Fair Trade: in or out the Market?

Claire Chambolle∗ and Sylvaine Poret†

December 2005

Abstract

This paper focuses on a sustainable perspective of fair trade con-

cept. We propose a simple model to provide some theoretical argu-

ments in the debate about the sale of fair trade goods in the large-scale

distribution. The main hypothesis of our model is related to the obser-

vation that some consumers are willing to pay a premium for fair trade

products. We underline that the key variable for a retailer is not the

percentage of consumers who are willing to pay for a fair good but how

much the fair trade likers are willing to pay for it. Moreover, for some

values of the parameters, it exists a unique equilibrium where the fair

product is the only good sold and where fair trade likers', traditional

consumers' and retailer's surplus are greater than the ones when only

the traditional product is sold by the retailer.

Keywords: D21, L22, L31.

JEL classi�cation: fair trade, ethical premium, second-degree price

discrimination, distribution.

∗INRA-LORIA and Laboratoire Ecole Polytechnique-CECO. INRA-LORIA, 65 boule-vard de Brandebourg 94205 Ivry-sur-Seine cedex - France. E-mail: chamboll@ivry.inra.fr.

†INRA-LORIA and CREST-LEI. INRA-LORIA, 65 boulevard de Brandebourg 94205Ivry-sur-Seine cedex - France. E-mail: poret@ensae.fr.

1

1 Introduction

With the growth of fair trade, there is today a large debate about the in-sertion of fair trade products into the conventional distribution network.1

Indeed, fair trade organizations have a double purpose: to help producersfrom the South improve their living conditions and to transform interna-tional trade relations.2 The issue of the distribution and more particularlythe insertion of the fair trade into the large-scale distribution prompt a de-bate on priorities within this movement.

Fair Trade de�nes itself as an alternative approach to conventional inter-national trade. It is a trading partnership, based on dialogue, transparencyand respect, which seeks greater equity in international trade. It contributesto sustainable development by o�ering better trading conditions to, and se-curing the rights of, marginalized producers and workers, especially in theSouth.

Fair trade started as a grassroots movement in the late 1960s in Eu-rope. The aim was to alleviate poverty in the South by building direct,sustainable relationships with disadvantaged producers and providing fairaccess to markets in the North, using a strategy of �trade, not aid�. Thealternative trade organizations (Non-Governmental Organizations) created aparallel distribution network with specialty stores managed as cooperativesand sta�ed by volunteers.3 In 1988, a working with a Mexican cooperativeof co�ee producers, who requested help in marketing their product in Eu-rope, and a collaborator of a Dutch NGO conceived the idea of a fair tradelabel (Renard, 2003). Products bought, traded and sold respecting fair tradeconditions would qualify for a label that would make them stand out amongordinary products on store shelves and would allow any company to get in-volved in fair trade. Thus, the �Max Havelaar� label was established in theNetherlands. In the late 1980s, similar non-pro�t fair trade organizationsbegan labelling fair trade products to facilitate their entry into conventionalmarkets (Raynolds, 2000). In 1997, the creation of the Fair trade LabellingOrganisation International (FLO-I) united many of these labelling initia-

1See, for example, the virtual symposium of some French fair trade actorsin march 2004, � La grande distribution : l'avenir du commerce équitable ? �http://www.changerdere.com/accueil/.

2IFAT (International Federation for Alternative Trade) http://www.ifat.org3Oxfam and Twin Trading (Great Britain), Stichting Ideele Import (Netherlands) or

Artisans du Monde (France) are some alternative trade organizations.

2

tives. FLO-I is now responsible for setting international fair trade standardsfor certifying production, trade and labelling of a certain number of products.

Fair trade concerns agriculture and handicraft products, but only agricul-tural ones more or less transformed (co�ee, cocoa, tea, honey, sugar, rice, ba-nanas, and orange juice) can be labelled. A third-party certi�cation agency,member of the FLO network who harmonizes the quality norms, professional-izes the control mechanisms and creates a certi�cation structure, is in chargeof making sure that fair trade criteria are respected.4. The movement is aglobal network of producers, wholesalers, retailers, certi�cation agencies, andconsumers who commit themselves to certain fair trade guidelines. Trans-parency is the core issue at all the levels of the production and distributioncircuit. It is fundamental to maintain reputation which is the crucial com-petitive factor for fair trade products and labels give informations aboutproducers. Contrary to the other certi�cation schemes, which focus strictlyon conditions at the point of production, fair trade label criteria is unique inthat it covers both trade and production conditions (Raynolds, 2000).

There are general criteria for what the label guarantees. The producershave to follow certain agreements to qualify the products they produce asfair trade. Small scale farmers/producers can only be certi�ed Fair Trade ifthey have formed cooperatives, associations or other organizational entitieswhich are democratically controlled and contribute to the social and economicdevelopment of its members. The producers must tend towards the useof environmentally sustainable agricultural practices and a production ofquality. Trading standards stipulate that importers have to pay a price toproducers that covers the costs of sustainable production and living, pay apremium that producers can invest in development, partially pay in advanceso that the producers can purchase the necessary raw materials to completean order without falling into debt, establish long-term working relationshipsand contracts with producers. Certi�cation costs are born by importers,not by producers like in organic certi�cation. Finally, there are product-speci�c standards for each good that determine minimum quality, price, andprocessing requirements that have to be complied with.5

4For example, in the United-States, fair trade products bear the �Fair Trade Certi�ed�label and the �Fair Trade Federation� logo. TransFair USA is the third-party certi�cationagency that places the �Fair Trade Certi�ed� label on co�ee, chocolate, cocoa, tea, bananas,and other fruits.

5Fair trade co�ee producers are guaranteed that they will receive at least US$1.26. Ifthe world market price is higher than this rate, importers pay a premium of US$0.05 per

3

The percentage of fair trade was estimated at a mere 0.01% of total worldtrade. The annual aggregate net retail value of all fair trade products sold inEurope through alternative channels and supermarkets would exceed =C260million, of which =C210 million for labelled products (EFTA, 2001). Fair tradeproducts are sold in the 2,700 or so world shops in Europe (18 countries) andthey are available too in more than 43,000 supermarkets throughout Europethanks to the labeling schemes. In North America (the US and Canada) andthe Paci�c Rim (Australia, New Zealand and Japan) fair trade market is laterand less developed than the European one, but it is expanding much morerapidly. Total fair trade sales in these regions increased 37% during 2002,from US$183 million to US$251 million (FTF, 2003). In 2004, there were433 certi�ed producer organizations. In addition, the number of registeredtraders increased from 297 from 2003 to 406 at the end of 2004 (see on thewebsite of FLO-I, http://www.fairtrade.net).

More and more consumers know the concept of faire trade and declarethey are willing to pay a higher price for a product with fair trade criteria. InGermany, according to recent market surveys, 40% of the population believefair trade is a good idea and 11% already buy TransFair-labelled tea or co�ee,although the market share for those products is just 2% and 1% respectively.Almost 37% of Germans are willing to pay a higher price for co�ee if itis guaranteed that the bene�ts are distributed to producers in developingcountries (EFTA, 2001). Nevertheless, farmers groups produce more quantitythan they can sell to fair trade markets, the outlets are still too low toabsorb the potential production. In a French study (AlterEco, 2004), to thequestion �why have you never bought fair trade products?� 39.1% of the 495questioned persons argue the lack of information, 36.7% say not to know aretail outlet, 20.3% advance a high price, and among the other arguments(32%), some respondents mention the lack of signing in retail outlets. Thus,fair trade is curbed by the lack of market opportunities and its future dependson the consumers' better knowledge.

The introduction of fair trade products in large-scale distribution is asolution to this double objective.6 But the viable niche markets created byfair trade attract dominant actors of the agri-food industry and of the large-scale distribution, who develop some strategies to bene�t from the image

pound more. Certi�ed organic co�ee gets a further premium of US$0.15 per pound.6For example, in France in 2003, 66.9% of food purchases were made in supermarkets

(INSEE, 2004).

4

associated with the fair trade values. At the moment, there appear someparallel labels based on weaker criteria, such as �Bio Equitable� in France.Some distributors create their fair private labels, certi�ed or not by fair tradelabel. Thus, incorporate the large-sale distribution represents some risks.Given the great power of the distributors, they may impose their conditions:wholesale price cutting pressure, creation of less strict labels, risk of delisting,lack of long term contracts, and capture of the most lucrative niche. For somefair trade organizations, as Fédération Artisans du Monde, it is impossible toparticipate in large-scale distribution, because the distributors are not ethicper se. Fair trade organizations should withdraw from large-scale distributionsince the multinationals that structure it are part of the problem that fairtrade is trying to �ght against. The debate is acute in fair trade organizations,because some fair traders exclusively deal with supermarkets.

We propose a simple model to provide some theoretical arguments inthis debate. We want to explain the motivation of a retailer to sell somefair products and to show how a fair trade agency can react to the di�er-ent strategies of the retailer's according her preferences between fair pricereceived by producers and quantities of fair product sold. The main hypoth-esis of your model is related to the observation that some consumers, calledfair trade likers, are willing to pay a premium for fair trade products, func-tion of producers' revenue. We underline that the key variable for a retaileris not the pourcentage of consumers who are willing to pay for a fair goodbut how much the fair trade likers are willing to pay for it. Moreover, itexists an equilibrium where the fair product is the only good sold and wherethe surplus of fair trade likers and traditional consumers are greater thanthe ones when only the traditional product is sold by the retailer. From atheoretical point of view, this paper is related to several branches of the lit-erature: vertical relationships, second degree price discrimination. Fair tradeis the object of some theoretical approaches (Adriani and Becchetti (2002),Becchetti, Solferino (2003), Immordino (2002)), but to our knowledge thesupply of fair products by large-scale distribution is not modeled.

This paper proceeds as follows. In the next section, we describe theassumptions of the model and the possible strategies of a retailer in front ofthe possibility to sell a fair product. Section 3 analyses the response of thefair trade certi�er in terms of fair wholesale price and consumers' surplus.Section 4 presents a conclusion.

5

2 The distributor in front of fair product

2.1 The model

On the basis of a sample of 808 Belgian citizens and using conjoint analysis,De Pelsmacker, Driesen, Rayp (2005) study the importance of a fair tradelabel in the co�ee buying decision and the willingness to pay for such a fairproduct. On average, a fair trade label is considered as the second most im-portant co�ee attribute, behind the brand and as important as aroma. Theauthors divide co�ee buyers into four segments on the basis of the relativeimportance they attach to various co�ee attributes (Brand, Blending, Fla-vor, Package, Fair Trade Label). 11% of the respondents expressed a highpreference for fair trade label and are named �fair trade lovers�. For 40% ofthem, called the �fair trade likers�, fair trade label comes out as the mostimportant attribute but not as the prominent one. The two other segmentsare the ��avor lovers� (24%) and the �brand lovers� (25%).

Based on this empirical study, we model the sell of fair trade productsas follows. A single retailer can launch a fair trade product, in additionto the traditional product.7 The both products are quality equivalent. Weassume that the retailer faces a unitary market consisting of two segmentsof consumers. A fraction λ of the market are fair trade likers and 1− λ aretraditional consumers or products seekers. Furthermore, fair trade likers givevalue to producers' welfare and they are willing to pay a premium for a fairtrade product that is related to its wholesale price wf . As so, we de�ne thedemand function as follows:8

d(p) =

{1 + αwf − p for the λ fair trade likers

1− p for the 1− λ product seekers.

The parameter α re�ects the intensity with which fair trade likers enhancethe value of the producers' revenue, that is, the wholesale price they receive.We assume that α ∈ (1, 0). Based on Italian scanned data and using theprice hedonic model, Maietta (2003) shows that the marginal value of the

7In many countries, the food product distribution chain is controlled by a very re-stricted number of actors, which gives a high market power to distributors (see Allain andChambolle (2003) and Renard (2005)).

8In a paper on the subject of private labels, Soberman and Parker (2004) distinguishtwo segments of consumers: product seekers and brand seekers. The latter are willing topay a premium for the national brand equal to the level of advertising realized by the �rm.

6

ethical content in co�ee consumption is equal to =C2.36 per kilo, with anaverage price of around =C7.5 per kilo for co�ee. De Pelsmacker, Driesen,Rayp (2005) evaluate the willingness to pay for a fair trade label. 35% ofthe respondents are willing to pay a price premium of 10%, 17% a premiumof 20%, and 10% a premium equal to the average actual premium (27%). Itis clear that greater the premium price lower the number of persons willingto pay it. The data exhibit a decreasing relation between the intensity withwhich fair trade likers enhance the value of the producers' revenue and thenumber of fair trade likers. Including this relation in the model does notchange results.

The game consists of three stages. In the �rst stage, the traditionalproducer and the faire trade certi�er specify their wholesale prices simul-taneously, wf and wt per unit purchased. In the second stage, the retailerdecides on the strategy of reference list. The retailer can supply one of thetype of products, fair trade (Strategy 3) or traditional product (Strategy 2),or both (Strategy 1). In the third stage of the game, the retailer sets pricesfor the product(s) he sells. To simplify the analysis without loss of generality,we normalize to zero the marginal cost of production and retailing for bothproducts.

Retailer's objective function according the strategy choose at the secondstage of the game are as following:

• Strategy 1: (T, F)

πR1 (pf , pt, wf , wt) = λ(1+αwf−pf )(pf−wf )+(1−λ)(1−pt)(pt−wt) (1)

with pf the retail price for the fair product, pt the retail price for thetraditional one and wt its wholesale price.

• Strategy 2: (T)πR

2 (pt, wt) = (1− pt)(pt − wt) (2)

• Strategy 3: (F)

πR31(pf , wf ) = [λ(1 + αwf − pf ) + (1− λ)(1− pf )](pf − wf )

when pf < 1

πR32(pf , wf ) = λ(1 + αwf − pf )(pf − wf )

when pf > 1

(3)

We solve the game by backward induction.

7

2.2 The analysis

We �rst characterize equilibrium at the stage 2 subgame in which the retailerchooses product(s) he refers.

Proposition 1. At the second stage, the choice of the retailer depends ontwo thresholds of the fair trade product wholesale price,

wf1/31=

1−√

1− (1− α2λ)wt(2− wt)

1− α2λand wf1/2

=wt

1− α.

(i). If wf < wf1/31(wt, α, λ), then the retailer sells only the fair product, at

a retail price such as all consumers buy this product (strategy 3 case 1- (F)).

(ii). If wf1/31(wt, α λ) < wf < wf1/2

(wt, α), then the retailer sells both prod-ucts (strategy 1 - (T,F)).

(iii). If wf > wf1/2(wt, α), then the retailer sells only the traditional product

(strategy 2 - (T)).

Proof. See the appendix A.

These results are relatively intuitive and are due to two e�ects, repre-sented by the derivative of the retailer's pro�t in the fair trade wholesaleprice.

∂πR1 (wf ,wt)

∂wf= λ

2[−(1− (1− α)wf ) + α(1− (1− α)wf )]

= [−qf + αqf ] < 0,(4)

with qf the quantity sold of the fair trade product in the strategy 1. The�rst e�ect is direct: a greater wholesale price for a product puts the retailero� distributing it.9 The second e�ect is the retailer's ability to discriminatebetween the both segments of consumers. This e�ect reduces the negativee�ect of the �rst one. As some consumers give value to the producers' rev-enue, that is, the wholesale price, the retailer can use that and discriminatesbetween consumers.

The equilibrium at the stage 2 subgame is represented by Figure 1 withwt = 0.5 and λ = 0.4. If the fair trade wholesale price is so low (wf <

9We also �nd this e�ect in the case of the traditional good:∂πR

1 (wf ,wt)∂wt

= −qt < 0, withqt the quantity sold by the traditional producer in the strategy 1.

8

1-wt 1a

1ÅÅÅÅÅ2

1

2

wf

wf1ê2

wf1ê2b

wf1ê31wf1ê31b

HTL

HT, FLHFLavec pf<1

Figure 1: Equilibrium at the stage 2

wf1/31(wt, α, λ)) that it is very inferior to the traditional product wholesaleprice wt (wf1/31(wt, α, λ) < wt), the retailer opts for the fair product only.The dominant e�ect is the direct one related to the wholesale price.

In the case of an intermediate value (wf1/31(wt, α, λ) < wf < wf1/2(wt, α)),the retailer sells both products, even if the fair wholesale price is lower thanthe traditional product one. Indeed, in the strategy 1,

pf − pt =1

2(wf − wt + αwf ) .

Even if the fair wholesale price is lower than the traditional product one, thefair retail price can be greater than the traditional product one, because of theethical premium and the possibility of discrimination for the retailer. Thisis all the more true since the consumer enhances the value of the wholesaleprice receives by producers (high α). In this case, the retailer fully takesadvantage of the discrimination between the both segments of consumers.

9

Even when the fair wholesale price is greater than the traditional productone, the bene�t of the discrimination is higher than its cost. This area isall the larger since consumers enhance the value of the fair trade productwholesale price and, in a lesser measure, since the number of fair trade likersis great. Intuitively, the two fair wholesale prices thresholds are increasingin the traditional wholesale price and wf1/2 is increasing in α. But wf1/31 isdecreasing in α and in λ. If the fair trade likers are willing to pay more forfair products and/or if the number of fair trade likers rises, a retailer who soldonly the fair product may, after this change, sell the two types of products.This is because the retailer can better discriminate between consumers andincrease the fair retail price.

Beyond the threshold wf1/2(wt, α), the retailer refers only the traditionalproduct. Even if the discrimination is related to the fair wholesale price, thislatter is too high.

In Figure 1, we compare the two fair wholesale prices thresholds to those,indexed by b, obtained using a simpler demand function

db(p) =

{1 + α− p for the λ fair trade likers

1− p for the 1− λ product seekers,(5)

where the valuation of fair trade is independent of producers' revenue. Withthis formulation, we do not obtain the discrimination e�ect, related to αin Equation 4. Indeed, with a demand independent of the wholesale price,∂πR

1 (wf ,wt)

∂wf= −λ

2[1 + α − wf ] = −qf < 0. In this way, we show that wf1/31

is always greater than wbf1/31 and that wf1/2 is greater than wb

f1/2 only whenα > 1− wt.

At the �rst stage of the game, the traditional producer chooses his whole-sale price, wt. If the retailer refers the traditional product, traditional pro-ducer's pro�ts are as follows. If the retailer chose the strategy 1, πT

1 (wt) =

(1− λ)wt(1−wt)2

and in the strategy 2, πT2 (wt) = wt(1−wt)

2. Thus, for all strate-

gies chose by the retailer, the best response of the traditional producer isw∗

t = 12.

3 The response of the fair trade certi�er

At the �rst stage, the fair trade certi�er sets the wholesale price of the fairproduct. With respect to her objective, we consider that she acts as a tradeunion.

10

3.1 The fair trade agency utility

We assume that fair trade certi�er utility, U(wf , qf ), depends positively onfair wholesale price, which determines the revenue of the small producers,and fair quantities sold, which represent the number of poor producers in thefair trade network. It is speci�ed in an utilitarian form:

U(wf , qf ) = (wf − wmin)β(qf )

1−β,

and β is a wholesale price preference parameter, with β ≤ 1 and wmin theminimum wholesale price. We assume that the latter is equal to the produc-tion marginal cost, speci�ed to zero. When β = 0.5, the utility of the fairtrade certi�er is equal to the total pro�t of small producers.

In strategies 1 and 3 case 2, the objective function of the fair trade certi�eris as follows

U1(wf ) = U32(wf ) = (wf )β

(λ(1− (1− α)wf )

2

)1−β

.

In strategy 3, when the fair retail price is lower than 1 (case 1), the fair tradecerti�er utility is

U31(wf ) = (wf )β

((1− (1− αλ)wf )

2

)1−β

.

These functions are concave in wf . The two �rst order conditions allowus to obtain the optimal wholesale prices for the fair trade agency withoutconstraint,

wf31 =β

1− αλ= arg max

wf

U31(wf ) and wf1 =β

1− α= arg max

wf

U1(wf ),

increasing in parameters α, β, and λ for wf31 .

3.2 The results

Given the fair trade certi�er's preferences and the retailer's strategies, we cande�ne the fair wholesale price that the fair trade certi�cation agency shouldpropose to the retailer in order to see the fair product on his shelves.

11

Proposition 2. The equilibrium fair wholesale price w∗f for the fair trade

certi�er and the equilibrium strategy for the retailer depend on the ethicalpremium parameter, α, and on the wholesale price preference parameter, β.

(i). If α < 12, the fair product is the unique product sold (Strategy 3 case 1

-(F)) and w∗f depends on β.

• If β < β, w∗f = wf31 < 1

2.

• If β > β, w∗f = wf1/31

= 2−√

1+3α2λ2(1−α2λ)

< 12.

(ii). If α > 12, then there exists a threshold value of β, β∗, such that

(a) if β < β∗, the equilibrium strategy is the strategy where the uniqueproduct sold by the retailer is the fair one (Strategy 3 case 1 -(F))and

• if β < β, w∗f = wf31 < 1

2,

• if β > β, w∗f = wf1/31

= 2−√

1+3α2λ2(1−α2λ)

< 12.

(b) If β > β∗, the equilibrium strategy is the strategy where both prod-ucts are sold by the retailer (Strategy 1 -(T,F)) and

• if β∗ < 1− α and β∗ < β < 1− α, w∗f = 1,

• if max{β∗, 1− α} < β < 12, w∗

f = wf1 > 1,

• if β > max{β∗, 12}, w∗

f = wf1/2= 1

2(1−α)> 1.

Proof. See the appendix B.

Figure 2 depicts these results. If the ethical premium parameter is low,the unique way for fair trade certi�er to see fair products marketed by theretailer is to propose a wholesale price lower than the traditional productone. The retailer can sell in any way the product that it is equitable ornot, because consumers buy the product even if it is not fair. Thus, if thelabel enables him to obtain the product at the wholesale price lower thanthe one proposed by the traditional producer, he sells only the fair product.When the fair premium parameter (α < 1

2) is low and when the wholesale

price preference parameter β is low, the fair trade agency can propose tothe retailer a fair wholesale price increasing in her wholesale price preferenceparameter. But, above a certain level of this parameter, β, the retailer refuses

to sell the fair product if the wholesale price is greater than wf1/31=

eβ1−αλ

.

12

1ÅÅÅÅÅ2

1a

1ÅÅÅÅÅ2

1

b

HT,FL_wf1

HT,FL_wf1ê2HFL_wf1ê31

HFL_wf1ê31HFL_wf31 HFL_wf31

bè

bèè

b*b*b

λ = 0,4

Figure 2: Equilibrium at the stage 1

The fair trade agency can propose a relatively high wholesale price (higherthan 1) only when the fair trade likers are willing to pay a high premium forfair trade products, because only in this case it is pro�table for the retailerto discriminate between fair trade likers and traditional consumers. Table 1gives equilibrium retail prices and retailer's pro�ts and we can see that inall equilibrium strategies retail price is increasing in α. In other words, theretailer can bene�t from discrimination only if the fair trade likers give ahigh valuation to producers' revenue whatever their number. Thus, the mostimportant parameter for the fair trade is not the number of fair trade likers,λ, but the value they give to small producers' revenue, α.

When the willing to pay an ethical premium is relatively high, the fairtrade certi�er faces two di�erent wholesale price strategies according to herwholesale price preference parameter. If the fair trade agency gives a relativehigh value to the producers' individual revenue, she sets a high fair wholesale

13

EquilibriumStrategy

w∗f

pf πR

(T, F )wf1/2

= 12(1−α)

12

+ 1+α4(1−α)

> 54

116

(T, F )

wf1 = β1−α

12

+ β(1+α)2(1−α)

> 34

116

< λ(1−β)2

4+ (1−λ)

16< 3λ+1

16

(T, F )wf = 1

54

< 1 + α2

< 32

116

< α2λ4

+ (1−λ)16

< 3λ+116

(F )

wf1/31= 2−

√1+3α2λ

2(1−α2λ)

34

<1+w∗f (1+αλ)

2< 7

8116

< [1−eβ]2

4< 1

4

(F )

wf31 = β1−αλ

12

< 12

+ β(1+αλ)2(1−αλ)

< 78

116

< [1−β]2

4< 1

4

Table 1: Equilibrium prices and retailer's pro�ts

price and the retailer markets both types of products. If she prefers guaranteehigher total quantities sold, she has to propose a low fair wholesale price toexclude her competitor.

When both products are sold at the equilibrium and when the wholesalepreference parameter is relatively low (β∗ < β < 1 − α), the fair tradeagency proposes a wholesale price equal to the unit. When the wholesalepreference parameter increases, the equilibrium wholesale price increases too,until a certain level of β. When β > max{β∗, 1

2} , the fair trade agency is

constrained in the o�er she can make to the retailer. If the agency's o�er wasgreater than wf1/2

, the retailer would sell only the traditional product (seeProposition 1 and Table 1), because otherwise his pro�t would be lower than116, retailer's pro�t when he sells only the traditional good. In addition, the

retailer's pro�t does not change or, in many cases, rises with the introductionof the fair good in the market.

The threshold value of β, β∗, is decreasing with the pourcentage of the fairtrade likers, λ, and decreasing with the ethical premium parameter, α. Morethe number of fair trade likers or more they give value to producers' revenue,more the retailer can discriminate between the two types of consumers andwant to market both types of products.

We compare these results to those, indexed by b, obtained using the

14

simpler demand function given by Equation (5). When the ethical premiumis independent of the fair wholesale price, results are qualitatively similarto those obtained with an ethical premium related to the wholesale price.Nevertheless, the equilibrium (T, F ) is less frequent in the model with thedemand equation (5) than in the model with the ethical premium functionof the fair wholesale price, especially when α is high. This is due to the factthat if the ethical premium depends on the wholesale price and if fair tradelikers give a high value to the fair products, the retailer can price discriminatemuch more between both types of consumers.

3.3 The consumers' surplus

As the consumers' surplus is the di�erence between what consumers are will-ing to pay and their actual expenditure, it is interesting to study how theintroduction of a fair trade product changes the surplus of both types of con-sumers. In Table 2 we summarize how the di�erent levels of fair wholesaleprice, according the equilibrium, a�ect the surplus of the fair trade likers,SCf , and the one of the product seekers, SCt.

EquilibriumStrategy

w∗f

SCf SCt

(T, F )wf1/2

= 12(1−α)

132

132

(T, F )

wf1 = β1−α

132

< (1−β)2

8< 1

8132

(T, F )wf = 1

132

< α2

8< 1

8132

(F )

wf1/31= 2−

√1+3α2λ

2(1−α2λ)

(1−w∗f (1+αλ−2α))2

8> 1

32

(1−w∗f (1+αλ))2

8< 1

32

(F )

wf31 = β1−αλ

(1−β 1+αλ−2α1−αλ )

2

8> 1

32

(1−β 1+αλ1−αλ)

2

8> 1

32

when β <˜β = 1−αλ

2(1+αλ)

Table 2: Consumers' surplus

We take the consumers' surplus in the case where only the traditional

15

product is proposed, that is, 132, as a referential level. The consumers' surplus

is a�ected by both a change in price and characteristic of the product sold bythe retailer. Whatever the equilibrium, when the fair trade product is sold,the fair trade likers' surplus is stable or enhanced, only because they givevalue to the characteristic and despite the increase in the price of the product.At the equilibrium where both products are sold and where w∗

f = 12(1−α)

, thefair trade agency succeeds in extracting all the fair trade likers' surplus relatedto their preference for the characteristic of the product and in giving to theretailer the pro�t he should have obtained from the only sale of traditionalproduct (see Table 1). In the other two cases where he sells both products,the surplus of the fair trade likers increases. The surplus of fair trade likersis always greater when the retailer sells only the fair product, because thefair retail price is lower and despite the fact that the fair wholesale price islower.

The surplus of the products seekers is stable with respect to the referentialsituation when both types of products are sold by the retailer, because thetraditional product is sold at the same retail price, pt = 3

4. When only the

fair product is sold, the product seekers' surplus can be greater than the onethey receive when the traditional product is sold. Indeed, in the equilibrium

where w∗f = wf31 , if β <

˜β (see Figure 2 and Table 2), the surplus of the

product seekers is greater than 132. At this equilibrium, each agent is in a

better situation than in a market without the fair product.

4 Conclusion

This article tries to give some arguments in the debate about the introductionof fair trade goods in the large-scale distribution. First, the most importantparameter for the fair trade is not the number of consumers who are willingto pay an ethical premium for a fair product but how much these consumersare willing to pay for this kind of good. Indeed, products from fair trade andlabelled are agricultural products already in the shelves of supermarkets intheir traditional format and a retailer is interested by a fair product only ifits wholesale price is lower than the one of the traditional good or if he candiscriminate between fair trade likers and other consumers with a relativelyhigh retail price for the fair product. This second solution is possible only ifthe fair trade likers give a high value to the ethical characteristic of a product.

Second, when the ethical premium parameter is high and when the fair

16

trade certi�er has a preference for the wholesale price in relation to thefair quantities sold, she can extract all the surplus related to the ethicalcharacteristic of the good.

Third, when the fair trade agency has a preference for quantities, she hasto propose a low wholesale price to the retailer. In this case, only the fairproduct is sold and we show that the surplus of all consumers is higher thanthe one they obtain with only the traditional product. Fair price to producersdoes not necessarily imply above average price to consumers (Bowen, 2001).Many products are highly competitive thanks to the lack of intermediariesin the distribution chain, the savings in advertising costs, and the work ofvolunteers in fair trade organizations.

The next logical step in the problematic of fair trade would be to verifythe hypothesis of Alternative Trading Organisations which suggest that bypaying a fair price for even a small part of production, a snowball e�ectappears on prices paid for the rest of production (Bowen, 2001).

A Proof of Proposition 1

At Stage 3 of the game, the retailer chooses retail prices of products accordinghis strategy.

• Strategy 1: (T, F)

The condition of positive quantity of fair product is wf < 11−α

.

The retail prices are pf =1+(α+1)wf

2and pt = 1+wt

2and the retailer's

pro�t is πR1 (wf , wt) = λ

(1+αwf−wf

2

)2

+ (1− λ)(

1−wt

2

)2• Strategy 2: (T)

pt = 1+wt

2and πR

2 (wt) =(

1−wt

2

)2• Strategy 3: (F)

Case 1: pf < 1

pf =1+(αλ+1)wf

2with pf < 1 ⇔ wf < 1

αλ+1= wf1 and

πR31(wf ) =

(1+αλwf−wf

2

)2

17

Case 2: pf > 1

pf =1+(α+1)wf

2with pf > 1 ⇔ wf > 1

α+1= wf2 and

πR32(wf ) = λ

(1+αwf−wf

2

)2

For Stage 2 of the game, we compare retailer's pro�ts, two to two, andwe obtain that:

- whatever wf , πR1 (wf , wt) > πR

32(wf ),

- if wf <1−√

1−(1−α2λ)wt(2−wt)

1−α2λ= wf1/31

, πR31(wf ) > πR

1 (wf , wt) > πR2 (wt),

- if wf1/31< wf < wt

1−αλ= wf2/31

, πR1 (wf , wt) > πR

31(wf ) > πR2 (wt),

- if wf2/31< wf < wt

1−α= wf1/2

, πR1 (wf , wt) > πR

2 (wt) > πR31(wf ),

- if wf > wt

1−α= wf1/2

, πR2 (wt) > πR

1 (wf , wt) > πR31(wf ).

B Proof of Proposition 2

We can easily show that

{U31(wf ) > U1(wf ) = U32(wf ) if wf < 1

U31(wf ) < U1(wf ) = U32(wf ) if wf > 1

and that

wf31 < wf1/31

⇔ β <1− αλ

1− α2λ

(2−

√1 + 3α2λ

2

)= β <

1

2

wf1 < wf1/2⇔ β <

1

2wf1 < 1 ⇔ β < 1− α

We compare the di�erent values of the fair trade agency objective functionaccording to α and β.

(i). When α < 12, that is, wf1/2

< 1, as U31(wf ) > U1(wf ) if wf < 1,the unique equilibrium which can be achieved with the sale of the fairproduct is the one where the fair product is the unique product sold(Strategy 3 case 1 -(F)) and the optimal wholesale price w∗

f , that thefair trade agency proposes, depends on β.

• If β < β, w∗f = wf31 < 1

2.

18

• If β > β, w∗f = wf1/31

= 2−√

1+3α2λ2(1−α2λ)

< 12.

(ii). When α > 12, that is, wf1/2

< 1, the equilibrium with both productssold by the retailer can emerge too.

• If β < β, then wf31 < wf1/31and wf1 < wf1/2

.

- If β < 1−α, we have to compare U1(1) and U31(wf31). Though,U1(1) = U31(1), thus U31(wf31) > U1(1). The equilibriumstrategy is the strategy 3 case 1 and w∗

f = wf31 .

- If β > 1 − α, we have to compare U1(wf1) and U31(wf31). Itexists a threshold of β,

β∗1 =ln λ

ln λ(1− α)− ln 1− αλ,

such as

if β < β∗1 , then U31(wf31) > U1(wf1) and the equilibriumstrategy is the strategy 3 case 1 and w∗

f = wf31 and

if β > β∗1 , then U1(wf1) > U31(wf31) and the equilibriumstrategy is the strategy 1 and w∗

f = wf1 .

• If β < β < 12, then wf1/31

< wf31 and wf1 < wf1/2.

- If β < 1−α, we have to compare U1(1) and U31(wf1/31). It exists

a threshold of β,

β∗2 =ln [2αλ(1−α2λ)]−ln [2(1−α2λ)−(1−αλ)(2−

√1+3α2λ)]

ln [αλ(2−√

1+3α2λ)]−ln [2(1−α2λ)−(1−αλ)(2−√

1+3α2λ)],

such as

if β < β∗2 , then U31(wf1/31) > U1(1) and the equilibrium

strategy is the strategy 3 case 1 and w∗f = wf1/31

and

if β > β∗2 , then U1(1) > U31(wf1/31) and the equilibrium

strategy is the strategy 1 and w∗f = 1.

- If β > 1 − α, we have to compare U1(wf1) and U31(wf1/31). It

exists a threshold of β,

β∗3 : U1(wf1) = U31(wf1/31),

such as

19

if β < β∗3 , then U31(wf1/31) > U1(wf1) and the equilibrium

strategy is the strategy 3 case 1 and w∗f = wf1/31

and

if β > β∗3 , then U1(wf1) > U31(wf1/31) and the equilibrium

strategy is the strategy 1 and w∗f = wf1 .

• If β > 12, then wf1/31

< wf31 and wf1/2< wf1 . Thus, we have to

compare U1(wf1/2) and U31(wf1/31

). It exists a threshold of β,

β∗4 =ln [λ(1−α2λ)]−ln [2(1−α2λ)−(1−αλ)(2−

√1+3α2λ)]

ln [(1−α)λ(2−√

1+3α2λ)]−ln [2(1−α2λ)−(1−αλ)(2−√

1+3α2λ)],

such as

if β < β∗4 , then U31(wf1/31) > U1(wf1/2

) and the equilibrium strat-egy is the strategy 3 case 1 and w∗

f = wf1/31and

if β > β∗4 , then U1(wf1/2) > U31(wf1/31

) and the equilibrium strat-egy is the strategy 1 and w∗

f = wf1/2.

All the thresholds of β are decreasing in α and

β∗3 = β∗1 when β = β

β∗3 = β∗2 when β = 1− α, with ˜α de�ned by β∗2(˜α) = 1− ˜α

β∗3 = β∗4 when β = 12, with α de�ned by β∗−1

4 (α) = 12

β∗2 = β∗4 when α = 12.

In addition, we can show that β∗2(α = 12) = β∗4(α = 1

2) > 1

2when λ < λ '

0.636215.We de�ne β∗ as follows:

when λ < λ, β∗(α) =

β∗1(α) when α ∈ [β−1(α), 1]

β∗3(α) when α ∈ [α, β−1(α)]

β∗4(α) when α ∈ [12, α]

when λ > λ, β∗(α) =

β∗1(α) when α ∈ [β−1(α), 1]

β∗3(α) when α ∈ [˜α, β−1(α)]

β∗2(α) when α ∈ [12, ˜α]

Thereof, when α > 12,

20

if β < β∗, the equilibrium strategy is strategy 3 case 1, when the fair productis the unique product sold by the retailer.

if β > β∗, the equilibrium strategy is strategy 1, when the retailer sells bothproducts.

References

Adriani F., Becchetti L. (2004), Fair trade: a "third generation welfare"mechanism to make globalisation sustainable, CEIS Working Paper, n. 62.

Allain, M-L., Chambolle C. (2003), L'Economie de la Distribution, EditionsLa Découverte, collection Repères, Paris.

Alter Eco (2004), Les consommateurs français et le commerce équitable,http://www.altereco.com.

Becchetti L., Solferino, N. (2003), On ethical product di�erentiation, CEISWorking Paper , n. 188.

Bowen B. (2001), �Let's go fair�! In EFTA, Fair Trade Yearbook 2001,www.eftafairtrade.org.

De Pelsmacker P., Driesen L., Rayp G. (2005), Do Consumers Care AboutEthics ? Willingness-to-Pay for Fair-Trade Co�ee , Journal of ConsumerA�airs, 39 (2): 363-385.

EFTA (European Fair Trade Association) (2001), FAIR TRADE IN EU-ROPE 2001. Facts and Figures on the fair Trade sector in 18 EuropeanCountries.

FTF (Fair Trade Federation) (2003), Report on Fair Trade Trends,http://www.fairtradefederation.com

FLO (Fairtrade Labelling Organizations International)http://www.fairtrade.net

Immordino G. (2002), Fairness, NGO activism and the welfare of the lessdevelopped countries, CSEF Working Paper n.101

INSEE (2004), Le commerce en 2003, INSEE Première 965, mai 2004.

21

Maietta O.W. (2004), The hedonic price of fair trade co�ee for the Italianconsumer, Collana Working Paper n. 2/2004.

Parker P.M., Soberman D.A. (2004), Private labels: psychological version-ing of typical consumer products, International Journal of Industrial Orga-nization, 22: 849-861.

Raynolds L. T. (2000), Re-Embedding Global Agriculture: The Interna-tional Organic and Fair Trade Movements, Agriculture and Human Values,17: 297-309.

Renard M.-C. (2005), Quality certi�cation, regulation and power in fairtrade, Journals of Rural Studies, 21 : 419-431.

Renard M.-C. (2003), Fair Trade: quality, market and conventions, Journalsof Rural Studies, 19(1): 87-96.

22