Adjustment and poverty in Mexican agriculture : how farmers' wealth affects supply response

POLICY RESEARCH WORKING PAPER 1494 Adjustment and Poverty By and large. it appears that Mthe goals of agricultural in Mexican Agriculture reform are being met in Mexico. But measures such as How Farmers' Wealth Affects decoupling income supports and price supports or Supply Response reorienting research and extension could help farmers who cannot afford access to Ram6n L6pez machinery and purchased John Nash inputs and services. Julie Stanton The World Bank International Economics Department International Trade Division August 1995  POLICY RESEARCH WORKING PAPER 1494 Summary findings L6pez, Nash, and Stanton report rhe results of a study of landholdings increased, the average farmer received Mexican farm households using 1991 survey data and a more credit (in real terms), more farm households earned smaller resurvey of some of the same households in income from off-farm work, and more farmers used 1993. purchased inputs. Asset ownership and educational One study goal was to empirically examine the attainment also improved modestly. relationship between assets and the output supply The very small low-Cl group in this sample fared as function. Using a production model focusing on capital well as, or better than, the other groups. True, their level as a productive input, they found that both the supply of educational achievement fell, and fewer of them had level and the responsiveness (elasticities) to changing off-farm income than in 1991. But their use of credit, input and output prices tend to depend on the farmer's irrigation, machinery, and purchased inputs increased net assets and on how productive assets are used. more than for other groups. Regression analysis using data from the surveys shows The limited data are not proof of a causal link, but the that farmers who use productive assets such as machinery fact that the goals are being met should at least ensure tend to be positively responsive to price changes, while that adverse conditions are not undermining reform. those with no access to such assets are not. Farmers that lacked access to productive assets did not Another study goal was to monitor the condition of respond as well to incentives or take advantage of the Mexican farmers in a rapidly changing policy opportunities presented by reform and may need environment. The 1991 survey data suggest that farmers assistance, particularly to get access to credit markets. with more limited use of capital inputs (the "low-CI" There may be a good argument for decoupling income group) were more likely to grow principally corn and to supports from price supports for farmers, since income grow fewer crops, on average, than the others. They also payments that are independent of the vagaries of had more problems getting credit and were less likely to production could provide a more stable signal of use purchased inputs, such as seeds, fertilizer, and creditvorthiness than price supports do. pesticides, or to use a tractor to prepare the soil. They Possibly reorienring research and extension services tended to be less well-educated, and their land tended to more to the needs of low-Cl producers could also be of lower quality. improve the efficiency with which the sector adjusts to Results from the panel data showed conditions new incentives. generally improving for the average farmer in the sample Hypotheses and tentative conclusions from this study area between 1991 and 1993, during a period when will be explored further when more data are collected in agricultural reforms were implemented. Cropping 1995. patterns were more diversified, the average size of This paper - a product of the International Trade Division, International Economics Department - is part of a larger effort in the department to investigate rhe effects of international rrade policy on individual producers. The scudy was funded by the Bank's Research Support Budget under the research project "Rural Poverty and Agriculture in Mexico: An Analysis of Farm Decisions and Supply Responsiveness" (RPO 678-23). Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Jennifer Ngaine, room R2-052, telephone 202- 473-7947, fax 202-676-1341, Internet address jngaine@worldbank.org. August 1995. (47 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas ahout development issues. An objective of the series is to get the fitidings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be used and cited accordingly. The findings, interpretations, and conclusions are the authors' own and should not be attributed to the World Bank, its Executive Board of Directors, or any of its member countries. Produced by the Policy Research Disseminarion Center  ADJUSTMENT AND POVERTY IN MEXICAN AGRICULTURE: HOW FARMERS' WEALTH AFFECTS SUPPLY RESPONSE Ram6n L6pez University of Maryland John Nash World Bank Julie Stanton University of Maryland The authors wish to thank the following people for useful comments and contributions: Louise Cord, Paul Glewwe, John McIntire, Martin Ravallion, and Abdel Senhadji.   I. Introduction I.1 Background: Recent and planned reforms in trade and agricultural policy The Mexican economy has become increasingly open over the last eight years. Since mid-1985, when almost all imports were subject to controls, protection has declined dramatically. Almost all imports are now free from controls, and the maximum tariff now is only 20 percent. Mexico acceded to the GATT in 1986, and recently completed negotiations for the North American Free Trade Agreement (NAFTA), further institutionalizing its commitment to trade policy reform. Until recently, however, the agricultural sector has been a prominent exception to the trade liberalization. Imports of maize, beans, wheat, and barley still require licenses and their levels of protection have remained high. The large devaluation of 1994-95, combined with support prices fixed in nominal terms, have reduced the level of domestic (support and "agreement") prices vis-a-vis border prices, but an increase in these domestic prices is under discussion. One obstacle to liberalizing the grain markets has been the fact that, although the poorest farmers are not net sellers, many poor rainfed farmers outside the subsistence sector rely on these commodities. In addition, a fall in grain prices could reduce the demand of rural labor by large farmers, thereby lowering the wages of the households who are net sellers of labor.Thus, it has been feared that a reduction in protection could significantly increase the incidence and severity of rural poverty. Politically powerful large-scale farmers also have benefitted from this protection and opposed liberalization. These obstacles notwithstanding, the Mexican government is considering moving even faster than required by NAFTA to liberalize the agricultural trade regime. It has announced that under the Programa -1-  Nacional de Modernization del Campo (PROCAMPO), most restrictions on agricultural trade will be eliminated, with income support payments to farmers based on historical (not current) production patterns and volumes. These payments would then be phased out over a 15-year period. The first steps in PROCAMPO began in 1993, but the large-scale delinkage of price and income support is now scheduled for Spring 1996. In addition to these significant changes in output markets that have occurred or are imminent, Mexican farmers also are being faced with monumental changes in input markets. Subsidies have been virtually eliminated for fertilizer use and the former parastatal monopoly on marketing has been opened to competition. Subsidies on electricity (used for irrigation pumps) and surface irrigation water have been reduced. Imports of agricultural machinery, which were restricted in order to protect domestic manufacturers, have been liberalized, as has the market for pesticides. Changes in rural financial markets have reflected the on-going restructuring of the whole financial sector in Mexico. In 1988 and 1989, limits on commercial bank lending to the private sector were first partially relaxed, then eliminated. Following this, the government removed other controls, such as interest rate caps and forced lending quotas. By July 1992, the commercial banking system was privatized. However, the reforms have not had as much effect in the rural sector as in others. Controlled on-lending interest rates, combined with subsidized credit guarantees and agricultural insurance, continue to require large government transfers (though not as large as in the past), and to impede the entry into the market by private institutions. Poor farmers still face restricted access to credit, though the previous practice on the part of governmental banks of only making loans for specific -2-  crops and specific input packages (Heath, 1990) is being phased out. In the land market, recent reforms have been truly fundamental. Much of Mexican agriculture has been organized since the revolution as "ejidos", organizations of farmers, each of whom may have usufruct rights to parcels of land, but not other ownership rights. Thus, they have been unable to pledge land as collateral, making them potentially poorer risks for lending than those with ownership rights. (This was one major raison d'etre for government rural banks.) This system also limited the incentives for investment in land improvements and impeded the consolidation of inefficiently small plots. In early 1992, the Constitution and applicable laws were amended to give ejidatarios much greater flexibility. In principle, they will now be able to rent land, enter into joint ventures with outsiders, sell land to members of the ejido, or sell it to non-members with the approval of a majority of ejido members. Land can now be pledged as collateral. All of these changes are intended to encourage the development of dynamic rural land and financial markets. However, these changes have been slow to be made operational. Only some of the ejidos have been given "certificates of ejido rights" (20 percent in early 1995, with another 50 percent in process), and even fewer have received full private title (de Janvry, e. al., 1995). 1.2. The purpose of this study The economic environment in rural Mexico, as in the rest of the country, has undergone wrenching change in the recent past, with more change yet to come. This mirrors the situation in many other reforming countries, though the changes in Mexico are arguably among the most revolutionary in the world, at least outside of the economies in transition -3-  from socialism. Two major questions facing the governments of reforming countries are whether citizens are able to readily avail themselves of the opportunities presented by the changes, and if not, what policies are needed to facilitate this? These issues are especially critical for the poor, both because they may be less capable than others of adapting to changes and because the consequences of non-adaptation may be more severe. This study has two purposes. One is to monitor the condition of Mexican farmers, especially the poor, and see how changes in the economic environment have affected them. While it will not be possible to prove a causal link between any particular reforms and the changes in practices or circumstances, we will be able to tell if these changes are consistent with the goals of the reforms. This monitoring is important in itself, since if the goals are not being met, it may signal the need to investigate why and to take corrective measures (e.g., accelerating or fine-tuning the reforms, or responding to obstacles to implementation). One lesson of adjustment experience in many countries is that sustainability is increased when there is the perception that reforms are having a pay-off. In establishing this perception, the counterfactual (whether things would have been better or worse without the reform) matters less than whether the goals are being met. The second purpose of this study is to investigate questions related to constraints on the ability of the poor to adjust. The myth that farmers in general fail to respond to changing incentive structures has long ago been discarded. They are neither ignorant nor tradition-bound, and can adjust their inputs, outputs, and mix of activities (both on- and off- farm) in response to changing circumstances. However, there are several reasons that lead one to expect that poor farmers may be -4-  less price responsive than others and, therefore, that both during the transition and in the new less static market environment, poor farmers may be less able than others to take advantage of opportunities. First, poor farmers are more likely to face credit constraints due, in part, to lack of collateral availability. Second, poor farmers are likely to be constrained by their low-quality natural resources, particularly land, which limits the range of crops that can be efficiently grown. Some land could be improved via erosion control, improved fertilizers, access to irrigation, or use of fallow periods, but these options require resources that the poor farmers generally do not have, and also take time to implement. Thus poor farmers may be unable to change their crop and production techniques. Third, unlike large farms which may face an essentially elastic supply of hired labor, small farms are more likely to depend on family labor which has few alternatives and which therefore can be considered quasi-fixed. Without much flexibility in hiring labor, these farmers may be expected to exhibit less price responsiveness. Fourth, the average poor farmer has achieved a lower level of formal education than other farmers. Without the skills needed to adopt new crops and production techniques or to move into non-agricultural activities, farmers may face a drop in income with any adverse change in relative prices. Finally, poor farmers by definition lack a reserve of financial resources for adjusting to shocks. A natural disaster could have immediate and lasting effects on farm productivity, which could not be offset by using more inputs, as it could by farmers with more financial resources. This paper focusses on the first of these arguments for lower price responsiveness by -5-  poor farmers, although we recognize that all may be important. Some of the others will be incorporated in a more general model and tested with a larger dataset as part of the ongoing followup to the current study. I.3. Approach of the paper The work described here will begin to answer the questions of how farmers in Mexico have adjusted to the recent structural changes, and what constraints they face, with a special focus on poor farmers. To do this, we first provide an empirical description of a sample of Mexican farmers, according to production, land and household characteristics. This includes a comparison of characteristics in 1991 and 1993 of the same households to illustrate how they have responded to, and been affected by, changes in the economic environment. We also divide the sample according to productive capital asset use to focus on the question of how relatively poor farmers differ from the other groups. (The way in which these groups are defined is explained in the Appendix.) We next formalize the hypothesis that farmers with lower wealth are less price responsive than richer farmers, in the sense that the former show not only lower levels of agricultural production at a given price, but also impaired responsiveness to changes is prices. We then empirically test the implications of the model in our sample. The final section of the paper discusses conclusions and policy recommendations. The empirical work in the paper is based on the results of two surveys. The first was conducted in 1991 in four states -- Sonora, Puebla, Tlaxcala, and Guanajuato -- for 881 farmers, including both ejidatarios and private farmers. Although the 1991 survey was not -6-  specifically designed to address the issues raised in the current study, it included many relevant questions, which we exploit here. It was not designed to be representative in all respects of Mexican agriculture as a whole. Sonora, situated in the northwestern part of Mexico, has received significant investment in its infrastructure, including irrigation and marketing facilities. The soils, except for recent concerns about salinization, are generally considered very fertile and the mix of production is concentrated in fruits and vegetables more so than the other states in our sample. Much of Sonora's production is intended for export to the United States. Puebla and Tlaxcala, on the other hand, have fewer production amenities. Irrigation has not been widely available and the land is of low quality. Farmers are much more dependent on grain production and government subsidies. Guanajuato falls somewhere near the middle of these extremes. In general, the lands and access to infrastructure are more heterogeneous than in the other regions: some farmers have more in common with their Puebla/Tlaxcala counterparts, while others compare to Sonora farmers. The Bajfo valley of our study area is noted for variability in fertility with few very poor quality areas. However, access to irrigation is not as common as in the northwestern states. In 1993, a limited re-survey for the current study was carried out in Guanajuato, which was chosen based on its heterogeneity. Although approximately 95 farmers were successfully located during the survey period, only 85 surveys have been accepted for analysis due to internal inconsistency or incompleteness of the others. The sample was selected via a random draw of the 300 farmers surveyed in Guanajuato in 1991, and consists -7-  in large part of ejidatarios. This rather lop-sided representation, along with the small sample size, makes it difficult to draw conclusions particular to tenure class. However, the vast majority of farmers in Mexico are ejidatarios, so the conclusions based on this sample are still broadly applicable. As Guanajuato is not one of the poorest areas, some characteristics of this sample may not reflect those of Mexico as a whole, especially with respect to the very poorest farmers. The organization of the remainder of the paper is as follows: Section II contains a detailed discussion of farmer characteristics, including an examination of how these have changed, and how their conditions and practices relate to their wealth. Section III presents a theoretical model emphasizing the way in which wealth affects the level and price- responsiveness of supply, and Section IV describes the empirical model used to test this theory. In Section V, we discuss the main empirical findings. Section VI concludes. II. Statistical Description of Farmers 11. 1 Overview A principal goal of this study is an understanding of the characteristics and practices of Mexico's farmers, especially the poor, and how these characteristics and practices have been modified by changing economic conditions. The analysis in the previous sections implies that in a period of rapidly changing incentives, the poor-- and especially those without access to assets-- would not be expected to fare as well as the non-poor. But it cannot predict to what extent this would produce an observable deterioration in their relative position in response to specific changes such as those in the Mexican economy in the recent -8-  past. In this section, we statistically describe the two samples according to various indicators of their socio-economic condition and farming practices. We compare these indicators across different groups within each year's sample and across years. This is intended not only to descriptively characterize the farmers in the samples, but also to suggest potential connections between observed changes in the sample and structural changes in the economy. II.2 Comparisons of groups based on access to productive capital inputs In this section, we construct an index of farmers' land holdings and access to other production-enhancing capital inputs using the full 1991 survey results. The purpose of this exercise is to gain a clearer understanding of the characteristics, conditions and practices of different economic strata of farmers, by distinguishing between three broad groups. We refer to these groups as "capital inputs groups" (CI groups). (See Appendix I for the definition of the groups and the distribution of farmers across the groups). The capital inputs to which a farmer has access is not the same as the value of all assets owned by a farmer (the variable "A" in the theoretical model below). Rather, the definition of CI groups in this section is intended to proxy the potential permanent income of each farmer by focussing on productive potential. Access to these productive assets is a more reasonable proxy for this than would be either current income or total assets, which includes assets that are not involved in producing future income (e.g., housing). Given the absence of any universally accepted indicator of relative welfare or poverty, we judge this to be the most reasonable for our purposes. In the Appendix, section A. 1, we give more details about the distribution of farmers across the criteria and motivate our definition of the groups. In this -9-  section, we discuss the resulting statistical characterization of the groups. Table 1 provides descriptive statistics for our groups on a wide variety of variables. With respect to cropping patterns, the importance of corn to producers drops significantly' as CI increase, with 77 percent of "low" farmers and only 11.5 percent of "high" reporting it as most important. This is consistent with the widely shared perception that poorer farmers rely more on corn and beans than other crops. Wheat, on the other hand, increases significantly in prominence as principal crop as CI increase, from 5.2 percent of "low" CI farmers to over 58 percent of the "high" group. Sorghum is most often reported as the principal crop by the "medium" asset group, and approximately 5 percent of the farmers in each group report some type of vegetable as their most important crop. Farmers in our "low" group produce significantly fewer crops than the other two groups, which do not significantly differ from each other. That is, the farmers who either have no access to irrigation and harvest machinery or whose land holdings are small produce fewer products on average than other farmers. An area of particular concern in Mexican agriculture is access to credit. The evidence from our sample indicates that there is a significant association between higher CI and higher credit per hectare. Further, the percentage of farmers who ask for credit that actually receive it rises with C1 group, from 58 percent for the "low" group to 79 percent for the "high," while the percentage that reports problems with obtaining credit drops ' Statistical tests are conducted in pairs of asset groups. Differences in means are tested for value variables using SAS' Proc Test. For frequencies, a method described in Fleiss (1973) is used. -10-  significantly as CI increase.2 The farmers receiving higher credit and reporting fewer problems in getting it are the ones with access to irrigation or harvest machinery or who own or work larger parcels of land.3 The dependence of farmers on the labor market is also shown in Table 1. A significantly larger proportion hires labor as CI increase, and the "high" group hires significantly more workers on average than either other group. This may reflect greater land sizes and therefore scale of operation, or perhaps a greater production of cash crops such as vegetables, which are often hand-picked. It may also just reflect a less binding financial constraint on hiring. An interesting result is the lack of statistical difference in the wages paid by all three groups, which seems to indicate a well-functioning rural labor market. The proportion of farmers who purchase seeds, use fertilizer, use pesticides, and prepare soil with a tractor increases with CI group. The percentage that intercrop, often thought to be associated with poorer farmers, decreases with CI group. Thus the farmers who have larger land holdings or better access to large-scale capital more often use other variable inputs that generally increase productivity. Many fewer "low" Cl farmers purchase seeds and use pesticides than in the other two groups. These are inputs not necessarily limited to large-scale operations for effectiveness, so farm size probably has less influence than the conditions that lead to farmers having no irrigation and no harvest machinery. On 2 A problem is indicated by any of several responses by the farmer, including: asked for credit, but was rejected; received less than requested; did not apply because interest rate is too high, or did not know how to obtain credit, or it was not available in the farmer's area. 3 Unfortunately, we are unable to extend this analysis to interest rates paid on credit because many farmers are unaware of the rate they actually pay. This phenomenon arises in part from the nature of most ejido loan arrangements and indeed many independent credit deals which lead the farmer to focus more on what amount is to be paid back than what the implicit rate is. - 11 -  the other hand, it is not very surprising that a significantly lower proportion of "low" CI farmers use a tractor to prepare the ground, as there are economies of scale for this. Ownership of assets such as tractors, farm implements and vehicles is also more common as we move from "low" to "high" groups. To some extent, this might be correlated -12-  Table 1: Comparison of indicators for capital input groups (values are percentages unless otherwise indicated) 1991 T nw_MP__IL__ _ T-T;_gh TCt Principal crop': Corn 77.3 27.0 11.5 a,b,c Wheat 5.2 32.8 58.1 a,b,c Sorghum 3.8 25.9 16.2 a,b,c Vegetables 4.7 5.4 5.3 No. of crops produced 1.6 2.0 2.1 a,b Credit indicators: Credit per hectare owned2 0.59 1.24 1.82 a,b,c Ask for credit 27.0 65.9 91.2 a,b,c Receive credit 15.7 44.6 70.7 a,b,c Receive who ask 58.0 67.9 78.5 a,c Report problems with credit 61.7 31.3 11.5 a,b,c Input usage: Hire labor 63.3 76.2 84.9 a,b,c No. of workers hired 3.1 3.0 5.1 a,c Daily wage (pesos) 19,986 18,656 19,928 Purchase seeds 26.3 84.7 95.3 a,b,c Use fertilizer 75.7 91.2 94.0 a,b Use pesticides 39.7 86.1 90.7 a,b Irrigate 20.9 84.0 91.3 a,b,c Use machinery inharvest 8.5 84.2 85.3 a,b,c Use tractor in preparation of soil 62.5 96.9 97.3 a,b Intercrop 19.3 5.8 2.0 a,b Farm assets: Own tractor 6.7 36.4 59.3 a,b,c Own farm implements 6.7 37.1 59.3 a,b,c Own vehicles 12.3 32.3 70.0 a,b,c Land characteristics: Km to nearest paved road 4.5 3.5 2.9 a,b pct. Irrigated 22.9 68.8 84.3 a,b,c pct. 'Very Fertile' 6.6 17.4 21.3 a,b pct. 'Fertile' 51.5 67.6 65.7 a,b pct. "Poor' 41.6 15.0 12.9 a,b pct. with Flat terrain 61.0 82.1 88.6 a,b,c pct. under Cultivation 82.5 95.1 90.9 a,b,c Education of principal: 36.0 39.8 18.0 a,c No education 57.7 52.4 40.7 a,c 4.0 3.4 19.3 a,c Up to completed primary 1.7 3.7 21.9 a,c Up to completed secondary Beyond_ secondary P Principal Crop refers to the crop which the farmer reports as being his most important. Credit is reported in millions of 1991 pesos. Ct Irrigated and other characteristics indicate the percentage of land area, so the figure is an average. Tests: a: low and high groups significantly different at least 10 percent confidence level. b: low and medium groups significantly different at least 10 percent confidence level. c: medium and high groups siguificantly different at least 10 percent confidence level. -13-  with reported "use", which is a criteria of our group definition. However, many fewer farmers own farm machinery than use (e.g. 37 percent of "medium" own while 84 percent use farm machinery), so the effect is not just an artifact of the way the groups are defined. Rather, this result may reflect the unique status of ejidatarios. (Generally speaking, the ejido collective owns the larger machinery available to ejidatarios, who either rent it from the ejido or share use of the equipment.) A significantly higher proportion of farmers own either a car or truck as CI group increases, indicating that "high" CI farmers also have access to transportation, implying lower transaction costs. Along these same lines, we find that "low" CI farmers must travel a significantly greater distance to the nearest paved road than either other group. "Low" CI farmers report a significantly lower proportion of their land as being "very fertile" or "fertile", and a higher proportion as "poor." Further, an average of 61 percent of the land area of the low CI farmers is flat, compared to 82 and 89 percent for the "medium" and "high" farmers, respectively. Finally, there also appears to be some positive correlation between 'high" CI and education, though not between the other two groups. II.3 Year comparisons Because of the recent changes in the economic environment, we would expect to see changes in the behavior of farmers and in the structure of the agricultural sector. In this section, we report some of the results from the sub-group of 85 farmers who were questioned in both surveys. We also relate these findings to the economic policy changes underway in Mexico. Both the findings themselves and conclusions regarding their causal relationship -14-  with the reforms must be regarded as only suggestive, given the small size of the sample and the other factors at work in the economic environment. As illustrated in Table 2, changes occurred in cropping patterns. There was a significant increase in the number of farmers that planted corn and wheat as the principal crop (government incentives for these crops were high in this period), while sorghum showed a large decline. Overall, the number of crops produced by the average farmer increased. More of the sample in 1993 produced vegetables, and the average output of these crops increased, as well. The greater diversity of crops was also reflected in an increase in intercropping. Diversification can have two effects. The first is to increase average income levels, when the result is higher production of high-value crops, especially fruits or vegetables. Levy (1991) found that "(d)iversification is one of the key mechanisms to increase the returns to land," and noted that in the North and Northwest regions, there was a great deal of this kind of diversification in response to market incentives, but not much in other parts of the country. Another effect is to reduce risk. Farmers' perceived risk may have been higher in 1993 than in 1991. One reason is the removal of the parastatal CONASUPO's guaranteed support prices for some crops. Another is the reform in the credit markets, which eliminated the de facto requirement that all farmers participate in the crop insurance program. Increased risk from these sources would be expected to increase farmers' incentives to self- insure by diversifying their crop mix, as they have apparently done. Other changes could be linked to the reforms in the land market. One of the reasons -15-  for the reform program was the hope that it would lead to consolidation of uneconomically small parcels and the development of a more fluid land market. And, in fact, between 1991 and 1993, the average size of the holdings in our sample increased by about 18 percent4. Moreover, land rentals doubled, and land area used to produce crops increased by nearly one hectare (12%) on average. Some changes are also evident in the factor markets for credit and labor. Recent reform would be expected to increase both credit demand and supply. The supply should increase because some ejidatarios can now use land as security, maling them better risks for commercial lending (Feder and Feeny, 1991). The demand for credit, long-term credit in particular, should increase, since more secure tenure rights enhance the incentives to invest in property. Also, a virtual write-off of much of the farm debt held by the government lending institutions in 1992 may well have increased the demand for credit by creating a perception that debt would not have to be re-paid in the future. With both supply and demand increasing, the equilibrium volume of credit should unambiguously increase as well. Indeed, the average credit received on a per hectare basis rose from 1.06 million pesos (about US$351) to 1.15 million (about US$381) in 1991 values, and the acceptance rate for credit requests of the farmers in our sample rose from 64 to 83 percent. Nevertheless, there is still a perception among our sample of farmers that credit is difficult to obtain. This is evidenced by a significant increase in farmers reporting problems in obtaining credit (from 29 to 53 percent). There was also a drop in the percentage of farmers asking for credit (from 68 to 57 percent.) While the fall in the number asking for credit, De Janvry, et. al. (1995) also reported this pattern of land consolidation in the ejido sector. -16-  Table 2: Comparison of indicators for sample of 85 farmers across years Indicator 1991 1993 Hectares owned 8.32 9.79 Hectares rented 0.46 0.93 Hectares used 7.98 8.95 Principal crop': Corn 30.5% 48.2* Wheat 3.7 21.7* Sorghum 47.6 ,24. 1* Vegetables 6.1 8.4 No. of crops produced 2.28 3.27* Credit indicators: Credit per hectare owned' 1.06 1.15 Ask for credit 68.3 56.6 Receive credit 43.9 47.0 Receive who ask 64.3 83.3* Report problems with credit 29.3 % 53.0* Labor characteristics: Hire labor 84.1% 69.9* No. of workers hired 3.88 8.15* Production characteristics: Purchase seeds 78.0% 69.9% Use fertilizer 90.2 94.0 Use pesticides 78.0 91.6* Irrigate 72.0 80.7 Use machinery in harvest 63.0 81.9* Use tractor in preparation of soil 90.1 92.8 Intercrop 13.4 22.9 Km to purchase seeds 10.51 8.83 Km to purchase fertilizer 10.73 9.14 Farrn Assets: Own home 95.1 100.0 Own tractor 41.5 44.6 Own farm implements 41.5 44.6 Own vehicles 28.0 33.7 Income-related Characteristics: Households with off-farm income 37.8 54.2* No. of family members with off-farm jobs 0.61 1.08* Education of principal: None 58.5% 51.8 Up to completed primary 37.8 42.2 Vp to completed secondcry 0.0 0.0 Beyond secondary 2.4 3.6 * Indicates that the value in 1993 is significantly different from the 1991 value at least a 10 percent confidence level. 'Principal Crop refers to the crop which the farmer reports as being his most important. 2 Credit is reported in millions of 1991 pesos. -17-  although small, is somewhat puzzling, these results overall seem to be consistent with a hypothesis that the availability of credit is improving.5 Labor market conditions have also changed. A smaller fraction of the sample hired labor in 1993, which is consistent with the finding of Heath (1990) that the fraction of ejidatarios that hire labor is decreasing. But at the same time, the average number of workers hired (by those that did hire some) increased greatly. Thus, the net employment of workers increased in 1993 to about 2.1 times its 1991 level. Also, farm households' participation in the off-farm labor market increased. The number of households reporting off-farm income rose from 38% in 1991 to 54% in 1993, while the average number of family members with this type of employment went from 0.61 to 1.08. One explanation of the increase in off-farm labor is that the rural and nonrural labor markets are becoming better integrated. Farmers' access to other purchased inputs seems to have improved. The distance reported to the nearest seed and fertilizer marketing outlets dropped between 1991 and 1993, as would be expected from the privatization and deregulation of marketing channels. The use of most purchased inputs and improved production practices (fertilizer, pesticides, irrigation, harvest machinery, tractor to prepare the soil) increased, though the rate of use of some was already quite high in 1991. (The sole exception was the use of purchased seeds, which declined slightly.) The increase in the incidence of use of these inputs is especially interesting in light of their higher cost (due to removal of fertilizer subsidies), the stronger efforts to recover costs of providing irrigation through higher user charges, and the relaxation of the link between credit and pre-determined input packages.6 This is consistent with other findings that farmers are willing to pay more for inputs when they have more I De Janvry, et. a]. (1995) found that access to credit increased modestly between 1990 and 1994, but that the volume of credit to the ejido sector declined as loans became smaller. 6 De Janvry, et. al. (1995) report that in corn production on ejidos, the use of most purchased inputs declined between 1990 and 1994. Their sample also showed substantially lower use of purchased inputs in both years than did our sample. -18-  control over the quality and conditions of provision (Knudsen and Nash, Postel). Because of problems with the survey, it is difficult to describe changes in the level of income received by the farmers. However, a number of alternative measures seem to indicate that, in many ways, farmers are better off in 1993 than they were in 1991. First, the ownership of assets-- including home, tractor, other farm implements, and vehicles-- has increased. Second, the educational achievement in the sample has improved, with the fraction reporting no education dropping from 59% to 52%.7 While gains in both areas have been modest, they must be interpreted in light of the short time elapsed between the surveys. In addition, the improved access to production technologies, as well as to off-farm work, appears to indicate that farmers in 1993 had more options, and were therefore arguably better off than before.8 II.4 CI Groups Across Years To get some preliminary indication of how changes between 1991 and 1993 in the conditions and practices depended on the farmers' access to productive assets, we divided the 85-household sample into CI groups according to the previously described criteria. Conclusions regarding changes over time for each of these groups are even more tenuous than those based on the sample of 85, principally because of the small size of the groups, especially the "high" group. Tables A3 and A4 in the appendix present figures comparing the groups in the two years. In the paragraphs below, we briefly summarize a few of the main patterns that seem to emerge. The evidence of increased diversification of cropping is similar for all groups. 7 The education level used in the calculations corresponds to the head of household. In some cases, between 1991 and 1993, the person designated as head of household changed (within the same family), usually from an elderly father to his son. Thus, while other household characteristics remain unchanged, the education level of the person responsible for most decisions could change dramatically. I It is not clear what is indicated by a change in the fraction of the farm population that receives off-farm income. Lustig interprets an increase in off-farm income sources as a sign of distress in farm households in the early 1980s, basically a supply-side explanation. We, on the other hand, tend to view this as a demand-driven phenomenon, with farmers being drawn to off-farm jobs because they pay better and are less risky than farm income. Certainly, this is what has occurred in other economies as they have developed and industrialized. In this case, an increase in off-farm income indicates an increase in welfare. -19-  However, the pattern is strongest in the high group, and weakest in the low group. Consequently, whereas in 1991 there was little difference among the groups, in 1993 the high group farmers were the most diversified (4.1 crops) and the low group the least (2.6 crops).9 There were no clear patterns of change among the groups with respect to intercropping. In the credit market, the credit per hectare increased for the low and high groups, while falling for the middle group. This is accompanied by an increase for all three groups of the acceptance rate of credit requests. The fraction reporting credit problems, on the other hand, stayed the same for the low group, while rising for the other two. The number of farmers who hired labor decreased in the lower two groups, while rising in the high group. The number of workers hired increased in the lower two groups and stayed virtually the same for the high group. The rate of participation in off-farm labor markets fell slightly for the low group, while rising most for the medium and rising slightly for the high. The improvement in production practices was most marked for the low asset group. This group showed large increases in the number of farmers purchasing seeds, using fertilizer, using pesticides, irrigating, using harvest machinery, and using a tractor to prepare the soil. The changes in the other groups were small, though they began from much higher bases in 1991. Consequently, the differences among the low, medium, and high groups narrowed substantially. The reported distance to purchase both seeds and fertilizer fell for all groups. There has been considerable discussion of whether structural reforms hurt the poor. (For a survey, see Berg, et.aL, 1993) Since our data only covers the period since 1991, which was well after many of the economy-wide reforms had taken place, they cannot shed much light on the effects of structural adjustment in general. However, many of the reforms that affected specifically the agricultural sector began about this time or later. An analysis of I De Janvry, et. a]. (1995) found that small ejido farmers changed more of their area from monocropped corn to intercropping than did larger farmers. By this measure, small farmers became more diversified to a greater extent than large. -20-  the change in welfare of the low-asset group from 1991 to 1993 may be informative. On balance, the low CI group seems to have been better off in 1993 than in 1991, though the indicators do not all point to this conclusion. On the negative side, the education level in this group fell slightly and fewer of these households had off-farm income. But, their production practices improved across the board, their credit per hectare more than doubled, and the rate of ownership of assets (tractors, farm implements, and vehicles) increased substantially. These results suggest, though they cannot prove, that Berg,