| |
| IDRC:
Resources: Books: Catalogue:
Cover Crops in Hillside Agriculture |
Chapter 7Factors Influencing Adoption of the Abonera SystemHypotheses regarding adoptionThe 1992 farm-survey data indicate that some two-thirds of hillside farmers in northern Honduras use the abonera system to grow second-season maize (see Chapter 4). Although this level of adoption is significant, it also implies that one-third of the hillside farmers do not. Differences between the two groups can be analyzed in light of the features of the abonera system that create costs and benefits. The probability of adoption is likely to be reduced by features that increase the farm-level costs of the abonera system, whereas features that increase the benefits can be expected to increase the probability of adoption. This analysis assumes with conventional economics that farmers' decision-making is based mainly on their objective of maximizing utility at the whole-farm level (Anderson et al. 1977). However, in contrast to adoption studies emphasizing the individual characteristics of farmers at one point in time, our analysis examines the role of broader market effects and ongoing changes in land-use patterns and land-tenure modalities in the technology-adoption process. We recognize that one decision criterion may be important to some farmers but not to others and that interactions between factors may influence farmers' behaviour in unforeseen ways. Below, we describe seven hypotheses regarding features of the abonera system likely to influence adoption, and in the next section we use survey data to test these hypotheses.
The costs of acquiring velvetbean seed and knowledge of its uses may have been high during the early stages of diffusion of the abonera system in northern Honduras; factors influencing this transaction cost, such as farmer characteristics (place of origin, age, education, etc.), may also have been relevant 20 years ago. Today, however, knowledge of the technology is very widespread, and the seed is readily available throughout the region; thus, the costs to farmers of gaining access to the information and seed needed to adopt the technology are undoubtedly very low and have little importance in explaining northern Honduran farmers' adoption of the abonera system. Versions of the adoption model that included farmer characteristics and other statistical tests of their relationship to adoption failed to reveal any significant associations. Consequently, farmer characteristics and other proxies for transaction cost are not included in our adoption model (presented below), although transaction costs may be relevant in areas where the use of the technology is fairly recent. Thus, in some areas, transaction cost could be considered in a general model for examining factors influencing adoption of the abonera system. Empirical analysisThe factors influencing adoption of the abonera system, outlined above, are complex and interconnected. To examine the combined effects of these factors on the adoption decision, we used logit analysis of data from the farm survey of 126 maize producers5 . The variables in the analysis, described in Table 46, follow the hypotheses formulated above. A binary variable for the adoption decision (CHOICE) was defined as 1 if the farmer grew at least some second-season maize under the abonera system during the 1991/92 winter cycle; 0, if the farmer did not. Results of the regression, presented in Table 47, indicate that the combined effects of the independent variables significantly explain adoption behaviour. Sensitivity analysis, presented at the end of this section, helps to rank the importance of each factor.
Land, labour, and markets To examine the impact of land ownership on the adoption decision, we used a binary variable distinguishing between landowners and tenant farmers. As expected, the results of the regression indicate that landowners are more likely to adopt the technology than tenant farmers (hypothesis 1; P <= 0.01). This tendency may reflect the greater potential for landowners to capture the benefits of investment in the abonera system, compared with farmers who depend on rented land. We initially analyzed the effect of land opportunity costs on the adoption of the abonera system through its relationship to a variable for total farm size (including both owned and rented land), on the assumption that the opportunity costs of land dedicated to the abonera system would decline with farm size (hypothesis 2). The results of the regression were not significant, however; farm size had no apparent effect on adoption behaviour. The logit regression was run again with variables distinguishing between the amount of owned and rented cropland and pasture, as opposed to an aggregate variable for farm size. We hypothesized that the opportunity costs of the abonera system may be subject to the amount of land available to farmers through either ownership or rental markets. Well-developed and low-cost land-rental markets in northern Honduras may make it less costly for very small landowners to adopt the abonera system on their own land while renting land for other crops. The effect of land-rental markets on land opportunity costs and subsequent adoption decisions should be most apparent during the first season, when land planted to velvetbean is not available for alternative crops. We created variables for land rented during the first season and land rented during the second season to analyze this effect. Results of the logit regression showed that land ownership and land-rental markets affect adoption in tandem. A variable for the amount of fallow and cultivated land owned by the household (CROPLAND) has a positive effect on the probability of adoption (P <=0.05). In addition, land area rented during the first season (RENT1ST), when the opportunity costs of the abonera system are greatest, also has a positive and significant effect on adoption (P <=0.05). The land area rented during the second season (RENT2ND), when the abonera system presents no opportunity costs, has a negative effect on the probability of adoption (P <=0.1). These results suggest that the relative availability of land resources, either through ownership or rental markets, reduces the opportunity costs of the abonera system and consequently enhances the probability of adoption. The seasonal effects of land-rental markets on land opportunity costs are also apparent. The effect of ranching activities on adoption of the abonera system (hypothesis 3) can be analyzed through its relationship to pasture production. A variable for the amount of pasture owned by a household (PASTURE) has a negative effect on adoption of the abonera (P <= 0.05). (A variable for the ownership of cattle was also tested in the logit model. By itself, it showed a significantly negative impact on the adoption decision, but the PASTURE variable was used instead because it more directly reflects the potential conflict between pasture-management practices and the abonera system.) This result supports the hypothesis that competition between pasture-management practices and the abonera system increases the costs of adoption for farmers engaged in ranching or pasture-production activities. Given the increasing importance of cattle ranching in the region, the apparent incompatibility of these two land uses may become a significant limiting factor in adoption of the abonera system — an issue discussed further, below. A variable for the percentage slope of the largest maize field cultivated by each farm household (LSLOPE) was included in the analysis to assess the impact of land type on the costs and benefits of the adoption decision. A nonlinear form was used for the percentage slope because farmers' concerns about landslides on very steep slopes planted in an abonera system can be expected to reduce the probability of adoption at the upper end of the distribution (see farmers' perceptions of the risks of the abonera system on very steep slopes, in Chapter 4). A positive effect on adoption was found (P <=0.05). As hypothesized (hypothesis 4), farmers with steeper maize fields are more likely to adopt the abonera system, possibly because the potential benefits of the technology (risk reduction) are greater under these field conditions. The impact of household labour constraints on the adoption decision was analyzed through its relationship to the availability of family labour (LABOUR), calculated as the sum of adult family labour (n = men + women). We hypothesized (hypothesis 5) that the labour-saving benefits of the abonera system would be greatest for families with smaller labour resources available for on-farm work. However, the effect of this variable on the adoption decision was not significant, and the sign of the coefficient for the variable was contrary to that expected. One possible explanation for this outcome is that labour-constrained and labour-abundant households may actually benefit equally from adoption of the abonera system, either because labour costs per unit of land are reduced or because labour resources are freed for investment in other activities. More fundamentally, however, the variable may simply not be sensitive enough to capture the effect of relatively small variations in the availability of labour resources within the sample population; the standard deviation for this variable is narrow. The effect of constraints on access to commercial fertilizer was, according to hypothesis 6, an important factor affecting adoption of the abonera system. On methodological grounds, however, the effect was indeterminable on the basis of actual fertilizer use during the second season, as the abonera system is perceived by farmers as a low-cost substitute for commercial fertilizers on second-season maize. As noted previously, farmers do not generally apply fertilizer to maize planted in an abonera system as they believe it is unnecessary for achieving reasonably high maize yields. Hence, fertilizer use during the second season would veil an endogenous relationship with adoption, the dependent variable. This problem of endogeneity does not hold, however, for farmers' use of commercial fertilizer on first-season maize. The application of commercial fertilizers to first-season maize has increased markedly during the last decade and is now widely considered a beneficial but expensive input for first-season maize. Thus, the use of commercial fertilizer on first-season maize (FERT1ST) can be considered a proxy for constraints on access to this input. When introduced into the logit regression, however, the variable is not significant, although the sign of the coefficient for the variable is negative, as expected. Consequently, no conclusion regarding this factor can be reached from this analysis. The effect of market orientation on the adoption of the abonera system (hypothesis 7) can by analyzed through its relationship to farmers' actual maize sales during the second season. The abonera system is ideally suited to the production of maize during the second season, a cycle when maize prices are at their highest. For market-oriented farmers, the potential benefits of adoption can be expected to be greatest during this period. To test this effect, we included in the model a variable (SALES2ND) measuring the proportion of the previous year's second-season maize harvest sold on the market (we used a value of 1 if the farmer sold half or more of the harvest; 0, if not). The sign of the coefficient for this variable is positive, as expected, and significant at P <=0.01, suggesting that the market orientation of farmers does influence the adoption decision. This finding supports arguments made earlier that policy changes affecting the seasonality of maize markets could have impacts on adoption of the abonera system. In sum, four types of factors have significant effects on farmers' adoption of the abonera system: security of access to land (hypothesis 1); influences on the opportunity costs of land, such as farm size, land-rental markets, and the management of pastures (hypotheses 2 and 3); land characteristics (hypothesis 4); and the market orientation of maize producers (hypothesis 7). Constraints on access to commercial fertilizers (hypothesis 5) and the impacts of labour resources (hypothesis 6) were inconclusive. Overall, the factors included in the model enable it to correctly predict 80.2% of the sampled observations, a compelling result for analyses of this nature (see Table 47).
Sensitivity analysis The qualitative and quantitative factors examined above do not all have the same level of impact on the adoption decision. The relative importance of the qualitative factors can be seen by examining the changes in probabilities that would result from changes in the values of these variables. To rank these factors, we defined a "typical farmer" by the most frequent values of the qualitative variables included in the model. Thus, a typical farmer is one who owns land (71.4%), does not apply fertilizer during the first season (71.4%), and sells less than half of the second-season maize harvest (51.6%). Table 48 shows the probability of adoption for this typical farmer and the effect of changing the values of the qualitative variables. Results are consistent with expectations. The probability of adoption for a typical farmer evaluated at the sampling mean of the quantitative variables is 64%, a measure virtually equal to the actual level of adoption indicated by the survey data. By contrast, farmers who are typical in all respects except that they sell more than half of their second-season maize crop have a much higher probability of adoption (an increase of 40% over that of the typical farmer). The probability of adoption among farmers with a typical profile but without land ownership decreases by 42%, a clear indication of the influence of this factor on the adoption decision. Finally, the probability of adoption among typical farmers decreases by 14% if they also fertilize first-season maize. A different approach is needed to measure the sensitivity of quantitative variables to changes in their values. The relative importance of the quantitative factors in the adoption decision can be seen by examining variable elasticities, defined as the percentage change in probabilities that would result from a percentage change in the value of these variables. These values are calculated for a typical farmer as described above, as well as for a typical farmer more oriented to the market. Table 49 shows the results for both types of farmer.
For a typical farmer, the opportunity cost of land, as measured by the variables CROPLAND and RENT1ST, has a sizable impact on the adoption decision. For example, an increase of 10% in the average amount of cropland owned increases the probability of adoption by almost 2.45%. Similarly, an increase of 10% in the area rented during the first season increases the probability of adoption by 3%. By contrast, the impact of the quantitative variables is much less among the market-oriented farmers, indicative of the high probabilities of adoption already found among farmers with this profile. Adoption and livelihood strategies The logit analysis of factors influencing adoption helps explain the pattern of adoption found among household groups identified in Chapter 3 (Table 50). The data show that subsistence workers are least likely to adopt the abonera. This may be due to their dependence on small parcels of rented land for maize production. The high rate of adoption among medium-scale farmers may be due to their being relatively free of land constraints and fully engaged in commercial maize production. Small-scale farmers are somewhere in between these two situations; they are land constrained but have some land of their own where an abonera can be established. Diversified farmers have only an average level of adoption, despite being relatively well endowed with secure land resources. This may be due to the competing demands on the land held by members of this group. Diversified farmers are struggling to become ranchers rather than farmers and may tend to emphasize pasture production over other land uses. The opportunity costs of the abonera system may consequently be higher for this group. Established ranchers, by contrast, can afford to dedicate a few hectares of land to the abonera system without significantly affecting their ability to acquire pasture for their cattle. Among this group, adoption of the abonera system is high.
Discussion and conclusionsThe analyses presented above indicate that land tenure, land distribution, competing land uses, and relative prices of outputs (maize) significantly influence farmers' adoption of the abonera system. These findings have implications for policymakers and researchers concerned with developing hillside agriculture. First, it seems clear that security of access to land is a fundamental condition for investment in productivity-enhancing, resource-conserving technologies in hillside environments. This general conclusion should be qualified, however, in light of several distinctive features of the land-tenure system in northern Honduras. As noted in Chapter 2, individual land ownership can take two forms in northern Honduras: titled property (dominio pleno) and squatters' rights (dominio útil). Titled property is fully recognized by the state, which conveys the right of use and unimpeded transfer of land. Squatters' rights also convey the right to use and transfer land, so long as annual municipal land taxes have been duly paid and the buyer assumes the obligation to continue paying these taxes. These rights are less flexible, however, than titled property because banks and other lending institutions do not recognize squatters' rights as guarantees against farm loans. This represents an unimportant limitation for most farmers, however, as farm credit is extremely limited throughout the region anyway. Survey data from 1990, distinguishing between titled property and squatters' rights, indicates that squatters and titled owners are equally disposed to adopt the abonera system; adoption rates for these two types of landowners are statistically the same (Buckles et al. 1991). This finding suggests that although land ownership is an important consideration in the adoption decision, the form of land tenure may not be. A legal tradition that recognizes the rights of squatters to use public lands seems to convey with it the level of security of access needed by farmers to invest in the abonera system. This experience contradicts the common assumption that formal land titles are the only form of land tenure consistent with the long-term planning horizon needed to support the adoption of technologies with long-term benefits. An implication of this finding is that policies reinforcing the rights of squatters might be just as effective in providing security of access to land and facilitating use of conservation technology as formal titling programs. Second, the lack of land ownership is not an absolute limitation on farmers' use of the abonera system. A third of the tenant farmers surveyed reported that at least some of their maize was planted in an established abonera rented from someone else. This is possible because abonera land-rental markets have developed throughout the region in recent years, as farmers with more land than they can cultivate themselves divert some of it to aboneras for rent or for use by family members. For landowners, an abonera is an improvement in the land that can be captured in higher rental rates; as noted in Chapter 6, farmers are willing to pay a premium of 60–70% for rights to cultivate maize on land with an established abonera, a clear indication of the potential the farmers perceive in the field. Thus, although land ownership is important, the development of abonera land-rental markets has facilitated the use of this technology by landless farmers as well. The adoption decisions made under the two circumstances are nevertheless distinct. For landowners, a decision to adopt the abonera system is relatively secure and enduring; they can expect to realize tangible benefits from the investment over an indefinite period of time. By contrast, tenant farmers decide whether to rent an established abonera, with the expectation that the field will be immediately more productive than lower-cost alternatives. Their decision is subject to the availability of established aboneras in uncertain land-rental markets, and their use of the system is potentially discontinuous (their use of the abonera system may be interrupted, as was noted in Chapter 4). Land-rental markets are important in providing access to aboneras not only to tenants but also to small-scale landowners. The logit analysis demonstrates that differences in the amount of land resources available to farmers, either through ownership or land-rental markets, modify the opportunity costs of the abonera system and consequently the probability of adoption. In tandem, land ownership and first-season land-rental markets seem to have a significant impact on the probability of farmers' adopting of the abonera system. An explanation for this result is that farmers with larger farms are more likely to adopt the technology; farmers with smaller farms of their own are likely to adopt it if they can rent the land for the first-season crops that the aboneras displace. Data for landowner groups are presented in Table 51; these data show high adoption rates, even among farmers with very little land of their own. These farmers adopt the abonera on their own land and rent land for other crops.
Land constraints on farmers' adoption of the abonera system are currently eased by a well-developed and low-cost land-rental market. This situation is subject, however, to changing land-use patterns affecting northern Honduras. Pasture production, stimulated by new markets for milk products, is expanding rapidly throughout the hillside area. As noted in Chapter 6, milk and cheese production is more profitable than annual crops and entails fewer risks. For farmers with enough resources to become ranchers, the shift from crops to pasture can improve their livelihoods. The people most likely to be in this situation are the diversified farmers, who work with numerous competing demands on their land resources. For many people in hillside households, the purchase of cattle of their own is beyond their means, leaving them the more limited option of renting out land for pastures or selling some land to finance the acquisition of cattle. Qualitative evidence (Humphries, in press; DB's field observations) suggests that larger-scale ranchers residing in coastal communities not included in this study are acquiring the more accessible and better-quality land in the hillsides for seasonal grazing of herds, displacing small-scale producers to more marginal lands. These developments increase pressure on land-rental markets, reducing the availability of land for first-season crops and consequently making it increasingly difficult for small-scale operators to dedicate land to the abonera system. In the Conclusion, we discuss this threat to the stability of the abonera system and the opportunities to enhance its productivity without undermining its ecological merits. 5 The assistance of Kenneth Mullen and Paul Heisey in the use of logit techniques is gratefully acknowledged. Errors in interpretation are, however, the sole responsibility of the authors. [BACK]
Table of Contents < Previous - Top - Next > | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||