Mexico - Rural Financial Markets (Vol. 2 of 2) : The Annexes

Document of The World Bank FOR OFFICIAL USE ONLY CONFIDENTIAL 14599-ME VOL. 2 MEXICO RURAL FINANCIAL MARKETS VOLUME OF ANNEXES August 25, 1995 FILE COPY Natural Resources and Rural Poverty Division Country Department II Latin America and the Caribbean Regional Office This document has a restricted distribution and may be used by reciplents only in the.performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. CURRENCY EQUIVALENTS Currency Unit = Mexican New Peso (N$) US$1 = N$3.41 (July 31, 1994) US$1 = N$6.15 (May 31, 1995) FISCAL YEAR January 1 - December 31 WEIGHTS AND MEASURES 1 meter (m) = 3.28 feet (ft) 1 kilometer (kIn) = 0.62 mile (mi) 1 hectare (ha) = 10,000 m' = 2.47 acres 1 square 'kilometer (km) = 0.38 square miles (mi?) = 100 ha 1 metric ton (m ton) = 2,205 pounds ABBREVIATIONS AND ACRONYMS AGEB Census Enumeration District (Area Geoestadistica BAsica) ANAGSA National Agricultural Insurance Company ASERCA Support Services for Agricultural Marketing (Apoyos y Servicios a la Comercializaci6n Agricola) BANRURAL National Rural Bank (Banco Nacional de Cr6dito Rural) BANXICO Bank of Mexico (Banco de Mexico) CANAFO Mexican Development Savings Bank (Caja Mexicana de Fomento) CETES Treasury Bills Rate CFI Chartered Nonbank Financial Intermediary CNB Superintendency of Banks (Comisi6n Nacional Bancaria) CONASUPO National Food Supplies Company (Compatiia Nacional de Subsistencias Populares) CP Credit Cooperative (Caja Popular) CV Coefficient of Variation FRE Farm Rural Entrepreneur FEFA Special Agricultural Trust Fund (Fondo Especial para Financiamientos Agropecuarios) FEGA Technical Assistance and Loan Guarantee Trust Fund for Agriculture (Fondo Especial de Asistencia T6cnica y Garantia para Cr6ditos Agropecuarios) FERTIMEX Mexican Fertilizer Company (Fertilizantes de Mexico) FICART Trust Fund for Rainfed and Irrigated Areas (Fideicomiso para Cr6dito en Areas de Riego y de Temporal) FIRA Trust Fund for Agriculture (Fondo de Garantia y Fomento para la Agricultura, Ganaderia y Avicultura) FIRCAVEN Trust Fund to Restructure Overdue Portfolio (Fideicomiso para la Restructuraci6n de la Cartera Vencida) FIRCO Trust Fund for Shared Risk (Fideicomiso de Riesgo Compartido) FOCIR Trust Fund for Rural Infrastructure) (Fideicomiso para Infraestructura Rural) FONAES National Fund for Solidarity Enterprises (Fondo Nacional de Empresas de Solidaridad) FOPESCA Fisheries Guarantee and Trust Fund (Fondo de Garantia y Fomento para las Actividades Pesqueras) GATT General Agreement on Trade and Tariffs GOM Government of Mexico INEGI National Institute for Statistics, Geography and Informatics (Instituto Nacional de Estadistica, Geografia, e Informitica) NAFTA North American Free Trade Agreement NAFIN National Finance Bank (Nacional Financiera) NFRE Non-farm Rural Entrepreneur NGO Nongovernment Organization PROCAMPO Farm Support Payments Program (Programa de Apoyos Directos al Campo) PRONASE National Seed Production Company (Productora Nacional de Semillas) PRONASOL National Solidarity Program (Programa Nacional de Solidaridad) RE Rural Entrepreneurs RFMs Rural Financial Markets ROSCAS Rotating Savings and Credit Associations SAP Savings and Loan Association (Sociedad de Ahorro y Pr6stamo) SEDESOL Secretariat for Social Development (Secretaria de Desarrollo Social) SHCP Secretariat of the Treasury (Secretaria de Hacienda y Cr6dito Piblico) TCs Transaction Costs UC Credit Association (Uni6n de Cr6dito) CONTENTS Annex A. Rural Entrepreneurs Annex B. The Characteristics of Available Credit Services in Three Selected Regions Annex C. The Participation of Rural Entrepreneurs in Financial Markets Annex D. Rural Credit Markets as Mechanisms to Cope with Risk Annex E. Case Studies of Nonbank Rural Lenders This report was written by Rodrigo A. Chaves (Task Manager, LA2NR) and Susana M. Sanchez (Consultant). It benefitted greatly from the comments of Michael Baxter, Silvia Castro, Louise Cord, Carlos Cuevas, Paul Holden, Frank Lysy, Michael Lubrano, William Maloney, David Nielson, Andrea Silverman, Jennifer Sobotka, and Jacob Yaron. The Division Chief is Michael Baxter, the Lead Economist is Frank Lysy, and the Country Department Director is Edilberto Segura. ANNEX A THE RURAL ENTREPRENEURS - A.2 - Economically Active Population 1. Approximately 56 percent of adults (16 years or older) in the regions surveyed are economically active. As figure 1 shows, 40 percent of adults are wage-earners, while 16 percent are entrepreneurs (REs). Of the working population, 80 percent are employed in nonagricultural enterprises, mostly as salary earners. The bulk of farm workers are jornaleros earning daily wages. Rural Etopronou Entrepreneurs Econondecy inactiva 44% Form 2. The *-' majority of REs, the focus of this study, are engaged in nonfarm activities. About "" 55 percent are 1 involved in Figure Al. Occupation of Adult Residents in Surveyed Areas nonfarm ventures, 37 percent in farm enterprises, and 8 percent in both. The most important nonfarm activities are commerce, services, and manufacturing. About 27 percent of REs are connected with trading and commerce, 17 percent provide services, and 13 percent have manufacturing or cottage industry ventures. A comparatively small group of REs are employed in construction, mineral extraction, fishing, and silviculture, in that order of importance. 3. RE 78% Nonindigenous 70 people make up 84 percent of the so regions' 4o population and 74 30 2 24% percent of REs. 2 Indigenous people constitute 16 Fomlo Mnato Totwe percent of the o indigonous O Non-Mndigonous population and 24 _E_I percent of REs. Figure A.2 Distribution of REs by Gender and Ethnicity About 27 percent of all REs are women. Female entrepreneurs are most numerous in Guanajuato, where they make up 31 percent of REs, and are least numerous in - A.3 - Veracruz, where 23 percent of REs are women. In the category of ethnicity and gender combined, 52 percent of REs are nonindigenous males, 22 percent are nonindigenous females, 20 percent are indigenous males, and 4 percent indigenous females. 4. Gender and ethnicity are associated with distinct patterns -- of occupation among REs. ~ ~~ Males, for example, account for 93 percent of 0 10 20 3D 40 sD W0 70 so 90 100 the REs in farming, while I_1___ U_0. ____F_7 females make up Figure A3. Distribution of REs by Economic Sector and 39 percent of Gender those in nonfarm businesses. Women are particularly active in commerce and trading, accounting for 46 percent of REs in this activity. Furthermore, 81 percent of REs who are indigenous and male are in farming, whereas only 19 percent of indigenous female entrepreneurs have gone into this activity. At the same time, 48 percent of REs who are nonindigenous and male have taken up farming, but only 10 percent of nonindigenous female REs have done so. 5. Most REs are small entrepreneurs, as is evident from the low number of employees they hire and their general lack of business records. Indeed, 85 percent of all nonfarm REs and 93 percent of farm REs run their operations by themselves without the help of permanent employees. Close to 52 percent of farm REs do not even use seasonal workers. Only 6 percent of all nonfarm REs have more than one employee, while only 2 percent of farm REs have more than one permanent employee. 6. Most RE None enterprises are 6 Formal Accouting simple enough or small enough not to require formal accounting O3%e records. Only 16 percent of R.Es keep such records. In view Personal Notes 17% of the nature and Figure A4. Accounting Practices Among REs A.4 - size of RE ventures, it is not surprising that 63 percent of REs keep no written records whatsoever and that 17 percent keep only informal personal notes. Most REs have small businesses and hire few or no employees outside their family circle. Because they make all the decisions themselves and have control of the assets (e.g., cash), they do not need internal control mechanisms or sophisticated information systems to keep track of their activities or inventory. 7. Of the pure farmers, only 4 percent keep formal accounting records. In contrast, 29 percent of merchants and traders maintain such records and are the largest group to do so. 8. About 80 percent of REs belong to organized groups such as guilds, unions, and chambers of producers; 73 percent also state that they are registered by a government agency such as a municipal treasury or the Secretariats of Commerce (SECOFI) and Health, while 30 percent are registered by the Secretaria de Hacienda y Cr6dito Piblico (SHCP). Farm Rural Entrepreneurs (FRIEs) 9. Forty-five percent of REs are involved in farm or agricultural ventures. The survey distinguished between landed and landless FREs. Landed producers account for 72 percent of all FREs. The most common landed FREs in the three regions as a whole are private smallholders (i.e., pequefios propietarios), who account for 38 percent of FREs, while ejidatafios account for 27 percent. 70 60 50 40 20. - . 10 0 - - Pog. Ejidatolo Oltho Sharocroppe Ranted Land Bonowed Otho, Propiotoio Londod LAnd Landless 0 Guanewoto O Puable O0 Ve,cwr 0 Total Figure AS. Tenancy Regime of Land for Farm REs by Region (Percentage of Farm REs) 10. The relative importance of the different land tenancy regimes varies significantly across the surveyed regions, however. Private smallholders represent only 12 percent of FREs in Guanajuato, for example, whereas in -A.5 - Puebla they represent 62 percent. In contrast, ejidatarios make up 49 percent of all FREs in Guanajuato and only 11 percent in Puebla. In Veracruz, FRE land tenancy regimes are more evenly balanced as private smallholder and ejidatarios constitute 39 percent and 21 percent of FREs, respectively. 11. The production technologies used by FREs are on average rather simple: 24 percent use mechanized equipment, 67 percent use only manual tools, and the remaining 9 percent use animal traction. The most commonly used inputs are chemical fertilizers, employed by 61 percent of FREs; herbicides and insecticides, applied by 40 percent of FREs; high-yield seed varieties, planted by 23 percent of FREs; and natural fertilizer, utilized by 18 percent of FREs. However, a significant proportion of FREs (23 percent) employ traditional agricultural technologies that require none of the inputs just mentioned. 12. There is a significant relationship between land tenancy regimes and the technologies employed by FREs. Fifty-three percent of ejidatarios use machinery in their activities, while only 12 percent of pequefios propietarios do so. Thirteen percent of ejidatarios employ traditional technologies without modern inputs (e.g., fertilizers, insecticides) while 23 percent of pequelios propietarios do so. 13. The average FRE holding is fairly small, since 85 percent of FRE farming businesses operate on less than 10 hectares. Thirty percent of these holdings are 1 hectare or smaller, and more than a third are between 1.1 and 5 hectares in area. Farm area is concentrated in a few FREs, however: the 30 percent of farms in the smallest category (1 hectare or less) account for only 3 percent of the farming area available to FREs, whereas the 7 percent in the largest group (25 hectares or more) account for 54 percent of the farming area. Irrigation is practiced mainly on medium and large farms. The 25 percent of farms from 5 to 25 hectares account for 45 percent of irrigated areas and the 8 percent covering 20 or more hectares account for 32 percent of irrigated land in the hands of FREs. Nonfarm Rural Entrepreneurs (NREs) 14. More than half (55 percent) of REs are involved in nonfarm enterprises, principally trading, services, and manufacturing. Forty-four percent are in trading, 27 percent in services, and 20 percent in manufacturing and cottage industries. Of the remaining 9 percent of NREs, 5 percent are in construction, 3 percent in fishing and silviculture, and 1 percent in nonmetallic mining. - A.6 - 20% 27' 44% O mnufactwe 0 construction 0 Trading 0 Services 0 othcr Figure A6. Distribution of Nonfarm REs by Economic Activity (Percentage of Nonfarm REs) 15. About 90 percent of the NREs generate profits in their operations. The proportion that do so is uniform across the main subsectors: 87 percent of traders and merchants; 91 percent of providers of services; and 90 percent of manufacturers. 16. Sixty-four percent of NREs do not have premises for their businesses. Premises are most common among traders and merchants: 50 percent of traders have such premises. The remaining 50 percent of traders work in improvised local markets (tianguis) (10 percent); the streets or in the houses of their costumers (13 percent); and at NRE homes (25 percent). Thirty-one percent of manufacturers have specific business premises for their ventures. The rest-roughly two thirds--operate cottage industries from their own homes. Twenty-nine percent of service providers have business premises. The remaining 71 percent operate from various places: improvised local markets (9 percent); the streets (this category includes transportation services) (17 percent); client homes (22 percent); and NRE homes (21 percent). 17. Equity investments are by far the most important source of funds for starting up nonfarm rural enterprises. Three-quarters of NREs financed their business setup costs with personal savings. In fact, only 13 percent of NREs received loans to start their businesses, and 80 percent of such loans were granted by friends or relatives. Commercial credit from suppliers and costumers was a significant source of startup funds for less than 4 percent of NREs. -A.7 - Did not Need 20% Other Sources 5% Commercial Credt 4% Own Resources 58% Loans 13% 0 10 20 30 40 50 60 70% Figure A7. The Most Important Source of Funds to Start Up Nonfarm Enterprises (Percentage of Nonfarm Rural Enterprises) ANNEX B THE CHARACTERISTICS OF AVAILABLE CREDIT SERVICES IN THREE SELECTED REGIONS - B.2 - The Relative ffmportance of the Formal and ffnformal Sectors 1. Three kinds of credit products or services were considered in this study: cash loans, which are credit transactions disbursed and repaid in cash; commercial credit, which includes all credit provided in kind and paid back in cash; and forward sales or sales with a down payment, which consist of payments received by the borrower in exchange for the future delivery of products and services. 2. Providers of financial services are classified as either formal or informal, according to whether they have a government-granted license to provide financial services or operate under the auspices of specific legislation. 3. The formal sector includes commercial banks, development banks, chartered nonbank financial intermediaries, and other registered institutions. Chartered nonbank intermediaries are nonbank financial institutions that have been granted an operational license by the Comisi6n Nacional Bancaria (CBN) or by the Secretarfa de Hacienda y Cr6dito Pi6blico (SHCP) and those intermediaries that have been allowed to operate without licenses but under the auspices of a particular piece of legislation. The most important examples are Uniones de Cr6dito (UCs), Sociedades de Ahorro y Prdstamo (SAPs), and Cajas Populares (Cps). Other registered institutions are organizations that provide credit but are not financial intermediaries. The main example in this category is the Instituto Nacional Indigenista (INI). 4. Informal lenders consist of moneylenders, who provide cash loans in exchange for explicit interest payments; friends and relatives, who provide cash loans without explicit pecuniary compensation; and traders, merchants, and processors, who engage in commercial credit or sales with a down payment as defined above. 5. The C30% Trannctlons 0% Amount relative market shares of 70 different credit products used by REs (e.g., cash loans, commercial credit) and 10 sectors (e.g., a Formcl Monaylandor Friends & Commorcid Forawd formal, informal) ss.o RokOvan Crdn sOO are measured in Figure B1. Relative importance of formal and informal sectors in three selected rural areas - B.3 - terms of the total number of individual transactions and in terms of the total volume of transactions. Individual transactions provide information on the coverage of REs and their access to the financial sector; and the volume of credit determines the weighted characteristics of the supply of credit available in the areas studied. Number of Individual Transactions 6. Individual transactions in the informal sector outnumber those in the formal sector by a margin of seven to one. The informal sector accounts for 87 percent of individual cash loans, commercial credit transactions, and forward sales. Figure 1 summarizes the shares of the informal and formal sectors in the rural credit markets of the regions studied. 7. Credit services consist largely of commercial credit (54 percent of transactions); cash loans from the informal sector (24 percent); cash loans from the formal sector (13 percent); and forward sales (8 percent). Credit services are provided by retailers (24 percent), wholesalers (16 percent), friends and relatives (14 percent), moneylenders (10 percent), commercial banks (6 percent), commercial establishments (6 percent), final users of services (5 percent), and chartered nonbanks (5 percent). Various other sources (e.g., processing and trading enterprises) have a combined share of 14 percent of individual transactions. Volume of Transactions 8. The formal sector provides 61 percent of the credit issued to REs, primarily in the form of cash loans. Such loans exceed the combined volume of commercial credit and forward sales by a margin of almost three to one. In the two years prior to the survey, 73 percent of the credit received by REs consisted of cash loans, 25 percent of commercial credit, and 2 percent of forward sales and sales with a down payment. 9. Fifty-five percent of the credit extended to REs is in the form of cash loans granted by commercial banks. The remainder consists of commercial credit from wholesalers (10 percent), cash loans from moneylenders (8 percent), cash loans from chartered nonbanks (6 percent), commercial credit from retailers (5 percent), cash loans from friends and relatives (3 percent), processing and trading enterprises (13 percent). Regional Differences 10. The relative importance of these different types of lenders varies from region to region, however. The number of individual transactions conducted - B.4 - by the formal sector, for example, is greatest in Guanajuato, which experiences almost twice the formal transactions conducted in Veracruz and Puebla owing to the significant market presence of chartered nonbanks in the state. Also, Guanajuato is the state in which "reciprocity lending" is most frequent. Friends and relatives have a considerably larger share in the number and amount of transactions in Guanajuato than in the remaining two regions. And in terms of amount, Guanajuato is the region in which moneylenders allocate the largest percentage of credit, and forward sales are most important. 11. The formal sector supplies the largest share of credit received by REs (68 percent) in Puebla. This is because informal loans in this state are on average significantly smaller than in the other regions (see below). Moneylenders account for the largest portion of individual transactions, although they are least important in terms of total credit allocated. Friends and relatives are less important in Puebla than in the other two regions, both in terms of number and volume of transactions. 12. The informal sector is most important in Veracruz, where it accounts for 90 percent of transactions and 47 percent of the amount of credit. This is due in large part to the fact that commercial credit is more abundant in Veracruz than in the other two regions. Characteristics of Cash Loans 13. The cash loans received by REs in the two years prior to the survey were examined to determine the terms and conditions imposed by the formal intermediaries, moneylenders, and friends and relatives. Particular attention was given to the amounts, terms to repayment, declared use of loan proceeds, collateral requirements, speed of disbursement, transaction costs, and repayment problems. A hedonic regression was used to explain the effective rates of interest paid by REs as a function of the characteristics of the loans, as well as the characteristics of the borrowers. The resulting information is presented in Tables 1 and 7. 14. Cash loans are defined as credit disbursed and repaid in cash. As already mentioned, cash loans are supplied by both the formal sector (banks, chartered nonbank intermediaries, other registered institutions) and the informal sector (moneylenders, friends and relatives) and represent the most important source of credit for REs, amounting to 73 percent of the total received. Even so, less than a quarter of REs obtained this type of credit. - B.5 - Amount of Cash Loans 15. As might be expected, banks provide the largest loans in rural credit markets. The average bank loan (N$36,000) is almost nine times larger than the average loan allocated by chartered nonbank intermediaries (N$4, 100). Charter nonbank intermediaries are second to banks in terms of average size of loans. Money lender loans, run at about N$3,100 and thus place third in rank, while friends and relatives offer the smallest loans, which average about N$875. 16. The differences in the average size of the loans granted by each type of lender are consistent with the distribution of their transactions when judged by loan amount. Banks, for instance, have a clear preference for comparatively large transactions: 85 percent of their loans were for N$10,000 or more. Loans of this size category accounted for 98 percent of bank disbursements. 17. Chartered % U* O mterd-nm 0 .0. P.md 0 Now/t.m. a d.m.b & nonbanks tend to serve borrowers of intermediate size as proxied by so the loan amount. Most of their loans run between " N$2,000 and ro 0 N$10,000. These NS1,000 - NS2,00 - > kinds of Loans NKOW NSZOW NSIO,OW NSIO.OW represent 54 Figure B2. Distribution by amount of the cash-loans granted percent of by different kinds of lenders (percentage of loans transactions and granted by each type of lender) about half of the amounts disbursed. 18. The amounts moneylenders provide vary greatly, although about 57 percent of individual transactions are for N$2,000 or less and there are comparatively few large loans. Indeed, loans for more than N$10,000 represent only 11 percent of individual transactions but 54 percent of the amount lent by moneylenders to REs. 19. Not surprisingly, friends and relatives tend to provide very small loans. Seventy percent of these loans were for less than N$1,000. - B.6 - 20. The average amount provided to different groups of REs also vary considerably. REs devoted exclusively to agricultural activities received loans averaging N$10,8000, which is twice the average received by REs focusing on nonagricultural activities (N$4,000). REs who combine agricultural and nonagricultural ventures received loans for an average amount of N$7,6000. The average lent to male entrepreneurs was N$8,300, which is almost four times larger than the amount lent to their female counterparts (N$2,200); meanwhile, the average amount granted to nonindigenous entrepreneurs (N$8,300) was six times the average amount received by indigenous entrepreneurs (N$1,300). Terms of Repayment Cash Loans 21. Repayment terms vary from lender to lender. Bank loans have the longest repayment terms, the average being 21 months. Weighted by amount, however, the average runs to 31 months. This means that larger cash loans to REs have longer repayment periods. Bank loans with repayment terms of 18 months or longer account for 16 percent of transactions but for 46 percent of the amount received by REs from banks. 22. The average repayment that chartered nonbank intermediaries require of REs is 10 months, or half the average of bank loans. However, these lenders tend to have more uniform terms than banks, are which are concentrated more or less around one year. 23. Loans from the informal sector (i.e., moneylenders and friends and relatives) have significantly shorter repayment periods and quite flexible repayment schedules. Moneylender loans are usually canceled within three to four months.' Moneylender loans together with chartered nonbank loans show the least variation in of their terms. 24. About one-half of the loans granted by moneylenders have flexible repayment schedules. That is, the maturity and amortization of such loans are contingent on the borrowers' ability and willingness to cancel the debt at a given point in time, since the repayment schedule is not explicitly established prior to disbursement. 25. Loans from friends and relatives offer the shortest terms and the most flexible repayment schedule. These kinds of loans are usually paid back in 2.5 This average corresponds to those moneylender loans in which a specific term was actually agreed upon with the borrower at the time of disbursement. As noted below, many loans from moneylenders and friends and relatives do not even specify the term of repayment. -B.7 - months, and 59 percent of the number of loans and 72 percent of the total amount have flexible repayment schedules. Flexibility does not mean long repayment periods. In fact, when repayment terms from moneylenders and friends are flexible, the loans tend to be paid back within about 4.5 months. 26. Flexible repayment terms are particularly helpful to borrowers because they reduce cash-flow uncertainty. As shown below, REs have been willing to pay significant premia for such flexibility. 27. Only the informal sector provides loans with flexible repayment rates, and hence these loans are rather small. One problem here is that state contingent credit contracts require a great deal of local information.2 Not only do the actual "states of nature" have to be verified (e.g., the cash available to the borrower at a given point in time), but in some cases the actions of the borrower must be monitored closely, especially when such states of nature depend on both exogenous random variables (e.g., high or low production) and on the actions of the borrower (e.g., diligence or negligence). 28. The informal sector, in contrast to the formal sector, is able to provide flexible terms on its loans because the costs of screening and monitoring their borrowers (i.e., state verification) are relatively low. Outsiders have realized that they would do best to rely on simpler, fixed-value debt contracts. 29. Repayment terms also vary with the type of borrower. For farm REs, the average is double the average term of loans by nonfarm REs. The average term of the loans received by farm REs is 11 months and the corresponding average in the case of nonfarm REs is 6 months. REs who combine farm and nonfarm activities enjoy the longest average repayment periods: 18 months. 30. Repayment terms also appear to be affected by gender . On average, REs have more than twice as much time to repay their loans (10 months) than their female counterparts (4 months). Ethnicity does not seem to affect the average repayment term of loans, however, which is about 9 months for both indigenous and nonindigenous REs. 31. As a whole, then, the majority of RE loans are fairly small with a short repayment period, although a few are large and have significantly longer terms. In individual transactions, 35 percent of the loans have flexible 2 Contingent credit contracts are those in which the present value of payments made by the borrower and/or the repayment schedule of the loan depends on the realized outcomes of pre-determined variables (i.e., states of nature). In this type of contract the repayment term, the interest rate, and/or the amount of principal repaid, depend, for example, on the borrower's realized income. - B.8 - repayment terms, 51 percent have to be canceled in less than one year, and only 14 percent have repayment terms of more than one year. When it comes to % the amounts lent, % n=nt 40 10 percent of the 4 total amount of o . loans have 26 flexible 2o repayment .. schedules, 48 1: % Trnsactions percent have to a be paid in less Floxiblo c 3 3-6 6-12 >-12 than ne yer, IMonths Months Months Months than one year, and 42 percent Figure B3. Relationship between Amount of Cash Loans and have a repayment Repayment Term term of more than one year. Furthermore, loans with terms of 18 months or more represent 6 percent of individual transactions but constitute 40 percent of the total amount granted. 32. The predominance of flow over stock credit services and the apparently limited access to long-term debt by only a few larger borrowers is probably due to a strong demand for short-term loans and a weak supply of long-term loans. 33. The demand for short-term credit is high because rural incomes are so variable, as a result of windfalls or shortfalls; unexpected consumption needs may arise, as in the case of medical emergencies; and cash flows are often unsteady because of the seasonal nature of farm activities. Thus short-term financial services are vital in order to help rural folk keep up with consumption levels across income cycles, particularly among the poorest households, whose average consumption levels are low to begin with. 34. Long-term credit is rather limited because both because of the general lack of lending institutions in Mexico and the current uncertainty regarding the effects of new trade policies on the country's rural areas. These factors, together with the country's history of price instability, no doubt deter most lenders from entering into long-term loans with fixed nominal values. Only banks and a few sophisticated borrowers are able to engage in loan contracts 3 Financial services--deposits and credit-services may be either flow or stock financial services. Flow financial services are characterized by loans that are repaid and savings that are withdrawn in one or few income cycles-normally less than a year. Stock financial services, on the other hand, have longer terms--normally several-income cycles. - B.9 - with adjustable interest rates. Because such loans are complex, both parties must be able to observe a reference rate (e.g., CETES), as is the case with several long-term loans reported in the regions studied. Reported Use of Loan Proceeds 35. During the survey, REs were asked what they did with the loan received. Their responses (Tables 4 and 5) indicate that 60 percent of the loans were used to finance business investments, and that 40 percent went into household consumption. As a share of the amounts borrowed, business investments were even higher, at 87 percent of these amounts. 36. These results should be interpreted with caution, however, because they do not necessarily reflect the activities actually financed by the lender at the margin. Given the fungibility of money, it is possible, for instance, that a loan actually invested in the borrower's enterprise may have also allowed household consumption to increase, because a portion of the RE's equity-which in the absence of the loan would have been invested in the RE's enterprise-could instead be used to finance consumption. Nonetheless, the results are interesting inasmuch as they seem to reflect general patterns of loan investment. 37. Rural entrepreneurs are more likely to use loans from the formal sector to finance business investments rather than household consumption. Ninety- two percent of the loans from banks and 70 percent of the loans from chartered nonbanks were used for such purposes. Similarly, 95 percent of the amount borrowed from banks and 74 percent of the amount from chartered nonbanks went into business activities. 38. About half the number of loans received from moneylenders and friends and relatives were used to finance household consumption. As for the amounts borrowed from moneylenders, slightly more was used to finance consumption (54 percent) rather than production (46 percent). In contrast, most of the amount borrowed from friends and relatives (73 percent) was invested in business activities. Collateral and Guarantees on Cash Loans 39. Instead of using tangible assets such as real estate and movable goods as collateral, lenders in the regions studied rely heavily on fiduciary contracts and informal agreements to guarantee repayment of the loans they allocate. Almost 90 percent of individual loans and roughly one-half of the total amount lent to REs are guaranteed by fiduciary contracts (e.g., letras de pago) or informal agreements (e.g., verbal promises). - B.10 - 40. Mortgages on land are used mainly by the banking sector. Banks required such collateral on a third of their loans. These loans account for 62 percent of the aggregate amount disbursed by banks to REs in the regions. Moneylenders also use mortgages as collateral. They received mortgages on 9 percent of their loans, which is equivalent to 5 percent of their disbursements to REs. 41. Movable goods are the least used form of collateral, appearing as a guarantee in only 4 percent of transactions. The combined amount of these transactions accounts for 5 percent of the total volume disbursed. 42. Guarantee requirements are stricter for formal sector loans, particularly bank loans, than for informal sector loans, especially those from friends and relatives. Lender groups can be ranked by their use of formal guarantees as follows: a. Banks have the the 10 strictest collateral requirements. 70 a 0 Chafr non-banics Forty percent of on individual loans so wol.,. disbursed by * aMomooder banks have Fif a 20. .Rolatives required tangible o assets such as o 0 Movablo Fiduciary- Rducdary Vorbal mortgages on Molvago GOc Cocoignor Florrw Prm.lo land and movable Figure B4. Collateral Required by Different Types of Lenders goods as collateral. These collateralized loans represent 64 percent of the bank disbursements to REs. Even so, the remaining 60 percent of bank loans were granted on fiduciary contracts. In fact, banks allocated a third of their loans-amounting to 25 percent of funds-on the fiduciary responsibility of the borrower alone, without cosigners. Loans guaranteed by cosigners accounted for 26 percent of bank transactions and 11 percent of disbursements to REs. b. Chartered nonbank intermediaries also count heavily on fiduciary guarantees for their loans, which have been used in 86 percent of their transactions, which is equivalent to 69 percent of their disbursements. Chartered nonbank intermediaries require cosigners for their loans more often than banks. Loans with cosigners accounted for 43 percent of their transactions and 40 percent of their disbursements to REs. - B.11 - c. Contrary to popular belief, moneylenders rely primarily on promises from their borrowers rather than on tangible collateral to enforce credit contracts. Fifty-five percent of the moneylender loans granted have been based on the fiduciary responsibility of the borrower (e.g., pagare). These loans account for 56 percent of disbursements to REs. The second most common guarantee accepted by moneylenders is a verbal promise. Informal agreements of this type account for 26 percent of moneylender loans and make up 10 percent of their disbursements. As a whole, tangible assets and cosigners were used-to guarantee payment on only 20 percent of transactions and account for 34 percent of the amounts disbursed to REs by such lenders. d. As might be expected, friends and relatives have the least stringent collateral requirements. Ninety percent of loans disbursed by this group are based on verbal promises, which also account for 86 percent of the amount disbursed. Only in 8 percent of the loans given by friends and relatives were borrowers required to sign a promissory note. These somewhat "stricter" loans account for the remaining 14 percent of the amount lent by friends and relatives. 43. Tangible assets (i.e., real estate and movable goods) are used as collateral for loans in 11 percent of transactions. In these cases, the ratio of the market value of the pledged asset to the amount of the loan (collateral/loan) is very high. When the collateralized loans are issued by banks, the market price of the pledged assets is on average six times larger than the amount of the loan. When the credit is obtained from moneylenders, the average value of the ratio increases to 26 times. 44. This seemingly excessive collateralization of loans is due to the costly and lengthy processes involved in repossessing assets pledged as collateral. Many rural lenders have blamed this situation on the inadequacies of the legal framework and the inefficiency of the courts (see Annex D). 45. In the face of such collateralization, large numbers of otherwise creditworthy REs are likely to stay out of the credit markets. Peculiar collateral requirements together with the shortcomings of the legal framework could make the borrowers' cost of defaulting disproportionate to the amount borrowed. This is because the rural areas under consideration have no functioning mechanisms with which to auction property given as collateral and to ensure that any amount remaining after lenders have been fully compensated will be returned to the borrower. Hence, those potential borrowers who would be required by lenders to pledge collateral may face an artificially skewed risk- return distribution for their investments, which may persuade them to drop out of credit markets. As indicated earlier, 20 percent of all REs have not requested a loan because they consider borrowing too risky. - B.12 - 46. This use of collateral is at odds with that of other financial markets, particularly those of developed countries. In the United States, for example, approximately 40 percent of credit is secured by liens on movable goods alone, yet such instruments are rarely used in the rural areas studied." There is no reason to suppose that the country's rural credit markets as a whole are any different. 47. The reliance on fiduciary contracts supports the hypothesis that rural credit markets in Mexico may be segmented. This is because such contracts, combined for example with the generalized lack of financial records on the part of REs make local knowledge vital to the effective provision of rural credit. Only those who already know potential local borrowers are able to provide credit services profitably. By way of example, the data on transaction costs of borrowing presented below indicate that 72 percent of all loans received by REs were granted by lenders located within 1 kilometer of the borrower's residence. In other words, REs apparently have to rely mainly on local resources, which are often limited. Therefore, a low initial endowment of resources in a locality would have a long-lasting effect on wealth accumulation and, hence, on poverty. 48. An efficient use of collateral would save lenders the costs of screening borrowers, not to mention the costs borrowers have to incur in order to establish their creditworthiness because collateral acts as a hostage pledged by the borrower to support the repayment promise implicit in credit transactions. The willingness to pledge a valuable hostage signals the borrower's commitment and incentive to deliver on the promise to repay the loan. Speed of Disbursement and Transaction Costs of Cash Loans 49. When choosing among potential lenders, borrowers pay close attention to the speed and timeliness of the disbursement, especially when cash needs are urgent or unexpected.' Borrowers tend to be more concerned with other variables (e.g., interest rates) when deciding about the sources of funding for, say, capital investments. For the most part, however, the speed and timeliness 4 See: Heywood, Fleisig et. al. "How Legal Restrictions on Collateral Limit Access to Credit in Bolivia", Private Sector Department, The World Bank, June 1994, pag. 6. As indicated in the next section, providers of commercial credit secure credit with liens on movable goods less often than providers of cash loans. The survey questionnaire was designed to obtain rough estimates of transaction costs. Nonetheless, estimated transaction costs for different types of lenders are consistent with studies elsewhere and with expected results-as reported. Note also that the calculation of transaction costs do not take into account the maturity of loans (i.e., not annualized). - B.13 - of disbursement together with the transaction costs of borrowing greatly affect RE decisions regarding whether to participate in credit markets. Thus, 13 percent of REs who did not request loans said they considered the process too lengthy and costly. 50. The speed W... of loan disbursement, measured as the 1 time elapsed a . between loan application and 4. actualI disbursement, seems to depend B..k. Car.ere other Mon.ylender Friends & on the size of the non-ftnks Forft loans. That is to Figure B5. Average Speed of Disbursement of Loans by say, those lenders Different Kinds of Lenders who provide the smaller loans are also the ones who disburse loans faster. Friends and relatives disburse their loans faster than any other group of lenders: they take three days on average. Next, moneylenders take about 10 days to disburse. Formal institutions, chartered nonbanks and banks, take an average 4 and 14 weeks, respectively. 51. Those lenders who are slower to disburse also impose higher transaction costs (TCs) on the borrowers. Thus the TCs of borrowing are usually highest with banks, which usually charge an average of N$700 per loan. 52. Transaction costs are much lower for loans from chartered nonbank intermediaries. These costs average about N$40 and represent 1 percent of the amount borrowed. 53. The informal sector (i.e., moneylenders, friends and relatives) offers borrowers the lowest transaction costs, about N$12 on average. The costs average about 0.7 percent of the amount lent from moneylenders, and 1 percent of loans from friends and relatives. The latter are slightly higher than the former because friends and relatives provide smaller loans than moneylenders. 54. The costs of borrowing vary greatly in part because of the geographic distance between borrowers and lenders. The average distance between REs and the moneylenders and friends and relatives from whom they borrow is 4 - B.14 - kilometers, whereas in the case of bank loans the corresponding distance is 31 kilometers. In the former case, creditworthiness is established through local information about the applicant's character and reputation and can be acquired at a low marginal cost. 55. Perhaps more important, transaction costs are also governed by the technologies lenders use to screen loan applicants. The cost of producing sufficient information to support a credit contract can be substantial and is normally the main component of transaction costs in such loans.6 Borrowers have to signal or demonstrate to lenders their ability and willingness to honor their debt. Lenders, on the other hand, have to interpret and verify the legitimacy of the signals provided by prospective borrowers. Loans completely collaterized (e.g., pawning) imply low information costs to lenders because the value of the pledged good provides all the information required to assess the probability of repayment. In other cases, the lender must gather all the necessary information, while the borrower's costs are limited to requesting the loan. As a rule, however, both lenders and borrowers will incur information costs. The modest TCs for informal sector loans suggest that in these cases borrowers do not have to incur significant expenses such as lengthy trips and hence sizable amounts of forgone income in order to establish their creditworthiness. 56. In contrast, the information banks rely on to screen loan applicants costs more to acquire. Such information consists of audited financial statements and property appraisal reports. The borrower normally covers the cost of producing this information. 57. The matching of borrowers and lenders usually depends on the costs of the transaction. For example, large businesses normally maintain accounting records of their economic activities not only because they aid management but also because, in most cases, ownership and control of such businesses are separated. Hence, accounting records and external audits are necessary instruments of internal control that would be used regardless of credit transactions. Once audited financial statements are available, the marginal cost of using them to prove creditworthiness is zero. Therefore, firms that prepare audited financial statements are more likely to request credit from financial intermediaries whose lending technology readily provides such information. 6 The elements of lending technology are analyzed in detail in Annex E. Particular importance is given to the information component of such technology (borrower screening,monitoring of projects, and enforcement of contracts). -B.15 - 58. The matching of borrowers and lenders is also governed by the amount of the desired loan from a given source. Small REs-who presumably demand comparatively small loans-would be inclined to request credit from those sources that would also imply small TCs (e.g., moneylenders). At the same time, REs investing in large ventures would not be as influenced by TCs because such costs tend to be invariant to the size of the loan. Repayment Problems of Cash Loans 59. The survey collected data on the repayment problems of both lenders and borrowers. To begin with, an attempt was made to determine the extent to which borrowers were able to meet all loan amortization and interest payments on time and whether an agreement re-scheduling the loan was negotiated with the lender. 60. Two imperfect proxy indicators of delinquency were devised. The first indicator is the amount and number of loans that experienced repayment problems as a percentage,of the total amount of loans and the total number of loans disbursed, respectively. The second indicator is the amount and number of loans that experienced repayment problems and that were not rescheduled. 61. It should be pointed out, however, that the information was collected from the borrowers themselves, who for various reasons may underreport delinquency. Also, the measures of delinquency overstate the problem because the amounts used to proxy arrears correspond to the amount of the loan and not to the outstanding balance at the time of the repayment problem. Nonetheless, the resulting information can be used to study systematic differences in repayment across groups of borrowers and lenders. 62. Thirteen percent of all FORMAL SECTOR loans fell behind B..k. their scheduled Chafered non-anks payments before Other Fomal7 maturity. INFORMAL SECTOR Among these Monr*ender problematic FIend and loans, 83 TOTAL percent were a 5 10 Is 20 2 30 S defaulted by the " % borrower. Figure B6. Repayment problems on cash loans Loans already defaulted at the time of the survey represented 11 percent of all loans received by REs, including those that had not matured at the time. Eighty percent of - B.16 - the loans that fell in arrears were initially rescheduled by the lender. However, re-scheduling does not seem to have improved significantly the likelihood of loans being repaid, since 80 percent of rescheduled loans were defaulted anyway. 63. The survey results suggest that repayment problems are more severe for formal sector lenders than for their informal counterparts, both in terms of the number of individual loans and the amounts lent. As table 6 shows, 26 percent of the individual loans received by REs from formal lenders fell in arrears; 30 percent of chartered nonbank loans and 27 percent of bank loans had payments overdue at some point. The corresponding percentage for informal sector loans was 7 percent: 11.5 percent of moneylender loans and 4 percent of loans from friends and relatives. The combined original amount of formal sector loans that fell in.arrears at some point equaled 47 percent of the total amount of loans granted by this sector. The corresponding percentage for the loans from the informal sector was 21 percent. 64. REs who combined farm and nonfarm ventures fell into arrears in 20 percent of their loans, REs who specialized in farm ventures fell into arrears in 15 percent of their loans, and REs who had nonfarm ventures in 12 percent of their loans. In terms of the amount in arrears, farm REs seem to pose a higher credit risk for lenders as the loans that fell in arrears amounted to 60 percent of the total amount disbursed to such borrowers. The corresponding measure for nonfarm entrepreneurs was only 14 percent. The Effective Interest Rate on Cash Loans 65. The principal concern in studying the nominal effective rates of interest on the cash loans received by REs was to determine the interest rates charged by different lenders and the factors affecting the interest rates paid by individual REs. Cash loans from friends and relatives were not examined because the focus of attention was the rates of interest charged by lenders who have profit motives. 66. The effective rates of interest used for the analysis correspond to the internal rate of return on observed loans. The calculations took into account all relevant attributes of loan contracts such as the frequency of payments and whether interest charges are based on a flat or a declining balance. Effective rates are required to compare the rates of the various kinds loans that take place in the study areas. 67. Nominal as opposed to real rates were used because the loans observed were disbursed at different points over the period from June 1992 to June 1994 and their maturities vary. Hence, the calculation of ex-post real rates would be -B.17 - cumbersome and would add little to the analysis as the rates of inflation were low and stable in the period. The changes in the consumer price index for 1992, 1993, and 1994 were 11.9 percent, 8 percent, and 7 percent, respectively. Furthermore, there is no reason to believe that different groups of lenders or borrowers had systematically different inflationary expectations at the time loans were granted. 68. The effective rates of interest charged by lenders vary greatly. Chartered nonbank financial intermediaries charge an average 40 percent, which is the lowest annual effective rate on loans.7 The weighted average of the rates charged on such loans is 35 percent, which suggests that these intermediaries tend to charge lower rates on their larger loans. Banks behave in a similar fashion, both in terms of the actual rates charged and in the lower rates attached to larger loans. The average of the annual effective rates charged by banks is 45 percent, while the weighted average is 33 percent. 69. As expected, moneylenders charge the highest rates of interest, which run many times higher than those of formal intermediaries. The average annual effective rates charged by moneylenders is 330 percent. The weighted average is 200 percent. Like the formal intermediaries, moneylenders also tend to charge lower rates on larger loans. 70. As already mentioned, rates vary considerably across the formal and informal and formal sectors. The rates charged by lenders within the same group also vary significantly, as is evident from the high coefficient of variation (CV) for the effective rates of loans within lender groups.8 The CV is highest in the case of moneylenders, where it approaches 150 percent for observed rates. Chartered nonbanks offer the least variable rates; the CV in this case amounts to 58 percent. The CV for the rates of bank loans is 102 percent. 71. In general, the interest rates paid by borrowers depend on the conditions of the loans received, the attributes of the borrowers themselves, the kind of economic activity they are engaged in, and the specific markets in which loans are granted. 7 This is because of the common practice of Cajas Populares, Uniones de Credito, and--to a certain extent-banks to demand deposits or compensatory balances as loan requisites to increase the effective rate of loans. Detailed information on the amount and rates paid on such balances is not available-however. The coefficient of variation (CV) is a measure of a variable's dispersion. The CV equals the standard deviation of the variable divided by its mean. -B.18 - 72. Table B7 presents the results of a regression that relates observed rates of interest and the characteristics of the loans granted, the attributes of the borrowers, and other variables of the locality in which borrowers reside. The results of the regression should be understood as a summary of the aggregate interest rate conditions at which the rural financial sector provides loans-with certain characteristics (e.g., term, amount)-to distinct types of REs (e.g., farmers, females) in different localities (e.g., population size). Caution should be exercised when deriving causality relationships among variables, however. 73. Several highly significant variables (95 and 99 percent confidence intervals) help to explain the interest rates. In the aggregate, borrowers who received larger loans and who had longer repayment periods paid lower rates. Also, some borrowers paid a premium on repayment flexibility, as indicated by a positive sign for the coefficient for the interaction dummy "flexibility in repayment. " 74. The effects of the significant explanatory variables included in the regression are summarized in table Bl.' The table shows that-controlling for other variables-an RE who borrowed an amount 1 percent greater than the amount borrowed by another RE is expected to pay a monthly interest rate that is 9 percent lower than the rate paid by the smaller borrower. In short, a 1 percent increase in the amount of a loan is associated with a 9 percent decrease in its monthly effective rate of interest (e.g., from 5 percent to 4.5 percent a month). The references made below regarding the effects of the explanatory variables on the monthly effective rate are expressed in terms of percentage changes in the interest rate and not on percentage points of change. That is to say that a 10 percent decrease in a monthly interest rate of, say, 5 percent per month corresponds to a reduction of 0.5 percent to a monthly interest of 4.5 percent. - B.19 - Table B1. Expected Differences in the Monthly Interest Rate Associated to Changes in Explanatory Variables Expected Change in Variable Monthly Rate of Interest Loan amount -0.0936 Log of repayment term (months) -0.1880 Log of repayment term (months), if flexible term 0.0300 Dummy: Actual use of the loan (1 =consumption, O=production) 0.3320 Log of RE's enterprise income (N$ and includes on farm consumption) -0.0713 b Dummy: Gender of RE (1 =male, 0=female) -0.3361 Dummy: Ethnicity of RE (1 = indigenous, 0= non-indigenous) -0.5038 Distance between borrower and lender (kilometers) -0.0159 Population in RE's locality of residence (thousands of people, -0.0401 1990-census) a. Percentage change in the effective interest rate associated to one percent change in the explanatory variable (i.e., elasticity). b. Percentage change in the expected effective interest rate when the dummy equals 1 (e.g. male) with respect to the situation when the dummy equals 0 (e.g., female) -ceteris paribus. The expected change in the rate of interest equals (expO - 1), where P is the estimated coefficient presented in Table B7. c. Percentage change in the monthly interest rate associated to a unit change in the independent variable (e.g., increase in a thousand inhabitants). 75. Similarly, interest rates are highly sensitive to the term for repayment of loans as a 1 percent increase in the maturity of the loan is associated with a reduction of 18 percent in the effective monthly interest rate (e.g., from 5 percent to 4 percent per month). 76. The actual use of the loan for household consumption or investment is also associated with the rate paid. Borrowers who reported having used the loan for household consumption paid-controlling for other variables-a monthly rate 33 percent higher than those who declared having invested the loan proceeds in their enterprises (e.g., from 5 percent to 6.5 percent per month). 77. The rates of interest borrowers pay on loans are significantly related to the amount of enterprise income and the RE's gender and ethnicity (95 and 99 percent confidence intervals). In particular, borrowers with low levels of enterprise income and female pay higher rates than borrowers with different personal attributes. 78. In particular, a difference of 1 percent in enterprise operational income is associated with an expected reduction of 7 percent in the monthly rate of interest (e.g., from 5 percent to 4.65 percent per month). 79. Rates differ even more significantly in relation to gender. The monthly rates of interest paid by male borrowers are almost a third lower than the rates paid by their female counterparts. - B.20 - 80. The interest cost of borrowing appears to be highly affected by ethnicity. Indigenous borrowers are charged half the monthly rates paid by their nonindigenous counterparts. In addition, the interest rates on loans are significantly related to the geographical distance between borrowers and lenders and the population in the localities where borrowers reside (99 percent confidence for both coefficients). Borrowers whose source of loans is located further away from their residence pay lower rates of interest, although borrowers who live in small towns tend to pay higher rates of interest. 81. Two obvious questions can be raised here. First, are rates of interest governed more by the characteristics of the loans (e.g., term to repayment) or by the type of borrower (e.g., female RE)? Second, are the observed differences in interest rates the result of the different degrees of access among borrowers (e.g., nonindigenous and male REs) to lenders who have very different lending policies (e.g., banks). 82. The suggestion that borrowers' specific rates of interest depend on access to given types of lenders is based on the observation that the characteristics of loans are correlated to the type of lender (e.g., banks lend at longer terms), and at the same time borrower-specific attributes (e.g., income, locality of residence) determine the type of lender to which REs have access (see next section). The hypothesis that the interest rate paid by individual borrowers is largely determined by whether they have access to the informal or formal sectors is studied by adding a sectoral dummy variable (1 = formal sector, 0 = otherwise) as an explanatory variable to the regression used in the analysis. 83. The econometric results (see table B2) confirm the hypothesis that the rates of interest paid by borrowers are largely related to the sector to which they have access. Individuals with access to formal sector loans pay 77 percent lower rates than they would pay on equivalent loans from the informal sector. 84. After controlling for the formality of the lender, all lenders levy higher rates for longer-term loans: a 1 percent increase in the term to maturity is associated with a 25 percent increase in the rate (e.g., from 5 percent to 7.5 percent per month). Larger ventures pay lower rates, since a 1 percent increase in enterprise income is associated with a reduction of 4 percent in monthly effective rates. The relationship of the borrower's gender to the interest rate paid was also highly significant (99 percent confidence) and actually increased after controlling for the lender's formality. Male borrowers pay monthly rates of interest that are 35 percent lower than the rates female borrowers would have to pay for loans with the same characteristics. - B.21 - Table B2. Expected Differences in the Monthly Interest Rate Associated to Changes in Explanatory Variables Including a Sector Dummy Expected Change in Variable Monthly Rate of Interest Log of repayment term (months) -0.2530 Dummy: Formal sector lender (1= lender b formal sector, 0=otherwise) -0.7404 Log of RE's enterprise income (N$ and includes a on farm consumption) -0.0518 b Dummy: Gender of RE (1 =male, 0= female) -0.3453 Distance between borrower and lender (kilometers) -0.0097 a, b, c: see Table BI. - B.22 - Table B3. Characteristics of Loan Transactions: Amount, Repayment Terms, and Interest Rates Characteristic Mean Std Dev' C.V.* Min Median Max Loan amount (new pesos) Formal sector 16,735.61 27,847.03 166.39 50 5,000 133,000 Banks 36,410.56 34,698.77 95.30 300 20,800 133,000 Chartered nonbanks 4,166.03 4,336.15 104.08 250 2,500 20,000 Other registered institutions 868.87 644.02 74.12 50 800 2,200 Informal sector 1,797.12 3,351.86 186.51 30 500 20,000 Moneylender 3,169.70 4,729.43 149.21 50 1,000 20,000 Friends and relatives 877.62 1,302.19 148.38 30 400 7,000 Traders 904.99 1,212.60 133.99 30 350 3,000 Total 7,164.80 18,362.91 256.29 30 1,000 133,000 Repayment terms (months)' Formal sector 14.11 20.68 146.58 0.23 6.00 120.00 Banks 21.44 30.06 140.25 1.00 5.00 120.00 Chartered nonbanks 10.40 5.94 57.13 2.00 12.00 36.00 Other registered institutions 6.39 3.80 59.48 0.23 8.00 12.00 Informal sector 3.25 2.70 83.07 0.23 2.00 18.00 Moneylender 4.13 2.42 58.73 0.70 3.00 12.00 Friends and relatives 2.58 2.73 106.00 0.23 1.25 18.00 Traders 2.40 2.29 95.37 0.23 2.00 6.00 Total 9.15 16.28 177.92 0.23 4.00 120.00 Annual effective interest rate (%) Formal sector 32.67 36.15 110.64 0.00 26.82 213.84 Banks 45.44 46.51 102.36 4.06 28.49 213.84 Chartered nonbanks 39.87 22.99 57.65 0.00 42.58 103.75 Other registered institutions 0.93 4.22 455.23 0.00 0.00 20.14 Informal sector 131.34 349.98 266.47 0.00 0.00 3,272.53 Moneylender 328.99 491.74 149.47 26.82 13.84 3,272.53 Friends and relatives 0.00 0.00 0.00 0.00 0.00 0.00 Traders 0.00 0.00 0.00 0.00 0.00 0.00 Total 98.01 289.36 295.23 0.00 7.73 3,272.53 Monthly effective interest rate (%) Formal sector 2.16 1.96 90.59 0.00 2.00 10.00 Banks 2.88 2.18 75.70 0.33 2.11 10.00 Chartered nonbanks 2.73 1.36 49.83 0.00 3.00 6.11 Other registered institutions 0.07 0.32 455.32 0.00 0.00 1.54 Informal sector 4.25 6.38 150.05 0.00 0.00 34.07 Moneylender 10.66 5.82 54.60 2.00 10.00 34.07 Friends and relatives 0.00 0.00 0.00 0.00 0.00 0.00 Traders 0.00 0.00 0.00 0.00 0.00 0.00 Total 3.55 5.41 152.52 0.00 0.62 34.07 a. This value was estimated assuming simple random sampling. b. Transactions with flexible contractual repayment term were not included in this particular calculation. Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. - B.23 - Table B4. Reported Use of Loans: Percentages of Individual Loan Transactions Sector Total Investments Consumption Formal 100.00 81.37 18.63 (35.84) (29.16) (6.68) Banks 100.00 92.57 7.43 (14.76) (13.66) (1.10) Chartered nonbanks 100.00 70.40 29.60 (13.36) (9.41) (3.96) Other 100.00 78.94 21.06 (7.72) (6.09) (1.63) Informal 100.00 48.58 51.42 (64.16) (31.17) (32.99) Moneylender 100.00 47.56 52.44 (25.61) (12.18) (13.43) Friends and relatives 100.00 49.12 50.88 (36.74) (18.04) (18.69) Traders 100.00 51.97 48.03 (1.81) (0.94) (0.87) Total 100.00 60.33 39.67 Note: Figures in parentheses indicate the percentage of total transactions represented by column headings (e.g., loans granted by banks and invested in the borrower's venture represent 13.66 percent of all transactions). Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. -B.24 - Table B5. Reported Use of Loans: Percentages of Amount of Loans Financial sector Total Investment Consumption Formal sector 100.00 93.08 6.92 (83.93) (78.13) (5.80) Banks 100.00 95.16 4.84 (75.20) (71.56) (3.64) Chartered nonbanks 100.00 74.05 25.95 (7.79) (5.77) (2.02) Other 100.00 85.16 14.84 (0.94) (0.80) (0.14) Informal sector 100.00 54.16 45.84 (16.07) (8.70) (7.37) Moneylender 100.00 45.98 54.02 (11.36) (5.22) (6.14) Friends and relatives 100.00 73.17 26.83 (4.48) (3.28) (1.20) Traders 100.00 88.02 11.98 (0.23) (0.20) (0.03) Total 100.00 86.83 13.17 Note: Figures in parentheses indicate the percentage of the total represented by column headings (e.g., loans granted by banks and invested in the borrower's venture represent 71.56 percent of the total amount borrowed). Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. - B.25 - Table B6. Proxy Measures of Repayment Problems by Source of Loans (percent) Transactions Amount Sector Total No Total No Transactions Arrears Arrears Amount Arrears Arrears Formal sector 100.00 25.90 74.10 100.00 47.29 52.71 (35.84) (9.28) (26.56) (83.93) (39.69) (44.24) Banks 100.00 26.74 73.26 100.00 50.66 49.34 (14.76) (3.95) (10.81) (75.20) (38.10) (37.10) Chartered nonbanks 100.00 30.12 69.88 100.00 20.18 79.82 (13.36) (4.02) (9.34) (7.79) (1.57) (6.22) Other formal 100.00 17.02 82.98 100.00 2.12 97.88 (7.72) (1.31) (6.40) (0.94) (0.02) (0.92) Informal sector 100 7.03 92.97 100.00 21.19 78.81 (64.16) (4.51) (59.65) (16.07) (3.41) (12.66) Moneylender 100.00 11.52 88.48 100.00 29.10 70.91 (25.61) (2.95) (22.66) (11.36) (3.31) (8.06) Friends and relatives 100.00 4.26 95.74 100.00 2.23 97.77 (36.74) (1.56) (35.17) (4.48) (0.10) (4.38) Traders 100.00 - 100.00 100.00 - 100.00 (1.81) (0.00) (1.81) (0.23) - (0.23) Total 100.00 13.80 86.20 100.00 43.10 56.90 Note: Figures in parentheses indicate the percentage of the total represented by column headings (e.g., loans granted by banks represent 35.84 percent of the total number of loans received by rural entrepreneurs). Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. - B.26 - Table B7. Results of a "Hedonic" Regression of Effective Interest Rates in Three Selected Rural Areas of Mexico Regression Std. Variable Coefficient Error Dependent variable: Natural logarithm of monthly effective interest rate Intercept 4.5039 0.5507 Log of loan amount (N$) -0.0936 0.0478 c Log of repayment term (months) -0.1880 0.0919 c Interaction term between flexibility in repayment and log of repayment term 0.2180 0.1137 * Dummy: Actual use of the loan (1 =consumption, 0= production) 0.2867 0.1636 * Log of RE's enterprise income (N$ and includes on farm consumption) -0.0713 0.0305 * Dummy: Borrower's economic activity (1 =only agricultural venture, 0=otherwise) 0.0070 0.1878 Dummy: Gender of RE (1= male, 0= female) -0.4097 0.2043 a Dummy: Ethnicity of RE (1 =indigenous, 0=non-indigenous) -0.7007 0.2509 * Distance between borrower and lender (kilometers) -0.0159 0.0041 ** Population in RE's locality of residence (thousands of people, 1990-census) -0.0401 0.0140 * On-going market wage in borrower's locality of residence (N$) -0.0346 0.0232 Number of transactions with the lender 0.0041 0.0077 Dummy: Type of collateral (1 =real collateral, 0=otherwise) 0.1421 0.1900 Percentage of borrowers' transaction costs of loan amount 0.0185 0.0160 R-square 0.6732 Adj R-sq 0.6006 */**/*** significant at the 90 percent, 95 percent, and 99 percent confidence intervals--respectively - B.27 - Table B8. Results of a "Hedonic" Regression of Effective Interest Rates in Three Selected Rural Areas of Mexico Including a Sector Dummy Variable Regression Std. Variable Coefficient Error Dependent variable: Natural logarithm of monthly effective interest rate Intercept 3.4590 0.5228 *** Log of loan amount (N$) -0.0057 0.0542 Log of repayment term (months) -0.2530 0.1157 ** Interaction term between sector and log of repayment term 0.2230 0.1455 Dummy: Actual use of the loan (1 =consumption, 0= production) 0.1164 0.1473 Dummy: Formal sector lender (1 =lender formal sector, 0=otherwise) -1.3432 0.3135 * Log of RE's enterprise income (N$ and includes on farm consumption) -0.0518 0.0263 Dummy: Borrower's economic activity (1 =only agricultural venture, 0=otherwise) -0.1158 0.1636 Dummy: Gender of RE (1 =male, 0= female) -0.4236 - 0.1761 ** Dummy: Ethnicity of RE (1= indigenous, 0=non-indigenous) -0.2530 0.2304 Distance between borrower and lender (kilometers) -0.0097 0.0039 *** Population in RE's locality of residence (thousands of people, 1990-census) 0.0007 0.0149 On-going market wage in borrower's locality of residence (N$) 0.0008 0.0205 Number of transactions with the lender -0.0030 0.0068 Dummy: Type of collateral (1= real collateral, 0=otherwise) 0.0087 0.1668 Percentage of borrowers' transaction costs of loan amount 0.0091 0.0141 R-square 0.7592 Adj R-sq 0.7009 * * significant at the 90 percent, 95 percent, and 99 percent confidence intervals -- respectively - B.28 - Characteristics of Commercial Credit 85. Commercial credit is defined as credit received in kind but that has to be repaid in cash. Commercial credit is the most widely used credit service in the areas studied. About a third of all REs received commercial credit in the two years prior to the survey. These transactions represented more than half of all credit transactions in which REs participated, including cash loans and forward sales, and accounted for a quarter of the volume of credit received by such entrepreneurs. 86.. In terms of coverage of REs, commercial credit is dominated by retail merchants, who provide 40 percent of individual transactions. In terms of the volume of credit, wholesalers are on top with about 40 percent of the total amount purchased on credit by REs. Goods Purchased on Credit 87. Seventy percent of commercial credit transactions were acquiring goods to be used in the debtor's business ventures. The purpose of the remaining 30 percent of transactions was to acquired goods devoted to household consumption. Merchants tend Trnobn to specialize in granting credit for either * household consumption or o- for enterprise purposes. For 0 example, goods Wholesale Retail tinerant Other Total tradera traders traders tradars purchased. on .. .o.....p.... EHousohold Consumption Business Invostmonta credit from itinerant traders Figure B7. Reported use of goods purchased on credit were evenly distributed between household consumption-52 percent of transactions-and enterprise investments. On the other hand, the goods acquired in 70 percent of the purchases from wholesalers were used in the debtor's enterprise. Of course, this specialization is highly related to the products sold by the merchant as farm supplies, for example, would likely be used for business purposes. - B.29 - Average Amount and 4o Financial Cost of Commercial Credit Transactions 15 88. The 5 average Interest Rate > 0 % All Transactions commercial comeci Cash-.oan O commercial CredIt credit transactions are Figure B8. Average monthly effective rate of interest charged on small, short- commercial credit and cash loans term agreements that are expensive for the borrower, at least when compared to cash loans. The average amount of these transactions was close to N$2,000 (less than a third of the average of cash loans), while the average term to repayment was about three months. On the other hand, commercial credit is a rather expensive service, especially in comparison with cash loans. The average effective cost across all commercial credit transactions was 16.25 percent a month. 89. The financial cost of commercial credit seems to be extremely high, particularly when one takes into account that only half of such transactions implied a financial cost for the debtor, at least that she was aware of. ' The average cost of commercial credit for those for whom financial costs were not reported to be zero was 36.5 percent a month. Table B12 also shows that commercial credit transactions that commanded financial costs were much larger (three times larger) and covered much longer periods (five times longer) than those transactions with no financial cost for the debtor. 90. The very high financial cost of commercial credit may be explained-at least in terms of the borrower's willingness to pay such high rates-by the short terms and relatively small amounts of the transactions. These two characteristics could make the absolute cost of using credit (i.e., equivalent in pesos) reasonable in comparison with the cost of finding a substitute source of funding (e.g., cash loans) or another supplier willing to sell the same good also on credit. In the long run, however, the financial costs of repeated transactions may accumulate to a sizable amount. 10 Differences between the "cash price" and the "credit price" are common instruments to collect financial charges on commercial credit. Often times debtors do not realize or are not able to quantify such differences in prices. -B.30 - 91. The question that remains, however, is why the seemingly lavish financial returns implicit in commercial credit are not bid down by other merchant-lenders, including potential entrants. One possibility is that in the areas studied there are no abnormal returns to selling goods on credit. If this explanation is to be plausible, delinquency rates need to be significant, so as to make large risk premia necessary; and merchant's transaction costs in screening potential debtors need to be high. The survey data on cash loan default rates and the anecdotal evidence from interviews with moneylenders and trader-lenders suggest, however, that credit losses are not the likely candidate to explain such high financial costs. 92. On the other hand, transaction costs are not a likely candidate either, because it is reasonable to expect merchant-lenders to incur costs similar to those of informal moneylenders, since both use similar technologies to assess the creditworthiness of potential debtors. Hence, only a portion of the difference between the average rates of moneylender loans and the average financial cost of commercial credit-a difference of 6 percent-may be attributed to transaction costs, given that purchases on credit are smaller than moneylender loans." However, a good portion of the high financial cost of commercial credit remains unexplained. 93. A plausible explanation for the observed high rates is that the market for commercial credit in the regions studied is not contestable in the sense that incumbent lenders are not subject to competition from potential entrants because only those who already know borrowers can provide credit services effectively. Competition among incumbent lenders over a given pool of borrowers, on the other hand, is not necessarily strong for the very same reason: for any given potential borrower, some lenders have an advantage regarding the costs for the borrower to prove her creditworthiness and the costs for other lenders to screen her.'2 Under such circumstances, informal lenders may behave like "location" monopolists over a pool of borrowers being able to profitably arbitrage on their cost advantages. 94. A regression was fitted to the data in order to examine the determinants of the effective rate of interest implicit in commercial credit This would be the case under the reasonable assumptions that screening costs are invariant to the amount of credit and that such costs are--as mentioned before--similar for merchants and moneylenders (i.e., they use similar lending technologies). 12 This is equivalent to saying that there are significant sunk costs and strong "learning by doing" effects in an informal lender's credit allocation technology. The case studies included in this report strongly suggests that such is the case. In fact, a portion of the cost differential between two potential lenders is due to plain serendipity in that it results from factors such as living close to the borrower and-in the words of an interviewed lender-"having friends in common". - B.31 - transactions." The regression results (see Table B16) indicate that after controlling for several other factors seven variables were found to be statistically significant in explaining the financial cost of commercial credit: namely, the characteristics of the contract, the type of goods purchased on credit, the attributes of the borrower's enterprise, and the size of the locality where she resides. In particular, commercial credit transactions for small amounts and short terms imply higher financial costs. This result is consistent with the corresponding findings regarding cash loans where larger amounts of loan and longer terms are also associated to lower rates. The decline in financial cost as the term to repayment and amount increase-which is also the case for cash loans-may be due, on one hand, to longer-term loans and large amounts that allow fixed transaction costs of lending to spread over a longer period of time and over a larger amount (this may reduce the necessary rate to be charged on credit). On the other hand, it may be that individuals borrow for longer terms and large amounts and thus search longer for loans as the absolute value of the savings resulting from finding lower cost credit may compensate the transaction costs of shopping around; this implies more substitutes (i.e., more elastic demand), which implies a lower interest rate (mark-up over marginal cost) charged by price-discriminating monopolists. 95. At the same time, those purchases on credit that were completed at low transaction costs for the borrower imply higher effective interest rates. This strongly suggests that commercial credit providers are able to arbitrage profitably on reduced transaction costs for the borrower and thus that the market for such services may not be competitive as lenders behave like location monopolists. 96. The kind of asset purchased also affects the financial cost of commercial credit. Productive inputs and goods acquired to trade are sold at a lower financial cost of credit than capital goods (e.g., machinery, equipment) and household consumption items. Finally, borrowers who do not have agricultural ventures, whose enterprises produce low levels of operating income, and who live in localities with low levels of population are charged higher effective rates than those REs with higher incomes and who live in larger towns. 97. The following table presents the expected differences in the monthly effective rate of interest associated to changes in the mentioned statistically 13 The econometric technique used is a "Two-Step Heckman Model" in which the dependent variable was the natural logarithm of the monthly effective financial cost on those commercial transactions on which such costs were observed. The Heckman model was chosen to correct for any selection biases resulting from the choice of the dependent variable. - B.32 - significant variables. The effects of the mentioned variables are rather important in terms of magnitude and very consistent with the ones found for the case of the rates of interest on cash loans. Table B9. Expected Differences in the Monthly Interest Rate of Commercial Credit Transactions Associated to Changes in Explanatory Variables. Expected Change in Variable Monthly Rate of Interest Amount (N$) -0.2869 a Repayment term (months) -0.4575 a (transaction costs/amount)*100 -0.0268 c Dummy: Assets purchased (1 = inputs,0 = otherwise) -0.5581 b Dummy: Assets purchased (1 = goods for trading, 0 = otherwise) -0.3698 b Dummy: RE's economic activity (1 = nonfarm ventures only, 0 = otherwise) 0.4366 b Log of RE's enterprise income (N$ and includes on farm consumption) -0.0936 a Population in RE's locality of residence (thousands of people, 1990 census) -0.0324 c a, b, c: See Table Bl. Source: Heckman estimation Collateral and Guarantees on Commercial Credit Transactions 98. The Movable Good data Verbal 2% Promises summarized in 43% Fiduciary- figure 9 signer regarding the use of collateral and legal documents are consistent with Fiduciary- credit markets 52 % that are not so tha te . ot Figure B9. Collateral Required on Commercial Credit competitive. Transactions. (percentage of all transactions) The figure indicates that commercial credit, like cash loans, is granted on the creditor's trust; 95 percent of commercial credit transactions were not supported by any form of collateral, including 40 percent of the transactions based solely on the verbal promise of the debtor. The generalized lack of collateral and legal instruments is both a reflection and a cause of highly personal relationships between debtors and creditors. This contrasts sharply with developed financial markets in which transactions are mostly impersonal. - B.33 - 99. It should also be pointed out that commercial credit is fairly inexpensive for REs in terms of transaction costs (TCs). The average imputed value of TCs incurred in receiving commercial credit was N$17 per transaction, while the average TCs represented 2 percent of the amount borrowed. These levels of TCs, albeit very low, are slightly higher than those incurred by REs who borrowed similar amounts from moneylenders and friends and relatives. Table B10. Shares of Different Types of Merchants in the Provision of Commercial Credit to Rural Entrepreneurs (Population Estimates) All Areas Guanajuato Puebla Veracruz Trans. N$ Trans. N$ Trans. N$ Trans. N$ Wholesale traders and merchants 30.23 38.29 25.72 17.27 32.55 45.72 31.85 36.81 Retail traders and merchants 43.29 19.65 47.11 53.62 40.83 7.20 42.30 22.57 Itinerant traders and other merchants 10.76 7.22 10.01 7.52 10.45 10.78 11.52 3.00 Other 15.71 34.84 17.15 21.59 16.17 36.30 14.33 37.63 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. Table 111. Goods Purchased on Credit and Use Given to Them' (Population Estimates Household Total Consumption Business Investments Home Goods for Fixed Durables Other Inputs Trading Assets Other Retail traders and 100.00 19.68 17.22 29.78 23.00 3.03 2.15 merchants (43.29) (8.52) (7.46) (12.89) (9.96) (1.31) (0.93) Itinerant traders and merchants without 100.00 14.36 37.51 14.85 25.49 0.00 3.49 shop (10.76) (1.54) (4.04) (1.60) (2.74) 0.00 (0.38) Other 100.00 13.81 4.64 28.72 42.06 9.70 - (15.72) (2.17) (0.73) (4.51) (6.61) (1.53) - Total (100.00) (13.29) (15.19) (23.01) (39.54) (4.20) (1.58) * Figures with no parentheses correspond to the percentage composition of each row. Figures inside parenthesis correspond to the percentage of total transactions represented by column headings (e.g., commercial credit granted by wholesalers to purchase home durables represents 1.05 percent of all transactions). Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. - B.34 - Table 312. Characteristics of Commercial Credit: Amount, Repayment Terms, and Implicit Interest Rate Characteristic Mean Std Deva C.V. Min Median Max Amount (new pesos) Wholesale traders and merchants 2,302.46 4,727.21 205.31 8 500 22,000 Retail traders and merchants 888.16 1,396.68 157.26 5 465 12,500 Itinerant traders and other merchants without a shop 1,176.38 3,689.73 313.65 35 220 20,000 Other 4,198.00 8,904.42 212.11 45 800 30,000 All transactions 1,880.08 4,806.22 255.64 5 490 30,000 Repayment terms (months)b Wholesale traders and merchants 1.77 2.64 149.10 0.03 1.00 12.00 Retail traders and merchants 3.85 5.11 132.91 0.23 2.00 30.00 Itinerant traders and other merchants without a shop 2.64 2.72 103.09 0.07 1.50 12.00 Other 2.94 4.36 148.07 0.03 1.00 18.00 All transactions 2.88 4.18 144.94 0.03 1.00 30.00 Monthly effective cost of credit (%)' Wholesale traders and merchants 18.30 73.49 401.52 0.00 0.00 593.21 Retail traders and merchants 11.15 18.88 169.34 0.00 3.67 102.57 Itinerant traders and other merchants without a shop Local 38.15 71.94 188.58 0.00 15.62 369.89 Other 9.74 26.12 268.23 0.00 0.00 115.76 All transactions 16.25 51.09 314.30 0.00 0.00 593.21 a. Estimated assuming simple random sampling. b Transactions with a flexible repayment term were excluded from this calculation. c. These effective costs correspond to the internal rate of return on the observed transactions. The calculation of such rates takes into account all relevant attributes such as the frequency of payments, whether interests are calculated on flat or declining balances, and any differences in prices between cash or credit purchases. This is required to compare the rates of the great diversity of transactions observed. Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. -B.35 - Table B13. Use of Collateral and Credit Guarantees In Commercial Credit' (Population Estimates) Movable Informal Total Goods Cosigners Fiduciary Agreement Wholesale traders & merchants 100.00 2.76 2.65 58.36 33.98 (30.23) (0.83) (0.80) (17.64) (10.27) Retail traders and merchants 100.00 1.31 3.77 48.96 32.48 (43.29) (0.57) (1.63) (21.20) (14.06) Itinerant traders 100.00 - - 37.94 62.06 (10.76) - - (4.08) (6.68) Others 100.00 5.30 1.97 27.90 54.76 (15.72) (0.83) (0.31) (4.38) (8.61) Total (100.00) (2.23) (2.74) (47.30) (39.62) Note: Figures in parentheses indicate the percentage of total transactions represented by the respective column heading (e.g., commercial credit granted by wholesalers using movable goods as collateral represents 0.83 percent of all transactions). Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. Table B14. Characteristics of Commercial Credit Transactions: Amount, Repayment Terms, and Financial Cost Std Characteristic Mean Dev* C.V.* Min Median Max Amount (nuevos pesos) Positive financial cost 2,860 6,237 218 35 800 30,000 Zero financial cost 1,131 3,103 274 5 300 30,000 Repayment term (months) Positive financial cost 5.21 5.03 96.53 0.03 4.00 30 Zero financial cost 0.92 1.75 189.44 0.03 0.50 18 Monthly rate of transactions with positive financial cost 36.48 71.55 196.12 1.02 17.33 593.20 Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. - B.36 - Table B15. First Stage of Heckman: Binomial Probit Model Variable Coefficient Std. Error Dependent Variable: 1 = transaction with reported positive interest rate, 0 = otherwise Intercept 0.1353 0.2787 Dummy: 1 = Wholesale, 0 = otherwise 0.2060 0.2850 Dummy: 1 = Retail, 0 = otherwise 0.4942 0.2673 * Dummy: 1 = Itinerant traders, 0 = otherwise 0.9185 0.3475 * Number of previous transactions with commercial credit provider -0.0291 0.0047 * Percentage of population working in primary sector in RE's locality of residence -0.9935 1.7940 Log-Likelihood -138.0405 Likelihood ratio test: Chi-Squared ( 5) 58.3437 Sample size 242 a. Significant at the 90 percent confidence interval. b. Significant at the 95 percent confidence interval. c. Significant at the 99 percent confidence interval. - B.37 - Table B16. Sample Selection Model: Two-Step Heckman Least Squares Regression Variable Coefficient Std. Error Intercept 5.8465 0.5512 * Log of trade credit amount (N$) -0.2869 0.0652 *** Log of repayment term (months) -0.4575 0.0804 *** Distance between RE and Commercial Credit Provider 0.0027 0.0033 (kilometers) (transaction costs/ amount)*100 -0.0268 0.0131 ** Dummy: Good bought (1 = inputs, 0 = otherwise) -0.8167 0.2808 *** Dummy: goods bought (I = goods for trading, 0 = otherwise) -0.4618 0.2383 ** Dummy: RE's economic activity (1 = only nonfarm venture, * 0 = otherwise) 0.3623 0.2165 Dummy: Gender of RE (1. = male, 0 = female) 0.2723 0.2019 Dummy: Ethnicity of RE (1 = indigenous, 0 = nonindigenous) -0.2126 0.2792 Education Level 0.0376 0.0375 Log of RE's enterprise income (N$ and includes on farm consumption) -0.0936 0.0492 * Log of wealth (real estate and bank deposits) 0.0260 0.0169 Population in RE's locality of residence (thousands of people, 1990 census) -0.0324 0.0132 *** I (Lambda) 0.1330 0.2646 R' 0.5146 Adjusted RI 0.4423 AIC 0.6902 Standard error corrected for selection 0.7852 p (Correlation of disturbance in regression and Selection 0.1694 Criterion) Sample size 109 F A 7.1186 a. Significant at the 90 percent confidence level. b. Significant at the 95 percent confidence level. c. Significant at the 99 percent confidence interval. - B.38 - Forward Sales and Sales with Downpayment 100. For the purposes of the survey, forward sales and sales with downpayment were defined as those commercial transactions in which REs received cash payments in exchange for future delivery of products and services. For ease of exposition, both terms-"forward sales" and "sales with downpayment"-are used interchangeably unless otherwise indicated. These transactions represent a source of funding similar to credit in as much as they allow REs to use resources not yet earned. This form of financing has been widely studied in several other settings, especially in the agricultural sector of Asian countries, where it has been generally termed "interlinked credit transactions. " 101. Among the three types of credit services included, forward sales and sales with downpayment are the least significant source of funding for RES in the regions studied. In particular, 5 percent of REs had access to forward sales, such operations accounted for 8 percent of all credit transactions in which REs participated, and the amounts involved represented 4 percent of the volume of credit received by REs. 102. A word of caution is in order concerning forward sales and sales with downpayment, however. Some REs use this type of transaction not as funding instrument (e.g., credit) but rather as a mechanism to induce compliance in their customers in certain agreements or contracts. This function of sales with downpayment has little to do with financial markets. Examples are the advanced payments required by tailors and bakers in order to produce goods that are more valuable to a particular individual than to the market at large (e.g., tailor-made garments and birthday cakes). Since "no-shows" inflict costs on the REs, a security deposit may be demanded. 103. Another, more subtle, example maybe found in forward sales of perishable crops. Producers may use such transactions to minimize price risk rather than for funding purposes. A producer may agree to deliver his crop at a fixed contracted price, below the price expected at harvest. Observed payments, rather than credit, may be a mechanism to make sure that the buyer will purchase the crop even if the spot price at the time of delivery is below the previously agreed upon price. In the absence of posted bond (i.e., advanced payment) the buyer may behave opportunistically by purchasing the crop only if the spot price at the time of delivery is greater than the contracted price. This possibility is discussed in detail as part of the case studies of lender technologies. - B.39 - 104. In any event, Table B17 classifies observed forward sales in several categories and according to whether the commodity is purchased by an intermediary or the final consumer. The most common transactions were those in which final users were involved (70 percent). On the other hand, transactions involving agricultural products were less common than advances on services to be performed and on small manufactured goods (e.g., furniture). 105. The overall scarcity of agricultural forward sales or interlinked transactions in the regions studied contrasts with the stylized facts that characterize rural financial markets in other developing countries, particularly in Asia. These financial arrangements may not have been developed in the areas surveyed because until very recently marketing of the main inputs and outputs has not only been heavily controlled but has also been carried out by the government. For example, the government monopoly FERTIMEX supplied fertilizer and production was marketed through agencies such as INMECAFE and CONASUPO. The continued reform of the agricultural marketing system should allow for competitive goods and financial markets to assist in the provision of rural credit and other services. Such reform should be directly associated with any strategy aimed at developing the country's rural financial markets. 106. Forward sales to intermediaries are quite different from sales with a downpayment to final users, at least in terms of the average amounts involved and the time that elapses between the cash advance and the delivery of the product (see the table below). Intermediaries provide advanced payments almost 10 times larger than final consumers, while the average period of time between cash disbursement and delivery of the goods sold is four times longer for intermediaries. 107. Forward sales and sales with downpayment fit the pattern of informality observed for most financial transactions in the regions surveyed. The vast majority of forward-sale contracts are rather informal and built around the trust of the parties involved, as indicated by the fact that 92 percent of them were based on verbal agreements. In 8 percent of transactions the sellers were required to sign a promissory note (pagare) for the amount received in advance, and only 2 percent of the transactions were formalized by a legal contract (contrato privado) establishing the commitments and responsibilities of the parties involved (e.g., date of delivery of the goods, quality requirements). - B.40 - Table 317. Forward Sales and Sales with Downpayment: Products Sold* (Population Estimates) Agric. and Cottage Indus. and Livestock Sea Small Scale Total Products Products Services Manufacturing Other Intermediaries' 100.00 55.38 14.72 - 11.72 18.17 (32.70) (17.87) (4.82) - (3.83) (5.94) Final consumer 100.00 2.17 - 38.51 52.33 6.99 (67.30) (1.46) - (25.92) (35.21) (4.70) Total (100.00) (19.33) (4.82) (25.92) (39.05) (10.64) Note: Figures in parentheses correspond to the percentage of total transactions represented by the respective column heading (e.g., forward sales of agricultural products to intermediaries represent 17.87 percent of transactions). a. Includes all those who were not the final user of the goods purchased with a downpayment, such as processors and traders Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. Table B18. Characteristics of Forward Sales and Sales with Downpayment: Amounts and Terms to Delivery of the Good Characteristic Mean Std Dev' C.V.* Min Median Max Amount (new pesos) Intermediaries ' 2,953.05 5,300.75 179.50 17 400 16,000 Final consumer 305.49 389.65 127.55 3 100 2,000 Total 1,159.89 3,271.12 282.02 3 200 16,000 Repayment terms (months) Intermediariesh 2.24 2.65 118.47 0.03 2.00 10.00 Final consumer 0.61 0.51 83.30 0.03 0.50 3.00 Total 1.07 1.65 154.10 0.03 0.50 10.00 a. This value was estimated assuming simple random sampling b. Includes traders and processors, and in general all those who were not the final user of the goods purchase with a downpayment. Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. ANNEX C THE PARTICIPATION OF RURAL ENTREPRENEURS IN FINANCIAL MARKETS 1. The participation of REs in financial markets is analyzed in three stages, beginning with credit markets. This first stage includes all three classes of credit products discussed previously (cash loans, commercial credit, forward sales) granted by any type of lender. In addition to descriptive statistics, a multinomial logit model was fitted to the survey data in order to summarize the most important determinants of whether a given RE would participate in credit markets at all and what class of credit service she or he would be more likely to obtain. 2. The second stage of the analysis focuses on cash loans, in view of the their magnitude (73 percent of the total volume of transactions) and because this market segment is where policy interventions are more likely to be relevant and effective.' Three issues are examined in detail: (a) what explains whether REs request and received cash loans from any source; (b) what factors influence the matching between certain groups of borrowers (e.g., farmers) with specific lenders (e.g., banks), considering only those REs who actually obtained loans; and (c) why do the large majority of REs not apply for loans from any source? Descriptive statistics and simple econometric exercises were used to analyze these issues. 3. The participation of RE in financial markets as depositors or savers is also examined. Summary statistics on the deposit services received by REs from formal financial intermediaries and on their participation in rotating savings and credit organizations (ROSCAS) are provided. ROSCAS are informal groups of individuals who collect a pool of resources that is then allocated among themselves. These groups are ubiquitous in Mexico, where they are known as tandas. Reference is made to policy interventions in credit markets. Any program aimed at improving the functioning of rural financial markets should include a component of savings mobilization. In fact, one of the advantages of cash-loans over other types of credit services is that there may be strong economies of joint production (i.e., economies of scope) between cash-loans and deposit services-not present in alternative combinations with other credit services-such as commercial credit. - C.3 - The Participation of Rural Entrepreneurs in Credit Markets 4. REs Coa...cCamCreial Frard participate insales participat inoan & Commercial Credit 8% credit markets only to a limited La Oer Combinations degree. The majority of REs did not receive credit in the two None years pnor to 1s6 the survey: only 1 45 percent of Figure C1. Participation of Rural Entrepreneurs in Credit them received Markets (percentage of all entrepreneurs) any type of such service-cash loans, commercial credit, or advanced payments on future delivery of goods or services. The rural credit markets of the regions studied differ significantly from those of other developing countries in at least two dimensions. First, the observed credit market participation in the areas studied is significantly lower than the rates reported in the rural areas of other countries. For example, studies in selected rural areas of Bolivia, China, Costa Rica, El Salvador, Nigeria, The Philippines, and Thailand report much larger rates of RE participation in credit markets, especially by farmers. The reported rates range from 66 percent to 97 percent and refer mainly to cash loans (see Table C3). In contrast, less than a quarter of REs received cash loans in the areas surveyed during the period under study. 5. Second, the general view, formulated primarily in the Asian countries, of rural and other small entrepreneurs having a portfolio of different credit services from alternative sources is inconsistent with the findings of the survey. The data indicate that in the regions studied the comparatively few REs who did receive credit services tend to be rather "monogamous" debtors as the majority of them receive only one type of service (e.g., commercial credit) from only one sector (e.g., informal), and from a single source (e.g., merchant). Furthermore, the data also suggest that creditor-debtor relationships are maintained through long periods of time. 6. The actual access to credit services (i.e., coincidence of demand and supply decisions) is determined by the interaction of several variables. Access to any of the three credit services considered in the survey varies according to the sector in which REs operate. FREs have the lowest credit market participation, as only 33 percent of them received any type of credit, mostly loans. Credit market participation increases for NFREs as 50 percent of them received credit services, mostly commercial credit. In addition, 62 percent of - C.4 - those REs who combined farm and nonfarm ventures received some form of credit, which gives this group the highest rate of credit market participation, mostly commercial credit. 7. The types of financial services to which REs have access are also associated with the economic sector in which they operate. A larger share of FREs (24 percent) received cash loans than corresponding NFREs (19 percent). In contrast, only 16 percent of FREs have commercial credit in comparison with 34 percent of NFREs. REs who combine farm and nonfarm ventures have the highest rates of access to both cash loans and commercial credit; 26 percent and 43 percent, respectively. 8. Access to credit services in the aggregate of the three regions is comparable for male and female REs. Very similar percentages of male REs and female REs report at least one credit transaction, close to 45 percent. This balance is not maintained across states, however. In the states of Guanajuato and Puebla, proportionately more male REs have access to credit services than female REs. Puebla is located at one end, where 42 percent of male REs and 33 percent of female REs have received credit. Veracruz is located at the other end in the sense that a smaller percentage of male REs (48 percent) have access to credit services than female REs (62 percent). 9. Credit market participation by male and female REs differs most significantly with regard to the kind of credit services to which they have access. Male REs have more access to cash loans than their female counterparts. In the three regions as a whole, 25 percent of male REs received cash loans while the rate for female REs is 16 percent. These differences are more pronounced in Guanajuato and Puebla, where the proportion of male REs with cash loans is twice that of female REs. In Veracruz, in contrast, the proportion of female REs with cash loans is basically equal to that of male REs: 22 percent. 10. In contrast, female REs have higher rates of access to commercial credit than male REs. In the three regions as a whole, 33 percent of female REs and 27 percent of male REs have received commercial credit. This difference extends to Guanajuato and Veracruz. In the case of Puebla, the rates of access to commercial credit are the same for both genders. 11. Ethnicity is associated with low rates of access to credit services. Whereas 48 percent of nonindigenous REs have received some form of credit, only 33 percent of their indigenous counterparts have. Access to alternative forms of credit is even more disparate across ethnic groups. Twenty-four percent of nonindigenous REs have received cash loans. The equivalent rate for indigenous REs decreases to 17 percent. Later it is shown, however, that - C.5 - rather than ethnicity itself, these differences have more to do with other variables also highly correlated to an individual's ethnicity, such as economic activity, wealth, and overall development of the locality of residence. 12. Commercial credit transactions are much more common among nonindigenous REs. This is indicated by the fact that the proportion of nonindigenous REs with commercial transactions is twice that of indigenous REs the three regions. Those rates are 32 percent and 16 percent, respectively. This disparity in access to commercial credit increases in Puebla, where the proportion of nonindigenous REs with commercial credit is almost three times greater than that of indigenous REs. 13. The analysis that follows examines more closely the personal, economic, and household characteristics of those REs who did not obtain credit and those who had access to any type of credit service: namely, commercial credit, advanced payments on future delivery of goods or services, cash loans from the formal sector, cash loans from moneylenders, and cash loans from friends and relatives. Tables C6 and C7 present the results of a regression fitted to the data for that purpose.2 In particular, Table C7 displays 10 variables that have statistically significant marginal effects on the probability that the average RE either did not participate in credit markets or that otherwise obtained a given type of credit. In other words, these estimated probabilities and the ensuing marginal effects refer to the likelihood that an actual transaction had taken place. Although elements of demand and supply are simultaneously involved, the results shed light on what makes borrowers request credit from a given source and what makes lenders willing to grant credit. 14. The results confirm that alternative types of lenders-particularly those who provide cash loans-specialize in providing loans to REs who are statistically different in terms of their economic activities, their income levels and wealth endowments, characteristics of the locality where they reside, and other demographic variables. By the same token, these different borrowers also specialize in obtaining loans from a particular type of lender. 15. Notwithstanding very low rates of credit market participation,those REs who did have access to credit conform to expected patterns of sorting or matching between borrowers and lenders as follows. Those REs widely regarded as better-off are the ones who have a better chance of gaining access to the formal sector. Hence, an individual who is wealthy, has a complex and large enough business to keep accounting records, who has established a credit 2 The econometric technique used is a multinomial logit model. The corresponding technical details are provided by Maddala (1993) and Judege et. al. (1991). -C.6 - history, and who is a farmer is the most likely of otherwise average REs to have received a loan from the formal sector in the last two years. On the other hand, the REs most likely to have received a loan from moneylenders are those who work outside agriculture, have large families, have borrowed in the past, and live in small towns and cities. Finally, the REs who have the best probability of having received a loan from friends and relatives are those who are younger, who experienced an economic crisis in the six years preceding the survey, have had credit in the past, and live in localities whose economies are less dependent on the primary sector of the economy (i.e., less agrarian). 16. As already mentioned, the formal sector is more likely to have lent to (a) REs who are wealthier-as proxied by the value of their real estate holdings and deposits in financial institutions. Wealthier individuals may not only be better credit risks (collateral availability) but also may demand comparatively larger loans-which formal intermediaries like. Overall, the formal sector provided 56 percent of its loans and 87 percent of the total amount of credit to individuals who belong to the wealthiest population quartile. Those percentages increase to 80 percent and 90 percent, respectively, in the case of banks. On the other hand, the formal sector represents 15 percent of all credit transactions and 65 percent of the volume of credit received by the top wealth quartile of REs. (b) REs who kept accounting records. Such records are an important ingredient in the traditional technologies used by formal lenders--especially banks-to screen borrowers. Hence REs who have formal accounting records of their ventures reduce the cost for a formal lender of establishing their credit- worthiness, which makes them more attractive clients. On the other hand, once a business has kept accounting records for its own purposes, the marginal cost of using them to signal creditworthiness to a lender should be very low, if not zero. This reduced marginal cost for the applicant should make bank loans attractive for potential borrowers when choosing their sources of credit. (c) REs who had borrowed in the past as proxied by having any form of credit six years prior to the survey. This variable may influence the probability of having received a loan in the last two years in several ways. Presumably, obtaining a loan should be less costly or difficult for those individuals who already have established their creditworthiness and know the -C.7 - screening techniques used by lenders. This is consistent with reports of lenders in the sense that a good reputation and availability of references are important conditions to grant loans (see lender case studies in Annex E). This variable is very significant-99 percent confidence interval-in explaining access to all five types of credit services studied. Not having a source of credit six years before the survey-on the other hand-also increases very significantly the probability of not having received credit in the last two years (see below). (d) REs in the agricultural sector. This effect is most likely explained-on the demand side-by the fact that the formal sector provides loans with term and repayment conditions that may be more amenable to the cash flows of agricultural and livestock investments; while on the supply side, most government-sponsored credit programs are aimed at agriculture (FIRA, BANRURAL) or at REs who are typically devoted to agriculture (indigenous REs, Instituto Nacional Indigenista). As a matter of fact, 70 percent of the individual transactions and 85 percent of the amount of credit granted by banks to REs were given to individuals who had agricultural ventures. On the other hand, bank loans are the single most important source of credit for farm REs as they accounted for 70 percent of the total amount of credit-from all sources-received by the group. 17. For their part, moneylenders are more likely (i.e., statistically significant marginal effects) to have lent to-otherwise average-REs who: (a) Have nonagricultural activities. The characteristics of the loans granted by moneylenders (e.g., short term, frequent installments) are definitively more consistent with the cash- flows cycles of nonagricultural activities (e.g., trading, services). Also, most moneylender loans were received by REs who are the lender's neighbors-92 percent of such loans were received by REs who live within 10 kilometers from the lender. In lieu of geographical diversification, moneylenders may choose to lend to individuals whose incomes are not as dependent on regional climatic conditions-as is the case for nonfarm REs. Moneylenders allocated 60 percent of their loans and 69 percent of their volume of credit to REs who had nonagricultural ventures. At the same time, moneylender loans accounted for 10 percent of the total credit amount received by REs with nonagricultural activity and for 20 percent of the - C.8 - credit received by REs who combined agricultural and nonagricultural ventures. (b) Have larger families. Larger families should haveeverything else the same-more cyclical (e.g., several children at the beginning of school year and christmas), and unexpected (e.g., medical emergencies) consumption needs. This feature may trigger recurrent demands for loans with: short term, rapid disbursement, and that can be freely used to finance household consumption. These three requirements are met by moneylender loans. Cyclical and unexpected demands for short term credit are even more likely for households which who do not keep cash deposits-for whatever reasons-to manage their cash-flows. This is the case of the vast majority of REs (see below). (c) Who live in smaller towns. Transaction costs of lending and borrowing are the prime suspects to explain this finding. Screening borrowers on the basis of their character and enforcing contracts locally may be cheaper for a moneylender who lends in a small locality. This is because the costs of gathering information about loan applicants may be just a by- product of geographical proximity. As indicated in Annex B, the large majority of moneylender loans are based on trust -80 percent of them were not collateralized. Such contracts require of the personal relationships between borrowers and lenders that are better fostered in a smaller town. The flip side of this-also related to transaction costs-is that smaller towns are also less likely to have a bank which-everything else constant-makes bank loans more expensive (i.e., traveling required). At the same time residents of small localities may also signal their credit-worthiness to their money-lending neighbors at lower cost relatively to a bank. The data presented in Annex B shows that in fact moneylender loans implied the lowest transaction costs for borrowers among all types of cash loans. (d) Have borrowed in the past. As mentioned before, this variable is associated with an increase in the probability that REs have received any type of credit-including loans from moneylenders. 18. Friends and relatives-the other lenders who provide cash loans-are more likely to have lent to otherwise average REs who: - C.9 - (a) Are male. Loans from friends and relatives are the only type of credit service in which the borrower's gender affects the statistical likelihood of the transaction. Loans from friends and relatives-which are at zero nominal interest-are basically granted because of altruistic or mutual insurance motives across households; thus it is likely that such loans are granted and requested by the heads of households. In turn, in the vast majority (85 percent) of the households surveyed, the respondents claimed that their households were headed by a male. Friends and relatives allocated 78 percent of the individual loans and 84 percent of the amounts of credit granted to REs to male borrowers. Loans from friends and relatives represented 3 percent of the total amount borrowed by male REs and 6 percent for female REs. (b) Are younger. The age of REs may capture determinants of credit demand related to life-cycle stage. Younger individuals may be more likely to obtain a loan from friends and relatives than otherwise average REs because of the "not-for-profit" nature of these loans. Younger individuals-albeit more entrepreneurial-may not be able to demonstrate creditworthiness beyond their circle of friends and relatives and may be higher risks as they are experimenting with new economic activities. An interesting related statistic is that the only lenders who granted cash loans to REs in order to start up enterprises are precisely friends and relatives. Friends and relatives represented the main source of funding to start up ventures in 12 percent of nonfarm enterprises, while the formal sector and moneylenders together accounted for about 3 percent of ventures. (c) Have experienced an economic crisis. An economic crisis may trigger a demand for credit in order to spread through time the effects of adverse realizations of disposable income. Yet altruistic or lenders for reciprocity are the most likely to lend to individuals who may not be very good credit risks anyway. (d) Live in towns and cities that are less dependant on the primary sector of the economy (i.e., agrarian) as measured by the percentage of a locality's population who work in the primary sector, according to census data. The effect is not surprising because of the highly correlated cash flows of agrarian economies. Correlation of cash flows hinders local financial intermediation of any kind. The sense of the - C.10 - argument, however, requires that friends and relatives of REs who live in agrarian towns also tend to either live nearby or in another agrarian town, which seems reasonable. As shown in Table ?, the variable has a very large-in absolute terms- estimated marginal effect, which suggests a rather rapid reduction in the probability of receiving loans from friends and relatives on the part of otherwise average REs who live in agrarian localities. (e) Have borrowed in the past. Friends and relatives conform to the behavior of other lenders in that they are more likely to have received a credit request and to have lent to REs who had borrowed in the past. 19. Few variables have statistically significant marginal effects in explaining access to the remaining two credit services (forward sales and commercial credit). Such is also the case with the REs who did not obtain credit at all. Advances on the future delivery of goods and services are more likely to have been received by individuals who work in nonagricultural activities and by those who had credit in the past. This is consistent with the evidence presented in Annex E regarding the use of advanced payments as mechanisms of contract enforcement primarily by REs working in cottage industries and small manufacturing (e.g., tailors, bakers, carpenters). Eighty- eight percent of all transactions in which REs received advanced payments involved REs who have at least a nonagricultural venture. 20. Commercial credit is more likely to have been received by individuals who had credit in the past. The probability that an otherwise average RE did not have any credit in the past two years increases with the individual's age and with not having had credit in the last six years. The effect of not having had any credit six years ago is substantial, as indicated by the large-in absolute terms-negative marginal effect presented in Table C6. The Participation of Rural Entrepreneurs in the Market for Cash Loans 21. Cash loans accounted for almost three-quarters of the aggregate amount of credit received by REs in the two years prior to the survey. The market for cash loans, on the other hand, is where policy interventions are more likely to be relevant. Therefore, in order to analyze the matching of borrowers and lenders in greater detail, the same econometric model (i.e., multinomial logit) - C.11 - previously fitted to the entire sample of REs was applied to the subset of REs who obtained cash loans. 22. The results, presented in Table C9, corroborate the described pattern of matching between creditors and debtors, specifically in loan markets. The model yielded other variables, in addition to most of those examined above, that also have statistically significant marginal effects on the probability that REs with certain characteristics had obtained a loan from one of the three possible kinds of lenders." In other words, basically the same set of variables already analyzed plus the ones described immediately below have significant marginal effects. 23. Rural entrepreneurs are more likely to have received a cash loan from the formal sector as the value of following variables increase: (a) enterprise income, (b) age of the RE, and (c) population of the locality in which they reside. The probability of a loan from the formal sector is also higher for those REs who did not suffer an economic crisis in the six years preceding the survey. These variables are significant in addition to (a) owning an agricultural enterprise; (b) being wealthier; and (c) keeping accounting records, as indicated above. These results confirm that those REs who were more likely to have received a formal sector loan were also those generally regarded as better-off. 24. Among REs who obtained loans, two additional variables turned out also to have significant marginal effects on the probability that they had obtained loans from moneylenders. In particular, REs whose enterprises generated less operational income and who live in towns with lower ongoing wages were more likely than otherwise average borrowers to have obtained their loans from moneylenders than from the other sources. The local wages used correspond to the ongoing daily wage for an unskilled worker (i.e., jornal) at the time of the survey. Such variables may proxy a locality's wealth (e.g., capital/labor). The other variables that yielded significant marginal 3 In this new regression only those REs who received cash loans are taken into account. Thus, the dependent variable includes three possibilities: cash-loans from formal lenders, cash-loans from moneylenders, and cash-loans from friends and relatives. The regression model applied to the entire sample included three additional options: advanced payments on future delivery of goods and/or services, commercial credit, and no credit at all. One of the reasons why there are additional variables whose marginal effects are significant now but were not significant before is that the marginal effects presented correspond to the partial derivative of the probability of each event (i.e., a loan from a type of lender) with respect to each of the independent variables evaluated at the mean value of the vector of independent variables. Clearly, the average value of such vector for the REs who obtained credit should be different to that of the entire sample. - C.12 - effects in this regression-economic sector in which REs work, and population of the locality where REs reside-maintained the direction of their effects, as already indicated. Similarly, lower wealth endowments and lack of accounting records made borrowers more likely to have requested and received loans from friends and relatives rather than from other sources. The other variables with significant marginal effects on the probability of receiving loans from friends and relatives were the age of REs and whether they have experienced an economic crisis. The effects of these variables are as analyzed before. 25. The matching of borrowers and lenders in the market for cash loans fits expectations about the types of REs who have access to certain kinds of lenders. The fact remains, however, that the large majority of REs did not have access to cash loans-most of them because they did not apply for such loans. This and other key issues regarding access to cash loans in the regions studied are summarized below: (a) i. Rural credit FORMALSECTOR markets are shallow. Less Mn-charo than a quarter of Or Formal REs received INFORMAL SECTOR cash loans from Monylonder any FrMends Relaftves source-formal . NON-8RROWERS or informal-in the two years % REs before the Figure C2. Participation of Rural Entrepreneurs in the Market survey. In for Cash Loans particular, (a) about 8 percent of REs borrowed from formal sector lenders-about 3 percent of REs had access to loans from banks, another 3 percent obtained loans from chartered nonbank intermediaries, and 2 percent received loans from other institutional lenders; 5 percent of REs received loans from moneylenders; and 9 percent of REs received loans from their friends and relatives. As indicated earlier, these rates of participation in loan markets are minimal when compared with those reported in selected rural areas of other developing countries. (b) Rural credit markets are very Tragmented and segmented. Rural entrepreneurs are highly specialized or "monogamous" borrowers-particularly in the market for cash loans. On the one hand, few REs borrow from both formal and informal lenders--only 1 percent of them did so (4 percent of those who - C.13 - borrowed). On the other hand, only 1.5 percent of REs obtained cash loans from more than one lender (6 percent of those who borrowed). In addition, credit transactions-once established-last for long periods of time and are repeated as borrowers repeat their applications to their previous lenders. On average, borrowers received their first loan from current sources of cash loans more than six years ago, while the number of previous loans was seven for banks and five for moneylenders. (c) Most REs who did apply for loans were granted credit by the corresponding lender-only 12 percent of loan applicants were rejected. This fact suggests that potential borrowers are fairly accurate at predicting the likelihood of being granted credit. This is important because the effective demand for loans is contingent on the perceptions that potential applicants have about the probability of being rejected, given the costs of the application process. As a matter of fact, 8 percent of REs declared that they did not apply for such service as they thought their applications would be rejected. (d) The large Is too Risky majority of REs-more than Did not need three quarters- Self-selected out did not request loans during the two years prior Cash-1-oan to the survey. owe These 0 8 18 24 32 40 entrepreneurs % REs were asked why Figure C3. Participation of REs in the Market for Cash Loans: they had not Reasons for no Requesting Loans, Denied Loan te Applications, and Kinds of Loans Received attempted to obtain cash loans. On the basis of their answers, these REs were classified in one of three major groups: REs who did not need a cash loan (22 percent of all REs), REs who consider borrowing to be too risky (20 percent of REs), and REs who self- selected out of credit markets because of the process involved in obtaining loans (27 percent of REs). This last group combines those REs who answered that they had not applied for loans because of their belief that their loan applications would be rejected (8 percent of REs); their perception that transaction costs of borrowing are too high (9 percent), and their uneasiness - C.14 - about dealing with financial institutions (9 percent). Another 6 percent of REs gave other answers. 26. The multiple characteristics of the respondents that fell into each group can be summarized using a multinomial logit model. Ten variables yielded statically significant average marginal effects on the probability that any given RE's answer would fall into one of the three possible categories. 27. The most likely REs to have answered that they did not need credit are: REs who have smaller families, REs who did not experience an economic crisis in the last six years, and REs who have more education. These results are consistent with previous results suggesting that REs who have larger families and who had experienced an economic crisis were more likely to have obtained a loan. 28. The REs more likely to have answered that they self-selected out of the market for cash loans are those whose ventures produce lower operational income (i.e., small enterprises), those who experienced an economic crisis in the last six years, those who are less educated, and those who live in towns and cities where the ongoing daily wage was low (i.e., poorer localities). This group of REs should be of special interest because they represent individuals who did not request a loan because of the process involved, even though they could have used a loan. Hence the demand for loans. on the part of these REs seems to be rather elastic with respect to the loan application process. 29. Finally, the most likely REs to have answered that borrowing is too risky are those who'are female, are indigenous, have presumably simple and small ventures that do not require of accounting records, live in locations in which higher percentages of the population work in the primary sector (i.e., agrarian localities), and live in larger towns and cities. It is interesting to note that indigenous REs were found to be less likely to have experienced an economic crisis (see Annex D) than their nonindigenous counterparts. This strongly suggests that indigenous REs may pursue investment strategies deliberately geared--certainly at a high cost-to minimize variations in their incomes (i.e., risks). 30. The fact that 20 percent of REs claim that even though they may had required a loan they did not apply for one because they considered borrowing as too risky-however difficult to interpret-implies that individuals who responded in such manner must have considered that defaulting on the loans they would had obtained would have meant personal losses for themselves. Loans that could be defaulted at no cost for the borrower are risky only for the lender. Borrowers experience a loss from default only if she endures one of the following: loss of personal property either pledged as collateral or seized - C.15 - by the lender through the judiciary, or acquire a bad reputation or damage a valuable relationship with the lender. The obvious implication is that binding loan contracts (i.e., those carrying a penalty for delinquent borrowers) may not be demanded by certain groups of REs (e.g., female, indigenous) unless the remaining elements of their economic reality do not change-including certain clear inadequacies in the functioning of credit markets. Participation of Rural Entrepreneurs in Financial Markets as Depositors and Savers 31. . The survey results confirm widely held notions that REs-as a subset of the rural population-have comparatively low rates of participation in financial markets as depositors of formal financial intermediaries. The survey also revealed that the majority of REs do not participate in informal rotating savings and credit associations (ROSCAS)-known in Mexico as tandas-which are an alternative mechanism for accumulating financial savings. 32. The percentage of REs who at the time of the survey had deposits in formal financial intermediaries was 9 percent. Guanajuato is the region with the largest percentage of REs having deposits in formal intermediaries (12 percent). The proportions of REs with deposits in Veracruz and Puebla were 8 and 7 percent, respectively (see Table C12). 33. Savings accounts are the most common deposit instrument held by REs: 54 percent of accounts. Checking accounts and certificates of deposit represent 26 percent and 14 percent of the number of accounts, respectively. On the other hand, banks have the largest share of accounts-67 percent-while credit cooperatives (i.e., cajas populares) have 19 percent of the three regions as a whole. The shares of other chartered nonbank intermediaries is for all practical purposes negligible. There are significant differences in the relative importance of alternative intermediary charters. For instance, banks have 96 percent of accounts in the Veracruz region and cajas populares have 36 percent in the Guanajuato region. 34. Seventeen percent of REs had deposit accounts in the formal sector but had closed their accounts at the time of the survey. That is to say, about two- thirds of REs who ever open a deposit account in a formal intermediary haver already closed it. This seems to be a very low retention rate of depositors. The large majority of accounts closed-87 percent-belonged to banks. The trend of individuals closing deposit accounts in the formal sector is consistent with the findings of a survey conducted by the Ministry of Finance (SHCP) - C.16 - which indicate that 15 percent of all respondents had held an account in the past whereas only 8 percent had an account at the time of that survey.s 35. At least 56 percent of depositors were net savers in financial markets because-while maintaining deposits-they did not borrow from any source.6 The percentage of REs who were net depositors for the formal sector was to 64 percent. 36. In order to summarize the characteristics of individuals more likely to have deposits in formal intermediaries a binomial logit model was fitted to the data.7 The results of the regression-presented in Tables C18 and C19-indicate that REs more likely to have deposits in formal intermediaries have the following characteristics: (a) Have large, sophisticated, agricultural enterprises. Each of the following variables increases by itself the probability that a RE would have a deposit account-everything else constant: amount of enterprise income (proxy for size of operation and demand for safe storage of cash), the number of permanent paid workers (proxy for size and need to manage cash flows), having formal accounting (proxy for complexity of the venture, e.g., internal control), and having an agricultural venture (may proxy for seasonality of cash flows, i.e., cash surplus at harvest). (b) Are wealthy, married, have fewer dependents, and live in localities where there are bank offices. Again, each of these variables increases by itself -holding everything else constant-the probability of observing that any given RE had a deposit account. It seems natural that wealth-proxied by the value of real estate holdings-should correlate with deposits in formal intermediaries' financial claims on formal intermediaries. The fact that REs who are married and have fewer children are more likely--ceteris paribus-to have deposit accounts is somewhat puzzling, however. The increased likelihood of having a deposit account for REs who live in Secretaria de Hacienda y Crdito Pblico (SHCP). "Encuesta de Servicios Financieros en Poblaciones Rurales y Semi-Urbanas". 1992. 6 The proportion of net savers in the financial sector may be greater as the average amoun of deposits may have been greater the average amount of indebtedness for some of the REs who deposited and borrowed simultaneously. 7 The dependent variable in this model assumes the value of 1 if a given RE had a deposit account at the time of the survey and a value of zero otherwise. - C.17 - localities where there are bank offices only confirms that the demand for deposit services is highly elastic to transaction costs, especially for small depositors. 37. As mentioned earlier, REs also participate in rotating savings and credit associations (ROSCAS) known in Mexico as tandas. ROSCAS are informal groups of individuals who periodically pay a fixed sum into a common pool so that each member-in rotation-can receive the amount collected. This mechanism allows all members-with the exception of the last individual to receive the pool of money-to borrow from the other members of the group. Given that the pool of funds allocated is made up of the periodic contributions of the members, the term to repayment and the amount borrowed by any given individual decreases as the number of disbursements of the pool increases. The member who collects the resources last gets back the sum of its periodical contributions. Tandas have been perceived as a mechanism by which individuals accumulate resources (i.e., save) in order to acquire assets that require lumpy expenditures of cash.8 38. About 10 percent of REs participated in at least one tanda in the period between June 1993 and June 1994. This includes tandas "in-cash" (9 percent of REs) and tandas "in-kind" (1 percent of REs).9 Participation in tandas varies significantly with the economic sector in which REs operate. Only 3 percent of REs who had agricultural ventures participated in tandas while the corresponding proportion for REs with nonagricultural business was 14 percent. The reason for this difference is that tandas require geographical proximity and uncorrelated and steady cash flows on the part of participants. These two characteristics are not common for agricultural ventures. 39. A summary of the personal and business characteristics of the REs most likely to participate in tandas was obtained by applying a binomial logit model to the data. The regression results-shown in Table C17-indicate that those REs most likely to have participated in tandas were younger, more educated, those whose ventures produced higher operational income, those who have nonagricultural ventures, those who additionally worked as salary earners, and those who live in locations with higher local daily wages. An increase in the value of each of these variables is associated with an increase in the probability that a given RE had participated in a tanda-everything else constant. It should be noted that variables widely believed to be highly associated to an individual's participation in tandas were not found to be See: Besley et. al. (1993) 9 Tandas "in-kind" are organized so that members can all acquire a specific asset such as-- for example-automobiles and household appliances. - C.18 - statistically significant: namely, gender, ethnicity, and the population of locality of residence and the percentage of people in that locality that work in the primary sector. Table C1. Participation of Rural Entrepreneurs in Credit Markets Percentage of Rural Entrepreneurs (Population Estimates) Vera- All Areas Guanajuato Puebla cruz Entrepreneurs with credit transactions 44.82 43.80 39.64 51.39 Loans only 12.83 14.16 12.31 12.02 Commercial credit only 19.20 14.86 17.75 25.19 Forward sales only 2.72 2.48 2.85 2.81 Loans and commercial credit 8.07 10.74 5.62 7.95 Loans and forward sales 0.78 0.60 0.51 1.27 Commercial credit and forward sales 0.57 - - 1.77 All three types of credit 0.65 0.97 0.60 0.37 Entrepreneurs with no credit Transactions 55.18 56.20 60.36 48.61 Total 100.00 100.00 100.00 100.00 Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. Table C2. Participation of Rural Entrepreneurs in Credit Markets Percentage of Rural Entrepreneurs-by Gender (Population Estimates) Total of Selected Areas Guanajuato Puebla Veracruz Male Female Male Female Male Female Male Female Entrepreneurs with credit 45.17 43.88 45.25 40.58 41.98 33.00 48.35 61.62 Loans only 14.79 7.43 18.81 3.80 13.38 9.27 12.57 10.21 Commercial credit only 17.36 24.28 11.75 21.80 16.55 21.17 23.32 31.48 Forward sales only 2.53 3.22 2.61 2.18 3.86 - 1.10 8.58 Loans and commercial credit 8.06 8.08 10.48 11.30 6.69 2.57 7.26 10.30 Loans and forward sales 1.07 - 0.86 - 0.69 - 1.65 - Commercial credit and forward sales 0.67 0.29 - - - - 1.98 1.06 All three transactions 0.67 0.58 0.74 1.50 0.81 - 0.47 - Entrepreneurs with no credit 54.84 56.12 54.75 59.42 58.02 67.00 51.65 38.38 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. - C.19 - Table C3. Comparison of the Participation of Rural Entrepreneurs in Credit Markets Across Coountries Borrowers No-Credit Total Formal Formal Informal and Sub- Only Only Informal total I Rural Surveys of Farmers Nueva Ecija Province, The Philippines", 1988-1989 3.9 63.0 26.8 96.1 3.9 100.0 Cochabamba Valley, Bolivia, 1990 - - - 52.2 47.8 100.0 Zaria, Nigeria, 1987-1988 - - - 65.0 35.0 100.0 Nakhon Rachasima Province, Thailand, 1984- 1985 15.9 32.0 9.7 57.6 42.4 100.0 Various Regions, Costa Rica', 1987 28.0 16.7 12.3 57.1 42.9 100.0 Guanajuato, Puebla and Veracruz States, Mexicob, 1994 6.7 25.8 5.6 38.1 61.9 100.0 Surveys of Non-Farm Ventures Iloilo, Negros Occidental, Cebu, and Bohol Provinces (Rural), The Philippinesb, 1991-1992 13.8 28.0 12.5 54.3 45.8 100.0 Zona Urbana del Valle Central, Costa Rica b (Urban), 1993 4.0 49.9 13.6 67.5 32.5 100.0 Guanajuato, Puebla and Veracruz States, Mexico", 1994 2.9 45.6 3.2 51.7 48.3 100.0 Rural Entrepreneurs Guanajuato, Puebla and Veracruz States, Mexicob, 1994 4.2 36.4 4.2 44.8 55.2 100.0. a. Does not include households with only Land-pawing contracts (2.4 percent) b. Refers to entrepreneurs Sources: The Philippines: Nagarajan (1992 Table 8, p. 90); Lapar (1994, Table 3.7, p. 60). Bolivia: Munoz (1994, Figure 7-1, p. 97)); Nigeria: Udry (1990, Table 2, p.255); Thailand: Siamwalla et al (1990, Table 3, p. 277); Costa Rica: Villalobos (1994, Table 14, p.63), Quiros (1991, Table 22, p.86 and Table 25 p.97) ; Mexico: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural, 1994 - C.20 - Table C4. Participation of Rural Entrepreneurs in Credit Markets Percentage of Rural Entrepreneurs-by Ethnicity (Po ulation Estimates Total of Selected Areas Guan4uato Puebla Veracruz Non- Non- Non- Non- Indigenous Indigenous Indigenous Indigenous Indigenous Indigenous Indigenous Entrepreneurs with credit 33.33 48.47 43.80 26.74 47.01 40.49 57.44 Loans only 13.14 12.73 14.16 12.36 12.28 13.99 10.94 Commercial credit only 12.58 21.31 14.86 8.98 22.75 16.48 30.02 Forward sales only 2.79 2.69 2.48 2.46 3.08 3.16 2.62 Loans and commercial credit 3.32 9.58 10.74 2.93 7.15 3.75 10.29 Loans and forward sales 1.04 0.70 0.60 0.80 2.18 0.77 Commercial credit and forward sales 0.45 0.61 0.93 2.23 All three transactions 0.85 0.97 0.94 0.57 Entrepreneurs without credit 66.67 51.53 56.20 73.26 52.99 59.51 42.56 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. - C.21 - Table C6. Rural Entrepreneurs and Financial Transactions Percentage of Rural Entrepreneurs (Population Estimates) Selected Areas Guanajusto Puebla Veracruz Farm Non-arm Both Farm Non-&rm Both Farm Non-&ar Bo&h Fam Non-far Both REs with credit 33.11 50.27 62.26 36.09 46.10 73.67 28.06 45.39 52.28 35.35 59.31 65.86 Loans Only 16.16 10.36 14.49 19.54 10.05 16.39 14.85 10.36 13.60 13.75 10.66 14.13 Commercial credit only 7.77 25.31 30.65 4.68 20.02 34.80 5.11 25.35 ,23.75 14.26 30.46 36.02 Forward sales only 1.12 3.40 5.54 1.23 3.65 2.01 2.03 11.08 4.54 2.93 Loans and commercial credit 7.46 7.98 11.59 10.64 9.52 22.48 6.09 5.60 3.85 5.36 8.89 12.79 Loans and forward sales 0.29 1.23 - - 1.09 - - 0.94 - 0.94 1.65 - Commercial credit and forward sales - - 1.04 - - - - - - - - 3.11 All services 0.32 0.96 - - 1.78 - - 1.11 - 1.04 - - With no credit 66.89 49.73 37.74 63.91 53.90 26.33 71.94 54.61 47.72 64.65 40.69 34.14 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. - C.22 - Table C6. Estiniated Coefficients of Multinomial LoAWt Model of Access to Credit Services (Number in Parenthesis are Standard Errors) Pcic IPH.A. FP NA/ ' P IEd/ P & "A P UM95p P ./P.UA - Variable CWsif. Coaff. C=sff. Couff. Coeff. Intercept -0.6741 -1.0974 -8.3670 -2.7273 0 -0.34494 (0.8844) (1.9220) (1.5570) (1.6090) (1.3610) Dummy: Agricultural vamure -1 .0362 0 ~ -2.8152 1.1995 -1.4689 -0.3442 (I -farm, 0=otherwise) (0.3181) (0.7753) (0.4442) (0.5160) (0.4451) Dommy: Gender -0.0185 0.7051 0.7068 0.6917 0.4543 (I = male. 0=fcmale) (0.2260) (0.4944) (0.5473) (0.5097) (0.3925) Dumnmy:Ethnicity (I-indigenous, -0.1425 0.2645 -0.0086 -0.7531 0.0204 O=non-idgenous) (0.2M9) (0.5063) (0.3072) (0.4932) (0.4112) Log income (includes imputed value of 0.1254 00 0.0816 0.1724 408004 -0.0390 on-farm cosumptiont) (0.0613) (0.1201) (0.0857) (0.0939) (0.0795) Log of wealth (realcamae and bank 0.0014 0.0116 0.1040 000.0344 -0.0170 deposits) (0.0213) (0.0450) (0.0449) (0.0428) (0.0337) Age of RE -0.0164 -0.0089 -0.0152 0.0033 -0.038800 (0.0070) (0.0142) (0.0125) (0.0125) (0.0120) Househtoldusize 0.0243 0.0422 0.0661 0.1751 0 ~ 0.0306 (0.0385) (0.0797) (0.0580) (0.0580) (0.0576) Dummy: Econom-dc Crisis 0.2401 0.4224 -0.2600 0.2262 0.832700 (I = experienced crisis. 0= otherwise) (0.208) (0.422) (0.3494) (0.3899) (0.3402) Dummy: Accounting 0.4740 0 -0.0111 1.206900 -0.1294 -0.2234 (I - Yes, 0=-No) (0.2666) (0.6m2) (0.4445) (0.5965) .(0.5299) Dummy: Credit history of RE 1.4453 00 1.3834 00 2.4535 0* 1.7029 1.8114 (I =credit six years ago, 0=otherwise) (0.2383) (0.4402) (0.3488) (0.3954) (0.3372) Percent pop. working in primary -4.9496 00 0.2646 2.0837 3.344 -4.7224 sector ia locality (2.5360) (4.3100) (4.1910) (3.8930) (3.8750) Population in locality of residence -0.0097 -0.0440 0.0756 00 -0.0888 -0.0140 (0.0223) (0.0521) (0.0393) (0.0571) (0.0373) Minimum Wage in locality -0.0249 -0.0694 0.0928 00 -0.0436 -0.0265 _____________ 1__ (0.02%6) (0.0736) (0.0453) (0.0592) ( 0 . 0 4 9 0 ) Log-Likelibojod -862.816 Likelihood Ratio Teat Chi-Squared (65) 326.3941 Number of Observations 788 * / * / ** significant at the 90%/95%/99% confidence interval respectively N.A. = No Access, C.C. =Commercial Credit Source: Multinomnial Logit Estnmation - C.23 - Table C7. Marginal Effects of Variables on Probability of Having Received Credit P No. C. ..c.0 P A&. Pay.m .. P, 's...,s.a. F & m .A da. Variable Coefmdent Coeident Comident Coeffisent Cosefident Coeffisent Dummy: Only Agricultural venture 0.2132 -0.1509 -0.0727 ** 0.0540 ** -0.0439 * 0.0003 (I =fam-RE, 0=otherwise) (0.1372) (0.1217) (0.0261) (0.0269) (0.0190) (0.0185) Dummy: Gender of RE -0.0600 -0.0256 0.0179 0.0212 0.0233 0.0232 * (0 =male. 0= female) (0.0794) (0.0854) (0.0156) (0.0175) (0.0154) (0.0139) Dummy: Ethnicity of RE 0.0318 -0.0199 0.0093 0.0015 -0.0275 0.0047 (I=indigenous,0=non-indigenous) (0.0871) (0.0883) (0.0152) (0.0171) (0.0173) (0.0155) Log of RE's Productive income -0.0154 0.0220 -0.0031 0.0051 -0.0043 -0.0042 (includes imputed value of on-farm consumption) (0.0198) (0.0217) (0.0028) (0.0037) (0.0033) (0.0034) Log of wealth (real estate and bank deposits) -0.0028 -0.0007 0.0002 0.0035 * 0.0012 -0.0014 (0.0072) (0.0077) (0.0013) (0.0017) (0.0013) (0.0012) Age of RE 0.0042 -0.0021 -0.0001 -0.0003 0.0004 -0.0021 * (0.00124) (0.0025) (0.0004) (0.0004) (0.0005) (0.0006) Household size -0.0109 0.0014 0.0007 0.0017 0.0062 * 0.0008 (0.0129) (0.0137) (0.0017) (0.0020) (0.0027) (0.0022) Dummy: Economic Crisis -0.0729 0.0265 0.0089 -0.0132 0.0042 0.0465 * (I =experienced crisis. 0= otherwise) (0.0692) (0.0726) (0.0108) (0.0140) (0.0130) (0.0144) Dummy: Accounting -0.0772 0.0763 -0.0040 0.0377 * -0.0100 -0.0228 (I =Yes. 0=No) (0.0921) (0.1030) (0.0129) (0.0231) (0.0198) (0.0165) Dummy: Credit history of RE -0.3844 ** 0.1793 *** 0.0223 ** 0.0637 * 0.0421 * 0.0771 * (I =credit history in last six years, 0=otherwise) (0.0900) (0.0774) (0.0098) (0.0192) (0.0139) (0.0207) Percentage of population working 0.7234 -0.8268 0.0426 0.1139 0.1774 -0.2306 * in primary sector in RE's locality (0.7851) (0.8018) (0.1238) (0.1372) (0.1604) (0.1409) Population in RE's locality of residence 0.0032 -0.0010 -0.0011 0.0028 -0.0033 * -0.0006 (0.0081) (0.0089) (0.0012) (0.0023) (0.0016) (0.0013) Minimum Daily Wage in E's locality 0.0047 -0.0038 -0.0018 0.0035 -0.0014 -0.0012 (0.0114) (0.0126) (0.0015) (0.0023) (0.0019) (0.0018) * I ** /*** significant at the 90 % / 95 % / 99 % confidence interval respectively a. Refers to the change in the probability of receiving any type of credit that results from a change in the independent variable -for the average RE (i.e., partial derivative evaluated at the mean value of the vector of explanatory variables) b. Correspond to quasi-elasticities (AP due to A% in productive income) Source: Multinomial Logit Estimates - C.24 - Table CS. Access to Cash Loans Percentage of Rural Entrepreneurs (Population Esthnates) All Areas Guanajuato Puebla Veracruz A. Borrowers Borrowed from formal and informal sectors 1.04 2.03 0.42 0.68 Borrowed from formal sector only 7.55 10.52 5.63 6.57 Borrowed from moneylenders only 5.14 3.88 6.03 5.47 Borrowed from friends and relatives only 8.61 10.03 6.96 8.90 B. Non-Borrowers 1. Requested a loan but application was rejected 3.12 3.27 3.02 3.08 2. Did not request a loan because: 74.45 70.26 77.95 75.01 a. Self-selected out of the market 26.77 25.47 26.09 28.92 b. Did not needed credit 21.89 22.59 23.55 19.42 c. Borrowing is too risky 19.82 17.34 20.72 21.39 f. Other 5.97 4.86 7.59 5.38 C. Total Rural Entrepreneurs 100.00 100.00 100.00 100.00 Source: Encuesta Regional de Servicios Financieros a Unidades de Produccidn Rural. Table C9. Marginal Effects of Independent Variables on Probability of having Received a Loan from Different Lenders a/ (Subset of Borrowers) Variable Coeffident Std. Error Coefidcent Std. Error CofdMent Std. Error Agric -0.1662 0.1371 0.3939 0.1451 **-0.2277 0.1161 * Gender -0.0776 0.1567 0.1624 0.1225 -0.0848 0.1000 Race -0.0680 0.1425 0.1688 0.1200 -0.1008 0.1078 LINC' -0.0141 0.0237 0.0461 0.0224 - -0.0320 0.0183 LWealth' -0.0288 0.0144 0.0284 0.0110 * 0.0004 0.0083 Age -0.0074 0.0039 0.0053 0.0030 0.0022 0.0027 HHSIZE -0.0130 0.0169 -.00056 0.0140 0.0185 0.0127 Crisis 0.2013 0.1195 0.2254 0.1006 c 0.0241 0.0779 Accounting -0.2914 0.1470 o 0.3597 0.1421 - 0.0683 0.1120 Populatoin -0.0020 0.0099 0.0225 0.0108 - -0.0205 0.0093 n Wage 0.0015 0.0146 0.0229 0.0147 -0.0245 0.0125 /* / 0* significant at the 90%/95%/99% confidence intervals, respectively Derivatives at the sample mean Correspond to quasi-elasticities Source: Multinomial Logit Estimation - C.25 - Table CIO. Estimated Coeffcients and Standard Errors of Multinonual Logit for Rural Entrepreseurs who did not Apply for Variable CoefRdient Std. Error CoeMient Std. Error Intercept -1.1260 1.0140 -2.9496 1.0300 0* Dummy: Only Agricultural venture (I -farm RE, 0=otherwise) 0.0858 0.3262 0.0777 0.3194 Dummy: Gender of RE (I -male, 0 -female) 4.2530 0.2539 -0.5064 0.2640 * Dummy: Ethnicity of RE (I -indigenous, 0 -non-indigenous) . -0.0476 0.2980 0.7392 0.2788 * Log of RE's Productive income 0.1080 0.0621 * 0.1234 0.0617 * Log of wealth (real estate and bank deposits) -0.0013 0.0232 0.0106 0.0236 Age of RE -0.0041 0.0079 4.0039 0.0080 Household size 4.0844 0.0452 * -0.0409 0.0427 Dummy: Economic Crisis (I -experienced crisis, 0-otherwise) 4.9785 0.2133 -0.2356 0.2160 Dummy: Accounting (I -Yes, 0=No) 4.0738 0.3073 -0.5979 0.3579 * Education index 0.1557 0.0600 * 0.0929 0.0629 * Percentage of Population working in primary sector in RE's -2.0612 2.7310 4.3116 2.3740 locality Population in RE's locality of residence 4.0060 0.0257 0.0535 0.0264 Minimum Daily Wage in RE's locality 0.0531 0.0319 * 0.0617 0.0339 Log-Likelihood -587.4136 Likelihood Ratio Test Chi-Squared (26) 85.1617 Number of Observations 590 * / ** / 0** significant at the 90%/95%/99% confidence interval respectively Note: The dummy variable for previous credit experience is not statistical significant. Source: Multinomial Logit Estimation - C.26 - Table ClI. Marginal Effects of Independent Variables on the Probabilities of Not Having Applied for a Loan due to One of Three Possible Reasons' P AN Iw.aI4 VM-1 -1 PT,r Variable Coefftcient Std. Error Coeffcient Std. Error Coeffident Srd. Error Dunmry: Only Agricutural venture (1 =frm RE, 0=otherwise) -0.0203 0.0758 0.0113 0.0606 0.0090 0.0534 Dummy: Gender of RE (I=fmale. O= female) 0.0944 0.0583 -0.0135 0.0480 0.0809 0.0445 0 Dummy: Ethnicity of RE (I= indigenous, 0= non-indigenous) -0.0857 0.0693 4.0621 0.0536 0.1478 0.0528 0+0 Log of RE's Productive income' -0.0288 0.0148 00 0.0124 0.0118 0.0164 0.0104 Log of wealth (real estate and bank deposits)' -0.0012 0.0054 -0.0010 0.0044 0.0022 0.0039 Age of RE 0.0010 0.0019 -0.0005 0.0015 -0.0005 0.0013 Houschold size 0.0156 0.0107 -0.0136 0.0084 O -0.0020 0.072 Dummy: Economic Crisis (I= experienced crisis, 0= otherwise) 0.1514 0.0503 -0.1752 0.0417 0oo 0.0238 0.0356 Dummy: Accounting (I=Yes, 0=No) 0.0833 0.0725 0.0282 0.0659 0.1115 0.0555 00 Education Index -0.0310 0.0137 0o 0.0239 0.0114 0o 0.0070 0.0103 Percentage of Population working in primary sector in RE's 0.2771 0.6341 -0.7122 - 0.4681 0.9892 0.4576 o locality Population in RE's locality of residence -0.0059 0.0059 -0.000 0.0049 0.0109 0.0046 o Minimum Daily Wage in RE's locality -0.0143 0.0073 0* 0.0060 0.0061 0.0083 0.0056 o 00 000 significant at the 90%I95%199% confidence interval respectively Derivatives at the sample mean * Correspond to quasi-elasticities Source: Multinomial Logit estimates - C.27 - Table C12. Particpation of Rural Entrepeneurs In Faandal Markets as Depositors of Formal Finandal Intermediaries (Percentages of Rural Entrepreneurs/Population Estimates) Rural Entrepreneurs Selected areas Guanajuato Puebla Veracrz 1. With accounts 8.97 11.99 6.76 8.23 2. With no acounts cuntly 91.03 88.01 93.24 91.77 a. Have had accounts in: 17.26 21.24 12.66 18.11 Conmercial Banks 15.03 18.89 10.38 16.04 Development banks 0.39 - 0.84 0.30 Uni6n de CrMdito 0.21 - 0.38 0.23 Sociedad de Ahorro y Pristamo 0.15 0.44 - - COas Populares 0.68 0.95 - 1.13 Cqyas de Solidaridad 0.16 0.48 - - Other 0.66 0.48 1.06 0.41 b. Never had an account 73.77 66.77 80.57 73.66 Total 100.00 100.00 100.00 100.00 Source: Encuesta Regional de Servicios Financeros a Unidades de Produccitn Rural. Table C13. Types of Financial Savings Accounts Held by Rural Entrepreneurs (Percentage of Number of Accounts/Population Estimates) Type of Account Selected Areas Guanajuato Puebla Veracruz NA 1.25 - - 4.13 Checking Account 26.37 13.99 17.59 51.91 Savings Account 53.95 69.16 51.57 32.87 Certificate of deposit 14.36 13.25 25.28 7.51 Other 4.07 3.61 5.56 3.59 Total 100.00 100.00 100.00 100.00 Source: Encuesta Regional de Servicios Financieros a Unidades de Produccdn Rural. - C.28 - Table Ci4. Distribution of Floancal Savinp Account by Finandal Institudon (Percentage of accounts/Population Estimates) Selected Areas Gnuato Puebla Veracruz NA 6.73 8.59 11.76 Commercial Banks 64.09 43.83 66.73 92.59 Development Banks 2.96 4.06 - 3.59 Unidn de CrAdito 0.63 - 2.65 - Cajs Populares 19.22 35.84 6.79 3.82 Cjas de Solidaidad 1.66 3.61 - Other 4.73 4.06 12.08 - Total 100.00 100.09 100.09 100.00 Source: Encuesta Regional de Servicios Fuanderos a Unidades de Product6n Rura. Table C15. Participation of Rural Entrepreneurs in ROSCAS (Percentae of Rural entrepreneur/Populadon Estnates) Participated in Selected Areas Guanajuato Puebla Veracruz ROSCAS Yes 10.36 11.87 7.16 12.23 'In-cash- 9.65 11.45 5.92 11.79 "in-kind' 0.71 0.42 1.24 0.44 No 09.64 88.13 92.94 o7.77 Total 100.00 100.09 100.00 100.00 Source: Encuesta Regional de Servicos Financieros a Unidades de Producci6n Rural. Table C16. Estimated Coefficient of Binomial Loot Model of Participation of ROSCAS Variable Coefficient Std. Error Intercept -5.1001 1.1299 Age of RE -0.0250 0.0105 c Log of RE's productive income 0.3022 0.0830 0e Number of income generating people in the household 0.1675 0.0905 * Education Level 0.1027 0.0629 0 Dummy: Agricultural venture only (I= farm-RE only, -1.1464 0.4921 00 0=otherwise) Dummy: Gender of RE (1 =male, 0=female) -0.0501 0.2833 Dummy: Ethnicity of RE (1=indigenous, 0=non- -0.1125 0.3985 indigenous) Percentage of population working in primary sector in RE's -2.4136 2.6524 locality of residence Minimum daily wage in the RE's locality 0.0600 0.0321 0 Dummy: Salary income (1 =RE works as salary 1.4831 0.6103 ** worker,0= otherwise) Log likelihood -227.9910 Likelihood ratio test(X ,-.o) 74.5610 AIC 477.9820 Sample size 788 Source: Logit estimation - C.29 - Table C17. Marginal Effects for the Binomial Logit of Participation in ROSCAS Std. Variable Coefficient Error Age of RE -0.0015 0.0006 ** Log of Productive Income 0.0176 0.0045 *** Number of RE's household that generated income 0.0098 0.0053 * Education Index 0.0060 0.0037 Dummy: Only farm Venture (1= only farm venture, 0=otherwise) -0.0668 0.0266 *** Dummy: Gender (I= Male, 0=otherwise) -0.0029 0.0165 Dummy: Ethnicity (1= Indigenous, 0= otherwise) -0.0065 0.0231 Percentage of population dedicated to primary activities in RE's locality of residence -0.1405 0.1536 On-going market wage in borrower's locality of residence (N$) 0.0035 0.0019 * Dummy: Salary Income (1 =RE works as salary workers 0= otherwise) 0.0864 0.0330 *** * / ** / *** significant at the 90 percent, 95 percent and 99 percent confidence intervals respectively. Source: Logit Estimation - C.30 - Table CI8. Estimated Coefficients of Binomial Logit Model of Access to Deposits in Formal Intermediaries Std. Std. Variable Coefficient Error Coefficient Error Dependent Variable: 1 = RE has a financial savings account, 0=otherwise Intercept -4.7872 1.0680 * -5.1274 1.1008 ** Age of RE -0.0062 0.0124 -0.0060 0.0125 Log of Productive Income 0.1907 0.0818 n 0.1994 0.0825 * Log of Real Estate Assets 0.0665 0.0366 * 0.0662 0.0367 * Education Index 0.0561 0.0762 0.0592 0.0771 Dummy: Only Non-farm Venture (1= non-farm venture, -0.9542 0.3685 o -0.9908 0.3709 * 0= otherwise) Dummy: Civil Status (1=Married, 0=oderwise) 0.8709 0.3897 c 0.8388 0.3902 * Dummy: Bank in Locality (1 =Yes,0= no) 0.5085 0.3171 Number of Dependents in household -0.1369 0.0730 a -0.1385 0.0734 * Dummy: Accounting (1= Yes, 2=No) 0.9900 0.3397 e 1.0162 0.3389 * Number of Paid Workers 0.4698 0.1360 ** 0.4357 0.1369 * Dummy: Gender (1 =Male, o=otherwise) -0.1566 0.4309 -0.1126 0.4347 Dummy: Ethnicity (1=Indigenous, O=otherwise) -0.2690 0.4425 -0.1843 0.4507 Population of Locality 0.0562 0.026 AIC 367.12 365.11 SC 427.82 425.82 Log-Likelihood -170.56 -169.56 Likelihood ratio test X2 (d.f.=12) 98.03 100.03 Sample size 788 788 * / ** / *** significant at the 90 percent, 95 percent and 99 percent confidence intervals respectively. Source: Logit estimation - C.31 - Table C19. Marginal Effects for the Binomial Logit of Access to Deposits in Formal Intermediaries Variable Coefficient Std. Error Age of RE -0.0003 0.0005 Log of Productive Income 0.0086 0.0034 *** Log of Real Estate Assets 0.0030 0.0015 ** Education Index 0.0023 0.0033 Dummy: Only Non-farm Venture -0.0383 0.0154 (1 =non-farm venture 0=otherwise) Dummy: Civil Status (1=Married, 0.0357 0.0157 ** 0 = otherwise) Dummy: Bank in l.cality 0.0237 0.0136* (1= Yes,0=no) Number of Dependents in household -0.0064 0.0031 ** Dummy: Accounting (1= Yes, 2= No) 0.0436 0.0154 Number of Paid Workers 0.0205 0.0067 * * / ** / *** significant at the 90 percent, 95 percent and 99 percent confidence intervals respectively. Source: Logit estimation ANNEX D RURAL CREDIT MARKETS AS MECHANISMS TO COPE WITH RISK - D.2 - 1. The importance of finding mechanisms to cope with risk is clearly evident from the economic distress suffered by REs in the past six years. The' discussion now turns to the causes of that distress, the likelihood that individual REs may suffer an economic crisis, and the ex-post strategies that REs use to cope with risk, including the role played by credit markets in such strategies. Economic Crises Suffered by Rural Entrepreneurs 2. Rural entrepreneurs were asked whether they had experienced an economic crisis-defined as a period of abnormally low available income--during the last five years and if so, to identify the year of the most difficulty.' The majority of REs-59 percent-answered that at least one of the years between 1989 and 1994 could be characterized as a year of economic crisis. A third of all REs said they had experienced such a crisis in 1993 and 1994 (see Table Dl). Sources of Risk for Rural Entrepreneurs 3. An Oth2r Low productin economic crisis Sicknoss of RE 23% may be due to a unexpected low levels of income, unexpected high and unavoidable Low pftos expenditures (e.g., disease), or Woakdomand for sericos Low salos to a combination 12% 21% of both. The Figure DI. Sources of Risk that Caused an Economic Crisis survey indicates for Rural Entrepreneurs (Percentage of Rural that the most Entrepreneurs who Suffered a Crisis) common source of economic crisis is negative income shocks-not unexpected expenditures. Three-quarters of those who had a crisis identified low income levels as the main cause of their crises. In particular, low levels of production (e.g., farm REs), low sales (e.g., traders), and weak demand for services were the most commonly cited causes of economic distress as indicated in Table D2. Low prices were mentioned as the cause of crisis by only 6 percent of REs, while It should be noted that the concept of economic crisis used does not include persistent poverty. For the purpose of the survey, economic crisis was defined as a particularly adverse period--relative to normal circumstances. -D.3 - forgone income due to the sickness of the RE resulted in distress for about 4 percent of REs. 4. The sickness of any family member-possibly including the RE- represented the most important cause of an economic crisis among those due to unavoidable expenditures. This type of event resulted in distress for about 18 percent of those who had a crisis (11 percent of all REs). 5. The table also shows that different REs are affected by different sources of risk. As might be expected, these sources of risk are highly correlated to the REs' economic activities, which in turn are related to their personal characteristics.2 Farmers have been more negatively affected by variations of yields and prices than REs in other activities who have been more exposed to low sales or weak demands for their services. In the same way, indigenous REs largely dedicated to agriculture are more susceptible to variations in production than nonindigenous REs. The opposite case is given by female REs mostly dedicated to nonagricultural activities who are distressed more often by reduced sales. The poorest two quartiles of REs are comparatively less susceptible to variations in output and prices than wealthier REs. 6. One very likely implication of the observed sources of realized risks faced by REs is that the more disadvantaged groups of REs may be more vulnerable to economy wide shocks (i.e., low sales, weak demand) while wealthier REs are more exposed to more localized or idiosyncratic sources of risk (e.g., whether, plagues). This empirical regularity becomes particularly relevant in view of the generalized economic crisis currently being suffered by Mexico's economy. The hypothesis is supported by the econometric results of the study. Strategies Used by Rural Entrepreneurs to Cope with Risk 7. Rural entrepreneurs were also asked about the most important strategies they had used to cope with economic crises. Their answers-summarized in Table D3 may be divided in three groups: (a) consumption smoothing intertemporally; (b) consumption smoothing across households; * reductions in household consumption or changes in occupational patterns. 8. In the case of risk-coping mechanisms that smooth consumption intertemporally the effects of bad realizations of income or unavoidable 2 Annex A shows that there is correlation between an individual's characteristics and her/his occupational patterns (e.g., indigenous and male REs are almost in their entirety farmers). - D.4 - expenditures are spread through time. The category includes loans that were obtained from a source that demands interest payments-mostly for-profit lenders; loans that do not require interest payments-mostly from friends and relatives; and depletion of savings or sales of assets that otherwise would not have been sold. Table D3 shows that in the aggregate this type of mechanism was cited by 21 percent of distressed REs. In particular, 7 percent received for-profit loans, 5 percent received interest-free loans, 1 percent delayed payments or defaulted on outstanding debt, and 8 percent sold assets that would not have been sold in the absence of the crisis. 9. The role of credit markets in helping REs to cope with risk is not homogenous-as REs have differential degrees of access to credit markets. Intertemporal mechanisms to cope with risk are mainly used by those REs generally regarded as better-off. There are clear differences-for instance, in the use of for-profit lenders between nonindigenous and indigenous REs; male and female REs; and wealthier (fourth quartile) and less wealthy REs (lowest two quartiles). The traditionally disadvantaged REs (indigenous, female, poor) use credit from moneylenders and formal intermediaries much less often than their counterparts. In particular wealthier REs use loans from for-profit sources almost three times more often than poor REs as a mechanism to cope with risk. This is also the case with the sale of assets which also presents marked differences. Indigenous and poorer REs sell assets to smooth consumption in the presence of a crisis less often than other groups, probably simply because they do not own as many assets and/or they self-consume those assets by which they store value (e.g., animals, corn stocks). 10. In the case of risk coping mechanisms that smooth consumption by sharing risks, the effects of a crisis on consumption are spread across households, especially among family members and friends. The category includes donations and transfers-in kind or cash-from friends and relatives and the government. These transfers are not required to be paid-back although they may entail a commitment for future reciprocity. Table D3 shows that this mechanism was cited by 13 percent of REs who had experienced a crisis: 12 percent received assistance from friends and relatives and 1 percent from the government or other agencies. 11. The remaining strategies followed by REs in coping with crises were to significantly reduce household consumption levels or change occupational patterns. The large majority of responses fell into this category: 59 percent. An increase in the supply of labor by household members outside the family business was reported to be the single most common reaction; it was chosen by 34 percent of REs and their families. Finally, almost a quarter of REs reported that their crises caused drastic reductions in household consumption. - D.5 - 12. The REs that are generally regarded as the most deprived groups tend to increase their labor market participation more than their counterparts in the presence of an economic crisis. That is the case especially of indigenous, female, and poor REs-as indicated by the percentage who responded that an increase in the labor supplied by households members was their main strategy to whether a crisis. On the other hand, those presumably better off (i.e., nonindigenous, males, wealthier) were more prone to reduce consumption levels than their counterparts. 13. The optimality-social and private--of these three general strategies to cope with risk depends on several elements. For instance, transitory income shocks may be dealt optimally through financial markets (e.g., savings and credit) while risk sharing may be appropriate when dealing with idiosyncratic shocks. What seems a matter of concern, however, is that 60 percent of REs who suffered an economic crisis had to reduce household consumption or increase household labor market participation. For instance, it is doubtful that the mostly small and confined markets in which REs function would be able to absorb the increased supply of labor that would result from aggregate shocks such as the country's current crisis. One negative consequence of the observed pattern of responses to a crisis may be labor-driven immigration to urban centers. 14. The probability that REs might experience an economic crisis is analyzed next. Table D4 summarizes the results of a regression that explains-ex-post-the probability that an individual RE had actually experienced an economic crisis as a function of the individual's attributes, the characteristics of her venture, and the locality where she resides. 15. As indicated in the table seven variables are significant in explaining the occurrence of an economic crisis, after controlling for a number of other factors. In particular, wealth levels (proxied by the value of real estate and bank deposits) are positively related to the occurrence of crises, nonindigenous REs were more likely to have experienced distress than indigenous REs, REs who have had access to cash loans (i.e., credit history) had a higher probability of crisis than those who did not have access, REs with larger number of dependents were more prone to a crisis, REs who lived in more agrarian towns (proxied by the portion of economically active population working in the primary sector) were more likely to have had a crisis, REs who live in more populated localities were also more likely to have a crisis, and residents of Guanajuato had lower probabilities of a crisis than residents of Puebla and Veracruz. S The econometric technique used is a binomial Probit model in which the dependent variable is given a value of one if an economic crisis was reported and zero otherwise. - D.6 - 16. These results suggest that the individuals more likely to have suffered a crisis are those who were likely to suffer a crisis because they consistently chose not to reduce or manage risks as intensely as other REs, and those who were likely to have a crisis because they could not reduce or manage the sources of risk that caused their crisis anyway. 17. The argument supporting the hypothesis is that the probability of a crisis is conditional on the ex-ante activities undertaken by REs to manage risk (e.g., diversification, insurance). That is to say, individuals who were systematically less willing or able to avoid risk should also have had a higher incidence of economic crises. On the other hand, the incentives to manage risk depend on risk preferences, the cost of the different strategies (e.g., insurance premia), and the mechanisms to cope with risk available to the individual in case a crisis actually occurs. It follows that-everything else the same-individuals who enjoy dependable mechanisms to cope with a crisis in case it happened may have less incentive to make costly investments to reduce risk ex-ante. 18. The REs who seem to have lower incentives to reduce risk ex-ante are wealthier (have real estate and bank deposits) and hence are more able to self- insure; are nonindigenous (they have less geographically localized networks of family and friends and they are more integrated to the rest of the economy, e.g., access to labor markets) than indigenous people; have had access to credit (they could spread the effects of shocks in time); and live in larger urban centers (they could change jobs and economic activities more easily). 19. In contrast, the REs with less ability to manage risk are those who have more dependents (they may be more vulnerable to, for example, sickness of family members); and those who live in more agrarian localities, where aggregate levels of economic activity are highly dependent on systemic factors such as whether or specialization in one or few regional crops. 20. This evidence from Mexico supports the general view that the need for consumption smoothing leads to a poverty trap for some groups of REs-in particular the poor, who are consistently less likely to be affected by negative realizations of income. Poor REs seem to be using risk reducing strategies that are low risk-low return or other fairly costly means to stabilize consumption. This is consistent with the empirical evidence from other countries indicating that production and investment decisions depend on how households can cope with income risk and that poorer households, in particular, appear to forgo potential earnings to reduce risk.4 4 See: Murdoch (1990) and Rosenzweig and Binswanger (1992). -D.7 - 21. In contrast, those who would be less affected by negative income realizations-who are in fact also more likely to have experienced them-may afford riskier investments with higher expected returns. In the long run, the gap between the not so poor and the poor would only widen if such is the case. This is where efficiency and equity issues merge. 22. Any intervention should consider that, in general, individuals choose a general risk-reducing strategy that combines portfolio diversification, savings and borrowing, and risk sharing, depending on the relative costs and benefits of each strategy. The costs and benefits of one strategy will affect the way other strategies are used, and these interactions should be important for policy design. By taking this into account. it is possible to conceive of a range of possible strategies to mitigate risk-one of which is increased access to safe savings and timely credit services. The efficacy of such strategy depends, however, on transaction costs and contract enforcement. The proposals put forth in Chapter 7 outline mechanisms that can be used to increase access to financial services so as not to erode the possible gains by transaction costs or by problems resulting from weak enforcement of contracts. - D.8 - Table Dl. Rural Entrepreneurs that Experienced an Economic Crisis in Three Selected Rural Areas of Mexico (Population Estimates of Percentage of Rural Entrepreneurs) Total Guangiuato Puebla Veracruz Entrepreneurs with crisis 59.19 56.83 58.94 61.88 Year of Crisis 1989 5.50 5.11 6.16 5.18 1990 7.55 4.95 9.20 8.46 1991 6.27 4.51 8.34 5.85 1992 7.66 6.52 8.27 8.18 1993 15.64 17.77 13.52 15.74 1994 16.57 17.96 13.46 18.48 Entrepreneurs with no crisis 40.81 43.17 41.06 38.12 Total 100.00 100.00 100.00 100.00 Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. - D.9 - Table D2. Reasons for having an Economic Crisis in Three Rural Areas of Mexico Percentage of Rural Entrepreneurs (Population Estimates) Ethnicity Gender Economic Sector Wealth Quartiles AU Non- Farm and Lowest two Third Fourth Indigenous Indigenous Male Female Farm Non-farm non-farm Quartiles Quartile Quartile Percentage of entrepreneurs within each group that had 59 61 55 62 52 63 56 60 51 66 66 crisis Low levels of income Low production 23.10 19.41 36.67 28.62 5.00 47.50 5.16 21.6 11.00 26.56 27.33 Low prices 6.00 5.50 8.03 7.53 1.06 11.20 2.50 3.78 3.68 3.71 11.27 Low sales (quantity) 20.80 24.36 9.20 17.20 32.40 1.12 34.16 32.70 20.15 20.54 22.20 Weak demand for services 11.90 13.81 5.69 12.00 11.60 3.41 18.28 13.11 18.20 8.46 6.92 Sickness of entrepreneur 3.90 3.9 4.00 4.27 2.78 5.50 3.03 1.70 3.00 7.16 2.19 Other causes of low income 9.20 9.61 8.45 9.70 7.82 10.65 8.92 5.43 11.60 6.40 8.96 Unexpected expenditures Sickness of entrepreneur or family member 17.87 17.33 20.23 15.80 24.44 18.25 17.85 16.55 19.60 21.60 14.60 Reparation or reposition of indispensable assets 1.00 0.95 0.51 0.80 0.84 0.29 1.41 0.00 0.50 1.60 0.63 Other abnormal expenses 6.11 6.12 6.30 3.30 15.11 1.60 9.30 4.95 74.50 4.45 5.95 Entrepreneurs with economic crisis 100 100 100 100 100 100 100 100 100 100 100 a. Wealth is proxied by the value of real state holdings and the average amount of deposits in financial intermediaries in the two years previous to the survey. Source: Encuesta Regional de Servicios Financieros a Unidades de Produccidn Rural. - D.10 - Table D3. Mechanisms to Cope with an Economic Crisis used by Entrepreneurs in Selected Rural Areas of Mexico Percentage of Rural Entrepreneurs (Population Estimates) Ethnicity Gender Economic Sector Wealth Quartiles All Non- Farm and Lowest two Third Fourth All_ _ Indigenous Indigenous Male Female Farm Non-farm Non-farm Quartiles Quartile Quartile Percentage of entrepreneurs within each group that had 59 61 55 62 52 63 56 60 51 66 66 crisis STRATEGY Financial Markets: 13 14.41 8.79 15.17 6.85 12.91 12.52 20.20 10.19 14.65 16.08 Loans with positive 7 8.21 3.41 8.90 1.53 7.30 6.96 8.51 3.94 7.63 11.15 interest rates Loans from friends 4.88 5.00 4.29 5.17 3.82 4.61 4.53 8.66 5.41 6.31 2.93 and relatives Delayed payments 1.20 1.20 1.09 1.10 1.50 1.00 1.03 3.03 0.84 0.71 2.00 of loans Sales of assets 8.22 9.67 2.92 8.54 7.05 10.33 7.26 4.28 6.27 4.03 14.62 Received donations: 13.44 14.42 9.57 11.68 18.93 10.76 16.48 7.37 14.00 15.36 11.21 From friends and 12.41 13.62 7.85 10.77 17.65 9.80 15.51 5.76 12.78 13.50 11.21 relatives From GOM and/or 1.03 0.80 1.72 0.91 1.28 0.96 0.97 1.61 1.23 1.86 0.00 NGOs Increased labor market 34.43 30.80 46.05 32.43 40.52 38.25 30.60 41.88 40.00 37.00 25.50 participation by entrepreneur and/or family member Reduction of consumption 24.70 24.47 24.80 25.75 21.00 22.41 27.51 19.00 22.52 23.70 28.83 Entrepreneurs with 100 100 100 100 100 economic crisis I Source: Encuesta Regional de Servicios Financieros a Unidades de Producci6n Rural. - D.11 - Table D4. Estimated Coefficients of Binomial Probit Model Estimated Marginal Effect at Variable Coefficient t-ratio Means * Dependent Variable: 1= experienced crisis, 0= otherwise Intercept -0.780 -2.23- -0.301 Dummy: Agricultural venture 0.177 0.94 0.068 (1= farm-RE, 0=otherwise) Dummy: Commercial venture 0.062 0.30 0.024 (1= commerce-RE, 0= otherwise) Dummy: Service venture 0.013 0.06 0.005 (1= services-RE, 0=otherwise) Dummy: Other type of venture 0.207 1.18 0.098 (1= other, 0= otherwise) Age of RE 0.003 1.02 0.001 Education level 0.021 0.82 0.008 Log of wealth (real state and bank deposits) 0.032 3.19** 0.012 Dummy: Ethnicity of RE -0.256 -1.93" -0.099 (1= indigenous, 0 = non-indigenous) Dummy: Gender of RE -0.007 -0.06 -0.002 (1 =male, 0=female) Dummy: Credit history of RE 0.322 2.94*' 0.124 (1= credit history in last six years, 0= otherwise) Number of dependents in household 0.042 2.07* 0.016 Percentage of population working in primary 1.735 1.60' 0.670 sector in RE's locality Population in RE's locality of residence 0.264 2.32" 0.010 Dummy: Guanajuato -0.259 -1.96* -0.100 (1= Guanajuato, 0=otherwise) Dummy: Puebla -0.370 -0.33 -0.143 (1= Puebla, 0= otherwise) LOG-L -515.81 Likelihood ratio (Chi-square with 15 D.F.) -47.31 Percentage of correct predictions 64 Number of observations 799 */**/*** significant at the 90 percent, 95 percent, 99 percent confidence intervals -respectively. a. Refers to the change in the probability of a crisis resulting from a change in the independent variable -for the average rural entrepreneur (i.e., partial derivative evaluated at the mean value of the vector of explanatory variables) ANNEX E CASE STUDIES OF NONBANK RURAL LENDERS - E.2 - 1. This Annex presents the main findings of 96 interviews with nonbank lenders (i.e., individuals and corporations) that provide credit services in selected rural regions of Mexico: namely, the Bajio region of Guanajuato, the northern mountain range of Puebla, the southern part of Puebla, the northern part of Veracruz, and the southern part of Tamaulipas. The interviews focused on the lending technologies, mechanisms to monitor borrowers, and the instruments for credit contract enforcement used by the interviewed nonbank lenders.' Lenders were selected by ad hoc methods-basically references from local residents and government officials. Although interesting per se, the findings are therefore not statistically significant. Background and Definitions 2. Technologies of Credit Allocation. Anyone involved in providing credit services-be it an individual or an institution-must employ a systematic method to do so. Such method may be regarded as a technology of credit allocation and on the operational side must include physical infrastructure, management information systems, and financial management techniques, while on the lending side it should include screening, monitoring, and enforcement mechanisms. 3. The particular technologies chosen by different lenders determine their cost structure (fixed and variable) of producing credit services. The business or operational component of the technology (e.g., management information systems, internal control) is closely related to fixed costs and its relevance increases with the number and complexity of transactions and with the existence of delegation of lending decisions on employees (agency problem). The argument could be made that the fixed cost structures resulting from traditional banking operational technologies are to a large degree to blame for the absence of formal intermediaries in rural localities. Those markets are too small to sustain large fixed cost structures, given reasonable intermediation margins. 4. The remaining information and enforcement components of a lending technology-very related to variable costs of lending-have to do with the screening of credit applicants, monitoring of borrowers, and the enforcement of credit contracts. Screening refers to those actions by which the lender gathers information regarding the probability that the borrower will repay the loan as originally promised. When a borrower repays a loan is because she is Annexes A and B present the market shares and detailed credit conditions at which the different lenders studied allocate credit to rural entrepreneurs in three selected regions. - E.3 - both-able-and-willing-to repay the loan. Screening has to do, therefore, with the ex-ante estimation of the probability that the borrower will be able to repay (i.e., generate enough cash income) together with the appraisal of the cost she incurs by defaulting on the loan (e.g., value of enforceable collateral) 5. Monitoring relates to the resources spent by lenders in ex-post loan supervision of those borrower's actions that may affect the previously estimated probability of repayment. Monitoring makes sense only if the lender is able to modify the behavior of the borrower when she finds-out that the borrower has engaged in activities that worsen the probability of repayment. 6. Enforcement has to do with the set of actions undertaken by a lender to increase the borrower's cost of default for any realized level of income. Regardless of realized borrower's cash flows and, hence, regardless of borrower's ability to repay, enforcement actions are directed to increase the cost of default for the borrower (e.g., foreclose on collateral). 7. There are two alternative technologies of information that can be used to provide financial services to rural entrepreneurs. The first one sustains screening, monitoring, and enforcement of credit contracts on standard data such as audited financial statements, real estate appraisal reports, and on legal instruments such as contractually established mortgages assets. This sort of information has a generalized value in the financial market because, once generated, the costs of using it are invariant to the identity and location of lenders as-for example-bank credit officers should be able to read audited financial statements. 8. The alternative technology builds the screening, monitoring, and enforcement of credit contracts on idiosyncratic information which refers to all repayment relevant information whose access and use imply differential costs to alternative lenders-depending on the lenders identity and location of operations. Some lenders (e.g., local moneylenders, friends) may generate idiosyncratic information at a very low marginal cost by, for example, taking advantage of the ordinary daily interactions among residents of the same locality and/or agents interacting in related markets. 9. The choice of a predominant type of information within a lender's technology influences: (a) the transaction costs of lending and borrowing; and (b) the competitiveness, structure, and performance of rural financial markets. This is due-simply-to the fact that,some lenders may have a comparative advantage (i.e., lower cost) on gathering information about borrowers while some borrowers may have a comparative advantage in signaling credit-worthiness to certain lenders. Such transaction costs differentials introduce segmentation in credit markets through their influence - E.4 - in the matching of certain types of borrowers with certain types of lenders. For example, large businesses normally maintain accounting records of their economic activities because they aid management and because, in most cases, ownership and control of such businesses are separated. Hence, accounting records and external audits are necessary instruments of internal control that would be used regardless of credit transactions. Once financial statements are available, the marginal cost for the firm to use them to prove credit worthiness is zero. Therefore, it is likely that firms which prepare audited financial statements would tend to request credit from financial intermediaries whose lending technology is intensive in such information-as confirmed empirically by the financial services survey to rural entrepreneurs. 10. The prevalence of either type of information will also affect the degree of competitiveness within the alternative segments of rural credit markets. This is because opportunity cost differentials between general-value and idiosyncratic forms of credit relevant information. By definition general- value information may be interpreted, at comparable costs, by other potential lenders. Hence, the general-value mechanisms by which potential borrowers signal credit-worthiness (e.g., financial statements, investment profiles) may be used to persuade several alternative lenders. In contrast, idiosyncratic information may not be interpreted, at comparable cost, by other potential lenders. This implies that borrower expenses incurred to signal credit worthiness with idiosyncratic information may be lender-specific in the sense that they have no opportunity cost outside the current credit relation-making the investments embodied in borrowers's idiosyncratic information signals sunk costs. This may allow lenders to behave as location monopolists within those market segments in which idiosyncratic information is pervasive and arbitrage on their clientele costs to signal credit worthiness to other lenders. Pure Moneylenders 11. For the purpose of these case studies pure moneylenders were defined as individuals who provide cash loans (i.e., credit disbursed in cash and repaid in cash) with a profit motive (i.e., charge interest on loans). The findings reported below are based on interviews conducted with 14 moneylenders.' 2 The issue of the competitiveness of the credit markets in the regions studied is analyzed in detailed in Chapter 6. 3 In the areas covered by the financial service survey to rural entrepreneurs, moneylenders provided 10 percent of all credit transactions--including cash loans, commercial credit, and forward sales--and accounted for 8 percent of the total volume of loans received by rural entrepreneurs in the period between June 1992 and June 1994. - E.5 - 12. All but one of the moneylenders interviewed were devoted to other economic activities besides money lending. The group included farmers, lawyers, politicians, traders, and merchants. This diversification of activities is due to the fact that most lenders have a limited number of borrowers whose attention-reportedly-would not required their full time. Reported numbers of borrowers vary from 25 to 100 with the exception of a completely specialized lender with 1,000 borrowers. 13. Fixed costs are negligible for most moneylenders as they-themselves-screen borrowers, disburse loans, keep records, and collect payments without paid personnel to assist them or special offices used just for their lending operations. In the very few cases in which employees were involved they were required for secretarial purposes 14. Reduced fixed costs--including the fact that lenders are not exclusively devoted to lending-is consistent with the comparatively reduced volume of their operations (i.e., few borrowers, small loans) which-even with high interest rates-could not support a large fixed cost structure. The question becomes "why moneylenders do not expand their operation by either becoming full time moneylenders and/or hiring paid personnel to assist them in their activities?."' Their answers related to the lack of "already known" credit-worthy individuals at any given point in time. This-in turn-has to do with the screening and enforcement components of the lending technology used by moneylenders-as follows. 15. Two general types of moneylenders were observed from a point of view of screening and enforcement technologies: (a) moneylenders who rely purely on the collateral offered by the lender; and (b) moneylenders whose screening and enforcement technologies are mostly based on the applicant's personal attributes and on local reputation effects. As shown below this is an additional cause and symptom of fragmentation and lack of competitiveness in the RFMs of the areas studied. 16. Moneylenders that rely exclusively on collateral use-as their single indicator of probability of repayment-the market value of the assets-mostly real state-pledged to support the promise of repayment on the part of the borrower. The value of collateral they reported is required for them to grant credit varies from 2 to 10 times the amount of the loan. These ratios are lower than those reported by REs surveyed-an average of 26 times the amount of the loans. By both accounts-however-the ratios of collateral to loans are too high. Most moneylenders answered that lack of capital was not a constraint to growth. In fact, most of them had access to loans form the formal sector and/or deposits in financial intermediaries earning much lower rates than the ones charged to their borrowers. - E.6 - 17. These high large ratios of collateral to loans result-in part-from inadequacies of the legal infrastructure as moneylenders justified them by the lengthy process to repossess real state-estimates varied from 2.5 years to 6 years. The argument is that over-collateralization is necessary to cover forgone interests in case judiciary actions became necessary. 18. Such large ratios of collateral to loan amounts may induce large numbers of credit-worthy REs to self-select out of credit markets. Peculiar collateral requirements together with the shortcomings of the legal framework could make the borrowers' cost of defaulting disproportionate to the amount borrowed. This is because in the rural areas studied there are no functioning mechanisms to auction property given as collateral and to insure that any amount remaining-after lenders have been fully compensated-will be returned to the borrower. In fact, some moneylenders consider default-rather than repayment-the favorable outcome of a loan as it implies obtaining assets at a very low price. Hence, those borrowers who would obtain loans by pledging collateral face an abnormally skewed risk-return distribution for their investments which may induce them to self select out of credit markets. 19. In view of the inadequacies of the legal system and of lack of competition some "collateral moneylenders" have adopted techniques to enhance the enforcement of credit contracts which have further negative consequences on the efficiency of RFMs by limiting entry of other potential moneylenders and by further increasing the number of REs who self-select out of financial markets. In particular, these lenders require-besides high collateral values-other repayment assurances that expose borrowers to their opportunistic behavior-for example: (a) some lenders do not require a mortgage but actual transfer of title on the real state pledged as collateral which will be returned if and when loans are canceled; (b) most moneylenders require their borrowers to sign blank and/or partially blank legal documents at the time of disbursement of loans.' Under such circumstances the credibility problem implicit in every credit contract shifts from the borrower-who has to repay the loan-to the lender who now that has to return the pledged assets when the loan is repaid and/or not to use a signed blank document to her undue advantage. 20. This implies that a reputation of not taking advantage of borrowers would be a condition of entry to certain segments of RFMs as a lender. This Partially blank documents are one of the ways in which moneylenders completely appropriate assets pledged as collateral without returning the remanent value after canceling all debt to the borrower by--for example--leaving the interests rate portion of documents blank. In case of default, an interest rate such that would bring the total debt to an amount greater or equal to the value of collateral is filled-in the document prior to taking the loan to court. - E.7 - reduces the pool of potential entrants and-hence-the contestability of the industry providing incumbent lenders with rents on the their reputation. Thus, it is clear that the real cost of the inadequacies of the legal system and its implied monopoly power may be that large numbers of REs may not borrow to finance profitable investments because of excessive collateral requirements and or fear that lenders will take advantage of them. The survey results indicating that 23 percent of all REs did not request a loan in the last two years because they consider borrowing to be too risky is consistent with this possibility. 21. Expectedly, collateral moneylenders reported that they did not care about the actual use given to loans, neither would they monitor the borrowers' activities after loans were granted. On the other hand, the rates of interest charged by these lenders are rather uniform across their borrowers. Uniformity in rates of interest makes sense as there are no risk differentials across lenders (i.e., completely collateralized loans) and information flows between lender and borrower are limited to the market value of collateral. This implies that collateral lenders are not able to price discriminate among borrowers as they do not have a basis to determine differences in willingness to pay (i.e., elasticity) across observationally equivalent applicants. 22. The other kind of pure moneylenders interviewed are those who base their lending decisions on the personal attributes of borrowers-"character- based lenders". Such lenders grant loans either on the verbal promise of their borrowers alone or on their fiduciary responsibility (e.g., pagare)-few times with cosigners. Such credit contracts are based almost on pure trust and-therefore-require of a very close-almost personalistic-relationships between borrowers and lenders. Screening of borrowers by these moneylenders is based entirely on idiosyncratic information on the borrowers character. This includes repayment relevant variables such as work habits, patterns of household expenditures, and the applicants involvement in economic transactions that require of a reputation of honesty in the market. Information on these characteristics of applicants is normally a by-product of location and proximity between applicants and lenders or results from a reference provided by-normally-another borrower trusted by the lender. Hence, character based lenders have a rather low marginal cost of estimating the probability that an applicant belonging to a limited pool of potential borrowers will be able to pay the loan. Hence, the applicant's willingness-given her ability-to repay the loan becomes the lender's main concern (i.e., pure repayment problem). 23. Moneylenders were observed to invest resources in assuring that borrowers take those actions that maximize the likelihood of repayment only in the absence of enforceable collateral. Actions reported by lenders included monitoring the borrower's actions to verify that the investment plan financed - E.8 - be actually carried-out and/or to avoid loan proceeds to be diverted to other uses. 24. A very interesting example is given by one of the largest pure moneylenders interviewed who finances agriculture and who invests resources in the incentive/monitoring scheme described below. The moneylender provides loans for up to 70 percent of the total investment-excluding all fixed assets-to produce wheat and other grains. The lender has established a plan of sequential disbursements. The first disbursement does not take place until applicants have actually invested their 30 percent contribution (e.g., soil preparation). The lender requires this equifty contribution to avoid morally hazardous behavior on the part of the borrower by introducing a deductible or co-payment of 30 percent in the venture. Such deductible aligns the incentives of the borrower to those of the lender in that the borrower-by protecting his own inyestment-will also maximize the probability of being able to repay the loan-which is a necessary condition to recover a un- collateralized loan. 25. In order for the described incentive system to function, the lender has to verify (i.e., monitor) that two events have taken place-namely: (a) the borrower has actually invested his contribution (30 percent); and (b) the borrower financed such investments with his own resources and not with additional debt from other sources. The lender-a lawyer and elected official (diputado)-uses paid employees to monitor that applicants have carried-out their equity investments. Such delegation of monitoring activities creates-in turn-an agency problem with two dimensions: (a) insure that his employees actually monitor investments; and (b) avoid collusion between supervisors and loan applicants. The lender has established a second system of incentives-this time-for his employees to solve this other agency problem. The system is based on random monitoring of employees, paying employees more than their opportunity cost (i.e., efficiency wages), and a credible threat of job termination. The lender audits-at random-the written supervision reports of his employees. In case of disparity between his own field observation and the corresponding supervision report, the employee is fired. The cost of negligence and/or collusion with the borrower for the employee is the net present value of the difference between the efficiency wages he collects working for the lender and the salary he would earned elsewhere. This-according to the lender-has proven successful. An important dimension of the scheme is that payments to supervisors are two tiered: (a) a piecewise (i.e., per visit) component; and (b) a profit sharing payment. By compensating employees in such way salaries become a variable-as opposed to a fixed cost. 26. Similar agency problems are necessarily faced by any lender-formal or informal-who delegates any component of the lending process to an employee or agent. The ability to solve agency problems is a necessary -E.9 - condition for success in any economic venture-particularly in rural credit markets where there is an abundance of possibilities for opportunistic behavior. The main lesson for both systems of incentives implemented by this lender is that rewards and punishments to agents (e.g., borrowers, employees) must align the incentives of the lender and that of agents in that agents-while pursuing their own interest-actually behave as the lender would prefer them to behave. 27. Once the lender has verified that the applicant has carried out the tasks corresponding to his equity contribution in the venture, he also attempts to make sure that such investments were financed with own resources and not with additional debt from other sources. One of the strategies used is to collect the legal titles of the borrower's real state. Such documents do not represent collateral because they are not legally enforceable but lack of possession makes it more difficult for the borrower to grant montages and-hence obtain credit from other lenders-mostly collateral lenders. The most commonly used strategy-though-is to share information with other local character-based moneylenders who have incentives not to cheat on the network of lenders as they will benefit from repeated interactions in the future. 28. In general, character based moneylenders solve the pure enforcement problem (willingness to pay) by using reputational effects and business-cum- social networks in which individual credit histories are assembled. If a borrower defaults on a lender and such lender is trust-worthy to other lenders the other lenders may deny future credit. Conversely, an individual's good "market reputation" would encourage not only lenders but other members of the network to do business with her. Other network members furnish reliable information about her character, her business, and how much credit can be profitably extended to her. Lenders can trust members of the group more than other potential borrowers who are not because they can use the network as a sanctioning mechanism should she cheat or default the lender. This is why all character based lenders interviewed require someone they trusted to vow that the applicant-besides a good farmer-had not defaulted on another borrower in the past. This systems requires reliable systems of communication among lenders that exist only in among very small groups of moneylenders who live close to each other. 29. This forms of collusion or networking was detected among character- based lenders only. These lenders do not seem to share information (e.g., applicants' credit references) with collateral-based lenders who neither seem to share information amongst themselves. The reason is that the "cartel" among character-based lenders is self-enforced in that they benefit-due to repeated future beneficial interactions in a very small locality-from the functioning of such information network-which requires not to cheat on the remaining members. Collateral based lenders do not benefit from the - E.10 - network-as their loans are more than covered by real assets-while actually may benefit from lying to other lenders by issuing a negative recommendation on a borrower in order provide the loan themselves. 30. The number of individuals who belong to the group or network over which any lender has a cost comparative advantage is limited-however. This implies that the number of loans that a lender can efficiently evaluate is also limited as there are severe diseconomies of scale in assessing personal behavior and in collecting non-standard information about individuals who do not belong to the network. This limits the potential growth of a lender's business or forces him to rely on the information possessed by other members of the group. Precisely for the same reasons the scope-geographical and otherwise-of character based lenders is limited further aiding in making RFMs fragmented. 31. One concern manifested by all rural lenders-not only moneylenders-is that the adjustments being experienced by the rural economy (e.g., removal of price supports, trade liberalization) at the time of the interviews were also disrupting the traditional ties and networks that supported some-albeit segmented-credit transactions. Just for example, increased participation in labor markets and immigration (see section ?) may make rural societies more anonymous by making the rural population more mobile. This will certainly weaken the role of social networks and their sanctions in enforcing credit contracts. A recurrent point made by moneylenders was that "people are not as trust-worthy as they used to be". The concern is that RFMs will become even more shallow and segmented as the established networks disappear way before new institutional alternatives developed.' The break-down of local credit networks seems to have been worst for product or market specific credit transactions (traders-lenders). The lending technologies of such lenders are analyzed next. Trader-Lenders and Agribusinesses 32. This part of the analysis examines interlinked credit transactions in which rural entrepreneurs receive credit services from agents who are interested in acquiring their production. Attention is given to different credit services and to the technologies used to produce them by individual trader- lenders and corporate agribusinesses. The section summarizes the main findings of 27 interviews-10 individual trader-lenders and 17 corporations-all of whom provide some form of credit service (e.g., inputs 6 Prior to the distress suffered by the banking system that originated with the devaluation of the Nuevo Peso in late 1994, two thirds of the country's municipalities did not have bank offices. If anything, the coverage of the formal sector will only decrease in the medium run due to the general conditions of the industry. - E.11 - on credit, cash advances) to rural entrepreneurs-mostly farmers. For ease of exposition the terms trader-lender, agribusiness, and lender will be used in this section interchangeably-unless otherwise indicated. 33. The lenders interviewed provide REs with three distinct types of credit/insurance services which are interlinked with input and output transactions. First, lenders provide REs with credit-both in kind and in cash-which the borrowers agree to repay with future delivery of production. The actual unit.price paid on the borrowers production is determined at the time of delivery of the output and it is usually calculated as the on-going spot price minus a percentage agreed upon at disbursement. Thus, the borrower's cost of using this kind of credit is made-up by a discount in price plus any explicit interest rate charged by the lender. These transactions are considered to be "pure" credit because borrowers conserve all price and production risk which implies that the value of their debt is fixed. It is possible-therefore-that delivery of total production to the lender be insufficient-either because of low yields and/or low prices-to cancel the loan. Pure credit transactions were observed to be interlinked with the markets for sugar cane, cereals and grains, horticultural products, coffee, citrus, sea food, vanilla, poultry, and rice. 34. The second type of financial products observed also involve disbursements to REs in exchange of future delivery of production. In these contracts-however-the agreed unit price on the borrower's production is completely specified at the time of disbursement and it is contractually independent of the on-going spot price at the time of delivery. Within this general group of contracts there is significant variation as transactions range from arrangements that are conceptually equivalent to a rental contract to price hedging contracts. The most common example of the rental-type of transactions occurs when traders pay for the production of a given plot of land prior-normally between one and six months-to harvest. In such cases the trader becomes the residual claimant of the production of that plot assuming-therefore-all price, yield, and fraud risks. 35. The price-hedging form of these transactions consists of a contract in which the producer agrees to future delivery of production to the lender at a fixed contracted price. The producer-however-retains the production risks. Observed payments-rather than being credit-are a mechanism to insure that the buyer will purchase the crop-even if the spot price at the time of delivery is below the previously agreed upon price. In the absence of the advanced payment (i.e., posted bond) the buyer may behave opportunistically by purchasing the crop only if the spot price at the time of delivery is greater than the contracted price. Hence, the main objective of the transaction-while allowing the producer to use resources not yet earned-may be to minimize price risk rather than funding purposes. - E.12 - 36. Regardless of the actual nature of each of the services provided-credit, rental, reduction of price risk-all contracts observed entail a promise on the part of REs who may benefit by repudiating their pledge to deliver products. In order to produce such contracts lenders have to-therefore: (a) sort-out those entrepreneurs more likely to comply from those more likely to default; (b) provide incentives and control mechanisms which make it more likely that producers undertake safer actions-thus-increasing the likelihood of repayment; and (c) enforce contracts in an environment in which the fragility of the legal system and the lack of assets to be pledged as collateral give ample allowance for opportunism (e.g., fraud, theft) on the part of producers. The systematic methods used by traders to achieve these tasks are-for all practical purposes -equivalent to a lending technology as defined above. 37. The large majority of trader-lenders do not rely on real assets pledged as collateral neither on information of generalized value to screen borrowers but rather make intensive use of idiosyncratic information in the form of: (a) trust acquired because of repeated past transactions; (b) requiring applicants to obtain a recommendation of credit-worthiness from a current borrower who would be screened-out of future contracts should the applicant cheat; and sometimes (c) requiring a reference from a local official (e.g., president of the municipality) where the borrower reside. These screening mechanisms allow only for slow growth in a lender's number of clients. In fact, the comparatively large numbers of borrowers reported by lenders are the result of many years in the activity. Trader-lenders and moneylenders alike associate rapid growth in their number of clients with more than proportional default problems. 38. The enforcement of interlinked contracts in the areas studied is a complex and very heterogenous endeavor which ranges from normal legal proceedings to repossess collateral to ad-hoc and mostly non-institutional instruments. Lenders combine the alternative instruments available to them according to the amounts of credit involved and the costs of using each alternative instrument. Nonetheless, the majority of trader lenders rely-at least in part-on the credible threat of excluding fraudulent borrowers from their own portfolios and-through information pooling networks among lenders-from the market at large. The threat of exclusion seems to provide enough incentives to honor their contracts to large numbers of borrowers. The ability to enforce contracts-however-is such a significant requirement in the effective provision of interlinked credit that it determines-to a significant extend-the actual products for which interlinked contracts are available (e.g., oranges but not peppers, barley but not corn). 39. Enforcement possibilities seem to determine the actual crops for which interlinked contracts may or may not be available in two different ways. The first is related to the purely technological-agronomical characteristics of the - E.13 - goods produced which make some crops more "enforcement-friendly" than others. Enforcement friendly crops are more likely to be financed by an interlinked credit contract. Lenders who trade in a variety of perishables and who offer interlinked credit contracts only to a subset of such crops responded that they provide credit to particular crops and not to others because of differences in the costs of enforcement. 40. For example, all three types of interlinked contracts were observed in the production of citrus while no contracts were offered to producers of a large variety of other fruits and vegetables (e.g., hot peppers, papayas, watermelons, mangoes). According to lenders, citrus offer reduced opportunities for moral hazard on the part of the borrower and increased possibilities for pure enforcement. Opportunities for morally hazardous behavior on the part of the borrower are reduced because once citrus orchards have blossomed and small fruits have loomed, yields will not be affected by the borrower's negligence-borrowers who received loans after a satisfactory blossoming will be-most likely-able to repay their loans. Such is not the case with other fruits and vegetables whose successful production and-implied loan repayment ability-depends on the borrower's diligence-until the last moment. On the other hand, there are differences even within citrus. Oranges have a comparatively short and fixed harvesting period which aids in enforcement as lenders know with precision the time at which harvesting will take place. Hence all three interlinked contracts are feasible due to an enforcement element. This contrasts with lime orchards where only "rental-like" contracts were observed because such orchards are harvested year round. 41. A second way in which easiness of enforcement influences the availability of interlinked credit contracts for certain products is through market structure. Products whose markets are less competitive are more likely to be the subject of interlinked transactions than those products whose marketing channels are more competitive. For example, large grain traders reported that they are more likely to engage in interlinked credit contracts with producers of barley than with producers of other grains and cereals because a monopsonic structure in the market for brewing malts makes enforcement of contracts very easy. The country's beer brewing companies have established a cartel (Impulsora Comercial S.A. de C. V.) for the collective purchase of their inputs. The cartel allocates barley production quotas to individual producers. Barley to be used in the production of brewing malts-on the other hand-commands a price at least twice that of barley to be used as forage. Hence, trader-lenders-who also have trading quotas-lend to those producers who obtained production quotas with no apprehension because defaulting on a loan implies having to sell their production as forage at half price. - E.14 - 42. Sugar cane mills have similar enforcement advantages as transportation costs of bulk sugar cane are very high and mills are located sparsely through-out rural areas. This makes the enforcement of interlinked credit contracts comparatively easier as-fraudulent borrowers would have to pay high transportation costs to sell their production to other mills. In any event, the reduced number of regional mills collude in order to enforce contracts. 43. The trader-lenders and agribusiness interviewed have significantly larger numbers of borrowers-spread over much larger geographic areas-than the moneylenders interviewed. The large majority of trader-lenders reported having hundreds of borrowers-up to 1,500 in one case. The geographic coverage of trader-lenders also seems more extended than that of moneylenders. One trader-lender located in the northern part of Puebla reported having borrowers scattered over the states of Sinaloa, Veracruz, Tabasco, and Oaxaca. 44. This significant outreach and geographical coverage requires extensive use of paid personnel to perform activities directly related to screening, allocation, and enforcement of loans and which require discretionary decisions on the part of employees. Trader-lenders -in contrast to most pure moneylenders-have been required to solve the agency problems that result from delegation of decisions and tasks on employees who in general are not the residual claimants of profits and losses-so they may shirk-and who could benefit from colluding with borrowers. The agency problems of large trader-lenders are particularly complicated because-for instance-credit decisions and monitoring activities cannot be replicated because they are based on idiosyncratic information. Hence verification of the appropriateness of decisions is impossible. In contrast, the lending technology used by traditional banks allows replication of credit decisions because they are based on information of general value. This allows banks for simpler disciplining of employees as their negligence and collusion can be determined somewhat more easily than in the case of trader-lenders. 45. Trader-lenders reported that they pay key employees a total remuneration that results from the combination of a fixed salary plus bonuses or incentive payments based on profits and/or recovery of credit. They also reported that the total amount paid to such employees was above what equivalent workers would demand to perform the same tasks (i.e., efficiency wages). The systems of incentives used by trader-lenders are similar to the ones used by the few moneylenders who also delegate on employees in that both imply a high cost for employees of being fired. For practical purposes trader-lenders make key employees their partners because their high compensations-relative to their opportunity costs-make them behave as if they had invested their own resources in the venture (i.e., equity contribution). Some trader-lenders reported that the fixed portion of the - E.15 - salaries paid to key employees is more than twice what those individual could make elsewhere. Clearly, such system of incentives will only work if the threat of employment termination is credible. 46. The diversity of interlinked contracts constitutes numerous responses to deficiencies of the institutional and property rights structures present in the rural areas studied (e.g., attenuated property rights on ejido land). Although such responses have allow REs who must likely would not had access to credit-otherwise; they have not deepen the market significantly as evidenced by survey results. 47. Large variegation of interlinked contracts and the comparatively large numbers of clients served by individual trader-lenders has not meant that in the aggregate large numbers of REs have access to such contracts. In fact, interlinkages between credit and other markets are scarce in the areas covered by the Survey on Financial Services to Rural Entrepreneurs. Only in 12 percent of the cash loans in which REs were involved the borrower had a connection with the lender in another market and in more than half of such cases the additional connection was in labor markets while only 2.5 percent of farm-REs had participated in interlinked credit/output contracts. The massive intervention of the government in the rural economy (e.g., CONASUPO, BANRURAL) has been a likely contributing factor in the scarcity of interlinked contracts and in the shallowness of RFMs at large because such interventions weaken the development of private markets. 48. Regardless of the relative importance of interlinked transactions it is more than likely that the process of adjustment resulting from the structural changes experienced by the rural economy (e.g., NAFTA, price liberalization)-and more recently the exchange rate crisis-will disrupt traditional ties in the rural economy. Just for example, changes in agricultural products will decrease the value of the relationships between agents that currently operate in parallel markets. These already established market interlinkages may disappear way before new interlinkages are developed. Strategies to Link the Formal and Informal Sectors 49. The informal sector has adapted to the institutional under-development and property rights deficiencies that characterize Mexican rural areas and has acquired a comparative advantage in borrower screening based on local information, monitoring, and the non-institutionalized enforcement of credit contracts. The formal sector-on the other hand-has a comparative advantage in intermediating funds over space-which is a specially important function in rural areas because of strongly correlated income flows and risks. A combination of these two comparative advantages becomes an intuitively appealing strategy to deepen and integrate RFMs. For instance, FIRA has - E.16 - established a program aimed at using a variety of enterprises (e.g., processors, input suppliers) as conduits for formal credit to the final users (Programa de Agro-Asociaciones). 50. There are two separate strategies that could be used to combine both sectors' comparative advantages: (a) recruiting informal sector agents to act as conduits of formal sector resources; or (b) induce the formal sector to behave-in certain dimensions-like the informal sector. It is - argued-however-that strategies based on recruiting informal agents are inferior to those alternative approaches to strengthen RFMs based on inducing the formal sector to behave more like the informal sector by adopting some basic elements of its lending technology-even if it is more costly to do so. 51. Any program aimed at improving the functioning of rural rmi cial markets should entail increased access to deposit services. Emphasis on recruiting informal agents to allocate formal sector loans is biased towards the credit dimension of financial markets. There are strong economies of joint production (i.e., economies of scope) between cash-loans and deposit services that could not be exploited if the strategy is based on recruiting informal agents. 52. One important shortcoming of promoting interlinked credit contracts is that it does not develop financial markets as such because-unlike the case of pure cash loans-interlinked credit contracts are contingent on the markets for specific products. Credit market relationships-so established-would be destroyed as producers change crops because-for example-it is likely that a former-say orange producer-will get a loan from a bank with which she conducted business in the past. Clearly such is not the case for those REs who received credit from trader-lenders. Financial market development is all about building trust and relationships among participants.' For example, the interlinked contracts observed in the production of barley may be destroyed in the-actually very likely-case that domestic production of such crop losses its competitiveness as the result of NAFTA. 53. In the Mexican current environment of structural change the best way to promote interlinkages is by increasing the general flow of capital to rural 7 Off-course the most direct mechanisms to build trust in financial markets are: (a) collateral by which borrowers pledge a valuable hostage providing credibility to their repayment promises; (b) improved institutional infrastructure which makes the collateral clause enforceable at low transaction costs; (c) information sharing networks between lenders--such as credit reference bureaus--which reduce the otherwise prohibitive cost of gathering information and improve enforcement by ostracizing defaulters from credit markets for ever. They would take long time an still will not necessarily improve access to credit-worthy REs who could be reached by other methods--however. - E.17 - areas-including the access of REs to unrestricted use credit-so that viable interlinkages may be developed endogenously. That is to say that if a firm desires to borrow in order to be able to provide commercial credit to its clients it should be able to do so without any special treatment. This contrasts with promoting one particular type of interlinked contracts or enterprises as conduits for formal sector loans. In fact, the nurturing of interlinked contracts may hinder the healthy development of financial markets at large. 54. There are doubts-anyhow-regarding whether REs would have access to increased amounts of credit and whether the cost of funds would be reduced. An increased supply of formal sector funds to informal lenders will not necessarily increase the availability of credit and/or reduce the cost of borrowing-particularly interest rates-for REs. Availability of funds may not increase because the source of success of interviewed moneylenders is precisely their informational advantage over a group of confined residents, the pool of individuals to whom they may safely lend is limited and has to grow very slowly. Therefore, there is no reason to believe that those who are currently providing credit services would increase the number of borrowers and/or the amounts lent as a result of their increased access to institutional credit. There is a strong possibility that lenders would replace-at least significant portions-of their resources with institutional credit to finance-on the margin-other activities. It should be remembered that most of these agents have other activities of their own. In any event, there is no evidence-anecdotal or otherwise-that wealthy lenders and agribusinesses are not currently able to access formal sector. 55. The extend to which increased access to formal sector funds by informal lenders is beneficial to rural entrepreneurs depends upon the market structure that characterizes the interaction of informal lenders. If informal lenders behaved monopolistic or are engaged in collusive behavior-as some of them do-any subsidies and/or increased amount of funds available to them will only increase their ability to capture rents accruing to such programs while not necessarily increasing the amounts lent to REs as they may replace the funds they have already committed by GOM funds. 56. Also, and in view of the rather high interest rates and other hidden charges imposed by traders (e.g., underpricing producers' outputs and overpricing supplier inputs) charged on commercial credit-for example-and the drain on REs net incomes makes it is necessary to examine the efficiency and equity implications of using the informal sector as conduit for institutional credit. Particularly if-as argued in Chapter 6-high interest rates do not necessarily correspond to risk differentials across individual borrowers. 57. The effects of using informal lenders as conduits for credit on the rates of interest charged on informal credit are not clear and depend on how - E.18 - the operational costs of moneylenders change and on the level of competition in the informal sector. Interest rates for certain borrowers may actually increase as a result of increased availability of credit to moneylenders if GOM programs increase competition for borrowers among lenders via induced entry. Induced entry resulting from GOM programs combined with a limited pool of confined potential borrowers (i.e., segmented market) may lead to excess capacity among lenders and higher average costs which-in turn-may require moneylenders to increase equilibrium rates. This requires-however-money lenders to have high fixed lending costs-which is the case only for some of them. 58. The recommended strategy for a more sustainable and neutral development of RFMs is to-off-course-correct underlying property rights problems-but even then-induce the formal sector intermediaries to behave more like informal agents-at least in certain aspect of their operation. The presence of the formal sector in rural areas and the access of the traditionally regarded as "nonbankable" small REs would be significantly improve by developing financial products, technologies, and incentives systems similar to those used by informal lenders-to be implemented in a conducive policy framework. An outline of such endeavor is presented in section ?. Chartered Nonbank Intermediaries in Mexico 59. Chartered nonbank intermediaries (CNFIs) are nonbank financial intermediaries which have been granted an operational license by the Secretaria de Hacienda y Cridito Pablico (SHCP) or by the Comisi6n Nacional Bancaria and those which have been allowed to operate without licenses but under the auspices of a particular piece of legislation. For the purposes of RFMs the most relevant CNFIs are Uniones de Cr6dito (UCs), Sociedades de Ahorro y Pristamo (SAPs), and Cajas Populares (CPs). The legal framework for UCs and SAPs is given by the Ley de General de Organizaciones y Actividades Auxiliares de Cridito and by a collection of specific rules issued by SHCP. The prudential supervision of UCs and SAPs is the responsibility of the CNB. The CPs-which are basically credit cooperatives-operate under the auspices of the Ley General de Sociedades Cooperativas and are not supervised by the government. There are questions regarding the legal status of the CPs because the cooperative law explicitly recognizes only producer and producer cooperatives-however. 60. Despite of the significant resources and preferential tax and regulatory treatment invested by GOM in the promotion of these private organizational forms, the number of these intermediaries is insignificant relative to Mexico's size and population. In fact, a large number of these organizations have resulted from the subsidies available to them rather than from business decisions of their owners. Furthermore, there are important risks imposed by - E.19 - these intermediaries to the government, depositors, and the overall development of financial markets-details as follows. Uniones de Cridito (UCs)' 61. Uniones de Cr6dito are borrower-owned organizations that engage in both financial intermediation and industrial/commercial activities. At the end of 1994, there were 366 UCs operating in the country with consolidated assets for N$19 billion out of which N$10 billion correspond to liabilities originated in portfolio rediscounts with GOM's development banks. 62. Large numbers of UCs were created as the result of increased availability of credit from GOM-mostly subsidized. The number of credit unions exploited from 40 in 1988 to 366 at the end of 1994. The reform of the country's development banking implied an increase in second-tier lending to the private sector. In particular, development banks lent directly and/or offered credit and guarantees to commercial banks so that they lent to UCs which-in turn-provided cash loans to their owners. In 1993 alone FIRA and NAFIN discounted more than US$2 billion to the sector. 63. There is no comprehensive data on the number of businesses or entrepreneurs financed through UCs. According to NAFIN discounted loans through UCs reached close to 20,000 enterprises in all sectors of the economy in 1993. FIRA-on the other hand-reports having discounted loans to 135 UCs in that same year. The coverage of UCs in the regions studied by the survey is rather limited as they allocated two per-cent of the number of cash loans received by REs between June 1992 and June 1994. 64. Available reports regarding the financial performance of UCs are very discouraging. For example, in the last semester of 1993 the SHCP required, NAFIN to classify its portfolios with UCs and make provisions against loan defaults due to the increasing financial problems and levels of delinquency in the sector. In particular, the CNB reports that-in 1994-42 percent of UCs had operational losses, seven percent became bankrupt, and 33 percent have arrears of more than 10 percent of their portfolios. 65. The lack of viability of UCs is due to the internal governance rules implicit in their legal framework which are inconsistent with their financial health. This problem has been aggravated by their relationship with GOM's development banks who have financed the borrower domination of UCs. More specifically the current legal framework allows UCs to: (a) lend to their owners; (b) have lower capital adequacy ratios than regular banks (i.e., The sources of the figures presented in this section are NAFIN and FIRA. Some of the data is summarized by Mansell (forthcoming). - E.20 - more leverage); (c) combine financial intermediation with commercial and industrial activities. These characteristics of UCs imply that for all practical purposes UCs have no capital. This results from the ability of these intermediaries to lend to its owners. In case of bankruptcy, owners can capture their equity investments by defaulting on their loans. Financial intermediaries have little incentive to avoid risks in the absence of capital. The problem is worsened by the large leverage ratios allowed to UCs through loans from GOM development banks which have implied weak incentives for their owners to exercise internal control and enforce credit contracts as they-as borrowers-may benefit from the organizations' losses. 66. Additionally, UCs are particularly difficult to regulate and monitor because they combine financial intermediation with other commercial and industrial ventures. It is difficult for bank examiners to monitor the risks involved in such diverse activities which makes their supervision particularly difficult and costly. Sociedades de Ahorro y Pristamo (SAPs) 67. Sociedades de Ahorro y Pr6stamo are-at least ostensibly-non-for- profit financial intermediaries that can collect deposits and provide loans only to its own members. In June 1994 there were 15 SAP licenses granted and 55 applications waiting to be approved. Some SAPs are rather large and complex organizations which have grown extremely fast. Within six months of having being licensed-in March of 1993-La Caja Mexicana de Fomento (CANAFO) had established more than a hundred branch offices in 14 states and had more than 70,000 depositors while La Caja Popular Mexicana-a recently approved SAP-consolidated 140 CPs into a single charter and began operations with 315,000 depositors. 68. There several problems imbedded in the legal framework for SAPs. In particular, SAPs are financial intermediaries which can be created and controlled by individuals who do not invest any capital in the organization. There are examples of very large SAPs (e.g., CANAFO) in which control (i.e., voting rights) has been acquired by groups with negligible equity investments. This is possible because depositors automatically sign over their voting proxies to management when they become members. Given that the law does not allow the trading of voting rights, it is almost impossible to oust an incumbent management team. The incentive structure resulting from standard SAP governance rules and property rights structure is, clearly, perverse as those individuals in control of these organizations are the residual claimants of their profits only and not of their losses. - E.21 - Regulatory Concerns Regarding Chartered Nonbank Intermediaries 69. The incentive problems implicit in the governance rules of UCS and SAPs are worsened because it would be difficult for GOM not to bail-out some of those organization in case of bankruptcy-which may contribute even more to morally hazardous behavior. On the one hand, such organizations have large numbers of depositors-more than 350,000 in one case-while on the other they have been chartered by the federal government (SHCP) and placed under CNB's supervisory responsibility. GOM may be assuming-therefore-important contingent liabilities with the chartering of such intermediaries 70. Another important regulatory concern is that-clearly-both UCs and SAPs represent opportunities to profitably arbitrage on regulation by choosing one such charter. 71. GOM should allow the creation of private small regional banks and eliminate UCs and SAPs by attrition. Regional banks would have lower initial minimum capital requirements which ought to be consistent with the size of the market they will serve. Regional banks would engage in simple operations (e.g., passbook accounts, certificates of deposit, working capital loans) and should-probably-maintain stricter equity ratios than nationally chartered banks. There are very favorable experiences with small regional banks in other countries (e.g., Indonesia and The Philippines). 72. The development of a system of regional banks with disparate minimum initial capital requirements should not cause regulatory apprehension. Regional banks would be subject to capital adequacy requirements in the form of minimum solvency or maximum leverage ratios. This represents sound regulation. In contrast, capital requirements as an absolute minimum amount of capital to enter the industry are, conceptually, anti-competitive regulation. The only justification for this latter type of regulation is the practical difficulty of supervising potentially large numbers of banks with small-scale operations. 73. In any event, the supervision of small regional banks would be-no doubt-an easier task than CNB's current responsibility of supervising close to 300 UCs engaged in very diverse activities and with perverse organizational incentives-as described above. CATALOGUE RS,/F I LE CONFIDENTIAL Report No: 14599 ME Type: