Labels Battle: Competition between NGOs as Certifiers
Sylvaine Poret∗
Preliminary version
Abstract
This paper studies the competition between the NGOs acting as standard-setting organizations. We consider a double duopoly model where NGOs compete to offer firms labels for
sustainability quality and where firms compete to sell consumers differentiated products. We
assume that NGOs’ objectives can be very different, some NGOs being mission-driven organizations, others market-driven organizations. We find that two stringent sustainability quality
labels cannot coexist if the consumers’ sustainability awareness is low and/or the costs related
to quality and certification are high. Moreover, the two NGOs have to offer two very different
labels in terms of sustainability quality criteria in order to be present on the labels market.
Keywords: labels, NGOs, vertical product differentiation, sustainability quality.
JEL classification: L13, L15, L31, Q01.
∗
INRA, UR1303 ALISS, Ivry-sur-Seine France; Department of Economics, Ecole Polytechnique, Palaiseau,
France. Postal address: INRA-ALISS 65 boulevard de Brandebourg 94205 Ivry-sur-Seine cedex France. Phone
number: +33(0)1 49 59 69 07. Fax number: +33(0)1 49 59 69 90. E-mail address: [email protected].
1
1
Introduction
An existing directory of certifications, the Ecolabel Index, currently tracks 458 ecolabels in 197
countries and 25 industry sectors, including AB (Agriculture Biologique) France’s national logo
for organic products, Fairtrade International an association of 25 organizations around the world,
Marine Stewardship Council(MSC) with a standard for sustainable fishing, or Non-GMO Project
an organization offering independent verification of testing and GMO controls for products in the
U.S. and Canada.1 Some labels are certified by a third-party, such as a public institute or a NonGovernmental Organization (NGO), but others arise from self-declarations by firms. According to
the 2010 Global Ecolabel Monitor, most eco-labels (58%) were run by non-profit organizations,
18%, by for-profit organizations, and 8%, by governments, with other types (industry associations,
hybrid for/non-profit partnerships, public-private partnerships) composing the rest (Big Room Inc.
and World Resources Institute, 2010). Moreover, about 92% of labeling programs required certification before they award an eco-label, whereas others require registration but no previous certification. Of those requiring certification, the majority (64%) were third-party certification systems.
On the corporate side, more than one-third of multinational companies have voluntary third-party
certifications for environmental or social standards (Kitzmueller and Shimshack, 2012). In this
way, NGOs as standard setters or certifying agencies appear as the preferred partners of firms in
many fields, including sustainable agriculture, fishing, packaging, supply chain management, labor
issues, renewable energy, forest resources, health, and safety. According to Gereffi et al. (2001),
NGOs are the strongest and most legitimate certifiers, because they are able to coordinate bodies
which set and oversee compliance.
The increasing presence of NGOs and the confidence they arouse afford them the power to
positively influence private sector behavior through constructive partnerships. Some NGOs’ motivations for such collaborations with the private sector are identifiable. The primary motivation
is the sinews of war: money. Indeed, the increasing scarcity of public funds and the increasing number of NGOs force NGOs to find new sources of funding. Because firms are institutions
with relatively easier access to financial resources, NGOs are motivated to establish alliances with
corporations. Another motivation for NGO collaboration with corporations is the rise of societal
1
Ecolabel Index was initiated in 2009 by Big Room Inc., a Vancouver-based company, and the World Resources
Institute, a Washington DC-based environmental think tank (http://www.ecolabelindex.com, accessed 15.01.14).
2
problems. Indeed, a partnership is a way to sensitize corporate clientele to an NGO’s cause. A
positive consequence of such partnerships is an increase in notoriety: the association with a firm
with a strategic position in the market is one way for an NGO to strengthen its reputation and
political influence (Selsky and Parker, 2005).2 Thus, NGOs have an incentive to work with large,
consumer-oriented, notorious companies. Consequently, a real market for NGOs’ services is created.
The development of sustainability labels borne by NGOs has created a real competition between them. A good illustrative example of this phenomenon is the Lipton-Rainforest Alliance
partnership. In May 2007, Unilever, the world’s largest tea company, announced plans to source
its entire tea supply sustainably. The objectives of this CSR activity as revealed by Unilever were
the following: to secure its source of supply by a long-term partnership; to rebuild Lipton’s market
share and to recover the potential additional supply cost through sales growth; and to obtain the
first mover gain on the tea market since none of the other main tea brands initiate the same kind
of approach. Due to these objectives, an alliance with a third party certification might be the best
solution for Lipton in order to establish an auditing and certification infrastructure in producing
countries. Then, Unilever needed to find a credible certification partner. Three labels developed
by NGOs have been studied by Unilever to make credible their approach: FLO (Fairtrade), Utz
Certified, and Rainforest Alliance. Unilever chose the Rainforest Alliance based on their common
approach, which are the Rainforest Alliance’s ability to work on an international scale and with
both large scale plantations and small farmers, and its market-based premium for farmers and its
ability to help move an entire industry (Poret, 2010).
In spite of the fact that tea was not in the product portfolio of the Rainforest Alliance, the
NGO agreed to cooperate. Thus, this shows its entry into certifying tea production in addition to
its long-established programs in coffee, cocoa, bananas, sustainable forestry, and tourism. Historically based in Latin America, by this alliance, the NGO strengthens its presence in Africa.
Another reason of the agreement is the common point of view about sustainability and mainstream
strategy.3 Furthermore, the Rainforest Alliance can benefit of the notoriety and the visibility of
2
For instance, the concept of fair trade has experienced an impressive expansion following the launch of the Max
Havelaar label (Fairtrade) awarded to brand-name products or private label products sold in large retail stores (Poret,
2010).
3
Tensie Whelan, the Rainforest Alliance Executive Director, said in 2007: “We are delighted to be working with a
company that understands the value of putting sustainability at the heart of its business”.
3
the brand Lipton, especially in Europe, where the NGO and its logo werre not well known. The
international NGO Rainforest Alliance aims are to conserve biodiversity and ensure sustainable
livelihoods by transforming land-use practices, business practices and consumer behavior. Rainforest Alliance sustainable agriculture certification, like the certification scheme UTZ Certified,
does not offer producers a minimum or guaranteed price, as it is required under the Fairtrade
scheme, therefore leaving them vulnerable to the market price variations. In 2005, a report by
Ethical Corporation Magazine compared Fairtrade and Rainforest Alliance certification programs
and it concluded coffee producers under the latter scheme received 21% less for their crop than
under Fairtrade. This explains why the Rainforest Alliance’s sustainable agriculture certification
has been called “Fairtrade Lite” by some critics. Therefore, Unilever would have chosen a less
demanding standard.
This kind of situations raise several concerns inside militants and organizations. Firms which
offer certified products pay license fees to the NGO in order to use its label/logo, which represent
a revenue source for it. Because of their large market share, volumes traded and sales made, large
mainstream companies are the largest revenue-raisers. This creates a dangerous dependency between NGOs and these conventional firms. In addition, some mainstream companies have created
certification systems in association with NGOs or certification organisations. This proliferation of
different certification systems and logos (Fairtrade, Bio-équitable, Rainforest Alliance) generates a
risk of confusion for consumers and of credibility loss for firms. Moreover, this abundance creates
a competition between labels. In this case, the risk is that “the bad label drives out the good one”,
that is, the certification system with less stringent and less costly standards becomes the leader on
the market and the benchmark.
Most probably to respond to the competition with other sustainability labels, in a mainstreaming approach, FLO recently proposed the development of a Fairtrade Sourcing Partnership (FSP)
with a new fair trade label. Currently, for a product to bear the Fairtrade logo, FLO standards
require that all ingredients that can be certified must be and that a minimum of 20% of the total
product comprise Fairtrade certified ingredients. The FSP would shift from this policy and use
a new logo to certify products containing only one certified ingredient - sugar, cocoa, or cotton
- even if the ingredient composes less than 20% of the total product. This new scheme aims to
increase the volume of commodities being purchased from Fairtrade certified farmers and to engage companies that do not want to commit to the full cost of certifying their products or that are
4
only interested in particular commodities. However, the introduction of a new certification mark
contributes to consumer confusion and possibly the erosion of credibility. Moreover, the new mark
has less stringent standards for “Fairtrade” products and is cheaper for companies to adopt. There
is a high risk that this mark will then devalue Fairtrade certification.
More and more voluntary labels with similar criteria being established by NGOs, this paper
proposes to analyze the competition between the NGOs acting as standard-setting organizations.
Our aim is to investigate the strategy set up by some NGOs, such as Fairtrade, consisting in reducing the level of its quality standards to face up to the fact that their labels are less chosen than the
others by firms.
Despite the importance of NGOs presence, economic analyses of the competition between
NGOs are rare. Competition between NGOs is mainly studied through their fundraising activities.
In Aldashev and Verdier (2009, 2010), horizontally differentiated NGOs act as producers of goods
and services in developing countries and compete with each other through fundraising. Aldashev
and Verdier (2009) explain the phenomenon of internationalization of major development NGOs,
similar in structure to the multinational firms. Aldashev and Verdier (2010) examine a model
where the NGOs allocate their time between working on the project and fundraising. They find
that with free entry of NGOs the equilibrium number of NGOs can be either larger or smaller than
the socially optimal one, depending on the efficiency of the fundraising technology.
The approach of this paper is closer to the literature on the economics of labels, see Bonroy
and Constantanos (2015) for a review on this topic. Ben Youssef and Abderrazak (2009) and
Harbaugh et al. (2011) model label competition in a context of uncertainty about the quality of the
labeled products. Ben Youssef and Abderrazak (2009) study a duopoly in which each label can
only be used by one firm; under complete information adding the second label always improves
environmental protection, but if consumers cannot tell which label is more stringent, the firms use
prices to signal qualities, and the second label actually reduces overall environmental protection.
Harbaugh et al. (2011) assume firms’ qualities and the stringency of labels are exogenous; they find
that even small amounts of uncertainty can create consumer confusion that reduces or eliminates
the value to firms of adopting voluntary labels.
There is a substantial body of literature comparing label levels according to the types of
standard-setting organizations - government, NGO, the industry - and their objectives (Heyes and
Maxwell, 2004; Bottega and De Freitas, 2009; Bottega et al, 2009; Manasakis et al., 2013). Fis5
cher and Lyon (2014, 2013) examine the competition between an NGO and a for-profit industry
association. Fischer and Lyon (2014) find that environnemental benefits may be smaller in the
presence of both labels than the NGO label alone. Fischer and Lyon (2013) show that when the
number of firms is fixed, the industry sets an ambitious binary standard and the NGO sets a basic binary standard. When there is free entry into the market, the NGO sets an ambitious binary
standard and the industry declines to offer a label at all. In all these papers, the objective of NGOs
is to maximize the environnemental quality or minimize a specific harm, usually related to some
externality. Bottega et al. (2009) consider two different objectives of third party certifiers to settle
the quality standard of the label : to maximize the high quality demand or to maximize the overall
quality of the market, independently of how many consumers will benefit from the high quality.
This study differs from previous works in several ways. Most importantly, we model rivalry
between two labels created by two NGOs, two organizations of the same type. We nevertheless
suggest that the NGOs’ objectives can be very different. Some NGOs are mission-driven organizations, whose activity is exclusively cause oriented, with a strong commitment to some values such
as fair trade or child labor. Their objectives are then clearly oriented towards the stringency of
their standards, that is, the quality of the label. Other NGOs work with large companies, often offer made-to-measure standards to them. These NGOs may be seen as market-driven organizations,
whose objective is to stimulate trade in order to fully promote their cause. We assume that NGOs
seek to maximize a preference function, a trade-off between the quantity and quality of labeled
products sold. We then consider a duopolistic vertical differentiation model with two sustainability labels whose the level of quality is endogenously chosen by each NGO. Our key finding is that
the coexistence of both labels is possible only if the objectives of NGOs in terms of sustainability
quality are extremely different.
This remainder of the paper proceeds as follows: Section 2 presents the model. Section 3
analyzes the firms’ choices in terms of prices and labels. Section 4 characterizes the equilibrium
labeling scheme for the two NGOs. Finally, section 5 concludes.
2
Model
We develop a Bertrand duopoly model of vertical differentiation. Sustainability quality is the
vertical differentiation variable. We consider this sustainability quality represents some composite
6
characteristic expected to objectively and synthetically measure the attention attached by the firm
mainly to the sustainable development. This sustainability quality is promoted by two NGOs, A
and B, which can propose two different labels. The following game is considered. In the first stage,
two NGOs acting as certifiers choose the level of standards required for obtaining their label, given
their respective objective. In the second stage, the two firms, 1 and 2, choose sequentially which
label to adopt. In a third stage, the firms set simultaneously prices. In the fourth stage, consumers
choose whether or not to buy good with sustainability label A or B.
2.1
Consumers
As in Motta (1993), Youssef and Lahmandi-Ayed (2008), Fischer and Lyon (2013) and Barry et
al. (2014), we adopt a framework of vertical differentiation consistent with Mussa and Rosen
(1978). Consumers have the same utility function u = θs − p (and zero utility if they do not
buy the differentiated good). They differ in their tastes for sustainability quality, described by the
parameter θ ∈ [θ, θ], θ being uniformly distributed with unit density. The higher the quality s of the
good, the higher the utility u reached by the consumers for any given price p. However, consumers
with a higher θ will be willing to pay more for a higher sustainability quality good. Considering
that the consumers’ awareness of sustainable development results in a higher willingness-to-pay
for a sustainable product than for a standard one, we consider that the parameter θ represents the
level of the consumers’ sustainability awareness.
In accordance with the literature on product differentiation, we assume that consumers can buy
at most one unit of the good. When a product is certified by a label, consumers know for sure that
the quality level is s, in the sense that the two NGOs acting as certifiers are trustworthy. The NGO
j offers the label j certifying the sustainability quality level sj , with sA > sB .
The following threshold values allow to characterize consumers’ choices:
pA − pB
,
sA − sB
pB
=
.
sB
θAB =
θ∅B
Consumers characterized by θ > θAB obtain a higher utility from consuming the good certified
by the NGO A rather than the alternative good certified by the NGO B; while consumers characterized by θAB > θ > θ∅B obtain a higher utility from consuming the quality set by the NGO B
7
rather than no consumption. The utility function implies the following demand functions for the
good labelled by NGO j, with j = A, B:
qA (pA , pB , sA , sB ) = θ − θAB
(1)
qB (pA , pB , sA , sB ) = θAB − θ∅B
(2)
Notice that the market is not covered, i.e. some consumers do not buy the certified variety of the
good.
When only one label is present on the market, the demand function for this labeled good is then
qj (pj , sj ) = θ −
2.2
pj
.
sj
Firms
On the supply side of the product market, we consider two symmetric firms, i = 1, 2. The firms’
marginal cost depends on the chosen label and is defined as
c(sj ) = (csj + k)sj
for j = A or B.
(3)
The marginal cost function c(sj ) has two components. The first component, cs2j , represents the
quadratic cost of providing sustainability quality for firms. Marginal cost of meeting the label j
requirements is independent of the quantity of good produced but strictly increasing and convex
in the sustainability quality sj . The second component, ksj , is the cost of certifying paid to the
NGO j. The parameter k represents the unit cost of controlling labelled criteria if the sustainability
quality level sj is viewed as a number of precise criteria certified by the label j, that is, the number
of criteria verified and controlled by the NGO j. To build up the model, we assume that the
consumer with the highest valuation for sustainability quality is willing to pay the unitary cost of
labeled criteria, θ > k.
If firms choose different labels, firm i profits are given by
πij (pA , pB , sA , sB ) = [pj − (csj + k)sj ]qj (pA , pB , sA , sB )
for j = A or B.
(4)
If firms choose the same label, the classic Bertrand game applies, which provides zero profit
for both firms. In this case, the total demand for the labeled product is shared equally by the two
firms.
8
2.3
NGOs
In the absence of public intervention on a sustainability issue, we want to investigate the impact of
the competition between labels supported by NGOs.
The label awarded by a NGO certifies that the product satisfies certain quality criteria. It
provides consumers with credible information on the sustainability quality of the labeled variant
otherwise unobservable. We assume that the certification technology is the same for the two NGOs,
that is, monitoring is almost perfect and the cost of monitoring is equal to zero.
The main hypothesis adopted concerns the objective of the NGOs. Aldashev and Verdier (2009,
2010) assume that NGOs maximize the impact of their respective projects, that is, the quantity of
services they produce towards their missions. Most of the literature on eco-labels (Heyes and
Maxwell, 2004; Bottega and de Freitas, 2009; Fisher and Lyon, 2013, 2014; Brécard, 2014) assumes that the NGO maximizes the environnemental quality. In this paper, we propose that NGOs
realize a trade-off between quality of their label and quantity of labeled products sold. This allows us to distinguish mission-driven NGOs and market-driven NGOs. To formalize this idea, as
Poret and Chambolle (2007), we assume that the objective is to maximize a utility function which
depends positively on the sustainability quality and on the quantity of labeled product sold.4 It is
specified in the following utilitarian form for the NGO A:
UA (sA ) = sαA (qA (sA , sB ))1−α ,
(5)
where α is the NGO A’s quality preference parameter, with α ∈ [0, 1]. When α = 0.5, the utility
of the NGO is equal to its benefits.
Likewise, the NGO B’s objective is to maximize the following utility function
UB (sB ) = sβB (qB (sA , sB ))1−β ,
(6)
where β is the NGO B’s quality preference parameter, with β ∈ [0, 1].
4
Poret and Chambolle (2007) specifically study the Fair Trade label and propose that the utility of the Fair Trade
certifier depends positively on the Fair Trade wholesale price, which determines the small producers’ revenue, also as
on the Fair Trade product quantity sold, which represents the number of small producers in the Fair Trade network.
9
3
Firms’ choices
The game is solved by backward induction. We start by solving the third stage of the game in
which firms choose their prices as a function of their preceding choice of labels.
3.1
Price choice
At this point two types of market outcome can be considered.
No Differentiation: when the two firms choose the same label, the classical Bertrand game
applies. The two firms offer the same price p = (csj + k)sj with sj = A or B, depending to
the label selected. This provides zero profit for the two firms and the total demand for the labeled
product is equal to q = θ − k − csj .
Differentiation: when the two firms choose different labels, they have some market power due
to product differentiation. Consider the firm i (i = 1 or 2) which has chosen the label A, it will
solve the following profit maximization program:
max πiA (pA , pB , sA , sB ) = [pA − (csA + k)sA ]qA (pA , pB , sA , sB ).
pA
(7)
The firm −i which has chosen the label B will solve the following profit maximization program:
max π−iB (pA , pB , sA , sB ) = [pB − (csB + k)sB ]qB (pA , pB , sA , sB ).
pB
(8)
First order conditions give following equilibrium prices:
sA
[2(sA − sB )θ + 2(csA + k)sA + (csB + k)sB ],
4sA − sB
sB
p∗B (sA , sB ) =
[(sA − sB )θ + (csA + k)sA + 2(csB + k)sA ].
4sA − sB
p∗A (sA , sB ) =
(9)
(10)
The expressions imply that the high quality firm (with label A) fixes a price higher than the price
charged by the low quality firm (with label B). Note also that the closer sB is to sA , the closer the
solution is to a Bertrand solution, that is, prices are equal to marginal costs.
The demand functions at the equilibrium prices in stage 3 are then:
sA
[2(θ − k) − (2sA + sB )c],
4sA − sB
sA
qB∗ (sA , sB ) =
[(θ − k) + (sA − sB )c] > 0.
4sA − sB
qA∗ (sA , sB ) =
10
(11)
(12)
These functions imply that the quantity of products labeled by NGO A, sold by the firm i, is
positive if
sA +
sB
θ−k
<
.
2
c
(13)
The presence of two sustainability labels on the market is possible only if the levels of the two
sustainability quality labels are relatively low compared to the demand/cost ratio of supplying a
labeled product θ−k
. Thus, at the price competition stage, two stringent labels can coexist if
c
the consumers’ sustainability awareness is high and the costs related to quality and certification
are low, that is, when firms and NGOs are cost-efficient to implement and control the sustainable
quality.
Respective profits of firm i and −i are:
2
s2A (sA − sB ) 2(θ − k) − (2sA + sB )c
πiA (sA , sB ) =
,
(4sA − sB )2
2
sA sB (sA − sB ) (θ − k) + (sA − sB )c
π−iB (sA , sB ) =
.
(4sA − sB )2
(14)
(15)
It is easy to check that the partial derivative of πiA (sA , sB ) with respect to sB is negative since sA >
sB and the condition (13) is satisfied. In a easier way, the derivative of π−iB (sA , sB ) with respect
to sA is positive. This means that the more differentiated the products in terms of sustainability
quality, the less intense the competition and the higher firms’ profits.
3.2
Label choice
As Bottega et al. (2009), we assume the choice of label is sequential. Bottega et al. (2009) argue
that in real world, labeling processes have usually relatively slow-motion patterns, as only few
firms adopt labels in the earlier stages of implementation. This may be explained by the fact that
voluntary labelling requires firms to adapt their production or trade methods. More, we assume
sequentiality in certification choice to avoid multiplicity of equilibria. This does not modify at all
the firms’ choice at equilibrium, but it allows the selection of an equilibrium when there is more
than one. See Ben Youssef and Lahmandi-Ayed (2008) for an identical assumption in a context
of quality choice. We assume that firm 1 is the Stackelberg leader and firm 2 the follower. The
11
extensive form of the certification game is depicted in Figure 1, with
2
s2A (sA − sB ) 2(θ − k) − (2sA + sB )c
π1A (sA , sB ) = π2A (sA , sB ) =
,
(4sA − sB )2
2
sA sB (sA − sB ) (θ − k) + (sA − sB )c
.
π1B (sA , sB ) = π2B (sA , sB ) =
(4sA − sB )2
(16)
(17)
(0,0)
Label A
Firm 2
Label A
Label B
(
,
)
(
,
)
Firm 1
Label A
Label B
Firm 2
Label B
(0,0)
Figure 1: The certification game
This allows us to state the equilibrium strategies in the following proposition.
Proposition 1 (Firms’ Equilibrium Strategies). In a duopoly model with vertical differentiation,
the results of the certification game are the following:
(i). if sA > max { θ−k
−
c
sB
, sB },
2
firm 1 chooses the label B and firm 2 also the label B - strategy
(B, B);
√
(ii). if max {f
sA (sB ), sB } < sA ≤
θ−k
c
−
sB
2
with sf
A (sB ) =
θ−k
c
−
sB
2
−
csB (4(θ−k)−3csB )
,
2c
firm 1
chooses the label B and firm 2 the label A - strategy (B, A);
(iii). if sB < sA ≤ sf
A (sB ), firm 1 chooses the label A and firm 2 the label B - strategy (A, B).
Proof. See the appendix A.1.
12
êê
q -k
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
c
sA
=
̅−
−
2
=
( , )
=
( )
( , )with
>
( , )
with
>
êê
q -k
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
3c
( , )
with
>
êê
q -k
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
3c
sB
êê
2 H q - kL
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
3c
Figure 2: The label choice
The results of the certification game are represented in Figure 2.
The coexistence of two sustainability quality labels is not possible if the levels of the labels
are relatively high compared to the demand/cost ratio of supplying a labeled product, that is, if
sA >
θ−k
c
−
sB
.
2
This is due to the fact that when both labels are stringent or when the the
consumers’ awareness (θ) is low, the market size is low. Hence, firms do not fight one other for
a market niche. We defined this effect as the “market coverage” effect. Firms have an interest
in avoiding direct price competition by differentiating their products with different labels only
if the market is fully covered, that is, if sB is relatively low. This first result means that two
stringent sustainability quality labels can coexist in a duopoly market with symmetric firms only
if the demand/cost ratio of supplying a labeled product θ−k
is relatively high, that is, if the
c
consumers’ sustainability awareness is high and/or the costs related to quality and certification are
low.
When the two sustainability quality labels can coexist on the market sA <
θ−k
c
−
sB
2
, firm
1 chooses the lowest quality label (label B) if sA > sf
A (sB ) and chooses the highest quality label
(label A) if sA < sf
A (sB ). Firm 2 always chooses firm 1’s opposite strategy. The choices can be
explained as follows. When the less stringent label, B, requires a low quality level, the market size
is potentially large. If the more stringent label, A, also requires a relatively low quality level, it is
13
profitable for firm 1 to choose the label A in order to fully take advantage of the market coverage
effect (qA > qB ). If the label A requires a intermediate quality level, it is again profitable for firm
1 to choose the label A. In this case, firm 1 takes advantage of the “product differentiation” effect
by setting a high price (qA < qB ). When the less stringent label, B, requires a intermediate quality
level, the market size is reduced and the product differentiation through the quality is also limited.
Thus, whatever the quality level of the more stringent label, the leader firm chooses the label B.
4
Labels competition
In this section, we analyze the strategic behavior of NGOs as sustainability quality certifiers
through the choice of the quality level required to obtain their label.
When only one label is present on the market, the NGO named C sets the sustainability quality
level to maximize its following objective:
UC (sC ) = sγC (qC (sC ))1−γ
with qC (sC ) = θ − k − csC .
The unique solution of this program is s0C = γ
θ−k
c
(18)
and the total quantity sold is equal to
qC0 = (1 − γ)(θ − k).
When both labels are present on the market, we assume that the NGOs simultaneously choose
the level of their sustainability quality standard.
The program of the NGO A is the following:
max UA (sA , sB ) =
sA
sαA
sA (2(θ − k) − (2sA + sB )c)
4sA − sB
1−α
.
From the first order condition, we obtain the best response function of NGO A as follows:
(i). when α < 25 ,

√
 4α(θ−k)−(3α−2)csB + YA
8c
sBR
A (sB ) =
 s
B
(ii). when α ≥
sBR
A (sB )
if sB <
θ−k
4α2
4−4α+3α2 c
if sB ≥
4α2
θ−k
4−4α+3α2 c
if sB <
2(4α−1) θ−k
3(1+2α) c
2(4α−1) θ−k
3(1+2α) c
2
5
=


√
4α(θ−k)−(3α−2)csB + YA
8c
 s
B
if sB ≥
14
(19)
with YA = (4(θ − k) − 3csB )(4α2 (θ − k) − (4 − 4α + 3α2 )csB ).
The best response of the NGO A is to differentiate its label from the rival label by setting a
higher sustainability quality standard when it is possible, that is, when the level of the NGO B’s
standard, sB , is relatively low.
The program of the NGO B is the following:
max UB (sA , sB ) =
sB
sβB
sA ((θ − k) + (sA − sB )c)
4sA − sB
1−β
.
From the first order condition, we obtain the best response function of NGO B:

1+2β θ−k


s
if sA < 3(1−β)

c
 A
√
1
(3+2β)csA −(1−2β)(θ−k)− YB
BR
1+2β
(i). when β < 4 , sB (sA ) =
if 3(1−β) θ−k
≤ sA ≤
2βc
c



 2 θ−k − 2s
if s > 1 θ−k
c
A
A
1+2β
(20)
1 θ−k
1+2β c
;
c
with YB = ((θ − k) − 3csA )((1 − 2β 2 )(θ − k) − (1 + 2β)(3 − 2β)csA )
(ii). when β ≥ 41 , sBR
B (sA ) = sA .
The best response of the NGO B is to differentiate its label from the rival label by setting a
lower sustainability quality standard when it is possible, that is, when the level of the NGO A’s
standard, sA , is relatively high. But, this strategy is feasible only when the NGO B’s quality
preference parameter, β, is weak. Indeed, if the objective of the NGO B is fully oriented towards
the sustainability quality of products it certifies, its best strategy is to set the level of its label at its
highest possible value, sA .
Proposition 2 (NGOs’ Equilibrium Strategies). In a double duopoly model with vertical differentiation, the results of the NGOs’ competition game are the following:
(i). When β ≥ 41 , then there are multiple equilibria such as sA = sB ;
(ii). when
√
7− 33
16
≤β<
1
4
and
(a) when α <
1
,
2(1−2β)
there are multiple equilibria such as sA = sB ;
(b) when α ≥
1
,
2(1−2β)
there is one unique equilibrium (s∗A , s∗B );
(iii). when β <
√
7− 33
16
and
(a) when α < α(β), there are multiple equilibria such as sA = sB ;
15
(b) when α(β) < α ≤
1
,
2(1−2β)
∗∗
∗
there are two equilibria (s∗A , s∗B ) and (s∗∗
A , sB ) with sA >
∗
∗∗
s∗∗
A and sB < sB ;
(c) when α ≥
1
,
2(1−2β)
there is one unique equilibrium (s∗A , s∗B );
Proof. See the appendix A.2.
Results of the NGOs’ competition game are represented in Figure 3.
a
1
=
(
∗,
1
2(1 − 2 )
∗)
è!!!!!!
33 - 1
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
8
=
= ( )
∗,
(
∗∗ ,
=
1
ÅÅÅÅÅ
3
∗
∗∗ )
= ( )
= ( )
è!!!!!!
7 - 33
ÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅÅ
16
1
ÅÅÅÅÅ
4
1
ÅÅÅÅÅ
3
b
Figure 3: The quality choice
In a double duopoly model where NGOs compete to certify differentiated products through
firms, the coexistence of both labels is possible only if the objectives of NGOs in terms of sustainability quality are extremely different. In this case, this allows firms to choose each a different
label, that is, the differentiation strategy too.
5
Conclusion
We have developed a simple model of competition between two NGOs which may have different
objectives in terms of sustainability. We found first that two stringent sustainability quality labels
16
cannot coexist in a duopoly market with symmetric firms if the consumers’ sustainability awareness is low and/or the costs related to quality and certification are high. Firms realize a trade-off
between a market coverage effect and the classical product differentiation effect. Second, we show
that the two NGOs, which compete to offer firms a sustainability quality label, may supply two
very different labels in terms of sustainability quality criteria in order to be present on this labels
market.
This model can be extended by assuming a sequential choice of their certification level for
NGOs and by assuming multi-tier labels as Fisher and Lyon (2013).
References
Aldashev, G. and T. Verdier. (2009). When Ngos Go Global: Competition on International Markets for Development Donations. Journal of International Economics, 79(2), 198-210.
Aldashev, G. and T. Verdier. (2010). Goodwill Bazaar: Ngo Competition and Giving to Development. Journal of Development Economics, 91(1), 48-63.
Barry, I., Bonroy O., Garella, P. G. (2014). Labeling by a Private Certifier and Public Regulation.
Working Paper GAEL 2014-07.
Ben Youssef, A. and C. Abderrazak (2009). Multiplicity of Eco-Labels, Competition, and the
Environment. Journal of Agricultural & Food Industrial Organization 7(2).
Ben Youssef, A. and R. Lahmandi-Ayed (2008). Eco-labelling, Competition and Environment:
Endogenization of Labelling Criteria. Environmental and Resource Economics 41(2): 133-154.
Bonroy, O. and C. Constantatos (2015). On the Economics of Labels: How Their Introduction
Affects the Functioning of Markets and the Welfare of All Participants. American Journal of
Agricultural Economics, 97(1), 239-59.
Bottega, L. and J. De Freitas (2009). Public, Private and Nonprofit Regulation for Environmental
Quality. Journal of Economics & Management Strategy 18(1): 105-123.
Bottega, L., P. Delacote, et al. (2009). Labeling policies and market behavior: quality standard
and voluntary label adoption. Journal of Agricultural & Food Industrial Organization 7.
17
Brécard, D. (2014). Consumer Confusion over the Profusion of Eco-Labels: Lessons from a
Double Differentiation Model. Resource and Energy Economics, 37(0), 64-84.
Fischer, C. and T. P. Lyon (2014). Competing Environmental Labels. Journal of Economics &
Management Strategy.
Fischer, C. and T. P. Lyon (2013). A Theory of Multi-Tier Ecolabels. WP
Harbaugh, R., J. W. Maxwell, et al. (2011). Label Confusion: The Groucho Effect of Uncertain
Standards. Management Science 57(9): 1512-1527.
Heyes, A. G. and J. W. Maxwell (2004). Private vs. public regulation: political economy of the
international environment. Journal of Environmental Economics and Management 48(2): 978996.
Manasakis, C., E. Mitrokostas, et al. (2013). Certification of corporate social responsibility activities in oligopolistic markets. Canadian Journal of Economics/Revue canadienne d’économique,
46(1), 282-309.
Poret, S. and C. Chambolle (2007). Fair Trade Labeling: Inside or Outside Supermarkets? Journal
of Agricultural & Food Industrial Organization 5(1).
A
Appendixes
A.1
Proof of Proposition 1
The result of the certification game depends on profits comparisons. It is trivial to see that
π1B (sA , sB ) = π2B (sA , sB ) > 0 for all values of sA > sB . Likewise, π1A (sA , sB ) = π2A (sA , sB ) >
√
csB (4(θ−k)−3csB )
θ−k sB
θ−k sB
0 when sA < c − 2 . More, π1A (sA , sB ) > π1B (sA , sB ) when sA < c − 2 −
=
2c
sf
f
A (sB ), with s
A (sB ) > sB when sB <
If sA >
θ−k
c
−
sB
,
2
θ−k
.
3c
π2A (sA , sB ) < 0 and since π2B (sA , sB ) > 0, choosing the label B is the
dominant strategy for firm 2, that is, firm 2 always chooses to certify its product with the label B.
Knowing that firm 2 always chooses the label B, firm 1 chooses also the label B since its gain with
label A will be lower than its gain with the label B (π1A (sA , sB ) < 0).
18
If sA <
θ−k
c
−
sB
,
2
π2A (sA , sB ) > 0 and since π2B (sA , sB ) > 0, firm 2 always chooses to
certify its product with the label opposite of the one chosen by firm 1. Knowing that strategy, firm
1 compares π1A (sA , sB ) and π1B (sA , sB ).
√
Since π1A (sA , sB ) > π1B (sA , sB ) when sA <
θ−k
c
−
sB
2
−
csB (4(θ−k)−3csB )
2c
chooses the label A when sA < sf
A (sB ) and it chooses the label B when
A.2
θ−k
c
= sf
A (sB ), firm 1
− s2B > sA > sf
A (sB ).
Proof of Proposition 2
Resolution of the NGO A’s program:
The program of the NGO A is :
max UA (sA , sB ) = sαA (qA (sA , sB ))1−α .
(A1)
sA
The FOC is the following:
α−1
sA
(qA (sA , sB ))−α
Since
∂qA
∂sA
=−
2(θ−k)sB +(8s2A −4sA sB −s2B )c
(4sA −sB )2
The second order condition
∂ 2 UA
∂s2A
∂qA
=0
αqA (sA , sB ) + (1 − α)sA
∂sA
(A2)
< 0, there exists at least one solution to the equation (A2).
< 0 induces the following condition:
sA >
sB (2(θ − k) − csB ))
.
4α(θ − k) − (3α − 2)csB
(A3)
The FOC is equivalent to the following quadratic equation in sA :
8cs2A − 2(4α(θ − k) − (3α − 2)csB )sA + (2(θ − k) − csB )sB = 0
(A4)
The discriminant of the quadratic equation
∆ = 4(4(θ − k) − 3csB )(4α2 (θ − k) − (4 − 4α + 3α2 )csB ) = 4YA is positive if and only if
sB <
4α2
θ−k
.
4−4α+3α2 c
The lower real root is a local minimum, since it does not observe the SOC (A3).
The higher real root is:
4α(θ − k) − (3α − 2)csB +
sA (sB ) =
8c
It is easy to cheek that sA (sB ) <
θ−k
2
−
sB
.
2
√
YA
> 0.
(A5)
The solution has also to observe the additional
condition, i.e., sA (sB ) > sB .
When α < 25 ,
19
• sA (sB ) ≥ sB , if sB ≤
4α θ−k
;
3(α+2) c
• sA (sB ) < sB , if sB >
4α θ−k
.
3(α+2) c
When α > 25 ,
As
• sA (sB ) ≥ sB , if sB ≤
2(4α−1) θ−k
;
3(1+2α) c
• sA (sB ) < sB , if sB >
2(4α−1) θ−k
.
3(1+2α) c
4α
3(α+2)
4α2
4−4α+3α2
>
>
2(4α−1)
3(1+2α)
when α <
2
5
and
4α2
4−4α+3α2
>
2(4α−1)
3(1+2α)
>
4α
3(α+2)
when α > 25 , we
obtain the best response function of NGO A as follows:

θ−k
4α2
 s (s ) if s <
A B
B
4−4α+3α2 c
2
BR
(i). when α < 5 , sA (sB ) =
;
θ−k
4α2
 s
if s ≥
B
B

 s (s )
A B
2
BR
(ii). when α ≥ 5 , sA (sB ) =
 s
B
if sB <
if sB ≥
4−4α+3α2
c
2(4α−1) θ−k
3(1+2α) c
2(4α−1) θ−k
3(1+2α) c
.
Resolution of the NGO B’s program:
The program of the NGO B is :
max UB (sA , sB ) = sβB (qB (sA , sB ))1−β .
(A6)
sB
The FOC is the following:
−β
sβ−1
B (qB (sA , sB ))
Since
∂qB
∂sB
=
sA (θ−k−3csA )
(4sA −sB )2
∂qB
=0
βqB (sA , sB ) + (1 − β)sB
∂sB
< 0 if and only if sA >
equation (A7) if and only if sA >
The second order condition
θ−k
.
3c
∂ 2 UB
∂s2B
If sA ≤
θ−k
,
3c
θ−k ∂UB
, ∂sB
3c
(A7)
there exists at least one solution to the
> 0, then sBR
B (sA ) = sA .
is negative when sA >
θ−k
,
3c
since
∂ 2 qB
∂s2B
=
2sA (θ−k−3csA )
(4sA −sB )3
< 0.
The FOC is equivalent to the following quadratic equation in sB :
βcs2B − ((3 + 2β)csA − (1 − 2β)(θ − k))sB + 4β(θ − k + csA )sA = 0
(A8)
The discriminant of the quadratic equation
∆ = ((θ − k) − 3csA )((1 − 2β)2 (θ − k) − (1 + 2β)(3 − 2β)csA ) = YB is positive whatever sA .
The higher real root is a higher than sA , then it is not a valid solution.
20
The lower real root of the quadratic equation (A8) is:
(3 + 2β)csA − (1 − 2β)(θ − k) −
sB (sA ) =
2βc
√
YB
> 0.
(A9)
The solution has also to observe additional conditions, i.e., sB (sA ) < sA and sB (sA ) < 2 θ−k
−2sA .
c
◦ sB (sA ) < sA if and only if sA >
1+2β θ−k
3(1−β) c
◦ sB (sA ) < 2 θ−k
− 2sA if and only if sA <
c
1
1+2β
>
θ−k
.
3c
1 θ−k
.
1+2β c
1+2β
3(1−β)
when β < 41 , we obtain the best response function of NGO B:

1+2β θ−k


s
if sA < 3(1−β)

c
 A
1
BR
1+2β
θ−k
1 θ−k ;
(i). when β < 4 , sB (sA ) =
sB (sA )
if 3(1−β) c ≤ sA ≤ 1+2β
c



1 θ−k
 2 θ−k − 2s
if sA > 1+2β c
A
c
Since
>
(ii). when β ≥ 14 , sBR
B (sA ) = sA .
By solving the FOC (A7) as a quadratic equation in sA , this solution can be rewritten as the sA
best response of NGO B as a function of sB :

θ−k


− s2B

 c
(i). when β < 41 , sB−BR
(sB ) =
sB
A
A (sB )



 s
B
(ii). when β ≥ 41 , sB−BR
(sB ) = sB .
A
√
(3+2β)csB −4β(θ−k)− YAB
B
>
with sA (sB ) =
c
θ−k
3c
if sB <
if
4β θ−k
1+2β c
4β θ−k
1+2β c
if sB >
≤ sB ≤
1+2β θ−k
3(1−β) c
;
1+2β θ−k
3(1−β) c
and
YAB = (4β(θ − k) − 3(3 − 2β)csB )(4β(θ − k) − (1 + 2β)csA ) > 0 if sB >
4β θ−k
.
1+2β c
Quality equilibria
To obtain quality equilibria, we equalize the two reaction functions.
(sB ) has two real solutions if and only if
The equality sA (sB ) = sB−BR
A
∆ = Y = (1 − 3α)2 − 4(4 + α(−5 + 3α))β + 4(2 − α)2 β 2 > 0, that is, if and only if
α < α(β) =
√
3+2β(4β−5)−4(1−β) 6β
(3−2β)2
<
1
3
or α > α(β) =
√
3+2β(4β−5)+4(1−β) 6β
(3−2β)2
In this case, the two real solutions are
s∗B =
√
β(3−10β)α2 +(3−5β+4β 2 )α−1+10β−4β 2 −(1−2β+αβ) Y θ−k
3(1−β(α−2+2(1−(1−α)α)β))
c
and
s∗∗
B =
√
β(3−10β)α2 +(3−5β+4β 2 )α−1+10β−4β 2 +(1−2β+αβ) Y θ−k
.
3(1−β(α−2+2(1−(1−α)α)β))
c
The associated values of sA are
21
> 13 .
√
2β(−3+2β)α2 +(3+5β−10β 2 )α+1−β+4β 2 +(1+β−2αβ) Y θ−k
6(1−β(α−2+2(1−(1−α)α)β))
c
s∗A =
and
s∗∗
A =
√
2β(−3+2β)α2 +(3+5β−10β 2 )α+1−β+4β 2 −(1+β−2αβ) Y θ−k
.
6(1−β(α−2+2(1−(1−α)α)β))
c
To end the proof we have to check that the equilibrium candidates satisfy the following four
conditions:
1
(i). s∗B > 0 and s∗∗
B > 0 ⇔ α > 3;
(ii). s∗A >
θ−k
3c
and s∗∗
A >
(iii). s∗A <
θ−k
c
−
s∗B
2
θ−k
3c
⇔ α > 13 ;
and s∗∗
A <
θ−k
c
−
s∗∗
B
;
2
these conditions are satisfied without any additional
conditions.
∗∗
(iv). s∗A > s∗B and s∗∗
A > sB .
√
(1−α)−(7−5α+4α2 )β+2(2−3α+4α2 )β 2 +(1−β) Y θ−k
.
2(1−β(α−2+2(1−(1−α)α)β))
c
• s∗A − s∗B =
If (1 − α) − (7 − 5α + 4α2 )β + 2(2 − 3α + 4α2 )β 2 > 0, that is,
√
−(1−5β+6β 2 )+ (1−2β)(1+β(8+β(−91+46β)))
=α
e(β), s∗A − s∗B > 0.
α<
8β(1−2β)
If (1 − α) − (7 − 5α + 4α2 )β + 2(2 − 3α + 4α2 )β 2 < 0, that is, α > α
e(β), a necessary
condition is that α >
∗∗
• s∗∗
A − sB =
1
.
2(1−2β)
√
(1−α)−(7−5α+4α2 )β+2(2−3α+4α2 )β 2 −(1−β) Y θ−k
.
2(1−β(α−2+2(1−(1−α)α)β))
c
∗∗
If (1 − α) − (7 − 5α + 4α2 )β + 2(2 − 3α + 4α2 )β 2 < 0, that is, α > α
e(β), s∗∗
A − sB < 0.
If (1 − α) − (7 − 5α + 4α2 )β + 2(2 − 3α + 4α2 )β 2 > 0, that is, α < α
e(β), a additional
necessary condition is that α <
Since α(β) ≤
√
β≥
7− 33
,
16
1
2(1−2β)
∀β, α(β) =
1
.
2(1−2β)
1
2(1−2β)
when β =
√
7− 33
,
16
and α
e(β) ≤
we obtain the following quality equilibria:
(i). If β > 14 , then there are multiple equilibria:

i
h
4α2
θ−k θ−k
 sB ∈
, c when α <
2
i
h 4−4α+3α c
sA = sB with
 s ∈ 2(4α−1) θ−k , θ−k when α ≥ 2
B
3(1+2α) c
c
5
√
7− 33
16
1
and α < 2(1−2β)
, then there are multiple equilibria:
h
i
1+2β θ−k θ−k
sA = sB with sB ∈ 3(1−β)
,
;
c
c
(ii). If
<β≤
1
4
2
5
22
1
2(1−2β)
when
(iii). If β <
√
7− 33
16
(iv). If β <
√
7− 33
16
and α < α(β) = 3+2β(4β−5)+4(1−β)
(3−2β)2
h
i
1+2β θ−k θ−k
sA = sB with sB ∈ 3(1−β)
, c ;
c
and α(β) < α <
1
,
2(1−2β)
√
6β
, then there are multiple equilibria:
∗∗
then there are two equilibria: (s∗A , s∗B ) and (s∗∗
A , sB )
∗
∗∗
with s∗A > s∗∗
A and sB < sB ;
(v). If α > α(β), then there are one unique equilibrium: (s∗A , s∗B ).
23