Mexico - Fourth Livestock and Agricultural Development Project

Document of The World Bank FOR OFFICIAL USE ONLY F\j E Report No. 2 5 7 7 PROJECT PERFORMANCE AUDIT REPORT MEXICO FOURTH LIVESTOCK AND AGRICULTURAL DEVELOPMENT PROJECT (LOAN 910-ME) June 29, 1979 Operations Evaluation Department This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.   FOR OFFICIAL USE ONLY Project Performance Audit Report MEXICO FOURTH LIVESTOCK AND AGRICULTURAL DEVELOPMENT PROJECT (Loan 910-ME) TABLE OF CONTENTS Page Preface i Basic Data Sheet ii Disbursement Table iii Highlights iv PROJECT PERFORMANCE AUDIT MEMORANDUM I. Summary 1 II. Main Issues 4 A. The Adequacy of the Project Design 4 B. Interest and Rediscounting Rates 7 C. Monitoring and Evaluation 11 Annex 1: Borrower's Comments 16 PROJECT COMPLETION REPORT I. Background 17 II. Project Description 17 III. Project Implementation 18 IV. Project Impact 36 V. Bank Performance 47 VI. Special Issues and Conclusions 49 Annexes 1-4 Map This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without IFC authorization.   - i - Project Performance Audit Report MEXICO FOURTH LIVESTOCK AND AGRICULTURAL DEVELOPMENT PROJECT (Loan 910-ME) PREFACE This is a performance audit of the Fourth Livestock and Agricul- tural Development Project in Mexico for which Loan 910-ME was approved in June 1973 in the sum of US$110 million. The final disbursement in respect of this loan was made on October 29, 1975. The audit report consists of an audit memorandum prepared by the Operations Evaluation Department and a Project Completion Report dated August 11, 1978. The PCR was prepared by the Latin America and Caribbean Regional Office on the basis of a country visit in November 1976. The audit memorandum is based on a review of the Appraisal Report (No. PA 133A) dated May 25, 1973, the President's Report (P-1240A) of May 30, 1973, the Loan Agreement dated June 18, 1973 and the PCR; correspondence with the Borrower and internal Bank memoranda on project issues as contained in relevant Bank files have also been consulted and Bank staff associated with the project have been interviewed. A copy of the draft report was sent to the Borrower on May 3, 1979. The response received is at Annex 1 of the PPAM. On the basis of this abbreviated review process, the audit finds no reason to question the accuracy and adequacy of the PCR with respect to the project's principal achievements and shortcomings. The points discussed by the audit have been selected because of their importance to this and other agricultural credit projects.   - ii - PROJi.ECT iF:RRFRMANCE AUIDIT RfEorT BASI:C DATA SllEE1T MEXICO FO0RTHII I_VESTOCK D)EVEL.OPME:NT PROJECT (LOAN 910-'E) iiEY PROJECT DATA Appraisal Actual or Item Expectation Current Estimate Total Project Cost (US$ millioni) 271.7 302.4 Overrun () 11 Loan Amount (1I'$ million) 110.0 110.0 /I Disbursed ) 110.0 Cancelled ) March 31, 1979 - 0 Repaid to ) - 5.8 Outstanding to ) - 104.2 Date for Cconpletion of P'hysical Components n.a. n.a. Proportion Completed by Appraisal Target Date (Z) - n.a. Proportion of Time Underrun ()- 60 Incremental Econiomic Rate of Return (Z) 29 18 OTHER PROJECT DATA Original Actual or Item Plan Revisions Current Estimate First Mention in l'iles or Timietable - - 05/11/77 CGoverionene t 's AppI icnt ion - - 05/11/72 Nh,-gt i:lt i..s 05/14/73 - 05/14i/3 Roard Approval 06!12/73 06/12/73 loan Agreemf-nt1 D.ate 06/1/73 - 06/18/71 Effectiv.neFs Date 10/21!73 11!20/73 10/,24/73 Closing Date 06/30/78 (18/'1 /75 10/274/75 Borrower Nacional Financiers S.A. Exercut ilng Agency Foo'do dIe (:l ant in y Ioe'.ento Fisc nl Y,-it' Ic, Lot r-wr Jan',:,ry I - I-:cvnlrber 31 Follow-oln) Project:-' Nanmes Fiftl Aj ni(lt irnIl & Live stock Dev. and Sixth Agric,-ltural Credit Loan Num'-r.c 121 7-MIE .:mJ I 561-'i &Asao,,,ts (UiS C., i I I i,n) li:SI25.0 ani i;.iiu Loan Agreenme-t Dates 03/11/76 wint 09/?7/78 MISSI0N PDlATA Monthh No. of No. of Dnat, of Mi.ssion /I Year W,-,ks Persols Menneets PIelpt- Appraisal 10/72 5 8 40 05/25/73 Spnpervision 1 06/73 2 1 2 06/18/73 S,.pervision II 11/73 2 4 8 01 /04/74 Su.pervision III 04/74 2 1 2 ol4/15/74 Suipervis ion IV 09/74 3 2 6 10/21/74 Sopereision V 12/74 4 2 8 12/23/74 S.pervisiont VI t)6/75 5 3 15 06/13/75 Supervisioin VII 05/76 1 ? 2 06/07/76 Supervisiwi VIII 11/76 3 2 6 01 /05/77 Complet iom 11/7t6 10 2 20 08/1J/7lt Total 69 COUIN'ritY EXCIIANGE RATES Name of Curreincy (Abbreviation) Peso (M.-xS) Year: Appraisal V-ir Averag.- Fxchang,: Rat,:: IT:ES;I 12.5 Intervsen ing Y Yns Average 1Is-,.] 1?.'. Com.pl Cetion Year Averi,ge iUS$1 = 12.5 /] P1; oc!. exrh::eg. aIjn.. tmee!t of 0.'34.   - iii - Project Performance Audit Report MEXICO FOURTH LIVESTOCK AND AGRICULTURAL DEVELOPMENT PROJECT (Loan 910-ME) Disbursement Table (US$ million, cumulative) Appraisal Actual Actual as % Period ending Estimate Disbursements of Estimate 12/31/73 4.0 0 06/30/74 13.9 - 0 12/31/74 29.2 57.9 198 06/30/75 54.8 78.3 143 12/31/75 77.1 110.0 143 06/30/76 94.3 - - 12/31/76 107.2 06/30/77 110.0   - iv - Project Performance Audit Report MEXICO FOURTH LIVESTOCK AND AGRICULTURAL DEVELOPMENT PROJECT (Loan 910-ME) HIGHLIGHTS The Fourth Livestock and Agricultural Development Project provided funds for medium and long-term lending for agricultural, livestock and agro- industrial investments. Funds were channeled through the Fideicomisos Institutidos en el Banco De Mexico en Relacion con la Agricultura (FIRA). The Fourth project contained two main subcomponents, one, to finance invest- ments judged feasible for commercial financing on a country-wide basis and two, a new program designed to provide development credit and extensive technical assistance to small farmers in poor agricultural regions. The project financed a large and diverse set of investments. Nearly 11,000 subloans benefitted 54,000 subborrowers in the two sub- components. Deviation from appraisal forecasts of investments, away from land improvements and towards more breeding stock, occurred in both sub- components. Inadequate attention to research, and an insufficient number of agricultural technical agents in Mexico, resulted in a lack of suitable applied technology for implementation during the project. The economic rate of return is tentatively estimated at 18 percent, lower than the appraisal estimate of 29%. Points of special interest are: - success in establishment of a new project subcomponent for lending to small farmers (PPAM paras. 2, 28, 37,. PCR para. 6.05); - Bank's acceptance of low interest rates on subloans to small farmers led to reduction in lendings and technical assistance to this group (PPAM paras. 25-39); - inadequacy of the project design for the fourth loan (PPAM paras. 14-24); - FIRA's progress in development of research and demonstration facilities led to reduction in production costs (PCR para. 3.10); - difficulties and high costs of establishing a monitoring and evaluation system for project subloans (PPAM paras. 40-55, PCR paras. 3.04-3.09, 5.02, 6.04); - FIRA's need to participate in project identification for agro-industrial subloans at an earlier stage (PCR paras. 3.27-3.28).   Project Performance Audit Memorandum MEXICO FOURTH LIVESTOCK AND AGRICULTURAL DEVELOPMENT PROJECT (Loan 910-ME) I. SUMMARY Background 1. Since 1965 the Bank has approved six loans totalling US$600 million equivalent to Mexico to support an agricultural credit program providing medium and long term funds for livestock, agricultural and agro-industrial development. All six loans have been channeled through the Fideicomisos Institutidos en el Banco de Mexico en Relacion con la Agricultura (FIRA), an administrative body providing common management for a group of three trust funds within the Bank of Mexico (BANXICO). These three funds jointly promote the flow of credit to the agricultural sector. 2. The fourth program was designed as a continuation of efforts financed under the first three projects. It contained two main sub- components: i) the General Sub-project (GSP) which supported livestock, crop production and agro-industrial investments judged feasible for commercial financing on a country-wide basis. Participating credit institutions, both public and private, extended loans to producers and received refinancing from FIRA; and ii) the Low Income Producers Sub-project (LIPSP) which initiated a new program to assist small farmers, both ejidatarios and individual proprietors, through the provision of development credit and extensive technical assistance. A small fund for agro-industrial financing was included in this subcomponent. In addition the project provided additional funds for 1) production oriented studies and demonstrations and 2) technical training for FIRA and participating banks' staff, and for project farmers. 3. FIRA continued to build upon its growing project experience and performed as an able and efficient institution. It strengthened the technical capacity of its staff through training abroad, transferred 130 technical staff to work full time on the PIPSP sub- component, added to its agro-industrial staff and continued the regional decentralization of its operations.  - 2 - 4. The Loan Agreement required FIRA to measure the on-farm and agro-industrial impact of project investments. This is an area in which FIRA has been relatively weak. After an initial slowness, largely based on its concern with the cost of the effort which was proposed by the Bank, FIRA took steps to fulfill its commitment. A monitoring and evaluation unit was established in early 1976. Its efforts have focused on sub-borrowers under the LIPSP subprogram, producing two useful reports. More effort is needed, however, particularly regarding the impact of subloans within the GSP subcomponent. As the cost of the monitoring and evaluation effort has been high, a re-examination is merited to clarify goals, methodology, and the incidence of the costs of evaluation. 5. A set of graduated interest rates were established for project sub-borrowers ranging from 10 to 12 percent for commercial borrowers under the GSP subproject. Interest rates to LIPSP sub- borrowers were set at 7.6 percent, the rate established by Government for low-income producers throughout Mexico. The Bank was concerned that this rate was too low, and that the funds used for interest rate subsidy would better be spent on agricultural research and technical assistance. The Bank did not press the Government sufficiently on this issue, either during negotiations or during implementation. 6. Because of rising inflation in Mexico, all sublending rates (both GSP and LIPSP) were negative by the time the project became effective. Rising costs, associated with inflation and higher technical assistance costs to LIPSP sub-borrowers, resulted in two small modifications of the interest and rediscounting rates for the GSP subcomponent. The rates for LIPSP sub-borrowers were not changed. All on-lending interest rates have remained negative throughout the last six years. 7. The Loan Agreement stipulated that FIRA encourage all agro- industrial sub-borrowers to obtain several price quotations from several countries for any contract for equipment and associated services financed out of the loan. The formula employed was designed to ensure inter- national "shopping" without the formal procedures of ICB. FIRA had difficulty meeting the project agreement formula and expressed the judgement that this regulation was both too inflexible and time consuming. The Bank agreed to relax the formula and allow approval when the sub-borrowers had evidenced reasonable effort at shopping. The new system seems to have worked well. 8. Disbursement of funds proceeded quickly and was completed in October 1975. The project financed 10,900 subloans, compared with 12,600 forecast at appraisal. The project reached 54,000  - 3 - beneficiaries under the two subcomponents. There were 30,500 beneficiaries under the GSP program, nearly triple appraisal predictions. A much greater than expected number of subloans were made to ejido groups, private sociedades, and to private grupos solidarios. This pattern reflects the income disparities which exist among ejidos, as well as the Government's policy to encourage loans to farmer groups. There were 23,500 beneficiaries under the LIPSP program, essentially the number forecast at appraisal. However, during project implementation the volume of lending to low income producers was reduced from appraisal forecasts by about 30 percent because the low rediscount rate and higher technical costs for this subcomponent placed a severe financial strain on FIRA. 9. The average size of the GSP subloans was about 50 percent higher than projected, at US$22,000. The average size of the LIPSP subloans was about 33 percent below the appraisal estimates, at US$19,200. Livestock operations absorbed 53 percent of the total investment items financed with project funds. Within livestock loans, 66 percent of the investments were for breeding stock. Thus, investments deviated from appraisal assumptions by placing much less emphasis on land improvement and irrigation investments (4 and 5 percent of livestock investments, respectively). For annual crops, which accounted for 23 percent of total project investments 40 percent of the subcomponent investments went to machinery, 39 percent to land improvement and 14 percent to irrigation. 10. The pattern of systematic deviations away from appraisal forecasts for investment by category reflects the lack of appropriate applied technology in Mexico for recommending to sub-borrowers, and the continuing reluctance of farmers and ranchers to invest in land improvements due to agrarian reform legislation and pressures. Data from FIRA's monitoring system indicate that project investments have caused additional land to be brought into production, but the investments have caused little increase in the crop yields or in livestock herd technical coefficients. 11. The economic rate of return estimated at appraisal was 29 percent. This was a weighted average of the returns to a diverse set of investments in different production activities. Insufficient evidence is available to permit a precise measure of actual investment impact, and some loans have not yet matured. The PCR tentatively estimates the rate of return at 18 percent. However, as beef cattle investments have had the lowest rate of return, and have assumed a greater than predicted importance, this estimate may be too optimistic.  - 4 - II. MAIN ISSUES A. The Adequacy of the Project Design 12. The first agricultural and livestock credit project was designed to increase private bank participation in lending to the agricultural sector. Because most private banks, in 1965, were not accustomed to working in the agricultural sector, had little or no staff specialized in this area, and viewed the sector as being highly risky and of low profitability, Bank funds were to be on-lent through FIRA, and the participating banks, to farmer-s. The participating banks, through the discount mechanism, were thus enabled to make a volume of loans much greater than the amount of funds committed from their own resources, and the rate of return on their own funds was sharply increased. Further, the appraisal of individual farm investments, originally carried out chiefly by FIRA agents, did much to reduce bank costs and to reduce the risk of the loans made. 13. The agricultural and livestock credit program has evolved considerably from this beginning. Public as well as private banks now participate. The agricultural staffs of the participating banks have grown markedly and improved in quality. These staffs now take responsibility for most on-ranch investment appraisal and supervision. FIRA technicians have become increasingly specialized, concentrating on staff training, program promotion, coordination, and new technology. FIRA has decentralized its activities into ecologically similar regions, and provided its local offices with greater independence. The scope of the program has grown enormously in terms of the number of loans made, the total volume of lending, the special distribution of loans, and the variety of investments for which funds are offered. In addition, FIRA manages other credit programs, some within the context of other Bank projects (with additional Bank support) as well as having growing amounts of domestic funds available. Both the private and, especially, the public banks also lend to agriculture from their own resources, i.e., funds which are not discounted through FIRA. 14. The growing complexity and volume of FIRA's lending, and the increase in total credit to the agricultural sector, suggest that a number of changes would perhaps be made in the way in which the agriculture and livestock credit projects are appraised and supervised. Project staff has been aware of this need for some time, and received, as early as 1976, strong opinions from FIRA and the Banco Nacional de Credito Rural (BNCR) that they were in agreement. However, several factors have impeded the design of a new approach acceptable to the Bank and the Governmentl/. 1/ The Region comments that at the time of appraisal the Bank was not willing to use a sector approach to agricultural lending. A region specific project as an alternative would not have been accepted by FIRA and BNCR.  - 5 - Both the fifth and the sixth projects followed a similar approach. The proper solution is not wholly clear, but several problems evident in the fourth project are related to what appears to be currently unsatisfactory design. 15. While the first project was limited in scope, it sought to induce private bank participation in the sector and, almost on a pilot basis, tried to introduce specific new techniques through on- farm investment. The fourth project came much closer to being a sector loan in substance. However, although not fully recognized in the first project, by the fourth project some Bank staff knew that Mexico 1) simply did not have adequate agricultural technology, 2) sector policy deficiencies were crucial to arresting the deteriorating food supply situation, 3) it was not longer sufficient to limit the Bank's focus to FIRA, but rather the agricultural credit system as a whole had to be analyzed, and 4) Bank staff were simply unable to provide proper supervision to the increasingly large and complex agricultural credit program. 16. Nonetheless, the Bank persisted in applying the same project design again during the fourth loan. The principal problems resulting are evident in the appraisal report design. The emphasis was placed on the elaboration of individual on-farm investment models instead of on limiting sector policies and deficiencies. As for those projects which preceeded it, the fourth appraisal report contained a set of models presented in the annexes. These models purported to indicate the impact of representative investment packages which would be carried out under the projec.t. The models indicated, in shorthand form, the technology which was to be applied and provided a standard against which actual project results were to be compared. In their aggregate, they also provided an expected project economic rate of return upon which Board approval of the project was partly based. 17. Given the diversity of Mexico's agricultural conditions, and the volume of FIRA's lending, a large number of separate models were required to cover the range of investment activities which were to be financed. The staff effort to develop all of these models was enormous. Moreover, and this was a much more fundamental problem, the technologies represented in many of the models were of questionable existence or were at least uneconomic under the conditions then prevailing in Mexico. 18. In appraising the fourth project, some Bank staff wished to avoid the preparation of the detailed farm investment models which they felt were largely artificial. The models required much effort, there was little evidence that such models actually reflected the investments which were being made, and Bank staff could, in the absence  - 6 - of having to prepare the models, work on more fundamental problems. However, the staff was told that the models formed an essential part of the appraisal process and were required for Board presentation. The models were therefore constructed again, although in somewhat abbreviated form. 19. The Bank has moved to increasingly utilize quantitative techniques in its work. This is important. However, it is clear that normal operational pressures on Bank staff can induce staff to follow the path of least resistance even if, as in this case, the normal procedures do not appear to have been the most appropriate. 20. The design of the appraisal report implied that Mexico had a set of viable technologies for agriculture and livestock which, when applied through on-farm investments under then current sector conditions (prices, farm management, technical assistance) provided very high financial and economic returns. The economic rate of return on the GSP and the LIPSP were estimated in the appraisal report at 26 and 41 percent, respectively. Thus, although project staff felt that the models were not representative of actual investment opportunities, they risked convincing other staff within the Bank, and the Government, to the contrary. 21. It is important to note that the appraisal missions for both the third and fourth projects (in 1970 and 1972) had tried to convince the appropriate officials of the Government to focus more on the need to develop technology appropriate to its d.iverse agriculture. The missions indicated that the lack of appropriate improved technology to recommend to sub-borrowers was limiting the effectiveness of the Bank-supported credit projects. Both missions returned feeling that, despite clear evidence from FIRA's experience that there was a need for well directed applied research, especially in livestock and rainfed agricultuy1, that virtually nothing was being done to improve the situation- . The missions indicated that the Bank might be prepared to help organize and finance such an effort. Nonetheless, neither in the Government, nor at other levels within the Bank, does there seem to have been a consciousness of the severity of the problem. Only after the crop shortfalls, and large imports of basic grains required in 1972-733, did the Government begin to re-examine its agricultural policy. The Bank did the same. 1/ FIRA did move to fill this gap, to the extent possible. It increased its budgetary and staff committment to applied and demonstration programs. Although not established as a research agency, it has achieved some very impressive results which are now recognized internationally. These results have been most significant in the area of low-cost livestock production systems, for both cattle and dairy units. Efforts by Bank staff to encourage and assist FIRA's work in this direction have been especially important.  - 7 - 22. It is clear, with hindsight, that a large number of other factors were also hindering growth in agriculture. Among the most important were low agricultural product prices, a shortage of technical assistance, neglect of small farmers, inadequate water pricing and water use efficiency on irrigation projects, inappropriate land reform legislation, and a need for rural infrastructure. Additional funds for the sector were needed, but for this credit to have been fully effective a set of policy changes was advisable. Accordingly, the fourth loan should have been treated more as a sector loan because of both its size and complexity, and also the importance of sector as opposed to project issues. Less emphasis was due to the elaboration of individual investment models and more to the design of the macro-system which was the principal determinant of whether the sector was to develop. 23. The need for sharply different agricultural policies has now been recognized and, in harmony with Government efforts, the Bank has moved to assist in this direction. A number of recent Bank reports on Mexican agriculture are impressive in this regard. However, the project design has so far been continued for both the fifth and sixth agriculture and livestock credit projects. Two alternatives would seem to offer potential improvement. In one, funds would be provided to FIRA for sector lending with only broad guidelines as to the expected activities and technologies to be financed. The Bank's efforts would be placed on institutional and sectoral policy improvement such as (i) the identification of research and extension priorities, (ii) the monitoring of investment impact on important variables like financial and economic rate of return, income distribution, employment, and (iii) important barriers to efficient investment like price distortions, and the need for improved land and water policies. 24. In the other, although not necessarily mutually exclusive, the Bank's lending program would be disaggregated into smaller projects, which would be appraised by type of activity and perhaps for different regions. Lending would be used to achieve more limited and more specific goals, like the introduction of individual technologies, the provision of credit to individuals under special conditions (small farmers), rural development, and the like. As the funds which Mexico will need from the Bank are likely to be smaller, relative to domestic resources, in the future, the latter approach would permit Bank staff to work cooperatively with Mexican officials and technicians more closely on specific development issues. B. Interest and Rediscounting Rates 25. As in previous projects, subloans were provided at graduated interest rates. Within the GSP subcomponent, larger loans bore higher interest rates, ranging from 10 to 12 percent, on the supposition that they Would be received by larger and wealthier farmers. The selection  - 8 - of one rate, for sub-borrowers within the GSP instead of the set of graduated rates by loan size, would probably have reduced administrative costs without significantly affecting sector equity or allocational efficiency. Within the LIPSP subcomponent, all subloans were made at 7.6 percent. The lower rate for this subcomponent was proposed by the Government as part of a general policy of reduced interest rates to small farmers. The Government argued that (i) the loan beneficiaries were "very poor" and merited the implied income transfers and (ii) potential loan recipients controlled valuable land and labor resources which could be brought into production only by offering low interest rates. 26. A substantial discussion took place within the Bank during loan preparation regarding the acceptability of the low rate proposed for LIPSP sub-borrowers. On the one side, it was argued that a higher rate, such as 10 percent, was preferable on both equity and efficiency grounds. The data available indicated that rural incomes in Mexico were skewed, with about 40 percent of the rural poor having (in 1969) per capita incomes lower than US$86. The definition of a small farmer, which FIRA then intended to use for those who would qualify for LIPSP subloans, was a net family income of less than US$2,000. This was about US$360 per capita, or four times as large as that earned by the lower 40 percent. Under the higher limit, 90 percent of the rural population qualified for LIPSP loans. Because the poorest of the rural population were mostly landless, and unlikely to be good prospects for credit, Bank staff expected that most LIPSP funds would go to borrowers having rela- tively high incomes by rural standards. Accordingly, subsidiza Ron on account of poverty as defined in the case appeared inappropriate- 27. It was also pointed out that there was great demand for credit among small farmers, that considerable amounts were currently being borrowed from money lenders at very high interest rates, and that substantially more credit would be sought by small farmers at 7.6 percent than the project could supply. Low interest rates were not needed to induce investment among small farmers. Instead, it was argued that higher interest rates would result in improved allocation within the limits of the credit which could be made available. 28. On the other side, it was argued that the Government was simply not willing to negotiate the interest rate for small farmers. A general policy announcement on this issue had just been made. Bank staff believed that the Government would prefer to forego Bank assistance for the LIPSP subcomponent (which had been designed independently by the Government 1/ The Region states that the project definition was not designed to serve the absolute poorest of the rural population. However, it did reach poor farmers.  - 9 - previous to its inclusion in the fourth project), rather than reverse its policy. Because of the importance of the project, it was argued that the Bank should not delay by entering into discussions, but rather proceed with the loan, evaluate the consequences, and subsequently return to the issue if it then seemed merited. 29. The issue within the Bank was not one of fact regarding the appropriateness of the low interest rate for the LIPSP subcomponent, but of strategy regarding what the Bank could accomplish. Some staff believed that the issue was so important that it should be pressed on the Government, while others were content to let it pass while concentrating on other problems. Prior to negotiations it was agreed, however, that the strong economic arguments against the use of low interest rates to LIPSP would be given to the Government. The Bank's argument would emphasize the limited resources then available for agricultural investment in Mexico, both in terms of credit and of research1 nd technical assistance, and the imperative need to do something about this- . However, if after a strong argument had been made to the Government, and pressed to a high level, the Government refused to change its position, the Bank was prepared to accept 7.6 percent. 30. During negotiations the issue of interest rates was raised perfunctorily, but was not strongly pushed to the Government. The Bank accepted the low interest rate for LIPSP borrowers without making a strong case. This was a different position than the Bank had initially agreed before or than it had assumed in negotiating agricultural or livestock credit projects with most other countries in the region. Particularly where inflation had been a serious problem, the Bank had worked strenuously for higher interest rates to sub-borrowers. Thus, an opportunity was passed for substantial discussion with the Government regarding the advantages and disadvantages of interest rate subsidies for achieving the goals which the Government had established and other related factors important to agricultural sector development were not raised. It is unlikely that higher interest rates for low income borrowers could have been obtained at the time, but the issue could have been discussed in considerably greater degree in the hope of influencing future policy actions. 31. As it turned out, the low interst rates charged to LIPSP sub- borrowers resulted in less actual lending to this subcomponent and to 1/ As is mentioned in the discussion of the design of appraisal report models, projects staff felt that there was very little new technology available in Mexico that would be used either to increase agricultural production, particularly in rainfed areas, or to increase livestock and dairy production.  - 10 - provision of less technical assistance than would otherwise have occurred. Thus, a policy which the Government hoped would benefit the rural poor worked to their disadvantage. The reasons for these developments are related closely to the high cost of making loans to this component, and to the relatively high level of inflation prevailing in Mexico during the project. 32. In 1972, during appraisal of the fourth project, inflation in Mexico was about 5 percent. Inflation began to increase significantly at the beginning of 1973, reflecting the lagged effect of large increases in public sector demand and in the money supply which had taken place in 1972. Thus, during negotiations, April/May 1973, inflation was rising rapidly and the interest rates being charged to LIPSP sub-borrowers turned negative in real terms. Following 1973, domestic inflation was exacerbated by international inflation. It averaged 21 percent between 1974-77. 33. As pointed out in the PCR (para. 3.13) two interest rate adjustments were requested by FIRA during the period of loan implementation. In May 1974, a supervision mission discussed with FIRA the fact that the latter was operating at a financial deficit on the amounts discounted for LIPSP loans. FIRA discounted participating bank's loans at 4.6 percent while borrowing part of the required funds from the Bank at 7.25 percent, with the balance from its own funds or from the Bank of Mexico at an average rate of about 4.6 percent. This deficit, of course, was inherent in the original design of the project. However, rising technical and administrative costs associated with promotion, organization, technical appraisal and supervision of the LIPSP loans were adding to FIRA's deficit. Government subsidies to cover the technical assistance costs to low income farmers had been agreed during appraisal and incorporated into the Loan Agreement (Guarantee Agreement, Section 2.03 and Loan Agreement, Schedule 2.D), but were proving inadequate. FIRA indicated that it did not wish to request a higher Government contribution to cover the higher administrative costs and resulting deficits. It feared that this transfer payment would grow rapidly (the more successful LIPSP became, the greater would become FIRA's deficit), and the Government would then either become unwilling to further underwrite the program or would wish to interfere more directly with FIRA operations. FIRA's concern for its operating independence was a major consideration. 34. FIRA and the Bank agreed subsequently that the only adequate solution was to try to obtain Government agreement to a general increase in interest rates to farmers of all categories, and of the corresponding discount rates to increase FIRA's margins. In December 1974, despite growing awareness of the need to raise interest rates, the Government decided that it did not wish to raise them for small farmers. FIRA was permitted to raise rates on loans within the GSP program. FIRA also informed the Bank at this time that it had abandoned its earlier intention to seek a reallocation of project funds from GSP to LIPSP because the financial strain imposed by the LIPSP program.  - 11 - 35. In August 1975, FIRA again requested an increase in on-lending rates, this time to improve the operating margins of participating banks. The Bank again agreed. The Government refused again to allow rates to be increased by the LIPSP, but rates within the GSP were further increasedl/ . 36. Although lending for the LIPSP subcomponent had originally occurred more rapidly than predicted at appraisal, and although FIRA had considered reallocating funds from the GSP to the LIPSP component, the final allocation, with Bank approval, was reversed in the opposite direction. Only 13 percent of funds in the fourth project were disbursed to small farmers, as opposed to the 19 percent allocated at appraisal. 37. FIRA made a determined effort to lend to small farmers. By early 1974 it had transferred 27 percent of its technical staff to work full time on the LIPSP subcomponent and made available most of the remaining staff to assist on a part-time basis. Nonetheless, FIRA was also concerned with its financial position. Once the institutional cost of lending to small farmers became too great, FIRA reduced this lending. 38. The low interest rate also resulted in less technical assistance being made available to borrowers. The Bank noted in its review of the fourth project (PCR mission, April 1977) that, although Loan 910-ME had been disbursed 12 months ahead of schedule, there had been a lowering of technical standards (relative to past projects) in order to expedite lending operations. Bank staff felt this was particularly serious in the case of loans made for beef development where investments in breeding stock, as opposed to infrastructure and pasture improvements, had risen disproportionately. 39. At least part of the reason for the decline in technical assistance provided is that FIRA was still being squeezed financially as a result of higher costs associated with the LIPSP program, and of inflation. It tried to compensate for higher costs by increasing its lending volume, although potentially reducing the productivity of the loans being made. In short, the lending program was harmed, not helped, by the fact that interest rates were fixed at low real levels, and poorer farmers, who needed additional funds and additional technical assistance, were hurt the most. The Bank did not pursue the case as strenuously as the evidence indicates was merited. C. Monitoring and Evaluation 40. Starting with the second project, FIRA was required to measure the impact of product investments on selected farms and agro-industries. The Bank requested information on the financial 1/ Interest rates were held at 7.6 percent for LIPSP sub-borrowers during the fifth project as well (Loan 1217-ME, effective August 1976), and interest rates to GSP borrowers were increased only marginally.  - 12 - and economic rates of return on different investments, and on the various factors affecting profitability, both of which might permit improved project planning and implementation in the future. 41. FIRA agreed, reluctantly, to undertake the requested evaluation. The analysis carried out during the second project fell short, however, of what the Bank desired, and the Bank pressed for an improved effort during the third project. Funds were made available in the third loan for a consultant who could assist FIRA in the design of a monitoring and evaluation system, as it was recognized that FIRA had no experience in the task that was being requested. The Bank also had no staff which it could make available on a continuing basis to assist in the evaluation. 42. During the third and fourth projects, FIRA conducted a detailed survey of a sample of farms/ranches which had received project subloans. A pilot system of permanent record-keeping on a sample of commercial farms was also initiated. The consultant assisted in both of these efforts. OED also sampled a set of farmslyhich had made investments in the second project, and analyzed this data- . The information on project impact which these various efforts produce9 were discussed in some detail in the OED evaluation of the third project- . In essence, subloan investments were found to have had a positive effect on farm production, but, although producing a satisfactory economic rate of return, profitability was significantly below that which had been predicted in the third appraisal report. Investments permitted an increase in production principally at the extensive margin, rather than at the intensive margin. New lands were brought into cultivation, or into livestock production, but crop yields were not significantly increased, nor were livestock technical coefficients on existing herds markedly improved. Within the sectoral context then prevailing, the path of development which occurred was probably the most economical. 43. The PCR provides a good description of progress made in monitor- ing and evaluating investments made during the fourth project. FIRA's efforts in this period were focused almost exclusively on sub-borrowers within the LIPSP program, and a record-keeping system for a sample of these farms was initiated. Of the two reports issued to date, one is especially useful. It contains an analysis of a sample of nearly 400 individual smallholders and ejido enterprises, and indicates candidly the significant degree to which operational difficulties affect these enterprises. The most important problems are identified, at least qualitatively. The information obtained in the effort will likely permit an even more productive flow of credit to this set of sub-borrowers in the future. 1/ Used in "Evaluation of Agricultural Credit Programs." OED Report No. 1357 dated November 12, 1976. 2/ OED Report No. 1573, dated April 27, 1977.  - 13 - 44. It is important, however, that no additional monitoring and evaluation has been done of investments carried out within the GSP subcomponent. This subcomponent contains the bulk of the funds disbursed in the successive projects to date, and most new funds are also going to this group. Despite previous evaluation efforts, a great deal of additional information is needed on the functioning and impact of investments financed within this subcomponent. There has been inadequate effort to draw policy implications from the evidence obtained, so that the feedback desired has not been fully achieved. Further, while the LIPSP subloans have been made more recently, and have not fully matured, there are many GSP sub- borrowers who received their first loans over ten years ago, and who have, in some cases, received repeater loans. Thus, subloan impact can now be measured more completely. 45. FIRA's recent emphasis on LIPSP sub-borrowers is explained largely in terms of how it perceives the costs and benefits of monitoring and evaluation to itself. The cost of the monitoring and evaluation program is significant. The special unit established to undertake this task commenced operation with a total of 32 technicians, 7 percent of FIRA's total technical staff. They were assigned to work full time on the project, and other staff assisted part time. Given that this evaluation staff has worked solely on small farmer loans during the fourth loan, and that the total FIRA staff assigned to the LIPSP sub- component totalled 130 technicians, the monitoring staff amounted to fully 20 percent of the staff assigned to the LIPSP program. 46. Monitoring and evaluation is important, but its costs, particularly to the on-lending agency which bears these can be quite large. The Bank has perhaps not given full consideration to this issue. For example, FIRA's original reluctance to undertake a major evaluation effort was made quite clear to the Bank. FIRA did not believe that the benefits which would accrue to it, as an institution, would justify the costs which it would have to bear. FIRA believed that the results, when obtained, would be valuable principally to the Bank, and thus that the Bank should pay. 47. Part of the problem is that the Bank did not explain to FIRA in sufficient detail the precise information which could and should be obtained in the monitoring effort, and how this information would be of use in planning future activities, and improving subloan appraisal and supervision. The Bank perhaps did not fully recognize the magnitude of the task which it was proposing. The concept of monitoring and evaluation was relatively new in Bank agricultural projects when FIRA was first approached. However, even after FIRA understood what was wanted, it remained reluctant to spend large amounts  - 14 - on an evaluation of the FSP subcomponent. It agreed to do this only when the Bank made it a condition of the third loan. FIRA became more positive towards evaluation only after it switched its emphasis to the smaller farmers in the LIPSP program. 48. FIRA's institutional position is an important reason for explaining its perspective on monitoring and evaluation. FIRA, as a banking agency depends on the Bank of Mexico, and yet enjoys considerable autonomy and separate prestige. It is a non-profit institution whose income from its portfolio is expected to only cover its current expenses. Nonetheless, FIRA believes that a small operating surplus is important to the continuation of its autonomy. A deficit, such as that which might be caused by a large expenditure on monitoring and evaluation, would require FIRA to request a subsidy from outside sources. Although it has accepted some funds to finance the additional costs of technical assistance provided to small farmer sub-borrowers, as have also been made available to the private and public banks operating in the agricultural sector, FIRA was unwilling to see this subsidy grow too large. Rather than risk political interference in its operations, it chose to limit the growth of the LIPSP subcomponent during the fourth loan. For similar reasons, it earlier resisted the major additional expenses which it believed the monitoring and evaluation of subloans to commercial producers would involve. 49. FIRA also did not perceive large benefits from such an evaluation. Its commercial portfolio was sound. Nearly all loans within the GSP sub- component appeared good and the collateral pledged against the loans was generally adequate. Further, as FIRA is a rediscounting agency, the participating banks stood to bear any losses, with FIRA suffering only if these banks themselves defaulted. Loan productivity might have been improved through evaluation, but FIRA perceived that an increased volume of lending was probably more important to it so long as at least satis- factory productivity was obtained on the loans which were being made. 50. The situation was different, however, with respect to the LIPSP subcomponent. First, the small farmer portfolio was a genuinely new activity, perceived as more risky, and certainly more costly. The efforts involved in organization of farmers, loan appraisal, and technical assistance were much more extensive than those directed to commercial producers. A large political emphasis in Mexico has been placed in recent years on increasing lending to small farmers, and FIRA knew that the Bank might direct an increasing proportion of its future project support in this direction. For all these reasons, FIRA felt it important to learn as much as possible 9 out this group, and the factors most important in lending to it- . 1/ The actual risk from LIPSP subcomponent is still extremely small. It composed only 13 percent of the fourth project and small farmer subloans comprise only about 7 percent of all the Bank-assisted subloans made since the first project began.  - 15 - 51. Second, the monitoring and evaluation of small farmers qualifies for subsidies established to reimburse the expenses associated with technical assistance provided to small farmers. Thus, FIRA could obtain support for its monitoring and evaluation efforts to this end, which it could not have done for similar efforts aimed at larger commercial producers. 52. The lesson is simple. The Bank must be increasingly sensitive to the high cost of monitoring and evaluation. It must also respect the institutional interests of the agencies through which it lends when monitoring and evaluation systems are desired. It may be appropriate to include funds within Bank loans to pay a larger part of any evaluation costs, thereby making the direct cost to the agency itself less onerous. 53. In the case of the Mexican agricultural and livestock credit program, however, additional evaluation of the GSP component is still desirable. In this effort, focus should be placed on subloans made for the purchase of beef cattle breeding stock. This is substantially the largest subloan item. The evaluations carried out indicate that beef cattle subloans have had little impact on herd technical coefficients, and a relatively low economic rate of return. The rate probably does not exceed 12 percent. This is only barely adequate and higher benefits might be achieved if funds which are now destined to livestock could be redirected to other more profitable investments. This is being attempted in the sixth loan in which a lower ceiling has been placed on the proportion of loan volume allocated for livestock and, within livestock, for breeding stock. 54. It is important that potential loan substitution is also highest for this subloan category. Loan funds placed in breeding animals are highly liquid and, once obtained, are extremely difficult to control. Ranchers may hold the new animals purchased, selling other animals previously owned on the pretext of financial need, and thus gain net funds at low interest rates for alternative investments. FIRA has done a good job of administering livestock loans, but even its controls are inadequate to deal with this problem. Now that real interest rates on subloans have been negative for six years in a row, the market mechanism can no longer be depended upon as a guarantor that the funds lent will be allocated efficiently. 55. It must be recognized, however, that the subloans for livestock development, including the purchase of breeding animals, are very important to FIRA's institutional viability under current conditions. Without the volume of lending which this activity generates, at relatively low staff cost in terms of appraisal and supervision, it is unlikely that FIRA would be able to provide intensive staff assistance to other agricultural sub- sectors in its lending program. Thus, a re-examination of FIRA's discount margins on subloans may be required to ensure that it is reimbursed appropriately to the cost which its bears on each type of loan.  Annex 1 NACIONAL FINANCIERA S.A. - 16 - Office of the General Manager TRANSLATION Official Financing June 7, 1979 Mr. Shiv S. Kapur, Director, Operations Evaluation Department International Bank for Reconstruction and Development 1818 H Street, N.W. Washington D.C. 20433, USSA Dear Mr. Kapur, I wish to refer to your letter dated May 3, 1979 requesting our opinion on the draft project performance audit report on the Fourth Livestock Development Program (Loan 910-ME) prepared by your Department. I would like to inform you that after having analyzed the document we agree in general terms with the Bank's opinion, in the sense that the stated project objectives were achieved with success. I consider it appro- priate, however, to emphasize some points of view presented in the report, which deserve special consideration. 1. In a strict sense, FIRA's interest rates policy is oriented towards guaranteeing thatthey permit the effective promotion of agricultural and livestock development; however, the rates are established so as to eli- minate the possibility of reducing the Fund's resources. The modifications and adjustments in the interest rates for the various credits granted under Loans 910-MR, 1217-ME and 1569-ME respond to such objective. 2. Regarding technical assistance, the Fund has tended to operate as a second tier bank, based on a large participation of the private banking sector in providing these services, which relatively diminishes the cost to the Fund on such account. This situation does not preclude, however, studying the establishment of a specific program of technical assistance of a larger scope. 3. Finally, I would like to emphasize the desirability of a more active and direct participation in future evaluation reports on the part of the technicians of the executing organization as well as of the Nacional Financiera S.A. This would allow the Bank a larger, direct experience which would serve as a basis for reorienting its evaluation policies from the point of view of the fulfillment of the stated project objectives. Hoping that our comments will be useful, I remain, Sincerely yours, Pedro Galicia General Manager  - 17- PROJECT COMPLETION REPORT MEXICO FOURTH LIVESTOCK AND AGRICULTURAL DEVELOPMENT PROJECT (LOAN 910-ME) I. Background 1.01 Since 1965, the World Bank 1/ has made a series of six loans totalling US$600 million equivalent to Mexico in support of an agricultural credit program providing medium and long term funds for livestock, agricul- tural and agro-industrial development. All six loans have been channeled through the Fideicomisos Instituidos en el Banco de Mexico en Relacion con la Agricultura (FIRA), an administrative body providing com mn management for a group of three trust funds within the Bank of Mexico (BANXICO), which jointly promote the flow of credit to the agricultural sector. The Fondo de Garantia y Fomento para la Agricultura, Ganaderia y Avicultura (FONDO) was the first trust fund established in 1955, to provide a combination of credit and technical assistance to the public and private banks and commer- cial producers. The Fondo Especial para Financiamientos Agropecuarios (FEFA) was created in 1965 specifically as a channel for loan funds from the IBRD and IDB. The Fondo Especial de Asistencia Tecnica y Garantia para Creditos Agropecuarios (FEGA) was set up in 1975 to defray the technical assistance costs associated with the financing of ejidatarios and other low-income producers and to guarantee, in substantial part, the recovery of past due principal balances of such loans granted by private credit institutions. 1.02 Due to the rapid implementation of the Third Livestock and Agricul- tural Development Project (Loan 747-ME), the Mexican government contacted the Bank in August 1972 to initiate the development of a fourth loan to support the on-going investment program under FIRA, and to include for the first time a component for low income producers. After discussions with FIRA personnel in Washington in September 1972, the appraisal of the Fourth Project 2/ took place in October 1972, followed by negotiations in mid-May 1973. The definition of an innovative component to reach low-income producers was accompanied by discussion centered on interest rate policy towards this group. The Bank agreed that the project should conform to the government's sectoral interest rate policy in Mexico, which set 7.6% as the interest rate for low-income producers throughout Mexico. II. Proiect Description 2.01 The Fourth Project was designed as a continuation of the program, financed under the first three livestock and agricultural development projects (Loans 430-ME, 610-ME, and 747-ME) to provide financing and technical services to the agricultural sector. This program was created to provide medium and long term funds to agricultural producers and processors, and simultaneously to widen the role of the private and official bankinxg system in technically assisted lending for on-farm development. In addi- tion, the Fourth project reflected closely the policy priorities defined by the Mexican government for the agricultural sector: (i) a rapid increase 1/ Referred to as the Bank throughout the report. 2/ Refers to the Fourth Livestock and Agricultural Project (Loan 910-ME) throughout the report.  - 18 - in the production of foodstuffs both for domestic and international markets and (ii) an improvement in the distribution of income. 2.02 The project consisted of two main subcomponents: (a) The General Sub-Project (GSP) which was designed to continue the investment program developed under the first three Bank- financed projects. Livestock, crop production and agro- industrial investments, judged feasible and sound by FIRA technicians were eligible for financing under the GSP on a country wide basis. Participating credit intitutions extended loans to producers and received refinancing from FIRA; and (b) the Low Income Producers Sub-Project (LIPSP) which initiated a new program aimed to assist ejidatarios and other small farmers through the provision of medium and long term credit and extensive technical assistance for on-farm development. A small agro-industrial sub-component was also included. In addition, the project provided for studies, training and demontration aimed at: (i) providing technical services to farmers particularly ejidatarios and other low-income producers; (ii) production-oriented studies and demonstra- tions; and (iii) technical training for FIRA and participating banks' staff and for project farmers. The project was declared effective on October 24, 1973 after a delay of several months needed in the finalization of legal agreements and documents specifying the rate of commission which NAFINSA would charge FIRA, during project implementation. III. Proiect Implementation Institutional Development 3.01 The Fourth project was implemented by FIRA, the administrative organization within the Bank of Mexico (BANXICO) responsible for the admin- istration of three trust funds for agricultural development (para 1.01). FIRA acts as both a technical assistance agency as well as the rediscounting arm of BANXICO. All project sub-loans were made on the basis of development plans evaluated by FIRA technicians. FIRA staff also provided on-farm supervision and training, as well as developed specialized demonstration units. 3.02 FIRA's performance as implementing agency for the first three Bank- financed livestock and agricultural development projects has been good. Under the Fourth project, FIRA continued to build upon its growing project experi- ence, and performed as an able and efficient institution. FIRA was able to successfully undertake the implementation of the new LIPSP sub-component. The rapid rate of fund commitment under the LIPSP was in part the result of FIRA promotion of the project among small farmers and ejidatarios six months before project effectiveness. FIRA also transferred some 130 technical staff to work full time on the LIPSP and made additional staff available on a part-time basis to assist in the promotion, organization and technical  - 19 - assistance of small-holders and ejidos. These measures were complemented by the training of staff in the Banco Nacional de Credito Agricola (BA) and the Banco Nacional de Credito Ejidal (BE) who were involved in LIPSP implementation, through courses organized by FIRA at the Chapingo Agricultural School, and through direct training in the field. However, despite these efforts, FIRA was faced with higher manpower per loan require- ments for the evaluation and supervision of LIPSP operations particularly for individual smallholders although less so for Ejido groups. Training 3.03 FIRA continued to strengthen the technical capability of its staff through training programs abroad. In addition FIRA moved to strengthen its agro-industrial technical staff, a department which had been judged weak at appraisal. The project agreement required, as a condition of effectiveness, that FIRA employ an agro-industrial specialist to evaluate projects and to provide staff training. FIRA met this condition without delay and also further strengthened its agro-industrial technical staff by recruiting five new staff members who were chiefly assigned to field offices and by sending two project technicians abroad for specialized training. In order to process the expanding program, as well as to continue the decentralization of operations, FIRA expanded its total number of rural offices from 99 to 122 and increased the total number of its technical staff. FIRA efforts to provide low-income producers with adequate technical back-stopping also were substantial. However, the increased proportion of sub-borrowers in the low-income category, increased FIRA costs as low-income producers required more technical assistance per sub-loan than commercial borrowers (para. 3.13). Monitoring 3.04 One weakness in FIRA's performance was the failure to fully imple- ment the monitoring activities required under the project agreement (Section 2.07(b), which were designed to measure the on-farm and agro-industrial impact of project investments. Initially, FIRA was slow to establish a permanent continuous record keeping system and had not fully complied with provision 2.07(b) of the project agreement when the loan was fully disbursed in October 1975. However, FIRA was taking steps towards fulfilling its commitment in the latter part of 1975. A small group of FIRA technicians were selected to initiate the monitoring program, and a consultant engaged to assist this core of technicians in establishing a data recovery system, drawing up the necessary forms, and advising on the selection of sample farmers. The project's closing date was extended to cover establishment of the unit. 3.05 The unit commenced operation in January 1976, with a field staff of 25 technicians in addition to 7 headquarters' staff technicians. Initially it was decided to concentrate on the low-income producers sub-sector (LIPSP). The decision to concentrate the initial monitoring efforts on the LIPSP was based on the judgement that the low-income producer component was a more problem prone and risky operation, and thus could most benefit from the feedback which early monitoring would provide. At the same time, this decision greatly increased the difficulty of establishing the system, as low-income farmers are the least likely to keep on-farm records. Initial  - 20 - monitoring activities centered on a stratified random sample of 276 indi- vidual small farmers (Pequenos Propietarios) and 117 collectives (ejidos), willing to participating in the program. These represented 12% and 17.5% respectively of the two LIPSP groups financed (during the 12 months July 1, 1974 to June 30, 1975), under IDB loan 298SF/ME and IBRD Loan 910-ME. 3.06 The initial system comprised keeping records on a continuing basis of the sample farms over the development period of the sub-loan and is based on (i) a daily diary kept by the borrower; (ii) frequent farm visits by the technician (every 2 to 4 weeks); (iii) three monthly and annual progress report to headquarters by each field technician; and (iv) coordination of the field activities and processing and interpretation of the data by the headquarters staff. Field Officers were located country-wide in 18 FIRA offices selected on the basis of those having the highest concentration of LIPSP. The unit was officially designated "Sistema de Registros Continuos y Asistencia Tecnica Empresarial" and was located within FIRA's Producer Organization and Cooperative Development. 3.07 Considering the lack of previous experience in on-farm evaluation by almost all the technicians appointed to the program 3/, the unit has functioned surprisingly smoothly and has made considerable progress in its operations. The units' staff are young and qualified and the field technicians have been quick to develop a rapport with the pequeno pro- pietarios and ejidatarios. In-service training of field technicians has been carried out on an informal basis during the course of visits by headquarters staff to the field offices. This is to be strengthened by the appointment of two field coordinating officers to the headquarters staff with specific responsibility for supervision of field staff activities. On a formal basis, in-service training was given through a one-week seminar held by the unit to discuss progress to date in September 1976 and attended by all its technicians. The future program calls for the expansion of the sample to concentrate in key regions of FIRA operations and main produc- tive activities. 3.08 One additional bottleneck in the unit's work has been in the programming of data and subsequent processing by FIRA's computer unit. The monitoring unit's demands are but one of many within FIRA and the computer is fully extended. FIRA has taken steps to correct this, however, and future data processing should be accomplished in a timely manner. Two studies were completed in June and August 1977 based on data and information obtained from the sample farms as follows: (i) changes in net worth and (ii) a subjective evaluation of the key technical and managerial aspects on on-farm performance (para. 4.09). 3.09 Thus, the establishment of a monitoring unit within FIRA proceeded more slowly than anticipated at appraisal. Although FIRA initially did not share the Bank's emphasis on the establishment of a continuous monitoring 3/ However, most technicians had previous operational field experience within FIRA.  - 21 - unit, commitment to the unit grew over the life of the project. Clearly, FIRA staff and resources were constrained by the heavy commitment necessary to implement on-going lending programs, particularly with the added staff requirements created by the growth of low-income sub-projects in FIRA's portfolio. In addition, the establishment of a comprehensive computerized monitoring unit required more time and effort than envisaged at appraisal. The Bank's insistance on the establishment of such a unit was sound, but more technical backstopping could have been provided FIRA at an earlier stage during project implementation in order to assure the success of the system. FIRA's program is both large and complex, comprising a large number of sub-loans of many different types. Therefore, the establishment of a monitoring unit requires the testing and evolution of both techniques and institutional experience. In addition, due to the relatively long period required for the maturity of on-farm investments, little meaningful data can be generated immediately upon sub-project financing. Several years of lead time result in more useful time-series data, more accurately reflecting longer term project impact. Research and Demonstration 3.10 FIRA made considerable progress in improving its research and demonstration facilities under the project. FIRA demonstration farms for perennial and annual crops were improved and operated smoothly. In addi- tion, FIRA financed and operated demonstration centers for low-cost milk production in wet and dry tropical and temperate zones are excellent. These demonstration farms were particularly innovative in providing evidence that milk production can be economically and financially feasible in the wet and dry tropical and temperate zones, by developing systems maximizing the use of improved grass and legume pastures thereby reducing the use of expensive concentrate feeds. Using the full FIRA technology package, milk production is feasible in any region in Mexico with reductions of about one-third of the traditional cost. On-lending Operations Participating Institutions 3.11 As under the first three Bank-financed projects, FIRA acted as the rediscounting facility for sub-loans granted by participating private and official credit institutions. At appraisal, it was forecast that GSP component funds would be channeled through private credit institutions and the Banks of the Banco Nacional Agropecuario (BANAGRO) system. LIPSP funds would be channeled through private and BANAGRO Banks, and in addition through the Banco Nacional de Credito Ejidal (BE) and Banco Nacional de Credito Agricola (BA) systems. During the project lifetime, two modifica- tions in the appraisal on-lending arrangements were made. First, in August 1974, the Borrower requested that the Financiera Nacional Azucarera (FINASA), a Government owned financiera with the sole and exclusive responsibility for meeting all the credit needs of cane farming and sugar production, be made eligible for rediscounting under the project. The Bank agreed to the amendment of project agreements, providing a modest  - 22 - contribution to the Government's planned sugar development program, by supporting medium- and low-income cane producers term credit needs for the replanting of cane, new planting of cane, new planting development and farm equipment. The Bank's agreement to FINASA participation was limited to US$4.5 million for on-farm cane production and subject to FINASA and FIRA both establishing separate accounts for project activities. At project closing only US$1.4 million had been disbursed for cane production through FINASA. The shortfall in actual disbursements was due to the low Government fixed prices for cane in an inflationary setting which resulted in a reduction on the demand for funds for these types of investments. 3.12 The second modification in on-lending arrangements took place in April 1975. According to Schedule I of the original Project Agreement, the Banco Nacional de Credito Agricola (BA) was an eligible channel for sub-loans under the LIPSP component. In November 1974, the BA requested the modification of the Agreement, granting BA eligibility for GSP funds. The Bank agreed to the change, as FIRA funds for LIPSP were fully committed at that time, and it was judged a reasonable mechanism for augmenting BA's funds thereby allowing the bank to serve additional farmers. Of a total of Mex$ 2,564 million rediscounted by FIRA under the project, Mex$ 846 million (33.5%) was channeled through the public banks participating under the project. A total of Mex$ 411.6 million was discounted through the public banking system to low-income producers; a total of Mex$ 447.1 million was rediscounted under the GSP component through the participating public banks. Interest and Rediscounting Rates 3.13 During the project lifetime two modifications in the interest and rediscounting rates occurred (Table 1). The first was requested by FIRA in December 1974, and approved by the Bank effective January 1, 1975. FIRA requested the first revision in rate structures to cover increased expenditures on staff salaries and transportation for field staff necessitated by the increasing proportion of FIRA operations oriented to low-income producers. The slow start-up of FEGA, designed to cover the technical assistance costs associated with low-income producers, contributed to FIRA's perception that an increase in interest rates was justified. Thus, the first revision increased the rediscount and on-lending rates for project beneficiaries in the GSP component in order to improve FIRA's overall financial position. A second revision was requested by FIRA and approved by the Bank in August 1975 and served to improve the operating margins of participating banks by increasing the on-lending rates to beneficiaries, without further revision in the rediscount rates. The Bank's judgment in approving the revision was that: (i) the investment being financed under the project had adequate financial rates of return to absorb an increase in interest rates without discouraging further applications; and (ii) higher interest rates to sub-borrowers reflected a positive movement towards real rates of interest in the agricul- tural sector, and were therefore attractive on overall policy grounds. However, interest rates to LIPSP producers remained unchanged at 7.6% through- out the life of the project.  MEXICO COMPLETION REPORT FOURTH LIVESTOCK AND AGRICULTURE DEVELOPMENT PROJECT LOAN 910-ME INTEREST AND DISCOUNTING RATES Rediscount Rediscount Rstes to Pwticipating Rates Interest Rates to Interest Ratoe to Interest Rates to Sze of Lo-nros Percantsg$ of Loan Institutlcns First Beneficiaries bensficiarlor BEneficiaries 2/ B(Pesos)-jrfe Discounted Appraisal Revision- Appraisal First Revisim Second Revisioir Up to 150,000 Up to 90 7.5 8.75 10.0 11.25 11.5 150,001-250,000 Up to 85 6.0 9.75 10.5 12.0 12.25 250,001-500,000 Up to 70 9.0 10.5 11.5 12.5 13.0 SO0,001-1,000,000 Up to 70 9.0 11.0 11.5 13.0 13.5 1,000,001-5,000,000 Up to 70 10.0 11.50 12.0 13.0 13.5 5,000,001-10,000,000 Up to 70 10.0 12.25 12.0 13.5 14.0 10,000,001-15,000,000 Up to 70 10.0 12.75 12.0 13.5 14.5 All LIPSP subloans Up to 90 4.6 4.6 7.6 7.6 7.6 1/ As of January, 1975 (proposed in December, 1974). 2/ As of September, 1975.  - 24 - Total Project Costs and Financing 3.14 Total project costs were Mex$ 3,780 million (US$302.4 million) as compared with appraisal estimates of Mex$ 3,396 (US$271.7 million). 4/ Table 2 Total Project Costs and Financing (US$ millions) 1/ GSP LIPSP Studies Training Appraisal Actual Appraisal Actual Appraisal Actual Producers 33.4 34.0 2.6 2.8 - - Participating Banks 34.3 55.1 5.3 3.7 - - FIRA 65.2 81.1 19.2 14.0 1.7 NA IBRD 82.8 89.0 25.0 19.8 2.2 1.2 TOTAL BY COMPONENT 215.7 259.2 52.1 40.3 3.9 2.9 TOTAL PROJECT: Appraisal 271.7 Actual 302.4 1/ At US$1=Mex$ 12.5 Source: FIRA. Increased project costs resulted, in part, from high rates of price increases for key investment items, experienced during the project period. 3.15 Financing of GSP and LIPSP components deviated only slightly from appraisal projections: 4/ Includes an estimated figure for total costs under Studies, Training and Demonstrations, which matches appraisal estimates and IBRD disburse- ments to this category. Thus, the total sum attributable to the category is Mex$ 36.3 million (US$2.9 million). Data on actual expenditures are not available as FIRA budgeting for these activities is on an annual rather than project-by-project basis.  - 25 - Table 3 Project Financing (% of Total) GSP LIPSP Appraisal Actual Appraisal Actual Producer 16 13 5 7 Participating Banks 16 21 10 9 FIRA 30 31 37 36 IBRD 38 35 48 48 TOTAL 100 100 100 100 The deviations can be analyzed by component; under the GSP, producers contri- buted slightly less to total project costs and participating banks 5% more than forecast at appraisal. The Bank loan financed a slightly lesser per- centage of total project costs, 35%, as compared with 38% forecast at appraisal. In the three previous loans 5/, one of the Bank's major concerns had been to encourage participating private banks to rediscount a smaller proportion of their loans with FIRA and thus make larger contributions to sub-loan financing. As the OED audit report of the Third Project (Loan 747-ME) noted, the Bank had utilized the gradual reduction in the proportion of funds discountable through FIRA as one mechanism for increasing the private banks contribution to project lending. This mechanism was repeated under the Fourth Project in which the maximum percent of the loan discountable was reduced from 80% (Third Loan) to 70% under the Fourth Project, for loans of between Mex$ 250,000 and Mex$ 500,000 (US$20,000-40,000). The average size of loans granted under the GSP component was US$25,337; therefore, the decrease in the percent of loans rediscountable by the private banks for loans of US$20-40,000, provides a partial explanation for the increase in participating bank financing of project lending. Under the LIPSP component, producers financed a slightly higher proportion of project costs (7% as compared with appraisal estimates of 5%), at the expense of participating banks and FIRA's contribution. Procurement 3.16 Under the Third Livestock and Agriculture Development Project (Loan 747-ME), the Bank had required that FIRA "encourage sub-borrowers to seek several bids from local suppliers and contractors wherever possible." 6/ 5/ Loans 430, 610 and 747-ME. 6/ Schedule 2C to Project Agreement, Loan 747-ME.  - 26 - In addition, all sub-loans exceeding US$400,000 were referred to the bank for approval. However, subsequent reviews of agro-industrial investments made under the Third Project revealed that however small, each sub-loan generally involved some imported machinery or equipment, and that procure- ment of this machinery had generally been made on the basis of one quotation. Therefore, the appraisal mission of the Fourth Project concluded that more extensive shopping by the sub-borrowers would be beneficial to agro- industrial investments. The Loan Agreement for the Fourth Project thus required FIRA to encourage all sub-borrowers to obtain several price quota- tons, preferably at least three, and for any contract for equipment and associated services for agro-industries to be financed out of the loan, and estimated to cost the equivalent of $100,000 or more, that the sub- borrower be required to obtain quotations from not less than 5 suppliers of such equipment produced in not less than three countries. The formula was adopted to ensure international "shopping" without the formal procedures of ICB. In addition, to assist FIRA in developing its technical capability to evaluate quotations, the hiring of an agro-industrial specialist by FIRA was a condition of effectiveness. 3.17 FIRA did hire the agro-industrial specialist and began to strengthen its agro-industrial department (para. 3.03). FIRA also agreed with the Bank's judgement that one of the obstacles for sub-borrowers to obtain several quotations was the lack of information on a range of manufacturers. To help solve this problem, FIRA prepared a catalogue listing foreign and local equip- ment manufacturers, indexed by agro-industry for distribution to all participa- ting banks. Unfortunately, this assistance was not immediately available at the beginning of the project period. In addition, FIRA technicians did relatively little in the promotion of agro-industries in various regions, instead relying on participating banks to bring clients to them. The partici- pating banks in turn did not advise prospective borrowers of the need to obtain several quotations and FIRA intervention was late in the loan cycle, at the stage when loan documents had already been prepared. Furthermore, many sub-borrowers had chosen machinery even before seeking a loan. As a result, FIRA submitted sub-loans for Bank approval which did not comply with the procurement regulations of the Project Agreement. FIRA clearly had difficulty meeting the project agreement formula, and expressed the judgment that this regulation was both too inflexible and time-consuming, causing considerable delays in sub-loan processing. In addition to prospective borrowers' tendency to have already chosen equipment on the basis of their own inquiries before approaching the Bank, FIRA defined as additional problems: (i) that equipment from certain countries is particularly suitable and well- developed for certain specialized activities; (ii) strong inflationary trends make quotations valid for only short periods of time, making firms reluctant to bid and increasing the time required and difficulty in getting the required five quotations; and (iii) some foreign manufacturers have no servicing facilities in Mexico, and are therefore not viable candidates for equipment contracts. FIRA therefore requested the relaxation of the procurement formula, on the grounds that it was both time-consuming and required too many bids under Mexican conditions. The Bank, faced with  - ii - delays in approving agro-industrial sub-loans due to non-compliance with procurement requirements, responded by relaxing the formula so that sub- loans would be acceptable even if they did not meet the five quotations from three countries rule, provided that a sufficient number of legitimate suppliers had been contracted and given reasonable time, (defined by the Bank as approximately one month) to make a response. The Bank indicated that provided that FIRA could detail why the five quotations from three countries could not be obtained, sub-loans would be approved with fewer quotations. After this relaxation of the procurement terms, which occurred in January 1975, FIRA provided information on number of bids received, and from'what sources,for agro-industrial sub-loan. Overall, Bank approval for these sub-loans was speedy. Although the vast majority did not meet the five bids-three country formula, FIRA did provide adequate evidence that a reasonable number of bids have been reviewed, and a justification existed for choosing from less than the required number of bids. 3.18 With hindsight, it appears that although sub-borrowers were not in fact receiving quotations from five firms in three countries, in most cases, the choice of machinery had been adequately researched by the sub- borrower. FIRA's role in loan processing was therefore essentially a passive one; technical evaluations were based on projects basically already prepared by prospective borrowers, including the choice of equipment. Prospective borrowers had considerable entrepreneurial ability and experi- ence, and in fact the greatest emphasis in GSP agro-industrial sub-loans was on the expansion and improvement of existing plants, rather than the development of new facilities. Within this context, it would appear that the Bank's insistence on a set formula for procurement of agro-industrial machinery was too inflexible and unsuited to the Mexican conditions. Clearly, sub-borrowers could well use some advice and assistance in seeking a variety of quotations for machinery purchases, in order to achieve poten- tial cost reductions. In the absence of such direction, prospective sub-borrowers were most likely to purchase machinery easily viewable and accessible in the United States. But the "five bids from three countries" formula was too rigid and proved too cumbersome to implement. Further, sub-borrowers had strong preferences and beliefs about the type of machinery they wished to purchase before applying for the loans. The Bank's subse- quent decision to modify this clause and allow approval of subloans in which the borrower had evidenced reasonable efforts at shopping for equip- ment, and made a prudent choice, was a sound and appropriate one. Accounts and Audit 3.19 At negotiations, assurances were obtained that FIRA would continue to submit certified statements of operating results and financial condition not later than three months after the end of each FIRA fiscal year. FIRA did comply with this assurance, and in addition produced quarterly reports on the progress of the Fourth Project. However, these often were received several months after the quarter closed. Audited balance sheets were well prepared and received on a timely basis.  - 28 - Disbursements and Reallocation of Funds 3.20 Disbursement of funds proceeded quickly, with project funds fully disbursed in October 1975, one year ahead of schedule (Annex 2). Two reallocations of funds were requested by FIRA during project implementation. The first, approved January 1, 1975, reflected the inclusion of FINASA as an institution eligible for project funds, for up to $4.5 million. The second reallocation approved September 30, 1975 was necessary due to overdrafts in categories 2a and 3a (GSP components for crop production and agro-industries) (Table 4). Project Lending Program 3.21 The lending program financed under the project totalled 10,881 sub-loans as compared with 12,589 forecast at appraisal. Thus the project reached some 54,017 beneficiaries under the two sub-components (Annex 4). The distribution of loan size reveals that 70% of the loans made under the GSP component were less than US$20,000, and 19% were between US$20-40,000. For the LIPSP component, 80% of the sub-loans were for less than US$20,000 and only 7% were between US$20-40,000. Thus, of the total beneficiaries receiving sub-loans about one-fourth under each sub-component received less than US$20,000 under the project; the average loan size within this category was US$10,300 for the GSP component, and US$5,600 for the LIPSP component. The average loan size for each component as a whole was US$25,337 for the GSP and US$19,162 for the LIPSP. The GSP Component 3.22 The total number of loans financed under the GSP component was slightly lower than appraisal estimates totalling 8,971 compared with a forecast of 10,843; consequently, the average size of investment per sub- loan was somewhat higher at US$29,100 compared with appraisal estimates of US$19,900. GSP - Number of Sub-loans Average Size of Investment (US$'000) Number of Sub-loans Average Size of Investment Enterprise Forecast Actual Forecast 1/ Actual Livestock 6,489 5,151 20.3 27.2 Crops 4,290 3,420 14.8 23.5 Agro-industries 64 114 318.8 312.8 Unclassified - 286 - 18.0 TOTAL 10,843 8,971 19.9 29.1 1/ Includes total price contingency of about 9%.  MEXICO COMPLETION REPORT FOURTH LIVESTOCK AND AGRICULTURE DEVELOPMENT PROJECT LOAN 910-ME ALLOCATIONS AND DISBURSEMENTS ($ Million) Original First Second total Category Allocation Reallocation 2 Reallocation X Disbursed 2 (January 1, 1975) (Sept. 30, 1975) I. Medium and long term loans for live- stock production: (a) GSP component 50.9 50.9 47 51.4 47 51.2 99.6 (b) LIPSP component 14.4 14.4 13 9.6 9 9.6 100.0 II. Medium and long term loans for crop production: (a) GSP component 24.5 21.2 22 25.3 23 25.3 100.0 (b) LIPSP component 8.9 7.7 8 8.6 8 8.6 100.0 (c) Part 1 bis (a)!' - 4.5 1.3 1 0.1) 31.0 (d) Part 1 bis (b) 0 1.3) rll. Medium and long term loans for agroindustries: (a) GSP component 7.4 7.4 7 12.3 11 12.4 101.0 (b) LIPSP component 1.7 1.7 1 .3 - 0.3 100.0 IV. Studies, training and demonstration 2.2 2.2 2 1.2 1 1.2 100.0 TOTAL 110.0 110 100 110.0 100 110.0 1/ Category II parts (c) and (d) were added to the project agreement October 31, 1974 to provide for NAFSINA's participption in the project.  - 30 - The average size of the loan financed under the GSP component was also higher than appraisal estimates. For the livestock category, the average sub-loan totalled US$23,600 as compared with appraisal estimates of US$17,300. In the crop category, the average sub-loans totalled US$19,700 as compared with forecasts of US$12,500. Agro-industrial loans reached an average size of US$296,491 as contrasted to appraisal estimates of US$254,700. None- theless, funds loaned as a percent of investment closely paralleled appraisal forecasts:  GSP - TOTAL INVESTMENT AND FUNDS LOANED FORECAST - ACTUAL (US$ Million) % of Loans Amount of Investment Amount of Funds Loaned Loans as % of Investment Enterprise Forecast Actual Fcrecast Actual Forecast Actual Forecast Actual Beef 39 41 74.2 96.7 63.1 84.8 85 88 Dairy 14 10 42.8 26.1 36.2 21;5 85 83 Other Livestock 7 7 15.1 17.5 12.8 15.5 85 88 Total Livestock 60 58 132.1 140.3 112.1 121.8 Annual Crops 32 30 50.1 61.9 42.6 51.8 85 84 1 Perennial Crops 7 8 13.2 18.3 11.2 15.6 85 86 Total Crops 39 38 63.3 80.2 53.8 67.4 A-roindustries 1 1 20.4 35.7 16.3 33.8 80 95 Unclassified - 3 - 5.1 .- 4.4 - 86 TOTAL 100 100 215.8 261.3 182.2 227.4 Ln  - 32 - The only substantial deviation from appraisal estimates was experienced under the agro-industrial sub-loan category in which loans constituted 95% of the investment, or an increase of 15% over appraisal estimates. 3.23 An analysis of project sub-loans by enterprise type reveals a sectoral distribution of sub-loans which closely parallels appraisal esti- mates. Thus, in terms of absolute numbers of sub-loans, livestock absorbed 58% of the sub-loans granted as compared with 60% forecast at appraisal, and crop development 38% as compared with 40% forecast at appraisal. However, certain deviations did occur within investment categories; thus 71% as compared with 65% forecast at appraisal of livestock sub-loans were for beef development, with a resulting decrease in the percentage of sub-loans for dairy development (17% vs. 23% forecast at appraisal). In terms of total funds loaned under the component, actual lending by investment category closely paralleled appraisal estimates. Thus, 54% of total funds on-lent were utilized for livestock sub-loans as compared with 60% forecast at appraisal. The crop development category accounted for 30% of on-lent funds as compared with 29% forecast at appraisal (Annex 1). The LIPSP Component 3.24 The total number of loans financed under the LIPSP component was slightly higher than appraisal estimates totalling 1,910 as compared with a forecast of 1,746. The average size of the investment per sub-loan was, correspondingly, slightly lower at US$20,600 as compared to appraisal fore- casts of US$29,800. LIPSP - Number of Sub-loans Average Size of Investment (US$ '000) Number of Sub-loans Average Size of Investment Enterprise Forecast Actual Forecast 1/ Actual Livestock 372 828 80.6 25.5 Crops 1,368 920 13.5 18.7 Agro-industries 6 1 600.0 80.0 Unclassified - 161 - 5.3 TOTAL 1,746 1,910 29.8 20.6 1/ Includes price contingency of about 9%.  - 33 - The average size of the loan financed under the LIPSP component was also lower than appraisal estimates at US$19,162 as compared with US$28,350 fore- cast at appraisal. However, within the component itself substantial devia- tions occurred. Due to the large increase in the number of sub-loans for livestock development, 828 as compared with 372 forecast at appraisal, the average loan size within this category was considerably below the appraisal forecast of US$76,700, instead totalling US$23,900. The number of loans for beef development registered the largest increase within the livestock category, totalling 505, or more than two and one-half times the appraisal total of 149. Dairy development sub-loans increased by approximately 60% to a total of 274 under the sub-project. The number of sub-loans for crop development totalled 920, as compared with appraisal forecasts of 1,386. The average, however, is skewed by the large decrease in sub-loans for perennial crop development, as sub-loans for annual crop investments approxi- mated appraisal forecasts. Sub-loans for perennial crops totalled only 147 with an average loan size of US$19,500; the appraisal report forecast sub-loans of an average of US$6,800 and totalling some 645 investment. The deviation away from crop development investments, particularly for perennials, in favor of livestock development, further reflects the sectoral problem of continuing uncertainty over Government land tenure policies and programs, resulting in a reluctance to make longer term fixed investments in land develop- ment. Total funds loaned, as a percent of investment, closely parallel appraisal forecast; thus, in both the livestock and crop development categories loans as a percent of investment were 93% as contrast to 95% estimated at appraisal. LIPSP - Total Investment and Funds Loaned: Forecast - Actual (US$ million) Percent of Loans Amount of Investment Amount of Funds Loaned Enterprise Forecast Actual Forecast Actual Forecast Actual Livestock 21 43 30.0 21.20 28.5 19.70 Crops 78 48 18.5 17.20 17.6 16.00 Agro-industries 1 - 3.6 .08 3.4 .08 Unclassified - 9 - .90 - .80 TOTAL 100 100 52.1 39.40 49.5 36.60 3.25 Total amount of investment and total funds loaned under the LIPSP components were approximately 76% and 74% of appraisal estimates, respectively. At appraisal, furthermore, it had been foreseen that the LIPSP sub-project would include 6 agro-industrial sub-loans for the development of logging and saw- mills, with a total investment of US$3.6 million, and total funds lent of US$3.4 million. However, the agro-industrial component of the LIPSP fell far short  - 34 - of appraisal expectations at approximately US$800,000 total funds loaned. The reasoa for the shortfall was due to the failure of the government agency involved to complete project preparation (para 3.26). ARro-industries 3.26 The Fourth Project included a sub-component for the financing of agro-industrial investments in both the GSP and the LIPSP. Total investments under the GSP agro-industrial sub-component were forecast at US$20.4 million, with Bank financing totalling US$7.4 million. A reallocation of funds occurred, however, in September 1975, increasing the Bank's contribution under the GSP to US$12.3 million, and decreasing the LIPSP to US$300,000. Actual loans for agro-industries totalled 114 under the GSP, as compared with appraisal estimates of 64. The average size of investment per loan was slightly below appraisal estimates, and the average sub-loan size slightly above as follows: GSP - ARro-industries (US$ Million) Forecast Actual Number of Loans 64.0 114.0 Amount of Investment 20.4 35.7 Amount of Funds Loaned 16.3 33.8 Loans as % of Investment 80.3 95.0 Average Size of Investment (US$'000) 318.8 312.8 Average Size of Sub-loan (US$'000) 254.7 296.5 Approximately 11% of the total IBRD loan under the project financed agro- industrial investments. At appraisal, it had been envisaged that six sub-loans would be processed under the LIPSP component for six logging and saw mills for a total investment of US$3.6 million. These projects were to be prepared by the "Forestal Vicente Guerrero" for investments in the state of Guerrero. However, preparation of requests for financing never materialized. Thus, less than one million dollars was on-lent for low-income agro-industrial investments. 3.29 FIRA financed agro-industrial projects are generally of good quality, and are only approved after an evaluation process carried out by FIRA technicians, which eliminates 10-20% of total loan applications. However, the success of the program is attributable not only to FIRA technical staff, but also to the fact that prospective borrowers have considerable entrepreneurial ability and are able to prepare sound investment packages. FIRA's role is basically limited to a review based  - 35 - on information prepared by prospective borrowers. As a result, FIRA financing has tended to concentrate on large established agro-industrial producers; intervention in the ejido sector is practically non-existent due to the ejidos lack of experience in identifying and preparing projects for submission to FIRA. Furthermore, the lack of useful and easily accessible data on existing agro-industries, collected according to uniform standards among all FIRA branches, is an area where considerable gains in time efficiency and analytical strength could be accomplished. 3.28 FIRA has a total staff of 36 technicians with expertise in agro- industries of which 35 are graduate agricultural engineers. The regional distribution is well matched to sectoral requirements: FIRA Agro-industrial Technicians Region Number of Technicians Central Office (Mexico City) 8 Northwestern 4 Northern 6 Northeastern 4 Western 3 Central 3 Southern 3 Guerrero-Michoacan I Yucatan Peninsula 1 TOTAL 36 Future strengthening of FIRA's agro-industrial unit should provide for FILRA's intervention earlier in the project cycle. Although FIRA technicians do an effective job in assuring that only viable projects are financed, earlier participation in project identification could not only provide technical assistance to borrowers in assessing alternate sources for equipment, but also could lead to less capital intensive projects which increase the potential for much-needed employment in Mexico. In addition, FLRA tech- nicians should follow-up on industries financed in order to develop a perception of the start-up and operational problems in sub-loan projects.  - 36 - IV. Project Impact On-farm Investment 4.01 FIRA records reporting investment items financed under the project reveal certain systematic deviations from appraisal forecasts. However, FIRA data have been revised in this report as FIRA includes certain plant materials in the breeding stock category, thereby tending to overinflate these figures. Based on a revised classification, the following pattern was observed: GSP Component Analysis of Revised Reported Investment Items (US$ Million) Land Machinery Constructions Breeding Enterprise Improvement Irrigation and Equipment Installations Stock Total Livestock 4.900 6.300 11.400 18.300 80.4 121.2 Annual Crops 20.100 7.300 20.600 3.700 - 51.6 Perennial Crops 8.600 2.900 2.800 1.300 - 15.6 Agro-industries 2.900 0.032 31.500 0.008 - 34.4 Unclassified 0.088 0.248 0.872 0.376 2.8 4.4 TOTAL 36.500 16.800 67.200 23.600 83.3 227.3 1/ Totals may not sum due to rounding. Of the investment financed with loan proceeds, 53% of the total is attribut- able to livestock operations; within these units, 66% of the investment financed breeding stock and 4% and 5% respectively were attributable to land improvment and irrigation. For annual crops, which accounted for 23% of the financed investment, 40% of the total is attributable to machinery, 39% to land improvement and 14% to irrigation. The total number of units financed under the component was as follows:  - 37 - GSP Component Units Financed Land Improvement 299,251 Irrigation 3,897 Machinery and Equipment 3,476 Construction and Installations 12,731 Breeding Stock 221,985 For livestock units, this represents a strong particularly deviation of investment away from land improvement and irrigation in favor of breeding stock. GSP Component Analysis of Reported Investment Items Forecast - Revised Reported Percent Livestock Enterprises Forecast Reported Land Improvement 21 4 Irrigation 4 5 Machinery and Equipment 7 10 Construction and Installations 13 15 Breeding Stock 55 66 TOTAL 100 100  - 38 - 4.02 A similar pattern occurred under the LIPSP sub-component: LIPSP Component Analysis of Revised Reported Investment Items (US$ Million) Land Machinery Constructions Breeding Enterprise Improvement Irrigation and Equipment Installations Stock Total Livestock 0.248 0.576 2.200 3.200 13.500 19.700 Annual Crops 2.900 4.500 5.400 0.312 - 13.200 Perennial Crops 2.300 0.248 0.256 0.096 - 2.900 Agro-industries 0.016 - 0.088 - - 0.104 Unclassified 0.016 0.032 0.272 0.008 0.400 0.816 TOTAL 5.400 5.400 .8.200 3.600 13.900 36.500 Thus, 54% of the total investment financed is attributable to livestock enterprises, and for these units, 69% of the total was attributable to the breeding stock category. 36% of the total LIPSP investment was classified as annual crop units; within these units, 41% of the investment financed machinery and equipment. Again, the deviation from appraisal forecasts is clear: LIPSP Component Analysis of Reported Investment Items Forecast - Revised Reported Percent Livestock Enterprises Forecast Reported Land Improvement 24 1 Irrigation - 3 Machinery and Equipment 17 11 Construction and Installations 21 16 Breeding Stock 38 69 TOTAL 100 100  - 39 - The total number of units financed through the LIPSP on-lending program was: 49,448 for land improvement; 528, irrigation; 1,170, machinery and equip- ment; 3,389, construction and installation; and 46,986, breeding stock. 4.03 The pattern of systematic deviation away from appraisal forecasts for investment by category reflects the continuing reluctance of farmers to invest in land improvements due to the uncertainty over the agrarian reform law. This trend was observed in previous projects and will continue so long as the sectoral issue of land tenure is largely unresolved in Mexico. In addition, as the LIPSP component was initiated by the project, it was more difficult to anticipate the investment pattern in these units. 4.04 Project loans were utilized to purchase 269,000 head of cattle worth approximately US$108 million equivalent (US$401 per head) or about 135,000 head worth US$54 million equivalent per year (Annex 3). Cattle imports during 1974 and 1975 totalled 57,000 head valued at US$34.6 million (US$532 per head) but complete data is not available to indicate FIRA financed imports. In 1972 the national beef and dairy cattle herd was estimated at 26.3 million head, and the annual increment from 1972 to 1975 is estimated to have averaged about 700,000 animals. This data suggests that FIRA was financing about 19% of the annual increment to the national herd. On the basis that 50% of the annual herd increment comprised breeding animals, on average FIRA would have financed about 40% of these. Based on the following assumptions: 40% cows in national herd, and an 18% heifer replacement rate, about 2 million heifer replacements would be needed in 1975 in order to maintain the national herd. Thus, FIRA financed cattle represented about 7% of annual heifer replacements. On-farm Development 4.05 Under the Fourth Project, a total of 10,881 sub-loans were granted on the basis of farm development plans prepared by FIRA technicians. The monitoring of the on-farm impact of the program was to be undertaken by the monitoring unit established under the project in FIRA. Establishment of the unit (para 3.04) proceeded extremely slowly, and methodological as well as analytical problems have been encountered in its operation. Consequently, data on the actual impact of on-farm investments are in a preliminary stage of development at the time of project completion. 4.06 FIRA completed a preliminary survey of on-farm technical and mana- gerial skills and services among small farmers and ejido groups. The study was particularly geared to assess: (i) the existing quality of on-farm record keeping; (ii) the quality and adequacy of technical assistance provided by lending banks to the farm; and (iii) the degree to which the farm development plan prepared by the technicians was being followed after sub-loan appraisal. The study was based on a total of 273 pequenos proprietarios and 117 ejido farms. The results of the survey closely parallel observations made in earlier evaluation efforts. 7/ The study showed most sample farms had inade- quate or non-existant systems for farm record keeping despite technical assist- ance received from the participating banks and FIRA at appraisal. Within the 7/ See OED report Sec M77-353.  - 40- pequenos proprietarios group, 81.% of the farms were classified as having no record control systems at all, or systems which were ineffective. Some 15% of the farms had systems which were classified as fair, while only 4% were considered to have systems regarded as satisfactory. Within the ejido group the pattern was similar--68% of the ejidos had either non-existent or in- effectual systems, while 38% were classified as fair and only 4% as satis- factory. Within both groups the standard of record keeping was best for annual crop farms and dairy enterprises and worst for perennial crop farms and beef enterprises. 4.07 The survey also investigated the quality and quantity of technical assistance provi-ded by participating banks to borrowers, particula-rly within the context of the implementation of on-f atm development plans. Within the pequenos proprietarios group, the level of technical assistance was poor-- only 15% of the farmers were judged as having received adequate technical assistance with the credit package, and only one farm out of the total was judged as having implemented its development program as prescribed by the investment plan. Ejido investments had a better record, receiving more technical assistance from FIRA and the participating banks--54% of the sample were classifie-d as having received satisfactory assistance, although only 7% were found to have carried out their development plan as prescribed. Collectivized ejido operations were more efficient in the use of scarce technlical manpower resources indicating that the grouping of low-income farmers might provide one mechanism for conserving limited technical assis- tance efforts in future operations. However, it is clear that FIRA, as a banking institution, is constrained in its ability to allocate resources and personnel for technical assistance to small farmers. FIRA continued its efforts to shift some of the technical assistance burden onto the participating banks, and provided training to BE and BA staff in sub-loan appraisal (para 3.02). However, the FIRA study clearly indicates that technical assistance is required and in its absence, considerable deviation from farm investment plans is likely to occur. The technical assistance required by low-income producers will increasingly have to be provided through the development of research and extension facilities complementary to FIRA. Such assistance will be required to ensure that the technical changes in productivity envisaged at sub-loan appraisal actually occur. 4.08 In September 1977, FIRA prepared a first report summarizing the monitoring unit's survey of 276 lown-income producers and 117 ejidos. The data are based on information from 1973-74, before the farms received credit, and December 1976, the second year of the respective sub-loans. These data are hardly indicative of the impact, at full development, of project sub-loans, as they reaflect a poinat in time relatively early in project implementation. However, they provide some financial information on low-income farms and data reflecting productivity.  - 41 - 4.09 The FIRA monitoring unit also generated data on the physical and technical progress experienced on sub-loan farms. The data on livestock opera- tions is far from complete, particularly for ejido operations, but it does provide information on basic changes in certain productive activities on- farm. Smallholder beef ranches increased their carrying capacity relatively little from 0.7 A.U./ha. to 0.8 A.U./ha; dairy farms did somewhat better increasing their carrying capacity from 0.5 A.U./ha. to 0.8 A.U./ha. (Table 7). Milk production per cow showed little change, although the value of milk and meat output showed substantial increases. The average herd size increased from 31 to 54 cows on the beef enterprise, and from 21 to 28 on the dairy farms. Ejido beef enterprises increased their average improved pasture area from 1,989.4 to 4,147 ha; however, carrying capacity rose only marginally from 0.7 A.U./ha to 0.9 A.U./ha. Ejido dairy farms were all new operations and therefore no "before project" comparisons can be made. Output per cow in ejido dairies averaged 10.3 liters per cow per day. 4.10 The data on crop farm investments indicates that the expansion in areas cultivated was substantial for all categories except ejido beef farms which registered a decline for both annual and perennial crops (Table 8). For smallholder enterprises the percentage increases were as follows: Smallholder Sample Percent Increase in Area Cultivated (ha) Enterprise Annual Crops Perennial Crops Beef 106 106 Dairy 13 -27 Annual Crops 12 212 Perennial Crops 20 140 Mixed 47 248  LOAN 910-ME: FOURTH LIVESTOCK AND AGRICUL1URAL DEVELOPMENT PROJECT ANALYSIS OF CROP INVESTMENT ACTIVITIES ON SAMPLE FARMS Value of Output Value of Output in Value of Increased labor use Area in In Annual Crops Permanent Crops Area in Permanent Crop Sales per farm (man/years) Farm Type Annual Crops (ha) (constant ,000 pesos) (constant .000 pesos) Crops (ha) (constsnt ,000 pesos) Family Hired Before After Before After Before After Before After Before After Permanent Temporary Smaliholders Beef Cattle 411.0 850.1 586.9 789 177.0 460.6 83.9 172.8 0.2 0.3 0.3 Dairy 258 291.2 418 312.9 50.0 75.4 20.4 16.0 - 0.5 0.6 N/A Annual Crops 2,205.9 2,466.5 3,298.5 3,683.7 84.8 167.7 18.1 56.5 0.1 - 0.2 Permanent Crops 55 66 119.5 103.2 530.4 812.1 78.9 189.6 0.5 - -0.2 Mixed 274.9 403.1 617.3 937.5 143.7 176.7 28.2 98.3 0.3 0.1 0.7 4- Average 3,204.8 4,076.9 5,040.7 8,569.0 985.9 2,389.2 229.5 533.2 5,936.8 7,353.0 0.2 0.2 0.2 FRidos Beef Cattle 800 150 1,482.5 37.6 650.7 28 96 29 1.9 0.2 3.1 Dairy 2,460 3,483 13,491.7 22,710.2 1,317.6 1,004.2 500 434 -1.4 1.2 4.9 Annual Crops 2,886 5,039 12,243.4 16,594.9 204.9 64.1 67 134 N/A 2.7 0.2 0.9 Permanent Crope 577 223 501.4 184.8 66.4 1,030.6 41 318 0.1 4.3 2.1 Mixed 178 176 286.7 164.9 - 117.5 61 83 -0.6 1.6 2.4 Average 6,902 8,971 28,007.7 39,693.0 2,239.6 2,244.6 704 998 29,613.5 30,644.0 1.2 0.8 1.9 Source: FtUO, Primer Informe Sobre la Evolucin de la& Empresas Acteditadas Balo el Programs de Productore. de Bae oo Ingreeo Segunda Porte, Agoeto, 1977.  MEXICO LOAN 910-ME: FOURTH LIVESTOCK AND AGRICULTUIAL DEVELOPMENT PROJECT Analysis of Livestock Investment Activities on Smple Farms Value of Milk and Heat and Milk Sales Total Pastures (ha) Carrying Capacity Mtlk Production Meat Output/ha per farm Average Herd Size before After (A.U./ha) (It/cow/day) (constant pesos) (constant ,000 pesos) (A.U.) Fare Type Natural lrvd Natural Improved Before After Before After Before After before After Before After Smaelholdera Beef hattle 52.2 23.7 55.1 28.7 0.7 0.8 2.7 2.6 337 516 32.4 51 31 54 Dairy 5.8 - 3.3 2.3 0.5 0.8 6.1 5.7 2,691 4,153 64.1 104.2 21 28 Annual Crops 1.8 2.1 232 350 16.2 24.8 19 23 Permanent Cropa N/A 2.1 501 916 - 47.9 - 50 Mixed 3.4 3.4 721 1,393 24.8 54.5 18 28 Kuidos Beef Cattle 5,623.1 1,989.4 4,442.0 4,147. 0.7 0.9 27.5 174.9 84 212 Dairy N/A 10.3 N/A 559.3 N/A 146 Annual Crops Permanent Crops hixed So rce FONDO, Primer Informe Sobre Ia Evelueain de la% Elpreas Acredttadas Bala el Programa de Productores de Bhaos Ingresos- Sesunda Parte, A8osto, 1977.  - 44 - Increases in cropped area were substantial then, on smallholder beef enter- prise as well as mixed, annual and perennial crop farms. Dairy enterprises performed worse registering only a 13% increase in annual crop acreage. The ejido farms showed a similar pattern, although the beef units registered a lesser increase, while dairy units performed relatively worse with an increase in perennial crop acreage of 36% and a slight decrease in annual crop acreage. Eiido Sample Percent Increase in Area Cultivated (ha) Enterprise Annual Crops Perennial Crops Beef -81 243.0 Dairy 42 -16.2 Annual Crops 75 50.0 Perennial -60 13.0 Mixed - 36.0 The ejido sample reflects more extreme changes in patterns of cultivation; however, total land under cultivation increased in all units except for beef enterprises and perennial crop farms. 4.11 Yield data are organized by crop and on the basis of a small sub-sample of both smallholders and ejido units. In general, changes in yields on smallholder farms were not large: Smallholder Sub-sample Yields (ton/ha) Enterprise Units Sampled Before After Crop: Maize Beef 13 2.1 2.1 Dairy 6 2.4 2.1 Annual Crops 26 1.4 1.6 Perennial Crops 5 1.5 1.2 Mixed 12 2.5 2.8 Crop: Beans Annual Crops 10 0.6 0.6 Crop: Sorghum Annual Crops 13 3.6 4.6 Crop: Alfalfa Annual Crops 8 49.9 38.2 Crop: Wheat Annual Crops 6 1.6 1.8  - 45 - It is clear from the proceeding table despite problems of the exact compara- bility of before and after situations that changes in yields have not been substantial; for sorghum, wheat, and maize some improvement was registered. However, it also must be noted that the project sample is extremely small, and was taken within the first two years of sub-project investments. In addition, as in the case of alfalfa, the sub-sample does not cover the most relevant type of sub-project, (dairy rather than annual crop units) where technical improvements in productivity would be most likely to occur. 4.12 With the ejido units, a similar pattern can be viewed: Elido Sub-sample Yields (ton/ha) Enterprise Units Before After Crop: Cotton Annual Crops 13 2.5 2.6 Dairy 9 2.3 1.8 Crop: Maize Annual Crops 6 2.9 2.2 Beef 5 1.3 0.7 Again it is clear that the data are both tentative and incomplete; however, changes in productivity have been minimal and even negative in some cases. 4.13 While the monitoring system has produced some useful data on financial performance and technical productivity, few comprehensive conclu- sions can be drawn from these preliminary results. The sample is a small one, taken early in the life of the project, and the data collected are expressed in financial terms which in the absence of more detailed tech- nical information on the physical progress of sub-loans are not adequate indicators of on-farm development. In addition, the deflation of figures by the consumer price index is a conceptually less than optimum mechanism for determining constant dollar changes in the value of input and output. Clearly these early data reflect only modest changes in on-farm productivity and tight financial and cash flow positions of low income producers. How- ever, whether these modest gains will improve during the development period, or not, cannot be directly extrapolated, particularly in the absence of supporting information analyzing why such patterns have occurred. 4.14 The system of data collection itself, as is clear from these early efforts, will require certain modifications in methodology. FIRA is planning to increasingly measure physical rather than financial measures of on-farm change, while limiting its sample to main areas of FIRA financing. Expansion  - 46 - of the sample to cover medium income farmers is also planned. In addition, technical coefficients for sub-borrower performance will have to be made more complete and linked to the utilization of loan funds and on-farm develop- ment plans prepared at the time of sub-loan appraisal. Rates of Return 4.15 The Appraisal Report estimated the economic rate of return of the project at 29%. In attempting to estimate the actual rate of return, yield and production data were taken from the latest information collected by the monitoring unit in 1977 from samples of ejido and smallholder units, and from a special study undertaken by the livestock department of FIRA on productivity in animal production units. Input and investment cost data are based on the information presently available which was also utilized in preparation of the Sixth Credit Project (Loan 1569-ME) which was appraised at the end of 1977. Given the incomplete nature of the data available from the monitoring unit and the fact that it is still too early to judge the results of some types of lending (such as perennial crops), the overall rate of return for the project can only be tentatively estimated; the best judgement is that it is about 18%. As under the Third-Project, investments in beef cattle continued to have the lowest rate of return, but to constitute the largest category of lending once again reflecting farmers' reluctance to invest in fixed assets such as land improvement in view of the unresolved situation with regard to the agrarian reform law. Beneficiaries 4. 16 Although the total number of GSP loans fell short of appraisal estimates, the number of beneficiary families under the component totalled 30,465, far exceeding appraisal forecasts of 11,000 families (Table 6). The increase in beneficiaries was due to a deviation away from appraisal expec- tations that under the GSP loans would nearly always be made to individuals. In fact, 68 sub-loans were made to ejido groups, 136 sub-loans were made to private sociedades, and 1,526 sub-loans to private grupos solidarios. Such a pattern reflects the income disparities which exist between ejidos, as well as the Government's policy during the project implementation period to encourage loans to farmer credit unions and other groups of farmers. Thus, some 40% of the total funds loaned under the GSP were received by some 23,220 farmers organized into groups. It is probable that the large number of members in sociedades is due, in part, to the inclusion of the agro- industrial component in the figures, which may tend to inflate the figures slightly. 4.17 Under the LIPSP, 792 out of a total of 1,910 sub-loans were attri- butable to group lending. Thus, 86% of the total funds lent were received by groups of various kinds. The component reached slightly more beneficiaries  - 47 - (23,552) than appraisal forecast at 22,600. The ejidatarios received 74% of the total, and accounted for 21,371 beneficiaries. Although the "Grupos Solidarios" received a larger number of loans (314), the "Sociedades Locales de Credito Ejidal" received the largest amount of funds, totalling Mex$ 236.9 thousand, and accounted for the largest number of beneficiaries, 16,456 or 70% of the total under the sub-project. The Government policy to organize large ejidos in the southeast region contributed to the importance of credit societies as a channel for medium term credit ejidatarios. 4.18 The GSP sub-project was envisaged as countrywide in scope at appraisal; in fact, the distribution of project costs was quite even, with the South accounting for the smallest proportion, 8%, and the Northeast the largest portion of total project costs at 22%. The LIPSP, in contrast, was expected to be concentrated in the "poorest regions", such as the states of Oaxaca, Guerrero, Queretaro, and Aguas Calientes. In fact, 36% of total project costs were attributable to the north region, 20% to the west, and 15% to the south indicating a slight deviation from appraisal forecasts. V. Bank Performance 5.01 The appraisal mission for the Fourth Project concentrated its efforts on the preparation of an innovative component for lending to low- income farmers, but also continued to address itself to FIRA's overall financial and institutional development. As in the Third Project, the Fourth Project appraisal report contained a number of detailed representa- tive farm models for beef, crop and agro-industrial investments. Supervision of the project absorbed a total of 47 manweeks, with missions focusing much of their attention on institutional development within FIRA, particularly the operation of the monitoring unit, problems encountered with the procure- ment of agro-industrial machinery (para 3.15) and the progress of the on- lending program to GSP and LIPSP sub-components. Supervision missions had neither the time nor resources to directly follow up on the progress of on-farm investments, nor to make field assessments of the level of tech- nical assistance being received by farmers and its impact on farming practices. 5.02 An important issue faced by the Bank during the otherwise smooth and rapid implementation of the project was the establishment and operation of the monitoring unit. The Bank's goal in requesting the creation of such a unit was based on experience in the previous three projects, which was marked by a lack of information on the impact and effectiveness of the investment program on-farm. Data collected through a consultant's evalua- tion in 1972 indicated that, although the financial gains experienced by producers appeared satisfactory, there was a need to gather more compre- hensive data on the physical productivity of sub-borrowing farms. Certainly, given the size and importance of the FIRA program, the creation of a monitoring unit within FIRA was a sound objective. However, the appraisal report did not take into account the practical difficulties inherent in the establish- ment of such a unit, particularly in following small scale farmers who keep few or no records. The report also did not attempt to outline the basic principles and objectives of such a program at the outset of the project  MEXICO LOAN 910-ME-FOURTH LIVESTOCK AND AGRICULTURE DEVELOPMENT PROJECT Loan Distribution by Type of Borrower Number Number of of Number of Beneficiaries Loan Average Average per Loans Beneficiaries per Loan Total per Loan Beneficiary