Mexico - Agricultural Projects

Document of The World Bank FOR OFFICIAL USE ONLY Report No. 8860 PROJECT PERFORMANCE AUDIT REPORT MEXICO FIFTH AGRICULTURAL AND LIVESTOCK CREDIT LOAN (1217-ME) SIXTH AGRICULTURAL CREDIT LOAN (1569-ME) SEVENTH AGRICULTURAL CREDIT LOAN (1891-ME) EIGHTH AGRICULTURAL CREDIT LOAN (2454-ME) AGRICULTURAL CREDIT LOAN (2610-ME) JUNE 29, 1990 Operations Evaluation Department Is document has a restricted distribution and may be used by recipients only In the peformance of thei oficial duties. Its contents may not otherwise be disclosed without World Bankau lim ABBREVIATIONS ACF Average Cost of Funds AID Agency for International Development ANAGSA National Agency for Agricultural Guarantees & Insurance BANRURAL/BNCR Banco Nacional de Credito Rural BTO Back-to-office Report CIANO Agricultural Research Center for the Northwest CPI Consumer Price Index DFC Development Finance Company FEFA Special Fund for Agro-Livestock Financings FEGA Guarantee and Technical Assistance Fund FICART Trust Fund for Credit in Irrigated and Rainfed Areas FIRA Trust Funds Instituted in Relation to Agriculture FONDO Guarantee and Development Fund for Agriculture, Livestock and Poultry FONEI Industrial Equipment Fund GIRA General Interest Rates Agreement IADB Inter-American Development Bank IBRD International Bank for Reconstruction and Development INIA National Agricultural Research Institute INIFAP National Agricultural, Livestock & Forestry Research Institute IPB Initial Project Brief LIBOR London Inter-bank Onlending Rate MIP Middle-Income Producers M&E Monitoring and Evaluation NAFIN/NAFINSA Nacional Financiera SA OED Operations Evaluation Department (IBRD) OP Other Producers PPA Project Performance Audit PPAR Project Performance Audit Report PCR Project Completion Report PBI/PIB/LIP Low Income Producer PIDER Integrated Rural Economic Development Program PLANAT National Rainfed Agriculture Plan PPU Planting & Programming Unit SAP Special Adjustment Program/Special Action Program SPN Supervision FOR OFFICIAL JS ONLY THE WORLD BANK Washngton. D.C. 20433 US.A. Ofte af Oksedwn-CGweal Opeseous twhosMenn June 29, 1990 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECTs Project Performance Audit Report on Mexico Fifth Agricultural & Livestock Credit Loan (1217-ME) Sixth Agricultural Credit Loan (1569-ME) Seventh Agricultural Credit Loan (1891-ME) - Eighth Agricultural Credit Loan (2454-ME) Agricultural Credit Loan (2610-ME) Attached, for information, is a copy of a report entitled *Project Performance Audit Report on Mexico Fifth Agricultural & Livestock Credit Loan (1217-ME), Sixth Agricultural Credit Loan (1569-ME), Seventh Agricultural Credit Loan (1891-ME), Eighth Agricultural Credit Loan (2454- ME), and Agricultural Credit Loan (2610-ME),* prepared by the Operations Evaluation Department. Attachment Thi doat ha a rWWid Mtibuto a be used by rcipials on* in the perfaan of thei oiiia dties 1n COMteIS aw no "rw be diclse witou Worl BAk autstin FOR OFFICIAL USE ONLY PROJECT PERFORMANCE AUDIT REPORT MEXICO FIFTH AGRICULTURAL & LIVESTOCK CREDIT LOAN (1217-MB) SIXTH AGRICULTURAL CREDIT LOAN (1569-NE) SEVENTH AGRICULTURAL CREDIT LOAN (1891-MB) EIGHTH AGRICULTURAL CREDIT LOAN (2454-MB) AGRICULTURAL CREDIT LOAN (2610-ME) TABLE OF CONTENTS Page No. Preface ..................... ....... * ........... ... . ...... ... Basic Data Sheets .............................................. 11 Evaluation Sunmary .....*............................. xi PROJECT PERFORMANCE AUDIT I. PROJECT BACKGROUND ........... 1 II. PROJECT DESIGN ..................... to ..... * ...... .... 6 A. PIRA 5 ..................***... . ...*..... ..... 6 B. FIRA 6 ....................... ......... ... 7 C. FIRA 7 ............................... ..... 9 D. FIRA 8 ................. .............. o. .* ....... . 11 E. FIRA A . ............................*... .......... 14 III. IMPLEMENATION . ............ ..........o....................... 15 A. FIRA 5 .... .................. .. ................. .. 15 B. FIRA 6 .................................... too** ....... 28 C. FIRA 7 .................o*...................... ...... . 20 D. FIRA 8 ............ *....................#............... 24 E. FIRA 8A ......#.*..... o...... *.... **....... . .. .. . *.. 26 IV. IMPACT OF LOANS ........................................ 28 A. Production Impact .................................... 28 B. Institutional Development ............................ 30 1. FIRA .o.t*............... *.... . . . . . . . . ...... 30 2. Commercial Banks ................................. 33 3. BANRURAL ...... ............... ...... 34 4. The Fisc ......................................... 35 V. BORROWER AND IBRD PERFORMANCE ... . .. ... ........ . .... .. 37 VI. ISSUES AND FINDINGS ..................... ........*.. .. 39 Th document has a restdcted distribution and may be used by recipients only in the pedormance of their official duties. Its contents may not otherwise be disclosed without World Bank authoriation. Table of Contents (Cont'd) Page No. Table 1 - ISRD Disbursements for Agricultural Credit in Mexico - by Quarter, 1976-1987 ...................... 47 Table 2 - IBRD Disbursements for Agricultural Credit in Mexico - by Month, 1976-1989 ................ ..... 48 Table 3 - Disbursements from Inter-American Development Bank Agricultural Credit Loans in Mexico, 1976-1989 ...... 51 Table 4 - Government Loss through Erosion of FEFA Equity ........ 52 Table 5 - Government Loss through Erosion of FONDO Equity ....... 53 Figure 1 - IBRD Agricultural Credit Projects in Mexico Quarterly Disbursements, 1976-1987 .................. 54 Figure 2 - Agricultural Credit Projects in Mexico Annual Disbursements by IBRD and ADB ............... 55 COMENTS FROM THE BORROWER Attachment - Comments from Nacional Financiera .................... 57 PROJECT PERFORANCE AUDIT REPORT NEXICO FIFT8 AGRICULTURAL & LIVESTOCK CREDIT LOAN (1217-W) SIXTH AGRICULTURAL CREDIT LOAN (1569MB) SEVENTV AGRICULTURAL CREDIT LOAN (1891-M) IGHTH AGRICULTURAL CREDIT LOAN (2454-ME) AGRICULTURAL CREDIT LOAN (2610-ME) PREFACE 1. This is the Project Performance Audit Report (PPAR) for five agricultural credit projects supported by five loans all to Nacional Financiera SA (NAPINSA), Mexico, and all executed by FIRA (Trust Fnds Instituted in Relation to Agriculture). The five were as followst Project Name Loan # Loan Amount Approval Closing Final Disbursement <US$ mill> <date> <date> <date> 5th Agricultural & Livestock Credit 1217 125 03109176 01101180 05117/79 6th Agricultural Credit 1569 200 05/16/78 06/30/82 06/12/81 7th Agricultural Credit 1891 325 08/07/80 12/31/84 02/27/85 8th Agricultural Credit 2454 300 06127/84 09/30/88 . 01/21/86 Agricultural Credit 2610 180 07/23/85 09/30/87 12/22/87 The full amount of each of these loans was disbursed without any cancella- tions. As of February 28, 1990, the Borrower had already repaid US$376.08 million of these five loans. 2. The PPAR is based on the Project Completion Reports (PCRs) of the 5th, 6th and 7th projects (Loans 1217, 1569 & 1891),,/ on a draft of the IBRD-prepared sections of the PCR for the last two projects (Loans 2454 & 2610), on the Staff Appraisal Reports for the first four projects, on the President's Reports for all five projects, on the loan, project and guaran- tee agreements, on the transcripts of the Executive Directors' meetings at which the projects were considered, on study of project files, and on dis- cussions with IBRD and IADB staff. 11 A single PCR for the 5th and 6th projects was prepared by Latin America and Caribbean Region, Agricultural Division 4 based on the Borrower's background information and a country visit in May 1982. That PCR was reviewed by OED and passed through to the Executive Directors on December 28, 1982 (OED Report No. 4240, SecM 83-13). The PCR for the 7th project was prepared by the Latin America and Caribbean Region, Agricultural Division C, a country visit by project staff taking place in April 1985. - ii - 3. An OED mission vitAted Mexico in June-July 1989 and discussed the effectivene:s of the 5th, 6th and 7th projects (Loans 1217, 1569 & 1891) with Mexican Government officials of NAFINSA, Ministry of Finance (Hacienda), FIRA, the National Banking and Insurance Commission, partici- pating commercial banks, farmers and food processors. Another OED mission visited Mexico in November 1989 and discussed the effectiveness of the last two projects (Loans 2454 & 2610) with officials of NAFINSA, FIRA, Ministry of Agriculture and Water Resources, and the National Institute of Agricul- tural, Livestock and Forestry Research. The fraternal cooperation and valuable assistance of all these officials, farmers, bankers and entrepre- neurs in the preparation of this report is gratefully acknowledged. 4. The PCRo of the 5th & 6th and of the 7th projects provide a reasonably accurate account of project experience, including performance of IBRD, of FIRA, the project executing agency, and of other participants in project implementation. The PPAR explores certain aspects of design and implementation of these projects in greater depth and extends this analysis to the next two agricultural credit projects. The Audit draws attention to the massive erosion of the equity which the Mexican Government invested in FIRA. It also focuses on: whether the projects lead to investment and growth in Mexican agriculture; whether the projects strengthened agricul- tural financial markets; whether the Borrower's low-interest-rate policy was wise; how successful the policy of earmarking and subsidizing credit for low-income producers was; whether commercial banks need FIRA support for lending to non-low-income producers; and the pros and cons of having an agricultural rediscount bank do research and extension. 5. Following standard OED procedures, copies of the draft PPAR were sent to the Borrower and the executing agency. The PPA was revised in the light of comments from the Borrower. These comments are also reproduced as an Attachment to the PPAR. (1솔i-!l『―!!-[&&.!&!-;&1’。，… .& Iv - PROJECT PERFOPIMCE AUDIT REPOIRT AM AeRMLUELIMMagaLgenu-eatm Dale Rm mm - 620 UT raggm 2LÅ am*~ Attwa ot a~ ag a at gettmta totteatad A«aal Å~ **1 zettimto P»~ C«ta (usmuu05) 637.0 396.1 90 Crott ~me (UsIDUU98) 200.0 im* im 0~ 9~ ~roval 05116178 Daga utecelycoms 01/12/79 Data ~ leal COD~ tg conplatod 0.4. 0.6. Pt~ztt- ~15 complaced 5.4. 9.4. cloalm utc 06/30/82 96~4 Ucc of k~ 28Z plog~ Utt of a~ U-3n M~ 09 Du~ maftficuftes (1981) 67.000 f=lucs 148~ 1/ ~ of credlc ä~ letattes (1980) 13,646 14.061 108.99 CAULA= Dtsfo&Ss"MS TY79 VM pni rm ».0 86.0 164.0 =.0 32.9 162.1 m o - åt~ co a 09 satt=ta im isa in% e~ of 7~ DlGhG»~ 86111/81 ft~ pal repaid te <Woddaliyr-Xmwuuon> ag of 06/30/82 1/ Výcl~ b««~« trom ~IaduauT. of x~ speciall~loce rwirfö~=. 1~ or NU~ Datt Pe~ ta FUM a*F~ed Ii Ra~.LI T~ 3/ PMU~ 4/ pga~ eloft 83/77 4 60 3/ 1, b, e - å999~ 10/77 7 140 51 1. 1, b - - affervloton 1 01/78 1 22 71 t 1 2 el T ~Tviston a 04/79 3 33 7/ et 4 1 1 94 T ftpuv~ 3 11/79 4 64 8/ C. 10 te 6 1 14 t tepmvui» 4 09/80 1 17 4 N, t 8496~ 64 5 01/81 2 34 71 1 C~ ~ 04181 a 36 101 2M Kwwr ma sarga~ 71=mien h~tin ap~ vm 31 at c=*~ (4h~ ton) n~ ou~ zxe~ Ratos .. ~ Teu ave~ 0$ 1.00 - 22.37 Intern~ T«t* av~Se 1.00 - 13.26 c~ tion u« Av*rag* 1.00 - 24.44 rolla~ pmjout e~ 7~ vii L~»~t 1~ im-w ~ ~ (malluoo) Un325 Date 8~ 8/13/80 ~lealt~ac; b-a~twal ~et; c-flua~ ~yat; d-ag. fleumdål ~ywt, ~Oft~st; ful~ ofilc*r; 8.uv«~ *Pecl~ ~= 6~11*t; 114~ad. met~ J-agro~stl ~Icultaral ~19 *~ l~fte-ttm « Moor prob~ Z-anda~ pnblam; ~ ~Jor V»h~- Iltmprovlog. b-statlo~. ad ,-t~«1: ~U~l; and Com~ vIgh lm. 91~ vd M of La. 1217-n. C*~ with Sa of la. 1217-M. cce~ with m or La. 1211-m. ca*~ nick sa or la. 1217-m md pro~tion of seventh agr. «d uvu~ cr. pr. C*1~ ulth SIN of La. 1217-M ad ko. 1891-M- 101 ca~ with ~lon of ta. 1211-4a. PROJECT PERFORMANCE AUDIT REPORT SEVUTI AORICULTURAL CREDIT PROJECT ASIC DATA gaET aT PiOJCT DAT Appraisal Actual or Actual s of EBtuatt e stiUated Actual Appraisal Estimate Proect Coats (US~tton) 1.179.0 800.1 68 m=RD Loan AMunt .25. 323.0 0 Date !oard Approval 817f60 817180 Date 8ffectiveness 10/80 11/17/80 Closig Date 3/31/84 12/31/84 so~e ate of letu 21 over 100 15-53% lpia~lal Rate of Retu 22-41% 19-272 umber of Direct Beneficiaries 140.000 families 393.765 281 CUNULATIVE DISSULIENETS "yl 182 FY83 "184 ms8 Appraisal Estimate (USS*000) 70.0 185.0 290.0 325.0 - Aetaal (USso1i1~) - 63.0 157.0 302.0 325.0 Asual as x of Estimate - 342 54% 93 1002 Datø of Fil Dtabursemat 9/31/84 Praia T~i omm"br198 S10 tle - vi - STAFP lIgPUTS Prappre;eel 28.3 28.2 Appraiset 81.5 S1. Segoleles 8.9 1.9 Sgpervileln 9.8 21.1 18.8 6.2 11.4 5.4 N0.4 Other .4 1.. . 2. 8ubte6eI .4 114.0 13.4 21.1 16.6 6.2 11.4 6.4 138.1 MISSION DATA No. of Kandays Specialigations Performfc. Typen of Mission Date Persons In Represented 1/ Rating 21 Trend 31 problem 41 TnZ.~/r.) Preparation 11/79 4 64 5/ cfa.e Appraisal 1/80 7 130 bgE,Jter.i.k Supervision 1 10/80 1 17 6/ e 1 2 ,? Supervision 2 2/81 2 34 '1 c,k T 2 K.T Supervision 3 10/81 2 10 cIk 1 2fT Supervision 4 12/81 2 22 a,c _ 2 fT Supervision 5 11/82 3 48 i axce n NlT Supervision 6 5/M 1 16 e 2 .N,? Supervision 7 10/83 l 12 e i Nj Supervision 8 8184 2 22 jk N,? Completion 3/85 k 15 '/ ag.e - Total (SPN) Borrower Nacional Financiera (KAPINSA) gencuting Agency FIRA Fiscal lear January 1 - December 31 Name of Currency (abbreviation) Iex$ Currency Exchange Rate: Appraisal Year Average US$ 1&00 - 22.9 Intervening Years Average USS 1.00 - 67.0 Completion Year Average US$ 1.00 - Follow-on Project: Name FIRA VIII Loan/Number 2454-NE Loan/Anount 2S illion) 325.0 Date Board Approval 6/7/4 / amagriculturalist; bagricultural economist; cufinancial analyst* d-ag. financial analyst; -eeconomist; f-loan officer; g-livestock specialist; h-pasturc specialist; Iwagro-ind. engineer; j-agronamis:; kwagricaltural credic adviser. 2/ 1nproblem free or minor problem; 2moderate problem; and 3major problem. I/ lwimproving; 2estationary; and 3-deteriorattng. Y/ FnfinancIal; amanagerial; T?technical; P-political; and Gmother. T/ Combined with SPN of FIRA V and VI (La. 1217-ME and 1569-ME). T/ Combined with SPN of FIRA VI (Ln. 1569-fE). 7/ Combined with SPN of FIRA V and VI (Ln. 1.17-4E and 1569-ME). I/ Combined with PRN of FIRA VIII. / Combined with SPIN of FIRA VIII (La. 2454-ME) and Identification of Agroindustries Project. 一雙1么－ 方戴縫馭露萬目豳刃勵騙馴讓瀾勵買劉口陷跑緊 闖畸啊螂華 蠶才馴口忽劇悶江中皿徊口閱么口口中認寫點鰓取汙 《馴U口忽鳥騷馮州寫） 到纏鑼糅壁型L韭國矗 鑒u，既之鰓。細回一•。，一頃，_& 勵馴團豳絀鱸加”鹹他紳唱”細禹購神•神，居神l開“ 妒叫“俗觔鴉中夠卹•巒雙祠劉騙．o啊亂‘儲 必細•自•“東俗勗唱騙．唱l騷l頗》細00.0馴繙•O名00 織叫勵rJ．屆馴闖俗以爾jl卹編妒細• 斗．他欄開‘馴開喲•h•“鳥•加編，以論認細，詣O認寫 C馴閱口誦鼠調唱州間闖權朧勵為網勵為C朧馴么抑個閱劇匯閱口屹口 向•開臘細d繃馴 名認禺寫屆常屆萬寫禺霄寫認當甚娥 。•馴自．•。州頃網••．•－--－藝迎日變嬰旦迎蠶總州L- ”一尸甲黝忽付計州 肉閱“紹磁由飽閱．唱饑O!，開遞l認臨O―細亂O―觀臨O! “一州stfl州鬥’鬥 細自．騷騙亂才讓鬍際01’一要一！一！ ”劉L州j州一，一―斤―一― O•目d祠口do臘口馴••麾勿團鯽戲。，馴磚 口州唱刀叫曉期騙 ••••••■•••••口••• 師邊么贏斗留 －••••■•．•．閱••．•目 州馴州．閱鑰闖“卹。細離一•“．闢． 戶’戶“齣開，k由．。’，馴目“. 觔戶”齣闕盒細糰。細‘一糾“戶綢 向，閱團•細瞼”肩’•”州瞭。．&.&& 細觀“閱同細勿間蘭間臨“向州•““•間闕” 齣•d搬平•潤l開臨。向團•“•“•闖闕” 劉，國纔•細•“•，“糾網”&, 細．膩朧細目”開•．綢．闕”參。馴“磚‘O劇瀾“ 仰編叫“閱騙“‘觔•編悶•戲“匕糾” 以‘”細．邸開•叩”•開•叩“• ．■•••••••••■•••． y••．•”‘•間••“綢•，黝－一劍卸•．•••．抑占•口卹戲個•· - viii - USE 2f STAFF MOMM STAFF ~~) FM ~ FR4 FM ~ TOM Iden~ La =74 4~ 81.62 6142 III ý& thm 348 am SU~ bn 0&7 1230 1 LM CU OM 2LOI T~ ~ W~ Offi IU4 71.33 1240 1 OM 4.= OM 117.06 OM mm~ Mes~ p ma inga~ IdUffle~ 11/82 3 Plopaum i om i om 1 2 f 11/83 6 14 elfA~ 2 4 40 01/0 3 ma lim i f 8~~= 4 01/87 2 f** 8 ~mo 1 0.2 INMO a -I 600n~ as a agelc~ I 8 m *grund~~_ f m fin~a~O m~ S=« Projed FRU Wid %WO MIL OTHER PROJECT DATA umlomt Flnaftelcra log 49~ FIM Fl~ ~ « borrö~ Oc~ P al m~ of curr~ (abbr~lötloft) ~ lem# ~ average (1=) UN1.0D a ma ~* mm log ~re ~rags (19~ u881.00 a 212.4 cofflahlgn ~ sverage (19«) um.00 a 811.8 Fé$ lo~ proj~ Agriculturet Cmdit Proj~ - Ix - PROJECT PErP~IANCE AUDIT REPORT HEEICO AORICULTURA. CREDIT PRoJECT (LOAN 2610-mE) UASIC DATA SHEKT KET PitSJECT DATA Appraisl Aual or AeMa5 as i of Ex9ectloen Curr~n6 Eslast Approitl Estiat6 Proje Co~t (1$ all Ion) 473.0 442.4 94 Lo AaAnm (USI alliton) 180.0 180. 100 Nubr of 0lImet B~eeflerlee (La 2454 I 2610) - Fa ile* 105,000 118,780 118 CUMULATIVE EBTIMAT AND ACTUAL D~IURSEMENTS (in mlon of Uss) Oumula~ Otidbursemnts WORLD BANK FISCAL YEARS (Asofn 30> 1m im 1 AppmaissiW mate 150.0 180.0 180.0 % et Total 83.4 100.0 - Acumi 92.3 138.6 180.0 %GTotl 51.8 71.9 100.0 Dale of Final Dlsbursement: Dember 22.1987 PROAlCT TMABLE DDat Atual Sags Planned Reviisgd Date Prsparation/Review Mission March 1985 March 1985 Loan Ngot~ations 134ay-8 30-May-SS BrApprovat 25-un-f 234~-u Signatur. Date 2-Sept-85 Date of Effetivmnems Aquat 1985 1-1an-8 17-Deo-85 Conplation Datg 1-June-S~ 30-Sept4 7 30-Sept-7 CMing Data 1.De-86 30-Sept-7 30-Sept-87 STAFFlNP'Tweells) FU85 FVSS PYS? PV88 Tota ldent/Prep. 1.96 6.0 7.96 Nega nl 6.7 6.67. Supervisaon 6.31 10.11 6.9 2L40 Total Sta f West. 1.96 11.9 10.11 6.96 37.03 Daga Namésrof 8tffulm mm~m • 3 1iY> pmusen inte FWed repreeuted Preparaim 09165 1 1.5 1 Supervson1 11165 1 0.8 f supervislona 0os8 3a 5 e,e, Supervisione 2esta8 1 a0 6 Suprvislin4 01187 2 1.8 e,f staurvilons 5 0ter 3a a s,e,f supervsons6(partlal) 06167 1 0.2 se •= uconomist, ss - agrlcutural economlst;a a sgriunluranlst; I f nnel analyst;o = credf specilist Sourc: Project Flis and Banis MIS Notu This is an interim <oan. hrs was no appraisal misslon. 0l11E5_Pit0JECT DATA borrewer Naelmenl Pp~.. glIPaSA) Exec-l1g age -y PIaA Fleeu p yer ef beres.. Janury 1 - aeeber 81 Mmet of erray (abbrevIa6I.n) Mmens Apel.. year mvrag <198 US51.OO u 28s.5 Intervsalai year "verage (18n) USc1.00 0 11.3 Caplea~ year Bverage (1g87) US$51.00 * 1878.2 Fellee-.. proj..6: Minth Agri.ulteral Cr.dl6 Pr.jes6 - xi - PROJECT PERORMANCE AUDIT REPORT MEXICO FIFTH AGRICULTURAL & LIVESTOCK CREDIT LOAN (1217-ME) SIXTH AGRICULTURAL CREDIT LOAN (1569-ME) SEVENTH AGRICULTURAL CREDIT LOAN (1891-ME) EIGHTH AGRICULTURAL CREDIT LOAN (2454-NE) AGRICULTURAL CREDIT LOAN (2610-MB1 EVALUATION SUMMARY Introduction 1. Through these five loans, which disbursed from late L976 through 1987, IBRD provided US$1,130 million to the Mexican Government's agricul- tural credit system. The loans, commonly known a2 FIRA 5, 6, 7, 8 and 8A, were approved in 1976, 1978, 1980, 1984 and 1985 respectively. 2. Prior to the five loans audited here, IBRD made four very similar agricultural-credit loans to Mexico for $275 million. These disbursed from 1966 through 1975. Subsequent to the five loans audited here, IBRD made another very similar agricultural-credit loan for $400 million which dis- bursed from 1987 through the present. A further loan for the sub-sector is currently being processed. Concurrently with these loans purely for agri- cultural credit, IBRD provided about another US$ 330 million to the Govern- ment's agricultural credit system through credit components of other agri- cultural loans. Thus IBRD has contributed US$2140 million of funding to Mexican agricultural credit since 1966. Findings 3. The Audit finds that, in a negative policy environment for the agri- cultural sector, the five loans helped to keep agricultural investment going. The projects could not keep sub-borrowers from transferring resources out of the sector, but, at least, agricultural investments were made with loan funds in most cases (before they were opened up to support recurrent-expense borrowing in 1983). On the whole, the investments were probably sensible, financially and economically. 4. The projects were immensely costly to Government and contributed to its foreign debt crisis. By deciding not to insulate FIRA (the Trust Funds for Agriculture) against inflation and not to let it atrophy, Government effected a massive resource transfer to a small and generally better-off part of the rural population. IBRD argued against these transfers. How- ever, when the alternative to accepting them was slowing disbursements, IBRD repeatedly backed down from its principles and from agreed legal posi- tions. 5. The projects were moderately successful in using FIRA, a specialized public credit institution, to get commercial banks to lend to a somewhat lower income stratum of farmers. Most poor Mexican farmers, however, - xii - continued not to be clients of the IBRD-supported Government credit system. The success in broadening the income strata serviced by commercial banks stems from the incentive subsidies ?IRA gave to participating banks and not from subsidized interest rates to the low-income producer (P81). These subsidized interest rates were very costly and created incentives for cheating by borrowers and participating banks. It was costly for FIRA to combat these incentives. 6. FIRA carries out extension services and even adaptive agricultural research that would normally be the function of the Ministry of Agricul- ture. In so doing, FIRA is filling an important gap in services, at least to that part of the farming community which it serves. 7. Credit institutions were strengthened through the projects. How- ever, savings mobilization by farmers and commercial banks was weakened. It is too early to say whether credit discipline, which is excellent, will survive the withdrawal of massive public subsidies. Background 8. Before the 1970s, Mexico had kept inflation strictly under control, but inflation edged up to the 5 percent annual range in 1970-2 and then to the 12-24 percent range during the implementation period of FIRA 4, 1973-5. This made the real interest rates paid by farmer-borrowers under FIRA 4 negative. Inflation worsened during the 1977-87 period when FIRA 5, 6, 7, 8 & 8A were being implemented. 9. Mexican agricultural policy has aimed to keep down food prices for the urban majority. Mexico is an efficient producer especially of sub- tropical and labor-intensive farm goods, but less so for mechanizable tem- perate-zone products, such as wheat, maize, soybeans and meat. When Government holds down farm prices to satisfy the urban consumer, production falls, which requires imports that use foreign exchange. Policy then swings to help Mexico's farmers. Policy has swung in this cycle for a century. Most of the time, agriculture has been taxed through administered prices to promote industrialization. 10. Nevertheless, Mexican agricultural growth in the 1960s and early 1970s was significant principally for two reasons: 1) increased specializa- tion in labor-intensive "stoop" crops like tomatoes, other vegetables and strawberries for export to the North American market; and 2) the "green revolutiong in wheat, which lowered production costs and allowed Mexico to compete with North America. Neither of these factors were much help to the majority of rural Mexicans who have very small plots, most of which are devoted to maize and beans, yields of which have stagnated. Obiectives 11. IBRD's early agricultural credit loans were designed to support technological innovations, most notably pasture improvement. Through the five loans being audited here (and their successors), IBRD sought primarily to strengthen rural financial markets, while still fostering investments that made good economic sense. The executing agency and the immediate object of IBRD's institutional-strengthening efforts was PIRA, composed of three Government-owned trust funds designed to promote agricultural lending through commercial banks. 12. IBRD sought to strengthen FIRA bys - increasing its capital (Except for FIRA 7, Government contributed IBRD loan proceeds to FIRA as capital; IBRD also required that Government augment FIRA's capital through the central bank); - getting FIRA to adjust the rediscount and on-lending rates on its funds to make them higher than inflation; and - improving FIRA's management by improving monitoring and evaluation (ME) and its operational manuals. 13. Because PIRA worked through commercial banks, the borrowers of FIRA-rediscounted loans were creditworthy farmers able to meet stiff colla- teral requirements--roughly the best-off 15-20 percent of the rural popula- tion. IBRD and the Borrower sought, from FIRA 4 onward, to induce comer- cial banks to lend to lower economic strata, the PB. The inducements, for borrowers qualifying based on family income, were preferential and subsi- dised interest rates; FIRA technical assistance to banks and farmers for project preparation, processing and implementation monitoring; and free repayment guarantees. 14. In working through FIRA, IBRD first nearly ignored Government's largest rural-credit institution--BANRURAL. Over 90 percent of this insti- tution's loans were short-term and almost all were to farmers unable to provide collateral. BANRURAL was financially unsound. Nevertheless, it entered project objectives in two ways. Firstly, with commercial banks unenthusiastic about lending to PBI, BANRURAL was an attractive vehicle for achievement of this mutually agreed objective (for which it was used start- ing with FIRA 6). Secondly, IBRD hoped to use the carrot of access to FIRA funds to induce the Borrower to reform BANRURAL. While this was an IBRD objective, it was not a mutually agreed one. Implementation Experience 15. The first four of the five projects were implemented quickly in the sense that the loans disbursed on or ahead of schedule. The main impedi- ments to even more rapid disbursement were twos - sluggishness la sub-loans to lower income strata (PBI) in the first four IBRD loans; and - the Borrower's agreements not to submit disbursement requests for FIRA commitments during periods when it did not reach agreement with IBID with respect to FIRA's interest rates as stipulated in legal agreements. Such periods were numerous during implementa- tion of FIRA 7, 8, and 8A. - xiv - 16. During the implementation period, Inflation accelerated further as followsa Implementation Period Average Annual Inflation (Project) (Percentage) FIRA 5 22 FIRA 6 24 FIRA 7 75 FIRA 8 60 FIRA 8A 123 The loan agreements for the five projects had a variety of provisions regarding interest rates. These reflected delicate compromises between IBRD's concern that real interest rates be positive and the Borrower's reluctance to raise them, coupled with a belief or hope that inflation was about to come under control. With accelerating inflation, PIRA's fixed Interest rates, negative in real terms under PIRA 4, became much more so. 17. During implementation of all five projects, IBRD repeatedly urged the Borrower to raise interest rates and to index them and to make them adjustable; the Borrower consistently showed reluctance to do so. There were numerous disagreements over the abstruse legal provisions which des- cribed the compromise on interest rates. IBRD frequently backed down, agreeing to interest rates that were still very negative in real terms and often inconsistent with what had been agreed earlier. 18. During PItA 7, when inflation accelerated, the Borrower thrice was unwilling to make the interest-rate adjustments legally called for and voluntarily suspended submission of disbursement requests for commitments made by participating banks during these periods. These periods covered 26 of the 39 months during which FIRA 7 was disbursing. Nevertheless, PIRA 7 was fully disbursed on schedule because IBRD again agreed to revised inte- rest rates that were well short of positive in real terms and also because of provisions of the late-1983 Special Assistance Program, under which IBRD agreed to relaxed disbursement rules for FIRA 7. Theve changes, which went counter to three of IBRD's objectives in the program, reduced the loan share allocated to PRI, reduced financing of investments by permitting rediscount of loans for recurrent expenses, and reduced the investment share of participating banks and farmers by increasing YID Is share. 19. During FIRA 8 and 8A, the interest-rate discussion was governed by GIRA, the General Interest Rates Agreement between IBRD and the Borrower and covering all IBRD credit lending to Mexico. In spite of GIRA, FIRA's rates remained well below inflation rates throughout implementation of FIRA 8 and 8A. One problem was that the Average Cost of Funds, a Government- controlled index to which interest rates were indexed, was itself below the inflation rate by from 3 percent to 28 percent over the period. Another problem was the Borrower's difficulties in conforming to GIRA, which, as under FIRA 7, led to repeated periods of agreement not to submit disburse- ment requests. There were disputes over income classes of borrowers, the definition of "basic crops, for which farmers got preferential interest rates, and farmers' contributions. In spite of all this, under FIRA 8A, - XV - still-negative but nevertheless rising interest rates finally started to curtail the demand for FIRA-rediscounted credits, particularly for non-PBI farmers who had to pay rates closer to the real cost of money and to put up more of the investment costs themselves. 20. The inducements to lend to PZH (see Para. 5 above) proved more enticing to the farmers than to the commercial banks. Disbursement of funds earmarked for PBI proved slower than that of non-earmarked funds under all but the last of the five loans, as it had in FIRA 4. On several occasions, the Borrower sought modifications in the definition of PSI to permit part or even most of the upper rural 15-20 percent to qualify; IBRD refused. Sometimes, the Borrower unilaterally modified the definition, leading to supervision disputes with IBRD. IBRD supervision missions cons- tantly suspected that richer farmers were masquerading as PBI to obtain the subsidies. Spot checks discovered evidence of such cheating and one PIRA 6 supervision mission discovered a substantial number of loan applications from manifestly non-qualifying farmers that had been submitted by partici- pating banks for rediscount and accepted by FIRA. (The Borrower arranged to replace these with other, qualifying loans.) The Audit does not know how widespread this phenomenon was, but notes that FIRA is structured and motivated to combat it. 21. During FIRA 5 implementation, IBRD funds not earmarked for PBI were exhausted whereas a large balance remained in earmarked categories. BANRURAL's customers were PBI, but IBRD and the Borrower recognized that BANRURAL was not a sound banking institution. Therefore, the legal agree- ments provided that FIRA not rediscount BANRURAL's loans without IBRD's determination of BANRURAL's eligibility. IBRD wished to encourage the Borrower to reform BANRURAL, but also wished to maintain the flow of dis- bursements and saw BANRURAL as a tool for doing so. IBRD set out its conditions for determining BANRURAL's eligibility in September 1976. A series of analyses, proposals and counter-proposals, and a temporary, limited determination of eligibility culminated when the September 1977 supervision mission reported that BANRURAL had not substantially improved. Nevertheless, the mission recommended that IBRD declare BANRURAL a partici- pating bank, eligible for FIRA rediscounts. In October 1977, IBRD did so. 22. Even with BANRURAL, FIRA's disbursements under PBI categories lagged until implementation of FIRA 8A, in part because BANRURAL had ample sources of financing other than IBRD (and also other IBRD loans). Under FIRA 8, participating-bank spreads were made much more attractive for loans to PBI than for loans to other producers. Participating banks responded and, during FIRA 8A, loans to PBI finally exceeded targets while those to others did not. This response was mainly on the part of commercial banks and not BANRURAL. 23. IBRD concurrence with the Borrower over speeding disbursements again became evident when FIRA 7 was included under the Special Assistance Program, sacrificing policy objectives in order to get PSI categories dis- bursed rapidly (see Para. 18 above). 24. FIRA 7 especially concentrated on management atrengthening through building up FIRA's Planning & Programming Unit, its operational manual, and - Zvi - its monitoring and evaluation unit. These were built up (the manual euch later than stipulated in the legal agreements). IBRD was never satisfied, however, with the manual or with the M&E system. In M&E's case, IBRD did not feel it shed enough light on the ultimate Impact of FIRA's operations. The Borrower feels that IBRD expectations were unrealistic. Furthermore, regarding the X&E system built up with considerable outside technical assistance, the Borrower asked IBRD for guidance in improving what IBRD considered weaknesses but felt that it got only the assurance that there was technical-assistance money left in the loans, not intellectual gui- dance, a position which IBRD staff disputes. In the Audit's opinion, FIRA management did not appreciate the uses M&E might serve but considered it an IBRD imposition. Results 25. Regarding the ultimate productive impact of the loans, the Borrower has confidence in FIRA's reporting and M&E system, IBRD's PCRs indicate some reservations, and IBRD's supervision reports were still more skepti- cal. The Audit is impressed with the difficulty of measuring the Oultimate impact' of agricultural loans under any circumstances, but particularly so for Mexico during the period in question. For most of the period, agricul- tural output prices were not being adjusted for inflation so that agricul- ture was being taxed heavily by inflation; the sector was stagnating; there was probably net disinvestment from agriculture to other sectors and to abroad. Under these circumstances, the farm models in FIRA's reporting system might give an accurate financial picture but the wrong economic one. That is because assets are fungible, especially for the relatively well-off borrowers whose investments FIRA was predominantly rediscounting. Even if the investments IBRD financed through FIRA are financially and economically sound, as the models suggest that they were, they might merely be the ins- truments that permitted borrowers to withdraw other resources from the sector. The Audit can not estimate the economic benefits stemming from the loans. 26. That much said, there is still a prima facie case that FIRA-redis- counted investments were pretty sound ones. Firstly, ?IRA supervised these loans, so they were not phantom investments. Since FIRA's technical agricultural staff was and is competent and highly motivated, these investments have been generally sound. Secondly, even though such highly subsidized loans left plenty of roon for inefficiency, the sectoral distortions discouraging agricultural investment were great. Crudely, they tended to offset the pro-investment distortions of FIRA's subsidies. Thirdly, if some PIRA loans were accompanied by their borrowers' transfers out of the sector, not FIRA but the sectoral environ- ment was responsible for that. There was probably net agricul- tural disinvestment in Mexico despite FIRA, but there would have been more if interest rates had not been subsidized. - xvii - 27. Regarding financial impact, the most serious result of the loans is their cost to Government. Credit discipline in the FIRA system (quite a contrast to the BANRURAL system) was excellent. Defaults and bad debts were not a problem as participating banks repaid FIRA and farmers repaid banks (perhaps in expectation of repeater loans at similarly negative real interest rates). The problem is that FIRA's capital eroded seriously. Since its interest rates were radically less than inflation, what it got back was worth less than what it had lent. The lose In value of the equity of FEEA, the most important component of PIRA for this purpose, from end 1976 to end 1986 was US$1150 million. (The Audit has not calculated the cost to Government of FIRA's other two components, FEGA & FONDO, but these are believed to be minor by comparison. The public cost of BANRURAL, by contrast was several times larger than that of FIRA; BANRURAL not only eroded its capital by lending at negative real interest rates but also had poor repayment discipline and very high administrative costs.) 28. Government could have forced FIRA to live with the consequences of the losses in purchasing power of its capital--a reduced real level of operations. It did not. With the help of US$1130 million from the five IBRD loans the proceeds of which were contributed to FIRA as capital (FIRA 5, 6, 8 & 8A) or as a loan (FIRA 7), and of other foreign borrowing and loans from the central bank, Government maintained the real level of FIRA's operations. FIRA and, in large measure, the participating banks, were shielded from the financial consequences of their loans, while those farmers fortunate enough to get them benefited at Government expense. 29. Government's recourses to foreign borrowing to maintain the value of FIRA's capital had, of course, to be paid in foreign exchange with Govern- ment absorbing the cost, including exchange-rate risk. In the case of the IBRD borrowings, assuming that present exchange-rate relationships continue to the end of the repayment periods, actual US$-denominated cost of funds will have been as follows (see PPA, Para. 135 for explanation)t Project Rate Relative to LIBOR (2) (2) FIRA 5 4.7 -5.0 FIRA 6 1.3 -6.5 FIRA 7 12.1 +4.4 FIRA 8 16.0 +7.3 FIRA 8A 10.4 +1.2 It is useful to distinguish these financing costs, as well as the costs of financing other foreign borrowing and recourse to the central bank to sup- port FIRA, from the underlying losses in value of FIRA's equity. 30. Regarding Institutional impact, the FIRA-cum-commercial bank system of lending to creditworthy farmers expanded thanks to Government support under the projects, partly financed by IBRD. There are more branches accessible to farmers. In many ways, the system is better than it was in the early 1970s. Branches offer more services. Credit is supervised; farmers get technical guidance on potential investments; loan maturities - xviii - and grace periods are made appropriate to the investments and to borrowers' circumstances. 31. These changes have occurred without deterioration in repayment discipline, which is excellent. In this regard, the FIRA system stands in marked contrast to Mexico's BANRURAL system and to government farm credit systems in many other countries. Would this discipline remain good if subsidies were removed and if :epayers did not expect to get a subsidized repeater loan? The answer will not be known until the system is put to the test. 32. FIRA's operating costs as a percentage of its average loan port- folio and of its annual volume of rediscounts were 3 to 4 percent for the period. This would be excessive for a pure rediscount operation. FIRA was more than that. Its operating costs were mostly "technical assistance," wherein it was performing research and extension functions normally provid- ed by agriculture ministries, at least for FIRA clients, and its FEGA loan- guarantee program for PB. 33. The truly administrative operating costs hover around 0.5 percent of loan portfolio and annual rediscount volume; the technical assistance and guarantees cost from 2.3 to 3.7 percent. These are reasonable, espe- cially given some of the special objectives FIRA had to try to realize. Because it had to try to channel more credit to lower-income farmers, for inscance, FIRA could not be content with rediscounting but also had to try to keep track of every loan rediscounted and of the income of every bor- rower. To do this, FIRA built a computerized record-keeping system which is now quite effective. But for the income-distribution objective, how- ever, this would have been superfluous; participating-bank records would have sufficed. 34. Besides improving its record-keeping, FIRA built a Planning & Programming Unit, an operational manual, and strengthened its monitoring and evaluation system during FIRA 7. These did not resolve the problems IBRD staff seem to have expected them to resolve. The Audit feels that those expectations were unrealistic. The M&E system, for instance, was still not providing results anyone trusted at the end of FIRA 7. (It has subsequently been improved and now contains a wealth of data and generates periodic reports which, however, do not provide answers to questions FIRA management is asking and do not, of course, document ultimate impact.) 35. When the projects began, virtually all commercial bank lending was to the best off 15-20 percent of the rural population. The projects induc- ed participating banks to seek out good credit risks with reasonable investment ideas from somewhat lower economic strata. Roughly half of their FIRA-rediscounted loans were to PBI. Granted that some of these were richer farmers in disguise, the banks have nevertheless undeniably broaden- ed the income classes to which they lend. 36. This success stems from incentives to banks: PBI quotas, free tech- nical assistance, and free loan guarantees. These are costly programs; there is no way now of knowing whether commercial banks will continue lend- ing to a somewhat broadened economic stratum once this stick and these - xix - carrots are removed. However, Interest subsidies to PMI borrowers played no role in this success. There never was a shortage of PBI who would have been happy to borrow at FIRA's regular subsidized interest rates, only a shortage of banks willing to lend to them. Besides having no role in broadening bank lending to less-rich farmers, the PRI interest rate subsi- dies were both costly and harmful because they tempted richer farmers and banks to cheat and cost FIRA money to build and maintain a system to try to prevent them from doing so. Sustainability 37. As per Paras. 27-8 above, the projects and the FIRA system as a whole could not sustain themselves. Government had to supply large quanti- ties of money principally from its central bank and its IBRD borrowings to keep PIRA from shrinking to negligible size. In the process, which took place in the context of serious economic discrimination against the sector, Government also displaced savings that would otherwise have been mobilized for investment by farmers andlor commercial banks. During the period, Government dependence on foreign borrowings for its transfers to sustain FIRA's equity increased from less than half (as low as 14 percent in some years) to 59 percent in 1984 with the Special Assistance Program in place. PROJECT PERFOR0NANCE AUDIT EXICO FIFTH AGRICULTURAL & LIVESTOCK CREDIT LOAN (1217-E) SIXTH AGRICULTURAL CREDIT LOAN (1569-16) SEVE,' AGRICULTURAL CREDIT LOAN (1891-1E) EIGTH AGRICULTURAL CREDIT LOAN (2454-ME) AGRICULTURAL CREDIT LOAN (2610-E) 1. PROJECT BACMROUND 1. The five agricultural credit loans under audit represent an IBRD investment in Mexican agricultural credit of US$1,130 million. This amounts to over US$100 million/year over the eleven years--1976 through 1987--when these loans were disbursing. That figure is higher--about US$1,420 million or over US$120 million/year--when you count the agricul- tural credit components of other IBRD projects disbursing over the same period.1/ So IBRD involvement in the agricultural credit sector with these loaas was major. 2. And, it did not start or end with the loans under audit. In the decade before the fifth credit loan, IBRD lent Mexico US$275 million for agricultural credit through four loans exclusively for that purpose (Loans 430, 610, 747, and 910), which averages to about US$27 millionlyear. 3. In the more than two years since the last of the three loans under audit stopped disbursing. IBRD has lent a further US$400 million Ninth Agricultural Credit Loan (No. 2837), as well as another US$40 million for agricultural credit as parts of other loans disbursed wholly or partly in this period from November 1987 to the present. That amounts to about US$165 millionlyear in purely agricultural-credit operations and more than US$180 million/year for agricultural credit in all operations. Even when these figures are adjusted for the depreciation of the US Dollar, they show a large and growing IBRD financing of Mexican agricultural credit--a histo- rical pattern of which the five loans under audit covstitute the middle period--which adds up to US$2,139.4 million to date. A further loan for the sub-sector is currently under consideration. 4. The Mexican agricultural credit sub-sector that has received these large infusions of IBRD financing was distinctly dualistic when IBRD involvement started in the mid-1960s, and still Is. Mexico has a highly developed comercial banking sector. Many of these banks began to reach out to lend to farmers and ranchers whom they judged creditworthy decades ago in response to commercial opportunity and to Government prodding. The creditworthy are not the poorest of the poor of course. Farmers who have established relationships with commercial banks almost always belong to 11 PIDER I & II (Lns. 1110 & 1466), Tropical Agricultural Development (La. 1553), Rainfed Agricultural Development (La. 1945), 76 percent of PIDER III (Ln. 2043) and 65 percent of Agricultural Marketing Project for Perishables (Ln. 2262) assuming evenly time-phased credit disbursements. -2- that minority of Mexican farmers generally called "commercial".2/ Even when their land holdings are small or when they are ejidatarios,31 they are almost inevitably specializing in crops or animal products for sale rather than on the consumer staples of maize, beans and chiles (which are, of course, marketed as well as consumed). The repayment experience of the commercial banks in the sector is good. The proliferation of bank branches even in quite small towns is quite remarkable. Furthermore, Mexico has an extensive rural bus system linking remote villages with commercial centers, so that farmers need not and do not limit their banking dealings with the bank that has the nearest branch. 5. The other mode of this dualistic structure is BANRURAL./ This bank, and its three predecessor banks which were fused to create it in 1975, caters mainly to the needs of ejidatarios and non-creditworthy land- owning farmers. BANRURAL inherited a long tradition of paternalism, cor- ruption and inefficiency, as a result of which farmers generally deal with it only when they cannot deal with the commercial banks. Since ejidatarios have no legal title to their ejidal land and cannot use it for collateral, they most often have no alternative to borrowing from BANRURAL; when they do have collateral (including land they told privately), they generally avoid BANRURAL. Consequently, BANRURAL's customers are a disadvantaged underclass, not generally landless laborers, but those who farm very small plots, which are too small to support a family. Therefore, they are also farm laborers or have non-farm jobs. Such people are easily the majority in rural Mexico. They grow mostly maize, beans and chiles, the food staples for which, for most of this century at least, Mexico has held down farmgate prices in order to keep down the urban consumers' costs. With few exceptions, BANRURAL's customer base is and has traditionally been both poor and the victim of Government discrimination. 2/ 'What this audit will call 'commercialO farmers and ranchers often appear under other names in IBRD-FIRA parlance, e.g., MIP (middle-income produ- cers/productores de ingresos medianos) or OP (otros productores). These terms generally refer to the roughly 20 percent of Mexican rural population who do not fit the definition of PBI (productores de bajos ingresos/low-income producers). This audit uses the term PBI to refer, as IBRD loan documents refer, to families deriving most of their livelihood from farming or animal husbandry arid whose net annual family income is less than 1000 times the official minimum daily wage in their region. They are, roughly, the louer eight deciles of the rural population. They are also referred to as PIB, LIPs, low-income producers, etc. 3/ Ejidos are a product of the Mexican revolution. Land belongs to the ejido, the community. The land is divided into parcels to which a member, an ejidatario, has a use right, which is passed on to one heir. The land may not be legally sold, rented, pledged as collateral, or sub- divided. Many ejidal lands are rented or sub-divided illegally. Ejidos were meant to perform many economic functions communally, but today collective ejidos are rare and the principal remaining (and very important) artifact of the reform is the special status of ejidal land. / The common name for Banco Nacional de Credito Rural (BNCR). 6. Such was, and is, the rural credit system through which IBRD attempted to effect agricultural change by means of its credit loans. Its essential characteristics were not much affected by the fusion of Banco Ejidal, Banco Agropecuario and Banco Nacional de Credito Agricola to form BANRURAL in 1975, nor by the nationalization of the previously private and semi-private commercial banks in 1982. 7. In the beginning, IBRD was trying to achieve discrete agricultural changes through its credit loans. These were principally in the livestock sub-sector, as indicated by the names of the early loans. Throughout Latin America in the late 1960s and early 19709, IBRD was fostering the tech- niques pioneered in New Zealand--pasture improvement to produce meat from grass (and legumes), not grain. Staff repeatedly expressed frustration that a large portion of Mexican farm credit lending was financing animal purchase--a transfer payment from the national point of view rather than an investment--with little for pasture improvement. 8. Over time, but especially with the loans under audit, IBRD objec- tives shifted from fostering designated concrete agricultural changes, to changing the structure of the agricultural credit systems procedures for screening credit applications, lending terms (including interest rate), and the income class of credit recipients. Particularly, regarding the latter, IBRD ained to get commercial banks to lend to a broader income class of farmers. 9. A final point of background critical for understanding the design of the five projects under review and their implementation is the trust funds through which the Mexican Government intervenes in the agricultural credit sub-sector. These are collectively known as PIRA (Trust Funds Ins- tituted in Relation to Agriculture/Los Fidecomisos Instituidos en Relacion con la Agricultura). 10. Government founded FONDO (Guarantee and Development Fund for for Agriculture and Livestock/Fondo de Garantia y Fomento para la Agricultura, Ganaderia y Avicultura) in 1955 to refinance commercial-bank short-term agricultural credits. In the early 1960s, Inter-American Development Bank (IADB) and U.S. Agency for International Development (AID) funds were passed through FOND to refinance longer-term agricultural loans. 11. PFPA (Special Fund for Agro-Livestock Financings/Fondo Especial para Financiamientos Agropecuarias) took over FONDO's long-term financing responsibilities in 1965. Government contributed USAID and IBRD funds (IBRD lending for Mexican agricultural credit started in that year) to FEFA as equity and allowed accumulated net income to augment equity too. 12. FEGA (Guarantee and Technical Assistance Fund/Fondo de Garantia y Asistencia Technica) was created in 1973 to encourage lending to low-income producers (ejidatarios and others) by subsidizing the banks' costs of loan evaluation and technical assistance, and by guaranteeing loan recovery (a free service until 1988). 13. FIRA is composed of FONDO, FEFA and FEGA. For purposes of these audits, FEFA is the trust fund of interest. With minor exceptions, IBRD loans exclusively for agricultural credit were disbursed through FEFA. For simplicity's sake, we refer to the loans as FIRA 1, 2, 3 and 4 (Las. 430- - 4 - ME, 610-ME, 747-ME and 910-ME); FIM S. 6 and 7, 8 and 8A--those audited here--(Las. 1217-ME, 1569-ME, 1891-ME, 2454-ME and 2610-ME); and FIRA 9 (Ln. 2837-ME). 14. There are many other public trust funds in Mexico, a number of which impinge on the agricultural credit sector. One, however, warrants mention here. FICART (Trust Fund for Credit in Irrigated and Rainfed Areas/Fideicomiso para Credito en Areas de Riego y Temporal) was created in 1976 to permit IADB to channel its lending to low-income farmers through BANRURAL while maintaining an arms-length relationship with BANRURAL, which could not have been pronounced creditworthy.5/ At the same time, IBRD continued to channel its credit through FIRA, while seeking to shift to low-income producers. This was the case for the five loans under audit here. At this time, FIRA rediscounted BANRURAL loans, among others, with IBRD funds. However, IBRD funds were then (from 1981) passed through FICART for the credit component of the Rainfed Agricultural Development Loan (PLANAT) (La. 1945-ME). Moreover, starting with Loan 2837-ME in 1987, towards the end of the disbursement period for the last of the five projects under review here, IBRD channelled some of its regular funds for Mexican agricultural credit through FICART. 15. FIRA 4, the Fourth Livestock and Agricultural Project (Loan 910- ME) was the credit project immediately preceding the five projects here audited and the last to be audited by Operations' Evaluation Department (OED) (Report #2577 of July 3, 1979). It was the first to earmark a part of the credit to low-income producers (PBI). The definition of PSI excludes an estimated 15 to 20 percent of rural families who are also, for the most part, the banks' traditional customers.6/ Nineteen percent of Project costs were allocated to PBI, and actual lending to them was esti- mated at 15 percent. These PBI borrowers were given subsidized interest rates-- initially 37 percent less than the interest paid by regular borro- wers--and participating banks were given subsidies through FEGA as an incentive to finding PBI customers and working out viable investment 5/ IADB financing for agricultural credit in Mexico has also been massive. For the period of this audit, 1977 through 1987, IADB disbursed US$865.3 million to the sub-sector through 11 loans, US$475.4 million through 6 loans to FICART and US$389.9 million through 5 loans to FIRA. In 1988 and 1989, IABD disbursed another US$249.4 million to the sub-sector through two additional loans and the residue of two of the earlier 8, making a total of US$1,114.7 million over the 13 years. See Table 3 and Figure 2 for details. Through the period under analysis, 55Z of IADB financing was channeled to BANRURAL through FICART so that its impact on IBRD farm credit financing was slight. IADB financing through FIRA was concentrated towards the end of the period under analysis. For the six years from 1978 through 1983, IADB financing through FIRA (US$60 million from its Loan 20-IC plus the residue of Loan 294-OC) amounted to 132 of joint IBRD-IADB financing for FIRA. For the four years from 1984 through 1987, IADB's US$310.45 million for through FIRA amounted to 1/3 of joint IBRD-IADB financing. During FIRA 8 & SA, there was competition between IADB and IBRD funds (see Paras. 36 & 71 below), and also informal cooperation between IADB and IBRD. 6/ See FIRA 4 SAR (#133a-ME), Para. 2.06. projects with them. IBRD agreed to the Borrower's request for these subsi- dies only after an intense internal debate in which Projects staff, regio- nal and central, reviewed the pitfalls and inefficiencies of trying to help the poor through interest-rate subsidies, only to be over-ridden by argu- ments of Programs' staff. 16. It was also during FIRA 4 that real interest rates to farmers became negative for the first time.71 Prior to the FIRA 4 period, that is, prior to 1973, Mexican inflation had been in the 2-5 percentlyear range, but in the 1973-75 period, it ranged from 12 to 24 percent. The interest rate to the project's general borrowers (12 percent) was negative in real terms and the subsidized rate to PBI (7.6 percent) even more so. 17. IBRD processing of FIRA 4 shows malor concern for reducing the share of proceeds going to large-scale farmers, for the low interest rates proposed by Government, and for the proportion of funds going into live- stock purchases (transfer payments) as opposed to pasture development and infrastructure (investments). Senior IBRD management took the position that, unless more could be achieved on these fronts, the amount of the loan should be reduced from the US$200 million then proposed. It was, to US$110 million. 18. The FIRA 4 PPAR focuses on interest rates, where it finds that, at negotiations, IBRD perfunctorily backed away from its insistence on higher interest rates. It also criticizes reliance on investment models which it considered artificial, laments that monitoring and evaluation is not bet- ter, and opines that technical standards of project screening have fallen and that borrowers were getting less technical assistance as a result of pressure to expedite lending. The PPAR concludes that, under these circum- stances, a sector approach would have been preferable.8/ 71 "Market" rates may, of course, be negative in real terms, while positive real interest rates may be far below the opportunity cost of capital. Below, we will note on numerous occasions IBRD's concern to get positive real FIRA rates. For the record, in Mexico during the period under analysis, real interest rates could have been positive and still well below market, hence not fully consistent with IBRD OMS No. 3.73. For more on this subject, see S. Ramachandran. 'Re-lending World Bank Fundst In Which Currency and at What Rate?6 Washington: IBRD, August 1988, unpublished, 9 pp. + annexes. 81 The PPAR did not point out, however, that reduction of the loan amount from the $200 million proposed to US$110 million had no practical effect. The reduction of 45 percent in the loan amount was matched, during implementation of FIRA 4, by a 45 percent reduction of the implementation period from the 44 months proposed to 24 months. IBRD financing of Mexican agricultural credit remained at US$55 millionlyear (US$200 million/44 or US$110 million/24). - 6 - II. PROJECT DESIGN A. FIRA S 19. FIRA 5 was appraised in May 1975, four to five months before FIRA 4's actual full disbursement, early enough to allow for effectiveness of the nw credit by the time the old one completed disbursing and an uninter- rupted flow of IBRD funds under normal circumstances. IBRD files record much discussion, and general agreement, to move towards a 'sector ap- proach., This meant de-emphasizing trying to predict the actual financial and economic Impact of the credit in the SAR by elaborating farm models and aggregating from them, trusting the executing agency more, as in a DFC operation, and increasingly limiting IBRD's relations to the financial intermediary. The Implicit recognition of FIRA's maturity was discussed and agreed. 20. Nevertheless, FIRA 5's design was not very different from those of its predecessors. The principal difference was increase in ear-marking of the loan for PBI from 22 percent of loan proceeds under FIRA 4 to 47 per- cent (61 percent of the amount for crops & livestock). The proposed increase in this income class of customers supposed a radical change for the commercial banks which had been the conduit for IBRDIFIRA operations, or the participation of BANRURAL. just then being created out of the fusion of its three predecessors (see Paras. 5 & 6 above). BANRURAL had no track record, but its predecessors did, and theirs was not good, certainly from a banking point of view, but also, most observers would agree, from the view- point of service to the poor. BANURAL was not to be permitted to partici- pate in FIRA 5 rediscounting until approved by IBRD. 21. During preparation, interest rates were much debated, both as to general level vis-a-vis inflation and as to subsidy to be given to lower- income borrowers.91 Subsidized interest rates for PBI had been a feature of FIMA 4, and all fixed interest rates on FIRA-rediscounted loans turned out, ex post, to be negative as a result of unprecedented inflation. For FIRA 5, IBRD accepted fixed interest rates proposed by the Borrower that 9/ The distorted exchange rate and distorted agricultural product and input prices were not debated. - 7 - turned out to be quite negative, as well as a more elaborate system of interest rate subsidies for poor borrowers.101 22. Other main adjustments to project design in FIRA 5 were: loosening of procurement rules for the agro-industrial sector, and provision for getting the monitoring and evaluation (M&E) system to actually work and to provide useful feedback to project managers. The major discussion within IBRD however, other than the interest rate question, was over size of the IBRD loan. US$150 million was in the lending program; the Borrower and the Appraisal Hission recommended US$200 million; senior management cut the amount to US$125 million, arguing, inter alia, that 'By the time we reach No. 5 in a series of loans we surely ought to have accomplished most of our institution building objectives" (internal memo of 1 July 1975) and that 'a very substantial amount of this loan would be devoted to the financing of commercial farmers who rank far above the Mexican average in terms of income'. Senior Management added, 'I would not consider it responsive if the project were left intact, and the reduction in the amount of the loan were to reflect merely a reduction in the time slice covered by the project.' (see footnote 6 above) 23. Finally, it should be noted that, in 1975 and 1976, IBRD staff and representatives of the Borrower devoted major efforts to reviewing environ- mental standards for agroindustries and to assuring that entities financed under the project would not pollute. B. FIRA 6 24. FIRA 6 was appraised in November 1977, when FIRA 5 had been effec- tive for 15 months and 18 months before it was fully disbursed. FIRA 5's funds for commercial farmers were already largely disbursed and both sides anticipated that funds for PB1 would disburse quickly following IBRD's acceptance, in October 1977, of rediscounting loans made by BANRURAL (see Paras. 35-8 below). Based on these projections, in order not to interrupt the flow of IBRD funds, after a post-appraisal mission in April-May 1978, 10/ The Appraisal Mission reported (Issue Paper, 13 June 1975, Para. 3) that prevailing commercial interest rates for medium-term loans were 15-17 percent and inflation for the most recent year was 14.3 percent. It then recommended interest rates zo commercial farmers of 11.75 to 14 percent (0.5 percent higher than what the Borrower proposed) and to low-income and very-low-income farmers of 9 percent and 7.6 percent, with a spread of 2 to 3 percent for the participating banks. In so doing, the mission recognized the principal distortionary effects of negative and discriminatory interest rates, including the fact that these rates were likely to erode FIRA's capital. Nevertheless, during the review process, IBRD decided not to push for the recommended 0.5 percent increase to commercial borrowers but to accept the Borrower's existing rates because they had recently been raised and it would be *difficult' to raise them again so soon. In the Decision Memorandum (27 June 1975, Para 5), IBRD Regional Management argued that Mexican inflation was running at "around 10 percent.' (A post-appraisal mission in July, however, reported that it was 17 percent (BTO of 22 August 1975, Para. 3.) - 8 - the SAR was prepared and the project approved by the Board on 16 May 1978 (see Implementation, Paras. 50-2 below for more). 25. The project designed was not much different from FIRA 5. The Project Brief (12 October 1976) describes it as having *similar lines to Agricultural Credit V," that *it is not expected that the Project will have innovative goals; rather it will aim at the 'fine tuning' of those activi- ties already implemented." '...the conditions of effectiveness for this new loan are in essence the same as those for Loan 1217-ME" (IBRD-to-FIRA of 20 September 1978). The SAR describes the project as 'a continuation' (Para. 3.01) *under arrangements similar to those which have proven to be successful under earlier projects.* (Para. 3.04) Four adjustments ist design are worth mentionings a) In light of the difficulty of disbursing the PBI component of FIRA 5, and also because other foreign credit for PBI was available both from the IADB and from IBRD itself under the PIDER loans, the share allocated to PB1 was reduced modestly from the 47 percent of FIRA 5 to 34 percent (see Loan Agreement, Schedule 1). The Ap- praisal Mission had recommended 20 percent (see Issues Paper, 8 December 1977, Para. 10). b) In view of the losses in the real value of FIRA's equity from lending at negative interest rates under FIRA 5, IBRD staff proposed that commercial borrowers be charged a rate adjusted twice a year to the average cost of funds to banks (ACF)11/ plus 2 percent, while P81 would continue to benefit from a subsidized and presumably negative fixed rate of 11 percent. The Borrower argued that inflation was coming under control and wished to continue with its recently enacted raising of interest rates, for *other producers' about 17 percent, from the rate that had applied since 1975 (when the CPI was 43 percent lowert). IBRD's Decision Meeting on 16 December 1977 reached consensus on the mission's recommendations, but pointed out that they could not be applied to outstanding FIRA rediscounts and agreed (Decision Memo of 23 December 1977, Para. 8) that rates to *other producers" would be adjusted only when there was "significant change' in the ACF, leaving the definition of 'significant change' to future discus- sions with Banco de Mexico. Nevertheless, during negotiations in April, IBRD agreed to accept the Borrower's recently-adjusted interest rates--11 to 13.5 percent for PBI and 16 to 17 percent for commercial farmers--and its agreement "to furnish a letter stating that it is their policy to review onlending arrangements, including interest charges, spread for the participating banks and proportion of loans to be rediscounted by FIRA. The Government would keep the Bank informed of the changes that take place as a result of these periodic reviews.' (internal memo of 17 April 1978, Para. 3). The SAR, revised to justify this decision, states, "In view of the Bank's projections of inflation rates in Mexico declining from the 1977 level of about 20 percent to around 111 The ACF, calculated and published monthly by Banco de Mexico, is the average cost of money to the financieras (DFCs). It is slightly higher than the average cost of funds to banks and lags behind the CPI. - 9 - 10 percent by 1981, many LIP sub-borrowers and almost all other sub-borrowers would be paying positive interest rates over the life of the project.* (Para. 4.06) In the instance, inflation rates as measured by the CPI were 29 percent in 1977, 17 percent fn 1978, 18 percent in 1979, 26 percent in 1980, and 28 percent in 1981. c) In view of concern that too much of IBRD funds were funding Orotating cows*--livestock purchases that were merely asset trans- fers in the national sense--this activity was to be limited to 53 percent of livestock financing (see Project Agreement, Schedule 2)* down modestly from the 65 percent actual figure in FIRA 5. d) The original proposal was for a loan of US$130 million. IBRD staff recomuended a loan of US$175 million, but this was increased to $200 million (FIRA had suggested US$260-310 million). There is no record of concern by IBRD senior management this time, as there was for FIRA 5, (see above, Para. 22), that loan size ought to be proportional to importance of institution-building objectives, and that five consecutive loans might have been presumed to have accomplished these, barring evidence to the contrary. There is a theme of implied concern for maintaining or expanding the volume of lending. IBRD staff resisted the Borrower's suggestion that the definition of PB1 be changed. The definition used in earlier IBRD credit projects, including those like PIDER and PLANAT with large credit components, made about 80 percent of the rural population PBI and the other 20 percent commercial producers. The Borrower proposed to move PBI, for purposes of FIRA 6, from the poorest 80 percent of the rural population co the poorest 97 percent, arguing that the former were well covered, for credit purposes, by an IADB loan, by other IBRD loans, and by the Borrower's own resources. Neverthe- less, the earlier definition was kept. C. FRA 7 26. FIRA 7 was appraised in January 1980, a year after FIRA 6 became effective and 18 months before its final funds were disbursed. IBRD staff tried hard to make FIRA 7 less a "continuation* than its predecessor was. They set out to change the seriously negative interest rates at which FIRA had been lending, as well as FIRA's imperfect report system that made it impossible to be sure just how small farmers' contributions to their investment projects were, how often wealthier farmers were disguising them- selves as PBI, and what the impact of the project on the economy ultimately was. The imperfections of this reporting system, and the deficiencies on which it was not giving complete or reliable information, were the consis- tent complaints of IBRD supervision missions for FlRA 6 (see FIRA 6 under Implementation below). In seeking these changes, IBRD staff were being consistent with the findings of the Project Completion Report for FIRA 5 and 6 of 3 July 1979. From files, it appears that the Borrower was not averse to continuity and that the impetus for change was coming from the IBID side. - 10 - 27. Regarding Interest rates, IBRD staff sought to index them to the ACP with periodic adjustments to assure that commercial farmers' rates were positive in real terms and that those to PBI (and to a new intermediate category, effectively the poorest of the old commercial farmers category) were not too negative.121 This was part of a larger dialogue between IBRD and Mexico involving, among others, FIRA 7 (Lu. 1891) and the Small and Medium Industry Project (Ln. 1881) being negotiated in parallel. 131 A side letter from the Borrower agreed to consultations whenever there would be a "significant* movement in the AC? and that, in case of non-agreement between Bank and Borrower in these consultations, disbursement requests would continue to be submitted until the annual adjustment came due.141 28. To strengthen FINA institutionally, FIRA 7 design emphasized esta- blishment of a FIRA Planning & Programming Unit (PPU) and of a procedures manual. IBRD's objective here was to curtail abuses that IBRD staff thought were going on: lending without any investment contribution from the farmer; lending to richer farmers disguised as PB1; lending for non-invest- ments e.g. herd replenishment. What these were to dolcontain was spelled out in great detail. There appears in retroqpect to have been a naive faith that if the right operating regulations were issued, if the real results of past operations were known, and if applications for loans were well analyzed, then not only would the extent of abuses be known but FIRA would confine participating banks and sub-borrowers strictly to the uses of FIRA funds intended under this and previous IBRD loans. 121 At negotiations, starting rates ranging from 14 to 21 percent--the ACF was then 17.9 percent--were fixed, with the proviso that the Borrower would, annually, *adjust <to the ACF> the rates of interest payable on <subsequent> subloans...to reflect the change in the average...prevailing interest rates in...the preceding 12-month period compared to the average...prevailing interest rates for the 12 months preceding the previous adjustment." (Project Agreement, Schedule 2, Para. 8) The lawyers who wrote this were subsequently to argue as to what it meant. When adjusted, were rates to PBI and the two categories of commercial farmers to be just as positive or negative in real terms after the adjustment as the starting rates had been? What did 'reflect" mean? Whatever. An attempt was made to introduce adjustable rates. 13/ For the audit of the Third and Fourth Industrial Equipment Fund (FONEI) Projects (Las. 1560-HE and 1712-ME) see OED PPAR #7859 of 13 June 1989. These trust-fund projects in the industrial sector exhibit many parallels to FIRA 5, 6, & 7 in the agricultural sector and are also indispensable background to understanding the problems of Ln. 1881. 14/ The Audit has not been able to find a copy of this side letter in files. There is, however, a draft of it, prepared by IBRD staff and sent to the Borrower, as well as many references to it. Because of clauses in other loans involving agricultural credit, any agreement under FIRA 7 would trigger changes in subsequent projects financed under Loans 1462, 1553, 1569 and 1945 (PIDER II, Tropical Agriculture, FIRA 6 and Rainfed Agriculture). - 11 - 29. There was one important issue not addressed in the design of FIRA 7 and also absent during supervision of FIRA 6 up to that time, though it had been discussed during preparation of FIRA 4 in 1972 and perhaps earlier --substitution of IBRD funds for those of the borrowers and of first-tier banks.151 The substitution issue is relevant to all centrally-funded financial-sector operations and could have been addressed by increasing interest rates or ameliorated by increasing the percentage of investment project costs financed by participating banks and farmer beneficiaries. During IBRD's Loan Committee review, a note from senior management suggests awareness of this issues 0I think we have done an excellent job with FIRA and the credit system generally. Much of the appropriate machinery is in place as are the policies. Some further assistance may be required in scaling up, but I do not believe that this needs to be, or can be associated with further large-scale provision of capital. ...if our advice is valued the associated lending can be more modest and focussed on special programs.' The note accompanied acceptance of a FIRA 7 design that did not heed this advice. IBRD provision of capital was large-scale--at US$325 million, the largest IBRD loan up to then--and very broadly focussed. Mandated partici- pation of banks from their own funds was not increased; that of farmersl ranchers was not increased much. 30. Review of IBRD files and conversations with numerous representa- tives of the Borrower and of IBRD who were party to the design and negotia- tion of FIRA 7 make it clear that all were highly optimistic at that time that they had reached important agreement and understanding. It was widely accepted in IBRD that the complicated legal language about FIRA's interest- rate policy in the legal agreement enshrined a major agreement on reform, and that detailed provisions for institutional strengthening (operational manual, PPU, ...) would lead to a stronger FIRA more concerned with the impact of its operations on the national economy, national income distribu- tion, etc. During implementation, this optimism proved unjustified. D. FIRA 8 161 31. FIRA 8 was appraised in November 1983, three years after FIRA 7 became effective and about ten months before its final disbursement. But in August 1982, when discussions about FIRA 8 began, it was still expected that FIRA 7 would finish disbursing by mid-1983. Therefore, FIRA's Direc- tor wrote to the Bank, "in order not to interrupt the continuity of your programs, it would be desirable, of course, that technical activities rela- tive to the 8th agro-livestock development project could begin.* The ini- 15/ This issue was eventually taken up by the January/February 1981 SPN mission after FIRA 7 became effective. 161 As of the completion of this Audit, the IBRD portion of a joint Project Completion Report for FIRA 8 & 8A (Loans 2454 & 2610) was still in draft. Hence, there are no references to it, although the Audit reviewed the draft informally. - 12 - tial project brief (IPB) which IBRD staff prepared at the end of 1982 for a US$400 million loan to finance a US$1,000 million project strongly stressed continxity and rapid disbursementslresource transfer. The IPB proposed one major innovation -- half of the money was to go through FICART and BANRU- RAL, and only half through FIRA and the commercial banks. This OBank par- ticipation would allow the Government to streamline and strengthen BANRURAL and, above all, to steer BANRURAL more towards development type lending.* At that time, FIRA 7 was experiencing repeated agreements-not-to-submit- disbursement-requests due to non-compliance with interest rate covenants and other problems discussed under Implementation below. 32. From the IPB to IBRD's Decision Meeting in December 1983 a year later when FIRA 8 took its final form, discussion over FIRA 8's design turned on three main points: its continuity; interest rates to borrowerst and BANRURAL's participation. 33. FIRA 81s IPB was attacked for its acceptance of continuity, with demands that FIRA 8 "be ... tied to specific policy reforms," particularly as regards prices and subsidies (staff memo of January 20, 1983). Never- theless, over the ensuing year, FIRA 8 became more like FIRA 7, not less. 34. The designers of FIRA 8 recognised the importance of interest rate policy, inter alia because of the decapitalization of the trust funds through lending at negative real interest rates. The Issues Paper (Decem- ber 8, 1983) reported that FIRA's and FICART's portfolios were earning an average interest of 19 percent in 1982 and 1983 when inflation was (at that time) estimated at 98 percent and 80 percent. IBRD missions recommended that "each individual interest rate should in future be linked directly to ACF, with the agreed level of subsidies to PBI being determined as a per- centage reduction of ACP.1171 In the event, interest rate questions were taken out of the FIRA 8 design process because they were being dealt with at the national level as part of what was to become GIRA, the General Inte- rest Rates Agreement. It was stipulated that FIRA 8 would conform to GIRA and would not become effective until GIRA did. Delays in negotiating, signing, ratifying, and applying GIRA delayed project processing. At cri- tical times when GIRA negotiations were not progressing well, it was decid- ed to continue processing FIRA 8, despite the *signal' that might send to the Borrower that IBRD was not really serious about interest-rate nego- tiations lest the IBRD lending program be disrupted. As an internal memo noted on October 31, 1983 at one of the points at which project processing was continued anyhow, "... delaying appraisal %auld probably mean that the project would slip out of the FY84 lending program.' 35. Enlisting BANRURAL's participation and reforming it was, by now, a familiar theme that had been tried in FIRA 5 and 6 (See Paras. 20 & 24 above), so that the difficulties should have been well understood. BANRU- RAL, once more under new management, was interested and had prepared a 7- volume project preparation report. The IPB was sanguine. In February 1983 its author was sent to investigate BANRURAL's support services for farmers and, though the mission seems to have failed to understand the salient weaknesses of BANRURAL, some of the weaknesses were reported, .,t only the 17/ Revised Project Brief of October 21, 1983, p. 7, Para. 26. - 13 - ineffectiveness of BANRURAL's support services but the fact that it employ- ed more than 27,000 people. 36. Another IBRD mission in September rewrote the Project Brief in October. While still up-beat about reforming BANRURAL, this mission cut the share of project costs to be channelled through BANRURAL from US$600 million to US$128 million and the loan amount to be so channelled from US$200 million to US$60 million. That brought the projected loan to US$260 million (since FIRA was not to be cut) and project size to US$570 million. From that point on, what was to be Loan 2454 grew a further 15 percent to US$300 million and the FIRA 8 project costs by a further 34 percent (mostly contingencies) to US$764 million. While generally more realistic about difficulties of reforming ANRURAL, the revised Project Brief makes some etraordinary arguments: that the Seventh Project (1891-ME) was performing satisfactorily; that all FIRA required was fine-tuning;18/ that the declining share of BANRURAL business FIRA would handle over the 1980-83 period proved that FICART was better placed to influence BANRURAL;19/ 201 and that a US$130 million IADB loan for FIRA then being processed2l/ would not affect FIRA's need for an IBRD loan (PB, Paras. 10-12, p. 3). 37. The prospective Borrower would not discuss with the November 1983 appraisal mission the Borrower's nearly-concluded study of BANRURAL and proposed reforms. When a Mexican delegation visiting Washington on December 12-4, 1983 refused to provide information on BANRURAL, the December 15 Decision Meeting decided to drop BANRURAL as an executing agency but allow it to borrow through FIRA as before. FIRA remained the sole executing agency and the loan to be channeled through it was raised to US$300 million. In reviewing the Decision Meeting results and in Loan Committee, IBRD management asked how the new project would achieve BANRURAL 18/ IBRD staff have pointed out that FIRA itself did only need fine tuning, and that FIRA 7 was performing satisfactorily in the sense that it generally conformed to policies agreed by IBRD, although those agreed policies were faulty. 191 IBRD staff argue that there was *some separation' between FICART and BANRURAL at board-of-director level, as well as some interest, at that time, both in IBRD and in FICART, in insulating BANRURAL's long-term lending from the "bad practices associated with' its short-term leading. 201 As commercial banks strengthened their rural network, the share of loans which FIRA had to channel through BANRURAL, the farmers' bank of last resort, declined, from 17 percent in 1980, to 16 percent in 1981, 12 percent in 1982 and 7 percent in 1983. In this way, FIRA was influencing BANRURAL. Commercial banks were giving more farmers choices, and they were using them to avoid BANRURAL. FICART, however, could have no influence in BANRURAL since its only function was to rediscount BANRURAL loans. 21/ Loan 133-IC (Mex), US$128,886,000 of which was disbursed in 1984. Adding disbursements from two other loans, IADB disbursed US$164.2 million to the Mexican farm credit sector during 1984, US$128.9 million of it through FIRA. See Table 3 and Figure 2. - 14 - reform with BANRURAL reduced to a participating bank again. Staff replied that, under the circumstances, not much was possible.221 So FIRA 8 became very much a continuation. Negotiations were uneventful. 38. Before Board presentation, one Executive Director asked why PB1 continue to receive credit subsidies. Staff replied that these would be reduced, but that we had "acquiesced" to them because loan payments would cause a cash flow problem, especially to poor farmers, in periods of high inflationt Staff's Board presentation stressed the bigger picture of agri- cultural-sector reforms. in credit, staff argued for 'Government's inten- tion to gradually reduce and in time eliminate most credit subsidies,' and the 'spirit' of the GIRA 'to reduce susidies and move towards positive real rates in accordance with an agreed timetable.' One Executive Director pointed out that FIRA had lost 60 percent of its capital in two years, seriously undermining its ability to raise commercial funds. Even with GIRA, FIRA would not cover its costs and would require large subsidies at the end of the loan. This situation, he argued, produced misallocation of resources and increased the substitution of funds at the producer and ins- titutional level. The loan was then approved. 39. There was one innovation in FIRA 8's design which was overlooked in these discussions: the major encouragement to participating banks to lend to PBI. Earlier loans had offered these banks greater spread on PB1 loans but without visible effect. Under FIRA 8, the spread for commercial producers was fixed at 2 percent; that for PBI at 6 and 6.5 percent 231 Possibly during implementation of this project and certainly during implementation of FIRA 8A, as we shall see under Implementation below, this provision significantly affected bank and farmer/borrower behavior. . FMIA BA 40. Loan 2610 was originally called Eighth Agricultural Credit Project --Phase II but had its name changed to simple Agricultural Credit Project. It is commonly called FIRA 8A or eight and one-half. Its design is as similar to that of FIRA 8 as its processing was brief. 41. The remarkably rapid processing of FIRA 8A occurred in the 7th month of disbursements of FIRA 8, which disbursed in less than 16 months at a rate of nearly US$20 million per month, one third of the four years of disbursement anticipated at appraisal. In fact, 2/3 of FIRA 8 disbursed in December 1984 to Harch 1985, its first four months of effectiveness, thus at US$50 million per month (see Implementation below). In these circum- stances, the first mention of FIRA 8A in files, a memo justifying proceed- ing without an appraisal, makes it plain enough that the purpose is conti- nuity and rapid disbursements: 221 For more than six years thereafter, IBRD made no further attempts to use BANRURAL as an executing agency for its rural credit projects. 23/ Project Agreement, Schedule 2, Para. 6, p. 11. Note that there was a middle MIP category at this time which was abolished again in December 1985. For simplicity's sake, this Audit refers only to IPB6 and to ecommercial farmers' (see note 12 to Para. 4 above). - 15 - *To allow a review of the status of Mexican agricultural credit and to prepare and appraise the ninth credit project, supporting further policy and institutional reforms, we would need about nine months. In order to avoid disruption in FIRA's lending and the enhance the implementation of the General Interest Rate Agreement in agriculture, we propose an interim operation extending the Eighth Agricultural Credit project by about nine months...." (staff memo of May 2, 1985) A one-man preparation and design mission was announced for June 17-27, 1985. On June 26, presumably the day before the mission returned to head- quarters (there is no B7O in files), the regional vice-president forwarded the draft grey cover President's Report to the Senior Vice President Opera- tions. That memorandum notes that negotiations, of which there are no minutes in files, had already been concluded. The indecent haste continued with distribution of documentation to the Board on July 2 and Board appro- val on July 23. 42. An Executive Director noted that Loan 2610 was 8a fairly quick operation in our terms." In its presentation, staff again emphasized the broad reforms in the agricultural sector as a whole and made much of the GIRA agreement and the fact that the ACF, and hence agricultural lending rates, were above the inflation rate. The ACF was running at 50-55 percent and, staff reported, inflation at 35-40 percent.24/ On cross-examination, staff admitted that this situation (exceptional, as it turned out) had only prevailed for three months, and that 80 percent of prospective borrowers were expected to be PB1 who would not be affected by the subsidy-reduction program since their interest rates would not be tied to the ACF. A number of Executive Directors were not impressed, commenting on the rapidity of disbursement of FIRA 8, the misallocation of resources implicit in the negative interest rates, the difficulty of mobilizing savings when interest rates are negative, the preferability of improving PBI access to credit instead of subsidizing their interest rate, and the financial weakening of Government as a result of these projects. When, asked one Director, could GIRA be strengthened? Staff countered that IBRD had acquiesced in interest-rate arrangements that were less than ideal to offset the disin- centives to agriculture flowing from the trade regime with credit and pric- ing subsidies to producers in agriculture. The loan was approved. III. IMPLEmENTATION A. FMA 5 43. In retrospect, the most serious consequence of FIRA 5 was the heavy loss to Government from lending at fixed interest rates that were quite negative with inflation running at from 16 to 29 percent. This, however, did not impede implementation. Beneficiaries were eager to borrow at such subsidized interest rates; they could make substantial financial profits on the loans. Demand from farmers and participating banks for the IBRD funds being channelled to them through FIRA on such attractive terms was very strong. 241 Annual inflation for 1985 turned out to be 58 percent. * 16 - 44. Judging by the files, the main implementation problem, which began by delaying effectiveness, was how to lead the 47 percent of the loan ear- marked for PaM.251 This problem is mentioned only very tangentially In the PCR (Paras. 3.07, 5.10 & 7.04). Very few PBI were known customers of commercial banks; the loans that made sense for them were usually smalll the 3 percent spread was not enough to interest the banks, given the high cost of servicing these small loans and of finding these new customers, most of whom the commercial bankers regarded as uncreditworthy. At the rates at which they were offered (7.6 to 9 percent interest with inflation running at 16 to 29 percent), there was no shortage of farmers who wanted these funds, but commercial banks were not too interested. 45. Hence IBRD's interest in BANRURAL. IBRD staff had been intrigued, during FIRA 5 project processing, with the development possibilities of helping Mexico turn the new institution into a sound development bank. However, the files show that there were no illusions on either side. Two of BANRURAL's constituent banks, Banco Ejidal and Banco Agricola, were particularly noteworthy for their quantity of bad debts, their low loan recoveries, their focus on short-term lending, their over-manning, and the widely alleged corruption of their staff. IBRD was already channelling agricultural credit through these banks under PIDER I & II (Lns. 1110 and 1462-ME) and FIRA 4 (Ln. 910-ME) by having the banks run parallel opera- tions and keep separate accounts of these operations. There is no record of problems. Mexico's initiative to consolidate public-sector agricultural lending to the poor in BANRURAL, however, seemed to offer an opportunity to clean up the operation and make it efficient. Indeed, this was the Bor- rower's justification for the initiative. IBRD, therefore, made channell- ing of FIRA 5 funds through BANRURAL conditional on IBRD's determination of BANRURAL's eligibility (Loan Agreement, Schedule 1, 2(d) and Project Agree- ment, Section 2.06). What IBRD would wish to see before declaring BANRURAL eligible was set out in Minutes of Negotiations and summarized in a letter of 17 September 1976, which also summarized what the Borrower had already done and what additional steps IBRD considered necessary to put BANRURAL on a sound footing within five years. By telex of 23 November the Borrower accepted IBRD's BANRURAL plan. As a result, by telex of 30 November, IBRD agreed to allow up to US$15 million of Loan 1217 funds to be used to redis- count BANRURAL loans between then and 31 March 1977. These funds were fully committed by 31 December 1976. 46. If IBRD wanted to turn BANRURAL into a sound developmental insti- tution, it also needed BANRURAL to get the PBI part of Loan 1217 disbursed. The November 1976 SPN mission notes, IFIRA is continuing to find that the task of making sub-loans to the low-income producers sub-category ... is more difficult and time consuming than for MIPS (medium income producers).* The March-April 1977 SPN mission reported that disbursement for commercial farmers through end 1976 had been six times that for PBX. That mission analyzed BANRURAL carefully. Recovery rate was about 60 percent. It concluded, inter alia, 0BNCR's financial position is not sustainable' and 251 This compares with 30 percent for regular customers, 15 percent for agro-industry, and 7 percent for institutional strengthening, agricultural research and extension, etc. With re-allocations, this distribution finally becames PBI - 47 percent; other farmers - 44 percent; agro-industry - 7 percent; other 2 percent. * 17 - "The fundamental point is that a large part of BNCR's activities are not strictly of a banking nature, but include provision of technical assis- tance, marketing services, infrastructure works and, for some very poor communities, what amounts to cash grants-in-aid.1 A new action program was agreed. IBRD management agreed with some reluctance. An internal memo (19 April 1977) reflects the unease of one senior manager with a broader pers- pective and a sense of history: 0I agree that the financial picture of BNCR looks very bad. Los- ses of the order of $500 million per year are a national macro-economic and fiscal problem far transcending the questions of Bank financed operations. *Of course, when we first went into credit operations in Mexico, it never occurred to us that we would get close to the always notoriously unsound Banco Ejidal and Banco Agricola. And when, in 1973, we made the first Fondo <FIRA> loan with a low income producers (LIP) component, and the Banco Ejidal and Banco Agricola were considered as possibly eligible vehicles, it was agreed to use them only on condition that our funds could be strictly segregated from the rest and insulated from the noncommercial atmosphere pervading these institutions. It was thought that the problems of these institutions had to be solved by the Mexicans in their own time and in their own way, i.e. that they were too large and too deeply imbedded in the national political structure to become part of our negotiations. 'The merger, in 1975 of the two *social* banks with the more busi- ness-like Banco Agropecuario was justified by the Mexicans precisely by the argument that the latter's orientation would begin to affect the operations of the former..... While I had my doubts--after all the opposite could happen equally well--it is not yet clear that this was wrong. ...our position cannot change; we must ensure ... that our funds be handled cleanly and effi- ciently, as, apparently, they have so far., 47. In July 1976, IBRD Programs' staff discussed BANRURAL's continued access to rediscounting facilities with Mexican officials. Staff recorded (BTO of 3 August 1977) that "Banrural is not yet in a position to commit itself to separating its loans into hard and soft windows', only one point of the action plan, but commended BANRURAL management's "sincere effort to comply with the request of the Bank to classify its portfolio, review staffing levels and needs, and prepare financial projections' (none of which had been done), and recommended that BANRURAL be declared eligible for rediscount of the PBI funds remaining in Loan 1217. Projects staff pointed out (internal memo of 5 August 1977) that *there has been little real progress in carrying out the proposals agreed with BANRURAL for the improvement of its accounts and financial position., 48. At this time, IBRD ceased disbursing against applications for non- PBI farmers because their categories were exhausted. The September SPN mission reported that commercial farmers' loans accounted for 82 percent of those rediscounted to date, those of PBI only 10 percent. BANRURAL's finances and reform plans were again examined in detail. The report is highly critical of both. "The lavish government assistance in recent years, considered by all as a subsidy or grant, is the main contributing factor to low recovery of loans in the past and will be in the future if nothing is changed., The presentation of the balance sheet, overdues - 18 - rescheduling, reporting, interest rate subsidies, the short-tern, non- investment character of 90 percent of BANRURAL's operations, all are strongly criticized. In spite of all this, the mission reached the remarkable conclusion 'that the mere fact that a basic change in the lend- ing policy of BNCR is being introduced should make BNCR eligible to parti- cipate in lending operations from Bank loans.* On receipt of a letter from BANRURAL promising the reforms discussed with the September SPN mission, IBRD accepted BANRURAL (by telex of 20 October 1977) as eligible for redis- counts under Loan 1217 without limits. 49. Compared to the above, other implementation problems pale by comparison. As the PCR explains (Para. 3.09), Mexico's interest rates for loans to non-PBI in the sugar sector were inconsistent with interest-rate agreements under Loan 1217. After much soul-searching, IBRD decided to stick to its agreement; the Borrower did not apply for sugar investment rediscounting. The US$10 million allocated for this purpose was reallo- cated to farm loans for commercial farmers. Staff correctly worried about the quantity of financing of cattle purchase. This had amounted to 72 percent of the livestock sub-loans in FIRA 4 and 71 percent of those in FIRA 5 up to end 1976 (65 percent for the whole project. See PCR, Para. 3.18). The possibility that such purchases were transfers was highlighted when SPN missions visited farms whose cattle purchases had been refinanced several times but the size of whose herds had not changed. When funds were reallocated, at least no funds were transferred into this category. IBRD and FIRA staff put a lot of effort into strengthening FIRA's ME system, including provision of foreign consultants (apparently financed by FIRA's own resources and as part of IBRD's SPN budget). However, as the PCR reports (3.27-8, 3.32-3, 3.35), the system was still unable to provide the information management wanted. FIRA has complained that it asked for assistance from IBRD in this regard without results (see Borrower's Com- ments on FIRA 7 PCR); IBRD staff counters that consultancy money for this purpose was available in the loan and was not used, and that FIRA manage- ment never perceived how M&N could be useful to it. In any case, we can have little confidence of how many sub-borrowers really were PBI, of how many rediscounted loans were repeater loans, of what the producers' contri- butions to investments was, etc. Government unilaterally changed the defi- nition of PBI from that in the legal agreements; SPN missions reported cases from spot checks where PBI did not qualify. The Audit notes that this evidence confirms experience elsewhere that it is difficult, perhaps futile, to try to earmark credit for certain classes of farmers, especially using income as a test. B. FIRA 6 50. FIRA 6 became effective on 12 January 1979 and about one-quarter of it disbursed concurrently with FIRA 5 before the latter fully disbursed on 17 May 79 (not in October 1978 as anticipated at appraisal of FIRA 6). Although FIRA could now rediscount BANRURAL's loans to PBI, FIRA 5's funds earmarked for PBI moved more slowly than anticipated because of the volume of external financing available to BANRURAL for the same customers from an IABD loan and from other IBRD loans. FIRA 6 as a whole disbursed much more rapidly than anticipated at appraisal--in two and a half years instead of three and a half. However, as before, disbursements to commercial farmers proved more rapid than those to PBI. On 27 September 1979, eight and a half months into the disbursement period, only 22 percent of the funds - 19 - earmarked for PBI has disbursed, as vs. 70 percent of those for comercial farmers. Categories for the latter were exhausted by February 1980 after 13 months of disbursement in a loan that was meant to last three years; disbursements for PB1 were stretched out over an additional 17 months until June 1981. 51. Initially, IBRD supervision staff took accelerated disbursements as a sign of project success. The report of the November 1979 mission hails 'disbursements surging at a dramatic rate (almost 200 percent of appraisal rate)' and *strong demand for credit by farmers and increased interest on the part of commercial banksu. Later. IBRD staff apparently reflected on the likely causes of the surging disbursements. The report of the February 1981 supervision mission says, *Strong demand for Project funds is partly due to the low interest rates and in some cases substitu- tion of borrowers' and participating banks' fundes. With its fixed and low interest rates in an environment of increasing and unprecedented inflation, FIRA was making farmers' and banks' an offer they could not refuse, but in the process, the real value of FIRA's capital was declining dramatically. 52. In terms of implementation, however, FIRA 6 went smoothly. Spe- cial legal covenants were few and, though supervision missions found that some were violated consistently, in retrospect it is clear that neither the Borrower nor IBRD management took them seriously. Each supervision report and subsequent letter to the Borrower notes the same non-compliances$ a) sub-borrowers are not contributing the share of investment costs prescribed by the legal agreements, indeed frequently not any- thing1261 b) there were irregularities in the definition of PBI, delays in the application of agreed definitions, and farmers who were passing themselves off as PBI who did not qualify by any definition; and c) FIRA's M&E system and other aspects of reporting and record- keeping were deficient in that they did not provide reliable information on the above phenomena or on the "ultimate economic impact* of the project. The stridency of IBRD Projects' staff's complaints increased with succes- sive supervision reports and the Borrower's non-resolution of the problems, except for the last report of May 1982 which followed the transfer of a project manager to another region. It is the opinion of the audit, further discussed below, that farmers and participating banks were behaving ratio- nally to protect themselves from or to exploit distortions created by their economic environment and by the design of the project. This behavior was 26/ This should hardly have been surprising in an environment of economic uncertainty where agriculture was being squeezed by not raising farm procurement prices while the price of everything else soared. Agile farmers (and not exclusively "commercial' ones) were investing as little of their own resources as possible in their farms, expatriating those resources in some cases, and replacing them with cheap Government funds where possible. - 20 - sometimes contrary to legal covenants and frustrated project objectives, but it did not impede implementation and it certainly was not irrational. C. FIRA 7 53. From FIRA 7's effectiveness (17 November 1980) to December 1983, when it was transformed by the Special Assistance Program (see Para. 62 below), its implementation was not smooth. Some of the impediments stemmed from IBRD's changing its lending rate in the face of world inflation; from IBRD's switch to its Pooling of Currencies arrangement; from the Borrower's attempt to submit for reimbursement agro-industrial investments made before the Loan Documents were signed; from FIRA's failure to observe the dated covenant for preparing an operational manual; and, later on, from the disbursement consequences of Mexico's devaluations. But the biggest impe- diments, as well as the most serious effects on Mexico's rural credit sys- tem, stemmed from Mexico's rapid inflation, how the Borrower wished to deal (or in certain instances, not to deal) with this inflation, and how IBRD reacted. Discussions were heavily weighted towards interest rates, but also included adjustments in categories of farmer loan recipients and the increasingly-short-term nature of credit as inflation blossomed. 54. There were 50 months between FIRA 7's Effectiveness and its clos- ing at the end of 1984. All disbursements actually occurred in the 39 months from July 1981 through September 1984. Mexican inflation, which had grown quite serious during the FIRA 5 and 6 periods, accelerated and was particularly disruptive during this period, especially in 1982 and 1983. Average annual increases in the consumer price index (CPI) and in the much- lower average cost of funds (ACF) were as follows: year CPI ACF 1980 26% 21% 1981 28? 29Z 1 I 1982 59? 40? I I FIRA 7 Disbursement Period 1983 102? 57% 1 I 1984 66% 51? 1 1985 58? 562 1986 86? 81? 1987 132X 95% During 26 of the 39 months--from 15 July 1981 to 15 January 1982, from 1 August 1982 to 24 November 1982, and from 5 August 1983 to 30 November 1983 --the Borrower recognized that agricultural credit interest rates were not in accord with agreements with IBRD, decided not to bring them into compli- ance, and, as a consequence, agreed not to "commit" funds during this - 21 - period. i.e. not to submit for reimbursement loans made during these periods.271 55. IBRD's first supervision mission after Effectiveness observed that the 'significant increases" in the ACF from October 1980 through January 1981 'fully justify' the revision of FIRA's rates as per the side letter (supervision letter of 9 March 1981). The ACF had moved from 22.42 percent in October 1980 to 25.46 percent in January 1981. The Borrower holds that the trigger mechanism of the side letter was never activated (see Borro- ver's comment #8 on PCR 3.08 (attached) for instance). Apparently, the side letter agreed to consultations with IBRD in the event that the ACF increased by 3 percentage points or more within three months but did not prescribe anything in the event of non-agreement between Borrower and Lender. (By contrast, Ln. 1881 (Small & Medium Industries) prescribed increases in interest rates in these circumstances.) Whatever the legal niceties of what was agreed, it is quite clear that the Borrower strongly and consistently wished to keep agricultural credit rates below inflation rates, and that IBRD staff, particularly Projects' staff, strongly and consistently wished to get those rates to or above inflation rates. Howe- ver, when push came to shove, IBRD, particularly Programs' staff, accepted rates below inflation rates, regarded as contrary to the Borrower's under- takings in the Project Agreement, Schedule 2, C/8, as spelled out by the elusive side letter. These contradictory intentions of the two parties caused clashes throughout the implementation period and involved numerous and lengthy discussions, not only on Ln. 1891 but also on other loans, most notably Ln. 1881. Agreements invariably came unstuck and/or were disre- garded by the Borrower. 56. In February 1981, before the first post-Effectiveness supervision of FIRA 7, IBRD and NAFIN (Nacional Financiera, the Borrower) had agreed on an interim interest-rate increase under Small & Medium Industries (Ln. 1881), but Hacienda (the finance ministry) refused. Clearly, discussions under FIRA 7 stipulated by the side letter were not likely to result in agreement, in which case no interest-rate adjustment would be made until the required 15 July adjustment stipulated by the Loan Agreement. There were general discussions between NAFIN and IBRD, including two NAFIN visits to IBRD HQ in March and May. NAFIN requested no adjustments, interim or regular, during 1981. On 16 June 1981, the Borrower agreed not to submit disbursement requests for loans "committed" after 15 July (the date of the regular annual adjustment) until the Borrower should have reached agreement with IBRD on new rates. During this Iagreement-not-to-submitI period, which lasted to 15 January 1981, and long thereafter, there were numerous and sometimes pointed exchanges between IBRD and FIRA trying to ascertain exactly which loans FIRA had committed before the cut-off date. FIRA evi- dently had difficulty providing this information; IBRD staff were evidently skeptical as to its accuracy. Nevertheless, in the July-September quarter, IBRD disbursed US$62 million against pre-15-July commitments and in the October-December quarter, US$2 million. 27/ Loans made during the third of these three periods were ultimately submitted and accepted under the Special Assistance Plan (see Para. 62 below). - 22 - 57. In October 1981, Banco de Mexico had authorized agricultural lend- Ing at rates linked to the ACF for the first time, but only for "other producers'--commercial farmers representing the highest roughly 2 percent of rural Mexicans by income--and only for loans for crops other than the basic five (maize, beans, wheat, rice & sorghum) covered by a special pro- motional program. In Washington meetings on Christmas Eve 1981, the Borro- wer proposed higher rates281 On this basis, IBRD agreed, on 5 February, to resume disbursements as of commitments after 15 January. At that time, US$264 million remained to be disbursed, of which only US$25 million (9 percent) was in *other producer* categories. The other 91 percent was clearly to be lent at non-indexed, below-ACF rates. (ACF turned out to average 29 percent for the year.) In the event, the share of loans sub- mitted for IBRD disbursement that were below ACF turned out to be nearly 100 percent because commercial farmers managed to take almost all their loans for the five basic crops, at 28 percent, rather than at the higher ACF + 3 percent.29/ 58. Even with subsidized interest rates for PBI and MIP, loans to these categories were not being made at the anticipated rate. Because of increasing inflation and static farm-gate prices, farmers were not increas- ing their investments in farms. For them, interest rate subsidies were only a minor offset to the crippling "taxes' levied on the sector by its terms of trade.30/ Smaller, poorer farmers, moreover, had fewer opportunities to shift their resources elsewhere to take advantage of inflation--replacing them with FIRA-rediscounted loans. Consequently, in April 1981, Banco de Mexico proposed a redefinition of FIRA 7 categories to put higher-income farmers in lower categories, as well as a revision of interest rates for 15 July. Borrower and Lender did not come to an agreement on interest rates and the moratorium on submission of disbursement applications was re-instituted by the Borrower on commitments after 1 August. 59. This hiatus lasted until 24 November when interest rates 2 per- centage points higher than those proposed in April were agreed to. By this time, the ACF had risen to 46 percent, so that the 26 percent and 33 per- cent being charged to PBI and MIP were strongly negative, as was the 36 28/ 19 percent for the poorest 80 percent; 25 percent for the next 17 percent, and 28 percent for "other producers' for production of the basic five commodities or the ACF + 3 percent for them for other enterprises. 29! 'Over 90 percent' according to the report of the September/October 1983 supervision mission, Annex 6, Para. 18. 301 Parts of 1981 and 1982 are an exception to this pattern, mentioned here and elsewhere. In response to its first negative agricultural trade balance in 1981--the result of squeezing the agricultural sector- -Mexico substantially increased farm gate prices and introduced a range of input subsidies. The high price of oil and discovery that Mexico's reserves were much larger than previously realized made these changes easier. Farmers responded and agricultural production increased by 8.5 percent for one year. However, the oil-price bust of 1982 ended this brief interlude of Government *largesse* to the farm sector. - 23 - percent for *other producers" for the five basic crops. IBRD staff then seem to have abandoned efforts to achieve interim adjustments (perhaps a tacit acceptance of the Borrower's legal interpretation), but the agree- ment-not-to-submit arrangement was resumed when the Borrower failed to make the annual adjustment again on 15 July 1983; this disbursement gap lasted from 5 August to 30 November, by which time the "narrow" concerns of agri- cultural credit were being subsumed under the broader concerns of the Spe- cial Assistance Program (SAP, Programa de Asistencia Especial) and, later, the General Interest Rate Agreement (GIRA). (see below) 60. There were other implementation concerns. Disbursements were far behind projections for all but the highest-income category of borrowers, both because of the long periods of agreement-not-to-submit (acknowledged by the SPN reports) and because farmers were reluctant to invest in their farms even at subsidized interest rates (not noted in supervision reports). The Borrower tried to solve this problem by further requests to redefine beneficiary income categories and to re-allocate funds from lower-income borrowers to higher. An example is NAFIN's request of 20 July 1982, accepted by IBRD on 2 August, to redefine categories and to re-allocate US$26.5 million from PBI to commercial farmers. In so doing, one Project Officer complains (internal memo of 16 August 1982) that IBRD "regressed glaringly." 61. The discussion over income categories of borrowers became more heated during the October 1981 supervision mission. At that time, FIRA's computer sub-loan monitoring system, under development for some time, had begun producing readily-accessible printouts on sub-loans. For the first time, it became relatively easy to check on borrower categories and quali- fications. The FIRA records showed that many sub-borrowers did not qualify for the interest rates they were getting and/or were not contributing any share of the investment. Subsequent missions redoubled efforts to get FIRA to tighten management controls in order to catch farmers or banks who/that were trying to beat the system. Also, FIRA tried (without success) to get IBRD to agree to let participating banks reduce the contribution of their own funds (17 August 1981), and to accept IBRD financing of used machinery (accepted 17 August 1983), repairs, and short-term credit. In sum, the Borrower sought various ways to expedite disbursements and to adjust the legal agreements to suit changes in FIRA's pattern of business as inflation accelerated; IBRD staff generally held to the original objectives of the agreements: divert more credit to lower-income farmers; finance invest- ments; encourage participating banks to use more of their own resources; avoid disbursements where verification that borrower shopped for and got the best deal is impossible (used machinery & repairs). Where IBRD staff retreated from these positions, they did so reluctantly. 62. The Special Assistance Program changed this. In SAP, recognizing Mexico's severe structural adjustment problems and the efforts being made to solve them, IBRD tried to come to Mexico's assistance without a structural-adjustment loan but rather by modifying existing loans. By letter of 19 July 1983, NAPIN requested inclusion of FIRA 7 in SAP. IBRD kept the proposed SAP modifications in abeyance as a lever for getting the adjustment in interest rates that had been due in July. On 30 November 1983, agreement was reached on new interest rates for farmers ranging from 27 percent to 53 percent. (At that time, the ACF was 55 percent.) With FIRA 7's inclusion in SAP: - 24 - - IBRD disbursed against 70 percent of the sub-loans rediscounted by FIRA instead of 40 percent; - disbursements were permitted against short-term working capital; - another reallocation--US$20 million--was made from PBI to commer- cial farmers. Although these changes were not legally formalized through amendment of the Loan Agreement until 28 August 1984, inclusion of FIRA 7 in SAP represented a major IBRD concession from some important principles agreed earlier in recognition of difficult circumstances. It speeded disbursements too. 63. This last interest-rate increase under FIRA 7 ranged from 1.5 percent to 9.5 percent for the four groups of beneficiaries, averaging 4.7 percent unweighted but only 2.5 percent when weighted by the current lend- ing patterns. IBRD staff was well aware that these changes did not comply with the interest rate covenant but argued the changes were "moving towards positive real interest rates' and that the increases needed to produce compliance, 16.8 percent, 6would be relatively large and impractical to introduce." (internal memo of 15 December 1983, Paras. 1,5) By this time, Lender and Borrower were looking beyond FIrA 7 to GIRA which, however, did not affect FIRA 7 because it disbursed ful'.y before GIRA came into effect. D. FIRA 8 64. FIRA 8 basically took its final form in December 1983 (see Para. 32 above) but was not declared effective until December 1984, six months after Board approval in the last week of the fiscal year. Its implementa- tion was held up by problems with GIRA. 65. GIRA was negotiated in the context of FIRA 7 but never applied to it. It was approved by IBRD's Loan Committee on April 10, 1983 and signed on August 7, 1984. FIRA 8, which had been held hostage, was signed the next day. GIRA governed, or was meant to govern, interest rates on all IBRD credit operations in Mexico, including FIRA 8 and 8A. GIRA interest rates were fixed as a percentage of the ACF, which was taken as a proxy for inflation.31/ GIRA provided a timetable for the increase of lending rates to or above ACF in semi-annual adjustments, except for low-income farmers, for whom rates were to attain 60 percent of ACF by October 1, 1984 and 80 percent by January 1, 1987. 31/ As the figures in Para. 54 above show, from 1980 through 1987, ACF averaged 80 percent of the inflation rate, but the average obscures variation from 3 percent over to 44 percent under. As the President's Report No. 4646-HE of February 22, 1988 for IBRD's later Agricultural Sector Loan (2918-ME) said (Para 80(b)): 'The ACF was chosen as a benchmark rate because it is the closest proxy for an index of a market borrowing rate that is readily available in the heavily regulated environment in Mexico. While the Government sets the rates that determine the ACF, its freedom to manipulate the rate is tightly constrained by the ease with which capital can be taken out of the country.* - 25 - 66. Just as FIRA 8's signing had been held up by Mexican second- thoughts about GIRA, so was its Effectiveness. After missing a GIRA-sche- duled interest-rate adjustment in October, the Borrower got into compliance with GIRA in November and FIRA 8 became Effective on December 3, 1984. The compliance was brief. For funds committed during the first three months of 1985, when rates again did not comply with GIRA, there was another agree- ment not to submit disbursement requests. In fact, throughout the imple- mentation of FIRA 8 and of its successor, the dominant note in files is over various failures to comply with GIRA. Other concerns seem almost insignificant. 67. Two weeks after Effectiveness, the IBRD Project Officer noted that the Borrower was being very efficient about disbursement requests. Requests for about US$100 million had been or were being submitted, and all of these were at the old, pre-GIRA rates.321 What this meant was reported by the March 1985 supervision mission. FIRA's average lending rate in 1984 had been 25.4 percent, whereas inflation had been 66 percent (the report said "about 60'). So the confident conclusion of the August 1984 supervi- sion mission-- 'Because of steps taken in FIRA 8 ... FIRA is no longer being decapitalizedw--had to be revised. Between December 3, 1984 and March 31 1985 there were only 29 calendar days in December when the agree- ment not to submit disbursement requests was not in effect. Nevertheless, during that period, under Loan 2454, IBRD disbursed US$54.8 million in December, US$99.9 million in January, US$5.4 million in February, and US$42.3 million in March, virtually all for loans rediscounted at manifest- ly very negative real interest rates. 68. A striking characteristic of almost all the supervision reports for FIRA 8 is how sanguine they are. FIRA is reported to be profitable (it was). FIRA's capital position is reported to have improved (because the Borrower was giving it the proceeds of the loan and other grants). The issue of what all this was costing the Government of Mexico was not men- tioned. There were amendments to the loan agreements to increase the amount of retroactive financing permitted (at lower-then-GIRA interest rates), from investment credit to harvest credit, and later, in FIRA 8A, to decrease the share of their own resources that borrowers and participating banks had to mobilize. The notion that the Mexican farm credit system should be moving towards self-finance and self-sustainability, with banks and farmers encouraged to save and mobilize their own resources, a notion that received IBRD consideration in earlier projects, is quite absent.331 69. Disbursement of the last third of FIRA 8's loan dragged out over ten more months of squabbles over compliance with GIRA, over submission of 32/ This December 17, 1984 memo incorrectly reports that FIRA 8 became effective on August 8, 1984. The US$100 million consisted of US$30 million for the special account, an equal amount of retroactive financing, and US$40 million of sub-loans rediscounted for the August 8-November period. 33/ IBRD staff concerned with the project at that time have emphasized that all of the above was known by all and that 'the compromises on interest rates were not at the level of supervision staff," and again that 'the interest rate debate cannot be dissociated from macro-policy.* - 26 - disbursement requests for loans made at non-complying rates, and/or in non- complying periods, or to sub-borrowers who manifestly did not belong in the category where their loans had been put. During this time, FIRA 8A was rushed in to maintain the disbursements. S. PIRA SA 70. Loan 2610 for FIRA SA became Effective December 17th 1985. Of its US$180 million, US$70 million disbursed during December and January, also last two disbursement months for FIRA 8. Thereafter a number of problems caught up with rapid disbursement, which, by this time, had become almost synonymous with implementation. Disbursements, which had been expected to take a year or less, stretched out over 25 months. 71. On June 24, 1986, six months into implementation, the Project Officer summed up FIRA 8A's implementation when handing over to his succes- sor* 'The single largest problem relates to GIRA and its implementation., Another US$23 million had disbursed by then, but there had been a flurry of misunderstandings and/or disagreements between the Borrower and IBRD over interest rates to farmers, banks' spreads, and who was eligible for which interest rate. If disbursement on schedule is equated with good implemen- tation, then things got worse after June 1986. There were no disbursements during seven of the next eight months, partly because FIRA 8A funds were displaced by IADB funds which came on better terms and, perhaps, with fewer hassles about conforming to a lot of covenants,341 partly because of interest-rate disputes, and partly because GIRA was starting to work. 72. What appeared to IBRD supervisions at this time as problems appear to the Audit to be signs of success. From the supervision mission of March 1985 on, there are definite signs that the sector no longer had an insa- tiable demand for FIRA-rediscounted loans. It reported (June 11, 1986, Annex 2, Para. 5, p. 2), mas at end April 1986 FIRA's actual lending," which for years had exceeded projections, "declined below the program.' Besides the drought, the mission and FIRA attribute this to 'the initial effect of higher interest rates and input costs, without the corresponding increase in selling prices; and ... an inadequate margin on discounts to the participating banks for subloans to OPs,9 i.e. commercial farmers. 'Also, the demand for short-term credit is now mush stronger than for medium- and long-term credit (70 percent short-term compared to about 60 percent targated).R 73. To some extent, the fall-off in demand for loans was a consequence of Government discrimination against agriculture. However, it was also evidence that the policy measures IBRD technical staff had been fighting for over two decades were starting to bite. The fall-off in demand was precisely in that part of FIRA's lending program where interest rates were least negative. Rates to commercial farmers at the time were at the ACF for commercial farmers for production of maize, wheat, beans, rice or sor- ghum, at ACF+2 when they produced anything else. Average ACF for 1986 34/ IADB disbursed a total of US$212.4 million for Mexican agricultural credit in 1986, with US$161.0 million for FIRA, US$160.6 million of it coming from Loan 197-IC (Mex), most of it late in the year. (see Table 3 and Figure 2). - 27 - turned out to be 81 percent, modestly below the CPI increase of 86 percent and further below FIRA's real cost of funds but still high enough to give borrowers pause. A further incentive to prudent borrowing was the require- ment that commercial farmers or participating banks put up half the invest- ment, encouraging more resource mobilization from farmers and banks, and less from Government and IBRD. Moreover, the bankers' spread of 2 percent to commercial farmers was less interesting to them than the 6.5 percent spread to PB1. For the first time, FIRA's rediscounts of loans to PBI were running ahead of those to commercial farmers, not vice versa. Rates to PBI were still more than 40 percent below the inflation rate, but even these rates were starting to ration credit demand a bit, screening out weaker investment prospects. These developments were, after all, important objec- tives of IBRD agricultural lending policy. 74. Nevertheless, the May 1986 supervision mission supported the Bor- rower's request for amendments to reduce the share of commercial farmers' sub-projects that farmers and their banks had to finance from 50 percent to 20 percent, to re-allocate US$25 million from investment to short-term credit, and to add to the list of five basic products for which commercial farmers received additional credit subsidies soya, safflower, sunflower, sesame, coconut, vegetable oils, milk, eggs, poultry, goat, sheep and beef. These modifications, and others later in implementation, such as enlarging participating banks' spread on loans to commercial farmers, helped maintain disbursements but did not foster IBRD's earlier major sectoral objectives, such as promoting resource mobilization by banks and the farmers them- selves, nor encouraging commercial banks to find creditworthy PBI custo- mers, nor encouraging investments in an environment that otherwise discou- raged them, nor even the objective of curtailing the still-large interest rate subsidies with their large fiscal cost and distorting effect on secto- ral investments. 75. From May 1986 to Loan 2610's Closing in October 1987, there were flurries of disagreements over interest rate adjustments and unilateral changes in participating-bank margins and definitions of borrower catego- ries and 'basic crops' favored by additional interest-rate subsidies. The common supervision theme was the weakening demand for FIRA-rediscounted agricultural credit, or what the Audit regards as the moderating demand for agricultural credit in response to serious inflation--the average annual CPI increase was 132 percent in 1987--and higher though still very negative real interest rates. In 1987, commercial farmers were paying 92-97 percent interest for FIRA-rediscounted long-term loans and PBI 67-72 percent. The annual average ACF was 95 percent. PBI rates had been scheduled to rise to 80 percent of this in January 1987 by the original GIRA, but IBRD agreed to a request to allow this adjustment to be postponed to 70 percent of ACF from January to April 1987 and to 75 percent from May to October. Again IBRD agreed to relax the covenants and postpone the date when interest rates, which were forever "moving towards positive real rates in accordance with an agreed timetable," would actually get there. 76. By the time of the May 1986 mission and thereafter, IBRD was focussing beyond FIRA 8A to the reforms that would be introduced in FIRA 9. The MHay 1986 mission already notified Government that its proposals for FIRA 9 would have to including halving of the interest rate subsidy to PBI that would remain once GIRA was fully implemented (Aide Memoire, Para. 2). Further hopes for sectoral reform were pinned on a study which Hacienda - 28 - (the ministry of finance) was to have carried out according to terms of reference to be agreed with IBRD and IADB. After Loan 2610 finished disbursing in October 1987 and FIRA 8A gave way to FIRA 9, in March 1988, inflation started waning and the ACF, which had generally lagged inflation as it rose, also lagged it as it fell. Under these circumstances, where application of GIRA would have produced a very positive real interest rate, the Borrower proposed alternatives to GIRA and IBRD effectively agreed to GIRA's non-application. But these are post-scripts to the audit of FIRA 5 through FIRA 8A that may be taken up in future when FIRA 9 is audited. IV. IMPACT OF LOANS A. Production Impact 77. The PCRs confirm the finding of supervision reports for FIRA 5, 6, and 7 in asserting that loans rediscounted under these projects generally financed sensible investments which generated increased production and farm income. IBRD reports usually enter caveats regarding the randomness and objectivity of FIRA samples. These seem to have been particularly suspect in the years of FIRA 5, 6 & 7. Moreover, it is recognized that the very best sampling of a very large number of farms has problems. Nevertheless, there is no reason to doubt that FIRA sampling results are generally cor- rect. The reasonableness of FIRA-financed investments has been confirmed by visits of IBRD supervision missions and by the Audit. And this conclu- sion also stands to reason. Firstly, FIRA has, over the years, been able to employ the cream of Mexican agricultural and livestock graduates; its technicians have high esprit de corps and generally take their job serious- ly. That alone has tended to weed out bad investments. Secondly, even though negative interest rates lessen borrowers' incentives to invest care- fully, they do not remove it altogether. The Audit's small sample of farmer-borrowers contained farmers who might have had to be more careful and to work more quickly without the gift of highly-negative real interest rates, but all were making basically sensible, technically sound invest- ments. 78. Also, the US$1130 million of IBRD lending under these five projects undoubtedly permitted the formal agricultural credit system to expand more rapidly than it would have in the absence of the loans. IBRD's participation clearly did not, however, bring more local resources into agricultural Investment. Because of the negative conditions in which Mexi- can agriculture had to operate during this period, farmers consistently avoided Investing their own funds, even to the modest proportions stipulat- ed in the legal agreements. Wherever possible, commercial banks substitut- ed FIRA funds for their own. Once borrowed and contributed to FIRA's capi- tal, IBRD funds are Government of Mexico funds. Still, with the SAP of 1983, IBRD and the Borrower agreed to substitute IBRD funds for those of Government. So IBRD lending permitted the more rapid expansion of the formal agricultural credit system, but it did not institute a weaning process whereby local savings and investment would replace dependence on foreign borrowing. Rather the reverse. 79. Expansion of formal agricultural credit does not translate automa- tically into expansion of Investment in agriculture. When the conditions in which the sector has to operate discourage investment, as was certainly - 29 - true for most of Mexican agriculture for most of the period under conside- ration351 , then some of the credit expansion will compensate for sectoral disinvestment by farmers and by banks. To actually quantify this phenome- non accurately would be horrendously difficult, but there is enough anec- dotal evidence to convince this Audit that there was quite a lot of such Odisplacement' going on in Mexico during PIRA 5, 6, 7. 8 and 8A. Spot checks of sub-loans, both through FIRA's record-keeping system and by field checks, turn up hints, such as farmers' not putting any of their own money into investments, ranchers' using loans to restock their land (presumably after selling stock and investing the proceeds elsewhere), etc. In such an environment, which is in no way the fault of the agricultural credit sys- tem, what government does to foster agricultural investment through sub- sidized interest rates is a pittance compared to the discouraging effects of an unstable economic environment--so critical for a long-term proposi- tion like agriculture--or of administered farm-gate prices that don't keep pace with inflation and/or don't reflect the opportunity value of farm produce in terms of border prices. 80. The experiences of FIRA 7 when inflation reached its peak in 198213 and of FIRA 8A when it peaked again in 1987 are instructive. All farmers tended to shift from medium- and long-term borrowing to short-term. For FIRA 7 this was especially so for the PBI; commercial farmers, with their greater resources and mobility, were more likely to maintain some investments, presumably using subsidized FIRA resources in place of their own. For FIRA 8A offering less subsidy to commercial farmers, there is less evidence of such maintenance. These rational farmer responses to the situation answer the question posed by an IBRD Programs' officer in an internal memo back in the period of FIRA 4 but even more true in 1976-87: *...agricultural production has stagnated and the agricultural sector is said to have decapitalized in spite of the considerable increase in institutional term credit. Have financing sources been substituted or has capital productivity declined or is there any other explanation?" 81. Given the above, there are serious problems knowing how much of the Investment the projects financed is a net Increment, a difference between with project and without project. That invalidates economic rates of return based on investment models. In other words, given fungibility and the conditions Mexican farmers faced during these projects, the Audit puts little faith in the economic rates of return as calculated in the PCRs. IBRD's more recent approach, recognizing the pitfalls of ERR calcu- lations based on models in farm credit projects and concentrating on insti- tutional development, seems justified. 82. Beyond the conundrum on fungibility and substitution, a note is in order about FIRA's technical role in agriculture, as distinct from banking. In addition to being a rediscounting window, FIRA is an agricultural exten- sion service and an agricultural research service. These are functions normally carried on by agriculture ministries; in Mexico, the ministry's extension service has never been worth much as a change agent, while the research arm (then INIA, now INIPAP), once relatively effective, has been 351 But see footnote 27 above. - 30 - increasingly hit by retrenchment to the point where virtually its entire budget is consumed by salaries. FIRA has stepped into the gap with its technical assistance, training programs, and demonstration centers, which actually do applied research.36/ Indeed, FIRA's cost relative to its redis- counted portfolio is much too high to be justified based on its banking functions (cf. Paras. 85, 135 & 140 below). Its technical services, howe- ver, fill a serious gap and contribute to output expansion. Quantifying their productive impact is beyond the scope of this Audit. Moreover, our knowledge of their costs is incomplete and imprecise. While we know that 'technical assistance" and guarantees' costs rose from 2.8 percent of the average value of the portfolio in 1980 to 4.2 percent in 1984, FIRA was unable or unwilling to furnish information for other years or to break down its expenditures further at the request of this Audit. Also, from the national point of view, it is an unfortunate effect of having FIRA do much of Government's agricultural research, training and extension that these technical services tend to get focused on the rural minority who borrow long-term from formal institutions. These are not a random cross-section and tend not to be the poorest of the poor. B. Institutional Development 1. FIRA 83. The PCR9, like the SARs, stress the institutional strengthening of FIRA. It is useful to look at this strengthening financially, technically, and managerially. 84. Financially, FIRA comes out of the period of these loans as strong as it went in, a victory of sorts for an institution lending at radically negative real interest rates in a period of financial confusion. The main- tenance of FIRA's strength, however, stems from two facts: firstly, its rediscounts to participating banks are recovered automatically so that it can have no arrears or bad debts; and, secondly, Government compensates for its real losses (as distinguished from its nominal profits) by making up for capital erosion due to lending at negative real interest rates. IBRD agricultural credit loans were donated to FIRA as capital.371 In some ins- tances, proceeds of IBRD agricultural credit loans have not been enough to maintain the real purchasing power of FIRA's capital and IBRD has insisted that the Mexican Government provide FIRA with additional capital, which it has done through the central bank at highly negative interest rates. In other words, FIRA has been very far from maintaining the value of its capi- tal from continuing operations. What is relevant here, and this is a ques- tion generally avoided by IBRD appraisal reports when they declare FIRA financially sound, is how much FPA's failure to self-finance its opera- tions is costing the fisc (see Paras. 101-103 below). 36/ To this date, FIRA has carefully allocated its research resources to those areas where INIA/INIFAP is unable to do its job. Thus, there is no FIRA demonstration center in the Northwest, where CIANO in Ciudad Obregon, funded substantially by farmers' contributions, is a highly effective center for applied research. 37/ Except for FIRA 7, which is carried as a loan from Government. - 31 - 85. What about FIRA's administrative efficiency? Its offices in the capital (mostly now being given up as most departments are transferred to Morelia) were sometimes the object of envious remarke they were not aus- tere, but they were not out of line compared to other Mexican banks. Operating costs as a percent of average loan portfolio were 3 to 4 percent in the years for which we have figures. This is too high for a rediscount bank. All but about 0.5 percent, however, is the cost of FIRA technical assistance and guarantees for PBI loans. Moreover, that half percent includes the considerable cost of trying to direct credit to PBI. So FIRA's tasks include a great deal more than rediscounting. In the perspec- tive of those various tasks, operating costs are not unreasonable. 86. Technically, FIRA has become perhaps the strongest Government agricultural organization in Mexico. During most of the period,38! FIRA offered more attractive terms of employment and career possibilities than other Government agencies. It got and generally retained the best agricul- ture graduates from the Chapingo National Post Graduate Agriculture School. There can be no question that its technicians affected for the better the farm and ranch investments that banks and borrowing farmers made, and, particularly, that they raised the agricultural consciousness of bankers. Moreover, through the policy of absorbing the technical-assistance costs of preparing and appraising loans to PBI, FIRA spawned a host of independent agricultural consultants. A Mexican farmer who can afford to pay for advice, or a PBI borrowing from a commercial bank which intends to redis- count the loan with FIRA can get advice from young entrepreneurs, usually agriculture or veterinary graduates. Some of these technicians are employ- ed directly by FIRA but, increasingly, they are independent consultants, approved by FIRA and hired by farmer-borrowers with FIRA funds. FIRA's technicians also ran FEGA, FIRA's guarantee scheme for PBI which was a free service during the period under audit but for which participating banks now Pay. 87. FIRA's technical output has not been limited to helping make loans. FIRA says that its many ademonstration centers' are establishing a link between research and farmers. In fact, they are doing more. They are doing adaptive research. In several instances that this Audit saw, they have done an excellent job of working out new, cost-cutting technologies well-suited to farmers. Two examples are lower-cost dairying through pasture improvement and low-tillage grain production to save water (and labor and machine costs too). Since 1981, as Government's austerity drive has cut the national agricultural research budget to the point where INIFAP can only pay salaries, FIRA's agricultural research, and that of CIANO (INIFAP's farmer-financed Cd. Obregon branch) are practically the only national sources of agricultural research. Similarly, FIRA's training of farmers (and commercial bank staff) goes far beyond banking and into tech- nical agriculture. It is extension by another name. Since it is closely linked to farmers' wants and to their financing, it has to be pretty effec- tive extension which, traditionally, most of the Agriculture Ministry's regular extension is not. 38/ With the possible exception of ca. 1981, the oil-boom PLANAT period, when major efforts were made to strengthen the research and extension services. - 32 - 88. Beyond these general statements, it is futile to try to quantify the results of FIRA's technical development. However, the appropriateness of having an ,.gricultural-credit trust fund do research and extensions what kinds of research and extension then get done and who benefits from them-- are discussed below under Issues. 89. Managerially, FIRA got a lot of attention from IBRD, especially in connection with Ln. 1891 (FIRA 7). Great hopes were placed in axiomatiza- tiont in getting agreed procedures for loan screening and approval into a manual; and in better internal communication and data handlings record- keeping on loans, and M&E that would keep track of the *ultimate impact* of FIRA (and hence of IBRD) operations. After much weeping and gnashing of teeth, FIRA generated and now even uses an operational manual; this has, no doubt, rationalized its operations. FIRA's computerized records' system is working; despite quite decentralized operations, central management, now in Morelia, can quickly call up all the accounting details on any loan. And, after many foreign consultants, FIRA has a working MS system; information from the reasonably accurate and rather unusual sampling system is fed into the Morelia computer and periodic reports are generated with the push of a button. 90. Alas, the answers the automated reporting system provides do not begin to exploit the wealth of data gathered. FIRA's management is both proud of this M&E system and frustrated with it. One of the Borrower's comments (see attachment to FIRA 7 PCR, #9) relates specifically to requests for IBRD's help in making the M&E system more responsive and IBRD's supervision missions' failure to provide that help."9/ The system does not tell us much about the 'ultimate impact" of FIRA's program, but a) to do that would probably be impracticably difficult anyhow given the fungibility of assets, and b) FIRA's management has more proximate and more pressing concerns. More seriously for management, the M&E system's pre- programmed reports do not address many of these more immediate concerns. Consequently, management has now assigned the working system a low prio- rity. 91. All of the above, however, failed to address what is, in the Audit's opinion, FIRA's central management weakness--FIRA is a rediscount bank all of whose senior managers are agricultural engineers. These mana- gers are in their element when discussing low-tillage agriculture or vil- lage dairy farms or fruit and vegetable farming, or when discussing farming with farmers. During the Audit, when conversation turned to banking topics, such as appropriate participating-bank margins, or to management topics, such as operating costs for FIRA's rediscounting and for its *tech- nical assistance" as a percentage of outstanding loans, interest waned. Answers grew vague. %,ne of the questions FIRA management wants its M&E system to answer, for Amstance, are banking questions. After ten IBRD loans, including the five under audit, FIRA is excellently staffed for 39/ IBRD staff replies that money for this purpose was provided in the loans and not used, and that expert guidance was provided. This is so. The Audit believes that FIRA management always perceived M&E as an IBRD imposition and never saw the utility of the H part of E in running FIRA. This probably stems, in part, from IBRD preoccupation with the 2 part, the elusive desire to know "ultimate impact'. - 33 - technical agricultural assistance, but not for technical rediscount banking. 2. Commercial Banks 92. As a result of the projects being audited, commercial banks expanded their rural network and their agricultural lending. They were able to do so in an inflationary period in part thanks to lending at nega- tive real interest, which were attractive to borrowers, without loss to themselves. The resulting losses were absorbed by Government through FIRA. Moreover, repayment discipline was good, perhaps in expectation of repeater loans at similar interest rates. For the period, bad debts and serious defaults were less than 2 percent; it appears to the Audit that refinan- cings were not used to disguise problem loans to any significant extent.401 So commercial-bank lending to agriculture, already impressive in 1976, strengthened. 93. But it did not strengthen in terms of capital mobilization. Under the projects, FIRA, with the help of Government, IBRD, IADB and other money, clearly substituted for commercial bank resources, leaving commer- cial bank farm lending more dependent on public resource mobilization and less on its own at the end of the period than it was at the beginning. 94. In the process, did the commercial banks get a sweetheart deal from FIRA as is sometimes suggested? It is true that FIRA technical assis- tance and guarantees for loans to PBI have been and, in some instances, still are gifts to the commercial banks. These and fixed margins provided income with little risk. But the Audit is not convinced that these subsi- dies have been unreasonable, particularly if they served the social purpose for which they were intended: to induce commercial banks to lend to a social stratum to which it did not previously lead, the PBI. 95. The question about results that inevitably arises, therefore, iss is FIRA-rediscounted targeting of lending by commercial banks to PBI a 'joke', as is often alleged? The Audit believes that it is not, but still wisht-s to reserve judgement. When the Borrower and IBRD first tried to push the commercial banks into lending to PBI in FIRA 4 and 5, the amount of such lending was negligible, and it did not grow very quickly. By 1989, commercial banks still have trouble filling their FIRA quotes for PBI (there is no shortage of commercial agricultural borrowers), but the volume is considerable and significant. And quite a few PBI have succeeded and "graduated* to become 'commercial producers.' More of FIRA's PBI lending goes through commercial banks and less through BANRURAL now than during FIRA 5. 96. There should be no confusion about who the PBI are; they are not the poorest 40 percent. According to agreements under all the projects, they are a stratum just below the top 3 percent of the rural population to 40/ The default-bad debt risk was shared by the rediscounting banks and FEGA (a constituent part of FIRA). Until recent years and during the period under review, the PEGA guarantee was provided free. - 34 - whom commercial banks normally lend.411 And, there are numerous allegations of the top 3 percent managing to pass for PBI by conniving with banks or by using prestanombres (borrowed names) in order to take advantage of FIRA interest rate subsidies. Some such fraud no doubt happens, but the Audit is convinced that FIRA loan targeting to PB1 is not a joke. As one FIRA Director told the Audit, FIRA's only real purpose is to channel loans to PBI; all of its staffing and personnel transfer system is designed to see that funds are channeled to PBI. Given the circumstances in which it works (in which richer borrowers have incentives to cheat and to suborn FIRA staff), it does the best job that can be expected, or better. 97. Supposing that the Audit is correct, the test of the success of Government policy of inducing commercial banks to lend to a lower social stratum would be to see whether they continue after subsidies are with- drawn. 3. BANRURAL 98. Let us recall that, before FIRA 5 began, indeed before FIRA 1 began, BANRURAL's predecessors were unsound. By the time IBRD was trying to elicit reform of BANRURAL in return for accepting FIRA rediscount of BANRURAL loans under FIRA 5, BANRURAL's losses--then of the order of US$500 million/year--were recognized as "a national macro-economic and fiscal problem far transcending the question of Bank financed operations.042/ While IBRD staff recognized the magnitude of the problem, a review of files shows the consistent expectation that IBRD could, through the FIRA projects, influence or get FIRA to influence BANRURAL for the better. A major conclusion of the FIRA 7 PCR (Para. 12 in the Evaluation Summary) is that OFIRA failed to exert a positive influence on BANRURAL as expected.* In order to achieve this, IBRD had the carrot of allowing FIRA rediscount of BANRURAL loans if the Borrower got BANRURAL to reform. What reforms IBRD wanted was specified in a letter of 17 September 1976. 99. In the event, IBRD needed BANRURAL to disburse FIRA PB1 funds more than BANRURAL needed FIRA funds. Consequently, IBRD gave away its carrot 41/ As mentioned earlier, criteria changed several times over the period in question. When the "middle income producer' category was abolished in July 1986 and added to *other producers' (Ocommercial producers' in this Audit), that category grew to about 20% of the rural population and PBI was reduced to about 80 percent. 42/ See Para. 46 above. - 35 - in November 1976 and October 1977 431 without getting any meaningful reform, as careful analysis of BANRURAL by successive supervision missions confirms.44! 100. In commenting on the FIRA 7 PCR, the Borrower has observed (PCR, Attachment 1, Comment 7) that IBRD's expectations of IRA influence on BANRURAL were not related to any legal obligation on the part of the Bor- rower. The Audit agrees. Once IBRD accepted BANRURAL's eligibility for FIRA rediscounts under Loan Agreement, Schedule 1, 2(d) and Project Agree- ment, Section 2.06, IBRD had no further legal leg to stand on. 4. The Fisc 101. The impact of the projects on the fisc was simples the public treasury lost huge amounts of money, or, more elegantly, saw the real value of its investments in its agricultural-lending trust fund, FIRA, decline sharply. Government covered those declines by putting into FIRA both more of its own money and money it borrowed from abroad from IBRD. IADB, Citi- bank and Bank of Montreal and which debts it has to service with hard cur- rency. The losses in question do not stem from farmers' failure to repay loans as they do in the case of BANRURAL; credit discipline within the FIRA system was excellent. Since FEGA, the FIRA credit guarantee scheme, was free to participating banks to encourage them to lend to PBI until 1988, the fisc did lose money on guarantees; however, these losses were insigni- ficant compared to the massive recent losses of ANAGSA because guarantees were paid on less than one percent of loans covered and because the opera- 431 See Paras. 45-8 above. 44/ Recent post-FIRA-7 IBRD studies show that BANRURAL continues to be 8a national macro-economic and fiscal problem far transcending the question of Bank financed operations.0 Over-manning and corruption are still common; administrative expenses are still a very high percentage of portfolio; bad debts and serious defaults are still very high. BANRURAL loan recovery has apparently improved with the growth of 'indirect recoveries.' These are recoveries from ANAGSA, the Government insurance scheme under which borrowers from BANRURAL must insure their crops. But this is only a paper transfer of public-sector losses from BANRURAL to ANAGSA (except that ANAGSA insurance involves real costs to farmers, since it insures the loan, not the crop, and since they must pay for it, both to ANAGSA and, allegedly, to ANAGSA's field agents). ANAGSA's records and field checks indicate that ANAGSA- insured farmers incur very much higher crop losses than farmers guaranteed by FEGA and others in the same district, and that ANAGSA provides crop insurance to large areas that are not farmed. IBRD staff estimate the fiscal cost of the BANRURAL-ANAGSA complex at three times the 1977 estimate. In 1990, during the processing of IBRD's 11th agricultural credit loan to Mexico, Government abolished ANAGSA. - 36 - tion was relatively efficient.451 Virtually all of the fiscal losses stemmed from Government policy of not allowing FIRA to cover itself against the risk of inflation. 102. There is ample evidence that the Borrower incurred these losses consciously and deliberately. It is the burden of the Borrower's principal comments on the FIRA 7 PCR (Attachment I, General Comment #1 & Specific Comment #1) that these negative real interest rates were part and parcel of Government's overall policy. This policy was pursued in the face of persistent and consistent IBRD urgings to the contrary. Disagreements within IBRD were not over whether interest rates should be really positive but over how much divergence from that ideal to accept and what to do when the Borrower did not make prescribed adjustments. In this regard, IBRD Programs' staff were clearly more tolerant of the Borrower's departure from IBRD's ideal and souciant of maintaining the flow of disbursement funds than were the generally-harder-line Project's staff. 103. In following the policy it did, Government effected a major trans- fer from the fisc (and whoever pays for it) to farmers lucky or clever enough to get FIRA-rediscounted loans when inflation was soaring. The Audit estimates that this transfer was US$1,150 million for the ten calen- dar years 1977 through 1986.461 Table 4 gives details. The losses, in millions of US$ are summarized below for the years for which figures are available: 45/ The Audit was unable to obtain budget data that would have permitted quantification of the net cost of FEGA during the period, or of its much-less-important analog for non-PBI borrowers, FONDO. Total FEGA cost in the most recent year available, well after the closing of FIRA 7, is about 5.5 percent of the amount guaranteed, of which only about 25 percent was payment of guarantees and about 2/3 was "technical assistance* which, of course, renders services well beyond the normal risk underwriting and management of a guarantee scheme. 46! These figures are not for FIRA as a whole but for FEFA, that is, the long-term lending part of FIRA excluding short-term operations (FONDO) and guarantee operations (FEGA). They are in current US Dollars converted from Mexican Pesos at the official exchange rates, which were a pretty good reflection of equilibrium during this period. FEFA's accumulated net income for the period and its capital increases (mainly proceeds of IBRD and other foreign borrowing) during the period are added to its closing capital on 31 December 1976 and its closing capital on 31 December 1986 is subtracted. End-of-year balances are converted to US Dollars at end-of-year exchange rates; accumulated net income and capital increases are converted at the average annual exchange rates. Figures are from FEFA annual audited financial statements. What we know about corresponding FONDO losses are reported in Table 5. - 37 - FEFA FONDO 1977 - 28.9 na 1978 - 1.5 na 1979 - 9.5 na 1980 - 26.1 na 1981 - 74.4 na 1982 -462.6 - 2.9 1983 - 65.8 -21.8 1984 - 53.7 >48.847/ 1985 -177.5 na 1986 -249.7 na Losses were highest when inflation was accelerating and, by Government policy, FIRA was not protecting itself adequately, notably in 1982 during implementation of FIRA 7 and in 1985 and 1986 (and probably 1987 for which figures are not available) during implementation of FIRA $A. As Table 4 shows, from 1981 on, Government gave FEEA's enough new capital to make up its Peso losses so that its real losses are not apparent until these are translated into US Dollars, here used as a proxy for constant Pesos. 104. Immense though these transfers were, it should be remembered that they are small compared to transfers being made at the same time through the BANRURALIANAGSA rural credit complex. BANRURAL is bigger than FIRA; its interest rates were not less negative; and, while the FIRA system's customers were repaying in depreciated Pesos, a substantial portion of BUNRURAL's borrowers did not repay at all. Moreover, the technical quality of the BANRURALIANAGSA transfers in terms of investments and production was undoubtedly much lower than FIRA's technical quality. 105. The social merits and demerits of this transfer, and how they may relate to savings, investment and growth, will be treated under Issues and Findings below. V. BORROER AND IBRD PERORMANCB 106. On the Mexican side, we have seen that Government cost itself a great deal of money as FIRA frittered away the real value of the money Government gave it by lending at negative real interest rates. We have the Borrower's word that this policy of transfer to borrowing farmers was deli- 471 Information on FONDO's net income and additions to its capital for 1984 were not found. - 38 - berate.481 Speculation on motives is inconclusive. Were interest-rate subsidies meant to offset disincentives to agriculture flowing from the foreign trade and pricing regime, as IBRD staff argued when presenting FIRA 8A to the Board?491 If so, they were a costly and inefficient way to do so. Did Government believe that getting inflation under control was imminent so that the damage would soon cease? This would explain why Government repeatedly made agreements about interest rates which it then found painful to implement and often did not implement. But then why not let FIRA protect itself against inflation as the nationalized commercial banks did, compensating for mandated negative real interest rates by requiring offsetting deposits of appropriate size? Failing other motives, politics is often invoked. However, visiting FIRA beneficiaries, it is hard to see why Government would have needed to buy the support of this particular group, and for what purpose? The motive mystery remains. We can only be sure that the policy was conscious, deliberate, and of very great fiscal cost. 107. Given the constraints of its economic and policy environment, the Audit finds that FIRA Implemented the projects very well. This overall judgement takes into account a variety of Imperfections from chicanery or carelessness in submitting non-qualifying loans in disbursement requests to lethargy in developing a serious operational manual or a trustworthy and useful ME system. 108. On the IMRD side, interest rates were also the major single pre- occupation. The seriousness of IBRD staff's concern is manifest in the attention the subject was given at all stages of the project cycle, and by the fact that some efforts were made to obtain compliance with agreements to the point of the frequent agreements not to submit disbursement requests in FIRA 7, 8, & 8A. In view of the documented fiscal cost to Mexico of not achieving positive real interest rates in these projects, the concern was well placed. 109. And yet real FIRA lending rates remained seriously negative throughout. Constant IBRD niggling probably did help raise interest rates over what they would have been otherwise by supporting those Mexicans who wished to reduce the fiscal loss. But even in FIRA SA, when rates got high enough to throttle down the immense credit demand, they were seriously negative. Reading the many statements in files that Government intended to reduce and in time eliminate most credit subsidies and to move towards positive real interest rates in accordance with an agreed timetable and seeing, over and over, that it did not happen gives the impression that staff were either hopelessly sanguine or dishonest. Most were neither. However, staff was not homogeneous on this point. Some, mostly Projects' technicians, were more interested in getting rid of subsidized-interest- rate distortions; they were generally more candid. Others, mostly Programs' staff and managers at different levels, were more concerned about Mexico's macro-economic predicament, about IBRD's lending program, and about disbursements; they were generally overly sanguine. In the events, real lending rates remained negative, and the frequent agreements not to 48/ See comments to FIRA 7 PCR referred to at Para. 102 above. 491 See Para. 42. - 39 - submit disbursement requests had only minor effects in slowing disburse- ments, mainly during FIRA 7. In the Audit's opinion, that the overly san- guine staff prevailed is not to IBRD's credit as a development institution. It might have been fruitful for IBRD staff to ask, given the major distor- tions in Mexico's agricultural sector, of which the interest rate was but the most prominent example, and given the Borrower's unwillingness to remove them even to the extent called for in legal agreements, whether 1-ding to the sector was appropriate. 110. When IBRD failed to stand by its principle on interest rates, the objective of making the rural credit system more self-sufficient also failed. But IBRD went further, agreeing to lower agreed farmer and participating-bank contributions on at least three occasions when the pace of disbursements seemed threatened. The self-sufficiency or weaning" objective, which is fairly prominent in discussions of FIRA 4 and 5, seems to fade from IBRD consciousness later. 111. Given that FIRA was trying to *swim upstream', i.e. to develop agriculture in an inauspicious environment. IBRD staff, in the Audit's opinion, had exaggerated expectations from operational manuals and M&E systems. In the Audit's opinion, IBRD staff had exaggerated expectations about reforming BANRURAL, something that would clearly have been in Mexico's national interest, when they could not even get the matter into legal agreements. 112. IBRD's performance in its objective of getting credit to poorer farmers is mixed. FIRA did get commercial banks to lend to PBI, but at great cost. In the Audit's opinion, the cost was unnecessarily high because some of the instruments for achieving this objective which IBRD and Borrower agreed to were pointless and costly (see Nos. 3 and 4 under Issues and Findings below). IBRD analysis should have made this clear beforehand and IBRD should have conveyed its findings forcefully to the Borrower. VI. ISSUES AND FINDINGS 113. This audit may be summarized as six issues and the associated findings, and a concluding remark. 114. Firstly, did the projects lead to investment and growth in Mexican agriculture? These, as distinct from getting the loans repaid or strength- ening the agricultural credit system, are the purpose for having the credit system. 115. More investment and growth compared to what? During most of the period in question, 1977 through 1987, Mexican agriculture was operating in a negative macro-economic context which discouraged investment in the sec- tor. The projects were trying to foster investment and growth in an envi- ronment of inflation and specific restrictions on agriculture through price regulations and trade barriers that created disincentives. These disincen- tives swamped the incentives created under the project. They were not the fault of the agricultural credit system and it should not be blamed for them. - 40 - 116. In this negative context for agriculture, FIRA's very attractive (to borrowers anyhow) interest rates, its technical assistance, and the expanding volume of credit and rural bank coverage must have increased agricultural investment relative to what it would have been without the projects in a context otherwise unchanged. FIRA's negative interest rates permitted some sub-optimal investments and mis-allocation of resources no doubt. The Audit does not regard this effect as terribly important, more than offset, as it was, by the investment disincentives of the sector's negative (but, of course, not uniformly and undistortedly negative) terms of trade. Therefore, it is reasonable to assume that investments made with PIRA rediscounts generally had reasonable financial returns, and probably higher economic ones. So to hold does not imply high credence in FIRA's and the PCR's models. 117. But because of the generally negative environment for the sector, investments associated with FIRA rediscounts were not necessarily net incremental investments for the economy. Through no fault of the agricul- tural credit system, because of the negative context, the projects also permitted some farmers to maintain their operations while shifting resources to other sectors (or abroad). It is not obvious what the without-project decapitalization of the sector would have been in such situations, making estimates of economic returns to the projects suspect. 118. Secondly, did the projects strengthen agricultural financial mar- kets? This objective is next only to fostering investment and growth because efficient agricultural financial markets will optimize investment and growth, subject of course to the sectoral context including availabi- lity of attractive new techniques, without prompting from Government or IBRD. 119. The answer is mixed. On the positive side, commercial banks became more involved in the sector. They increased the volume of their operations and expanded their branch network and the technical quality of their staff dealing with agriculture, all while maintaining excellent cre- dit discipline and collections. Few developing countries can match that record. Moreover, their repayment periods and grace periods are not rigid- ly determined by internal considerations but are flexibly adjusted to the investments being financed and the repayment capacity of the borrower. PIRA grew while remaining strong technically; it improved its internal management, at least as regards things its management is interested in. 120. BAURURAL did not improve its collection ratio, its ratio of admi- nistrative expenses to portfolio, or its responsiveness to farmers' wants, but since these were poor in 1976, the result is a neutral one that cannot really be blamed on the project. While IBRD staff entertained hopes of stanching the transfer of Government resources through BANRURAL and/or of improving its efficiency, these were not really part of the projects; the Borrower fairly objects to having the FIRA 7 project blamed because FIRA did not influence BANRURAL for the better. 121. On the negative side, the agricultural financial system became less financially self-sufficient. PIRA's failure to protect its capital against inflation has made it dependent on the central bank and on Govern- ment's largesse with the proceeds of foreign borrowing to maintain its influence. Hence FIRA's operations, as supported by IBRD, contributed in a - 41 - significant way to Mexico's debt problem. For the commercial banks, FIRA's terms and, for PBI, technical assistance and free guarantees (during the projects under audit) put them on a cost-plus contract; they took FIRA funds whenever they could, using their own funds when they could not. FIRA provided them with incentives to expand and improve their agricultural lending, but not to mobilize their own resources to finance it. Moreover, since demand for credit at negative interest rates from creditworthy com- mercial farmers was strong (no big surprise), commercial banks were helped to behave very conservatively, requiring, for instance, collateral worth several times loan amount. While there is no way of quantifying this phe- nomenon, banks sometimes required maintenance of off-setting, low-interest- bearing balances in return for loans, thus capturing part of FItA's inte- rest rate subsidy from their borrowers. However, banks seem to have been indifferent to requiring borrowers to invest their own resources in the sub-project investments as required by the projects' legal agreements. Such practices, consequences of subsidized interest rates, are good for the banks but not for efficient channelling of resources into the best invest- ments. Moreover, essentially no new savings fed the agricultural credit system except those of the Government. 122. The rural coimmercial banking system and FIRA are strong and well- articulated. Appropriate policies would provide them incentives to Increase their self-financing and to reduce their reliance on government grants. Future Government grants to the system, with or without IBRD financing, ought to be predicated on such chazges. However, without quite radical changes, it is hard to see how BANRURAL could significantly increase its self-reliance. 123. Thirdl9, was the Borrower's low interest rate policy wise? There can be no doubt that it strengthened the demand for rural credit during years when rapid inflation and, for most of the period, farm-gate prices that were not allowed to keep pace with it were discouraging farm and ranch investment. It effected a large resource transfer from Government mainly to those farmers who received FIRA-rediscounted loans.50/ 124. No doubt, Government did not plan on inflation so high. Conse- quently, it did not plan to make a transfer so big. Nevertheless, Govern- ment's consistent opposition to IBRD urgings for higher interest rates and to indexing and automatic adjustments manifests a will to effect a transfer to the farm sector, both in recognition that it was being hurt by inflation and fixed farm-gate prices and in an effort to reverse the resulting disin- vestment in the sector. 125. The policy was manifestly costly and only partly successful. The major policy problem with the transfer is that, while subsidized interest rates stimulate borrowing, not all of this translates into net new invest- ment in a sector that is depressed and discriminated against. Under the circumstances, not all Mexican farmers who got subsidized loans increased their investment in the sector (although good supervision by FIRA assured 501 If borrowing farmers captured 85-90 percent of the US$ 1150 million transfer during 1977-1986 (see Para. 103 above), (the rest being captured by the participating banks), then the transfer they received was roughly one billion US Dollars. - 42 - that the loan proceeds at least were invested in the sector). These parti- cular Mexican farmers were also the more worldly and often better-off ones so that a) they were more likely to be able to transfer resources out of the sector and b) they were less obviously deserving beneficiaries of transfers on social grounds (See *Fourthly" below). In fairness, the size of the transfer was certainly not planned, because the high rates of infla- tion were not planned. However, there was plenty of awareness of the possibility of inflation and opportunity to hedge the PIRA system against it--opportunity that was consciously not taken. 126. This Audit is convinced that there were cheaper and better alter- natives. Interest rate subsidies are a roundabout way to spur agricultural investment and/or to transfer income to the sector and probably introduce more undesirable allocative distortions than desirable benefits. They impact a small proportion of sectoral producers and a very specialized group at that. And, especially because of who the beneficiaries are, inte- rest rate subsidies can leak out of the sector. A more direct approach would have been to allow farmers the prices their products would have com- manded without Government efforts to keep staples cheap for urbanites, without quotas on meat exports to keep all Mexican producers from getting North American prices, etc. If such policies proved unacceptable, presum- ably because urbanites could not be charged the full price for their market basket and the fiscal cost would therefore be exorbitant, then the policy objectives could be realized more efficiently by subsidizing something used by a larger percentage of farmers than use credit, by a more representa- tive, less elite group of farmers, and ideally something less fungible than credit. Seed, fertilizer, pesticides, and even tractor fuel are common objects of subsidization. Such subsidies have their problems too, but, in Mexico, they are used by a broader spectrum of farmers than those who get FIRA rediscounts, and it is pretty hard to use them except in the farm sector. 127. Fourthly, how successful was the policy of earmarking and subsi- dising credit for low-income producers? The policy included four instru- ments: interest-rate subsidies, bank-by-bank quotas for PBI lending, FEGA guarantees, and technical assistance. Those low-income producers were not and were not meant to be the lowest 40 percent but rather the lowest ca. 80 percent of rural households, provided that no higher-income households succeeded in cheating to take advantage of the incentives. 128. The Audit's review of records and interviews with bankers and farmers convince us that the PBI policy has succeeded in getting commercial banks lending to an economic stratum of farmers just below their traditio- nal customers. Over the years, many of the old PBI have *graduated' to OP, commercial farmers, regular customers. The incentives keep the commercial banks looking for new PBI, new customers whose modest income and lack of connections would not automatically make them good credit risks, but for FIRA's incentives. So far, so good. 129. But, like all targeted subsidies, these create 'moral hazards," incentives for banks and non-PBI borrowers to cheat. FIRA knows this well and has taken elaborate steps to frustrate such cheating. Not everyone agrees, but the Audit believes that they have been as successful as one could expect in the circumstances. But that success has a rather high cost --the cost of FIRA's having to check up on the family income of sub-borro- - 43 - were, and of PIRA's managers having to check up on its staff who check up on sub-borrowers' family income, inter alia, transferring them frequently lest they become corrupted by non-PBI who want access to PBI inducements. A great deal of FIRA's contact with farmers is not as much technical assis- tance as auditing and policing made necessary because of targeted subsi- dies. 130. The Audit ventures a suggestion for achieving the policy objective at lower cost. It would be easier for FIRA to police the 150-odd redis- counting banks than the vast number of sub-borrowers. Of the four policy instruments, it is the interest-rate subsidies that induce farmers to dis- simulateg the commercial banks' quotas and guarantees do not interest them, and FIRA's technical assistance interests them perhaps less than it should. Genuine PBI first-time borrowers would be happy to pay OP interest rates or perhaps more if they had to to get into the formal credit system and esta- blish their creditworthiness, which is a thing of value. Therefore, the policy instruments might be focussed on the participating banks. Expanding the formal credit system to lower rural income strata depends principally on getting the commercial banks to look harder for non-traditional borro- were with viable investment projects. There is no evidence of a shortage of would-be PBI borrowers, but there is evidently no surplus of such persons with viable projects or of bank enthusiasm for finding them and incurring the high administrative cost of servicing their usually-small loans. Here FIRA's technical assistance and FEGA guarantees (now, post FIRA 8A, financially self-sustaining and still sought after by the parti- cipating banks), and even FIRA's PBI quotas, unpalatable though these are to economists, clearly serve their purpose. Perhaps they could serve the same purpose at lower cost, or commercial banks could be induced to share more of the risk by reducing the share of the loan that FEGA guarantees, but that is beyond the scope of this audit. Interest-rate subsidies for PBI do not serve the purpose of stimulating banks to find more PB1 loans. 131. Under FIRA 8, participating-bank spreads were made substantially more attractive for lending to PBI than for other lending (see above, Para. 39). There is evidence that, during FIRA SA, this step was working--redi- recting bank interest to PBI (see above, Para. 73), at least until IBRD agreed to change the spreads in an effort to maintain disbursements. 132. A study has been carried out by Hacienda, financed by FIRA 9, to find out, inter alia, how much more it costs banks to administer small loans than large, and to find and lend to new customers than to relend to known customers. Based on the results of this study, commercial banks could be compensated for finding and lending to PB1 on the basis of average historical costs. Such a subsidy would still cause some of the problems subsidies always cause, but it would be based, notionally at least, on actual costs and carefully focussed on the economic objective. 133. Fifthly, do commercial banks need PIRA support for non-PBI lendin? As a consequence of the third and fourth issues and findings above, the Audit finds that they do not now, although they may have during the period under audit. At present, and thanks in part of the projects under audit, the banks that deal with FIRA have a demand for FIRA funds for OP borrowers and borrowings that far exceeds FIA's supply. At a minimum, this suggests that these banks need less FIRA interest-rate subsidy as an inducement to such business, probably none at all. There is every reason - 44 - to believe that the comercial banks are strong enough in rural areas to be weaned from FIRA. This situation contrasts with that of PBI where, even with the multiple FIRA inducements cited above, the demand for FIRA PBI funds and the supply are roughly in balance. 134. Sixthly, what are the advantages and disadvantages of hajgj an agricultural rediscount bank do research and extension? These particular FIRA functions make Mexico unique for they are carried out by agriculture ministries in most countries. FIRA has stepped in to partially fill a partial vacuum created by the poor quality of Agriculture Ministry exten- sion and, in recent years, research (except in the Northwest where the latter is farmer-supported). 135. The unique arrangement has its advantages. FIRA extension (asistencia tecnica') and research (centros demonstrativos") are very closely linked to farmers and their wants, and to banks and their resources. It is precisely the weakness of such linkages that make most countries' ministerial extension and research services so futile and waste- ful. They are not virtues to be taken lightly. Moreover, the Audit is impressed with the quality of FIRA's work in research and extension. PIRA was unable to isolate the costs of these operations for the Audit, but we do know that FIRA's technical assistance and its guarantee costs have ranged between 2.3 and 3.7 percent of the value of its portfolio. We are still several steps away, then, from forming a judgement on cost-effective- ness of FIRA's extension and research. 136. Nevertheless, FIRA's research and extension do not fully do the job of national services through the ministry especially because of whom they serve and to whose wants they respond. The farmers FIRA serves are not a cross-section of Mexican farmers but the cream. More often than not, they will be the innovators, so that there is strategic sense in trying to reach them and hoping that innovations will trickle down. But trickle-down has its limitations. There will be factor-mix biases in what gets researched based on the interest of the generally larger, more prosperous, more progressive farmers and ranchers, and of their bankers--as, in the Bajio, FIRA's excellent work on reduced tillage faraing which, among other things, replaces labor with pesticides to cut costs. 137. It would seem imprudent to abandon research and extension that work, at least until Mexico builds a superb ministerial research and exten- sion system such as very few countries, rich or poor, possess. However, certain adjustments appear warranted to make the most of FIRA's capabili- ties and to compensate for their biases. Firstly, more should be done to disseminate FITRA's research and extension to farmers who don't borrow with FIRA rediscounts and to the official ministerial research and extension community. Secondly, a stronger ministerial research and extension func- tion would help compensate for FIRA's inevitable bias. 138. In conclusion, did PIRA 5. 6, 7, 8 & BA advance Mexican economic and social development? Because certain key information is missing as noted above, the Audit is a qualified one. We just cannot be sure. Still, a few general facts are clear. Mainly by maintaining negative real inte- rest rates during these five loans, Government cost itself in the order of US$1,150 million. If the social purpose of this transfer to farmers whose loans got FIRA rediscounts (and to a lesser extent, to participating banks) - 45 - was to offset the negative environment agriculture was subjected to--the distortions of artificially low farmgate prices and of inflation--and to keep some farmers investing in the sector in spite of that environment, then the projects achieved some modest success, though at a high public cost. Surely they were an inefficient way to further this purpose relative to several alternatives. Moreover, their effect was not nearly enough to reverse the general decline of Mexican agriculture, which decline, however, was not the fault of the agricultural credit system. 139. A second social purpose was to keep the richest farmers and ran- chers from getting everything by earmarking credit by income class of bor- rower and by subsidizing credit to the less-wealthy. Predictably, this created incentives to cheat; FIRA's structure and personnel policies are designed largely to control cheating. This costs. However, over the period in question, commercial banks did broaden the economic stratum they lent to a bit. 140. Mexico's commercial banks' rural lending and FIRA were strength- ened over the period, despite the negative conditions facing the farm/ranch sector. (BANRURAL, in contrast, grew bigger, but not better.) FIRA was still a reasonably efficient institution in 1987. Its operating costs had risen from 2.8 percent of average portfolio in 1980 to 4.2 percent in 1984- -too high for a rediscounter, but not unreasonable for a rediscount bank that was also its own extension and research service. The operating costs that are not clearly related to technical assistance or guarantees hover around 0.5 percent of average portfolio. That would be high for a pure rediscounter, but it is reasonable in view of the income-distribution task that FIRA management regards as its most important charge. The policy of distorting credit markets to favor PBI has administrative (and other) costs. 141. Over the five-projects' period, the system's dependence on govern- ment largesse increased rather than decreased. Macroeconomic conditions did not foster mobilization of savings by farmers and commercial banks for investment in agriculture. Cheap FIRA credit reinforced this tendency. FIRA could go on doing this, and could get institutionally stronger, only because Government gave it the money. 142. Government of Mexico chose to finance this operation partly through recourse to the central bank and partly through foreign borrowing, most prominently from I'AD. If present conditions persist until the end of the loan payment period, what Mexico will have paid in current US Dollars for the part of the projects financed by IBRD borrowings will have been 4.7 percent for FIRA 5, 1.3 percent for FIRA 6, 12.1 percent for FIRA 7, 16 percent for FIRA 8, and 10.4 percent for FIRA 8A. Compared to the 3-month LIBOR, these rates are -5.0, -6.5, +4.4, +7.3, and +1.2 percent respec- tively. In the lottery that governments play when they accept exchange- rate risk, Mexico did very well on two of these loans, not so well on three, though by that time international financial markets did not regard - 46 - Mexico as a prime risk borrower.511 The proceeds of that borrowing were probably not used so badly from a development point of view, although the Audit has to qualify that statement with two caveates that the fungibility issue makes the statement a good deal less than certain (see Para. 79 above); and that it is not hard to imagine ways of getting better develop- ment results with the resources. The main point, though, is that those who benefitted, a select group of farmers and ranchers mainly, were not the ones who had to repay the loans. Because of the way Government of Mexico used the loan proceeds, far from generating resources to repay the loans, it was constantly losing its Investment in FIRA over the period, to the tune of US$1,150 million. This loss and the foreign borrowing that foster- ed it contributed to Mexico's fiscal and foreign-payments problems of the 1980s. 51/ For a detailed elaboration of the assumptions behind these calculations and the arithmetic involved, see OED Report No. 7924. Project Performance Audit Report. Hungary. Grain Storage and Agricultural Mechanisation Project (Loans 2316-HU, B-2-RU and B-3-RU), June 30 1989, Annex 1. -47 Table 1 IBRD DISBURSENENTS FOR AGRICULTURAL CREDIT IN MEXICO - SY GUARTER, 1976-1987 (CURRENT US DOLLARS) QUARTER FIFTH - L.1217 SIXTH - L.1569 SEVENTH - L.1891 EIGHTH - L.2454 INTERIN - L.2610 ALL 5 PROJECTS 76.4 I 1,1,513 1.172.513 I 77.1 37.36,sas II 37,836.3881 I 77.2 I 19,648,934 I I | | 19,668,934 | I 77.3 I 94,Ms,784 I I I I 4,993,7841 I 77.4 11.7,2,s6a I I I I 11,7a2368 I 78.1 I 1.S70,99 I I I I I 11,570,998 78.2 6,088,991 | | 6,088.991 1 I 78.3 I 8,95s,649 e " I I 89s,649 j 78.4 I 15,355777 I I I I I 15,3557r I 79.1 4,4,528 I 5,946,3s I I I I 10,403,873 I 9.2 I 3174,072 I 6,9o, I I I SO,0,703 I -0i I oj 0 I I oj 79.4 I 1 49,670,767 j f I I 49,670,767 1 80.1 I I 28,248 OY I I I I 28248,707 I I 80.2 I j 31,341,941 I j j j 31,341.941 80.3 I I 22,561,420 | | 1 22.561,4201 I 80.4 I I 9,755,IS j I I I 9,755156 I 81.1 I I 1,318,376 I j I 1,318,376 I 81.2 I 4,56,658 I I 4,S6,658 I 81.3 I I 61,283,139 I | | 61,283,139 j 81.4 I I 00 I I of 82.1 I I I 1,5m5.90 I 1,575,090 82.2 I I I a7na,s5 I I 27.778.554 I 82.3 I I 17.3SO.346 I I I 17,350,346 I 82.4 I I I 24,670,047 I | 24,670,047 I I 83.1 I I I 6,352,947 I I 6,352,947 j 83.2I I I 18,040,227 I I 18,0,227| 83.3 I I I 5,467,151 I I I 5,467,151 I 83.4 I I 27.63.991 I I 27638991 I 84.1 I I 42,302,914I I I 42,302,914 j 84.2I I I 69,a0,717 I 69,880,717 84.3 I I I 22,5390630 22,539,630 84.4 I I I (845256)1 s 4 ao I I 46,320,884 I I 85.1 I I I 8,sra,as j 147,611,370 i156,184,151 I 85.21 I ,I ,298,078 26,298,078| I 85.3 I I 46.157,24 I I 46,157,254 I I 85.4 I I I 25,063,933 j 49,218,885 74,282,816 | 86.1i I I I 36s,68 9,9e I 36.6sa,935 1 I 86.2 I I I I I 6,938,713 I 6,938713 I I 86.3 I I I I I 22,838,590 I 22,838,590 I 86.4I I I I I 0 0 I 87.1 I I I I I 20,223,472 I 20,223.472 I 87.2I I I I I 935,1811 935,1811 .I I I 38,002,591 j 38,002,591 I 874I 5.s47,590 5 s,547.5 I TOTALS 10--02------ - *----1-------1- - 11 |TOTALS ( 125,000,002 | 200,000,001 j 324,999,998 | 300,000,001 | 180,268,010 I1,130,268,012( Table 2 48 - Part 1/3 IBRO DISBURSEMENTS FOR AGRICULTURAL CREDIT IN NEXICO - BY NONTN, 1976-1989 (CURRENT US DOLLARS) NORTH FIFTH - L.1217 SIXTH - L.1569 SEVENTH - L.1891 EIGHTH L.2454 INTERIN - L.2610 ALL S PROJECTS I 7612 I 1,172,513 | | 1,172,513 I not I 23,05802s I I I I as,ass025, I neaMw 750,408 I I I 750,408 7703 n 14,027,9ss I I I I 14,027,95S I I nom 0o I I I I 01 I 770s I 11,3sa4o2 I I j 11,352,402 | normI 8,316sss1 I I I I 8,316,S32 70 213,1I5 I I I I 213.1551 I nosm1 4,780,629 |I| 4,780,6291 I 77091 0I I I 0 oj I me 0o I I I I 0l I mi 36sea,m1 I I I I 368,736 I 7712 j 11,353,632 | | | | | 11,353,632 | I 0m01 oj 0 | | of I 7802 1,4ae.as I I j I 1,488,8 7803 j 10,oea,140 I I I I I 10,082,140 I I 7804 0o I I I I 01 I 7805 0a I I I I 0 I 7806 I 6,088,991 6 I I I I ,088,991 I 787 01o I I 0 I 7e08 IIo I I 0 I 7809| 8,958,69 I | I 8,958,649| I 7810 79,469 | | 709,4691 I 7811 9,859,629 | | | | 9,859,629| I 78121 4,786,679 I I I I 4,786,679I I 7901 I 2,999,1s I I I I 2,999,157 I m 79021 01 3,419,035 | | 3,419,035 7903 I 1,458,371 | 2,527,310 | | | 3,985,681 | 7 w4 I 2,65S.134 1,250,157 I | | 3,905,291 I 7905 j 518,938 I 4,ws,.os I I I I 5,315546 I w7906 40,853,866 I I I I 40,ss3,866 7907 I 01 I I I 01 I 79as I o1 I I I 01 I 7909 0 I | 0| I 7910 j 31,839,141 | | | 31,839,141 | 1 7911I I 0, 0 I 7912 I I 17,831,626 I I I I 17,831,626 I 8001 I 28,248,707 I I I 2,248,7o I I soo2i 01 o I I I 01 I 8003 1 0 | | | 01 I 8004 I 01 I I I 0 8005 | 17,212,975 | I | | 17,212,975 8006 I I 14,128,966 | | | 14,128,966 I 80071 I 01 I 0 18soo81 I oj I I I o1 8009 | 22,561,420 I | | I 22,561,420 | | 8010 j 4,516,314 | | | | 4,516,314 | SUBTOTAL: 125,000.002 189,186,125 0 0 0 314,186,127 (contimed..) i！〕召〕11〔〔萬屆〕唱〕〕i奮i〕〕11屆〕屆〕i〔〕〕〔〕〔〕義〕〕〕〔〔i〕〕奮召亂， 〕！!！、！！、！、！、！〕〕，〕；_ ！一：〕〔！〔！!!！〔〔〔！〕〔〕！〔〕〔〔〔！!！〔〔！〔！〔〕〔〕〔〔〔〕〔〔！！〔〕！〔〔！〕―i-!.! Table 2 50 - Part 313 lom DISSURS~Ts FOR AIWICULTURAL CREDIT 16 MX1CO - ST $MR, M6-1M (CURRM U$ DOLLARS) KWN FIFTH - L.1217 SIXTH - L.150 8~ 6 - L.1891 EIGHTR - L.2454 INMIN - L.2610 AU 5 PROJECTS 8410 loo.m loo.m 8411 20.245 20.245 8412 0,572,78DI 54,773,420 1 46.2W,639 i 8501 <701 99.%2.468 1 99,942.398 1 8502 8.572.851 1 5.395.009 13,967.860 1 8503 42^ 893 42.273.893 8504 596,322 596.322 8505 39,812 39.812 8506 25.661~ 25,661.944 8507 46.157.54 46.157,254 8508 0 0 8509 0 0 8510 15.527,038 15,527.038 8511 0 0 8512 9.536.M 1 49,218.883 1 58.755,778 8601 95.946 20~ .997 20,,680.943 8m2 0 0 mo 15,977,992 15.977,992 2.238,862 2,238.862 -om 0 0 4.699.851 4.699.851 8607 0 0 am 0 0 22.08.»0 22.08.590 8610 0 1 0 1 8611 0 0 1 8612 0 0 8701 0 0 8-m 0 0 870 20.223.4n 1 20,223,472 8704 106.915 106,915 8705 0 0 8706 =.267 =.267 8707 9,046.350 9.046.350 8708 0 0 8709 28.956,241 2B.956.241 8710.1 l i l l 0 0 8711 5.0w,ooo S.OW,000 8712 547.590 547.590 1- 1 : mi TOTALS 1 125,000.002 200,000,001 324.999.998 300.008.001 1 180,268,010 1,130,268,012 1 ,51一望跑hl極3 、〕〕！!!!!!：〕’., 、！ Program Perfor~e Audit Memor~ MEXICO FIRA 3, 6. 7, a & 8A FEGA Government Loss through froslon of FEFA Equity 1977-1986 ^x$ or US$ 91 t t f om) Veer Equity a 31 Dec Additions to Capttet Retalned Het Ineme ~ t CPl ~.Cost Exch"e Rate Loss for veer From Government Increase of F~ ~ Ave.) (9 Year End) <UM (110) <UM (x) M (USVKØXS> (US$IKØXS) (Hex%) (UM 1976 4634.3 231.7 11.84 15.4 20.0 19" 6616.7 291.4 1982.4 87.7 20.7 0.9 tg.i 12.88 22.6 22.7 -20.7 -28.9 1978 8628.4 380.1 2011.6 88.2 44.1 1.9 17.3 15.13 22.8 22.7 -44.1 -1.5 1979 12173.2 533.9 3544.8 155.5 178.1 7.8 18.2 16.35 22.8 22.8 -178.1 -9.5 1980 15127.7 649.2 2M.4 128.5 M.1 13.0 26.4 20.71 23.8 23.3 -298.2 -26.1 1981 16099.7 614.4 6n.o 27.4 300.0 12.2 27.9 28.58 24.5 26.2 0.0 -74.4 1982 18131.1 187.8 831.7 14.7 1199.7 21.3 58.9 40.40 56.4 96.5 0.0 -462.6 1983 20 m .5 145.9 2255.5 18.7 608.9 5.1 1013 56.65 120.1 143.9 0.0 -6ý.s 1984 4M6.3 222.8 14M.4 86.7 7372.4 43.9 65.5 51.08 167.8 192.6 0.0 -53.7 1985 101231.3 zn.3 40760.6 158.6 17554.4 68.3 57.7 56.07 M.9 371.7 0.0 -177.5 1986 258871.8 280.3 111787.0 182.7 45853.5 74.9 86.2 80.88 611.8 923.5 0.0 -249.7 1987 131.8 94.64 1378.2 2209.7 1988 114.2 67.64 2273.1 2281.0 TOTAL$ ~ > 181348.6 ~ .0 TOTAL -541.11 -1149.75 (UM 948.9 249.4 .......... . ...................................... .................................................................................... ............ Intrenseldecreme In Equi ty: <Nox$) 254237.3 <US$) 48.6 ................... ... ..................................... . ...................................................................................... Government LOSS In FEFA ~ty: <~$) -541.1 (UM -1149.7 Progras Performace Audit Neaorandma ENIXICO FIRA 5, 6, 7, 8 8 A Go~erient Los Through Eroson of R=NO EIpty 1981 through 1984 (MexS or US$ mftions) year Equity 8 31 Dec Addlttons to CapitaL Retafned Net Incuem Au~t CPl Ave.Cost Exchange Rate From Government Increas of Funds (Annuat Ave.)( Vear End) (Nex4) (US~) (>exS) (US9) (Mex ( > M <US$/Nex (U/Nex 1976 11.84 15.4 20.0 197 29.1 12.88 22.6 22.7 1978 17.3 15.13 22.8 22.7 1979 18.2 16.35 22.8 22.8 1980 417.0 18.3 26.4 20.71 23.0 23.3 1981 1628.0 71.4 41.0 1.8 98.0 4.3 27.9 28.58 24.5 26.2 1982 1767.0 75.8 21.0 0.9 147.0 6.4 58.9 40.40 56.4 96.5 1983 1935.0 73.9 22.0 0.9 464.0 18.9 101.5 56.65 120.1 143.9 1984 2421.0 25.1 65.5 51.08 167.8 192.6 1985 57.8 56.07 256.9 371.7 1986 86.2 80.88 611.8 923.5 1987 131.8 94.64 1378.2 2209.7 1988 114.2 67.64 2273.1 2281.0 TOTALS (MoKS> 501.0 709.0 (USM) 21.9 29.6 increas/decrease In Equity: (fex8) 793.0 <u8> -46.3 Govenment Loss In FONDO Equity: (fexl> -417.0 <US$) -97.8 IBRD AGRICULTURAL CREDIT PROJECTS IN MEXICO QUARTERLY DISBURSEMENTS, 1976 - 1987 YEAR1QUARTER 1976 1977 1978[ 19791L 19801 19811 1982 [*I 1983 1984 1985 [....... 19861 1987 (20) 0 20 40 60 80 100 120 140 160 DISBURSEMENTS MILLJONS FIRA 5 FIRA 6 FIRA 7 Fam1 FIRA 8 rn FIRA 8A Z L.1217 L.1569 L.1891 u L.2454 Li L.2610 AGRICULTURAL CREDIT PROJECTS IN MEXICO ANNUAL DISBURSEMENTS BY IBRD AND IADB 1977-1987 350 300 - -. -- - ----- 250 ............... .. . . . . .. ...... . . . 1~200 S 150 .... . ....... 50 100 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 YEAR I WORLD BANK INTER-AMERICAN DEVELOPMENT BANK _ 57 - Attachment Page 1 of 3 COMMENTS FROM THE BORROWER NACIONAL FINANCIERA Office of the Manager for Multilateral Financing Mr. Graham Donaldson May 9, 1990 Chief Agriculture, Infrastructure and Human Resources Division World Bank Washington, DC CEC-752-V/90 I refer to the Project Performance Audit Report on Loans 1217-ME, 1569-ME, 1891-ME, 2454-ME and 2610-ME, which this institution, acting as the Federal Government's financial agent, onlent to FIRA in order to finance its 1976-87 credit program. I enclose the comments of the agencies participating in project execution (i.e. FIRA, the Federal Government and Nacional Financiera), so that they can be incorporated into the report. Yours, etc. /8/ Jos4 Garcia Torres Manager Enclosure: as above cc: Lic. Guillermo V&zquez, Director of Planning for Credit, Finance and Administration, FIRA. Lic. Antonio Cervera, Director for International Financial Agencies, SHCP. Lic. Antonio Benites, Director for Agricultural and Fisheries Banking, SHCP. Lic. Jorge Delgado Benitez, Deputy Director for International Agencies, Nacional Financiera, SNC. Lic. Luis Nava, Executive Representative, Nacional Pianciera, SNC, Washington, DC. - 58 - Attachment Page 2 of 3 COMMENTS ON P_OgCT PERFORMANCE AUDIT REPORT (WORLD BANK LOANS 1217-ME. 1569-ME. 1981-ME. 2454-ME AND 2610-ME) 1. The document provides a very detailed overview of FIRA activities from 1976 to 1987, with special emphasis on the project implementation process, the provision of resources to beneficiaries, and the impact on FIRA' s performance and institutional development. 2. In general, the opinions expressed regarding FIRA's performance are favorable, since it efficiently fulfilled the purpose for which it was established and the objectives of each of the projects. 3. One of the lessons of the report is that FIRA must play a more active role in proposing ways of increasing its participation in the sector, instead of passively waiting for strategies to be decided by other government agencies. 4. Nevertheless (and even though the following comments may be considered of marginal importance), these clarifications should be made: (a) Page xii, para. 5 (TN: sic; in fact, para. 6] states that the Government augmented FIRA's capital through the central bank. In fact, the central bank financing consists of loans on which FIRA is paying interest, and the debt is accumulating. (b) Page xiii, para. 12: The Borrower did not voluntarily suspend submission of disbursement requests. The World Bank did not accept them because interest rates were not revised in time. (c) Page xiv, para. 18: The World Bank also failed to specify what results it expected from monitoring and evaluation. (d) Page xvi, first word ("support"): This should read "loans." (e) Page xvii, para. 28: The last three lines should be omitted, because--as the report itself states--FIRAhas developed an expensive system for checking the authenticity of PBI, with severe penalties for attempting to disguise other producers as PBI. (f) Page 3, para. 13, second and third lines: Should read "FEFA." (g) Page 17, para. 49, final part: Whenever supervision missions expressed doubts as to whether PBI qualified, the necessary checks were made and the Bank was informed of the results. No cases of misclassification were identified. (h) Page 18, para. 52 (a): In all cases, sub-borrowers have contributed their shares. It was difficult for FIRA to design a system for recording these contributions. (i) Page 29, first paragraph: Cost breakdowns were provided whenever requested; see the report by Mr. Paul Stroh. - 59 - Attachment Page 3 of 3 (J) Page 29, para. 84: See Note (a) above. (k) Page 31, para. 91: With regard to economic and financial issues, FIRA receives support and supervision from Banco de Mexico, so that the reference to FIRA's "management weakness" is inappropriate. (1) Page 42, para. 133: The Government's current financial policies are designed to achieve this purpose. S. FIRA is regarded as a model organization in this report, in spite of the serious decapitalization problems it has faced. These resulted from the macroeconomic difficulties and sector policies affecting the rural sector. May 8, 1990 JAMC*1ms