Choosing the" right" tools to assess the economic costs and benefits of growing landraces: an example from Bara District, Central Terai, Nepal

ARTICLE Plant Plant Genetic Genetic Resources Resources Newsletter, Newsletter, 2003, 2003, No. 134: No. 134 41-44 1 Choosing the ‘right’ tools to assess the economic costs and benefits of growing landraces: an example from Bara District, Central Terai, Nepal Devendra Gauchan1 and Melinda Smale2! Nepal Agricultural Research Council (NARC), PO Box 5459, Kathmandu, Nepal International Plant Genetic Resource Institute, Rome, Italy, and International Food Policy Research Institute, 2033 K St. NW Washington, DC 20006, USA Email: M.Smale@cgiar.org 1 2 Summary Résumé Resumen Choosing the ‘right’ tools to assess the economic costs and benefits of growing landraces: an example from Bara District, Central Terai, Nepal Choix du meilleur outil pour l’évaluation des coûts et bénéfices de la culture de variétés locales : exemple du district de Bara, Terai central, Népal Over the past decade there has been renewed scientific interest in strategies for in situ conservation of crop genetic resources, with calls for concise, empirical estimates of the relative costs and benefits of growing landraces. Economists often use marginal analysis based on partial budgets as a tool for estimating the economic returns farmers might expect from using (or choosing not to use) a new practice. However, caution must be exercised when applying this tool in semi-commercial agriculture and especially in analysing the costs of benefits of growing landraces. In semi-commercial agriculture, incomplete markets cause the effective input and output prices actually faced by farmers to diverge within a band defined by producer and consumer prices. In addition, markets may be partially absent for landraces or market prices may fail to reflect their distinctive attributes. Here, we illustrate and expand these points with an analysis that compares the costs and benefits of growing landraces instead of modern varieties in Nepal, a center of rice diversity. We also suggest other types of economics tools that may be of use in assessing the costs and benefits of growing landraces, and in addressing issues related to design, implementation and monitoring of projects to conserve crop biodiversity on farms. Au cours des dix dernières années, on a assisté à un regain d’intérêt scientifique pour les stratégies de conservation in situ de ressources phytogénétiques, rendant nécessaire des estimations simples et empiriques des coûts et bénéfices relatifs liés à la culture de variétés locales. Les économistes utilisent souvent l’analyse des coûts marginaux, fondée sur le concept de budgets partiels, afin d’estimer le bénéfice que les agriculteurs peuvent escompter en adoptant (ou non) une nouvelle pratique. Cependant, il faut être prudent en appliquant cette méthode à une agriculture semi-commerciale et, en particulier, à l’analyse des coûts et bénéfices liés à la culture de variétés locales. Dans le cas d’une agriculture semi-commerciale, du fait de la commercialisation incomplète, les prix des entrées et des sorties réellement supportés par les agriculteurs fluctuent entre les limites fixées par les producteurs et les consommateurs. En outre, les débouchés commerciaux des variétés locales peuvent être en partie absents ou les prix pratiqués peuvent ne pas refléter les caractéristiques propres à ces variétés. Dans cet article, nous illustrons et développons ces points par une analyse comparative des coûts et bénéfices de variétés locales cultivées en substitution aux variétés modernes au Népal, centre de diversité du riz. Nous suggérons également d’autres types d’analyses économiques utilisables pour évaluer les coûts et bénéfices de la culture de variétés locales. Nous abordons aussi les questions relatives à l’élaboration, la mise en oeuvre et le suivi des projets destinés à maintenir la biodiversité des plantes cultivées sur le site de l’exploitation. Selección de los instrumentos adecuados para evaluar el costo económico y los beneficios del cultivo de variedades nativas: un ejemplo del Distrito de Bara, Terai Central, Nepal Key words: partial budget analysis, on-farm conservation, genetic resources, rice diversity, on-farm conservation En el último decenio se ha renovado el interés científico por las estrategias para la conservación in situ de recursos genéticos de cultivos, y se han pedido estimaciones empíricas concisas de costos y beneficios comparados del cultivo de variedades nativas. Los economistas recurren a menudo al análisis marginal basado en presupuestos parciales como instrumento para estimar los rendimientos económicos que pueden esperar los agricultores si usan (o si optan por no usar) una práctica nueva. Pero hay que ser cautos al aplicar este instrumento en la agricultura semicomercial y en especial al analizar los costos y beneficios del cultivo de variedades nativas. En la agricultura semicomercial, lo incompleto de los mercados hace que los precios efectivos de insumo y producto con que operan de hecho los agricultores oscilen dentro de una banda definida por los precios del productor y del consumidor. Además, los mercados pueden faltar parcialmente para las variedades nativas o los precios del mercado pueden no reflejar sus características distintivas. Aquí ilustramos y ampliamos estas cuestiones con un análisis en el que se comparan costos y beneficios del cultivo de variedades nativas en lugar de variedades modernas en Nepal, un centro de diversidad arrocera. Proponemos también otros tipos de instrumentos económicos que pueden ser útiles para el cálculo de costos y beneficios del cultivo de variedades nativas y para tratar cuestiones relativas a diseño, ejecución y supervisión de proyectos de conservación de la biodiversidad en las explotaciones agrícolas. Introduction Plant genetic resource conservation efforts have long emphasized the collection of samples of biological diversity for storage seed ex situ (off site) rather than in situ (on site). This emphasis largely reflected the concern that potentially valuable landraces might be lost to natural disasters, wars or the rapid diffusion of modern varieties during the initial phases of the Green Revolution (Frankel and Bennett 1970; Frankel and Hawkes 1975). Over the past decade, scientific interest in techniques for in situ conservation of genetic resources has re-emerged. In situ conservation involves the maintenance of genetic variation at the location where it is 2 Plant Genetic Resources Newsletter, 2003, No. 134 encountered, either in the wild or in traditional farming systems, as part of a farming system, agroecosystem or habitat. From the perspective of the genetic resources that the two strategies seek to conserve, in situ and ex situ conservation can be viewed as complementary (Maxted et al. 1997) or as ‘intergrading phases of a continuum’ of conservation methods (Bretting and Duvick 1997). A form of in situ conservation, ‘on farm’ conservation is the choice by farmers to continue cultivating genetically diverse crops in the agricultural systems where the crops have evolved (Bellon et al. 1997). Landraces are often the focus of onfarm conservation efforts, since they tend to be more heterogeneous in a number of traits than modern varieties (MVs), which are bred to conform to certain standards required for commercialized agriculture. At the beginning of the Green Revolution, many viewed farmers’ decisions to replace landraces with modern varieties as the inevitable price to pay for economic development. Since superior modern varieties would sooner or later replace landraces, in situ conservation of crops was considered infeasible and researchers advocated collection for safe ex situ storage (Frankel and Bennett 1970; Frankel and Hawkes 1975). Encouraging farmers to continue cultivating landraces would deprive them of development opportunities, relegating them to the poverty they seek to leave behind. However, as argued by Brush (1995) and others (Altieri and Merrick 1987), de facto conservation of landraces persists even for highly bred, major food crops, such as rice, wheat and maize, because some farmers in some environments choose to continue growing them. The design, implementation and monitoring of in situ conservation projects involves the following economics issues. First, the factors that significantly affect the likelihood that farmers will continue to choose growing diverse crops as economic changes occur must be identified, along with the farmers most likely to do so. These relate to opportunity costs. Second, the relative costs and benefits they perceive from growing these diverse crops must be estimated, and compared to the benefits perceived by other members of society, such as consumers in nearby or distant locations. These concern private and social values. Ultimately, policy instru- ments must be identified that most effectively support the maintenance of crop biodiversity (the benefits of in situ conservation) while compensating for differences between farmers’ and society’s perceptions (the social costs of in situ conservation), in sites where the likelihood that farmers will continue growing diverse crops is greatest (farmers’ opportunity costs are least). So far, no definitive estimates of the relative costs and benefits to farmers of growing landraces vs. modern varieties have been proffered to enlighten policy debates. Why? The short answer is that the marginal net benefits of growing landraces relative to modern varieties (or vice versa) are specific to farmers, the germplasm available to them, the type of agricultural system in which they farm, and points in time. Furthermore, it may not be possible to measure net benefits in terms of a single trait, such as grain yield at harvest. Finally, these represent only the private costs and benefits to farmers, and do not include the social net benefits. In this paper, we begin by using a marginal analysis of partial budgets to explore the relative costs and benefits of cultivating rice landraces compared with modern varieties in a Nepalese community. The next section describes research methods. We then present the results of the analysis, and discuss the limitations of the tools using survey data from the same location. We suggest some alternative approaches in the concluding section. The purpose of this exercise is to provide an illustrative example for those who are concerned with economic aspects of on farm conservation—be they applied economists or plant genetic resource managers. Research methods This paper employs data from two sources. The first is a baseline survey conducted by Nepal’s in situ conservation project (summarized in Rana et al. 2000). Three ‘eco-sites’ were purposely selected to represent a major physiographic region (Mountain, Hills and Tarai) of Nepal (Figure 1). The term ‘eco-site’ refers to a watershed area that includes a cluster of communities or villages. Criteria used to select eco-sites include agro-ecological features and the importance of crops targeted for on-farm conservation in the farming system, as well as their diversity. In Nepal, these 2200-3000 msl 0 modern rice varieties 21 rice landraces 600-1400 msl 6 modern rice varieties 63 rice landraces 80-90m msl 20 modern rice varieties 33 rice landraces Figure 1. Map of Nepal showing the location of three eco-sites and their main features. Plant Genetic Resources Newsletter, 2003, No. 134 3 include rice, barley, buckwheat, pigeonpea, taro, cucumber and spongeguard. Rice is the major crop in the food economy. Before the baseline survey was conducted, households in each site were listed and stratified according to wealth-related criteria. A self-weighting sample of households was drawn using a constant sampling fraction of 22% across sites and sampling from wealth strata proportionate to the probability of selection within each site. Sample sizes were 202, 206 and 180 households in Bara, Kaski and Jumla eco-sites, respectively. The survey instrument covered: (1) social, demographic and economic features of farm households; (2) cultivation of landraces and modern varieties; (3) farmers’ perceptions of traits associated with landraces and modern varieties; (4) farmers’ use of purchased inputs; and (5) farmers’ access to information and support services. Bara was then chosen as the eco-site in which to examine more closely the relative costs and benefits of growing landraces versus modern varieties, for several reasons. The Bara eco-site is located in the low altitude (80–90 m asl), sub-tropical, fertile zone of the Indo-Gangetic plain (Tarai region) on the southern border with India. Farmers in this area have better access to inputs, modern technologies and marketing opportunities than in Kaski or Jumla eco-sites. Rice is grown in subsistence to semi-commercial production systems, and 68% of the total rice area is under seasonal or partial irrigation. Market towns are fairly uniformly distributed throughout the area and are accessible by dirt roads. A total of 53 different varieties of rice have been identified in the farming communities of the Bara eco-site, of which 33 are landraces and the remaining 20 are modern varieties (Rana et al. 2000). Farmers in Bara eco-site grow a larger number of modern varieties than in either Jumla or Kaski eco-sites. A set of structured household interviews was then implemented with a sample of 40 households selected at random in the Bara ecosite for both marginal and favoured environments. The purpose of the second survey was to collect cost-of-production data for marginal analysis of economic returns to cultivating rice modern varieties and landraces. Partial budgets were constructed as indicated in CIMMYT (1988). A combination of on-farm experiments and farmer surveys are the ideal source of data for such analyses, but on-farm experiments were not available for the illustrative case presented here. The yields reported are subjective yield estimates of farmers. Marginal analysis based on partial budgets Partial budgets organize data about the costs and benefits of alternative agronomic treatments or technologies for the purpose of establishing recommendations for farmers. Marginal analysis compares the treatments or technologies in terms of benefits net of costs that vary. Costs that vary are the costs per hectare of purchased inputs, labour and machinery that differ between the treatments or technologies in question. Resources are valued by their opportunity costs—that is, the value of the resource in its best alternative use (CIMMYT 1988). Marginal analysis with partial budgets lends few insights into variety choices unless, in addition to yield differences between varieties, farmer management and other costs associated with the seed also vary by type. While we would expect that management might differ between modern and landrace seed types, often farmers manage their landraces in a similar fashion. Partial budgets examine only a single activity, abstracting from other resource-use decisions on the farm. In the case of variety choice within a crop, however, it is often reasonable to assume that farmers allocate labour and land resources initially among crops, and given that decision, choose among varieties of crops (Barkley and Porter 1996). Hence we would not expect the results for seedtype comparisons to differ in a meaningful way between wholefarm and partial budget analyses. One notable exception would be a choice of variety that enables or reduces costs of a rotation with another crop because the variety matures earlier or can be planted later, releasing land or labour resources in a timely way. For example, in the case of cotton–wheat rotation in southern Punjab of Pakistan, Byerlee et al. (1987) used data on both crops to analyse the economic benefits from newer wheat varieties that could be planted later, enabling an additional cotton harvest. Table 1 compares partial budgets for two rice varieties when grown in the favourable environments of Bara eco-site. Favourable environments are defined as those in which farmers have access to irrigated productive farmland (mid-wetland), seed of modern varieties, information about rice production technology, and markets. Market-traded inputs and outputs are valued at prevailing market price at the farm gate. Family labour is valued at the prevailing wage rate in the local market. The wage good (rice grain paid in kind) is quantified by adding the costs of snacks that farmers serve to labourers while they conduct farm operations. Rice by-products such as straw are valued by the local market price at harvest. Table 1. Comparing the net benefits to farmers of growing the rice landrace (Nakhisaro) and modern variety (China-4) in favourable production environments of Bara eco-site Benefit/cost category Nakhisaro China-4 Benefits Grain yield (kg/ha) Grain price (Rs/kg) Straw yield (kg/ha) Straw price (Rs/kg) Gross field benefits (Rs/ha) 2200 7.5 4200 0.5 18600 3150 8.0 3400 0.5 26900 Costs that vary Seed (Rs 10/kg) Urea (Rs 8/kg) Bullock Labor (Rs 50/man-day) Pesticide (100ml) Total costs that vary (Rs/ha) 500 720 3900 10100 0 15220 550 1200 3900 10050 100 15800 3380 11100 7720 Net benefits (Rs/ha) Marginal net benefit of growing China-4 (Rs/ha) Marginal rate of return to growing China-4 Source: Field Survey (1999). Note: US $1=NRs 68 in 1999. 13.3 4 Plant Genetic Resources Newsletter, 2003, No. 134 In favourable environments the grain yield advantages of the modern variety China-4 over the landrace Nakhisaro are pronounced though straw yields are somewhat lower. The same amounts of bullock and human labour are used per hectare to grow either variety, though pesticides and some additional urea are applied to China-4. The price earned by selling China-4 is also slightly higher, which likely reflects the uniformity of grain type that is appealing to commercial buyers who can market it in larger volumes at lower costs. In the budget shown in Table 1, for every additional Rupee invested in growing China-4 rather than Nakhisaro, farmers earn 13.3 Rupees. The marginal rate of return exceeds by over 10 times the minimum rate of return that empirical evidence has shown is acceptable to farmers (50–100%, according to CIMMYT 1988). In other words, in the favourable environments of Bara ecosite, the economic opportunities foregone by planting a hectare in the rice landrace Nakhisaro rather than the modern variety China-4 appear to be enormous. As a consequence, farmers in such environments have replaced, or are in the process of replacing, landraces such as Nakhisaro, Sotwa and Mutmur with modern varieties like China–4. For purposes of comparison, taking Rs. 7720 as the opportunity cost of growing a hectare of Nakhisaro in favourable environments, the private costs to the farming community of planting all of its 136 hectares in this landrace would be over Rs. 1 million per season (over $US15 000, 1999). This result supports the hypothesis that in favourable production environments, when farmers have access to high quality land, irrigation (which mediates production risk), seed of locally adapted modern varieties, and well-functioning markets, there are likely to be high opportunity costs associated with encouraging them to continue grow landraces rather than modern varieties. The italics, however, are the most crucial aspect of the conclusions we draw from the findings presented in Table 1. First, we know that many farmers have no access to high quality, irrigated land. Indeed, the partial budgets shown in Table 2 demonstrate that in particular niches such as poor upland soils or swampy land of the Bara eco-site, certain landraces can be profitably grown. No modern varieties currently available in this location can compete with them. Bhathi has a benefit–cost ratio of 3.9:1 as compared to 1.4:1 for Mutmur, primarily because it earns a higher price on the local market. Farmers forego no development opportunities when they grow these landraces in poor environ- ments. At present, conservation of these landraces in specific production niches occurs at zero opportunity cost to farmers and no public funds are required to compensate them. Secondly, we know that access to seed and product markets, as well as related information, is not equal among farmers. Nor do all farmers have the objective of maximizing profits, since many in the Bara eco-site remain subsistence-oriented. The next section uses farm survey data to explore what this means for the way in which we use partial budgets as an economics tool. Farmers’ choices of rice types in Bara eco-site The data in Table 3 illustrate the breadth of choices that farmers themselves find optimal in Bara eco-site. Rice farming households include: (i) growers of landraces only; (ii) growers of modern varieties only; and (iii) growers of both landraces and modern varieties. Farmers also allocate their rice area among types in Table 2. Marginal net benefits to farmers of growing Mutmur and Bhathi landraces in poor environments of Bara eco-site Benefit/cost category Mutmur (upland) Bhathi (pond) Benefits Grain yield (kg/ha) Grain price (Rs/kg) Straw yield (kg/ha) Straw price (Rs/kg) Gross field benefits (Rs/ha) 2100 7.5 3200 0.5 17350 1500 13 1400 0.5 20200 Costs Seed (kg/ha) Seed price (Rs/kg) Urea (Rs 8/kg) Bullock Labour Tractor Total costs (Rs/ha) 55 10 480 3600 7875 0 12505 50 15 0 800 3240 375 5165 Net benefits Benefit to cost ratio 4850 1.41 15035 3.9 Source: Field Survey (1999). Note: USD $ 1= NRs 68 in 1999. Table 3. Farmers’ choice of rice types and rice area allocation in Bara eco-site Variable No. farmers growing Percent farmers growing Mean number of varieties grown Total rice planted area (ha) % Rice area in landraces % Rice area in modern varieties Rice types grown Landraces only Modern varieties only Both modern varieties and landraces 14 7 1.28* 0.39 100 0.0 100 51 1.84** 0.52** 0.0 100 83 42 4.0*** 1.21*** 25 75 Note: Pairwise t-tests show significant difference of means between categories at the 0.05 level with two-tailed test: *grow landraces only–grow modern varieties only; ** grow modern varieties only–grow both; ***grow landraces only–grow both. Source: computed from Baseline Survey Data, In Situ Project Nepal. Plant Genetic Resources Newsletter, 2003, No. 134 5 different proportions and grow different numbers of cultivars. Farmers growing only landraces are a minority and have smaller rice areas. Growers of both landraces and MVs tend to farm larger areas and maintain more rice varieties (an average of four in total). Half the households grow only MVs and many of these grow more than one. The data shown in Table 3 show that even though the marginal analysis suggests that modern varieties dominate landraces in the best environments, many farmers in the study site choose to grow some combination of modern varieties and landraces, and in varying proportions. What explains the discrepancy between the evidence presented in the choices farmers make and the evidence presented in Tables 1 and 2? Marginal analysis as represented in partial budgets is based on neoclassical economic theory. Neoclassical economic theory predicts that a farmer will choose to grow only the variety with the highest expected profits, offering no explanation for why a farmer grows two or more varieties simultaneously. Economists have proposed extensions to the expected profit maximization framework to explain similar empirical findings to those presented in Table 3, often termed ‘partial adoption’. These extensions to the neoclassical model have evolved over time, reflecting new paradigms in the theory of agricultural development. Often, new approaches represent slightly different ways of expressing or viewing the same problem. The economic models of the early Green Revolution period (1970s) typically embodied an assumption that the new seed varieties were superior to the varieties then grown by farmers. In this paradigm, partial adoption reflected farmers’ need to adjust to new technology. With time and the improvement in skills, technical and allocative efficiency was expected to increase and farmers would plant more area to the new seed variety (for example, Kislev and Shchori-Bachrach 1973; Hiebert 1974). In a final equilibrium state, ‘efficient’ farmers would plant all of their crop area to modern varieties. The theory of farmer decision-making under risk was a dominant paradigm in the variety choice analyses of the 1980s. In this paradigm, farmers’ decisions were motivated by their attitudes toward risk, and their choices made sense given that modern varieties often demonstrated higher average yields but greater yield variation. Farmers were not ‘inefficient’, but allocated land following principles consistent with those of the portfolio theory of investment (for example, Feder 1980; Just and Zilberman 1983). In the 1990s, economists argued that partial adoption could be attributed to any one of a number of competing explanations, including attitudes toward risk, the differential costs that farmers face when transacting in imperfect markets (de Janvry et al. 1991), or by agroecological heterogeneity such as soil type differences on farms (Bellon and Taylor 1993). It was recognized that traits other than yield affected the adoption decisions of farmers in semi-commercial agriculture (Adesina and Zinnah 1993; Renkow and Traxler 1994). In most cases, however, it may be difficult in an empirical study to identify which single explanation or combination of explanations produces the observed pattern of partial adoption (Smale et al. 1994). One of the major transaction cost approaches that is widely applicable to semi-commercial agriculture is that of the economic model of the farm household (de Janvry et al. 1991). In this approach, the farm household maximizes utility from a set of production outputs, choosing the amounts to consume on farm and sell in local markets. When all markets function well, only prices and production conditions on the farm determine their choices and the optimal choice is the one that maximizes profits. However, when markets are incomplete the effective input and output prices actually faced by farmers fall within a band defined at lower and higher extremes by the sellers’ and purchasers’ prices. These unobservable ‘shadow’ prices reflect market access and other transactions costs that vary according to the unique social and economic characteristics of each farm household. Since not all outputs are tradable, prices are said to be ‘endogenous’ rather than ‘exogenous’ (given, external) to the household. The farm household theoretical framework predicts that the characteristics of farm households shape their access to new technology, input and output markets, determining the shadow prices that govern their decisions, and through these prices, variety choices. Descriptive data from the Bara eco-site supports several of the competing explanations for partial adoption of modern varieties. First, the importance of agroecological heterogeneity is clearly demonstrated in the comparison between Tables 1 and 2. Second, a study of rice markets in the project sites suggests that grain markets for the landraces and their attributes are incomplete (Gauchan et al. 2001). Implicit in this statement is the recognition that, in part because of the heterogeneity of landraces, their local adaptation and cultural specificity, researchers have poorly understood them and inadequately documented their distinctive traits. Third, the information presented in Table 4 and Figure 2 is also consistent with the hypothesis that the characteristics of farm households affect the shadow prices that govern their variety choices. Table 4 relates household characteristics of rice farmers to their wealth rank. Farm size (total, irrigated and rice farm), access to irrigation, human capital status (literacy level), livestock holdings, and the number of months the household expects to be self-sufficient in rice are positively and significantly related to wealth rank. Higher caste households (Brahmin–Chhetri) were more likely to be classified in the richer and medium wealth groups, while lower caste households were more often poor. Medium caste households were fairly evenly distributed across wealth ranks. None of the lower caste households were classified as rich. Mean family sizes do not differ significantly by wealth rank of the household. Figure 2 shows that the simultaneous cultivation of different rice types is related to wealth status, as confirmed with a Chisquared test. A higher percentage (>60%) of households classified as rich or medium in wealth choose to grow both landraces and modern varieties simultaneously, while about two-thirds of poor households grow only modern varieties. This finding may appear counter-intuitive since the adoption of modern varieties is typically associated with access to cash and credit to purchase inputs, and these are often associated with wealth endowments such as farm size and access to irrigation (Feder and O’Mara 1981). One hypothesis would be that wealthier farm households are better able to absorb the additional labour and management costs associated with a more diverse and complex pattern of rice 6 Plant Genetic Resources Newsletter, 2003, No. 134 Table 4. Mean characteristics of farm households by wealth rank, Bara eco-site Farm household characteristic Wealth rank of farm household Family size (no. persons) Cultivated farm size (ha) Rice area per farm (ha) Irrigated rice area per farm (ha) Livestock holdings (no. large ruminants) Food sufficiency (months after harvest) % Access to irrigation (+) % Literacy status (+) % Upper caste (Brahmin Chhetri)(+) % Medium caste and ethnic groups % Lower disadvantaged castes (+) Rich (N=23) Medium (N=71) Poor (N=103) All (N=197) 6.8 2.239* 2.08* 1.40* 1.869 10.98 83 91 44 56 0 7.0 1.123** 0.989** 0.73** 1.929** 9.91** 69 55 48 46 6 5.94 0.437*** 0.385*** 0.27*** 0.88*** 5.16*** 58 21 23 58 19 6.45 0.89 0.80 0.61 1.38 7.34 65 42 34 54 12 Percent of farmers Note: pairwise t-tests show significant difference of means between wealth rank at the 0.05 level with two-tailed test: *Rich– Medium; **Medium–Poor; ***Rich–Poor. Chi-squared test shows significant difference in percentage distribution according to wealth rank, at 0.05 level (+). Source: computed from Baseline Survey Data, In Situ Project Nepal. 70 Growing only landraces 60 50 Growing only modern varieties 40 30 20 Growing both landraces and modern varieties 10 0 Rich Medium Poor All cultivation. In any case, the result has implications if the conservation strategy under consideration is to target particular households as conservators. Finally, research in the Bara eco-site indicates the importance that farmers place on certain agronomic and consumption attributes that are specific to particular landraces (Table 5). In summary, the descriptive analyses shown here signal the importance of: (1) agroecological heterogeneity; (2) market imperfections, and therefore household characteristics; and (3) nonyield variety attributes in predicting whether farmers will continue to grow rice landraces or not, and in which combinations. Figure 2. Choice of rice types by wealth status of farm household. (Source: computed from Baseline Survey Data, In Situ Project Nepal.) The relative private costs and benefits of growing rice landraces vs. modern varieties therefore vary among farmers, and we cannot summarize them accurately in a single partial budget. Conclusions This example shows us some of the insights we can gain from partial budget analysis. We see clearly that in favourable environments in Bara eco-site, when farmers have access to irrigation that mediates yield risk, to high-quality land and to markets, the opportunity cost of growing landraces instead of well-adapted Table 5. Farmers perception of major agronomic and consumption traits specific to each rice landrace, Bara eco-site Landrace Trait 1 Trait II Trait III Trait IV Trait V Mutmur (n=10) Adapted to poor, rainfed soils Adapted to poor, rainfed soils Religious value (Hindu culture) Early maturity Lower use of chemical fertilizers Lower use of chemical fertilizers Adapted to poor, rainfed soils Pest resistance Good fodder yield Good eating quality Higher price in informal market Nakhisaro (n=7) Sathi (n=4) Insect resistance Good eating quality n=Number of responding farm households Source: Field survey (1999). Good fodder yield Insect resistance Plant Genetic Resources Newsletter, 2003, No. 134 7 modern varieties is likely to be great. In particular niches such as poor upland soils or swampy land of Bara eco-site, certain landraces can be profitably grown and no modern varieties currently available in this location can compete with them. Our example also illustrates the dangers of using partial budgets to analyse the variety-choice decisions of farmers in semi-commercial agriculture—particularly when our goal is to link these decisions to strategies for on-farm conservation. Marginal analysis with partial budgets can be used to compare alternative agronomic treatments or technologies in terms of benefits net of costs that vary. In variety-choice comparison, partial budgets reveal little unless there are major differences in expected yields and prices received for the two varieties, or major differences in management costs. Moreover, partial budgets examine only a single activity, abstracting from other resource-use decisions on the farm. When the choice of variety is relating to large shifts in labour or land allocation (such as in a system with multiple crops), then whole farm analysis is more appropriate. CIMMYT (1988) outlined a number of extensions to the partial budget format, and reviewed the scope and limitations of marginal analysis. Alternatives that seem to be promising both agronomically and economically may have other drawbacks that only farmers are capable of identifying. Moreover, calculations are always based on estimates derived from experiments or farmer surveys, with accompanying measurement errors and variability. It is possible to incorporate yield variation or production risk into the analyses. Input and output prices also vary, and this might be addressed with sensitivity analysis in which results are recalculated using new parameters and incremental changes compared. But the limitation we have emphasized in this example relates to farmers’ objectives in semi-subsistence agriculture where the effective prices that govern their decisions are shadow prices that may be unique to each farm household. Survey data demonstrates that the choices made by farmers in this changing economic setting are complex. There are significant relationships, for example, between variety combinations and a number of household characteristics, suggesting that the net benefits farmers earn from growing landraces varies considerably among them. To test this hypothesis, however, we would need a more detailed data set and multivariate analysis. In study sites where on-farm conservation programmes are designed and implemented, baseline survey data that enable us to test such hypotheses, describe and monitor changes in farmers’ choices is essential. Table 6 summarizes the fundamental economics questions we see in the design, implementation and monitoring of on-farm conservation projects, with suggestions for appropriate analytical tools. The tools required to analyse the cost of on-farm conservation will include, but should not be limited to, marginal analysis based on partial budgets. Most of these are much more data intensive and time-consuming to employ, and therefore more costly. Such tools include multivariate econometric analyses based on farm household models and other valuation techniques, such as revealed and stated preference methods. Examples of econometric applications of farm household models to the study of farmers’ decisions to grow landraces include the analysis of potato in Peru (Brush et al. 1992), wheat in Turkey (Meng et al. 1998), the milpa system in Mexico (Van Dusen 2000), and maize in Mexico (Smale et al. 2001). Examples of the application of revealed and stated preference methods to estimate the marginal value of variety traits (which might also be appropriate for landraces) include hedonic analyses of rice in Asia (Unnevehr 1986), sorghum in India (Von Oppen and Rao 1981), and conjoint utility analysis of groundnuts in Niger (Baidu-Forson et al. 1997). While partial budget analysis and most of the methods used in these references focus on direct, private use value to farmers, methods such as contingent valuation or choice experiment methods (Hanley et al. 1998) offer the possibility of not only assessing the private use value but also private non-use values, as well as the social values associated with landraces. In the ideal case, findings obtained with one tool, such as a revealed preference Table 6. Economics questions in on-farm conservation and related tools Economics question Relevant approaches and tools • Which factors significantly affect the likelihood that farmers in a given site will continue to grow diverse crops? • Which farmers are most likely to continue to grow diverse crops? • What is the opportunity cost of growing diverse crops in these sites? • Partial budget analysis of data from farmers purposively selected to represent enterprise types • Whole farm analysis of data from farmers purposively selected to represent farm types • Descriptive and multivariate analysis of baseline and monitoring data, based on microeconomic theory of the farm household • What are the private use and non-use values farmers attribute to diverse crops? • To what extent to markets value the distinctive attributes of landraces? • What are the social values attributed to diverse crops? • Environmental valuation approaches, including revealed and stated preference methods, such as contingent valuation and choice experiments • Approaches from marketing literature, such as conjoint utility analysis • Hedonic approaches for estimating price variation associated with traits • What are the costs and benefits associated with proposed policies and other interventions intended to support on farm conservation • Cost-benefit analysis, based on a combination of the above approaches 8 Plant Genetic Resources Newsletter, 2003, No. 134 method, might be compared with those generated by the application of another method, such as a stated preference method. Acknowledgements This paper is a part of a study on economic valuation and onfarm conservation of rice landraces diversity in Nepal. The authors are grateful to Dr Pablo Eyzaguirre, Senior Scientist, IPGRI, for his guidance and assistance in designing the field study and granting financial support from IPGRI Policy Support Project. Special thanks are extended to Dr Bhuwon Sthapit, Scientist, IPGRI-APO for his guidance, review and logistic support. We also acknowledge the moral and logistic support of Dr MP Upadhayay, NPC, Nepal, and Dr Devra Jarvis, IPGRI-Rome, for this study. Mr P. Choudhary, and his local field team of Bara Eco-site including the National Multidisciplinary Group (NMDG) of Nepal In Situ Project deserve special thanks for making this study successful. We thank Ekin Birol, University College London, for her insights on application of environmental valuation methods. References Adesina AA, Zinnah MM. 1993. Technology characteristics, farmers’ perceptions and adoption decisions: a Tobit model application in Sierra Leone. Agricultural Economics 9:297–311. Altieri MA, Merrick LC. 1987. In situ conservation of crop genetic resource through maintenance of traditional farming systems. Economic Botany 41:81–96. Baidu-Forson J, Ntare BR, Waliyar F. 1997. Utilizing conjoint analysis to design modern crop varieties: empirical example for groundnut in Niger. Agricultural Economics 16:219–226. Barkley AP, Porter LA. 1996. The determinants of wheat variety selection in Kansas, 1974 to 1993. American Journal of Agricultural Economics 78:202–211. Bellon MR, Taylor JE. 1993. Folk soil taxonomy and the partial adoption of new seed varieties. Economic Development and Cultural Change 41:763–786. Bellon MR, Pham JL, Jackson MT. 1997. Genetic conservation: a role for rice farmers. Pp. 263–289 in Maxted N, Ford-Lloyd BV, Hawkes JG, eds. Plant Genetic Conservation: The In-situ Approach. Chapman and Hall, London, UK. Bretting PK, Duvick DN. 1997. Dynamic conservation of plant genetic resources. Advances in Agronomy 61:1–51. Brush SB. 1995. In situ conservation of landraces in centers of crop diversity. Crop Science 35(2):346–354. Brush SB, Taylor JE, Bellon MR. 1992. Biological diversity and technology adoption in Andean potato agriculture. Journal of Development Economics 38:365–387. Byerlee D, Akhtar MR, Hobbs P. 1987. Reconciling conflicts in sequential double cropping patterns through plant breeding: the example of cotton and wheat in Pakistan’s Punjab. Agricultural Systems 24:291–304. CIMMYT. 1998. From Agronomic Data to Farmer Recommendations: An Economics Training Manual. International Maize and Wheat Improvement Center (CIMMYT), Mexico, DF, Mexico. Feder G, O’Mara ??. 1981. Farm size and the diffusion of green revolution technology. Economic Development and Cultural Change 30:59–76. Frankel OH, Bennett E. 1970. Genetic Resources in Plants—their Exploration and Conservation. Blackwell Scientific, Oxford, UK. Frankel OH, Hawkes JG. 1975. Crop Genetic Resources for Today and Tomorrow. Cambridge University Press, Cambridge, UK. Gauchan D, Smale M, Chaudhary P. 2001. An analysis of market incentives for on farm conservation of rice landraces in Cen- View publication stats tral Terai, Nepal. IPGRI, Rome, Italy. [Unpublished manuscript] Hanley N, Wright RE, Adamowicz W. 1998. Using choice experiments to value the environment. Pp. 413–428 in Environmental and Resources Economics 11 (3–4). Kluwer Academic, Netherlands. Hiebert D. 1974. Risk, learning and the adoption of fertilizer responsive varieties. Amererican Journal of Agricultural Economics 56:764–768. de Janvry A, Fafchamps M, Sadoulet E. 1991. Peasant household behavior with missing markets: Some paradoxes explained. Economics Journal 101:1400–1417. Just RE, Zilberman D. 1983. Stochastic structure, farm size and technology adoption in developing agriculture. Oxford Economics Papers 35:28–37. Kislev Y, Shchori-Bachrach N. 1973. The process of an innovation cycle. Amererican Journal of Agricultural Economics 55:28– 37. Maxted N, Ford-Lloyd BV, Hawkes JG. 1997. Complementary conservation strategies. Plant Genetic Conservation: the In situ Approach. Chapman and Hall, London, UK. Meng ECH, Taylor JE, Brush SB. 1998. Implications for the conservation of wheat landraces in Turkey from a household model of varietal choice. Pp. 127–143 in Smale M, ed. Farmers, Gene Banks and Crop Breeding: Economic Analyses of Diversity in Wheat, Maize, and Rice. Kluwer Academic, Netheralnds and CIMMYT, Mexico. Rana RB, Chaudhari P, Gauchan D, Khatiwada SP, Sthapit BR, Subedi A, Upadhaya SP, Jarvis DI. 2000. In situ crop conservation. Findings of agro-ecological crop diversity and socioeconomic baseline survey of Kachorwa eco-site, Bara, Nepal. NP Working Paper NP. 1/2000. NARC/LIBIRD/IPGRI. Renkow M, Traxler G. 1994. Incomplete adoption of modern cereal varieties: the role of grain–fodder tradeoffs. Selected paper, American Agricultural Economics Association Annual Meetings, San Diego, August 7–10, 1994. Smale M, Just RE, Leathers HD. 1994. Land allocation in HYV adoption models: an investigation of alternative explanations. American Journal of Agricultural Economics 76:535– 546. Smale M, Bellon M, Aguirre Gomez JA. 2001. Maize diversity, variety attributes and farmers’ choices in Southeastern Guanajuato, Mexico. Economic Development and Cultural Change 50(1):201–225. Unnevehr LJ. 1986. Consumer demand for rice grain quality and returns to research for quality improvement in Southeast Asia. American Journal of Agricultural Economics. 68(3):634–641. Van Dusen E. 2000. In situ conservation of crop genetic resources in Mexican milpa system. PhD thesis. University of California, Davis, USA. Von Oppen M, Rao PP. 1981. A market-derived selection index for consumer preferences of evident and cryptic quality characteristics of sorghum. [AUTHOR: SOURCE/PUBLISHER?]