DO AGRICULTURAL PROGRAMS IN MEXICO ALLEVIATE POVERTY?
EVIDENCE FROM THE EJIDO SECTOR^*

LOUISE CORD^**
QUENTIN WODON^**

^* This research was completed as a background paper jointly for the Mexico Ejido Study and the Mexico Poverty Assessment at The World Bank. We are grateful to the staff of Secretaria de Reforma Agraria for providing us with the data. These are the views of the authors, and need not reflect those of the World Bank, its Executive Directors, or the countries they represent.
^** World Bank, Washington, D.C.
JEL classification: H5, I3

ABSTRACT

Following the economic and legal reforms of the early 1990s, the Government of Mexico has developed new agricultural support programs that promote productivity and facilitate the adjustment process, while not interfering with the market economy put into place by the reforms. In this paper, we use a panel survey for the the ejido sector for the years 1994 and 1997 to analyze the impact on poverty of two programs: PROCAMPO, a cash transfer per hectare of cultivated land, and ALIANZA, a matching grant program for investments in agriculture. Our evaluation method takes into account the endogeneity of placement. We find that participation in PROCAMPO significantly reduces the likelihood that a ejido household will be poor, but that ALIANZA has no significant impact on poverty. It is also shown that PROCAMPO may have a positive multiplier effect on household income.

RESUMEN

Después de las reformas económicas y legales de principios de los años 90, el gobierno de México ha desarrollado un nuevo apoyo agrícola a los programas que promueven la productividad y facilitan el proceso de ajuste, mientras no interfieran con la economía de mercado que pusieron en marcha las reformas. En este trabajo usamos una encuesta de panel para el sector del ejido para los años 1994 y 1997, para analizar el impacto en la pobreza de dos programas: PROCAMPO, una transferencia al contado por hectárea de tierra cultivada, y ALIANZA, un programa de fondos (matching) para inversiones en la agricultura. Nuestro método de evaluación toma en cuenta la endogeneidad de la colocación. Encontramos que la participación en PROCAMPO reduce significativamente la posibilidad de que una familia del ejido sea pobre, pero que ALIANZA no tiene un impacto significativo sobre la pobreza. Se muestra también que PROCAMPO puede tener un efecto multiplicador positivo en el ingreso familiar.

Keywords: Agriculture, Poverty, Multiplier.

 

1. INTRODUCTION

Up until the late 1980s, the Government played a dominant role in production and marketing decisions in Mexico's agricultural sector and particularly in the ejido sector. Under the ejido system, the Government granted land and water resources to a community of producers (known as the ejido). The community's members, or ejidatarios, had usufruct rights over the land they cultivated, but were not allowed to engage in sale, rental or sharecropping contracts. They were also prohibited from hiring wage labor and absences from the ejido of more than two years led to the loss of land rights. By the early 1990s, the ejido sector accounted for approximately half of Mexico's farmland and three quarter's of the nation's producers and provided a critical instrument for the Government to implement its production and marketing policies for the agricultural sector.

However, with the economic and institutional reforms that began in the late 1980s, the relationship between the ejidatarios (the households living in the ejidos) and the state underwent a dramatic change. Restrictions on the sale and rental of ejido land as well as on the hiring of labor were lifted and the state no longer told the ejidatarios what to grow and how to grow and market their output. Similarly, as part of the overall reform package, the Government also no longer provided widespread technical assistance, input and output subsidies, and government marketing channels. As a result, by 1993, while the ejido sector had more freedom to allocate its resources, it was in an institutional vacuum without much governmental support to facilitate the adjustment to a market economy with rapidly changing incentives (de Janvry et al., 1997; Cornelius and Myhre, 1998).

To fill this vacuum, the Government developed between 1992 and 1996 new agricultural programs that were designed to ease the adjustment process for producers, while not interfering with the emerging rural market economy. Two of the programs, PROCAMPO and ALIANZA, aim to increase investment and productivity in the agricultural sector without distorting production incentives. Using household data from recent panel surveys of the ejido sector conducted in 1994 and 1997, this paper analyzes the impact of PROCAMPO and ALIANZA, on poverty and income in the ejido sector. To that end, we develop a model that takes into account the endogeneity of program participation. Following Ravallion and Wodon (1998, 2000), the model uses the decentralization property of program placement to identify the impact of the two programs on poverty, controlling for endogenous participation.

Specifically, we estimate a system of participation and poverty probit regressions in which participation at the household level in the two programs depends on program availability at the state level and poverty at the household level depends on participation at the household level but not on program availability at the state level (a number of state-level geographic controls are also included in the model). The analysis indicates that participation in PROCAMPO significantly reduces the likelihood that a ejido household will be poor, while the more recently introduced ALIANZA program has no significant impact on the household's poverty. It is also shown that estimates that do not control for endogeneity are seriously biased. Next, to explain the strong impact that PROCAMPO has on poverty, we use panel data to examine whether or not PROCAMPO has a multiplier effect on household income. It is shown that there is apparently such a multiplier: one Peso from PROCAMPO results in two Pesos for the household.

The structure of the paper is as follows. Section 2 describes the two Government programs. Section 3 analyzes their impact on poverty. Section 4 tests for the existence of a multiplier effect. In that section, various potential explanations are briefly discussed to explain this multiplier effect. Assuming a constant propensity to invest, the higher income resulting from the PROCAMPO payment may lead to a net increase in household investment, encouraging the producers to purchase additional inputs and to obtain more agricultural income than they otherwise would. The PROCAMPO payment could also result in a higher propensity to invest by providing a more stable and secure stream of income to the household. It could also encourage the household to invest in riskier and higher yielding investments. Another possibility is that the program injects income into the local community, which translates into increased product sales and job opportunities (in a classical Keynesian fashion). A conclusion with some policy recommendations follows.

2 PROGRAM DESCRIPTION: PROCAMPO AND ALIANZA

2.1 PROCAMPO

The Program of Direct Payments to the Countryside (PROCAMPO) is an income support program for agricultural producers that was initiated during the Fall/Winter 1993/94 agricultural cycle. The program aims to ease the transition towards a market economy and specifically the sector's adjustment to the removal of guaranteed prices and market supports for key grains and oilseeds. The program provides agricultural producers (those with the legal usufruct rights over the land) with a fixed payment per hectare that is delinked from current production trends. The number of hectares per producer is determined by the number of hectares the producer had devoted to the production of one of the nine PROCAMPO crops (maize, beans, wheat, cotton, soybeans, sorghum, rice, barley, safflower and barley) in one of the three agricultural cycles preceding August 1993. The payments are made per hectare for each crop season and, for greater transparency, are fixed at the same level across the country.

PROCAMPO is a transitional program expected to last for 15 years and terminate in 2008. During the first ten years the payment was to remain constant in fixed terms and then was to be phased out. In reality, the payment per hectare in real terms has declined an average of 5 percent per year since 1994. The requirements to receive the PROCAMPO payments have changed several times since the program's inception. To be eligible for PROCAMPO payments in 1994, farmers could allocate land to any job creation activity. In 1995, payments were restricted to farmers growing one of the nine eligible crops. As from the fall/winter 1995/96 crop season, farmers are now allowed to devote PROCAMPO land to any crop, livestock or forestry activity, or place it in an approved environmental program, but they are not allowed to leave the land idle or use it for a non-specified activity. Any environmental activity must be designed in coordination with the SAGAR and the Secretariat of Environment, Natural Resources and Fisheries (SEMARNAP).

Nationally, according to government data, PROCAMPO covers 90 percent of Mexico's cultivated area. In the fall-winter season of 1997, almost three million agricultural producers received PROCAMPO payments totaling 7.5 billion pesos (US$ 984 million) and covering almost 14 million hectares. Just under 90 percent of the producers benefiting from PROCAMPO cultivated less than five hectares and received about half of the total amount of payments. Of the area covered, roughly 80 percent was rainfed farm land and about 45 percent belonged to producers with five hectares or less. The regional distribution of the program was fairly similar to the regional breakdown of agricultural GDP for Mexico.

In the ejido sector as well, according to our data, the large majority of the ejidatarios (84 percent) participated in PROCAMPO, receiving a payment for 5.2 hectares on average. The participation rates were lowest for ejidatarios renting out all of their land (perhaps because the PROCAMPO benefit was given to the renter) and for subsistence producers with less than two hectares. In line with the national data, the share of ejidatarios with five hectares or more benefiting from PROCAMPO was 49 percent. The principal reason why 16 percent of the ejidatarios did not participate was because they believed that they did not comply with the requirements (60 percent) or because they did not know about the program (22 percent).

While the PROCAMPO program is widely known and has a positive image in the ejido sector, there remain concerns about the program's management at the local level and the quality of the information which is provided to the farmers. For example, the ejidatarios provided a wide variety of responses on the cultivation requirements for participation in the program. The lack of understanding of these requirements may in part reflect the fact that the cultivation requirements have changed three times since the program's inception in 1994. Moreover, the selection process to identify the qualified beneficiaries is not always clear to some ejidatarios. During qualitative case studies, the ejidatarios indicated that they were not sure how the certification process took place (in some cases, the padron de productores has been updated yearly and in other cases there has been no modification to the padron.)

The principal aim of PROCAMPO is to provide income support to producers of basic crops in Mexico. In this respect, Government has been relatively successful since the average annual value by household of payments in the ejido sector is 2517 pesos (US$317) in our survey. We find that on average PROCAMPO payments provide about 8 percent of ejidatario household income across all income deciles, but the program's contribution represents up to 40 percent of the income in the poorest decile. Given its sheer size, PROCAMPO can be expected to have a positive impact on poverty reduction. The question is: just how large is this impact? And more importantly: does PROCAMPO provides benefits beyond the pure effect of a cash transfer? It is to these questions that we will turn after briefly describing ALIANZA.

2.2 ALIANZA para el Campo

In mid-1996, the Government launched another national rural development program —ALIANZA para el Campo— aimed at fostering agricultural productivity through productive investments (under a matching-grants scheme) and the provision of support services (research, extension, information, training). In practice, ALIANZA offers a menu of 24 federal and at least ten state and regional sub-programs to producers. These are designed to: (1) induce investments in human capital, technology, infrastructure and equipment; (2) support the transformation of the productive structure towards areas where agricultural has a comparative advantage; and (3) promote the insertion of agricultural producers into the marketing chain and world economy.

Today, ALIANZA is the third largest program managed by SAGAR, the Mexico's Ministry of Agriculture (the first two programs are PROCAMPO and CONASUPO), accounting for just under ten percent of the Ministry's programmed expenditures. The major sub-programs within ALIANZA are ferti-irrigation, mechanization, rural equipment, pasture improvement, and "kilo for kilo". Together these sub-programs represent half of ALIANZA's outlays. The ferti-irrigation program aims to increase productivity in irrigated areas by providing financial support for the installation of joint irrigation and fertilization systems. The mechanization program provides supports purchases of tractors and tractor parts. The rural equipment program assists farmers to purchase low cost equipment and technical assistance, while the pasture improvement programs help producers improve the quality of their pasture through improved seeds and infrastructure investments. The "kilo for kilo" program provides farmers with one kilo of certified seeds for the price of normal seeds.

A cornerstone of the ALIANZA program is its decentralized approach and the delegation of administration and decision-making to the states. Almost all ALIANZA programs require cofinancing by state governments as well as by producers. In 1996, producers contributed an average of 50 percent, the federal government 32 percent and the state governments 18 percent. The producer's contribution varies, but may be as high as 75 percent for programs that support the purchase of large scale machinery and as low as 0 percent for the popular improved seed programs.

The distribution of the federal ALIANZA resources across geographic regions is fairly similar, with the exception of the Center region which received almost one third of all federal resources. The remaining regions each received just under 20 percent of the total planned resources in both 1996 and 1997. There were just over one million producers nationwide participating in ALIANZA in 1997 (as of September), the majority of which were private producers, despite the fact that ejidatarios and comuneros were expected to be the primary beneficiaries. The low participation level for ejidatarios could reflect the fact that they have insufficient resources to participate, lack interest in the program, or are simply not aware of the program. In line with the national administrative data which indicates limited program participation among ejidatarios, our survey reveals that only 11 percent of the ejidatarios benefited from ALIANZA in 1997. Participation was highest in the Gulf region and for ejidatarios cultivating more than 18 hectares. Indigenous and non-indigenous ejidatarios had roughly similar participation rates.

3 IMPACT OF PROCAMPO AND ALIANZA ON POVERTY

3.1 A model taking into account endogenous participation

To analyze the impact of PROCAMPO and ALIANZA on Poverty, we need a model that can take into account the potential endogeneity of program participation. That is, while participation in PROCAMPO and ALIANZA may affect poverty, a household's poverty level may also affect participation, and therefore naïve regression estimates are likely to be biased. The model presented below uses the decentralization property of program placement to identify the impact of programs on poverty, controlling for endogenous participation. Specifically, we use a system of participation and poverty probit regressions where program participation at the household level depends on program availability at the state level (for which we have separate data), as well as on other household and regional characteristics. Poverty at the household level conditional on program participation at the household level does not depend on program availability at the state level (for details on the use of this strategy for estimation, see Ravallion and Wodon 1998, 2000, Wodon, forthcoming, and Wodon and Minowa, 2001). The identifying variable at the state level is a relative measure of program availability defined as the percentage of program funds going to a state divided by the percentage of potential beneficiaries (in this case the share of all ejidatarios) living in the state. To take into account non-linearity in the impact of program availability on participation, we use this variable as well as its squared value in the model. We also include a number of geographic controls.

Denote the per capita income level of household i living in area j by the latent variable Y*_ij, and assume that we do not observe Y*_ij, but only the poverty status of the household, denoted by Y_ij. Program participation to Procampo and ALIANZA are denoted, respectively, by the latent variables PR*_ij and AL*_ij to which correspond the observed dichotomic variables PR_ij and AL_ij. Program participation depends on a vector of household characteristics X_i (including a constant), and on the relative availability of the program in the area in which the household lives, denoted by RPR_j and RAL_j (and their squared values; these are continuous variables). Participation depends also on other geographic characteristics of the area in which the household lives, including ejido level variables and a set of regional dummies (defined at a more aggregate level than the state). All these geographic controls are denoted by the vector of geographic characteristics Z_j. As for the probability of being poor, it depends on the same household (X_i) and area (Z_j) characteristics, and on household level program participation PR*_ij and AL*_ij. The system is:

Equations (1)-(2) are the outcome probit for the probability of being poor. Equations (3)-(4) are the PROCAMPO participation probit, and equations (5)-(6) are ALIANZA participation probit. In equations (1)-(2), the endogeneity problem occurs because PR*_ij and AL*_ij are likely to be correlated with the error terms e_Yij, if only because unobserved household characteristics affecting poverty may also influence program participation. This problem is solved by instrumentalizing PR*_ij and AL*_ij using a two stage procedure and the decentralization property of the programs. That is, following Mallar (1977; see also Maddala, 1983: 246), we estimate consistently the above system by first running the participation equations (3)-(4) and (5)-(-6), and then using the index functions (the right hand side values of the first stage regressions) obtained from the first stage regressions in the second stage outcome regression (1)-(2).

3.2 Results for the participation equations

The probit regression for the participation in PROCAMPO in Table 1 (first three columns) reveals that controlling for other household and geographic characteristics, indigenous households are marginally (at the 10 percent level) more likely to participate in PROCAMPO than non-indigenous households. Larger households are also more likely to receive PROCAMPO. While participation rates are lower for households owning a high share of irrigated land, they are higher for households who cultivate a high share of irrigated land. This may reflect the fact that payments go to the household cultivating the land rather than to the owners. Residence in the North Pacific, South Pacific and Center regions had a negative impact on PROCAMPO participation, as compared to the North and Gulf regions. These results reflect the fact that regional participation rates are highest in the North and Gulf regions (88 percent) and lowest in the North Pacific region (74 percent). Importantly, the relative measure of availability of the program in the state has a significant and positive impact on household participation, which is key for our identification strategy. As could be expected, this relationship is non-linear, with a decreasing marginal impact of a higher availability rate (negative coefficient on the squared term).

The results of the probit regression on participation in ALIANZA are given in the next three columns of Table 1. They reveal a very similar pattern for the impact of the share of irrigated land (owned and cultivated) on participation. The data also suggest that ejidatarios residing in more developed and less isolated ejidos (as defined by ejido access to a telephone and dirt roads _ as opposed to paved rodas) are more likely to receive ALIANZA benefits. Ethnicity, education, migration variables do not appear to have a significant effect on participation in ALIANZA. Some regional differences are observed as was the case for PROCAMPO, but this time with households living in the Gulf region being more likely to participate. Again, the relative measure of program availability at the state level strongly affects participation.

Several reasons are advanced to explain the comparatively low outreach of ALIANZA. First, the program had only existed for one year at the time of the survey and had not yet had time to fully penetrate the rural sector. Second, ALIANZA does not appear to have been targeted to the ejido sector. Despite the fact that ejidatarios were 72 percent of the planned beneficiaries in 1997, they were only 11 percent of the actual beneficiaries, suggesting that the program did not target this sector as planned. Indeed, a majority of the ejidatarios (67 percent), and particularly those cultivating small parcels of land, explain that they did not participate in the program because they simply did not know about it. Third, about ten percent of the ejidatarios indicate that a lack of resources to meet the counterpart funding requirements is a constraint for participation in ALIANZA. Fourth, several structural features of the ALIANZA program may limit its impact and therefore the demand for the program on the part of potential beneficiaries. We will come back to these features below in discussing program impact.

3.3. Results for the outcome equations

The impact of PROCAMPO and ALIANZA on the probability of being poor are given in the second part of Table 1. Consider first the (biased) results obtained without control for the endogeneity of program participation. That is, these results were obtained by running equations (1)-(2) without the first stage estimations for equations (3)-(4) and (5)-(6). These naive regressions indicate no impact of ALIANZA on poverty, and a positive impact of PROCAMPO which reduces the probability of being poor by about 10 percent. Compare now these results with those obtained when suitable controls for endogeneity are introduced. While there is still no impact of ALIANZA, the impact of PROCAMPO is much stronger.

We should not be surprised that PROCAMPO reduces poverty. After all, the program provides cash transfers. But how can we account for the apparent lack of impact of ALIANZA? We can suggest a few tentative explanations. The apparent lack of statistically significant impact of ALIANZA on a household's likelihood of being poor may reflect the fact that the program was introduced in 1996, only one year before our survey was implemented. It may well take more than one year for the program's investments to have a beneficial impact on income. But there may also be reasons related to the structure and management of the program behind its insignificant relationship with income.

First, the value of the ALIANZA transactions are small as the ejidatarios tend to select sub-programs that require little counterpart funding as well as low levels of investment (this is the case for example of the "kilo por kilo" and rural equipment sub-programs). Indeed, the majority of the ALIANZA beneficiaries in the survey received less than 1000 pesos (US$117) and also contributed also less than 1000 pesos. While the value of the transactions (both the amount paid in and the producer's contribution) rise dramatically with the amount of land used per producer, this has no impact on the poor who are not large landowners. Interestingly, the value of the ALIANZA transactions of indigenous producers was lower than that of non-indigenous producers, suggesting that while the level of indigenous participation equals that of the non-indigenous population, the indigenous farmers receive a proportionately smaller share of the program's financial benefits (this in turn reflects the indigenous' lower level of wealth and assets).

Next, the lack of impact of ALIANZA may be due to structural aspects of the program which undermine its effectiveness. We can point to several potential problems here. The requirement for many ALIANZA programs that ejidatarios form a group to access the ALIANZA resources was noted as a possible deterrent to ejidatario participation during qualitative case studies. Group efforts tend to have higher transaction costs and past experience with producers' groups have discouraged ejidatarios from embarking on new collective enterprises. Even in communities with strong existing organizations, some ejidatarios noted that they felt uneasy because they perceive they have little control over the decision process and their organizations do not have the skills to manage this process (therefore they depend on others/outsiders to secure the ALIANZA resources). Another problem with the group requirement of some ALIANZA programs is that ejidatarios willing to participate in ALIANZA may not find other ejidatarios interested in the same sub-program. To request support, they may then have to join a group that does not have the same priorities, or choose not to participate at all.

Moreover, during the qualitative case studies ejidatarios expressed concerns about the quality and the price of the inputs purchased under ALIANZA _factors which may reduce the impact of and the overall demand for the program (Gacitua-Mario, 1998). The rationale for the Government (rather than independent producers) to procure the goods and supplies used, in many of the ALIANZA sub-programs is that this may, in theory, reduce the transactions costs of the purchases and lead to better pricing. But this type of organization may also create a whole range of new coordination problems. In one ejido the producers had requested (and contributed money for) hoses for irrigation. Instead they received smaller hoses that could only be used to bring water to individual homes (which for the most part they already had). In other ejidos, there were complaints over the quality of the fertilizer procured by the Government officials under ALIANZA. Ejidatarios also maintained that the distributors selected by the Government overcharged for purchases compared to their own distributors, in effect increasing transaction costs. These incidents have undermined the benefits of the program and led farmers that were considering applying to loose interest as they see little benefits readily accessible.

Incidentally, the fact that the Government tends to procure the inputs used under ALIANZA may also undermine the development of local markets and the sustainability of the program. If producers are to purchase many of the inputs supported by ALIANZA in future years with their own resources, it is important that they strengthen their ties with local suppliers and that the latter become better aware of the specific needs of the producer, so that they will be more likely to stock the relevant items in following years. In a nutshell, the intense involvement of the Government in ALIANZA may do more harm than good.

4. MODELING THE IMPACT OF PROCAMPO ON HOUSEHOLD INCOME

How can we explain the very strong impact of PROCAMPO on poverty? One possibility could reside in the existence of a multiplier effect whereby transferring one Peso to a household through PROCAMPO could result in a gain of more than one Peso. At least three explanations could be proposed to explain such a multiplier effect. First, the income effect of the PROCAMPO payment may lead to a net increase in household investment, encouraging the producers to purchase more inputs than they otherwise would. The expanded use of (better) inputs and investments would in turn lead to a higher net return from the beneficiaries' agricultural activities. For example, ejidatarios with PROCAMPO might purchase more fertilizer than they otherwise would, thereby increasing the profitability of their agricultural activities. This explanation is supported by the fact that about 70 percent of the program beneficiaries claim to use at least part of their PROCAMPO payment to purchase inputs. The multiplier may also reflect opportunities for increased income from the asset endowments that these households have, particularly irrigated land (Sadoulet, de Janvry and Davis, 1999).

Second, the PROCAMPO payment may provide a constant stream of income to the beneficiary households, which encourages the household to invest in riskier and higher yielding investments. It has been shown that given their limited ability to handle risk, poor rural households in Mexico allocate their limited labor, land, and capital in such a way as to avoid risk and income fluctuations (World Bank, 1995). In such a context, the assurance of a basic stream of income may reduce the risk aversion of the poor ejidatarios, making them more likely to shift out of maize and beans into higher yielding and riskier asparagus and strawberries. Third, there may also be a Keynesian multiplier effect, whereby the increased PROCAMPO income generates additional local consumption demand, which translates into increased product sales and work opportunities for households. The strength of the Keynesian multiplier effect depends on how much leakage there is for external products. To analyze this hypothesis further, household expenditure data would be required, which is not available.

To test whether there is indeed a multiplier effect in PROCAMPO, we can use the panel structure of the ejido data sets. While we did not use this in the previous section devoted to poverty, it is in fact feasible to track down the changes in household income observed between 1994 and 1997 for the households in the survey, and to relate these changes in income to the changes in the characteristics of the households, including PROCAMPO participation, as well as to the changes in the characteristics of the areas in which the households live. (Note that we will use here household income as our dependant variable in this section, while we used per capita income in the previous section to determine the poverty status of the households.) Starting from the equation (1), and focusing on PROCAMPO for which we have reliable information on the transfer received from the program, we first specify a fixed effect household level income and participation model as follows (with time subscripts):

Given that the PROCAMPO program did not exist before 1994 and that we have only two surveys, PR*_ijt (the value of the PROCAMPO transfer) is necessarily zero in the first year and the first difference is DPR*_ijt = PR*_ijt. Thus we can eliminate the fixed effects in the participation equation, which collapses to the PROCAMPO equation of the previous section; the only difference being that we use here the continuous value of the transfer rather than a dichotomic participation indicator. We then estimate:

In Table 2, equations (9)-(10) have been estimated using 3SLS. For comparison purposes, equation (9) has also been estimated using simple OLS. While in the OLS estimation there is no multiplier effect (the estimated coefficient is not significantly different from one at the 5 percent level), in the 3SLS estimation there is a multiplier effect. As was the case for the probit estimations of the previous section, not taking into account the endogeneity of program placement results in downward bias in the estimates of the impact of PROCAMPO. With controls for endogeneity, one Peso transferred through PROCAMPO creates two Pesos of additional income for the beneficiary households. This is a strong and encouraging effect, but we would caution that more research, perhaps using other surveys should be conducted in order to establish this finding on firmer grounds. As an aside, it is also interesting to see that an increase in household size or in the amount of land owned also had a positive impact on household income (which is not surprising), while an increase in non-family migrant relationships in Mexico and an increase in cultivated land both have a negative impact on household income (which is less trivial). Sadoulet et al. (1999) also identify a multiplier effect with PROCAMPO of a similar order of magnitude. They find that the multiplier is higher for households with medium and large farms, low numbers of adults in the households, and non-indigenous backgrounds.

Despite this preliminary evidence of a strong multiplier effect with PROCAMPO, the structure and management of the program may still limit the impact of the program. In other words, there remains room for improvement. For example, in almost all cases (95 percent) the PROCAMPO payment arrived after planting, which may decrease the incentive to use the payment to purchase inputs and thus reduce the program's impact. Moreover, the declining value in real terms of the PROCAMPO payment and its variability may reduce the certainty associated with the program, therefore undermining its risk reducing aspects.

Finally, in very few cases do ejidatarios use their PROCAMPO payment as a direct instrument to obtain credit in cash or in kind from a trade, which would encourage further farm or off-farm investment. The data shows that in less than 10 percent of the cases did producers sign over their PROCAMPO check to an input supplier or even have it deposited in a bank. The hesitancy of ejidatarios to use the PROCAMPO payment as collateral for credit may reflect a persisting low threshold for risk, but it may also reflect program features. As currently designed, if producers wish to use their PROCAMPO payment as a collateral, they must sign over their full payment to a bank or input supplier, which denies the household the flexibility to use the payment for a variety of purposes, as is currently done. Another possible explanation for the lack of use of PROCAMPO as collateral may be that banks and traders are unwilling to use the expected future PROCAMPO payment as collateral, given the variable amount of the payment, the uncertainty as to whether or not the local authorities will certify the ejidatario as being eligible to receive the payment for a particular year, and the uncertainty as to whether the government will actually release the payment in case of an economic crisis imposing fiscal and budgetary constraints.

5 CONCLUSIONS AND TENTATIVE RECOMMENDATIONS

5.1 Procampo

PROCAMPO has a very broad outreach in the ejido sector, including among indigenous ejidatarios, and at least half of the beneficiaries are small farmers. Our analysis also suggests that PROCAMPO has important economic benefits for its beneficiaries. As was to be expected with a cash transfer program, PROCAMPO reduces poverty even though it was not originally designed as an anti-poverty program. More importantly, there is evidence (which should be confirmed by further research) that PROCAMPO has a multiplier effect: transferring one Peso may lead to benefits of two Pesos. Hypotheses have been advanced to explain this multiplier effect. Either through the direct effect of cash availability and the lifting of ejidatarios liquidity constraint, or through a reduction in the risk aversion of the beneficiaries, the program may increase the household's investments and facilitate the choice of riskier investments which have higher rates of return. Alternatively, the program's income transfers may stimulate the local economy, raising the demand for local goods and services, thereby creating new income generating possibilities. The program also offers efficiency gains when compared to the previous system of guaranteed Government prices, especially when PROCAMPO payments are delinked from the cultivation of specific crops so that producers make production decisions based on market signals rather than government rules. Moreover, the administrative costs and leakages of PROCAMPO are likely to be lower than those associated with guaranteed prices. Still, despite these benefits, and in the absence of a full cost-benefit analysis of the program which would take into account administrative costs and potential leakage, we are reluctant to draw firm conclusions as to the overall transfer efficiency of PROCAMPO as compared to alternatives programs which could be envisioned.

What could be done to further improve the impact of PROCAMPO on farmers? To enhance program participation and to magnify program benefits, some tentative recommendations (some of which would have to be backed up by further research to ensure their full validity and viability) could be to:

· Distribute the payment earlier in the crop cycle or at least announce the amount of payment prior to planting (even if the actual payment is made after planting). This would add more certainty to the producers' short term income stream, facilitate the purchase of inputs (either on a cash or credit basis) and encourage investment by providing a more secure expected income.

· Allow ejidatarios to designate partial beneficiaries _ either banks or traders _ that could receive part of the PROCAMPO payment. By recognizing that ejidatarios use PROCAMPO as a risk management tool, the Government could promote the use of the PROCAMPO payment, or part of it, as collateral.

· Eliminate inefficient land use requirements _ this would add certainty to the producers' long-term income stream, simplify the eligibility criteria, allow for the earlier payment of PROCAMPO, lower the program's administrative costs and ensure the realization of the full efficiency gains arising from a delinkage between PROCAMPO and production activities. This does not mean that all requirements should be waived, but that the requirement structure be streamlined as much as can be.

· If it is not possible to eliminate the planting requirements, then disseminate these requirements more widely and increase the transparency of the certification process (i.e., make the list open to the public) _ this would again add more certainty to the ejidatario's income stream, clarify the eligibility criteria and increase the likelihood that banks and local traders will accept PROCAMPO as collateral.

· Consider providing technical assistance to ejidatarios with small plots to encourage them to use their PROCAMPO payments to invest in agriculture.

5.2 ALIANZA

The participation of ejidatarios in ALIANZA hovered at just over ten percent. Several factors have been mentioned to explain this low participation rate, including the program's recent introduction at the time of the survey, a lack of knowledge about the program, a lack of resources to meet the counterpart funding requirements, and a number of structural aspects which undermine program effectiveness. In addition, the program does not seem to have any impact on poverty. While this may reflect in part the small value of the ALIANZA transfers and the need to evaluate the program's income effect over a longer time frame in light of its investment nature, there are also other possible explanations. First, the requirement for group membership in some sub-programs may well raise the transaction costs (especially for the poor) of applying to receive ALIANZA, as well as render it more difficult for individual producers to participate in the sub-programs of most interest to them. Second, the fact that government officials often procure the inputs instead of the ejidatarios has led to some concerns that the quality and the price of the inputs received are not optimal. The issue of government procurement also undermines the long term sustainability of the program, since ejidatarios do not have the ability at the moment to strengthen their ties with local distributors and the latter are not made aware of what are the main demands of the producers (so that they can stock these products the following year).

Tentative recommendations which could be proposed for ALIANZA (some of which may need more substantiation from further research) in order to raise participation and program impact are as follows:

· Increase the dissemination of the program in the ejido sector so that a larger proportion of ejidatarios are aware of the program, understand its objectives and are clear on how to access the resources.

· Eliminate the remaining requirements for group participation to access ALIANZA resources. Allow the menu concept of the program to function and permit individual ejidatarios to select the sub-program most suited to their needs. Also allow the ejidatarios to purchase their own inputs directly from local distributors, even if this implies that the purchase price may (but also may not) at first be higher.

REFERENCES

Cord, L., E. Gacitua-Mario and Q. Wodon (1998), Social Capital and Land Titling in Mexico's Ejido Sector, Background Paper in "Mexico Ejido Reform: Avenues of Adjustment _ Five Years Later," World Bank, 1999.         [ Links ]

Cornelius, W. A. and D. Myhre (1998), The Transformation of Rural Mexico. Reforming the Ejido Sector. La Jolla: Center for US-Mexican Studies, University of California, San Diego.         [ Links ]

De Janvry, A., G. Gordillo, and E. Sadoulet (1997), Mexico's Second Agrarian Reform. Household and Community Responses. 1990-1994. Center for US-Mexican Studies, Transformation of Rural Mexico Series 1. University of California at San Diego/La Jolla, San Diego, California.         [ Links ]

Instituto Nacional de Estadística, Geográfica e Informática, INEGI. 1991. Censo Agrícola, Ganadero y Ejidal, Mexico.         [ Links ]

Gacitua-Mario, E. (1998), SRA/World Bank Ejido Study: Qualitative Ejido Case Studies.         [ Links ]

Lanjouw, P. (1999), Poverty and the Non-Farm Economy in Mexico's Ejidos: 1994-1997 Background Paper in "Mexico Ejido Reform: Avenues of Adjustment _ Five Years Later," World Bank, 1999.         [ Links ]

Maddala, G.S. (1983), Limited-Dependent and Qualitative Variables in Econometrics, Econometric Society Monographs, New York : Cambridge University Press.         [ Links ]

Mallar, Ch. D. (1977), The Estimation of Simultaneous Probability Models, Econometrica, Vol. 45, Nº 7, 1717-1722.         [ Links ]

Ravallion, M. and Q. Wodon (1998), Evaluating a targeted social program when placement is decentralized, Policy Research Working Paper Nº 1945, World Bank, Washington DC.         [ Links ]

Ravallion M., and Q. Wodon (2000), Does Child Labor Displace Schooling? Evidence on Behavioral Responses to an Enrollment Subsidy, Economic Journal, 110: C158-175         [ Links ]

Sadoulet, E., A. de Janvry and B. Davis (1999), Cash Transfer Programs with Income Multipliers: PROCAMPO in Mexico, Mimeo, University of California at Berkeley, October.         [ Links ] Wodon Q., forthcoming a, Energy Assistance and Disconnection in France, Applied Economics Letters.         [ Links ]

Wodon Q., and M. Minowa (2001), Training for the Urban Unemployed: A Reevaluation of Mexico's Probecat, in S. Devaradjan, F. Hasley, and L. Squire, editors, Economists' Forum 1999, World Bank, Washington DC.         [ Links ]

World Bank (1995), "Mexico Rural Financial Markets", Report Nº 14599-ME, 1995.         [ Links ]