The Ambivalent Impact of Coffee Certification on … · The Ambivalent Impact of Coﬀee...

World Development Vol. 77, pp. 277–292, 20160305-750X/� 2015 Elsevier Ltd. All rights reserved.

www.elsevier.com/locate/worlddevhttp://dx.doi.org/10.1016/j.worlddev.2015.08.021

The Ambivalent Impact of Coffee Certification on Farmers’ Welfare:

A Matched Panel Approach for Cooperatives in Central Kenya

BART VAN RIJSBERGENa, WILLEM ELBERS a, RUERD RUBENb and SAMUEL N. NJUGUNAc,*

aRadboud University Nijmegen, The NetherlandsbLEI – Wageningen University and Research Centre, The Hague, The Netherlands

c Jomo Kenyatta University of Agriculture and Technology, Nairobi, Kenya

Summary. — Certification is promoted to improve rural welfare through better market access and improved agricultural practices.We compare net effects of Fairtrade- and Utz-Certified coffee production in Central Kenya, using a matched panel from 218farm-households that belong to three cooperatives and were visited twice in 2009 and 2013. We distinguish between effects at field,farm, household, cooperative, community, and market levels. Both certification regimes improved coffee returns, but Fairtrade was moreeffective in coffee processing, whereas Utz contributed to productivity. Under stagnating coffee prices, Fairtrade farmers increased theircoffee specialization, while Utz farmers reduced coffee areas but increased yield.� 2015 Elsevier Ltd. All rights reserved.

Key words — certification, coffee, cooperatives, panel data, Kenya

*

1. INTRODUCTION

Certification of coffee producers is frequently suggested as auseful strategy for improving the position of smallholders inthe market. It started with the launch of the Fairtrade labelin 1988 and was followed by several other standards that arepromoted either by voluntary agencies and/or by private cof-fee companies. The label of Utz Certified has been launched in2002 with the aim to enhance responsible coffee certificationby established private companies. In recent years, other com-pany coffee labels were established by Starbucks (C.A.F.E.Practices) and Nestle (AAA). 1 The Common Code for theCoffee Community (4C) provides a baseline standard forstakeholders in the coffee sector to address sustainability issuesin production and sourcing at a pre-competitive manner.Various coffee labels rely on rather different strategies for

enhancing sustainable production and/or responsible trade.Whereas Fairtrade strongly focuses on reinforcing cooperativeorganization and membership representation, Utz Certifiedgives more attention to on-farm improvement of coffee prac-tices for upgrading coffee quality. In a similar vein, Fairtradeguarantees producers a minimum price (and a premium pay-ment for the cooperative); whereas Utz Certified relies on freemarket prices that recognize coffee quality improvements. Theprocedures for supporting farmers’ welfare are thus focusingon different entry points of their local livelihoods (Ruben &Verkaart, 2011). In this article, we are therefore interested totrace impact pathways of these different strategies for farmersthat have been involved in such labeling regimes for a substan-tial period of time.Earlier studies on the impact of coffee labeling rely mostly

on descriptive analysis and case studies regarding farmers’ per-ceptions (Bacon, Mendez, Gomez, Stuart, & Flores, 2008;Jaffee, 2007; Ronchi, 2002). These studies tend to concludethat Fairtrade strengthens producer organizations and rein-forces farmers’ well-being, but usually do not consider a coun-terfactual of non-certified producers. Recent studies that relyon broader statistical data sets and matching procedures aremore critical and usually find only modest income effects

277

(Jena, Chichaibelu, Stellmacher, & Grote, 2012; Valkila &Nygren, 2010; Ruben, 2008). To increase net cash returnsfor coffee growing households, improved yields seem to bemore important than price premiums (Barham & Weber,2012). Participation in Fairtrade networks reduces exposureto price variations, mitigates risk aversion, and enhancesinvestment attitudes. A comparative analysis of a matchedsample of different types of certified coffee producers in North-ern Nicaragua finds that Fairtrade provides better prices butprivate labels support higher yield and better quality perfor-mance. Fairtrade can thus be helpful to support initial marketincorporation, whereas private labels offer suitable incentivesfor subsequent quality upgrading (Ruben & Zuniga, 2011).Coffee certification in East Africa is of a rather recent nature

but has been rapidly expanding, representing currently 26% ofthe world’s sustainable certified coffee supply. 2 Empiricalstudies on the effects of standards for smallholders providediverse results. Riisgaard et al. (2009) compare the perfor-mance of different certification schemes of coffee growers inUganda, Kenya, and Ethiopia and find only slight differencesin crop quality and productivity performance and revenue out-comes. Bolwig, Riisgaard, Gibbon, and Ponte (2013) demon-strate that most coffee standards achieved impact underrather restricted conditions and suggest that more selectivesupport and better-tailored interventions are required to reachtangible welfare outcomes. More positive findings are fromChiputwa, Spielman, and Qaim (2014) who find that Fairtradecertification increases household living standards by 30% andreduces the prevalence and depth of poverty. Ruben andVerkaart (2011) show that coffee standards in Kenya andUganda provide rather different functions, with Fairtrade con-tributing to improved farmer organization and initial marketaccess and Utz-certified enhancing the incentives for qualityupgrading.This article is based on a new and unique balanced panel

dataset from coffee producers that belong to three differentcooperatives located in Kiambu County, Central Kenya. The

Final revision accepted: August 21, 2015

http://dx.doi.org/10.1016/j.worlddev.2015.08.021

http://crossmark.crossref.org/dialog/?doi=10.1016/j.worlddev.2015.08.021&domain=pdf

278 WORLD DEVELOPMENT

sample includes farmers certified with Fairtrade (since 2011),double certification of Fairtrade- and Utz-Certified (since2009) and non-certified farms as a control group. The 218 farmsin the sample were visited twice (in 2009 and 2013) and detaileddata are collected on their production and farming activities,commercial engagement and linkages at cooperative and villagelevels. The survey focuses on 6 different impact levels: (1) coffeefields (land, labor, and input use in coffee, coffee yields, coffeerenovation, etc), (2) farming systems (returns from other crops,off-farm work), (3) household livelihoods (income, expendi-tures, health, savings), (4) cooperative organization (technicalassistance, trade services), (5) village and community networks(water and sanitation, social capital) and (6) value chains (coffeeprices, sales). This multi-level framework enables us to disen-tangle the effects of certification and to identify strong andweakpoints of each labeling regime.The remainder of the article is structured as follows.

Section 2 provides an overview of earlier research on theeffects of coffee certification on smallholder welfare, consider-ing registered impact at micro (farm and household), meso(cooperative and village) and macro (value chain/market)level. Due attention is given to common fallacies in impactmeasurement related to self selection and endogeneity.Section 3 introduces the field sample in Kiambu County andoutlines the methodological approach for estimation impactwith balanced panel data. We rely on a matched difference-in-difference framework to assess the net impact ofFairtrade- and Utz-certified coffee production compared tonon-certified producers. Section 4 discusses the empiricalresults and identifies major differences in impact between thecoffee labels. The results are furthermore embedded in the dis-cussion on different characteristics of certification regimes andrelated to general tendencies in local market development.Finally, Section 5 provided conclusions and policy recommen-dations for enhancing the potential impact of coffee certifica-tion on farmers’ welfare.

2. IMPACT PATHWAYS OF COFFEE CERTIFICATION

Studies on the effects of certification for coffee production inrural communities and for the wellbeing of local farmers wereinitially based on anecdotal evidence, relying on qualitativeinterviews to capture stakeholders’ perceptions. Some morequantitative data have been generated through comparativecase studies of certified and non-certified farmers (Bacon,2005; Jaffee, 2007) and sector-wide analyses that focusedattention on price advantages and market shares of certifiedproducts (Raynolds & Wilkinson, 2007). In a similar vein, sev-eral studies compare different certification regimes and outlinetheir effects for the adoption of improved farm managementpractices, coffee quality upgrading, and net marketing margins(Ruben & Zuniga, 2011; Valkila, 2009).Rather recently, more robust studies regarding the socio-

economic impact of certification have been published that cor-rect for selection bias (i.e., active, better-off farmers are likelyto be among the first to participate in certification schemes)and also consider the likelihood of substitution effects (e.g.,with the focus on certified crops, production of other cropsor engagement in other non-farm activities might decrease).Most studies rely on evidence from South- and CentralAmerican countries where coffee certification started early.Field studies from (East) Africa are more recent and paydue attention to competing certification regimes.A major difficulty for fully understanding the net effects and

dynamic implications of coffee certification refers to the fact

that simultaneously different mechanisms are in force thatmay influence several dimensions of farm-household welfare.Certification not only provides incentives for adjusting coffeeproduction systems (plot level), but could also lead to changesin land use (farm-level) and in labor allocation (householdlevel). In addition, most certification systems rely on socialenforcement through engagement with farmers cooperatives(group level) and also involve external relationships with otherneighboring farmers (community level). Finally, supply chainnetworks with traders and processors influence prices andmarketing of certification regime (value chain/market level).This complex multi-layered framework of different interlinkedmechanisms trough which certification may influence farm-household welfare deserves further analysis.In order to be able to better disentangle the net effects of cer-

tification we developed an analytical framework that clearlydistinguishes these different impact levels. Figure 1 providesa systematic picture of the different pathways through whichcertification could influence farm-household welfare. Key dif-ferences in impact pathways between coffee certificationregimes are indicated through their intended incidence oneither coffee production systems and market signals (UtzCertified) or on cooperative organization and communitydevelopment (Fairtrade).The original Fairtrade proposition offers farmers guaranteed

minimum prices and an additional premium for community-level investments. Key attention is given to training for capacitydevelopment (often in partnerships with NGOs) in order toreinforce farmers’ loyalty with the cooperative organizationand to strengthen the cooperative bargaining position vis-a-vis traders and processors. Utz Certified—sometimes labeledas Responsible Trade—relies on strictly market-based prices,but intends to enhance farm-household welfare through coffeesystems upgrading at plot level (e.g., improved agriculturalpractices; tree renovation, etc. with reliance on Farmers FieldSchools) that is expected to result in better quality coffees thatcan receive a premium price at the market. In essence, Fairtradeexpects that exogenous price certainty provides incentives tofarmers to enhance their market integration, whereas UtzCertified expects that farming system intensification results inendogenous market price improvements. Both standards donot engage in providing access to finance and limit their atten-tion only to the coffee plots. The essential difference betweenboth approaches has been characterized by Petkova (2006) as’top-down’ vs. ’bottom-up’ certification.We rely on the indicated distinction in impact levels to dis-

cuss the effects of coffee certification as reported in the avail-able literature. This approach is considered as a welcomeaddition to the existing comparative overview studies (see:Blackman & Rivera, 2010; Le Mare, 2008; Nelson & Pound,2009) but permits more thorough insights into the—sometimescontradictory—effects that might occur at different systemlevels. Distinguishing between impact pathways also enablesa more balanced appraisal of the net welfare effects that speci-fic types of (small-scale) producers might expect from engage-ment in each of the certification regimes.The first impact level concerns the coffee plot. Utz-Certified

(and organic) certifications usually require investments for cof-fee upgrading and improved plot management practices. 3

Valkila (2009) and Valkila and Nygren (2010) point to therequirements of input use and labor intensification for obtain-ing certification that lead to higher costs and lower net margins.Renard (2005) and Ruben and Zuniga (2011) argue that Fair-trade coffee certification—compared to private labels—offersfew incentives for quality upgrading. Apparently, guaranteedprices are not sufficient to mitigate prevailing risk-aversive

C. Household

A. Coffee Plot

B. Farm

E. Village / Community

F. Market

D. Cooperative

Fairtrade

Utz Certified

Figure 1. Impact pathways of coffee certification.

THE AMBIVALENT IMPACT OF COFFEE CERTIFICATION ON FARMERS’ WELFARE 279

behavior, and targeted training programs offer better prospectsfor enhancing the adoption of good agricultural practices.Mostplot-level investment are, however, limited to improving vari-able input use and optimizing tree density, whereas in-depthinvestments in coffee tree renovation are still scarce (Ruben &Verkaart, 2011; Rueda & Lambin, 2013).The second impact level focuses on farm-level effects. Since

most impact studies are limited to plot-level information,insight into wider farm-level effects is generally limited.Ruben (2008) registers that certified farms tend to specializemore on coffee production and may neglect food crops and/or reduce engagement in off-farm employment. In a similarvein, Philpotts, Bichier, Rice, and Greenberg (2007) provideevidence of scarce biodiversity and limited shading on certifiedcoffee farms, whereas Van der Vossen (2005) shows that certi-fied coffee production competes for labor and nutrients withother farming activities. While specialization could reinforceinternal economies of scale and scope, net benefits remain lim-ited due to simultaneous reduction of other income-generatingactivities. The usually modest and sometimes non-significantincome effects from certification are thus likely to be associ-ated with land and labor substitution.The third impact level refers to household-level effects, where

the generation of income and decision on expenditures takeplace. Several empirical studies conclude that net incomeeffects of certification are rather modest, partly due to higherproduction costs and otherwise occasioned by low marketprices (Jena et al., 2012; Barham, Callenes, Gitter, Lewis, &Weber, 2011). Even with small net income effects, Rubenand Fort (2012) find significant livelihood improvements incredit access and durable expenditure that are related toreduced risk aversion. Arnould, Plastina, and Ball (2009)and Becchetti and Costantino (2008) observe positive changesin education and health outcomes related to Fairtrade certifi-cation. Bolwig et al. (2013) use a Heckman selection model tounravel the revenue effect from certification and conclude thatthe largest effects are due to engagement in coffee processing.Improvements in intra-household relationships have beenrarely observed (Kasente, 2012; Lyon, 2008).The fourth impact level is related to the cooperative. The cer-

tification agreement is almost always reached at group leveland the cooperative organization is the linking pin betweenthe individual farmers and the market. Arnould et al. (2009)distinguish the impact of Fairtrade certification from that of

cooperative membership and conclude that most direct eco-nomic effects are related to cooperative membership and lessinfluenced by market exchange arrangements. Saenz Seguraand Zuniga-Arias (2008) also find that membership of cooper-atives is the single most important factor driving income differ-ences between certified and non-certified farmers. Elder,Zerriffi, and Le Billon (2012) further analyze the effects ofFairtrade on social capital and conclude that factors like trustin leadership, democratic internal organization and women’participation are crucial for obtaining positive outcomes fromcertification.The fifth impact level concerns the community. Local partic-

ipation in certification schemes is sometimes biased towardbetter-off farmers, and hired workers depending on wage pay-ments do not always equally benefit. An important side-effectfrom certification that is frequently registered refers to spatialexternalities accruing to neighboring farmers. Such spilloversare found by Ruben and Fort (2012) as a consequence ofknowledge exchange networks at community level that leadto rapid diffusion of improved cultivation practices, but alsooccur due to the fact that the Fairtrade premium is usuallyinvested for collective goods at village level (road improve-ment, schools and health care facilities, water and sanitationfacilities) that equally benefit non-certified producers. Similareffects—albeit with more unequal village-level distribution—are registered by Arnould et al. (2009) with reference toschooling and health outcomes.The sixth and final impact level refers to the relations with

the market and supply chain networks. The pressures andopportunities created by global markets have a decisive effecton whether certification results in livelihood benefits (Weber,2011; Rueda & Lambin, 2012). Mendez et al. (2010) demon-strate that Fairtrade-certified farmers receive better pricesand higher coffee revenues, but sales to certified markets werefar too limited for reaching sustainable livelihood effects. Thisis confirmed by De Janvry, McIntosh, and Sadoulet (2012)showing that in Central America—due to over-certification—less than a quarter of the certified coffee canbe sold under premium conditions. Whereas farmers incursubstantial costs for obtaining certification it becomes increas-ingly important to guarantee access to premium outlets(Neilson, 2008). 4 Certification has brought, however, littlechanges into agency relations throughout the value chainand the nature of contracts (i.e., distribution of value added


shares) has not been substantially modified (Muradian &Pelupessy, 2005; Raynolds, 2009; Valkila, Haaparanta, &Niemi, 2010).

3. MATERIALS AND METHODS

To assess the impact of Utz and Fairtrade certification, weconducted a panel-study including three cooperatives inKenya. Kenya produces some of the best coffee in the worldnotably the ‘‘fully washed mild” more flavorful Coffeaarabica. This is attributed to the well-distributed rainfall,high altitude and moderate temperatures and deep red vol-canic soils. A large number of smallholder farmers dependdirectly upon coffee as a key source of income. But manyfarmers are also engaged in food production (maize) andlivestock (mainly dairy) and occasionally work as temporarylaborer off-farm. Household revenues are therefore ratherdiversified in order to be able to spread risks (Salami,Kamara, & Brixiova, 2010).In recent years, coffee certification schemes such as

Fairtrade (FT) and Utz have triggered a number of develop-ments in the Kenyan smallholder coffee sector. These includeincreased awareness of good agricultural practices, upgradingof processing practices, attention for environmental concerns,improved working conditions, record keeping, traceability andcooperative governance. The demand for certified coffee, how-ever, remains considerably lower than the supply. In 2012, lessthan one third of the certified coffee in Kenya (28% for Utz,30% for FT) was sold as certified coffee (i.e., using theFairtrade- or Utz-Certified label) (personal communication,representative Kenya Coffee Traders Association, February 1,2013).The three coffee cooperatives in our study—Ndumberi F.C.

S. Ltd, Kiambaa F.C.S. Ltd and Mikari C.G.C.S. Ltd—arelocated in Kiambu County in Central Kenya. The cooperativesare engaged either in single (Fairtrade) or double certification(Fairtrade and Utz) and compared with otherwise similar non-certified coffee producers. This region is located close toNairobi and has a fairly good road network that is trafficablethroughout the year. The volcanic soils are rich in organicmatter and the climate is ideal for growing quality coffee.The coffee growing zone in Kiambu spreads on the foothillsof the Aberdare ridge with altitude ranging from 1.520 to2.200 m.a.s.l. There are twenty-two active cooperative soci-eties with a total active membership of over 50,150 coffeefarmers (KCTA, 2012). 5

Coffee production in Kenya suffered during the study periodfrom secular declining margins, mainly due to the progressiveliberalization of coffee trade and the proliferation of coffeemarketing agents. Sales through the central coffee auction inNairobi increased in importance (capturing up to 90% of pro-duction), thus reducing options for directly negotiated sales.Consequently, with declining coffee volumes, prices for qualitycoffee increased. However, also costs for fertilizers, wagelabor, and other purchased inputs raised substantially, thusaffecting net coffee returns. Moreover, macroeconomic imbal-ances resulted in currency devaluation (due to rising importdeficits) and high domestic inflation (up to 20% annually) thataffected net export margins (World Bank, 2013). Compared tocoffee, returns from (wage) labor and revenues from domesticfood production (maize; dairy) became relatively more attrac-tive, especially since coffee yields and quality were affected bymaturing trees and deficient input applications, and foodprices strongly increased due to growing domestic scarcityand rising dependency on imports.

We selected three cooperatives from the regional sample intwo steps. First we interviewed experts knowledgeable on theKenyan coffee sector and reviewed secondary sources to estab-lish a shortlist of potential cooperatives to be taken up in thestudy. Second, we interviewed the management committees ofthe shortlisted cooperatives and made a selection based onsimilarities in location, agro ecological conditions andresource endowments. Hereafter, a random household samplewas taken from the membership listings of each cooperative.The sampling frame prepared for each of the cooperativeswas based on all members delivering coffee to the wet millsbelonging to the respective cooperative. To facilitate suitablematching, the number of coffee trees owned by members wasused as an indicator to guarantee uniformity in wealth. Werandomly selected households proportionately to the distanceto the cooperative wet-mills and their tree density. 6

The fieldwork is based on two rounds of data collection, vis-iting the same farmers in 2009 and 2013. The initial sample sizefrom each of the certified cooperatives included 80–100 farmhouseholds from both treatment groups. 7 One of the treat-ments—Ndumberi cooperative—became early engaged withcertification and carries now both Fairtrade and Utz labels,while the second treatment Kiambaa cooperative received somethree years ago its Fairtrade certification. Given the need formatching, a larger number of 120 households were selectedfrom the non-certified Mikari cooperative as control group.For reasons of comparability, the selected non-certified cooper-ative had similar characteristics in terms of number of wet mills,membership and governance structures. For the second roundof data-collection in 2013, all farmers that participated in thefirst round were revisited. From the 300 farmers that were inter-viewed in the first round, a total number of 218 could be revis-ited. Table 1 offers an overview of the sample composition andthe certification background of each of the cooperatives.Primary data were collected through single farm visits using

structured questionnaires. Field enumerators administered thequestionnaire while interviewing key respondents (usually thehousehold head). Households were defined as the social entitythat collectively makes productive and consumptive decisionsand often eats from the same granary. The first round of datacollection took place at November 2009 (covering the 2008coffee year) and the second round was carried out in January2013 (covering the 2012 coffee year). 8

The questionnaire used to assess the certification pathwaysis based on a detailed reconstruction of key socio-economicindicators at the plot, farm-households, cooperative-villageand market levels. We started to assess the farm-householdportfolio of production activities, recording yields, input useand prices of major agricultural activities that enable subse-quently the calculation of gross and net profit margins. Thisis aggregated to household level income by including othernonfarm and off-farm activities (including remittances), thusproviding information on the level and composition of totalhousehold income and possible trade-offs or complementari-ties between coffee production and other household activities.Full household revenues are calculated from the expenditurebudget, based on the standard procedures used in rural pov-erty studies (Deaton, 1997). This also permits disaggregationinto several spending categories (e.g., food consumption,transport, education, housing, health care, etc.) to identifypossible differences in income use between FT-, FT/Utz- andnon-certified households.Hereafter, attention is given to the available capital

resources (fixed assets, value of cattle, household durables,savings) and possibilities for borrowing (credit access) thatprovide an indication of relative household wealth. Current

Table 1. Certification regime and sample size in selected cooperatives of Kiambu region

Cooperative 2009 N 2010 2011 2012 2013 N

Ndumberi FT + Utz 80 FT + Utz FT + Utz FT + Utz FT + Utz 61Kiambaa None 100 None FT FT FT 71Mikari None 120 None None None None 86


and past investments for house upgrading or land improve-ments are specifically registered to account for changes inthe wealth status. Finally, farmers are asked to indicate theirsubjective appreciation of past and expected future welfareperceptions, for example, whether they consider that theirposition improved, remained the same or has deteriorated.In addition to these standard welfare indicators, it is partic-

ularly important to detect behavioral responses, as well aschanges at the level of farmers’ organization that may resultfrom FT and/or Utz involvement. We therefore included someadditional dimensions of involvement in certification, like atti-tudinal response related changes in risk attitudes, willingnessto invest and self esteem, as well as institutional effects relatedto the satisfaction with cooperative service provision, thedegree of identification with the cooperative firm (Ashforth,Harrison, & Corley, 2008), and the perceived bargainingpower (force index) of the cooperative. 9 Moreover, weincluded some questions regarding changing intra-householdrelationships and gender roles (particularly concerning keyhousehold decisions made alone or together) and on short-and long-run investment in sustainable land use practices.Finally, the survey explicitly considered spillover effects of

FT and Utz engagements toward other aspects of rural liveli-hood strategies (i.e., changes in crops choice, adjustments inlabor use within/outside the household) and externalities ofcertification that might accrue to all households in the commu-nity (related to premium use and changes in regional coffee orlabor markets). Combining all these aspects into an integratedapproach permits us to generate simultaneously insights intothe effectiveness and constraints of targeting efficiency (i.e.,inclusion of the poor households and possible leakages tothe non-poor), as well as the broader socio-economic implica-tions of certification at different scale levels (e.g., individual,household, farmers’ organization and village/region).The registered primary data are subsequently used for fur-

ther analysis (see Section 4). Data analysis of this balancedpanel is based on a matched difference-in-difference approach.We combine a ‘‘with and without” assessment of certificationcomparing FT-, Utz- and non-certified cooperatives, and a‘‘before and after” analysis of certification comparing baselinewith ex-post survey data. The diff-in-diff estimator compareschange in outcomes from the treatment group before andafter the intervention to the changes in outcomes in the con-trol group. By analyzing differences, the estimator controlsfor characteristics that do not change over time within thetreatment and control groups. The registered net change inthe control group is an estimate of the true counterfactual i.e., what would have happened to the intervention group ifcertification was not implemented. As a robustness check,we also computed the minimax approach (Dehejia &Wahba, 1999) that provided roughly similar outcomes. 10

We use propensity score matching to correct for the selec-tion bias that is often inherent in the ‘‘with and without”and in the ‘‘before and after” estimates. Propensity scorematching gives more objective judgments regarding the impactof an intervention and corrects for possible initial selectionbiases (Rosenbaum & Rubin, 1983; Heckman, Ichimura, &

Todd, 1997). To operationalize the propensity score matchingtechnique, variables or covariates that could influence theprobability of a household participating either in FT and/orUtz certification should be determined. In order to get anunbiased sample for certified and non-certified farms, we reliedon a set of exogenous variables that are not influenced by par-ticipation in any certification programs. We included the fol-lowing variables in the estimation of the propensity score:distance to the nearest wet mill (km), distance to the nearestmarket (km), years living at current premises, years of farmingexperience, years of education of household head, femaleheaded household (0/1), acreage, dependency ratio (nr. of chil-dren divided by nr. of adults within household). Hereafter, aprobit regression (treatment = 1; 0 = otherwise) was per-formed on the covariates to determine variables that influ-enced participation of household in the certification program(see Table 2). The Propensity Score (p-score) from the ProbitModel is then used to predict farm-household’s likelihood ofhaving FT/Utz certification. The region of common supportis established after eliminating the observations from thenon-certified group with a p-score lower than the minimump-score in the certified group, and from the certified groupwith a p-score higher than the maximum p-score of the non-certified group (see Figure 2). The final diff-in-diff comparisonsof the panel data are only made between matched farmers thatbelong to FT- or FT/Utz-certified cooperatives (treatmentgroup) with otherwise similar matched farmers belonging tocooperatives that did not join any of the certification program(control group) but would in principle qualify for it.

4. RESULTS

Tables 3–5 give the descriptive statistics for each of the threematched groups at the baseline and during the follow-up sur-vey. We follow the classification into six impact pathways asoutlined in Section 2. FT coffee farms are slightly larger andderive a somewhat higher and increasing income share fromcoffee sales. Both FT and FT/Utz farmers have a long experi-ence with coffee production, but the former operate farmingsystems with higher tree density and more input use. Neverthe-less, initial coffee yields were higher on non-certified farms,and all farm types suffer from a substantial yield decline insubsequent years. FT farms increased their specialization incoffee, while non-certified farms reduce their coffee depen-dency with almost half.At plot-level, coffee yields per hectare show a strong decline,

particularly on non-certified farms but to a minor extent alsoon certified farms. Nevertheless, the area devoted to coffee stillslightly increased (particularly on non-certified farms) and allfarmers made investment in coffee renovation in order tocounteract the tendency of declining tree productivity. Conse-quently, coffee remained the main income source and invest-ments into other crops are still of minor importance.At farm-household level, income derived from other crops

and livestock activities, as well as from off-farm employmentbecomes increasingly important for maintaining rural

Table 2. Probit regressions on treatment. (a) Probit on treatment (FT) vs. non-certified

b se

Dependency ratio 0.01 0.23Acreage �0.01 0.05Female headed household (0/1) 0.42 0.26Years education household head 0.05 * 0.03Years farming experience 0.02 * 0.01Years living in current location 0.01 0.01Distance to wet mill 0.44 *** 0.10Distance to market �0.12 ** 0.06Constant �2.31 *** 0.61

***p < 0.01; **p < 0.05; *p < 0.1; n = 157; LLH = �87.63; chi2 = 40.95; Prob > chi = 0.00; Pseudo R2 = 0.19.

Table 2b. Probit on treatment (Utz/FT) vs. FT

b se

Dependency ratio �0.21 0.25Acreage �0.06 0.05Female headed household (0/1) 0.06 0.25Years education household head �0.05 * 0.03Years farming experience �0.01 0.01Years living in current location 0.01 0.01Distance to wet mill �0.28 *** 0.09Distance to market 0.13 ** 0.05Constant 0.82 0.68


Table 2c. Probit on treatment (Utz/FT) vs. non-certified

b se

Dependency ratio �0.09 0.20Acreage �0.01 0.05Female headed household (0/1) 0.43 * 0.24Years education household head 0.00 0.02Years farming experience 0.01 0.01Years living in current location 0.01 0.01Distance to wet mill 0.15 0.11Distance to market 0.02 0.05Constant �1.31 ** 0.55



livelihoods. Heads of certified households are older, but non-certified households have a higher dependency rate and thushire wage labor more frequently. Food expenditures—particularly of the FT and Utz households—increased substantially,while transport costs to the factory decreased. All householdsmanifested an improvement in their economic situation, butonly FT/Utz -muticertified households also register anincrease in their savings.At the cooperative level, most indicators regarding satisfac-

tion with service provision and trade assistance show somedecline. Farmers with certification report slightly higher satis-faction rates, but differences between both labels are hardlyregistered. Also in terms of village-level social conditions nomajor differences are observed. Over time, engagement withother organizations tends to decrease, particularly amongFT/Utz-certified farmers.Finally, value chain level indicators show an improvement

in processed (dry) coffee prices that is particularly strong for

certified farms. Also cherry coffee prices increased, but far lessfor non-certified farms. The amount of cherry coffee sold wasstrongly reduced, while dry coffee sales remained fairly stable.This indicates that certified farm-households were able toimprove their situation mainly through better prices for pro-cessed coffee, with declining differences between Utz and FTcertification but increasing margins compared to non-certified farm-households.We analyzed the difference-in-difference effects of three

matched pairs of cooperatives in order to capture the neteffects of Fairtrade (comparing FT with non-certified farms),Utz-certified (comparing Utz/FT-certified farms with onlyFT-certified farms) and multi-certification (comparing FTand Utz-certified farms with non-certified farms). In the fol-lowing, we discuss the effects of each of these certificationregimes for registered changes in key parameters at farm/plot, household, cooperative/village and value chain/marketlevel.

After matchingBefore matching

Fai

rtra

de /

Non

-cer

tifi

ed

0.5

11.

52

kden

sity

psc

ore1

0 .2 .4 .6 .8 1x

kdensity pscore1 kdensity pscore1

.51

1.5

2kd

ensi

ty p

scor

e1

0 .2 .4 .6 .8x


Fai

rtra

de &

Utz

/ F

airt

rade

0.5

11.

52

2.5

kden

sity

psc

ore1

0 .2 .4 .6 .8 1x


0.5

11.

52

2.5

kden

sity

psc

ore1

0 .2 .4 .6 .8x


Fai

rtra

de &

Utz

/ N

on-c

erti

fied

01

23

kden

sity

psc

ore1

.2 .4 .6 .8 1x


.51

1.5

22.

5kd

ensi

ty p

scor

e1

.2 .4 .6x


Figure 2. Distribution of p-scores before and after matching.


(a) Impact of Fairtrade

Fairtrade certification gives most significant effects at plot/farm and value chain level (see Table 3). While acreage andnumber of coffee trees remains unchanged for FT farmers,non-certified farms increased both the acreage under coffeeand tree density. Moreover, non-certified farms increased theuse of hired labor for coffee production compared to FT

farms, thus freeing family labor for engagement in otherincome-generating activities. Both groups raised their invest-ments over time, but non-certified farmers at a far higher rate.FT farmers invested nearly twice as much in tree renovation in2009 compared to non-certified farmers, while in 2013 non-certified farms took the lead in new coffee investment.Nevertheless, both groups experienced substantial absolutedeclines in their coffee yields; non-certified farmers registered

Table 3a. Fairtrade-effect: matched difference-in-differences. Fairtrade farmers vs. non certified

Coffee plots 2009 2013 Diff-in-diff

NC FT Diff NC FT Diff

b se b se b r�x b se b se b r�x b r�x

Coffee area (acres) 0.59 0.04 0.65 0.04 0.06 0.05 0.93 0.11 0.64 0.08 �0.28 ** 0.14 �0.34 ** 0.15

Number of mature coffee trees 313.00 18.30 326.00 20.90 13.30 27.80 461.00 43.10 323.00 49.70 �138.00 ** 65.80 �152.00 ** 71.50

Number of coffee trees per acre 544.00 11.70 514.00 16.50 �30.10 20.20 560.00 29.60 603.00 55.80 42.30 63.10 72.40 66.30

Coffee yield (kg/acre) 1.74 0.18 1.25 0.13 �0.49 ** 0.22 0.68 0.09 0.95 0.23 0.27 0.25 0.75 ** 0.33

Coffee yield (kg/tree) 3.17 0.29 2.43 0.25 �0.74 * 0.39 1.33 0.17 2.62 1.16 1.29 1.17 2.03 1.23

Gross coffee income (�1,000 Kes) 18.80 2.04 19.60 2.61 0.86 3.31 10.20 1.28 14.70 2.79 4.52 3.07 3.66 4.51

Gross dry coffee income (�1,000 Kes) 2.26 0.52 2.72 0.62 0.45 0.81 1.51 0.33 2.02 0.74 0.52 0.81 0.06 1.14

Gross cherry coffee income (�100 Kes) 16.90 1.66 17.70 2.43 0.73 2.94 9.76 1.35 12.10 2.23 2.37 2.61 1.64 3.93

Input costs coffee (�1,000 Kes) 4.38 0.70 3.01 0.52 �1.37 0.88 3.29 1.07 3.31 0.76 0.02 1.31 1.39 1.58

Hired labor coffee (yes/no) 0.79 0.05 0.70 0.06 �0.09 0.08 2.19 0.23 1.26 0.27 �0.93 *** 0.35 �0.83 ** 0.36

Investment new coffee (yes/no) 0.15 0.04 0.28 0.06 0.13 * 0.07 0.61 0.06 0.49 0.07 �0.12 0.09 �0.25 ** 0.12

Means and Standard Errors are estimated by linear regression. Robust Standard Errors. Number of respondents = 157 (FT:71, NC:86). ***p < 0.01;**p < 0.05; *p < 0.1. Kernel-based Propensity Score Matching diff-in-diff. Estimation on the common support.

Table 3b. Fairtrade-effect: matched difference-in-differences. Fairtrade farmers vs. non certified

Farm 2009 2013 Diff-in-diff



Gross income other crops (x 1000 Kes) 3.89 0.59 1.95 0.52 �1.94 ** 0.78 0.84 0.47 0.18 0.13 �0.66 0.49 1.28 0.92Input costs other crops (�1,000 Kes) 3.07 0.36 2.23 0.34 �0.83 * 0.49 3.15 0.56 2.85 0.69 �0.29 0.88 0.54 1.01Seed costs other crops (�1,000 Kes) 2.93 0.31 2.36 0.22 �0.57 0.38 1.20 0.16 1.92 0.57 0.72 0.59 1.29 * 0.70Number of Livestock 9.56 0.89 5.49 0.75 �4.07 *** 1.16 8.48 0.95 4.17 0.65 �4.32 *** 1.15 �0.25 1.64Costs livestock (�1,000 Kes) 41.70 5.24 22.50 3.72 �19.20 *** 6.42 48.10 6.04 23.80 4.85 �24.20 *** 7.75 �4.97 10.10Land-related investments (�1,000 Kes) 13.70 3.06 9.17 3.73 �4.57 4.82 18.80 6.79 12.90 5.12 �5.98 8.50 �1.41 9.77Risk index (1–3) 2.14 0.04 2.13 0.04 0.00 0.06 2.14 0.06 2.18 0.05 0.04 0.08 0.04 0.10


Table 3c. Fairtrade-effect: matched difference-in-differences. Fairtrade farmers vs. non certified

Household 2009 2013 Diff-in-diff



Gross income 304.00 42.20 153.00 19.50 �151.00 *** 46.50 246.00 29.20 124.00 24.70 �122.00 *** 38.30 29.10 60.20

Share income from coffee (%) 23.30 3.13 24.00 3.70 0.74 4.84 15.10 3.33 37.00 6.15 21.90 *** 6.99 21.10 ** 8.51

Gross off farm income (�1,000 Kes) 94.90 17.60 95.30 14.50 0.40 22.70 114.00 14.30 68.00 20.80 �45.70 * 25.30 �46.10 34.00

Credit (�1,000 Kes) 43.10 9.22 16.80 5.85 �26.30 ** 10.90 29.00 9.12 28.90 16.00 �0.13 18.40 26.10 21.40

Savings (�1,000 Kes) 53.40 7.21 60.50 14.10 7.04 15.80 16.00 5.10 23.50 7.92 7.46 9.42 0.42 18.40

Total value of assets (�1,000 Kes) 136.00 21.60 80.00 17.20 �56.30 ** 27.60 133.00 51.50 104.00 30.00 �28.40 59.60 27.90 65.60

Expenditure food (�1,000 Kes) 4.87 0.22 4.21 0.34 �0.66 0.41 5.94 0.46 7.78 1.08 1.83 1.17 2.49 ** 1.24

Expenditure education (�1000 Kes) 15.30 2.70 8.37 1.66 �6.92 ** 3.17 11.50 3.39 11.30 3.40 �0.18 4.80 6.75 5.75

Expenditure energy (�1,000 Kes) 2.12 0.27 1.88 0.29 �0.24 0.39 3.56 0.42 5.31 0.96 1.76 * 1.05 1.99 * 1.12

Expenditure transport (�1,000 Kes) 4.38 0.57 4.62 0.80 0.25 0.98 3.53 0.48 1.19 0.32 �2.33 *** 0.57 �2.58 ** 1.13

Expenditure total (�1,000 Kes) 56.90 6.58 48.60 7.07 �8.29 9.66 51.90 8.23 32.70 5.11 �19.20 ** 9.69 �10.90 13.70

House investments (�1,000 Kes) 1.74 0.05 1.67 0.06 �0.07 0.08 1.81 0.05 1.68 0.07 �0.14 * 0.08 �0.06 0.12

Econ. situation vs. �5 years (1–3) 1.90 0.10 1.85 0.12 �0.05 0.16 1.48 0.09 1.38 0.10 �0.10 0.13 �0.05 0.21

Econ. situation vs. +5 years (1–3) 1.55 0.09 1.56 0.11 0.00 0.15 1.29 0.07 1.37 0.10 0.07 0.12 0.07 0.19

Male gender index 18.80 0.53 18.20 0.57 �0.57 0.78 13.00 0.73 14.10 0.55 1.11 0.91 1.67 1.20

Female gender index 19.10 0.55 18.90 0.92 �0.20 1.07 15.40 0.89 15.60 1.52 0.16 1.77 0.36 2.06

Days lost due to poor health (log) 0.64 0.16 1.12 0.22 0.48 * 0.27 0.70 0.17 0.88 0.22 0.19 0.28 �0.29 0.39



Table 3d. Fairtrade-effect: matched difference-in-differences. Fairtrade farmers vs. non certified

Cooperative 2009 2013 Diff-in-diff



Satisfaction tech. assistance (1–5) 3.07 0.17 3.30 0.18 0.22 0.24 2.57 0.16 3.04 0.13 0.47 ** 0.21 0.25 0.32Satisfaction trade assistance (1–5) 2.94 0.16 3.44 0.16 0.51 ** 0.23 2.69 0.16 2.78 0.12 0.09 0.20 �0.42 0.30Identification index 2.58 0.07 2.63 0.07 0.04 0.10 2.55 0.06 2.37 0.07 �0.19 ** 0.09 �0.23 * 0.13Force index 2.41 0.07 2.44 0.08 0.04 0.11 2.07 0.06 2.17 0.07 0.10 0.09 0.07 0.14Total value of side selling 5.67 2.96 3.35 2.25 �2.32 3.71 22.90 22.00 2.72 2.20 �20.20 22.10 �17.80 22.40


Table 3e. Fairtrade-effect: matched difference-in-differences. Fairtrade farmers vs. non certified

Village 2009 2013 Diff-in-diff



Have piped water (yes/no) 0.04 0.02 0.35 0.07 0.31 *** 0.07 0.42 0.06 0.43 0.07 0.01 0.09 �0.30 *** 0.11Have improved latrine (yes/no) 0.55 0.06 0.50 0.07 �0.05 0.09 0.63 0.06 0.51 0.07 �0.12 0.09 �0.07 0.13Number of organizations 1.51 0.06 1.19 0.05 �0.32 *** 0.08 0.49 0.06 0.53 0.07 0.04 0.09 0.36 *** 0.12


Table 3f. Fairtrade-effect: matched difference-in-differences. Fairtrade farmers vs. non certified

Market 2009 2013 Diff-in-diff



Coffee sold in cherry form (kg) 839.00 80.20 762.00 110.00 �76.30 136.00 457.00 71.50 301.00 43.20 �156.00 * 83.50 �79.90 159.00

Dry coffee sold (kg) 57.40 8.17 51.70 6.04 �5.68 10.20 43.80 6.58 33.80 8.84 �10.00 11.00 �4.33 15.00

Cherry coffee price (Kes/kg) 20.50 0.71 24.60 0.63 4.06 *** 0.95 27.30 1.52 49.60 4.53 22.40 *** 4.78 18.30 *** 4.87

Dry coffee price (Kes/kg) 33.30 1.63 41.00 0.65 7.64 *** 1.76 46.40 4.38 157.00 14.40 110.00 *** 15.10 103.00 *** 15.20



better yields at baseline, whereas FT farmers reach betteryields at endline. In addition, while prices improved for bothgroups over time, FT farmers register significantly higher cof-fee prices. In summary, this points to significant positive pro-duction effects of FT certification through intensification ofcoffee cultivation and further coffee processing (e.g., increasedshare of dry coffee sales).At household level, the share of income generated from cof-

fee decreased in non-certified farms from 24% to 15%, while itincreased in FT farms from 26% to 38%. Given the fact thatrelative coffee prices declined compared to prices of othercrops and livestock, non-certified farmers generate substan-tially higher incomes compared to FT farmers. Moreover, alsoincome from off-farm employment drops more in FT-certifiedfarms that tend to use more family labor in coffee production.Household income of non-certified farmers remains twice ashigh as income of certified farms, and—despite the positiveproduction effects—net income effects of certification arehardly registered. Consequently, only some minor shifts inexpenditure patterns occur, with increasing food and energyexpenditures and decreasing transport expenditures for bothgroups, at slightly stronger intensity for FT farmers.Finally, certification effects at cooperative level are rather

limited. FT farmers report a decreasing degree of identification

with their cooperatives, even while almost no effects arereported for the satisfaction with cooperative service provisionin the fields of technical assistance or marketing. At villagelevel, non-certified farmers increased their access to pipeddrinking water up to the same level as FT farmers. Moreover,both groups strongly reduced their involvement with othercommunity organization and tend to focus on limited socialnetworks. Other behavioral effects related to changes in genderroles or risk attitudes are not found.Certification brings FT farmers toward greater specializa-

tion into coffee production and further engagement in coffeeprocessing, enabling higher tree productivity and rising coffeerevenues. However, non-certified farmers seem better able toadapt to changing market conditions and still earn twice asmuch household income.

(b) Impact of Utz certification

The matched comparison of farmers with the combinedUtz/FT label with farmers that only have FT certification pro-vides insights into the net effect of Utz-certified coffee produc-tion (see Table 4). At first sight, differences between bothlabels are rather limited, and diff-in-diff estimates hardlyregister any significant effects at farm, market/chain and


cooperative level over time. Only for some variables at house-hold and village level changes can be identified.In economic terms, Utz-certified farmers received initially

substantially higher prices for cherry and particularly for drycoffee deliveries, but these price differences disappeared overtime as all certified coffees increase in prices. Farmers withUtz-certification slightly raised their cultivated coffee areaand input use, but coffee yields remained stagnant. Otherwise,Utz farmers had initially more credit and capital assets,whereas FT farmers relied more on their own savings. Thesedifferences tend to even out over time when investments forboth labels gradually decrease. Consequently, only minor dif-ferences in household expenditure patterns are registered.At cooperative and village level, Utz-certified cooperatives

stay behind in terms of access to safe drinking water, whereasFT cooperatives made substantial improvements in this field(presumably financed from FT premium revenues). On theother hand, Utz cooperatives increased access to latrines toreinforce sanitary conditions. Nevertheless, there are morelabor days lost due to poor health conditions in Utz-certifiedcooperatives. They also reduced their affiliation to other orga-nizations and thus became more focused on their internalsocial network.

(c) Impact of multi-certification

The comparison of double (FT and Utz)-certified with non-certified farms provides insights into the effects of multi-certification (see Table 5). We observe positive and significanteffects of multi-certification at several impact levels, mostnotably for coffee yields, coffee prices and income from coffeeproduction. Investments in coffee renovation and upgradingof coffee production systems took place on multi-certifiedfarms, enabling them to raise yields and to increase theirincome share derived from coffee production. These differ-ences are, however, less pronounced if we realize that the rel-ative decline in coffee income at non-certified farms iscompensated by substantially more revenues from other crop-ping and livestock activities. Moreover, multi-certification inNdumberi cooperative coincides with a longer period of beingcertified, and therefore part of the registered effects could bedue to learning.Multi-certified farms are better able to market their coffee

at favorable terms: they receive higher prices in both yearsand the difference with non-certified farms increases overtime.

Table 4a. Utz-effect: matched difference-in-diffe

Coffee plots 2009

FT FT/Utz Diff

b se b se b

Coffee area (acres) 0.67 0.04 0.53 0.04 �0.14 **

Number of mature coffee trees 327.00 20.30 275.00 19.90 �51.70 *

Number of coffee trees per acre 501.00 16.00 552.00 22.10 51.00 *

Coffee yield (kg/acre) 1.12 0.10 1.16 0.13 0.03

Coffee yield (kg/tree) 2.25 0.23 2.24 0.27 �0.01

Gross coffee income (�1,000 Kes) 19.20 1.95 19.20 2.72 �0.02

Gross dry coffee income (�1,000 Kes) 2.54 0.60 2.60 0.57 0.06

Gross cherry coffee income (�100 Kes) 16.70 1.90 16.70 2.48 �0.02

Input costs coffee (�1,000 Kes) 2.47 0.35 2.01 0.35 �0.46

Hired labor coffee (yes/no) 0.68 0.06 0.67 0.06 �0.01

Investment new coffee (yes/no) 0.25 0.06 0.12 0.04 �0.13

Means and Standard Errors are estimated by linear regression. Robust Standar**p < 0.05; *p < 0.1. Kernel-based Propensity Score Matching diff-in-diff. Est

The reduction of the income share from coffee at non-certified farms (from 24% to 15%) marks a large differencewith multi-certified farms that get to one third of householdincome from coffee. However, expenditures on food purchasesincrease more on multi-certified farms that sacrifice land use offood crops for cash crop production. Initially, non-certifiedfarmers saved twice the amount of multi-certified households,while at the end multi-certified households set aside almostdouble the amount of savings compared to non-certified farm-ers. This reflects their interest in future investments for farm-ing systems intensification.We do not find a significant impact of multi-certification

on most variables at the cooperative level, even while thedegree of satisfaction with service provision slightlydecreased. Otherwise, the increased bargaining power forth-coming from multi-certification is clearly appreciated. More-over, improvements in village-level drinking water services(for non-certified households) and latrines (for certified farm-ers) are registered.In summary, and in line with other research (Blackman &

Rivera, 2010; Ruben & Zuniga, 2011), multi-certification tendsto enhance further specialization in coffee and has a clear pos-itive price effect. Rising income shares from coffee do not nec-essarily result in higher household income, since non-certifiedfarms maintain strong investments in other cropping and live-stock activities and further engagement in off-farm employ-ment. Whereas certified farms thus diversify coffee outlets,non-certified farms diversify their activity pattern.

5. DISCUSSION AND OUTLOOK

Multi-level analysis of impact pathways of coffee certifica-tion provides unique opportunities for identifying in a moredetailed way the likely effects of different certification regimes.Our comparative analysis permits to draw conclusions aboutthe relevance of coffee certification at plot and farm level forparticular types of producers, the implications of certificationfor adjustments in resource use at household and village level,and the limitation and constraints that remain at higher sys-tem levels (community and value chain level).First, it appears that specific certification regimes have dif-

ferent effects on farming systems development. Whereas Fair-trade clearly enhances further specialization in coffeeproduction and more engagement in dry coffee processing,

rences. Utz/Fairtrade vs. Fairtrade farmers

2013 Diff-in-diff

FT FT/Utz Diff

r�x b se b se b r�x b r�x

0.06 0.67 0.09 0.62 0.08 �0.05 0.12 0.09 0.13

28.40 350.00 52.30 295.00 45.10 �55.50 69.10 �3.73 74.70

27.30 634.00 57.30 557.00 60.40 �76.30 83.30 �127.00 87.60

0.17 1.09 0.26 1.07 0.44 �0.02 0.51 �0.06 0.54

0.36 3.03 1.42 1.63 0.35 �1.41 1.46 �1.40 1.50

3.35 16.90 3.03 23.30 5.32 6.42 6.12 6.44 6.98

0.83 2.22 0.76 3.71 0.78 1.49 1.09 1.43 1.37

3.12 13.40 2.45 19.60 5.11 6.22 5.67 6.25 6.47

0.49 3.62 0.84 4.05 1.70 0.44 1.90 0.90 1.96

0.09 1.31 0.27 1.77 0.26 0.47 0.37 0.48 0.38

0.07 0.46 0.07 0.41 0.07 �0.04 0.09 0.09 0.12

d Errors. Number of respondents = 132 (FT/Utz:61, FT:71). ***p < 0.01;imation on the common support.

Table 4c. Utz-effect: matched difference-in-differences. Utz/Fairtrade vs. Fairtrade farmers


FT FT/Utz Diff FT FT/Utz Diff


Gross income 159.00 20.40 156.00 29.80 �2.65 36.10 126.00 18.90 174.00 33.20 48.20 38.20 50.90 52.60

Share income from coffee (%) 26.10 3.84 37.90 5.06 11.90 * 6.36 35.10 5.62 33.90 5.22 �1.25 7.67 �13.10 9.96

Gross off farm income (�1,000 Kes) 134.00 18.30 168.00 46.80 33.90 50.20 87.50 20.40 79.40 24.00 �8.11 31.50 �42.00 59.30

Credit (�1,000 Kes) 11.40 3.01 38.40 11.00 27.00 ** 11.40 6.46 3.44 20.00 12.00 13.50 12.50 �13.50 16.90

Savings (�1,000 Kes) 56.90 13.40 29.40 7.09 �27.50 * 15.10 21.50 7.25 30.20 8.81 8.70 11.40 36.20 * 19.00

Total value of assets (�1,000 Kes) 88.30 18.00 174.00 57.90 85.90 60.60 115.00 31.80 233.00 68.30 118.00 75.30 32.30 96.70

Expenditure food (�1,000 Kes) 4.17 0.32 4.51 0.36 0.34 0.48 7.98 1.04 8.73 0.91 0.74 1.38 0.40 1.46

Expenditure education (�1,000 Kes) 6.75 1.24 8.96 2.77 2.21 3.03 8.72 2.58 8.39 3.40 �0.33 4.27 �2.54 5.24

Expenditure energy (�1,000 Kes) 1.86 0.28 2.35 0.42 0.48 0.50 4.97 0.83 5.28 0.80 0.31 1.15 �0.17 1.26

Expenditure transport (�1,000 Kes) 4.49 0.71 3.61 0.56 �0.89 0.90 1.12 0.30 2.36 0.50 1.24 ** 0.58 2.13 ** 1.07

Expenditure total (�1,000 Kes) 49.80 6.94 45.20 7.61 �4.64 10.30 31.20 4.84 38.90 6.58 7.61 8.17 12.30 13.10

House investments (x 1000 Kes) 1.67 0.06 1.62 0.06 �0.05 0.09 1.68 0.06 1.71 0.06 0.03 0.09 0.08 0.13

Econ. situation vs. �5 years (1–3) 1.81 0.12 2.14 0.13 0.33 * 0.17 1.34 0.09 1.38 0.09 0.04 0.13 �0.29 0.22

Econ. situation vs. +5 years (1–3) 1.54 0.10 1.47 0.10 �0.07 0.15 1.31 0.08 1.22 0.07 �0.08 0.11 �0.01 0.18

Male gender index 18.30 0.43 19.40 0.97 1.05 1.06 14.30 0.61 14.20 0.86 �0.13 1.06 �1.18 1.50

Female gender index 19.90 0.51 21.00 1.36 1.10 1.45 16.00 1.04 14.20 0.77 �1.78 1.29 �2.88 1.94

Days lost due to poor health (log) 1.14 0.22 1.01 0.19 �0.13 0.29 0.98 0.23 1.69 0.26 0.71 ** 0.35 0.84 * 0.45

Means and Standard Errors are estimated by linear regression. Robust Standard Errors. Number of respondents = 132 (FT/Utz:61, FT:71). ***p < 0.01;**p < 0.05; *p < 0.1. Kernel-based Propensity Score Matching diff-in-diff. Estimation on the common support.

Table 4d. Utz-effect: matched Difference-in-differences. Utz/Fairtrade vs. Fairtrade farmers




Satisfaction tech. assistance (1–5) 3.37 0.17 3.57 0.18 0.20 0.25 3.02 0.12 3.02 0.15 0.00 0.19 �0.20 0.31Satisfaction trade assistance (1–5) 3.47 0.16 3.26 0.19 �0.22 0.25 2.79 0.11 2.96 0.15 0.17 0.19 0.39 0.31Identification index 2.63 0.06 2.61 0.07 �0.01 0.10 2.39 0.07 2.47 0.08 0.09 0.11 0.10 0.15Force index 2.41 0.08 2.54 0.08 0.12 0.11 2.16 0.06 2.37 0.07 0.21 ** 0.10 0.09 0.15Total value of side selling 2.64 1.27 2.63 2.17 �0.01 2.51 2.64 2.24 0.00 0.00 �2.64 2.24 �2.63 3.37


Table 4b. Utz-effect: matched difference-in-differences. Utz/Fairtrade vs. Fairtrade farmers




Gross income other crops (�1,000 Kes) 2.34 0.63 1.78 0.42 �0.57 0.75 0.17 0.14 0.50 0.39 0.33 0.41 0.90 0.86Input costs other crops (�1,000 Kes) 2.23 0.34 1.83 0.27 �0.40 0.43 2.63 0.64 1.14 0.37 �1.49 0.74 �1.10 0.86Seed costs other crops (�1,000 Kes) 2.38 0.21 2.09 0.22 �0.29 0.30 1.85 0.47 1.28 0.42 �0.57 0.63 �0.28 0.70Number of Livestock 5.09 0.69 7.38 1.42 2.29 1.58 4.31 0.72 7.06 1.47 2.75 1.64 0.46 2.28Costs livestock (�1,000 Kes) 22.70 3.90 22.60 3.52 �0.18 5.25 24.40 4.64 33.80 6.41 9.35 7.91 9.52 9.49Land related investments (�1,000 Kes) 9.35 4.16 10.20 3.34 0.81 5.33 12.10 4.66 26.80 7.86 14.70 9.13 13.90 10.60Risk index (1–3) 2.12 0.04 2.10 0.04 �0.02 0.06 2.17 0.05 2.17 0.06 �0.01 0.08 0.02 0.10



Utz-certified enhances input-intensification of coffeeproduction and multi-certification opts for coffee renovationand the diversification of coffee outlets. This may point towarda life cycle of certification that starts with enabling farmers toproduce larger volumes for the Fairtrade market and gradu-ally shifting emphasis to improving yields and quality that

are demanded by Utz-certified outlets. Multi-certificationbecomes an option to overcome problems of over-certification and to diversify their outlets.Second, household welfare and livelihood effects of coffee

certification remain generally rather disappointing. Utz-certified farmers receive initially substantially higher prices

Table 5b. Multicertification-effect: matched difference-in-differences Fairtrade/Utz vs. non certified farmers


NC FT/Utz Diff NC FT/Utz Diff


Gross income other crops (�1,000 Kes) 4.04 0.59 1.90 0.49 �2.14 *** 0.76 0.74 0.45 0.14 0.11 �0.60 0.47 1.54 * 0.89Input costs other crops (�1,000 Kes) 3.19 0.37 1.85 0.26 �1.34 *** 0.46 3.02 0.60 1.30 0.44 �1.73 ** 0.74 �0.38 0.87Seed costs other crops (�1,000 Kes) 2.97 0.31 2.06 0.23 �0.91 ** 0.38 1.22 0.16 1.27 0.45 0.05 0.48 0.96 0.61Number of Livestock 9.34 0.86 6.90 1.44 �2.44 1.67 8.42 0.87 7.37 1.53 �1.05 1.76 1.39 2.43Costs livestock (�1,000 Kes) 43.60 5.08 22.80 3.42 �20.80 *** 6.13 50.40 6.16 31.00 5.80 �19.40 ** 8.46 1.36 10.40Land-related investments (�1,000 Kes) 12.80 2.63 10.70 3.84 �2.03 4.65 16.50 5.55 20.20 5.68 3.71 7.94 5.74 9.21Risk index (1–3) 2.13 0.04 2.09 0.04 �0.04 0.06 2.14 0.06 2.18 0.07 0.04 0.09 0.08 0.10

Means and Standard Errors are estimated by linear regression. Robust Standard Errors. Number of respondents = 147 (FT/Utz:61, NC:86). ***p < 0.01;**p < 0.05; *p < 0.1. Kernel-based Propensity Score Matching diff-in-diff. Estimation on the common support.

Table 4e. Utz-effect: matched difference-in-differences. Utz/Fairtrade vs. Fairtrade farmers




Have piped water (yes/no) 0.34 0.06 0.16 0.05 �0.19 ** 0.08 0.47 0.07 0.17 0.05 �0.30 *** 0.08 �0.11 0.12Have improved latrine (yes/no) 0.55 0.06 0.53 0.07 �0.01 0.09 0.50 0.07 0.81 0.05 0.32 *** 0.08 0.33 *** 0.13Number of organizations 1.22 0.05 1.36 0.06 0.14 * 0.08 0.51 0.07 0.24 0.06 �0.27 *** 0.09 �0.41 *** 0.12


Table 4f. Utz-effect: matched difference-in-differences. Utz/Fairtrade vs. Fairtrade farmers




Coffee sold in cherry form (kg) 720.00 80.50 583.00 84.10 �137.00 116.00 303.00 41.80 361.00 69.80 57.70 81.30 195.00 142.00

Dry coffee sold (kg) 50.10 5.06 34.40 6.63 �15.60 * 8.34 38.00 10.40 31.00 5.46 �7.07 11.70 8.58 14.40

Cherry coffee price (Kes/kg) 24.70 0.58 27.60 0.64 2.95 *** 0.86 50.60 4.40 52.30 5.48 1.75 7.02 �1.20 7.08

Dry coffee price (Kes/kg) 40.90 0.64 84.20 3.97 43.40 *** 4.02 163.00 14.30 168.00 11.90 5.41 18.60 �38.00 ** 19.10


Table 5a. Multicertification-effect: matched difference-in-differences Fairtrade/Utz vs. non certified farmers

Coffee plots 2009 2013 Diff-in-diff



Coffee area (acres) 0.60 0.04 0.56 0.04 �0.04 0.06 0.98 0.12 0.62 0.09 �0.36 ** 0.14 �0.33 ** 0.15

Number of mature coffee trees 317.00 18.10 287.00 21.00 �29.20 27.70 486.00 46.70 296.00 47.40 �190.00 *** 66.50 �160.00 ** 72.10

Number of coffee trees per acre 547.00 11.50 535.00 17.00 �11.70 20.50 572.00 28.10 566.00 64.00 �6.16 69.90 5.52 72.90

Coffee yield (kg/acre) 1.75 0.18 1.16 0.14 �0.58 *** 0.22 0.71 0.09 1.09 0.45 0.39 0.46 0.97 * 0.51

Coffee yield (kg/tree) 3.18 0.30 2.30 0.28 �0.87 ** 0.41 1.36 0.18 1.63 0.35 0.27 0.39 1.14 ** 0.57

Gross coffee income (�1,000 Kes) 19.60 1.98 21.40 2.98 1.78 3.58 11.30 1.37 24.70 5.88 13.40 ** 6.04 11.60 * 7.02

Gross dry coffee income (�1,000 Kes) 2.31 0.49 2.85 0.62 0.53 0.79 1.49 0.33 3.82 0.84 2.33 ** 0.91 1.79 1.20

Gross cherry coffee income (�100 Kes) 17.30 1.63 18.50 2.71 1.25 3.16 9.80 1.32 20.80 5.67 11.00 * 5.83 9.74 6.63

Input costs coffee (�1,000 Kes) 3.69 0.64 2.06 0.35 �1.63 ** 0.72 2.91 1.05 4.10 1.69 1.20 1.99 2.82 2.11

Hired labor coffee (yes/no) 0.80 0.04 0.68 0.06 �0.12 0.08 2.19 0.23 1.68 0.26 �0.51 0.35 �0.39 0.35

Investment new coffee (yes/no) 0.15 0.04 0.15 0.05 0.00 0.06 0.62 0.05 0.41 0.07 �0.21 ** 0.09 �0.21 ** 0.11



Table 5c. Multicertification-effect: matched difference-in-differences Fairtrade/Utz vs. non certified farmers




Gross income 314.00 41.30 160.00 31.80 �154.00 *** 52.10 250.00 30.40 174.00 35.60 �76.70 46.80 77.30 70.00

Share income from coffee (%) 23.00 3.02 37.50 5.11 14.50 ** 5.93 13.20 2.92 38.10 5.76 24.90 *** 6.46 10.30 8.77

Gross off farm income (�1,000 Kes) 134.00 24.10 158.00 42.90 23.10 49.20 113.00 15.80 75.20 21.80 �37.60 26.90 �60.70 56.10

Credit (�1,000 Kes) 42.40 9.88 40.10 16.00 �2.30 18.80 28.00 9.65 27.50 15.50 �0.49 18.30 1.82 26.30

Savings (�1,000 Kes) 56.00 7.27 31.50 7.68 �24.40 ** 10.60 18.90 5.53 34.60 11.40 15.60 12.70 40.10 ** 16.50

Total value of assets (�1,000 Kes) 136.00 20.20 182.00 62.00 46.10 65.20 134.00 52.90 209.00 68.40 75.70 86.40 29.60 108.00

Expenditure food (x1000 Kes) 4.91 0.21 4.70 0.37 �0.20 0.43 5.93 0.44 8.69 0.88 2.76 *** 0.99 2.97 *** 1.08

Expenditure education (�1,000 Kes) 15.30 2.58 9.15 2.81 �6.18 3.82 11.50 3.37 5.59 2.08 �5.94 3.96 0.24 5.50

Expenditure energy (�1,000 Kes) 2.23 0.27 2.49 0.45 0.25 0.53 3.47 0.40 5.13 0.78 1.65 * 0.88 1.40 1.03

Expenditure transport (�1,000 Kes) 4.37 0.56 4.16 0.76 �0.20 0.94 3.49 0.45 2.61 0.54 �0.88 0.70 �0.68 1.17

Expenditure total (�1,000 Kes) 58.40 6.57 52.40 9.40 �5.94 11.50 55.50 8.57 37.50 6.36 �18.00 * 10.70 �12.00 15.70

House investments (�1,000 Kes) 1.76 0.05 1.63 0.07 �0.13 0.08 1.81 0.05 1.69 0.06 �0.12 0.08 0.00 0.11

Econ. situation vs. �5 years (1–3) 1.89 0.10 2.11 0.13 0.23 0.17 1.47 0.09 1.37 0.10 �0.10 0.13 �0.32 0.21

Econ. situation vs. +5 years (1–3) 1.52 0.09 1.50 0.11 �0.02 0.14 1.29 0.07 1.19 0.07 �0.10 0.10 �0.08 0.17

Male gender index 18.50 0.48 19.00 0.90 0.46 1.02 13.10 0.72 13.20 0.90 0.13 1.15 �0.33 1.54

Female gender index 19.20 0.66 21.20 0.88 2.08 * 1.10 14.80 0.80 15.30 0.56 0.49 0.97 �1.58 1.47

Days lost due to poor health (log) 0.65 0.16 0.92 0.20 0.27 0.26 0.65 0.15 1.44 0.26 0.80 *** 0.30 0.53 0.40


Table 5d. Multicertification-effect: matched difference-in-differences Fairtrade/Utz vs. non certified farmers




Satisfaction tech. assistance (1–5) 3.03 0.17 3.54 0.18 0.51 ** 0.24 2.56 0.16 3.10 0.17 0.53 ** 0.23 0.02 0.34Satisfaction trade assistance (1–5) 2.89 0.16 3.28 0.19 0.39 0.25 2.65 0.16 3.04 0.16 0.39 * 0.23 0.00 0.34Identification index 2.56 0.07 2.61 0.08 0.05 0.11 2.57 0.06 2.47 0.09 �0.11 0.10 �0.15 0.15Force index 2.40 0.07 2.55 0.08 0.15 0.10 2.07 0.06 2.40 0.07 0.33 *** 0.09 0.19 0.14Total value of side selling 5.87 3.16 3.20 2.48 �2.67 4.02 25.00 24.10 0.00 �25.00 24.10 �22.40 24.50


Table 5e. Multicertification-effect: matched difference-in-differences Fairtrade/Utz vs. non certified farmers




Have piped water (yes/no) 0.04 0.02 0.15 0.05 0.11 ** 0.05 0.40 0.05 0.19 0.05 �0.22 *** 0.08 �0.32 *** 0.09Have improved latrine (yes/no) 0.59 0.05 0.56 0.07 �0.03 0.09 0.64 0.05 0.83 0.05 0.19 *** 0.07 0.22 * 0.12Number of organizations 1.52 0.06 1.37 0.07 �0.15 * 0.09 0.50 0.06 0.24 0.06 �0.26 *** 0.08 �0.11 0.12



for cherry and particularly for dry coffee deliveries, but theseprice differences disappear over time as all certified coffeesincrease in prices. Certified farms were able to improve theirsituation mainly through better prices for processed (dry)coffee, with declining differences between Utz and FT certifica-tion but increasing margins compared to non-certified farm-households. However, non-certified producers were also ableto adjust their coffee farming practices over time and gener-ated in addition a sizeable income from other farm- andnon-farm activities. The effect of certification on household

income thus remains fairly limited, since on average no morethan one-third to a quarter of farm revenues are derived fromcoffee sales and the realized coffee sales under certified condi-tions only represents up to ⅓ of the total produced volume.The expected net income effect from certification thus referto less than 10% of household income. 11

Third, none of the certification regimes has been able to gen-erate substantial changes in the structure of the coffee valuechain. Primary coffee producers still receive no more than6–8% of the consumer price 12 and options for providing

Table 5f. Multicertification-effect: matched difference-in-differences Fairtrade/Utz vs. non certified farmers




Coffee sold in cherry form (kg) 858.00 78.80 628.00 91.00 �230.00 * 120.00 453.00 60.50 397.00 79.60 �55.70 100.00 174.00 156.00

Dry coffee sold (kg) 59.60 7.54 37.10 6.92 �22.50 ** 10.20 43.90 6.08 33.70 5.73 �10.20 8.36 12.30 13.20

Cherry coffee price (Kes/kg) 20.50 0.70 28.90 0.62 8.37 *** 0.94 27.20 1.56 53.70 5.63 26.50 *** 5.84 18.20 *** 5.92

Dry coffee price (Kes/kg) 33.80 1.69 84.20 3.78 50.40 *** 4.14 47.60 4.72 166.00 11.80 119.00 *** 12.70 68.40 *** 13.40



pre-finance (as part of the contract) have hardly improved.Coffee cooperatives could slightly reinforce their service provi-sion to members and their bargaining position with traders,but many farmers increasingly rely on time side-sales (’hawk-ing’) to outside agents to overcome market outlet constraints.Over time, non-certified coffee producers demonstrate to beable to reach similar output and yield levels and initial advan-tages of certification tend to disappear gradually. Moreover,some improvements in village service provision are registeredparticularly for Fairtrade-certified producers, but linkages of

coffee farmers with other social or community organizationsgenerally deteriorated.In perspective, an ambivalent picture emerges from the

experiences with coffee certification in Central Kenya that israther far away from the image that is created by the labelson the back of the coffee package. While the different stan-dards certainly provide incentives for coffee farmers toimprove production, yields or quality, little attention is givento more substantial changes in the pre-finance and competitivemarket conditions. There is still a much to do for fair trade!

NOTES

1. Civic organizations like Rainforest Alliance provide certification andsupervision to several company-operated schemes.

2. African coffee production includes both Arabica and Robusta varietiesthat are mostly grown by smallholders and marketed through cooperativesocieties. Production of coffee declined in East Africa from 19.5 millionbags in 1997 to only 17.5 million bags in 2008, mainly due to poormanagement practices, aging of the coffee trees and losses due to damageby pests and diseases. Coffee prices are strongly fluctuating and show ageneral declining tendency during the last five years.

3. Many of these trainings were provided within the framework of so-called Farmers Field Schools (FFS).

4. Certification costs may involve initial payments of €1,500–3,000followed by yearly tariffs of € 700–1,200.

5. The cooperative sub-sector in Kenya is made up of 422 cooperativeunions representing about 570,000 smallholders cultivating about85,000 ha, equivalent to about 0.2 hectares a piece (KCTA, 2012).

6. We conditioned households in Mikari and Kiambaa to fall within theoverall range of number of trees for Ndumberi (after removing outliers),with a minimum of 50 and a maximum of 700 trees.

7. Despite all necessary prior arrangements, some respondents wereeither absent on the date of interview or unable to participate for variousreason (leaving questionnaires partially filled). For this reason, in additionto the original sample selected, a replacement sample of up-to 20% wasavailed to the enumerators. The replacement sample was also randomlygenerated from each cooperative’s sample frame less the selected sample.In a few cases a second replacement sample was necessary. This wasoccasioned by deaths or out migration and also due to problems withregisters of cooperative membership.

8. In the second round, 82 farmers could not be traced again andthus dropped out of the panel sample. Annex 1 provides an analysis

of potential attrition bias that appears to be mostly related tofarmers with inferior coffee production conditions, low savings, andreduced membership of local organizations. The remaining samplethus can be considered as an upper boundary of certificationoutcomes since more marginal farmers are subject to outmigrationfrom the area.

9. The indices used were constructed in the following way:The Forceindex is constructed using the mean score (1 = Disagree, 2 = Neutral,3 = Agree) of the replies to the following five statements: 1. Theassociation/cooperative is an efficient organization. 2. The association/cooperative is an organization that reacts efficiently in face of events. 3.The association/cooperative is a profitable organization. 4. Regarding themembers of the association/cooperative it is peaceful. 5. The association/cooperative is trustworthy.The Identification index is constructed usingthe mean score (1 = Disagree, 2 = Neutral, 3 = Agree) of the replies tothe following five statements: 1. I feel that the association/cooperative ismine. 2. A lot of the associates think that the association/cooperative isours. 3. I feel proud to belong to the association/cooperative. 4. I feel verycommitted to the association/cooperative. 5. My values and the values ofthe association/cooperative are the same.The Risk index is constructedusing the mean score (1 = Disagree, 2 = Neutral, 3 = Agree) of thereplies to the following six statements: 1. Every day I get more convincedthat whoever does not risk, does not earn. 2. In order to make somemoney, I am willing to risk and lose. 3. I only invest when I am certain thatI have a good return (reversed). 4. Investing in new crops is very risky, I’drather not do it (reversed). 5. I prefer to invest on something safe, althoughI don’t earn a lot, before investing on a risky asset, where I can earn a lotbut lose everything as well (reversed). 6. I would borrow money if I wasconvinced that investing in a business would give me good profits.TheGender indices (separately for male and female) are constructed using thesum score (1 = totally disagree, 5 = totally agree) of the replies to thefollowing five statements: 1. The biological nature of women makes themrestrict their tasks to the house keeping (reversed). 2. Family planning is aresponsibility of both men and women. 3. Men tend to get most positionsin the cooperative (reversed). 4. Women are better administrators of goodsand services. 5. There is normally equal treatment between men andwomen. For the asset index respondents were asked at what price they


would you sell the following items: Tractors, Trailers, Vehicles, Carts,Donkeys, Wheelbarrows, Ploughs, Borehole, Well, Sickle, Hand hoe,Chaff cutter, Spray pumps, Diesel pumps, Water tanks, Grinders,Beehives, Motorcycle, Bicycle, Radio/Tape recorder, Batteries, TV,Mobile Phones, Furniture, Mosquito nets, Solar Panels, Milking churns.The index was constructed using the sum score of the aforementionedvalues.

10. Results are available from the authors upon request.

11. De Janvry et al. (2012) find similar evidence of over-certification inCentral America, where only a quarter of the certified coffee can be soldunder premium conditions.

12. Calculations using the average price for dry coffee in our sample(KES 73.16/kg equals €0.61/kg) and consumer prices for coffee in TheNetherlands being €8.34/kg for ground coffee (http://statline.cbs.nl/StatWeb/) resulting in a 7.3% of the Dutch retail price going to the coffeefarmer. See also: Morisset (1998) and Johannessen and Wilhite (2010).

REFERENCES

Arnould, E. J., Plastina, A., & Ball, D. (2009). Does Fair Trade deliver onits core value proposition? Effects on income, educational attainment,and health in three countries. Journal of Public Policy & Marketing, 28,186–201.

Ashforth, B. E., Harrison, S. H., & Corley, K. G. (2008). Identification inorganizations: An examination of four fundamental questions. Journalof Management, 34, 325–374.

Bacon, C. (2005). Confronting the coffee crisis: Can fair trade, organic,and specialty coffees reduce small-scale farmer vulnerability in north-ern Nicaragua?. World Development, 33, 497–511.

Bacon, C. M., Mendez, E. V., Gomez, M. E. F., Stuart, D., & Flores, S. R.D. (2008). Are sustainable coffee certifications enough to secure farmerlivelihoods? The millennium development goals and Nicaragua’s FairTrade cooperatives. Globalizations, 5, 259–274.

Barham, B. L., Callenes, M., Gitter, S., Lewis, J., & Weber, J. (2011). FairTrade/organic coffee, rural livelihoods, and the ‘‘Agrarian Question”:Southern Mexican coffee families in transition.World Development, 39,134–145.

Barham, B. L., & Weber, J. G. (2012). The economic sustainability ofcertified coffee: Recent evidence from Mexico and Peru. WorldDevelopment, 40(6), 1269–1279.

Becchetti, L., & Costantino, M. (2008). The effects of fair trade onaffiliated producers: An impact analysis on Kenyan farmers. WorldDevelopment, 36, 823–842.

Blackman, A., & Rivera, J. (2010). The evidence base for environmentaland socioeconomic impacts of ‘‘sustainable” certification. Resourcesfor the Future: RF-DP-10-17.

Bolwig, S., Riisgaard, L., Gibbon, P., & Ponte, S. (2013). Challenges ofagro-food standards conformity: Lessons from East Africa andpolicy implications. European Journal of Development Research, 25,408–427.

Chiputwa, B., Spielman, D. J., & Qaim, M. (2014). Food standards,certification, and poverty among coffee farmers in Uganda. WorldDevelopment, 66(1), 400–412.

De Janvry, A., McIntosh, C., & Sadoulet, E. (2012). Fair Trade and freeentry: Can a disequilibrium market serve as a development tool?September, (http://irps.ucsd.edu/assets/001/503924.pdf, accessed May2014).

Deaton, A. (1997). The analysis of household surveys: A microeconometricapproach to development policy. World Bank Publications.

Dehejia, R., & Wahba, S. (1999). Causal effects in non-experimentalstudies: Reevaluating the evaluation of training programs. Journal ofthe American Statistical Association, 94(448), 1053–1062.

Elder, S. D., Zerriffi, H., & Le Billon, P. (2012). Effects of Fair Tradecertification on social capital: The case of Rwandan coffee producers.World Development, 40, 2355–2367.

Heckman, J. J., Ichimura, H., & Todd, P. (1997). Matching as aneconometric evaluation estimator: Evidence from evaluating a jobtraining programme. The Review of Economic Studies, 64(4),605–654.

Jaffee, D. (2007). Brewing justice. Fair trade coffee, sustainability, andsurvival. Univ of California Press.

Jena, P. R., Chichaibelu, B. B., Stellmacher, T., & Grote, U. (2012). Theimpact of coffee certification on small-scale producers’ livelihoods: Acase study from the Jimma Zone, Ethiopia. Agricultural Economics, 43(4), 429–440.

Johannessen, S., & Wilhite, H. (2010). Who really benefits from fairtrade?An analysis of value distribution in fairtrade coffee. Globalizations, 7(4),525–544.

Kasente, D. (2012). Fair Trade and organic certification in value chains:Lessons from a gender analysis from coffee exporting in Uganda.Gender & Development, 20, 111–127.

KCTA. (2012). Kenya Coffee Directory. Nairobi: Kenya Coffee TradersAssociation.

Le Mare, A. (2008). The impact of fair trade on social and economicdevelopment: A review of the literature. Geography Compass, 2,1922–1942.

Lyon, S. (2008). We want to be equal to them: Fair-trade coffeecertification and gender equity within organizations. Human Organi-zation, 67, 258–268.

Mendez, V. E., Bacon, C. M., Olson, M., Petchers, S., Herrador, D.,Carranza, C., ... Mendoza, A. (2010). Effects of Fair Trade and organiccertifications on small-scale coffee farmer households in CentralAmerica and Mexico. Renewable Agriculture and Food Systems, 25,236–251.

Morisset, J. (1998). Unfair trade? The increasing gap between world anddomestic prices in commodity markets during the last 25 years. TheWorld Bank Economic Review, 12(3), 503–526.

Muradian, R., & Pelupessy, W. (2005). Governing the coffee chain: therole of voluntary regulatory systems. World Development, 33,2029–2044.

Neilson, J. (2008). Global private regulation and value-chain restructuringin Indonesian smallholder coffee systems. World Development, 36(9),1607–162.

Nelson, V., & Pound, B. (2009). The last ten years: a comprehensivereview of the literature on the impact of Fairtrade. Natural ResourcesInstitute, 1–48.

Petkova, I. (2006). Shifting regimes of governance in the coffee market:From secular crisis to a new equilibrium?. Review of InternationalPolitical Economy, 13, 313–339.

Philpotts, S. M., Bichier, P., Rice, R., & Greenberg, R. (2007). Field-testing ecological and economic benefits of coffee certification pro-grams. Conservation Biology, 21, 975–985.

Raynolds, L. T., & Wilkinson, J. (2007). Fair trade: The challenges oftransforming globalization. London/New York: Routledge.

Raynolds, L. T. (2009). Mainstreaming Fair Trade coffee: From partner-ship to traceability. World Development, 37(6), 1083–1093.

Renard, M.-C. (2005). Quality certification, regulation and power in fairtrade. Journal of Rural Studies, 21, 419–431.

Riisgaard, L., Michuki, G., Gibbon, P., Bolwig, S., Warring, N., & Rantz,L. (2009). The performance of voluntary standard schemes from theperspective of small producers in East Africa. Traidcraft, Gateshead,UK: Danish Institute for International Studies report prepared forTraidcraft (UK).

Ronchi, L. (2002). The impact of Fair Trade on producers and theirorganizations: A case study with Coocafe in Costa Rica. Sussex: PolicyResearch Unit: University of Sussex.

Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of thepropensity score in observational studies for causal effects. Biometrika,70(1), 41–55.

Ruben, R. (2008). The impact of fair trade. Wageningen, The Netherlands:Wageningen Academic Publishers.

Ruben, R., & Fort, R. (2012). The impact of fair trade certification forcoffee farmers in Peru. World Development, 40, 570–582.

Ruben, R., & Verkaart, S. (2011). Comparing fair and responsible coffeestandards in East Africa. In A. H. J. Helmsing, & S. Vellema (Eds.),Value chains, social inclusion and economic development (pp. 61–81).London, UK: Routledge.

http://statline.cbs.nl/StatWeb/

http://statline.cbs.nl/StatWeb/

http://refhub.elsevier.com/S0305-750X(15)00202-8/h0005









































http://irps.ucsd.edu/assets/001/503924.pdf












































































































b r P-Value

HouseholdExpenditure food (�1,000 Kes) �0.02 0.04 0.54Expenditure transport (�1,000 Kes) �0.02 0.02 0.37Days lost due to poor health(logarithm)

0.00 0.06 0.99

Share income from coffee (%) 0.00 0.00 0.43Economic situation vs. 5 years ago(1–3)

�0.11 0.10 0.27

Savings (�1,000 Kes) 0.00 0.00 0.52

FarmCoffee yield (kg/acre) 0.04 0.25 0.86Coffee yield (kg/tree) �0.03 0.14 0.81Coffee income (�1,000 Kes) 0.00 0.00 0.93Income other crops (�1,000 Kes) 0.02 0.02 0.16Input costs other crops (�1,000 Kes) �0.04 0.04 0.41Seed costs other crops (�1,000 Kes) �0.06 0.05 0.24Hired labor coffee (yes/no) 0.01 0.19 0.96Coffee area (acres) 0.28 0.30 0.36Investment in new coffee (yes/no) 0.03 0.22 0.90

CooperativeIdentification index 0.12 0.18 0.52Force index �0.13 0.17 0.47Satisfaction technical assistance (1–5) 0.06 0.09 0.49Satisfaction trade assistance (1–5) 0.00 0.08 0.95

Chain


Ruben, R., & Zuniga, G. (2011). How standards compete: Comparativeimpact of coffee certification schemes in Northern Nicaragua. SupplyChain Management: An International Journal, 16, 98–109.

Rueda, X., & Lambin, E. F. (2012). Linking globalization to local landuses: How eco-consumers and gourmands are changing the colombiancoffee landscapes. World Development, 41(1), 286–301.

Rueda, X., & Lambin, E. F. (2013). Responding to Globalization: impactsof certification on Colombian small-scale coffee growers. Ecology &Society, 18.

Saenz Segura, F., & Zuniga-Arias, G. (2008). Assessment of the effect ofFair Trade on smallholder producers in Costa Rica: A comparativestudy in the coffee sector. In R. Ruben (Ed.), The impact of fair trade(pp. 117–135). Wageningen, The Netherlands: Wageningen AcademicPublishers.

Salami, A., Kamara, A. B., & Brixiova, Z. (2010). Smallholder agriculturein East Africa: trends, constraints and opportunities. Working PaperSeries no. 105. Tunisia: African Development Bank, Tunis.

Valkila, J. (2009). Fair Trade organic coffee production in Nicaragua—Sustainable development or a poverty trap?. Ecological Economics, 68,3018–3025.

Valkila, J., Haaparanta, P., & Niemi, N. (2010). Empowering coffeetraders? The coffee value chain from Nicaraguan fair trade farmers toFinnish consumers. Journal of Business Ethics, 97, 257–270.

Valkila, J., & Nygren, A. (2010). Impacts of Fair Trade certification oncoffee farmers, cooperatives, and laborers in Nicaragua. Agricultureand Human Values, 27, 321–333.

Van der Vossen, H. (2005). A critical analysis of the agronomic andeconomic sustainability of organic coffee production. ExperimentalAgriculture, 41, 449–473.

Weber, J. G. (2011). How much more do growers receive for fair trade-organic coffee?. Food Policy, 36(5), 678–685.

World Bank. (2013). Time to shift gears – Accelerating growth and povertyreduction in the new Kenya. Nairobi/Washington: The World Bank.

Coffee sold in cherry form (kg) 0.00 0.00 0.80Dry coffee sold (kg) 0.00 0.00 0.53Cherry coffee price (Kes/kg) �0.02 0.02 0.32Dry coffee price (Kes/kg) 0.00 0.00 0.36

VillageHave piped water (yes/no) 0.35 0.22 0.11Have improved latrine (yes/no) �0.14 0.19 0.47Number of organizations �0.17 0.19 0.35

Constant 0.03 0.72 0.97Log pseudo-likelihood �154.84Number of observations 299Wald chi2 (26) 15.36Prob > chi2 0.95Pseudo R2 0.05

APPENDIX A. ATTRITION ANALYSIS

The baseline survey in 2009 included 300 farmers while in2013 during the endline survey 218 of the same farmers couldbe traced again. This means 82 farmers dropped out of thepanel sample. If the members who dropped out of our paneldiffer systematically from those who stayed in, then the datasetof continuing farmers is no longer representative of the origi-nal population. Thus results might be affected by attritionbias. We analyze the possible causes of this attrition with aprobit analysis (0 if no attrition and 1 if farmers droppedout of the sample) using several socioeconomic variables asexplanatory variables.

The pseudo R-squared from the attrition probit suggests thatbaseline variables explain about 4.8% of panel attrition during2009–13. The P-values show none of the variables in the attri-tion probit are statistically different from zero. We alsoperformed a Wald-test to see whether the variables jointly

equal zero, resulting in a Chi-squared statistic of 15.36 with26� of freedom (P = 0.9503). Attrition bias does not seem tothreaten the representativeness of our results.

ScienceDirectAvailable online at www.sciencedirect.com











































The Ambivalent Impact of Coffee Certification on … · The Ambivalent Impact of Coﬀee...

Documents

Transcript of The Ambivalent Impact of Coffee Certification on … · The Ambivalent Impact of Coﬀee...