Potentials of green consumerism for landrace conservation: evidence from eggplant production sector...

Environment and Development Economics 15: 127–151 C© Cambridge University Press 2010doi:10.1017/S1355770X09990155First published online 18 January 2010

Assessing the potential of labelling schemes forin situ landrace conservation: an example fromIndia

VIJESH V. KRISHNAInternational Maize and Wheat Improvement Centre (CIMMYT), CG CentreBlock, NASC Complex, DP Shastri Marg, New Delhi 110012, India. Tel:+91 (0)11 25842940. Fax: +91 (0)11 25842938. Email: [email protected]

UNAI PASCUALDepartment of Land Economy, University of Cambridge, 19 Silver Street,CB39EP, Cambridge, UK. Email: [email protected]

DAVID ZILBERMANDepartment of Agricultural and Resource Economics & GianniniFoundation, University of California, Berkeley, CA 94720-3310, USA.Email: [email protected]

Submitted January 23, 2009; revised July 23, 2009; accepted October 25, 2009

ABSTRACT. This paper addresses the potential of in situ crop landrace conservation,employing market-based instruments, which pre-requires that (1) consumers holdpositive use-value for the landrace attribute and (2) their willingness to pay covers both thetransaction cost of implementing these instruments and the opportunity cost of landracecultivation. The empirical examination is based on two closely related analyses of eggplantproduction and consumption sectors of India. At present, the vegetable markets of southIndia provide the landrace cultivators with a price premium adequate enough to coverthe opportunity cost of not opting for high-yielding modern varieties. However, we detectan underutilized consumer demand for landrace products. The wide margin that existsbetween the price premium farmers currently obtain for the landrace attribute and whatconsumers are willing to pay for it is indicative of the unexploited potential of labellingand certification schemes as an emerging agrobiodiversity conservation strategy.

1. IntroductionThis paper examines the potential of market-based instruments inconserving landrace varieties in situ, when they compete for acreage with

The financial support by the United States Agency for International Development,the Eiselen Foundation (Ulm) and the SV Ciriacy-Wantrup Foundation aregratefully acknowledged. We are also indebted to the three anonymous referees,Matin Qaim and Melinda Smale for comments and inputs to earlier drafts of thispaper. Any remaining errors are ours. The usual disclaimer applies.

128 Vijesh V. Krishna et al.

higher-yielding modern varieties (MVs). There exists an active debatesurrounding the relationship between adoption of MVs and erosion ofcrop varietal diversity. The conventional ‘genetic erosion hypothesis’ –that is, the adoption of few genetically uniform high-yielding varietieserodes landraces and eliminates the diversity of indigenous crop varieties,as proposed by Harlan (1975) – has obtained significant support in theliterature (e.g., Fowler and Mooney, 1990; Berg et al., 1991; Pretty, 1995;Food and Agriculture Organization [FAO], 1996; Brush, 1999). However, theassociation of MVs with the diversity erosion is rather indirect and complex.According to Smale (1997), the impact of diffusion of new varieties ongenetic diversity would depend on the chances of loss of genes from the cropspecies, which in turn would be determined by the nature of coexistencebetween MVs and landraces. The existence of ‘creolized varieties’ – theproducts of hybridization between landraces and MVs, either by design orby accident (Bellon et al., 2006) – further complicates the aforementionedhypothesis. Keeping these findings in view, the current paper proceedsupon the premise that MV introduction detrimentally impacts the acreageunder landraces (and we do not address its effect on the crop geneticdiversity status in toto) and attempts to examine the prospects of monetizingthe consumption preferences for their in situ conservation.

Landraces, which are often known as traditional or local varieties,constitute an important component of crop genetic resources and are rawmaterials for plant breeding because of their high capacity to tolerate bioticand abiotic stresses (Bellon, 2004).1 The conservation of landraces involvestwo main approaches, ex situ (off-site) and in situ (on-site), which are nowrecognized as complimentary (Brush, 1999: 3–28). For crop species, in situconservation refers to the sustained cultivation of crop populations on farm,where they have evolved under natural and farmer selection processes(Bellon et al., 1997). One of the main arguments in favour of this conservationapproach is that in contrast with the ex situ approach, it is dynamic withcrops adapting to changing environmental conditions. Furthermore, alongwith the genetic raw materials of alleles and genotypes, the decentralizedmanagement skills and traditional knowledge associated with crop geneticresources are also conserved in a co-evolutionary fashion with in situconservation of landraces (Brown, 1999; Brush, 2004: 196–206; Poudel andJohnsen, 2009). Despite these benefits, events of deliberately planned (e.g.,through public policy) in situ conservation efforts are uncommon (Fowlerand Hodgkin, 2004).

In areas where landraces are better adapted to the local agro-climaticconditions over MVs, cultivating them would generally entail relativelylower per-unit costs (Smale et al., 2004). In addition, the existence ofstrong non-market (viz., cultural) traits and values for the landracecultivator households provides further incentives to their de facto on-farm

1 In the present paper, we follow the simple definition of landraces proposed byBirol and colleagues (2008) that they are ‘variants, varieties, or crop populationscomprised of plants that are often highly variable in appearance, but whose geneticstructure has been shaped by farmers’ seed selection practices and management,as well as by natural selection processes, over generations of cultivation’.

Environment and Development Economics 129

conservation. Even in the absence of such incentives, market segmentationfor landrace products with a significant price increment may also facilitatetheir conservation (Gouchan et al., 2005). As landraces form an integralpart of crop genetic diversity, and the sustainable cultivation of landraces isone of the in situ conservation approaches, landrace-derived food productscould be eco-certified to provide biodiversity-conservation-related benefits,which may include significant non-use and option values. Markets forsuch specialized food products and eco-certification programmes areincreasingly popular, owing largely to favourable conditions because ofeconomic growth (Hamilton and Zilberman, 2006).

Much work has been done regarding the valuation and identification ofconsumer preferences for agricultural products with specific attributes suchas genetically modified (GM) foods (e.g., Kontoleon and Yabe, 2006; Krishnaand Qaim, 2008a), environment friendly practices (e.g., Moon et al., 2002;Larson, 2003) and region of origin (e.g., Scarpa and Del Giudice, 2004; Scarpaet al., 2005) in food production. While there has been recent efforts to betterunderstand the incentives for management of on-farm crop diversity byaddressing farmers’ perceptions and choices regarding varietal traits (e.g.,Van Dusen and Taylor, 2005; Birol et al., 2006, 2008), the literature tendsto neglect the crucial role of non-farmer buyer preferences in developingcountries, especially in the informal market chains. Bridging this knowledgegap would help design and support incipient ‘market friction-reducinginstruments’ (MFRIs), linked to local food markets, that in turn may providefarmers the right incentives for in situ conservation of landraces.

The objective of the present paper is to assess the potential of MFRIs,such as labelling and certification, in assisting the landrace productmarkets of developing countries emerge in the aforementioned context.Amongst the different types of market friction reductions, this paper dealswith market creation for inputs/outputs associated with environmentalquality, alongside information programmes that work through exogenouseconomic factors that are pivotal in determining consumer preferences.Conventionally, the potential to fully exploit niche markets has often beenconsidered as limited to the wealthy nations (Grote, 2002; Basu et al., 2003),where compartmentalization of consumer preferences with labels is ofteninstituted by changes in food policy.2 Against this widespread perception,we reassess the potential consumer demand for landraces in the emergentdeveloping economy of India.

We take eggplant production and consumption sectors of India for theempirical analysis. At present, food product differentiation in India mainlyoccurs without formal certification and labelling schemes. However, in thecase of a few crops such as eggplant, there exist certain distinguishable

2 For example, the European Union’s recognition of territorial specificity throughregulations 2081/92 and 2082/92, which led to the origin of denomination labels(protected denomination of origin and protected geographic indication), aretypical examples for such transformation in the developed world (Scarpa and DelGiudice, 2004; Scarpa et al., 2005). With respect to products similar to landraces,the recent regulation EC 509/06 (traditional speciality guaranteed) label is alsoequally noteworthy.


phenotypical characteristics indicative of landrace origin. In such cases, theotherwise credence and/or experience attribute of landrace origin turnsto be a search attribute for consumers. Across the vegetable markets ofIndia, the landrace products are informally (without labels) sold underthe category of ‘desi’, which can be translated as indigenous, folk or localvariety. A detailed examination of the market price structure of such cropsis expected to reveal the potential for a more formal and wider provision ofinformation (with labelling and certification schemes) regarding landraceorigin or form of landrace cultivation/conservation. Here, the case ofeggplant in India is used on the basis of two complementary analyses. Thefirst deals with the supply side (farm households) in order to estimate themarket (use) value currently harnessed because of the landrace attribute ofcultivated eggplants. The second one addresses the demand side (consumerhouseholds) to estimate the full consumption value of landraces.

The rationale behind selecting an eggplant production system in Indiais elaborated in the next section, while the testable hypotheses and theanalytical framework are presented in section 3. Section 4 describes theprimary data sets used in the empirical analysis. The main results of theempirical analysis are provided and discussed in section 5. The last sectionconcludes and addresses the main policy implications.

2. The context: landrace versus modern variety eggplants in IndiaIndia has made significant progress in recent years towards settingup a legal regime for the management of its plant genetic resources(Biber-Klemm et al., 2005). At the same time, agricultural developmentpolicies are increasingly focused on the development and disseminationof high-yielding MVs. While the exact impact of these MV disseminationprogrammes on genetic diversity is yet unknown, they limit the acreageunder landraces because of competition for land. This situation exemplifiesthe agricultural development versus resource conservation politicaldichotomy or clash of interests (e.g., agribusiness versus agro-ecology).This is the case for the eggplant crop, of which India is the second largestproducer in the world (FAOSTAT, 2007).

Traditionally, farmers have maintained and supplied the eggplant seeds,resulting in special varieties adapted to the region’s environment as well aslocal consumers’ tastes.3 In addition, as a food crop, a number of wild croprelatives of eggplant are used in indigenous medicine systems (Daunay et al.,2000).4 Furthermore, eggplant landraces and their wild relatives have beendocumented as having a significant option value because of their resistanceagainst pests and diseases (Sridhar et al., 2001). Lastly, some landraces have

3 The three most common cultivated eggplant varieties in India are Solanummelongena var. esculentum (round/egg-shaped fruits), S. m var. serpentinum (long,slender fruits) and S. m var. depressum (dwarf plants) (Bose et al., 2002).

4 Here, the in situ conservation approach gains special relevance, as the humanselection and management activities involved are multidimensional. Indigenousknowledge associated with not only the cultivation but also the utilization (in thetraditional medicine system of India, viz., Ayurveda) of diverse landrace varietieshelps maintain the cultural heritage based on Ayurveda in India.


been associated with non-use values; for example, the Matti gulla variety ofeggplant grown in villages of Karnataka is associated with religious beliefs.In the latter case, such non-use values provide an edge to their conservationoutside the market realm. However, given the increased dominant role offood markets in the informal agricultural sector, relying on cultural or non-use values for landrace conservation may be considered a weak strategy.Hence, we focus on the narrower, but significant, consumptive value oflandraces, as they can be harnessed through the more dynamic marketforces.

It is not surprising to find that consumer preference for eggplant fruits areexpressed according to the characteristics such as taste, colour, size, spinycalyx and shape. Such preferences become complex to analyze becauseof the large combination of the fruit’s characteristics that in turn hashistorically led to the cultivation of eggplant varieties with very diversephenotypes. Furthermore, in the face of such diversity, there has also beena significant adoption of eggplant hybrids in India. Since the 1980s, anincreasing number of the F1 eggplant hybrid varieties bred by private seedcompanies are being commercialized. More specifically, hybrids are beingwidely cultivated in the southern states of Karnataka, Andhra Pradesh andMaharashtra. By contrast, in eastern parts of India (especially the states ofWest Bengal and Orissa, which together account for around 50 per cent ofthe total eggplant area in the country) the adoption of hybrids is marginal,probably because landraces are more adapted to the local soil and climaticconditions (Krishna and Qaim, 2007, 2008b).

Apart from the conventional breeding sector, modern biotechnologyposes new challenges to on-farm landrace conservation, alongsideproviding opportunities to realize high levels of yield and reduced levelof insecticide use. Recognizing the economic relevance of resistant breedsin eggplant, GM hybrids and open-pollinated varieties (OPVs) are currentlydeveloped under a unique public–private sector research collaborationin India (Krishna and Qaim, 2007, 2008b). Although recent studies haveindicated (ex ante) that the Indian consumers have a generally positiveattitude towards GM foods (Anand et al., 2007; Krishna and Qaim, 2008a),there is a possibility that once the GM eggplant comes to be marketed inIndia, consumers’ knowledge and perception may alter significantly. Thepresent study gains significance in this context by providing an insightinto the potential welfare changes that consumers’ perception towards aproduction technology may bring.

3. Framework and hypothesesThe conceptual framework of the paper proceeds with a set of workinghypotheses that connects the supply and demand sides of the landraceattribute in crop production. In the literature, there are a number ofstudies that consider the welfare impacts of quality changes associatedwith technology adoption (cf. Alston et al., 1998: 243–246). However, mostof them only deal with the product quality enhancement associated withtechnology adoption. The present paper, in contrast, deals with a declinein consumption utility with adoption, which in turn helps provide a set of


hypotheses regarding the potential of on-farm landrace conservation usingmarket-based instruments.

The supply side of the problem can be characterized as follows: Farmeri chooses the privately optimal land allocation between landraces (r) andMVs (m) for a given crop (e.g., eggplant). Assuming the simple two-varietycase and following the technical explanation of joint production in thecontext of varietal choice (Smale et al., 1994), the assumptions of short-run fixity of cultivated land and diseconomies of scale are adopted. Giventhe stochastic yields of the landrace and MVs, farmer i chooses the levelof inputs to apply into growing variety j ( j = m, v) that maximizes herexpected profit. The associated production function is yi j = f (xi j ), whereyi j and xi j are the output and input (including land under cultivation)vectors respectively.

When production costs across the different varieties are undifferentiated,the relative performance of MV is given by the difference between themarginal value products (MVPi j ) with respect to xi j of modern and landracevarieties, i.e., (pim.

∂yi∂xim

− pir .∂yi∂xir

), where pi j stands for unit output priceobtained by farmer i on variety j. Assuming that the cost functions ofgrowing modern and landrace varieties are linear and of equal slope, themarginal cost (MC) of cultivating the MV is given by MCim = MCir +μim, where μim is the incremental factor cost of cultivating MVs. Theland allocation decision between varieties depends on the direction andmagnitude of the opportunity cost of cultivating a given variety. Theopportunity cost of landrace cultivation when MVs are available is givenby

OCir = (MVPim − MCim) − (MVPir − MCir ) . (1)

The MVP j can be estimated by employing a production function approach,keeping the farm and household characteristics at the corresponding meanlevels. The farmer would cultivate the landraces only if OCir ≤ 0 andMVPir ≥ MCir . Under a uniform output price ( p), OCir = p.( ∂yi

∂xim− ∂yi

∂xir) −

μim. Whether there exists a positive opportunity cost of landrace cultivationis an empirical matter, that is, whether p.( ∂yi

∂xim− ∂yi

∂xir) ≥ μim. We test this

hypothesis assuming μim = 0 and estimating a production function withinteraction terms (between hybrid/MV technology and xi ). The nullhypothesis thus becomes ∂yim

∂xim/

∂yir∂xir

≥ 1.When pim �= pir , OCir = (pim.

∂yi∂xim

− pir .∂yi∂xir

) − μim and a price premiumfor farmers for cultivating landraces, equivalent to ∂yim

∂xim/

∂yir∂xir

in percentageterms when μim = 0, may balance out any productivity disadvantage oflandraces against MVs.

In the case of perfect information about landrace attributes of food crops,a positive consumer preference for a landrace attribute would facilitate insitu landrace conservation (provided the price premium is channelled backto the farmers to increase production of landrace-based food products).However, the use of a price premium for landrace products requiresa priori product differentiation in the market. This is possible only ifthe landrace origin is expressed as a search attribute – either throughphenotypic characters or through labelling and certification. This is because


the price increment obtained for the landrace products may not be due tothe ‘landrace status’ alone, but also to those product attributes that arecommonly associated with landrace products. In other words, the marketprice of a food product is a function of a bundle of its characteristics,z = {z1, · · · , zC }, where zc is the characteristic c of the good and C is thenumber of characteristics (e.g., colour, taste, shape, origin and landracestatus).5 This implies that even when pir

pim≥ 1, the marginal impact of the

landrace status on the market price could be zero, and the price advantagemay be due to the associated phenotypical characteristics (e.g., colour).In this case, plant breeding techniques that can develop MVs with theseattributes can seriously challenge the on-farm conservation of landraces.

In order to impute the marginal value that farmers realize for thelandrace attribute of the crops they grow, a hedonic modelling approachcan be applied. Since the seminal work of Waugh (1928), a number ofstudies have employed hedonic pricing to examine the price structure ofdifferent agricultural products (e.g., Unnevehr, 1986; Parker and Zilberman,1993; Dalton, 2004; Edmeades, 2007; Amegbeto et al., 2008). However,often because of data limitations, hedonic functions do not includefarm–household variables (Hi ) that are often important in developing-country contexts. Such household-level characteristics may be seen asproxies for the outcome of the bargaining process between buyers(consumers/retailers/middlemen) and sellers (farmers) for prices. In ourcase, the hedonic price function is given by

pi j = h(zi j , θ

); zi j = {

G j , Hi , Di}

, (2)

where θ is the vector of parameters describing the shape of the hedonicfunction and Gj and Di are the vectors of characteristics of the food productand the regional and seasonal factors, respectively. The major estimationissue associated with h(.) is that its functional form is unknown a priori(Halstead et al., 1997). Following Haab and McConnell (2002), this is rectifiedby employing the Box–Cox transformation of the dependent variable. Thisallows estimating the marginal implicit price (MIP) or partial derivativeof the explanatory (dummy) variable associated with whether the foodproduct derives from the use of MV. If its sign is negative, it would indicatea price advantage obtained by farmers for the landrace attribute.

However, without a labelling or certification scheme in place for thelandrace attribute, farmers may only obtain a fraction of the consumers’willingness to pay (WTP) for the landrace attribute, i.e., pir

pim= η. p, where p

is WTP for landraces and η indicates the share of such WTP that is effectivelytransmitted to farmers (0 ≤ η ≤ 1). Unless η = 0, consumers’ positive WTPfor landraces ( p ≥ 0) could in principle lead to a price premium for landraceproducts, ( pir

pim≥ 1) obtained by farmers, thus providing an economic

incentive to cultivate landraces instead of their MV counterparts. Howeverthe WTP value and the realized price premium need not coincide. Wehypothesize that generally η < 1, indicating that consumers’ WTP for the

5 According to the theory of Lancaster (1966) of the demand for ‘characteristics’or intrinsic quality features and consumption is an activity in which goods andservices are inputs.


landrace product is greater than the actual price premium obtained byfarmers.6 Further, it is hypothesized that consumers’ WTP is positivelyaffected by the level of information (I) available to them regarding thelandrace characteristics of the product: ∂ p

∂ I ≥ 0. If this hypothesis is rejected,then the use of labelling and certification would be ineffective as anincentive for farmers to effectively increase the acreage under landraces.

Consumer’s WTP for the landrace product can be estimated using statedpreference valuation approaches, such as contingent valuation (CV) undera random utility framework. A major criticism of WTP estimation usingCV is that the estimates may be ‘biased upwards due to the hypotheticalnature of the payment commitment’ (Bateman et al., 2002: 439). However,the hypothetical bias is much lower when survey respondents make choiceson marketable goods familiar to them (as in our case, i.e., local landracevegetables) than when they are confronted with hypothetical, non-marketand non-local goods.

At first, consumer households were asked to state their attitude towardsproducts of landrace versus MVs at uniform prices, and only those thatstrictly preferred landraces were requested to state whether they wouldbe willing to purchase the landrace product at different price premiums.If the hypothetical purchasing decision is estimated independent ofthe preference (towards landraces) decision, estimates may suffer fromselectivity bias. This problem can be circumvented by estimating these twodecisions jointly, using a bivariate probit model with partial observabilityas specified by Meng and Schmidt (1985):

d∗k1 = γ ′

11 Hk + ek1, dk1 = 1 if d∗k1 > 0,

dk1 = 0 otherwise, (3a)

d∗k2 = γ ′

21 Hk + γ ′22 Bk + ek2, dk2 = 1 if d∗

k2 > 0 and d∗k1 > 0,

dk2 = 0 otherwise, (3b)

where d∗k1 and d∗

k2 are latent variables representing preference and purchasedecisions of consumer household k regarding a landrace product; Hk isthe vector of the household and regional factors (including informationalvariables); Bk stands for the bid value against which the willingness topurchase is elicited; γ11, γ21 and γ22 are the vectors of coefficients to beestimated; and ek1 and ek2 are the error terms with variance equal to one,(ek1, ek2) ∼ bivariate normal [0, 0, 1, 1, ρ], whereρ = Cov(ek1, ek2).

4. The dataThe data on eggplant cultivation in India were collected in 2005 from a crosssection of 240 farm households from Andhra Pradesh and Karnataka, two

6 It is only realistic to assume that when consumer preferences are strong enough,the system would gradually evolve, developing suitable market institutions, inorder to permit the percolation of these preferences. However, we do not examineinstitutional ways of affecting η.


leading eggplant-producing states in south India.7 The survey covered themajor eggplant-growing tracts within the selected states, most of which arelocated in the river belts. Using a stratified random sample, 6 districts and 13taluks (revenue subdivisions within each district) were selected purposivelyon the basis of the area under eggplant cultivation. Villages and farmhouseholds were selected randomly. The farm economic data were gatheredfrom these households, which included yields, variable production costsand farm prices for eggplant fruits. Information about attributes of themarketed eggplant fruits was also gathered from each farmer, includingthe skin colour and presence of spines in fruit calyx, amongst others. Suchinformation is complemented with the available data on socio-economiccharacteristics of farm households through structured surveys.

The average land holding of the sampled farms is 4.96 acres, and theaverage size of eggplant plot is 0.81 acres. Of the sampled households, 64per cent adopt F1 hybrids (marketed by more than 10 private companies butmainly by Maharashtra Hybrid Seed Company or MAHYCO and AnkurSeeds), and the rest cultivate popular landraces or unnamed varieties.Eggplant growers use either seeds or purchased seedlings as plantingmaterial (71 per cent of the hybrid adopters and 98 per cent of OPVgrowers use seeds for raising the crop, and the rest buy seedlings).8 Sixmajor eggplant landraces were recognized in the study area: Telupulu,Javeri, Rangapatanam, Sarpavaram, Desavalli and Singampalli. The unnamedvarieties are also being cultivated in the region for a long period of timeand show high local genetic adaptation. Farmers consider all of them asdesi varieties and value them on par with the aforementioned landraces.Furthermore, it should be pointed out that because of their close proximityto perennial rivers, the surveyed eggplant farmers do not face majorproblems of water shortages. Regarding the socio-economic characteristicsof farm households, the mean annual per-capita income is around 20,000rupees (US$417) with respondents having limited formal education (theaverage being less than five years of schooling).

Now we turn the focus towards the consumption–demand side ofeggplants. Both hybrid and landrace fruits are widely used in Indian dishes(e.g., Bharta curry) and rice (e.g., Vangi bath) preparations, and as a vegetableits popularity is consistent across different regions of India. In order togauge the consumption value for the landrace attribute, data from vegetable

7 Although the survey also covered the state of West Bengal, we have excluded thisregion from the present analysis. The adoption of MVs is marginal in this state,making a study on the MV–landrace interactions in product market not feasible.

8 Average quantity of hybrid seed used is 159 g/acre which costs Rs 1096. Onthe other hand, 285 g of OPV seeds were used to cultivate an acre, with anaverage cost of Rs 109. Many farmers use the farm-saved OPV seeds, makingit low cost in comparison. The marginal cost of hybrid adoption is Rs 987. Averagecost of hybrid seedlings is Rs 1969/acre, whereas the cost was Rs 2333/acre forlandrace (Sarpavaram) seedlings. Although the higher price of OPV seedlings maybe cited as the potential for seed market development for landraces, we refrainfrom generalization because of the very small sample size of landrace seedlingadopters.


consumers were also collected during 2006 from five important urbanlocations in India: New Delhi, Bangalore, Kolkata, Kolar and Barddhaman.The first three are amongst the largest cities of India and are administratedby municipal corporations. Kolar and Barddhaman are two districtheadquarters in the states of Karnataka and West Bengal respectively,which are in close proximity to important eggplant production regions.A stratified random sample design was followed for data collection. Eachlocation was first divided into zones. Corporation wards and consumerhouseholds were selected randomly from each of these urban zones. In total,the sample consisted of 645 households from 61 corporation wards, and thepresent study employs a subset of 629 observations avoiding householdsreporting zero consumption of eggplant fruits. No significant difference insocio-economic characteristics was observed between consumer and non-consumer households. In comparison with the farmer survey, consumerrespondents showed higher levels of education and income, with an averageof about 10 years of formal schooling and an average of about Rs 30,000(US$625) as annual per-capita income.

The survey was designed to gather information about consumers’preferences and attitudes towards different eggplant characteristics,including those of landrace fruits. In a second stage, for those individualswho indicated a clear preference for landraces over hybrids, a dichotomous-choice question on their willingness to purchase landrace products wasposed against hypothetical price increments, in order to estimate theconsumption (use) value of eggplant landraces.

5. Results and discussion

5.1. Supply side and price analysisThis subsection includes two complementary analyses. One deals withtesting whether hybrid varieties of eggplants in India are more productivethan farmers’ varieties given the inputs used in their cultivation. The otheraddresses the question of whether and to what extent there is an outputprice difference between hybrids and landrace eggplants.

Productivity of eggplants (hybrid vs landraces)Table 1 provides the basic information regarding the economics of eggplant(landraces versus hybrids/MVs) cultivation in south India. No significantdifference is observed in the total variable cost of cultivation of landracesagainst hybrids/MVs. This confronts the conventional wisdom that theper-acre variable cost of landrace cultivation is lower.9 Unsurprisingly,eggplant hybrids show a marked superiority over landraces regardingyields. The average marketed yield is 95 versus 112 quintals (Q) per acre

9 However, the cost structure varies. The main difference in the cost structure ofhybrids and landraces in the study area is that the latter is mostly cultivated in theleased-in land and is therefore associated with a higher rent. We moreover couldn’tfind any difference between eggplant hybrid and landrace plots with respect tosoil quality and irrigation facilities. In the case of hybrid production, however, thecost of material inputs (especially chemical fertilizers) was comparatively high.

Environm

entandD

evelopmentE

conomics

137

Table 1. Comparative economics of landrace and hybrid eggplant production.

Landraces (N = 86) Hybrids (N = 154) Difference of hybrids over landraces

Mean SD Mean SD Percentage Hypothesized sign p value

Variable cost (thousand Rs/acre) 18.98 12.43 17.29 7.38 −9.77 + .91Yield (Q/acre) 95.28 58.33 111.97 78.21 14.92 + .04Per unit cost of production (Rs/Q) 199.13 330.81 154.44 125.12 −28.94 − .06At constant market price:Net revenue (thousand Rs/acre) 23.18 26.99 32.24 32.70 28.10 + .01Market price (Rs/Q) 501.25 151.03 383.47 147.39 −30.71 − .00At differential market prices:Net revenue (thousand Rs/acre) 28.79 33.94 25.65 31.75 −12.24 +/− .47

Note: Level of significance in one-tailed t-test if the hypothesized sign of difference is either + or − and in two-tailed t-test ifit is +/−; Rs 48 = US$1 (on July 2009) and 1 Q = 100 kg.


for landraces and MVs respectively (farm households’ consumption ofeggplant is negligible). Owing to this yield superiority, the per-unit costof hybrid cultivation is lower by about 29 per cent. Moreover, at uniformoutput prices, hybrid eggplants generate significantly higher net returns tofarmers.

For empirical estimation of productivity impact of hybrid adoption, ageneralized Cobb–Douglas production function with interaction terms isestimated:10

y = α +∑

i

βi · xi + γ · H +∑

j≤i

δ j · xj · H +∑

k

φk · S +∑

m

ωm · D + u,

(4)where y is the eggplant yield transformed into its natural logarithm; xiis the quantity of ith input also transformed to its natural logarithm; His the dummy variable which takes the value of 1 for hybrid plots and iszero otherwise; S is the seasonal and state dummies; D stands for a set ofhousehold characteristics; β, γ , δ, φ and ω are the vectors of coefficients tobe estimated; and u is the error term.

Because of the small plot size, very few eggplant farmers cultivatemore than one variety, and none of them cultivates both hybrids andlandraces for the output market. Despite this, a comparison of hybridand landrace production systems is valid because (1) these systems arepositioned adjacent to each other geographically in the study area, and nodifference in the soil and climatic factors between hybrid and variety plotswas reported by the sampled farmers during the survey, and (2) althoughdifferences in the farmer household characteristics exist, they are beingcontrolled in the model estimation.

In order to avoid endogeneity bias, both pesticide application andhybrid adoption are instrumented, with district-level average pesticideprice (Rs/kg) and information-related variables (dummy variables showingmass media and government extension exposure) respectively. Theseinstrumental variables were proved relevant in the input-use and adoptionmodels but do not influence the per-acre yields, indicating that theproduction function is properly identified. Predicted insecticide amountsand probability of hybrid adoption were included in the productionfunction. The model is run with and without the interaction terms betweenthe instrumented hybrid adoption and other input variables. The interactionterms with fertilizer and plant protection chemicals (PPCS) are statisticallyinsignificant, indicating that the ratio ∂yim

∂xim/

∂yir∂xir

= 1 for these two inputscannot be rejected. Hence, these variables were excluded from the modelestimation, and the final results are shown in table 2.

The difference between models I and II is the additional hybrid–labourinteraction term, which nevertheless provided substantial insights intothe role of hybrid technology adoption in farm productivity. In model I,

10 This functional form was selected as the best fit for the production data, aftertesting different other functions (viz., quadratic, translog). Similar functionalforms are widely used to estimate the farm-level productivity analyses (e.g., Qaim,2003; Matuschke et al., 2007).

Environm

entandD

evelopmentE

conomics

139

Table 2. Production function estimates of eggplant cultivation in south India.

Model I Model II

Variable Description Coef. SE p -value Coef. SE p value

Fertilizer Logarithm of fertilizer amount in kg/acre 0.222 0.120 .07 0.208 0.118 .08Labour Logarithm of human labour days per acre 0.266 0.116 .02 0.819 0.208 .00PPC Logarithm of cost of PPCs per acrea 0.112 0.288 .70 0.093 0.283 .74Hybrid eggplant Adoption dummya 0.139 0.399 .73 5.005 1.580 .00Experience Farmer’s experience in eggplant cultivation

in years0.003 0.006 .63 0.003 0.006 .57

Education Schooling obtained by farmer in years 0.035 0.010 .00 0.034 0.010 .00Rabi Season dummy (versus kharif) 0.301 0.145 .04 0.252 0.143 .08Summer Season dummy (versus kharif) 0.349 0.164 .03 0.309 0.161 .06Andhra Pradesh State dummy (versus Karnataka) −0.106 0.273 .70 −0.077 0.268 .78Hybrid–Labour Interaction term −0.918 0.289 .00Model intercept 1.098 0.784 .16 −1.713 1.172 .15

Model statistics F value F(9, 230) = 5.25 F(10, 229) = 4.96

Prob > F 0.00 0.00Adjusted R2 0.14 0.17

Note: Dependent variable in both models is logarithm of marketable yield of eggplant (Q/acre); N = 240.aInstrumented variables.


the coefficient of hybrid adoption is largely insignificant, indicating thatceteris paribus, hybrid technology does not increase eggplant yield. Theinsignificance of hybrid–fertilizer interaction coefficient, which led to theexclusion of this term from the final model estimation, was interesting, ashybrid cultivation is fertilizer intensive with 4.98 Q/acre against the 3.95Q/acre for the landraces. In other words, the genotypic effect of hybridsallows higher level of fertilizer application and helps the farmers reap theassociated yield benefits.

Also noteworthy is the positive coefficient of education; hybrid adoptersare better educated (6.6 years of schooling) compared with the non-adopters(4.5 years). This difference also contributes to the mean yield differencebetween hybrids and landraces as indicated in table 1. Nevertheless, noneof these findings is as intuitive as in the case of hybrid technology interactionwith human labour in eggplant production process.

The introduction of the interaction term resulted in largely positive andsignificant coefficient for hybrid adoption (cf. model II in table 2). At thesame time, the interaction term of hybrid adoption with human labour isnegative, making ∂yim

∂xim/

∂yir∂xir

< 0 for labour input. At first glance, hybrids seemto be less responsive to labour inputs than their landrace counterparts.11

This observation is also associated with the zero difference across hybridand landrace production with respect to mean number of labour days usedin eggplant production. Also there is no significant difference with respect tothe number of hired/family or male/female labour associated with hybridadoption.

However, there seems to be an important difference with respect to labouruse between landraces and hybrids with regard to two activities – manualweeding and harvesting. The labour used for weeding is significantly highin landrace cultivation, compared with the hybrids (79 versus 38 workingdays, respectively).12 However, owing largely to the higher yield, hybridfarmers employ more labour for harvesting activities (additional 44 daysover landrace farmers). Nevertheless, between these two activities weedingis the one that enhances crop productivity most, and hence the dearthof labour used for weeding in hybrid cultivation results in the estimatednegative hybrid–labour interaction term. One of the reasons behind thislabour allocation disparity could be the budget constraints. Comparedwith the landrace cultivator, the hybrid-based farmer needs to invest morein fertilizers and harvesting; thus she is bestowed with less money tospend on weeding. All in all, the current labour allocation prevents farmersfrom reaping the genotypic yield effect associated with hybrid technologyadoption.

11 A similar finding was reported by Ramasamy and colleagues (2003) for the caseof hybrid rice cultivation in south India. Furthermore, taking the case of hybridwheat in central India, Matuschke and colleagues (2007) found no impact of hybridseed adoption on the productivity of human labour.

12 In both hybrid and landrace production sites, weeding is mostly done manually,and chemical management of weeds is highly sporadic.


Implicit prices of eggplants (hybrids versus landraces)Farm price obtained by the landrace cultivators (Rs 501/Q) is around31 per cent higher in comparison with that associated with hybrids(Rs 383/Q). There exists no significant output price difference betweenidentified landraces and unnamed varieties. This higher price of landracescircumvents their low mean yield that arises because of farmer and input-use differences. This ultimately leads to insignificant marginal revenuedifference between hybrids and landraces (cf. table 1). However, thesignificant question is whether the price differential is due to the landracestatus itself or to the associated product characteristics (e.g., skin colour),which is tested by means of a hedonic price function. If the price differenceis due to the associated product characteristics, the hybrids bred to expressthem would significantly reduce the price increment for landraces.

The price function estimation is carried out in two steps: first includingonly the product, regional and seasonal factors (G j , Di ) alone andlater adding the household characteristics (Hi ) to the model, as it isunconventional to include the latter in hedonic pricing estimation. Theresults are presented in table 3. We adopt the Box–Cox transformationfunction, and the MIPs are computed (cf. table 3). Six farm householdcharacteristics (viz., years of experience of the head of the household ineggplant farming, mass media exposure and contact with the governmentextension agencies, formal education status of household head, time takento reach market with the produce and size of farm owned) are included inthe model.13

The analysis, without the inclusion of household characteristics, revealsthat the landrace status is associated with an MIP of Rs 65/Q, which is55 per cent of the current absolute price difference between landraces andMVs (cf. table 1). The MIP of control variables is interesting. As expected,the share of fruits affected with borer pest drastically reduces the marketprice. Similar observation was made by Amegbeto and colleagues (2008)on the impact of pest infestation on market price of yams. Other eggplantfruit characteristics, viz., green/white skin colour and presence of spines inthe calyx, also increase the market price.14 All in all, the aggregate negativeimpact of hybrid status and fruit colour may be largely determining theprice difference observed.

Farmer attributes are also included in the model estimation in orderto account for the impact of farmers’ bargaining skills on product prices.

13 Since the mode of transport (e.g., manual transport, transport by bullock cart,transport by tractor) also plays an important role in determining the remotenessof the farm, the time taken to transport the product is included instead of thedistance to the closest market.

14 Regarding the spiny calyx attribute, 34 per cent of hybrids against 6 per centof landraces possess spines on the calyx, but surprisingly, consumers commonlyassociate the spiny attribute with landrace status. Adverse selection owing to thisinformation asymmetry may be one of the reasons for the associated high marginalimplicit price of spiny calyx. This can be interpreted as ‘disguise’ benefits (cf.Hamilton and Zilberman, 2006) to MV farmers, owing to the absence of effectiveproduct differentiation.

142V

ijeshV.K

rishnaetal.

Table 3. Hedonic price function.

Model I Model II

Variables Description Coef. χ 2 p value MIP Coef. χ 2 p value MIP

Experience Farmer’s experience ineggplant cultivation in years

0.021 4.43 .04 2.24

Transportation Time to transport the produceto market in hours

−0.043 1.64 .20 −4.58

Extension Information dummy −0.008 0.00 .97 −0.83Mass media Information dummy 0.382 3.04 .08 41.08Education Schooling obtained by farmer

in years0.002 0.02 .90 0.26

Farm size Farm size owned by thehousehold in acres

−0.036 4.09 .04 −3.84

Hybrid eggplant Adoption dummy (versuslandraces)

−0.583 4.73 .03 −65.01 −0.644 6.16 .01 −69.31

Infestation Share of pest infested fruits inthe marketed lot (0–1)

−3.322 10.41 .00 −370.53 −3.072 9.55 .00 −330.55

Purple fruit Fruit skin dummy (versusgreen/white)

−0.493 3.96 .05 −55.03 −0.501 4.38 .04 −53.94

Spiny calyx Dummy for spines in the fruitcalyx

0.724 10.20 .00 80.73 0.649 8.50 .00 69.80

Rabi Season dummy (versus kharif) −0.498 5.95 .02 −55.58 −0.561 7.45 .01 −60.40Summer Season dummy (versus kharif) −0.809 10.26 .00 −90.30 −0.871 12.10 .00 −93.70Andhra Pradesh State dummy (versus

Karnataka)0.787 9.60 .00 87.83 0.356 1.45 .23 38.33

Model intercept 13.221 13.230θ 0.223 0.117 .06 0.217 0.116 .06

Model statistics Log likelihood −1508.45 −1503.46LR χ 2 72.20 84.19p value .00 .00

Note: Dependent variable in both models is farm price of eggplant (Rs/Q); MIP stands for marginal implicit price or partialderivative of the corresponding variable.


Table 3. (Continued) Testing the transformation parameter, θ (N = 239).

Model I Model II

H0

Restrictedlog likelihood LR χ 2 p value

Restricted loglikelihood LR χ 2 p value

θ = −1 −1559.34 111.76 .00 −1564.32 111.73 .00θ = 0 −1505.22 3.53 .06 −1510.30 3.69 .06θ = 1 −1524.96 43.00 .00 −1529.38 41.86 .00

When the household characteristics are included in the model, the MIPgap between landraces and MVs widens to Rs 69/Q, whereas the samebecause of borer infestation and spiny calyx shrinks. In addition, farmerswho consider mass media as a major source of farm information are able toobtain higher price (by about 10 per cent). Experience in eggplant farmingis also found to be positively affecting the farm price, possibly becausefarming experience provides better information on the complex eggplantmarketing system and associated price structure, eventually influencingfarmers’ bargaining power. Surprisingly, no such positive and significanteffect is found to be associated with formal education. Farmers owninglarger farms obtain lower market price for their products, owing probablyto the selection of buyers. Small farmers sell their products mostly directlyto the consumers, whereas procurement by the wholesale agents is commonfor the larger quantities.

There also exists high seasonal and regional variation in eggplant prices.Farmers obtain a lower price during the summer and rabi seasons incomparison with the kharif season. The lowest price is obtained in thesummer season. However, it should be noted that only 15 per cent of hybridadopters cultivate eggplant in this season. On the other hand, the share oflandrace-based farmers, who select the summer season for cultivation, iscomparatively higher (29 per cent). The better adaptability of landracesto drought may be a factor attributing to this varietal choice, even at aprice disadvantage. Similar inter-seasonal variation of perishable productshas been observed by Parker and Zilberman (1993) and Amegbeto andcolleagues (2008). There is also variation in market price across productionlocations. For instance, in Karnataka, where the productivity of eggplant iscomparatively high, farmers obtain a lower price compared with those ofthe nearby state Andhra Pradesh.

All in all, these results indicate that the landrace attribute itself providesa significantly higher price for cultivators and that the product market isdifferentiated to a certain degree for catering to the needs of consumers.Even under the currently limited information level, in absence of formallabels and certification schemes for landraces, the price difference of Rs69/Q owing solely to the landrace status is relatively high and at leastsufficient to cover its disadvantage of its higher per-unit cost of production(difference of Rs 45/Q). Hence, we cannot reject the hypothesis that theprice premium currently realized in the informal markets for the landrace


attribute compensates for its lower productivity relative to its hybridcounterpart.

Information asymmetries and market imperfections exist in the currentscenario of no labels for landrace products. The transaction costs involved inkeeping the eggplant market segmented for landraces would be associatedwith the increase in the number of market agents involved. This may createa drift in the supply function and thus transfer only a fraction of consumers’WTP to the hands of the farmers. In addition, a percentage of consumersis not able to differentiate between landraces and hybrids, implying thatdespite the higher price for eggplant landraces, the scenario may not whollyreflect consumers’ preferences. The next section deals with ascertainingconsumers’ preferences for the landrace products by directly eliciting theirWTP for the landrace attribute. Information from urban households of Indiais used for this purpose.

5.2. Consumers’ stated WTP for landrace attributeIn the consumer survey, the majority of respondents (79 per cent) statedthat they could identify eggplant landrace fruits as being different fromtheir hybrid counterparts. When presented a uniform price scenario, 74per cent of all sampled consumers preferred landrace eggplant fruits overthat of MVs, whereas only 13 per cent showed a preference for the latterand the rest were indifferent. Preference for landraces is stronger in Kolkataand Barddhaman, the cities surrounded by landrace-growing tracts of land.The relatively large preference for hybrid eggplants is seen mostly in NewDelhi and Bangalore, which either are located far away from the productionlocations or are surrounded by hybrid eggplant-growing farmlands.

Following Nelson (1970), consumer preference for landrace products canbe compartmentalized into search, experience and credence.15 No searchattribute was mentioned by consumer households as a reason for thepreference for landrace products. The experience attributes include bettertaste (54 per cent of sampled respondents cited it as one of the reasonsfor preference), cooking quality (15 per cent) and desirable texture (10 percent). Credence attributes include high nutrition value (mentioned by 25per cent of respondents), reduced level of agro-chemical residues (23 percent) and innate medicinal properties (16 per cent). About 60 per cent ofrespondents mentioned at least one of the experience attributes as a reasonfor preferring landrace products, whereas 47 per cent indicated at least onecredence attribute.

Alongside understanding the consumer preferences towards landraces,it is also necessary to analyze how these preferences are translated intowillingness to purchase these products under different prices. Consumers’preferences can be translated as their willingness to purchase landraceproducts when there exists a uniform market price for landrace and MVproducts. On the other hand, the purchase model analyzes the factorscontributing to consumers’ WTP as an increase in price (premium bid

15 Search attributes of products are observable before consumption, and experienceattributes become apparent after consumption. However, credence attributescannot be verified directly (cf. Scarpa et al., 2005).


amount) for the same attribute. The key difference between these twomodels arises from consumers’ ability to pay as all the respondents whoprefer landrace product (at uniform price) may not be able to purchaseit at a given price premium. In addition, consumers’ social, culturaland psychological characteristics that determine the strength of thesepreferences also differentiate between these two models; strong preferencesare expected to affect the WTP for the attribute in question. The estimationresults of the bivariate (preference–purchase) probit model appear in table 4.

The preference model suggests that younger consumers have a morepositive attitude towards landraces. Similarly, per-capita income is found toraise the preference for landraces (albeit at a decreasing rate). Furthermore,consumers who describe themselves as capable of distinguishing landraceeggplants from MVs show a positive attitude to the former. This is linkedwith the fact that consumers from east India (that is to say from Kolkata andBurddhaman), where landraces are extensively grown, are more favourabletowards its consumption.

The purchase model assesses the strength of the attitudes towardslandraces using a stated preference valuation method. It addressesthe hypothetical purchase decisions (willingness to purchase) by thoseconsumers who indicate a positive preference towards landrace products.The estimation results of this model appear in the right column of table 4.Interestingly, older people are associated with a higher probabilityof purchasing landraces at a given price. Income and education alsosignificantly enhance the probability that the consumer with a positiveattitude towards landraces would be willing to pay a price premium forthe landrace eggplant. This result is in line with that of other valuationstudies on organic products (e.g., Florax et al., 2005). Similar is the impact ofinformation, which is represented here by both the education status and theability to identify the landrace products in the market. In stark contrast tothe preference model, consumers of West Bengal showed the lowest WTPamongst the favourable households from other parts of India. This paradoxmay be because consumers face no felt-need for differential markets in theseregions, as the existing local vegetable markets supply mostly the landraceproducts.

As would be expected, consumers’ ability to differentiate landraceproducts is associated positively not only with preference for landracesbut also with their willingness to purchase at a given price premium also.The marginal impact of information on probability of purchase is 15 percent. This positive association between information and consumers’ WTPis indicative of there being a potential for future market developmentfor landrace-based products, upon provision of reliable informationthrough labelling and certification. This result accentuates the necessity ofsimultaneous emergence of the two types of friction-reducing instruments –both market development and information programmes, as suggested byStavins (2003) – in order to effectively translate consumer preferences tofarmers’ decision making. It also supports the idea that the availability ofinformation can foster the development of green markets as in the caseof eco-labelled products in developed countries (Institut fur AngewandteVerbraucherforschung [IFAV], 2001).

146V

ijeshV.K

rishnaetal.

Table 4. Partial observability bivariate probit estimates on eggplant consumer preference for and willingness to purchase landrace products(N = 629).

Preference Purchase

Variable Description Coef. Robust SE p value Coef. Robust SE p value

Female Sex dummy −0.265 0.293 .37 −0.223 0.256 .38Age Age of respondent in years −0.137 0.063 .03 0.084 0.033 .01Square of age 0.002 0.001 .01 −0.001 3.E−04 .00Education Years of schooling obtained by

respondent0.010 0.022 .65 0.034 0.021 .10

Identification Dummy for respondent’s ability toidentify the landrace products

0.575 0.308 .06 0.417 0.242 .08

Occupation Dummy for respondent’s involvementin an income generating occupation

−0.210 0.291 .47 −0.243 0.233 .30

Household Number of members in the household 0.009 0.053 .87 0.067 0.056 .23PCAI Annual per-capita income of the

household (thousand Rs)0.040 0.023 .09 0.052 0.021 .01

Square of PCAI −3.E-04 2.E-04 .12 −3.E-04 1.E-04 .08East India Dummy for respondent from Kolkata

and Burddhaman6.757 0.624 .00 −1.319 0.680 .05

South India Dummy for respondent from Bangaloreand Kolar

0.337 0.433 .44 −0.389 0.641 .54

Bid Bid quoted in willingness to purchasequestion (Rs/kg)

−0.336 0.040 .00

Model intercept 1.614 1.918 .40 −1.592 0.545 .00

Model statistics ρ −.65Log likelihood −329.07Wald χ 2(23) 1510.06


Figure 1. Estimated demand for landrace eggplant products amongst eggplant-consuming households in urban India. ‘Preference = 1’ indicates the set of householdswith preference towards landrace consumption, and ‘Overall’ shows share of householdswith positive preference times share of households with a willingness to purchase at givenprice premium.

The median WTP is estimated as Rs 6.00/kg for the consumer householdsfavouring landrace products and as Rs 4.50/kg for all households(figure 1). These price premium values respectively are 57 per cent and43 per cent of the eggplant price in the urban consumer market duringthe time of data collection, indicating a significant potential for developingsegmented markets for the landrace attribute. This can be confronted withthe information from the supply analysis.

The current informal markets provide farmers a price premium of 31 percent (Rs 1.18/kg; cf. table 1), while the increment attributed specifically forthe landrace attribute is only 16 per cent (Rs 0.69/kg; cf. table 2). On theother hand, about 50 per cent of urban consumers are willing to pay upto Rs 4.50/kg for the same landrace product. Thus, the potential consumerpremium is more than six times greater than the price premium currentlyrealized by farmers, which indicates the scope for a more organized marketsystem with labels and certificates to conserve landraces in situ.

However, it would be unrealistic to assume that the consumer WTPcould be transferred entirely to the farmer, as this would imply no extratransaction costs (e.g., because of being more intermediaries in the market).Furthermore, following Parker and Zilberman (1993), the marketing marginis expected to increase with the level of quality characteristics. Gouchanand colleagues (2005) also indicated significant difference in gross and netmarketing margin across varieties in the case of rice. Having said this, ifthe MFRIs can generate a farm-price premium at least at half of the median


WTP value, they could result in a significant increase in farm profits, thuscreating incentives for farmers to increase the supply of landrace products.

These results suggest that the evolution of reliable marketing channelsalongside a formal labelling system for landrace products could helpincrease their acreage. It is quite plausible that consumptive values forlandraces by urban consumers exist for other major agricultural productsalso. Imparting market information on product origin to consumers couldhelp conserve landrace resources in these crops, provided the pricepremium hence generated is greater than the transaction cost of informationprovision.

6. ConclusionsThe present paper has aimed at contributing to the empirical literatureon the analysis of novel market (friction-reducing) instruments thatcan be applied to achieve the goal of in situ landrace conservation indeveloping countries. The conditions of market segregation for the in situcoexistence of landraces and MVs have been discussed, and the importanceof information dissemination under a potential labelling or certificationsystem for landraces has been addressed. The argument has been furtherillustrated with a congenial case of eggplants in south India, where theconventional hybrid technology diffusion and its associated yield impact islarge.

Currently, it is observed that the higher implicit price associatedwith the desi eggplant attribute (without any formal labelling system)is able to compensate the yield advantage of hybrid eggplant seedsfrom the farmers’ perspective. However, we also identify that thereexists an ever greater consumptive use value and demand for variouseggplant landrace attributes. An analysis of urban consumers’ preferencestowards landrace eggplants suggests that the price premium currentlyrealized by landrace cultivators is only a small fraction of whatconsumers would be willing to pay for the landrace attribute. By notfully translating consumers’ preferences to farmers, the current marketsystem jeopardizes the landrace conservation. The market imperfectionmay be ameliorated by using reliable information provision schemesby means of labels and certificates denoting landrace origin of theeggplant produce. Of course, the higher price premium associated withthe labelling scheme would need to cover the transaction costs. However,a wide margin exists between consumers’ WTP and the price incrementcurrently realized for landrace products. This indicates that even undersome additional transaction costs such labelling programmes may wellwork.

Two broad policy issues are worth mentioning with respect to the findingsof the current paper. Firstly, the hybrid technology diffusion can be seenonly as a partial success, since the perception about the quality of theresultant products is inferior from the consumers’ perspective. We thusreinstate the importance of taking into account the consumption prioritiesin developing agricultural technologies. Equally important here is thedissemination of unbiased information about the technology attributes insociety. Secondly, there appears to be a significant scope for developing


a formal marketing system with labelling and certification to differentiateproducts of landrace origin in developing economies such as India with alarge urban population base. The design of low-cost marketing channelshaving potential to transfer urban consumers’ consumptive value back tothe cultivators could help them sustain the supply of green products. In caseof landraces, it would contribute to the conservation of agrobiodiversity insitu. Strongly recommended in this milieu is the provision of legislativesupport for such quality indicators.

ReferencesAlston, J.M., G.W. Norton, and P.G. Pardey (1998), Science under Scarcity: Principles

and Practices for Agricultural Research Evaluation and Priority Setting, Cambridge:CAB International.

Amegbeto, N.K., V.M. Manyong, O. Coulibaly, and R. Asiedu (2008), ‘Estimatingmarket demand for fresh yam characteristics using contingent valuation:implications for crop breeding and production choices’, Agricultural Economics39: 349–363.

Anand, A., R.C. Mittelhammer, and J.J. McCluskey (2007), ‘Consumer response toinformation and second generation genetically modified food in India’, Journal ofAgricultural and Food Industrial Organization 5(1): 1138.

Basu, A.K., N.H. Chau, and U. Grote (2003), ‘Eco-labeling and stages ofdevelopment’, Review of Development Economics 7(2): 228–247.

Bateman, I.A., R.T. Carson, B. Day, M. Hanemann, N. Hanley, T. Hett, M. Jones-Lee,G. Loomes, S. Mourato, E. Ozdemiroglu, D.W. Pearce, R. Sugden, and J. Swanson(2002), Economic Valuation with Stated Preference Techniques: A Manual, Cheltenham,UK: Edward Elgar.

Bellon, M.R. (2004), ‘Conceptualizing interventions to support on-farm geneticresource conservation’, World Development 32(1): 159–172.

Bellon, M.R., M. Adato, J. Becerril, and D. Mindek (2006), ‘Poor farmers’ perceivedbenefits from different types of maize germplasm: the case of creolization in low-land tropical Mexico’, World Development 34(1): 113–129.

Bellon, M.R., J.L. Pham, and M.T. Jackson (1997), ‘Genetic conservation: a role for ricefarmers’, in N. Maxted, B.V. Ford-Lloyd, and J.W. Hawkes (eds), Plant Conservation:The in situ Approach, London: Chapman and Hall, pp. 263–289.

Berg, T., A. Bjørnstad, C. Fowler, and T. Skrøppa (1991), ‘Technology options andthe gene struggle’, Occasional Papers (Series C), NORAGRIC, As, Norway.

Biber-Klemm, S., T. Cottier, P. Cullet, and D.S. Berglas (2005), ‘The current law ofplant genetic resources and traditional knowledge’, in S. Biber-Klemm, T. Cottier,and D.S. Berglas (eds), Rights to Plant Genetic Resources and Traditional Knowledge,Oxfordshire: CAB International, pp. 56–111.

Birol, E., E. Rayn-Villalba, and M. Smale (2008), ‘Farmer preferences for Milpadiversity and genetically modified maize in Mexico: a latent class approach’,Environment and Development Economics 14(4): 521–540.

Birol, E., M. Smale, and A. Gyovai (2006), ‘Using a choice experiment to estimatefarmers’ valuation of agricultural biodiversity on Hungarian small farms’,Environmental and Resource Economics 34(4): 439–469.

Bose, T.K., J. Kabir, T.K. Maity, V.A. Parthasarathy, and M.G. Som (2002), VegetableCrops, Kolakatta, India: Naya Prakash.

Brown, A.H.D. (1999), ‘The genetic structure of crop landraces and the challengeto conserve them in situ on farms’ in S.B. Brush (ed.) Genes in the Field: On-FarmConservation of Crop Diversity, Rome, Ottawa, ON, Canada, and Boca Raton, FL:IDRC/IPGRI/Lewis Publishers, pp. 29–48.


Brush, S.B. (1999), ‘The issues of in situ conservation of crop genetic resources’ inS.B. Brush (ed.) Genes in the Field: On-Farm Conservation of Crop Diversity, Rome,Ottawa, ON, Canada, and Boca Raton, FL: IDRC/IPGRI/Lewis Publishers, pp.3–28.

Brush, S.B. (2004), Farmers’ Bounty: Localizing Crop Diversity in the Contemporary World,New Haven, CT, and London: Yale University Press, pp. 196–206.

Cleveland, D.A., D. Soleri, and S.E. Smith (2000), ‘A biological framework forunderstanding farmers’ plant breeding’, Economic Botany 54(3): 377–394.

Dalton, T.J. (2004), ‘A household hedonic model of rice traits: economic values fromfarmers in West Africa’, Agricultural Economics 31(2/3): 149–159.

Daunay, M.C., R.N. Lester, J.W. Hennart, and C. Duranton (2000), ‘Eggplants: presentand future’, Capsicum Eggplant Newsletter 19: 11–18.

Edmeades, S. (2007), ‘A hedonic approach to estimating the supply of varietyattributes of a subsistence crop’, Agricultural Economics 37(1): 19–28.

FAO (1996), ‘The state of world’s genetic resources for food and agriculture’, Paperpresented at the Background Documentation for the International Technical Conferenceon Plant Genetic Resources, 17–23 June 1996, Leipzig, Germany.

FAOSTAT (2007), available at http://faostat.fao.org/ (verified October 2009).Florax, R.J.G.M., C. M. Travisi, and P. Nijkamp (2005), ‘A meta-analysis of the

willingness to pay for reductions in pesticide risk exposure’, European Reviewof Agricultural Economics 32(4): 441–467.

Fowler, C. and T. Hodgkin (2004), ‘Plant genetic resources for food and agriculture:assessing global availability’, Annual Review of Environmental Resources 29: 143–179.

Fowler, C. and P. Mooney (1990), The Threatened Gene: Food, Politics, and the Loss ofGenetic Diversity, Cambridge: Lutterworth Press.

Gouchan, D., M. Smale, and P. Choudhary (2005), ‘Market-based incentives forconserving diversity on farms: the case of rice landraces in Central Tarai, Nepal’,Genetic Resources and Crop Evolution 52(3): 293–303.

Grote, U. (2002), ‘Eco-labelling in the agricultural sector: an internationalperspective’, Paper presented at the High-Level Pan European Conference onAgriculture and Biodiversity, 4 March 2002, European Council, Strasbourg, France.

Haab, T.C. and K.E. McConnell (2002), Valuing Environmental and Natural Resources:The Econometrics of Non-Market Valuation, Northampton: Edward Elgar.

Halstead, M.J., R.A. Bouvier, and B.E. Hansen (1997), ‘On the issue of functional formchoice in hedonic price functions: further evidence’, Environmental Management 21:759–765.

Hamilton, S.F. and D. Zilberman (2006), ‘Green markets, eco-certification, andequilibrium fraud’, Journal of Environmental Economics and Management 52(3): 627–644.

Harlan, J.R. (1975), ‘Our vanishing genetic resources’, Science 188: 618–621.IFAV (2001), Recherche: Verbraucherverhalten beim Lebensmittelkauf , Cologne, Germany:

IFAV.Kontoleon, A. and M. Yabe (2006), ‘Market segmentation analysis of preferences

for GM derived animal foods in the UK’, Journal of Agricultural & Food IndustrialOrganization 4(1): 1–38.

Krishna, V.V. and M. Qaim (2007), ‘Estimating the adoption of Bt eggplant in India:who benefits from public–private partnership?’ Food Policy 32(5/6): 523–543.

Krishna, V.V. and M. Qaim (2008a), ‘Consumer attitudes towards GM food andpesticide residues in India’, Review of Agricultural Economics 30(2): 233–251.

Krishna, V.V. and M. Qaim (2008b), ‘Potential impacts of Bt eggplant on economicsurplus and farmers’ health in India’, Agricultural Economics 38(2): 167–180.

Lancaster, K. (1966), ‘A new approach to consumer theory’, Journal of PoliticalEconomy 74(2): 132–157.


Larson, B.A. (2003), ‘Eco-labels for credence attributes: the case of shade-growncoffee’, Environment and Development Economics 8(3): 529–547.

Matuschke, I., R.R. Mishra, and M. Qaim (2007), ‘Adoption and impact of hybridwheat in India’, World Development 35(8): 1422–1435.

Meng, C. and P. Schmidt (1985), ‘On the cost of partial observability in the bivariateprobit model’, International Economic Review 26(1): 71–85.

Moon, W., W.J. Florkowski, B. Bruckner, and I. Schonhof (2002), ‘Willingness to payfor environmental practices: implications for eco-labeling’, Land Economics 78(1):88–102.

Nelson, P. (1970), ‘Information and consumer behaviour’, Journal of Political Economy78(2): 311–329.

Parker, D.D. and D. Zilberman (1993), ‘Hedonic estimation of quality factors affectingthe farm-retail margin’, American Journal of Agricultural Economics 75: 458–466.

Poudel, D. and F.H. Johnsen (2009), ‘Valuation of crop genetic resources in Kaski,Nepal: farmers’ willingness to pay for rice landraces conservation’, Journal ofEnvironmental Management 90(1): 483–491.

Pretty, J.N. (1995), Regenerating Agriculture: Policies and Practices for Sustainability andSelf-Reliance, Washington, DC: Joseph Henry.

Qaim, M. (2003), ‘Bt cotton in India: field trial results and economic projections’,World Development 31(12): 2115–2127.

Ramasamy, C., A. Janaiah, K.N. Selvaraj, and M. Hossain (2003), ‘Hybrid rice in TamilNadu: evaluation of farmers’ experience’, Economic and Political Weekly 38(25):2509–2512.

Scarpa, R. and T. Del Giudice (2004), ‘Market segmentation via mixed logit:extra-virgin olive oil in Urban Italy’, Journal of Agricultural and Food IndustrialOrganization 2(1): 1–18.

Scarpa, R., G. Philippidis, and F. Spalatro (2005), ‘Product–country images andpreference heterogeneity for Mediterranean food products: a discrete choiceframework’, Agribusiness 21(3): 329–349.

Smale, M. (1997), ‘The Green Revolution and wheat genetic diversity: someunfounded assumptions’, World Development 25(8): 1257–1269.

Smale, M., M.R. Bellon, D. Jarvis, and B. Sthapit (2004), ‘Economic concepts fordesigning policies to conserve crop genetic resources on farms’, Genetic Resourceand Crop Evolution 51(2): 121–135.

Smale, M., R.E. Just, and H.D. Leathers (1994), ‘Land allocation in HYVadoption models: an investigation of alternative explanations’, American Journal ofAgricultural Economics 76(3): 535–546.

Sridhar, V., O.P. Vijay, and G. Naik (2001), ‘Field evaluation of brinjal (Solanumspp) germplasm against shoot and fruit borer (Leucinodes orbonalis Guen.)’ InsectEnvironment 6(4): 155–156.

Stavins, R.N. (2003), ‘Experience with market-based environmental policyinstruments’, in K.-G. Maler and J.R. Vincent (eds), Handbook of EnvironmentalEconomics 9(1): 355–435.

Unnevehr, L. J. (1986), ‘Consumer demand for rice grain quality and returnsto research for quality improvement in South-East Asia’, American Journal ofAgricultural Economics 68(3): 634–641.

Van Dusen, M.E. and J.E. Taylor (2005), ‘Missing markets and crop diversity: evidencefrom Mexico’, Environment and Development Economics 10(4): 513–531.

Waugh, F. (1928), ‘Quality factors influencing vegetable prices’, Journal of FarmEconomics 10: 185–196.

Potentials of green consumerism for landrace conservation: evidence from eggplant production sector...

Documents

Transcript of Potentials of green consumerism for landrace conservation: evidence from eggplant production sector...