more or less returning to 1987 levels in real terms, so that the overall picture is one of stagnation, rather than decline, during the project period. 3.6 It is difficult to discern the impact of FIRA/FICART credit on agricultural production and productivity, and thus of the World Bank project on Mexican agriculture. However, the Audit does not agree with the PCR's contention that the project had no demonstrable impact because the World Bank financed only a small share of FIRA/FICART's lending and agricultural production was stagnant during the project period. As long as all credit is extended to sub-borrowers on the same terms, the share of FIRA/FICART credit that is rediscounted by the World Bank has no bearing on the impact of that credit." Furthermore, the stagnation observed in Mexico's agricultural economy during the project period does not rule out the conclusion that credit had a positive impact, since the relevant counterfactual may have been negative growth. This point holds a fortiori if one observes that conditions in the late 1980s were particularly difficult for less viable farmers, who had little or no access to FIRA/FICART funding, whereas asset and equity holdings of a representative sample of FIRA clients were found to have increased in real terms throughout the project period (see Table 3).19 3.7 While the impact of the project on agricultural production in Mexico cannot be quantified, it is feasible to consider whether or not the investments that the World Bank helped to finance were financially viable. To this end, Annex 1.13 presents figures on prices to farmers and costs of production per irrigated or rainfed hectare for wheat, maize, rice, beans and sorghum. Cost data were obtained from FIRA's SUECO database and price and yield data from the Secretariat of Agriculture (SARH). For each crop, financial prices and yields were used for the state that accounts for the largest share of Mexico's total output, e.g. Tamaulipas for sorghum and Sonora for wheat. The Audit concludes that investments in irrigated crops were financially viable, although not necessarily economically sound (because of the distorted policy environment). Gross returns to farmers were positive for irrigated maize, rice and wheat, and increased in real terms over the period 1986-1990 for all three crops, even though production costs rose more rapidly than output prices for maize and rice. The picture is less encouraging for rainfed production. Gross returns on rainfed 18. It is, however, true that the additionality and therefore the impact of the Bank's funds, as opposed to credit per se, is difficult to establish. Regional staff noted in an internal communication, dated November 30, 1994, that "without FIRA IX, FIRA would have resorted, as it actually did once the credit line was disbursed, to increase rediscounts from BANXICO" (sic.). This line of argument regarding the additionality/impact of Bank funding is clearly also applicable in sectors other than agricultural credit. 19. The data are drawn from FIRA's SERC database. For details on the methodology, see the Boletin Informativo: Evaluacion de Resultados de Credito, Etapa 1988-1990, prepared by FIRA's evaluation unit. Unfortunately the SERC data does not include a control group, therefore the difference in performance between FIRA and non-FIRA clients cannot be quantified. Furthermore, it is probably no coincidence that FIRA clients performed better than others, since they would in all likelihood have been offered credit precisely because they were good credit risks. This further vitiates attempts to isolate the "impact" of FIRA's credit. 16 sorghum were marginal and returns were generally negative on rainfed beans in Zacatecas until 1990.' These findings are less sanguine than the FIRA findings presented in Part III, Table 7 of the PCR. FIRA's models compare appraisal with re-estimated investment plans for rainfed annual crops, irrigated annual crops, beef cattle ranching and dual purpose cattle ranching. The farm investment plans suggest that re-estimated net benefits at full development were positive for all investments, and exceeded appraisal estimates in the case of irrigated annual crops and of beef cattle ranching. Table 3: Increments in Assets and Equity during Project Years for a Stratified Sample of FIRA Borrowers Sample Size Base Value in Value in Increment 1988 Mex$ '000s PBIs OPs Year Base Year 1990 Annual Crop Producers: 195 94 1988 Average Assets 330 473 44% Average Equity 303 412 36% Perennial Crop Farmers: 46 59 1985 Average Assets 316 616 95% Average Equity 295 560 90% Dual Livestock Producers: 89 55 1986 Average Assets 326 507 55% Average Equity 316 465 47% 3.8 The Audit mission met with larger farmers and groups of farmers in Guanajuato and Sinaloa and found that yields had increased following the opening of the Mexican market to competition from foreign producers. In particular, maize yields doubled to 7-8 tons per hectare for farmer groups with irrigated land in Sinaloa. Yields for wheat and barley increased by about 50 percent to 6 tons/ha for farmers influenced by the conservation tillage techniques disseminated by FIRA's demonstration centre in Guanajuato. However, yields increased far less for small, individual producers and were essentially unchanged for those operating on rainfed land. 3.9 Efficiency of Resource Allocation: Although the investments supported by the World Bank were financially viable, it is very likely that many of them were inefficient from an economic point of view. The price of fertilisers, water and electricity were highly subsidised, while prices to the producer were fixed by GOM for most crops until 1990, often at levels above world prices. For example, the PCR for AGSAL II estimates average nominal protection coefficients (NPCs) ranging 20. Between 1986 and 1989, yields on rainfed beans in Zacatecas were well below the 0.6 tons/ha required for an acceptable return to the farmer. If yields in Zacatecas had equalled the national averages of 0.45 to 0.6 tons/ha during 1986-90, gross returns would have been small but positive in four of the five years. 17 from 1.18 for sorghum to 1.65 for maize during 1988-92.1 It should be noted that the primary sources of inefficiency in this highly distorted environment were trade, exchange rate and input/output price factors, rather than credit at below-market rates of interest. These factors have been addressed quite comprehensively via trade and agricultural sector adjustment operations.' One concern that the Audit mission was not able to address was the question of whether below-market rates of interest had induced excessively capital-intensive development by sub-borrowers and therefore accelerated migration to urban areas. This concern merits investigation by Mexican research institutions. 3.10 Diversion of Sub-Loans: Diversion of loans appears to have been rampant among BANRURAL clients during the project period.' For example, the Secretariat of Finance (SHCP) estimates that BANRURAL financed about 800,000 ha of non-existent crops during 1988.' The Audit Mission was repeatedly informed by different sources about the high incidence of diversion of BANRURAL loans. The problem was attributable primarily to the poor design of the compulsory insurance scheme managed by ANAGSA and to weak supervision by BANRURAL. Low rates of interest were at best a minor explanatory factor, since pressure to repay the loan was minimal. The typical ruse involved diversion to consumption purposes of at least part of the BANRURAL loan, claims to ANAGSA for "crop damage" and reimbursement by ANAGSA to the farmer and to BANRURAL. According to a World Bank study, an incredible 68 percent of the area insured by ANAGSA was damaged or destroyed in 1989.' However, ANAGSA and BANRURAL agents are believed to have benefitted considerably under the scheme. 3.11 In sharp contrast, off-farm diversion of loan funds appears to have been fairly modest for FIRA/commercial bank clients. FIRA could not present data for the project period, but affirmed that there had been little variation in the incidence of "irregularities" since 1987. Their data for the third quarter of 1993 indicate that irregularities were found in 11 percent of cases surveyed. The incidence was higher for long term credit (16 percent) than for short term credit (5 percent) and for wealthier producers (13 percent) than for low income producers (9 percent). Thirteen percent of the irregularities were classified as diversions; this amounts to only 1.4 percent of the sample surveyed by FIRA. 3.12 Of course, the mere existence of a sub-project investment does not rule out the possibility that the sub-loan might have substituted for other funds which would otherwise have been used for the investment - these funds would be freed up for non-farm purposes as a result of the sub-loan. 21. Draft Project Completion Report: Mexico - Second Agricultural Sector Adjustment Loan (3357-ME), dated June 28, 1994. Note: the nominal protection coefficient measures the ratio of domestic to import parity prices. An NPC greater than unity implies a subsidy. 22. Trade Policy Loan 2745-ME, Second Trade Policy Loan 2918-ME, Agricultural Sector (AGSAL I) Loan 2918-ME, and Second Agricultural Sector Adjustment (AGSAL II) Loan 3357-ME. See OED's forthcoming Performance Audit Report on AGSALs I and II. 23. The Audit could not obtain data on whether the problem was less serious for sub-loans rediscounted by FIRA/FICART than for other loans in BANRURAL's portfolio. 24. See "Mexico - The Banrural Agricultural Credit System", LA2AG, 16th January, 1990. 25. The problem was also openly acknowledged in the press. For example, a front page headline in El Diario de Sinaloa (14th March, 1994) stated that "Diversions swelled the debt" and quoted a government official as saying that "a large part of the overdues of the ejidal sector is the result of diversions in the use of loans". 26. Page 10 of the BANRURAL study (see footnote 22). 18 Although conclusive evidence is not available one way or another regarding this issue, the Audit mission met with wealthy farmers in Guanajuato and Sinaloa who did not borrow when interest rates were highly negative (before 1988) but are currently borrowing in spite of high real interest rates, or are continuing to rely on own funds. Indeed, if loans had been taken out during 1986-90 at short- term OP rates to cover 80 percent of the working capital costs for selected irrigated crops (maize, rice and wheat -- see Annex 1.13), the resulting ex-post returns on farmers' equity contributions would have exceeded the market reference rate. Investments in rainfed crops would, ex-post, have performed poorly until 1990, relative to the market reference rate. The Audit found no support for allegations of widespread capital flight using FIRA/commercial bank loans. Allegations of this type had surfaced in the course of Bank operations in the early 1980s. Figure 1: Agricultural Lending by FIRA, BANRURAL and Commercial Banks 10 14- Commercial Banks 10- C rR 0 BANRURZAL 1986 1987 1988 1989 1990 1991 1992 3.13 Crowding-Out of Commercial Bank Lending: One type of substitution of funds that the Audit believes to have been important is the provision of FIRA-rediscounted sub-loans to wealthier commercial farmers who could have qualified for short-term credit at market rates using the commercial banks' own resources. This echoes the finding of the PAR on the previous five Bank- supported agricultural credit project, which notes that "[w]herever possible, commercial banks substituted FIRA funds for their own"." There is indirect evidence to support this postulate in the increase in commercial bank funding for agriculture that followed the introduction of the tasa mezcla system in 1989 (see para. 3.29). This system effectively liberalised interest rates on short term loans 27. OED Performance Audit Report No. 8860, paragraph 78. 19 to OPs by allowing commercial banks to blend funding at unrestricted rates with funds at controlled rates for an overall loan rate to OPs that roughly equalled market lending rates. That year, commercial banks lending to agriculture increased 139 percent in real terms, compared to 34 percent for HRA and a real decline for BANRURAL (see Figure 1).' Agricultural lending also grew more rapidly in 1989 than lending to other sectors (see Annex 1.5, 1.6). However, the problem of increasing arrears on agricultural loans soon curtailed the rapid growth in agricultural lending. B. Institutional Development 3.14 The project contributed to the institutional development of FIRA in three ways: it funded equipment for FIRA's demonstration centres, it provided training to FIRA staff and it supported the construction of FIRA's new headquarters. FICART's institutional development component was not financed by the project and little information is available about it since FICART has been disbanded. 3.15 Demonstration Centres: The Audit mission visited FIRA's demonstration centre "Villa Diego" in the State of Guanajuato. It had been transformed from a livestock centre to an agricultural centre in 1987, as agriculture became more important regionally and good results had been achieved with livestock. Villa Diego is one of six remaining demonstration centres administered by FIRA - there were nine at project completion, but three centres, owned by the State Governments of Tlaxcala, MichoacAn and Guerrero, have since been closed. While the Audit mission found that the centre was having a noticeable impact on the production of nearby farmers, it did not find that the project had a significant effect on the activities of the demonstration centre. Furthermore, much of the dissemination is conducted via meetings held on over 400 demonstration farms that cultivate experimental plots with FIRA guidance and support, but civil works were restricted to the nine centres because Mexican law precludes permanent investments on demonstration farmers' lands. FIRA and GOM restrictions also constrained investment in equipment and vehicles - less than 30 percent of the World Bank funds allocated for these items were disbursed. 3.16 Training and Technical Assistance: External training and the use of consultants were also restricted under regulations introduced by the new Administration in 1988-89. Only one foreign consultant was hired, to evaluate bidding documents. Greater reliance on consultants in construction management would have been helpful. Foreign training consisted mainly of study tours, seminars, English language courses, conferences and trade shows. Only one foreign postgraduate degree was financed with project funds. However, 631 events were organised during the project period to provide training to FIRA's technical and administrative staff. A total of over 7,200 participants benefitted from courses including computer technology, strategic planning and financing, economics, and credit supervision.' An additional 226 events were organised for training small farmers, 28. In 1989, the entire banking system's agricultural portfolio rose by an average of 45 percent in real terms, compared to 27 percent for the system's non-agricultural portfolio. The exceptionally rapid expansion in agricultural lending reflects increases in relative, rather than absolute, expected returns on agricultural loans. The tasa mezcla is an important explanatory factor for the relative increase in expected returns on loans to agriculture. The expansion of FIRA/PB credit also allowed commercial banks to issue more bridging loans in the form of preswmos quirografarios to their agricultural clients. These bridging loans were at higher interest rates and were often recovered from the rediscounted portion of the FIRA/PB loans. Finally, while interest rates were high and spreads on agricultural loans were expressed as a percentage of ACF, it was more attractive to rediscount agricultural loans with FIRA. However, the decline in the ACF reduced the spread on rediscounted sub-loans until these spreads were well below those obtainable using commercial banks' own funds without FIRA rediscounts (see Table 5), i.e. substitution of FIRA funds for own funds became less attractive. (This eventually led to the de-linking of spreads on rediscounted sub-loans and the ACF.) 29. The total number of participants is higher than the number of beneficiaries, since many beneficiaries participated in more than one event. 20 benefitting about 6,600 participants.' The long term benefits of this training has not been evaluated by FIRA and cannot be determined by the Audit. 3.17 FIRA Headquarters: The new FIRA headquarters are still incomplete. The site is impressive, located on 28 hectares of land on the outskirts of Morelia, and the large main structure appears to be well constructed. Although the designs provided an already generous 7,700m2 of office space for 350 staff, this was grossed up about 100 percent to over 15,000m2, instead of by the usual 35 to 45 percent grossing factor, to cover conference rooms, halls and so on. During the Audit mission's visit in March, 1994, preparations were being made to elicit bids for landscaping and completion of the interior. FIRA staff eventually moved into the new facility in October, 1994. 3.18 Commercial Banks and BANRURAL: During the project period, commercial banks also benefitted from institutional development, notably in the form of FIRA technical assistance and training for their agricultural staff. However, since the privatisation of commercial banks in 1991, the number of agricultural loan officers has declined steadily. This decline is not unwarranted, since the use of private agricultural consultants as intermediaries between banks and borrowers has increased considerably. The cost of their assistance is covered under the sub-loans and is frequently subsidised via FEGA. One concern, however, is that their technical assistance focuses on financial intermediation rather than on improved cultivation practices. Another new service that has eased commercial bank processing of agricultural sub-loans is on-line access to a computerised FIRA database on agricultural investments. This service was introduced after Project Closing and is not attributable to the project, but it is highlighted as it has been well received by commercial bank staff. The Audit also found that significant institutional development of BANRURAL began during the project period (although not with project funding), particularly in the form of training programmes and upgrading of computerised systems. C. Impact on Financial Intermediaries 3.19 FIRA: Section 2.03 of the Guarantee Agreement required the Government to provide adequate funding to FIRA and FICART "so as to prevent any material capital erosion, in real terms, which FIRA or FICART may have incurred during the immediately preceding year on account of their lending operations after January 1, 1986". Government fulfilled the covenant and by 1991, FIRA's equity was 20 percent higher in real terms than in January, 1986. FIRA also bore no credit risk on its loans, because its trustee, Banco de Mexico, ensured that FIRA's loans to commercial banks were repaid automatically on their due date from commercial bank reserve funds with BANXICO. 3.20 Clearly, FIRA could not have achieved its strong financial position without GOM's support. A subsidy dependence index (SDI) was calculated to assess the extent of FIRA's reliance on subsidies (see Table 4 below and Annex 1.9).' It suggests that if FIRA had borrowed at market rates of interest and had recorded no Government transfers in its income statement, it could have earned a market return on its equity by charging interest rates 85 to 121 percent higher than the rates it actually charged. For example, in 1990, FIRA earned 23.9 percent on its loan portfolio. If FIRA's 30. Part 11 of the PCR reports just 3,907 domestically trained staff and PBI, since it only covers the period 1987-89. 31. The SDI was applied to the consolidated balance sheet of FONDO, FEFA and FOPESCA (FEGA was excluded from the analysis). For details on the methodology see J. Yaron: "Assessing Development Finance Institutions - A Public Interest Analysis", World Bank Discussion Paper 174 and OED: Jamaica - Export Crops Project (Loan 2414-JM), Performance Audit Report No. 10656. For a quick overview of the SDI approach, see Box 8.1 in the World Bank, Operations Policy Department: "Handbook on Financial Sector Operations", 1993. 21 average on-lending rate had been 85.4 percent higher, i.e. 44.4 percent instead of 23.9 percent, FIRA would have been self-sufficient, because the resulting spread of 4.3 percent over the market reference rate of 40.1 percent would have been enough to cover administrative costs and earn FIRA a market return on its equity. With inflation at 29.9 percent in 1990, the nominal rate of 44.4 percent required for self-sufficiency would have equalled 11.1 percent in real terms." Table 4: Subsidy Dependence Indices for FIRA (Figures in Percent) Panel 1: SDI for FIRA, Including Costs of TA 1988 1989 1990 1 1991 11992 1993 11 FIRA's average on-lending rate to PBs 43.5 30.8 23.9 14.2 11.0 11.5 21 Subsidy Dependence Index for FIRA 94.0 90.3 85.4 102.9 120.9 85.5 31 Minimum on-lending rate for self- 84.4 58.6 44.4 28.8 24.2 21.4 sufficiencya 41 Market reference rate 78.4 52.9 40.1 25.6 21.8 17.9 51 Spread required for self-sufficiency, [31 - [41 6.0 5.7 4.3 3.2 2.4 3.5 [Panel 2: SDI for FIRA, Excluding Costs of TA 1988 1989 1990 1 1991 1199211993 61 Subsidy Dependence Index for FIRA 88.6 82.8 78.4 93.3 109.5 74.1 71 Minimum on-lending rate for self- 82.1 56.3 42.7 27.4 23.0 20.1 sufficiency' 8] Spread required for self-sufficiency, [71 - [4] 3.6 3.4 2.6 1.8 1.2 2.2 a. This is found by multiplying the average on-lending rate in line [1] by (1+SDI), where the SDI is given in line [2]. E.g. for 1988, 84.4% = 43.5% x 194%. b. The minimum on-lending rate for self-sufficiency excluding TA is derived by multiplying the actual average on-lending rate in line [1] by (I+SDI), where the SDI is now given in line [6]. E.g. for 1988, 82.1% = 43.5% x 188.6%. 32. The market reference rate was determined by adding a risk margin of 3 percentage points to a base rate and adjusting the resulting rate for reserve requirements. The base rate used by the Audit was determined by the GIRA. For 1988 and 1989 it was the average cost of funds (see footnote 5); thereafter, it was the 28-day Mexican treasury bill (CETES) rate. FIRA's SDI was calculated as the ratio of total subsidies to interest income. For example, in 1990 subsidies to FIRA (excluding FEGA) were MexS1,223 million and interest income was Mex$1,432 million, resulting in an SDI of 85.4 percent. That year, total subsidies comprised Mex$1,051 million as a result of access to liabilities at an average cost of 14.6 percent instead of the market reference cost of 40.1 percent, and Mex$172 million as a result of access to equity at a below-market rate of return (the return on equity was only 32 percent). No grants or subventions were recorded in the income statement for 1990. See Annex 1.9 for annual figures on the subsidy to FIRA, excluding FEGA, broken down by source of subsidy. 22 3.21 Since FIRA's activities reach well beyond simple financial intermediation to include a substantial technical assistance programme, the SDI was also calculated without FIRA's income and expenses related to technical assistance. The Audit finds that on-lending rates would have had to increase by less, namely by 74 to 110 percent. At just 2.6 percent, the spread over the market reference rate required for self-sufficiency in 1990 would also have been lower in the absence of FIRA's technical support activities. 3.22 An important finding is that FIRA became a more efficient institution in the course of the project. The spread required for self-sufficiency declined from almost 6 percent in 1988-89 to around 3 percent during 1991-93. If FIRA's technical assistance activities are excluded, the required spread would have been between 1.2 percent and 2.2 percent during 1991-93, down from about 3.5 percent in the early years of the project. This was achieved by halving FIRA's ratio of administrative costs to risk assets. It should be noted that, notwithstanding a sharp fall in the SDI in 1993, relative to 1992, FIRA actually became less efficient in 1993, insofar as its margin for self-sufficiency widened by one percentage point. This increase is almost entirely attributable to increased intermediation costs rather than to higher technical assistance costs (see rows [5] and [8] in Table 4). 3.23 FEGA: FIRA's trust fund for guarantees and technical assistance performed less well than the trust funds that primarily provided credit (FONDO and FOPESCA also provide loan guarantees, but these are for OPs and fisheries, rather than for low income agricultural producers). The opening of the economy to competition from foreign agricultural producers, together with a sharp increase in real interest rates in 1988-89, led to an increase in arrears and therefore higher guarantee payments by FEGA (see Annex 1.8 and the PCR, Annex A). Furthermore the area indemnified by agricultural insurers fell sharply with the dissolution of ANAGSA, leading to a greater call on FEGA guarantees on loans to LIPs. FEGA's premia are insufficient to cover its payouts and net payments (after recoveries on paid guarantees) increased more than five-fold in real terms over the life of the project. 3.24 FICART/BANRURAL: The real value of FICART's equity was maintained after 1988, (although it had been allowed to decline during the first two years of the project) and in 1991 it was 9 percent higher than in January, 1986. The increment (1991 Mex$133 million) is equal to a mere 10 percent of the real value of total Government transfers to FICART during 1986-91." With little rationale for its existence and high operating costs, FICART was dissolved as of December 1992. 3.25 BANRURAL made significant losses every year and the real value of its equity fell 20 percent in the first year of the project. Thereafter, massive capital injections (almost US$3,000 million in 1990 alone) contributed to an overall increase in real terms in BANRURAL's stated equity over the project period. The Audit did not attempt to calculate the subsidy dependence index for BANRURAL because audited financial statements for the project period conceal an extremely serious problem with loan recoveries. BANRURAL's assets and equity are significantly overstated and its operating losses understated. It is estimated, for example, that in 1988 direct recoveries from borrowers amounted to no more that 10-15 percent of the volume on-lent and that indirect recoveries from ANAGSA only increased the overall recovery rate to around 60 percent of annual lending.' 33. Government transfers to FICART during 1986-91 totalled 1991 Mex$1,343 million, or about 1991 US$330 million. 34. See LA2AG: "Mexico - The Banrural Agricultural Credit System", 16th January, 1990. 23 Furthermore, the arrears figures were kept down by rescheduling and rolling over as much as one- third of the portfolio. In 1990, around 40 percent of BANRURAL's portfolio was written over to FIRCAVEN and to PRONASOL, which inherited BANRURAL's weakest clients. FIRCAVEN's recoveries were negligible. There are no data on PRONASOL recoveries for former BANRURAL clients, but PRONASOL's overall recovery rate on its unsecured loans is around 55 percent. 3.26 During project implementation, the Bank and GOM began an intensive dialogue on BANRURAL that contributed to positive measures aimed at addressing the above problems. Since 1990, the Government has allowed BANRURAL to shrink considerably and bear the losses it makes on bad loans, in an attempt to increase efficiency and reduce the cost of lending to small farmers. In turn, BANRURAL has attempted to counter the culture of non-payment more consistently by denying credit to delinquent borrowers and by applying a policy of zero tolerance for irregularities by BANRURAL staff. As an added incentive for timely repayment or prepayment of sub-loans, BANRURAL is offering borrowers interest rebates of up to 10 percent of the total interest paid. Local pressures on BANRURAL staff to continue with the old ways are still considerable, but BANRURAL's senior management appears to be supporting rural managers' attempts to apply sounder credit policies. In February, 1994, BANRURAL's recovery rate had reportedly risen to around 80 percent of current dues. 3.27 Commercial Banks: The US$300 million FIRA/commercial bank component of the loan amounted to less than one-half of one percent of Mexican commercial banks' assets in 1989. However, the World Bank's influence on PB's lending to agriculture was fairly significant, because GOM consulted with the World Bank during project implementation regarding spreads on agricultural loans. The Bank had hoped to base its recommendations on the Study's findings regarding the adequacy of the financial margins accorded under the GIRA (10 percent of ACF for loans to LIPs and 7 percent of ACF for loans to OPs), but the Study never examined the costs of intermediation or the adequacy of spreads for commercial banks. No data are available on the profitability of agricultural lending by commercial banks." 3.28 In July, 1988, the World Bank refused to agree to an increase in spreads to 12.5 percent of ACF on loans to LIPs and 8.5 percent of ACF on loans to OPs, because the increase would have been at FIRA's and FICART's expense and would therefore have increased subsidisation to the rural financial sector. However, following a sharp decline in the reference rate (ACF), the World Bank concurred with GOM's decision to change the spread from a given percentage of ACF to given percentage points of interest, 6 on loans to LIPs and OPs who produced basic products (hereafter "basic OPs"), and 5 percentage points on loans to other OPs.' Commercial banks had proposed spreads of 10 to 12 percentage points, and project staff estimates suggest that administrative costs for processing agricultural loans are about 8 to 10 percent of amounts lent, but the 5 to 6 percentage point spread on agricultural loans is augmented by PBs in several ways: first, they receive transaction cost subsidies from FEGA; second, they assess several fees for services they provide, e.g. registering the contract and transmitting the loan; third, they often require that compensating balances be maintained with them (notwithstanding SHCP regulations); fourth, they often advance bridge loans 35. Commercial banks did provide elaborate models to support their request for a large increase in spreads on agricultural loans, but these are likely to have been inflated. Hard data on administrative costs and returns on agricultural loans were not provided. This deficiency suggests that reporting requirements for participating banks may have been inadequate. 36. Basic products are 15 essential crop and livestock products for Mexican consumers, including maize, beans, rice, milk and eggs. 24 at commercial rates of interest, informing borrowers that the FIRA loan has not yet come through (even though the bulk of the delays occur at the commercial bank level), and then recover the bridge loan through the FIRA loan." Interviews by the Audit mission at the farm level and the FIRA level provided independent confirmation of the use of this approach, which is not peculiar to Mexican banks.' Table 5: Example of the Tasa Mezcla's Effect on Subsidies and Interest Rates for a Hypothetical Short Term Loan to a "Basic OP" Share of Cost of On-lendinT Spread1 Total Loan Funds Rate __ I Commercial bank participation on-lent at FIRA rate 20% 18.0% 21.0% 3.0% FIRA participation on-lent at FIRA rate 40% 15.0% 21.0% 6.0% Extra commercial bank funds on-lent at unrestricted rate 40% 18.0% 31.0% 13.0% Overall (short-term) loan to a "Basic OP" 100% 16.8% 25.0% 8.2% Assumptions This is a hypothetical example of a short-term commercial bank loan of Mex$100 to an OP who produces basic products and therefore qualifies for FIRA credit at CETES + 3 percentage points. The commercial bank could rediscount 80 percent of the loan with FIRA and make only the 20 percent mandatory contribution. Instead it decides to rediscount 40 percent of the loan and cover the remaining 40 percent with its own voluntary contribution, which it can on-lend at an unrestricted rate of interest. Suppose that ACF = CETES = 18 percent, and that this is the cost to the bank of its own resources. Since GOM accords a spread of 6 percentage points to commercial banks on loans rediscounted with FIRA, and FIRA's rate to sub-borrowers is CETES + 3 percentage points, or 21 percent, the cost to the commercial bank of the FIRA resources is 15 percent. Finally, suppose that the average on-lending rate in the market for loans to other sectors is ACF plus 7 percentage points, i.e. 25 percent. Findings By charging an interest rate of 31 percent on the unrestricted portion of the loan, the commercial bank can obtain an overall return on the loan that is equal to the market rate of 25 percent. The subsidy of 4 percentage points intended for the sub-borrower (i.e. the difference between the controlled rate of CETES plus 3 percentage points, or 21 percent, and the market rate of ACF plus 7 percentage points, or 25 percent) is absorbed by the bank. Without the FIRA funds, the bank's spread would have been 7 percentage points (since the market rate is 25 percent and the ACF is 18 percent). With the FIRA funds and the tasa mezcla, the spread increases to 8.2 percentage points, i.e. the difference between the market rate (25 percent) and the average cost of funds for this loan (16.8 percent). Note: without the tasa mezcla, the bank could have rediscounted 80 percent of the loan with FIRA, entailing an average cost of funds for the loan of 15.6 percent, and on-lent 100 percent of the resources at the FIRA rate of 21 percent, for a spread of 5.4 percent. 37. According to figure for 1992-93 from FIRA's SIM database, participating banks on average take almost three months (81-86 days) to process sub-loan applications. 38. See Jamaica - Export Crops Project, Loan 2414-JM, PAR no. 10656. Other features of Mexican commercial bank lending to agriculture that are observed elsewhere are collateral requirements well in excess of the loan amount as well as undervaluation of collateral. 25 3.29 However, the key factor that increased commercial banks' spreads on agricultural loans was the introduction of the tasa mezcla, or mixed rate on short-term loans to OPs, in June 1989. This scheme allows commercial banks to charge unrestricted interest rates on any funds that they provide over and above the 20 percent participation required on short term loans to basic OPs. For example, these sub-loans might be funded by a 20 percent mandatory participation at controlled rates by the PB, an additional voluntary contribution of 40 percent by the PB at unrestricted rates, and a 40 percent (rather than 80 percent) participation at controlled rates by FIRA (see Table 5 and accompanying notes for an example). In practice, the unrestricted portions of the loans have been provided at rates well above market rates, bringing the average interest rate to sub-borrowers approximately to market rates. 3.30 The World Bank's initial reaction to the scheme was cautious, because it was concerned that the application of the scheme only to short term loans to OPs would increase fragmentation of rural financial markets and reduce the transparency of subsidies." Project staff recognised that the FIRA rate to sub-borrowers (and thus to participating banks) would no longer determine the final rate to sub-borrowers on the entire loan (which would approximate market rates), so that the PBs could appropriate the subsidy intended for borrowers. However, Bank staff were more concerned with the overall level of subsidies than with the allocation of those subsidies between financial intermediaries and ultimate borrowers. They did not comment on the inappropriateness of the tasa mezcla scheme as an instrument for targeting subsidies.' The World Bank at first agreed to the scheme as a temporary measure, calling for further improvements at a later stage, but during preparation of the proposed follow-on project, the World Bank accepted the tasa mezcla as a longer term arrangement and the scheme is still in place. 3.31 An important concern regarding agricultural lending, and indeed all lending by PBs, has been the rapid rise in arrears as a percentage of the loan portfolio (see Figure 2 below). In 1992, total commercial bank arrears exceeded the value of their equity (excluding revaluation of assets). Furthermore, the arrears figures do not include loans that have been restructured. 3.32 The problem is particularly severe in the agricultural sector. During the Audit mission in March 1994, GOM and the Mexican Banking Association (AMB) were negotiating a massive programme of restructuring of agricultural loans. It was estimated that commercial banks' contaminated agricultural portfolio amounted to Mex$5,000 million (US$1,540 million), of which about half was overdue. Most of the overdues are on commercial bank funds, rather than FIRA (or World Bank) funds, since commercial banks advanced bridging loans, known as pr6stamos quirografarios, to agricultural producers at unrealistically high rates of interest in the late 1980s. Commercial bank staff attribute the arrears to the very rapid growth in indebtedness (a fivefold increase in real terms during 1987-93), a sluggish economy, poor risk management, weak monitoring, and provision of short term loans for sub-projects that should have been financed with long term loans. Indeed, the Audit mission met with large agricultural and aquaculture producers in Sinaloa 39. Combined Back-to-Office Report and Full Report, dated July 11, 1989, for a Partial Supervision of the Ninth Agricultural Credit Project and Preparation of the Proposed Rural Financial Subsector Project. 40. The subsidies inherent in the scheme are unrelated to loan size. This deficiency would hold even if transparency were to be enhanced by combining the two PB contributions, as suggested by project staff, and if fragmentation concerns were to be addressed by ertending the scheme to all categories of sub-borrowers. 26 who were experiencing liquidity problems but were continuing to invest in recovery of adjacent lands for production." Since 1990, commercial banks have been considerably more selective in their lending, and credit to the agricultural sector has declined as a share of their total lending. Figure 2: Ratio of Arrears to Loan Portfolio for Commercial Banks on Loans to Agriculture and All Sectors 0.0 7%- ;5.5% 6%- 5.5% c5%- 0 ..6 IZ4%- 0 2.6% .2% .0% CL 2%-1.2% 19% Agric. Portfolio 0% Total Portfolio 1988 1989 1990 1991 1992 D. Impact on Sectoral Reforms 3.33 Key Issues: The World Bank's principal concern regarding the Mexican rural financial sector was that FIRA and FICART/BANRURAL were providing credit to a small fraction of Mexican farmers at subsidised rates of interest. The subsidies tended to accrue to wealthier farmers and to encourage inefficient use of sub-loan funds (or of the equity that they replaced). Furthermore, the programme was extremely expensive for the Government. The project therefore adopted a two- pronged approach to reduce interest subsidies and encourage further reforms in the rural financial sector. First, it mandated increases in interest rates to sub-borrowers in accordance with the GIRA, and second, it called for sectoral reforms based on the recommendations of the Agricultural Financial Sub-sector Study. 3.34 Reduction in Interest Subsidies to Farmers: At the Decision Meeting held in January, 1986, it was agreed that "subsidy control and targeting would involve a saving to the Treasury in the order 41. Another factor explaining the arrears is the use of moneylender credit during the planting season to proceed with land preparation and sowing before the commercial bank or Banrural sub-loan arrives. The delays in the formal bank loan are attributable to slow processing by PBs, as well as to late applications by sub-borrowers. 27 of 0.4 percent of GDP, and that this impact would be sufficient to support the proposed loan". Between 1987 and 1991, interest subsidies to final borrowers through BANRURAL fell by an estimated 0.52 percent of GDP (5.9 percent of agricultural GDP, see Figure 3 below and Annex 1.12). Subsidies on loans via FIRA fell from 0.22 percent of GDP in 1988 to 0.12 percent of GDP in 1991-92 (or from 2.7 to 1.4 percent of agricultural GDP). Thus, the objective of reducing interest subsidies through these institutions by 0.4 percent of GDP to was clearly achieved during the project period. This success is of course offset to some extent by increased budgetary commitments to PRONASOL, whose beneficiaries include former BANRURAL clients. Figure 3: Interest Subsidies on Loans via FIRA and BANRURAL as a Percentage of GDP' 0.7% 0.6% ....... .. 0.5%- .20.4%- :2 (n0.3% - .S 0. 2% - <0.1% a. FIRA 0.0% n. BANRURAL 1987 1988 1989 1990 1991 1992 3.35 Other Reforms in Agricultural Finance: Prior to the project, the World Bank's policy objectives centred on the reduction of interest subsidies to final beneficiaries. During the project period, the World Bank formulated a broader policy agenda that was crystallized in the preparation of the proposed follow-on operation. The policy agenda included greater transparency in the provision of subsidies to FIRA and BANRURAL; self-sufficiency for the two institutions; increased savings mobilisation and better lending practices by BANRURAL; increased competition in the agricultural insurance market; more incentive-compatible insurance policies and practices at ANAGSA, and the assessment of actuarially fair premia by the Government-owned agency. 42. Note: there are no figures for FIRA in 1987 and Banrural in 1992. 28 3.36 Considerable progress was made towards these objectives during the project period, although not necessarily as a result of the project. The 1989 Convenio de Derivaci6n de Fondos transferred the onus for repayment of external loans from GOM to development banks and trust funds that had previously received the loans as equity injections. Any differences between the development banks' repayments to GOM (at the CETES rate) and the actual interest plus foreign exchange risk liabilities to the external donor were specifically budgeted by GOM. This increased the transparency of subsidies to FIRA and BANRURAL considerably, although the World Bank remains concerned about the provision of hidden subsidies to FIRA via concessional loans from BANXICO.' Progress on self sufficiency for FIRA and BANRURAL has been slower: while the SDI for FIRA excluding FEGA fell marginally during the project period (from 94 percent to 91 percent), subsidies to FEGA have increased considerably. The increase is attributable to external shocks to the agricultural sector, as well as to higher real interest rates, which affected both sub-borrowers' capacity and their willingness to repay. Precise data on BANRURAL are unavailable, but its administrative costs (including write-offs and provisions for depreciation) have risen from 21 percent of the loan portfolio in 1986 to almost 32 percent in 1992, without a corresponding increase in its financial margin." Nevertheless, progress has been made in sanitising BANRURAL's portfolio, recording overdues more realistically and improving BANRURAL's lending practices (see para. 3.24 ff and Annex 1.10). 3.37 Very slow progress was made on savings mobilisation: the volume of deposits mobilised by BANRURAL rose by only 4 percent in real terms between 1986 and 1992, but their share of total liabilities rose from 4 percent to 25 percent during the period. It would arguably have been inappropriate for BANRURAL to have mobilised substantial savings without first improving the quality of its portfolio. Finally, ANAGSA was replaced by AGROASEMEX in 1990 and FICART was absorbed by BANRURAL in 1992. AGROASEMEX is operating more commercially than its predecessor and its premium income exceeds its indemnity payments. However, AGROASEMEX is still dependent on Government subsidies and has a significant monopoly of the agricultural insurance market. 3.38 The Role of the Agricultural Financial Sub-sector Study: Many of the World Bank's sectoral objectives outlined above were expected to be supported by findings in the Agricultural Financial Sub-sector Study. Indeed, GOM officials suggested to the Audit that the World Bank approached the rural financial policy dialogue with orthodox preconceptions and used the Study to confirm their beliefs, rather than as a diagnostic tool. In practice, the Study could not fail to provide evidence of costly distortions that all parties were fully aware of, and project staff argue that it was their quantification of the costs of distortions, e.g. those attributable to the BANRURAL/ANAGSA nexus, that prompted the reforms undertaken by GOM. 3.39 While the PCR appropriately downplays the overall value of the Study's analysis, it credits the Study with a greater impact than it probably had (see PCR para. 57)." The Secretariat of Finance had already begun to study the BANRURAL system in 1986. Besides, much of the policy dialogue between the World Bank and GOM during the project period was conducted in the context of a proposed Rural Financial Sub-sector Project, for which an Initial Executive Project Summary was 43. For example, in 1992 FIRA's average cost of funds (7 percent) was less than half the CETES rate of 14.9 percent. 44. BANRURAL's financial margin reportedly rose from 8.2 percent in 1986 to 19.8 percent in 1989, before falling back to 8.4 percent in 1992. 45. The Study is credited, inter alia, for increases in PBI rates, the adoption of the tasa mezcla, the restructuring of Banrural, the closing of ANAGSA, the elimination of FICART and the introduction of a new transaction cost subsidy scheme for commercial banks that is designed to encourage increased lending to LIPs. 29 issued in September, 1989. FIRA, FICART, BANRURAL and SHCP were jointly required to prepare a feasibility report for the follow-on operation, including historical institutional analysis complemented by field work by SHCP. Finally, the reform-minded Administration that assumed office in 1988 probably would have undertaken several improvements in the sector even without the Bank-supported Study, in some cases out of strictly budgetary concerns rather than considerations of economic efficiency. For example, the introduction of the tasa mezcla was driven by the need to increase commercial bank participation in agricultural credit, due to GOM's budgetary constraints. 3.40 The Study cannot be credited for the elimination of FICART, which was clearly duplicating FIRA and BANRURAL work and for which an analysis of audited financial statements (plus some readily available SHCP and BANXICO data) was sufficient to reveal the costs of operation. Nor can the Study be credited for the transaction cost subsidy scheme introduced in 1992 to encourage greater commercial bank lending to LIPs, since the topic was neglected in the Study's analysis.' However the Audit does believe that research related to the Study influenced GOM's decision to increase interest rates to LIPs beyond GIRA requirements. Together with SHCP's related research on BANRURAL, data from the Study inclined the Government to close down ANAGSA and rationalise BANRURAL's operations, although there were also other incentives at work.47 The Study also enabled project staff to conduct a more informed and influential policy dialogue with the Government, thereby fulfilling a key objective from the World Bank's point of view (see para. 2.20). 3.41 Related Developments in the Broader Financial Sector: The reforms in agricultural finance took place in the context of a broader reform of the financial sector: controls on deposit interest rates were eliminated as of 1988; the reserve requirement was abolished in 1989; other types of forced lending were phased out by 1991, including a 30 percent liquidity requirement; commercial banks were reprivatised during 1991-92; new banking charters were granted in 1993, including to affiliates of United States banks, and in 1994 the Central Bank of Mexico became fully autonomous. Furthermore, banking supervision was upgraded, a new loan classification and provision system was introduced, commercial bank capital requirements were brought in line with guidelines issued by the Bank for International Settlements (BIS), and several development banks and trust funds outside the agricultural sector were merged or liquidated. Several of these reforms were supported by the World Bank's FSAL. These developments have increased the efficiency of financial intermediation in Mexico, but are likely to entail reduced lending to agriculture in the short run, relative to other sectors, largely because of continuing controls on on-lending rates to agriculture.' The most important development that benefitted agricultural financial intermediaries was the control of the inflation rate, which fell from 159 percent in 1987 to 20 percent in 1989 and 12 percent in 1992. As a result, even controlled interest rates to agriculture have been positive in real terms since 1988, except on loans to LIPs in 1992 (see Annex 1.7). 46. The idea of subsidising transaction costs was proposed in OED's Performance Audit Report of FIRA 5 through FIRA 8A in June, 1990 (Report No. 8860, para. 132). 47. First, restructuring of Banrural and ANAGSA were key actions to be taken before disbursement of the second tranche of the Bank's Financial Sector Adjustment Loan (FSAL, Loan 3085-ME, approved in June, 1989); second, the Bank and the IDB made a combined US$700 million for a follow-on operation contingent on substantial reform of the two agencies; third, there were important budgetary considerations (see Annex 1.9 for data on the cost of GOM transfers to the two agencies). 48. Other factors include the on-going arrears situation, problems with titling of land and its use as collateral, as well as commercial banks' aversion to "pulverising" their loan portfolio by granting a large number of relatively small loans. Thus liberalising agricultural interest rates (or raising them to market rates) is not sufficient to guarantee substantially higher commercial bank lending to LIPs, but it is a necessary condition. Put another way, as long as interest rates remain controlled and below market rates, there can be no sustainable solution to the problem of ensuring access to commercial bank lending (of their own resources) for a majority of viable low income producers. The same statement holds for long term loans to OPs. 30 IV. FINDINGS AND ISSUES A. Overall Assessment of the Project 4.1 Project Rating: The project was largely implemented as planned at appraisal: investment and incremental short-term credit was provided to farmers; training, TA and civil works were provided to strengthen FIRA's productive support mechanisms, albeit less than envisaged in the SAR (FICART's productive support was financed with other sources of funds); interest subsidies were reduced in accordance with the GIRA; the Government eventually prepared an Agricultural Financial Sub-sector Study with World Bank support and instituted several reforms, although they were not necessarily based on the Study's recommendations; and the Government ensured that budget provisions were adequate to materially maintain the capital structure of FIRA and FICART. The project clearly achieved its (albeit limited) objectives regarding interest subsidies and maintenance of FIRA's and FICART's financial integrity. Available data suggest that investments in irrigated crops financed during the project period were financially viable, although not necessarily economically sound. The picture is less sanguine for rainfed crops. Overall, FIRA's M&E results suggest that FIRA clients enjoyed substantial increases in asset and equity holdings. While diversion of funds was significant among BANRURAL clients, FIRA/commercial bank clients generally undertook the planned sub-projects. Finally, compliance with covenants was acceptable. Overall, the project is rated as satisfactory. 4.2 Institutional Development: The Audit believes that in the absence of the project, policy dialogue with GOM would have been minimal during the late 1980s. Indeed, since the suspension of the follow-on operation, the World Bank has had little or no influence on developments in Mexican rural financial markets. The existence of the Ninth project, coupled with the promise of both World Bank and IDB funds under the proposed follow-on operation, provided the World Bank with an important voice at a time when the Government was willing to consider far reaching reform of financial markets in general and agricultural finance in particular. The World Bank's insistence on thorough analysis of the rural financial sub-sector, both in the context of the Study and of the feasibility report for the follow-on, yielded data and findings that catalysed key reforms, such as rationalisation of BANRURAL and elimination of ANAGSA. While there is clearly a problem of attribution of results and the Audit believes that certain reforms (e.g. the reformulated transactions cost subsidy scheme) were spurred by the promise of a follow-on rather than disbursement of the Ninth project, institutional development was substantial during the project period. 4.3 Sustainability: On the one hand, the Government has been reluctant to undertake several of the reform measures proposed by the World Bank and the IDB as a condition for a follow-on operation. On the other hand, GOM appears to be committed to the policies supported under the project and was prepared to consider some additional measures. Furthermore, there is wider recognition among farmers that the earlier paternalism of the State has given way to a competitive environment in which the key to viability is increased sales and productivity. At the institutional level, three of the nine demonstration centres assisted under the productive support component have been shut down. Also, although FIRA and BANRURAL are far from self-supporting, total GOM transfers to the two institutions have fallen sharply (almost 80 percent in real terms - see Annex 1.9) and the value of their equity has increased in real terms in spite of the reduced support. Overall, the project's benefits are considered sustainable. 31 B. Key Findings Regarding FIRA/FICART 9 4.4 Disbursement Objectives versus Development Objectives: The unstated project objective of providing the Mexican Government with rapid access to foreign exchange was achieved." Over US$370 million were disbursed within two years, as was the 1986 commercial bank package to which the project was linked. At the macroeconomic level, the project was part of an overall country strategy involving several World Bank projects that was extremely successful. However, the short term liquidity objective was not entirely consistent with the long term development objective. First, the major deficiencies in the appraisal report presented to the Board (para. 2.12) are explained by lending pressure. Second, hasty processing of the loan in May/June, 1987, precluded a thorough appraisal of the FIRA headquarters component. There were several difficulties with this component during implementation. Third, lending pressure led to postponement of a sectoral approach with meaningful conditionality in favour of the same project orientation of the previous nine operations: there was no linkage of project funds to policy improvements beyond the GIRA conditionality, which itself was weakened to expedite the project. In particular, there was no linkage between project funding and completion of the Study, although the Study was the principal vehicle for the project's policy objectives. Indeed, the World Bank could at first not even obtain a copy of the full report, because the wording in the Guarantee Agreement did not require GOM to submit it. Notwithstanding its adherence to the letter of the Guarantee Agreement, GOM was clearly not interested in a wide-ranging dialogue with the World Bank on the rural financial sector during the early stages of the project. It is doubtful that much progress would have been made on the Study beyond the first phase without the World Bank's insistence on it as a condition for a follow-on. Finally, since policy dialogue between the World Bank and the Government is most effective before project funds are fully disbursed, the lack of control on the rate of disbursement greatly reduced the supervision staff's time for effective discussions with GOM." 4.5 The Project's Farm LUvel Impact. FIRA/FICART 9 reached more beneficiaries than expected at appraisal and probably exceeded SAR targets with regard to job creation. The project's impact at the farm level cannot be quantified. However, FIRA's M&E unit has tracked asset and equity holdings for a sample of beneficiaries and has found significant increases in real terms during the project period. Unfortunately there was no control group with which to compare the findings, because farmers who were not loanees were unwilling to supply information. Second-best solutions may be to compare clients with both investment and working capital loans to those with only seasonal credits, or to accept some measurement error and subsidise a control group. The viability of investments supported by the project was highly dependent on access to irrigation, and those investments that were financially viable within the distorted economic environment of the mid-late 1980s may not all have been economically warranted. Since most of the trade and price distortions have been removed in the past 7 years, financially viable agricultural investments are generally more warranted from an economic point of view today than they were during the project period. While it is too late to alter the production impact of investments financed under FIRA/FICART 9, future increases in agricultural production and exports in these less restricted markets will "depend largely on continuing rural credit reforms to stimulate future agricultural investment", as noted in the PCR (para. 8). 49. This success did not affect the project rating, since the Audit rated the project against its stated objectives. 50. Ninety-three percent of the loan was disbursed by June, 1989, rather than December, 1990, as hoped for at appraisal. 32 4.6 Misallocation of Sub-Loans: There was significant diversion of sub-loans by BANRURAL clients during the project period, primarily because there was little pressure on clients to repay, (loan funds could thus be used for activities that yielded little or no cash return). Under the circumstances, the interest rate to small farmers hardly mattered. Diversion of funds was much rarer for FIRA/commercial bank clients, as both FIRA and PBs monitored the loans and the banks made serious efforts to enforce repayment. There are no data on substitution of project funds for other sources, but there are reasons to believe that it was not a major factor (para. 3.10 ff). The lesson to be drawn from the project is that there must be a credible commitment by all intermediaries to monitoring of sub-projects and to recovery of sub-loans. 4.7 The Problem of Wealthy "Low Income" Producers: The problem of OPs masquerading as LIPs was rarely uncovered by FIRA's monitoring staff - penalties for commercial banks are severe (three times the ACF) if it is detected. The incentive for such masquerading has declined as LIP rates have approached OP rates." The main difficulty lies in the definition of LIPs: GOM has treated all ejidatarios as LIPs, and both non-farm incomes and borrowers' relatives' incomes have often been ignored. The World Bank held to the increasingly restrictive definition of LIPs (namely farmers with an annual net family income not exceeding 1,000 times the minimum daily rural wage for the region) for three reasons: 80 percent of farmers still fell below the cutoff, World Bank staff hoped to negotiate new targeting criteria based on loan size under the follow-on operation, and the graduation of loanees to the OP category brought interest rates on their loans closer to market rates. 4.8 Crowding-out of Commercial Bank Lending: The tasa mezcla scheme introduced in 1989 effectively allowed commercial banks to charge market rates on short term loans to at least some OPs. Commercial bank lending to agriculture rose sharply that year, suggesting that FIRA had crowded out short term lending by PBs to wealthier farmers. Furthermore, since the tasa mezcla effectively insulated the FIRA rate to borrowers from the actual rate to borrowers, it permitted commercial banks to capture the subsidy intended for their clients. As this scheme applied only to OPs, there was less of an incentive to lend to LIPs; indeed, lending to LIPs fell sharply for this and other reasons. FIRA management is aware of the need for their clients to graduate from FIRA support to a direct relations with commercial banks. This should be initiated by freeing interest rates on all loans to OPs. The Government's willingness to consider the extension of the tasa mezcla scheme to long term credit for OPs suggests that this is not infeasible. FIRA still has an important role to play at the level of LIPs (and of less wealthy OPs). The provision of pr6stamos quirografarios to FIRA clients by commercial banks can be taken as an indication of FIRA's success in extending the formal financial frontier. It should lead to a scheduled reduction in FIRA rediscounts of short term sub-loans to established clients (whether they are LIPs or OPs), so that FIRA's intermediation activities focus on the dual challenge of introducing new, potentially viable clients to commercial sources of credit and of encouraging PBs to extend longer term loans to agriculture. 4.9 The Volume and Transparency of Subsidies for Rural Credit: The World Bank maintained that the project would be justified if interest subsidies fell by around 0.4 percent of GDP. In practice, they fell by at least 0.6 percent during the project period (para. 3.34). Progress was also made in 51. In the case of PRONASOL, loans are interest-free, so the incentive for rent sceking behaviour by non-targeted farmers remains significant. Indeed, one member of the Audit team met with an OP who was developing his farm with an interest-free loan of about US$15,000 through PRONASOL. Thus the elimination loans at below-market rates through 13ANRURAI 4commercial banks may simply lead to the provision of cheap credit via alternative channels (viz. the experience in Tunisia - OED Performance Audit Report for the Third and Fourth Agricultural Credit Projects, Loans 1885-TUN and 2865-IN, Report No. 11977). 33 increasing the transparency of the significant subsidies that FIRA and BANRURAL still enjoy. Transfers to the two agencies (particularly to FIRA) could be made more transparent by charging CETES plus a zero to three percent margin as appropriate on BANXICO loans and increasing explicit subsidies through the income statement to offset the higher cost of finance. In addition, accounting of BANRURAL's financial situation still needs to be improved, particularly with regard to provisions for doubtful loans. BANRURAL's financial statements have regularly overstated profits and therefore equity, as well as the value of the portfolio, by including inadequate provisions for doubtful loans. Greater transparency, coupled with the use of the SDI, would go a long way towards helping GOM to evaluate the subsidies it provides against the merits of FIRA's and BANRURAL's development activities. Figure 4: Area Insured for Selected Crops 7 6- 3- 0 c2- 01 1986 1987 1988 1989 1990 1991 1992 E Beans [ ] Maize Rice E Sorghum [ Wheat 4.10 Loan Guarantees and Agricultural Insurance: The replacement of ANAGSA with a more commercially oriented agricultural insurer was a significant institutional development." With the disappearance of ANAGSA, the insured crop area plunged dramatically (see Figure 4). Although a substantial part of the area insured by ANAGSA probably never existed, coverage should be increased, particularly for crops that are more vulnerable to shocks such as droughts or frosts. An important precondition for this is raising AGROASEMEX's premia towards actuarially fair rates and increasing competition in agricultural insurance markets. The coverage of FEGA loan guarantees will also need to be expanded as FIRA draws more LIPs into banking relations with commercial 52. Unlike ANAGSA, Agroasemex is chartered under a general agricultural insurance law, rather than a specific law for it alone. 34 lenders and FEGA's guarantee premia should be increased to reflect FEGA's true exposure to risk. Current premia are inadequate to cover payouts and the cost to Government has grown rapidly in recent years. C. Subsequent Developments and Outstanding Issues 4.11 The Preparation of the Follow-on Operation: The proposed Rural Financial Sub-sector Project would have made the break from the project-oriented approach of the earlier ten operations to the sector-oriented approach recommended by OED's Audit of the Fourth project." The change in approach was appropriate, as was coordination between the World Bank and the IDB, since lack of collaboration between the two agencies had hampered efforts to institute reforms under earlier projects.' Given the substantial volume of lending under other World Bank projects during 1989-92, there was also little pressure to lend for the proposed follow-on. The proposed World Bank loan of US$300 million and IDB loan of US$400 million gave the donors leverage in discussions on policy reforms that far exceeded the donors' importance in FIRA's and BANRURAL's lending programme. The promise of World Bank/IDB funding and their stance on reforms also helped reform-minded parties in GOM to persuade more conservative ones to accept further change in the rural financial sector. To conclude, the promise of funding was essential for constructive dialogue, and significant policy measures proposed under the follow-on operation were adopted even though the project was never approved (e.g. the transaction cost subsidy scheme and the closure of FICART). Notwithstanding this success, the above can clearly not be regarded as a sustainable or replicable approach to achieving policy reform. 4.12 The Suspension of the Proposed Follow-On Operation: In October 1992, GOM and the World Bank established final positions on policy matters that were considered irreconcilable. The key issue that was not negotiable for the Government was liberalisation of interest rates to LIPs." Other key points of disagreement were the weighted average rate to commercial banks on FIRA rediscounts, and how to handle non-commercial farmers who were too good for PRONASOL (GOM maintained that targeted interest subsidies were an appropriate instrument for reaching them). Perhaps a greater loss than the funding (which hindsight suggests was not strictly "needed") was the reduction in the intensity of the World Bank-Government dialogue, which had been very constructive during 1989-91. Since early 1993, an exchange of ideas has been maintained in the context of a joint 53. Mexico - Fourth Livestock and Agricultural Development Project, Loan 910-ME, OED Project Performance Audit Report No. 2577, dated July 3, 1979. 54. See OED's Performance Audit Report of FIRA 5 through FIRA 8A, Report no. 8860, dated June 29, 1990- 55. The Audit agrees with regional staff that a firm stance on policy matters was "warranted after ten operations and recurrent requests of our Board in previous projects to avoid subsidies" (sic., internal correspondence dated November 30, 1994). Nonetheless, while GOM and Bank staff ascribe a particularly doctrinaire stance to IDB negotiators, draft Bank documents and correspondence tiles on the proposed project contain categorical statements that suggest a rather rigid interpretation of the rural financial market policy embodied in the Levy Report and in OD8.30. For example: "There is no particular reason for the government to be lending for agriculture.." (draft Report on the Rural Financial Sector Project, dated 28th December, 1990), or "Government development banks and trust funds dominate the Mexican rural financial market, having crowded out private intermediaries" (Final Eecutive Project Summary, dated 28th December, 1990) [emphasis added]. 'There are good reasons for government lending to agriculture, and while this Audit argues that there was clearly some crowding out, it is doubtful that Mexico's (publicly-owned) commercial banks would have offered substantially more long term credit or would have had a much larger short term lending programme tor LlPs - the FIPS simply overstates a good case for reform. Note: The Levy Report, which was presented to the Board in August, 1989, is formally known as the Report of the Task Force on Financial Sector Operations. Operational Directive 8.30 on Financial Sector Operations was released in February, 1992. 35 study of informal rural financial markets. However, with the election of a new Administration in August, 1994, and with the World Bank reexamining its policy on rural finance as it recasts OD8.30, it is expected that a broader and more intensive dialogue on rural financial sector reform can be resumed in the new sexenio. 4.13 The Future of FIRA as an Institution: With the establishment of an autonomous central bank, the future of FIRA has become a subject of debate. Within the next 18 months, BANXICO must relinquish trusteeship for its numerous trust funds to other trustees. It is as yet undecided whether FIRA should fall under NAFIN (the Borrower for this project), BANRURAL or another agency, or whether FIRA should become a separate development bank. At this point, the Audit favours establishing FIRA either as a separate, second-tier development bank or alternatively as a trust fund under NAFIN.' In either case, the quality of FIRA's able technical staff must be maintained, as must their focus on activities to encourage commercial bank lending to viable agricultural producers. The technical assistance support continues to be valuable and warranted for the immediate future, both for the commercial banks' agricultural staff and for final borrowers. Nevertheless, the merits of FIRA's TA should be assessed continuously in the light of its cost implications and of the quality and availability of alternative, private sources of technical support. 4.14 FIRA's Future Mission: The Audit believes that FIRA must assume greater risks in order to fulfill its mission of expanding the formal financial frontier. It should discontinue its short term lending operations for wealthier farmers within the next two or three years and phase out long term lending for them more gradually. This could be done by setting a cap on the Peso value of FIRA rediscounts for any given loan and loanee, in addition to the 80 percent limit on FIRA rediscounts. The Peso cap would be reduced over time as commercial banks begin to extend the terms of their loans in response to liberalised interest rates (see below). Note that the Peso cap would be most binding on commercial bank loans to large borrowers. Loans to LIPs and relatively poor OPs might still qualify for FIRA rediscounts of up to 80 percent of the loan. The focus of FIRA's activities should be on drawing viable and potentially viable LIPs (as they are currently defined) into relations with the formal banking sector. In addition, FIRA should continue to encourage commercial bank lending to poorer OPs. Commercial banks should not be allowed to obtain rediscounts from FIRA for given clients for more than e.g. five years. Thereafter, the banks' continuing concerns regarding the riskiness of LIP clients should be addressed via FEGA guarantees at actuarially fair rates. An expanded programme of guarantees may thus be required. To conclude, FIRA has an important mission to reach out to new clients and address information constraints which commercial banks may be reluctant to confront. There can be little justification for continued subsidisation of an established clientele, or for crowding out lending by the private sector. 4.15 BANRURAL's Future Mission: This is articulated well in draft documents for the proposed Rural Financial Sector Project. BANRURAL should use its extensive network of branches to mobilise deposits and to lend on the basis of stringent criteria, with an adequate margin to cover default risks and most administrative costs. To this end, BANRURAL and other participating banks should be allowed to charge unrestricted interest rates to sub-borrowers (see paragraph 4.18). Any remaining administrative costs could be covered by transparent, time-bound transaction cost subsidies from GOM. These subsidies should not be peculiar to BANRURAL, but rather should be extended to any participating bank for the first few (e.g. two or three) loans that it offers to new, potentially 56. BANRURAL needs to be strengthened considerably before it can be considered as a potential trustee. 36 viable clients." The Audit believes that producers who are not regarded as potentially viable should continue to be directed to PRONASOL sources of funds. On the other hand, BANRURAL should expect to lose some of its better clients through competition from FIRA-rediscounted commercial bank loans. Competition between commercial banks and BANRURAL for wealthier clients should not be ruled out. BANRURAL should also seek to diversify its lending, gradually increasing the proportion of non-agricultural, rural sector loans in its portfolio as BANRURAL staff gain experience with the evaluation of new types of credit. Finally, banking supervisors should hold BANRURAL to the same banking standards as other participating banks. 4.16 The Problem of Commercial Bank Arrears: Commercial bank lending to all sectors increased five-fold in real terms during 1987-1993. It increased more than six-fold in real terms to agriculture between 1987 and 1992. The rapid increase in indebtedness and numerous trade and macroeconomic factors (including high real rates of interest in 1988-89) led to a sharp increase in arrears on commercial bank loans to all sectors. The problem was particularly pronounced in agriculture (para. 3.31), and more severe for PB-funded credit than for FIRA-rediscounted sub-loans. In March, 1994, the Mexican Banking Association and GOM negotiated a programme of restructuring of some 18,000 agricultural loans totalling Mex$5,500 million. Payments are to be restructured over 15 years with interest payable at inflation plus 4 to 4.5 percentage points, though there is a five-year grace period for the 86 percent of clients whose loans amount to less than Mex$200,000." The programme needs to be monitored closely to establish the implications for returns to FIRA and commercial banks, and for borrowers' incentives to repay future credit. More generally, Bank supervision of any future financial intermediary loans in Mexico should include an annual review of the financial performance of all participating banks, with a particular focus on the profitability and performance of sub-loans that are supported by the World Bank. 4.17 The Vision for the Sector and the World Bank's Role: The opening of new trade opportunities through the North American Free Trade Agreement offers Mexican agricultural producers a great opportunity, as well as a great challenge, to increase their efficiency and serve a vastly expanded market. To this end, rural financial reforms must keep pace with advances in trade and pricing policies, as well as in the broader financial markets. As a result of bold measures taken by GOM, considerable progress was achieved during the Ninth Agricultural Credit Project and during the preparation of the proposed follow-on operation. However, much remains to be done. For example, there is little justification for subsidising short term commercial bank loans to wealthier OPs. Therefore the tasa mezcla should be discarded in favour of liberalised interest rates and a higher FIRA rate to commercial banks. Transaction cost subsidies would still be provided for sub-loans below a given size. The same approach should be pursued on long term loans to OPs (for which GOM has already consider applying the tasa mezcla). 57. The amount of these transaction cost subsidies should be minimised by funding only the diflerence between t.e 0n-lending rate required for an acceptable return to PBs on agricultural loans and the maximum feasible on-lending rate to clients, wAhere the latter is determined by calculating the expected return on sub-borrowers' equity for farm investments ani assessmg the maximum interest expense that could be supported while guaranteeing an acceptable return on equily lor the sub-!or!ower. The aii should be to cover as much of the administrative costs of lending to 'Bis as is possible via interest income raler than v;,' Governmei grants. 58. A programme of restructuring of Mex$18,000 million in overdues on industrial loans was also being negotiated during the Audit mission. 37 4.18 There is no shortage of demand for credit even at higher interest rates. The Audit mission found that LIPs were willing and able to borrow from moneylenders to ensure timely soil preparation and planting of seasonal crops. The main problem is access to formal credit for LIPs and poorer OPs, since they generally require small loans and commercial banks tend to avoid small loans because of high transaction costs per Peso lent. Thus a more effective way of providing immediate assistance to LIPs is to subsidise commercial banks' transaction costs on small loans, instead of providing interest subsidies to farmers. Interest subsidies to LIPs should be phased out according to a preestablished schedule over the short to medium term (four years or less)." At the end of this period, interest rates to all agricultural borrowers should be fully liberalised. Any further transaction cost subsidies should be limited to the first few loans to new clients (see para. 4.15). The Audit team believes that subsidies for subsistence or marginal farmers are better targeted via investments in physical and social infrastructure, particularly improved roads, rural electrification, education and health services. The Government's commitment to PRONASOL is evidence of GOM's awareness of the merits of such an approach.' Other measures to be undertaken include simplifying the rate structure on FIRA loans and replacing the income-based criterion for targeting with alternative criteria based, e.g. on loan size. 4.19 There has been a hiatus both in GOM's reform process and in World Bank involvement in Mexico's formal rural financial sector during the past two years. Full disengagement was correctly regarded as an ineffective way for the World Bank to ensure that Mexican rural financial intermediaries attain the high standards established in OD8.30. Therefore, responding to explicit GOM interest, the World Bank attempted to remain engaged in dialogue on rural financial markets by directing its attention to the heretofore neglected informal financial sub-sector. It is expected that, now that the new Administration has assumed office, the World Bank and GOM will begin a broader and more intensive dialogue on policy reforms for the entire rural financial sector. The scale of any future World Bank support for the sector should be tailored to the Government's willingness to consider further necessary reforms. 59. One way to do this would be to extend the asa mezcla to all agricultural loans over the transition period. If necessary, a capped tasa mezcla could be considered as an interim measure, with e.g. twice the spread allowed on any commercial bank contributions exceeding the minimum requirement on loans to LIPs. The cap would be raised and removed before full liberalisation of interest rates. 60. On the other hand, during implementation of the two AGSAIs, the Bank repeatedly expressed concern regarding the adequacy of public investment in agricultural infrastructure (see OED's forthcoming PAR on the AGSAL).  1986 1987 1988 1989 1990 1991 1992 [1] The Macroeconomy Nominal GDP (Mex$ millions) 79,191 193,312 390,451 507,618 686,406 865,166 1,033,224 GDP Deflator (1991 = 100) 10.6 25.3 50.5 63.5 82.2 100.0 116.2 Real GDP (1991 Mex$ millions) 750,032 763,969 773,471 799,285 834,918 865,166 888,921 Population (Millions) 79.6 81.2 82.8 84.5 86.2 87.8 89.5 Real GDP per caput (1991 Mex$ '000s) 9,423 9,408 9,341 9,459 9,686 9,854 9,932 . Growth in real GDP per caput (%) -0.1% -0.7% 1.3% 2.4% 1.7% 0.8% Unemployment Rate (narrow definition, 2) 3.9% 3.6% 3.0% 2.8% 2.6% 2.9% Consumer Price Index (year end, 1991 = 100) 13.8 35.7 54.1 64.8 84.2 100.0 111.9 Inflation Rate (Z) 86.2% 159.2% 51.7% 19.7% 29.9% 18.8% 11.92 1986 1987 1988 1989 1990 1991 1992 0 [21 Agriculture Nominal Agricultural GDP (Mex$ millions) 7,466 16,825 30,690 39,246 54,810 66,682 87,295 * Real Agricultural GDP (1991 Mex$ millions) 65,406 66,310 63,806 62,368 66,052 66,682 66,601 4 Agriculture/Total GDP (%) 9.4% 8.7% 7.9% 7.7% 8.0% 7.7% 8.4% Growth in Real Ag. GDP (%) 1.4% -3.8% -2.3% 5.9% 1.0% -0.1% $. (Lowest) Nominal Regional Min. Daily Wage (Mex$) 2,060 5,395 6,670 8,405 9,920 11,115 11,115 0 Real Regional Minimum Daily Wage (1991 Mex$) 14,959 15,116 12,323 12,973 11,784 11,115 9,930 1986 1987 1988 1989 1990 1991 1992 (3) Trade Exchange Rate (average annual, Mex$/US$) 0.61 1.38 2.27 2.46 2.81 3.02 3.09 6 Debt Service Ratio (%) 55.3% 40.12 47.3% 36.9% 26.7% Agricultural Exports (Hex$ millions) 647 1,231 2,354 2,521 4,690 5,571 Agricultural Imports (Mex$ millions) 554 1,291 3,780 4,601 5,480 5,808 Surplus or Deficit (Mex$ mitlions) 93 (60) (14426) (2,080) (790) (237) 1986 1987 1988 1989 1990 1991 1992 [4) Finance Deposit Rate (%) 90.2% 115.7% 31.0% 32.5% 22.6% 14.7% 17.9% ACF (average annual, 2) 80.9% 94.6% 67.6% 44.6% 37.1% 22.6% 18.8% CETES (%) 99.5% 122.5% 52.3% 45.0% 34.8% 19.3% 15.6% f Reserve Requirement (%) 10.0% 10.0% 10.0% 10.0% 2 Harket Reference Rate (2, *) 93.2% 108.4% 78.4% 52.9% 40.1% 25.6% 21.81 *: Defined as fb+3Z]/[1-r], where b u the base rate (ACF for 1986/89, CETES thereafter); r the reserve requirement. 1z (Continued..) 1986 1987 1988 1989 1990 1991 1992 (5) The Banking System's Non-Agricultural Portfolio Banking System's Portfolio (Excl loans to GOM, Mex$m) 24,819 58,837 80,496 122,486 188,563 274,139 399,958 Real Growth in Portfolio (2) -8.5% -9.8% 27.11 18.5% 22.4% 30.3% Loan Portfolio in Arrears (Excl loans to GOM, Mex$m) 757 1,185 1,627 6,233 8,171 9,578 17,231 System's Arrears/Loan Portfolio (%) 3.1% 2.0% 2.0% 5.1% 4.3% 3.51 4.3% Coamercial Bank Arrears/Loan Portfolio (%) 0.9% 1.2% 2.0%1 3.2% 5.5% . Banking System's Industrial Portfolio (Mex$ millions) 10,552 24,234 29,388 40,702 56,620 79,593 112,851 Industrial Portfolio / Total Non-Govt. Portfolio (%) 42.5% 41.2% 36.51 33.2% 30.0% 29.0% 28.2% Real Growth in Industrial Portfolio (%) -11.4% -20.0% 15.7% 7.1% 18.3% 26.7% Industrial Portfolio in Arrears (Mex$ millions) 338 476 653 1,714 2,226 2,352 4,301 Industrial Arrears/Industrial Portfolio (t) 3.2% 2.0% 2.2% 4.2% 3.9% 3.0% 3.8% Banking System's Housing Portfolio (Mex$ millions) 1,241 2,834 6,128 8,523 12,453 15,484 18,569 Housing Portfolio / Total Non-Govt. Portfolio (%) 5.0% 4.8% 7.6% 7.0% 6.6% 5.6% 4.6% Real Growth in Housing Portfolio (%) -11.9% 42.6% 16.2% 12.5% 4.7% 7.1% Housing Portfolio in Arrears (Mex$ millions) 18 8 19 24 23 191 389 Cr Housing Arrears/Housing Portfolio (%) 1.5% 0.3% 0.3% 0.3% 0.2% 1.2% 2.1% Banking System's Loans for Services (Mex$ millions) 6,923 17,430 21,753 33,291 62,360 99,585 154,761 Services Portfolio / Total Non-Govt. Portfolio (Z) 27.9% 29.6% 27.0% 27.2% 33.1% 36.3% 38.7% i. Real Growth in Services Portfolio (%) -2.9% -17.7% 27.9% 44.2% 34.4% 38.8% Services Portfolio in Arrears (Mex$ millions) 96 183 293 574 1,232 1,713 5,153 S Services Arrears/Services Portfolio (%) 1.4% 1.0% 1.3% 1.7% 2.0% 1.7% 3.3% Banking System's Comercial Portfolio (Mex$ millions) 2,378 5,327 8,634 20,332 31,911 49,431 75,584 0 Commercial Portfolio / Total Non-Govt. Portfolio (%) 9.6% 9.1% 10.7% 16.6% 16.9% 18.0% 18.9% Real Growth in Commercial Portfolio (2) -13.6% 6.9% 96.7% 20.8% 30.4% 36.6% Commercial Portfolio in Arrears (Mex$ millions) 113 180 183 523 1,320 2,851 3,729 'e Commercial Arrears/Commercial Portfolio (%) 4.8% 3.4% 2.1% 2.6% 4.1% 5.8% 4.9% 1986 1987 1988 1989 1990 1991 1992 a 16) The Banking System's Agricultural Portfolio I Banking System's Agricultural Portfolio (Mex$m, *) 1,818 3,691 8,862 15,381 21,956 26,844 33,393 Agric. Portfolio / Total Non-Govt. Portfolio (2) 7.3% 6.3% 11.0% 12.6% 11.6% 9.8% 8.3% Real Growth in Agricultural Portfolio (%) -21.7% 58.3% 45.0% 9.9% 2.9% 11.1% Loans to LIPs / Agricultural Loans (%) 48.9% 49.7% 40.9% 26.9% 19.1% Agricultural Portfolio in Arrears (Mex$m, *) 149 227 395 1398 2786 2363 3357 System's Agric. Arrears/Agricultural Portfolio (%) 8.2% 6.2% 4.5% 9.1% 12.7% 8.8% 10.1% Commercial Bank Agric. Arrears/Agric. Portfolio (%) 2.1% 2.6% 3.6% 5.5% 7.0% Commercial Banks' Agric. Portfolio (Mex$ millions) 377 824 2,416 6,901 12,199 14,118 16,192 Commercial Banks' Share of Agricultural Loans (%) 20.7% 22.3% 27.3% 44.9% 55.6% 52.6% 48.5% FIRA's Portfolio (Hex$ millions) 597 1,044 2,411 3,871 6,535 9,789 13,184 FIRA's Share of Agricultural Loans (Z) 32.9% 28.3% 27.2% 25.2% 29.8% 36.5% 39.5% Z BANRURAL's Portfolio (Mex$ millions) 844 1,823 4,035 4,609 3,223 2,937 4,017 ' BANRURAL's Share of Agricultural Loans (%) 46.4% 49.4% 45.5% 30.0% 14.7% 10.9% 12.0% *: Excludes mining, fisheries & forestry (Annex Continued..) 1986 1987 19887 1989 F 1990 19 1991 196 [7] Agricultural Interest Rates Nominal Rates to LIPs - Short Term (%) 58.2% 74.8% 65.6% 43.3% 33.8% 18.7% 15.1% Nominal Rates to LIPs - Long Term (%) 56.6% 66.3% 64.3% 42.4% 33.1% 18.3% 14.8% Nominal Rates to "Basic OPs" - Short Term (%) 82.9% 99.4% 70.6% 47.6% 37.82i 22.3%i 18.6%j Nominal Rates to "Basic OPs" - Long Term (%) 80.9% 90.9% 68.6% 45.6% 35.8% 21.3% 17.621 Nominal Rates to Other OPs - Short Term (%) 82.9% 101.2% 74.6% 51.6% 4l.82 26.3t 16.7%i Nominal Rates to Other OPs - Long Term (2) 80.9% 92.7% 72.6% 49.6% 39.8% 24.3% 16.4% M Real Rates to LIPs - Short Term (%) -15.0% -32.6% 9.2% 19.7% 3.0% -0.1% 2.9% Real Rates to LIPs - Long Term (%) -15.9% -35.8% 8.3% 19.0% 2.4% -0.4% 2.6% Real Rates to "Basic OPs" - Short Term (%) -1.8% -23.1% 12.5% 23.3% 6.1% 3.0% 6.0% Real Rates to "Basic OPs" - Long Term (%) -2.8% -26.3% 11.2% 21.6% 4.5% 2.1%; 5.1% Real Rates to Other OPs - Short Term (%) -1.8% -22.4% 15.1% 26.7% 9.1% 6.3%' 4.2% f Real Rates to Other OPs - Long Term (%) -2.8% -25.6% 13.8% 25.0% 3.8% 4.6% 4.6% ' 1986 1987 1988 1989 1990 1991 1992 [83 Agricultural Insurance -M Beans - Area Covered by Insurance ('000s Hectares) 949 1,023 1,080 755 239 29 32 Maize - Area Covered by Insurance ('000s Hectares) 2,911 3,179 2,998 1,932 375 108 141 Rice - Area Covered by Insurance ('0008 Hectares) 169 155 144 181 20 9 5 Sorghum - Area Covered by Insurance ('000s Hectares) 1,224 1,193 909 756 254 57 87 Wheat - Area covered by Insurance ('000s Hectares) 732 581 499 507 186 42 95 0 Total Area Covered by ANAGSA Insurance ('000s Ha) 7,100 7,300 6,600 4,900 Total Area Indemnified ('000s Hectares) 5,400 5,500 5,200 3,100 Area Indemnified / Area Covered M) 76.1% 75.3% 78.8% 63.3% GOM Transfers to ANAGSA/AGROASEMEX (Mex$ millions) 215 476 1,149 991 1,848 2 65 (8]Sbsdst Agricultural suranc FIRA Incmea From Lova Suranes (Mex$ mecaos s) 0 843 9,96 23 2,6 264 a uaran Coe eid to Bnks (Mex$) 28 6,4 8, 51,918 11912 R- Ne OutfovEredy on ansuranes( e rs) 153) 9 ( (1) [9a Subsidies oered by Inurace (000 Hectaes) Lend FoA (excl. FEGA): bya A Sbsdins r eTA (millions) 7,0100 1,22 1,313 1,0 Toa Subsid nemnceidn(a00 Liailtres) 5,4008 64 1,5 118,0 GO Trasfers oEut (x millions)AEME 753x$06illions))215 FIR Suarantes Aplid to Bancom (Mex$ millis) 000 18 5 1986 FIRA (excl. FEGA): So excluding TA (e) 88.6% 82.8% 78.4% 93.3% 109.5% FEGA Subsidy on Equity + Grant Income (Mex$ millions) 2 4 22 30 197 (panel [9] Continued..) 1986 1987 1988 1989 1990 1991 1992 (9] Subsidies to Agricultural Lenders (..Continued) GOM: Total Fiscal Transfers to FIRA (1991 Mex$m) 1,637 998 1,415 595 319 330 GOM: Total Fiscal Transfers to FICART (1991 Mex$m) 386 282 451 149 39 36 GOM: Total Fiscal Transfers to BANRURAL (1991 Mex$m) 4,566 3,712 5,799 4,096 9,916 949 1986 1987 1988 1989 1990 1991 1992 (10) BANRURAL's Key Financial Ratios (*) Ratio of Equity to Loan Portfolio (%) 15.8% 15.1% 53.9% 72.4% 51.7% 49.8% 58.9% Loans Classified as Overdue / Loan Portfolio (%) 3.0% 2.4% 3.7% 8.9% 7.0% 16.5% 16 2% Ratio of Deposits to Total Liabilities (%) 3.6% 3.7% 4.6% 9.8% 9.0% 9.3% 25:8% Ratio of Total Liabilities to Equity (Units) 17.8 20.3 3.7 2.6 1.6 1.7 1.4 (Interest Income-Interest Expense)/Avg. Portfolio (%) 8.2% 14.0% 19.2% 19.8% 16.0% 10.8% 8.4% Admin. Costs Excl. Provisions/Avg. Loan Portfolio (%) 13.2% 15.3% 15.1% 13.9% 15.3% 18.2% 19.4% W Total Financial & Admin. Expenses/Avg. Portfolio (%) 55.4% 70.6% 63.8% 60.3% 55.3% 40.5% 38.3% Return on Equity, before GOM Income Transfers (%) -81.1% -44.2% -10.9% 17.4% -29.4% -37.6% -40.5% Return on Equity, after GOM Income Transfers (%) -1.2% -1.5% -0.8% -0.3% -27.0% -30.2% -30.3% f CData should be used with caution since earlier financial statements understate the volume of non-performing loans. 1986 1987 1988 1989 1990 1991 1992 111) FIRA Interest Subsidy Passed to Commercial Banks Intended for Farmers '- Minimum On-lending Rate for Viability of FIRA (, )84.4% 58.6% 42.0% 25.3% 20.9% 0 Actual Average FIRA On-lending Rate (%) 43.5% 30.8% 23.9% 14.2% 11.0% interest Subsidy / FlRA's Portfolio (6) 40.9% 27.8% 18.1% 11.1% 10.0% Real Value of Interest Subsidy (1991 14ex$ millions) 1,554 1,574 1,283 1,002 1,10 Interest Subsidy Agricultural GDP (-) 2.7% 2.6% 2.0% 1.5% 1.4% Interest Subsidy /DP (%) 0.22% 0.20% 0.16% 0.12% 0.12% *: Actual average on-lending rate multiplied by [i+SDI]. 1986 1987 1988 1989 1990 1991 1992 (12] BANRURAL's Interest Subsidy to Farmers Estimated Minimum Rate for Viability of BANRURAL (% 136.5% 66.3% 59.0% 43.3% 25.3% Actual (Reported) Interest / Loan Portfolio () 47.3 43.2% 36.1% 32.0%1 18.12% Interest Subsidy / BANRURAL's Portfolio (%) 89.2% 23.1% 22.9% 16.8% 15.2% 1 Real Value of Interest Subsidy (1991 Mex$ millions) 4,732 1,596 1,684 1,072 832 1 Interest Subsidy / Agricultural GDP (%) 7.1% 2.5% 2.7% 1.6% 1.2%1 Interest Subsidy / GDP (%) 0.62 0.21%; 0.21% 0.13%1 0.10% 0 EsEstimated at CETES + 14% (Annex Continued.) 1986 1987 1988 1989 1990 1991 1992 [131 Returns to Farmers on Selected Crops Jalisco, Irrigated Maize: Cost (Mex$/hectare) 152 323 838 1,042 1,257 Jalisco, Irrigated Maize: Yield (tons/hectare) 4.9 3.0 3.5 3.2 4.0 Jalisco, Irrigated Maize: Price (Mex$/ton) 131 242 373 432 636 Jalisco, Irrigated Maize: Revenue (Mex$/hectare) 635 735 1,308 1,365 2,538 Jalisco, Irrigated Maize: Gross Return (Mex$/hectare) 483 412 470 323 1,281 Sinaloa, Irrigated Rice: Cost (Mex$/hectare) 193 354 757 1,112 1,434 Sinaloa, Irrigated Rice: Yield (tons/hectare) 3.8 4.3 4.1 4.3 4.2 Sinaloa, Irrigated Rice: Price (Mex$/ton) 98 238 450 476 420 Sinaloa, Irrigated Rice: Revenue (Mex$/hectare) 376 1,023 1,851 2,035 1,754 Sinaloa, Irrigated Rice: Gross Return (Mex$/hectare) 183 669 1,094 923 320 Sonora, Irrigated Wheat: Cost (Mex$/hectare) 146 284 676 894 1,141 Sonora, Irrigated Wheat: Yield (tons/hectare) 4.4 5.3 5.2 4.7 5.3 Sonora, Irrigated Wheat: Price (Mex$/ton) 58 120 310 380 484 Sonora, Irrigated Wheat: Revenue (Mex$/hectare) 256 637 1,601 1,801 2,556 Sonora, Irrigated Wheat: Gross Return (Mex$/hectare) 110 353 925 907 1,415 Tamaulipas, Rainfed Sorghum: Cost (Mex$/hectare) 92 144 270 370 468 Tamaulipas, Rainfed Sorghum: Yield (tons/hectare) 1.6 1.5 1.8 1.2 2.3 Tamaulipas, Rainfed Sorghum: Price (Mex$/ton) 58 117 225 319 320 Tamaulipas, Rainfed Sorghum: Revenue (Mex$/hectare) 92 174 409 370 736 Tamaulipas, Rainfed Sorghum: Gross Return (Mex$/ha) 0 30 139 0 268 Zacatecas, Rainfed Beans: Cost (Mex$/hectare) 85 203 380 490 690 Zacatecas, Rainfed Beans: Yield (tons/hectare) 0.4 0.5 0.3 0.1 0.6 Zacatecas, Rainfed Beans: Price (Mex$/ton) 224 507 793 1,180 1,742 Zacatecas, Rainfed Beans: Revenue (Mex$/hectare) 82 234 266 165 965 Zacatecas, Rainfed Beans: Gross Return (Mex$/hectare) (3) 31 (114) (325) 275 CL [14] Sources of Data LA2AG: Project Completion Report, Ninth Agricultural Credit Project (Ln.2837-ME) PCR LA2AG: Draft Agricultural Sector Memorandum, 1994 Correspondence Files for Loan 2837-ME Financial Statements for FIRA, FICART & BANRURAL IMF: International Financial Statistics INEGI: Anuario Estadistico de los Estados Unidos Mexicanos, Edicion 1992 Banco de Mexico: Indicadores Economicos, various issues SARH: Anuario Estadistico de la Produccion Agricola de los Estados Unidos Mexicanos, various issues Comision Nacional Bancaria: Banca Multiple, Diciembre 1982 - Diciembre 1992 Comision Nacional de Salarios Minimos FIRA Databases (incl. SUECO) and direct communications    IBRD 26569 UNoTED STATES OF AER" U N I T E D S T A T E S o--ç MEXICOC A EXCO 3W S -- o • •• YUOATAN RKo .go.1o. si.N T H c& oSn H •,2.A.•• K'OMETERS - HONDURAs 04 WÉSTG ....... Ø 1 20FIRA: Y ý TN -5-, A Regiaoal Offices ~* " State OFfices CO *æ. VEILACRU - Rogianal Boundariøs ICHOA I, State Bo.ndafies jaZ Inlte,tiorial BouIdøos i åLZ O O 200 300 400 500UTEAL KILOMETERS TA.,,/HONDURAS DECEMINER 1994