nyu stern school of business - the role of the beneficiary...

A

purewara©

K

becjagpnbascusama

r

h0

Journal of Retailing 90 (2, 2014) 255–274

The Role of the Beneficiary in Willingness to Pay for SociallyResponsible Products: A Meta-analysis

Stephanie M. Tully, Russell S. Winer ∗Stern School of Business, New York University, 40 West 4th Street, New York, NY 10012, United States

bstract

Many companies have made significant investments in socially responsible production practices for their products. Environmentally safe cleaningroducts, fair trade coffee, and sustainable seafood are just a few examples. In this paper, we conduct a meta-analysis of over 80 published andnpublished research papers across a large number of product categories to better understand differences in willingness to pay (WTP) for sociallyesponsible products. In particular, we are interested in whether the beneficiary of the social responsibility program—humans, animals, or thenvironment—affects WTP. We use two dependent variables: the percentage premium people are willing to pay and the proportion of respondentsho are willing to pay a positive premium. We find that the mean percentage premium is 16.8 percent and that, on average, 60 percent of respondents

re willing to pay a positive premium. Importantly, across both dependent measures, we find that WTP is greater for products where the socially

esponsible element benefits humans (e.g., labor practices) compared to those that benefit the environment. Implications for retailers, manufacturers,nd future research are discussed.

2014 New York University. Published by Elsevier Inc. All rights reserved.

ent; F

r2

muitbiaVepet

eywords: Meta-analysis; Social responsibility; Willingness-to-pay; Environm

Socially responsibly produced products and services areecoming more important than ever for retailers as their pres-nce continues to increase dramatically. Environmentally safeleaning products, fair trade coffee, and sustainable seafood areust a few examples of this growing trend. Recently, Levi Straussnnounced a line of jeans with a pitch “These jeans are made ofarbage.” The Waste < Less jeans are composed of at least 20ercent recycled plastic (BusinessWeek 2012). Although compa-ies and retailers offering socially responsible products provide aenefit to society, economic incentives are often a catalyst for, ort least an input into, the decision by a firm or retailer to provideocially responsible products (Karnani 2012). Understandingonsumers’ willingness to pay (WTP) for products producedsing socially responsible practices is important for the futureuccess of such endeavors. Despite the growth in this product

rea and previous research on WTP, we found few studies inarketing that have directly addressed this issue, with almost

ll focused on WTP for either coffee or apparel with socially

∗ Corresponding author. Tel.: +1 212 998 0540; fax: +1 212 995 4006.E-mail addresses: [email protected] (S.M. Tully),

[email protected] (R.S. Winer).

bWib(

up

ttp://dx.doi.org/10.1016/j.jretai.2014.03.004022-4359/© 2014 New York University. Published by Elsevier Inc. All rights reserv

air trade; Animal rights

esponsible attributes (e.g., De Pelsmacker, Driesen, and Rayp005; Ha-Brookshire and Norum 2011).

Despite other disciplines examining this issue, there are stillany unanswered questions about consumers’ WTP for prod-

cts produced using socially responsible methods. For instance,t is unclear how much more consumers on average are willingo pay for socially responsible products in general. As mighte expected from the rise in socially responsible product offer-ngs, many studies have found that consumers are willing to pay

relatively large premium for these products (e.g., Aguilar andlosky 2007; De Pelsmacker, Driesen, and Rayp 2005; Saphores

t al. 2007). However, a smaller number of studies have reportedremiums closer to zero (e.g., Grönroos and Bowyer 1999) orven to be negative in rare cases (e.g., Akkucuk 2011). In addi-ion, the retail sales for some socially responsible products haveeen slow. For example, Clorox’s widely heralded line of Greenorks cleaning products has had considerable difficulty gain-

ng traction with retailers and consumers. Importantly, price haseen suggested to be one of the top barriers to green consumption

Gleim et al. 2013).

Beyond the average WTP for socially responsible prod-cts, factors that influence WTP for socially responsibleroducts are still relatively unknown. Research has examined

ed.

http://crossmark.crossref.org/dialog/?doi=10.1016/j.jretai.2014.03.004&domain=pdfdx.doi.org/10.1016/j.jretai.2014.03.004mailto:[email protected]:[email protected]/10.1016/j.jretai.2014.03.004

2 l of R

is1reicepoancsv

rgMkacdms

WLcowasrpeswfcwpp

antirsrt

c

sdabnhowwf

B

gadtrmsutucpdmtt

tmmosbcre(pTnsfiptwfiaim

56 S.M. Tully, R.S. Winer / Journa

ndividual differences that can explain variations in WTP forocially responsible products (e.g., Balderjahn 1988; Roberts996). Relatively less is known about the factors of sociallyesponsible products themselves that may contribute to differ-nces in willingness to pay. Understanding the economic value ofmplementing different types of socially responsible practices toonsumers is of interest to society, manufacturers, and the retail-rs who distribute the manufacturers’ products. For instance, areroducts advertising the use of fair wages to employees morer less likely to increase a consumer’s WTP compared to thosedvertising the use of environmentally friendly tactics? This isot a trivial matter. A recent article in Forbes said that manyorporate leaders, realizing lackluster return on social respon-ibility investments, want to know the best way to increase thealue of these investments (Klein 2012).

To examine systematic differences in WTP for sociallyesponsible products, we employ a meta-analysis of a largeroup of studies that have explored WTP for such products.eta-analyses have been widely used in marketing. Some well-

nown examples are in the areas of advertising (Assmus, Farley,nd Lehmann 1984; Sethuraman, Tellis, and Briesch 2011), pri-ing (Bijmolt, van Heerde, and Pieters 2005; Tellis 1988), andiffusion models (Sultan, Farley, and Lehmann 1990). Theseeta-analyses have helped the marketing field develop a large

et of empirical generalizations (Hanssens 2009).Meta-analysis has also been used to understand variation in

TP for socially responsible products (Cai and Aguilar 2013;agerkvist and Hess 2011). However, past meta-analyses haveoncentrated on one product category (e.g., wood products) andne type of social responsibility (e.g., the environment). Byidening the set of studies included in the meta-analysis, we are

ble to make broader generalizations about average WTP forocially responsible products and can examine factors not yetigorously tested. Using each study in a paper as one data pointermits an analysis of multiple product categories using differ-nt types of social responsible practices that differ in strength ofocial norm. In doing so, our meta-analysis is able to examinehether consumers are likely to pay a larger or smaller premium

or socially responsible products that benefit the environmentompared to those that benefit human working conditions andhether differences in social norms for different products canredict variation in willingness to pay. See Table 1 for a com-arison of the current research to previous meta-analyses.

The goal of this meta-analysis is to generate a set of findingsbout consumer WTP for socially responsible products that areot conditional on the particulars of any single study, productype, or social responsibility type and to provide researchers, pol-cy makers, and retailers with a concise synthesis of the researchesults. Moreover, we aim to test moderators of the WTP forocially responsible products that may be of particular interest toetailers such as the type of socially responsible beneficiary andhe social norms associated with a socially responsible product.

Main hypotheses

In the current work, we examine whether variation in WTPan be explained by factors associated with the product and

p

wh

etailing 90 (2, 2014) 255–274

ocial responsibility type as well as general characteristics of theata collection method. More explicitly, we suggest that the vari-nce in WTP for socially responsible products may be influencedy the beneficiary of the social responsibility and the socialorms for the socially responsible product. We also examineow product domain, product certification and characteristicsf WTP elicitation such as the year the study was conducted,hether the method of elicitation is incentive-compatible, andhether respondents are allowed to respond with negative values

or WTP influence WTP estimates.

eneficiary of social responsibility

The ISO 26000, an organization tasked with standardizinguidelines for social responsibility, defines social responsibilitys the “responsibility of an organization for the impacts of itsecisions and activities on society and the environment, throughransparent and ethical behaviour” (2010, p. 3). Thus, socialesponsibility is a broad term to describe anything a companyight do that benefits society at large. Similarly, socially respon-

ible products provide a benefit to society at large. However, bysing the term socially responsible products, we are referringo products which benefit society through business practicessed in the creation of the product. Therefore, the manufacturinghoices a company makes (through the labor they employ, theackaging they choose, the distribution network they use) areone in a way to provide benefits to society. Choices a companyakes outside of the business practices employed in producing

he good (e.g., giving profits to charity) fall outside the scope ofhe current research.

Broadly speaking, socially responsible products can benefithree types of beneficiaries: the people of a society, the ani-

als in a society, or the environment. The current literatureakes it difficult to identify differences in WTP across type

f social responsibility. The vast majority of studies on WTP forocially responsible products focus on one socially responsibleeneficiary so whether the type of social responsibility impactsonsumers WTP is unclear. Only three articles in our literatureeview yielded studies that looked across multiple types of ben-ficiaries of socially responsible products. Loureiro and Lotade2005) look at WTP for fair trade coffee (which benefits peo-le) and shade-grown coffee (which benefits the environment).he estimates of WTP across these two types of products areearly identical. Carlsson, Garcia, and Löfgren (2010) foundimilar results. However, Hustvedt, Peterson, and Chen (2008)nd that there is a greater interest in socially responsible woolroducts which benefit animal rights than wool products benefit-ing the environment. Given the small range of studies examiningillingness to pay across beneficiaries, and the differences inndings across the few studies, it is currently difficult to makeny generalizations. Thus, if a retailer has the choice of stock-ng a product that provides fair wages to its employees or one

ade from renewable resources, it is currently unknown which

roduct will allow for a greater price premium.

We hypothesize that the beneficiary of social responsibilityill impact WTP. Although environmentally friendly productsave garnered much hype in recent years, we suggest that

S.M. Tully, R.S. Winer / Journal of Retailing 90 (2, 2014) 255–274 257

Table 1Comparison of previous meta-analyses to current meta-analysis.

Meta-analysis Social responsibilitybeneficiary

DV Main findings

Cai and Aguilar (2013) Environment Percent premium Mean WTP = 12.2 percent. Consumers are willing to pay a greaterpercent premium for products with lower base prices, for productsthat are more frequently purchased, when the elicitation method isindirect rather than direct, for contingent valuation compared toconjoint analysis, and for more recent studies.

Lagerkvist and Hess (2011) Animals Percent premium Younger age and greater income is positively associated with WTP.Consumers were found to be willing to pay a lower percentpremium for products with legal regulations, when cheap-talkscripts or double-bound dichotomous choices are used.Peer-reviewed papers had a lower overall WTP.

Current research Environment, people,and animals

Percent premium andproportion of people

Mean WTP = 16.8 percent. 59.7 percent are willing to pay apositive amount. Products that benefit people compared to thosethat benefit the environment increase WTP both in terms of thepercent premium and the proportion of people willing to pay a

slcsaWdUgasgosnfippltpbabth

D

dvswimp

trd

S

snlfwtu

natamspsmerp

ntGnoT

ocially responsible products that benefit people may be moreikely to increase sales. While polls suggest that many peopleare about the environment (Nielsen 2011), as noted previously,ales of “green” products have often proven lackluster (Cliffordnd Martin 2011; UNEP 2005). Additionally, if an increase inTP for socially responsible products is viewed as a form of

onation toward the beneficiary, charitable contributions in thenited States show that only 3 percent of charitable donationso toward animals and the environment combined compared to

much larger percentage going to causes that benefit peopleuch as education and human services (Giving USA 2013) sug-esting that people are more likely to pay for things that benefitther humans. Therefore, our main hypothesis is that WTP forocially responsible products that benefit people will be sig-ificantly greater than those that benefit the environment. Thendings of Hustvedt, Peterson, and Chen (2008) suggest thatroducts benefitting animals may well elicit greater WTP com-ared to those benefitting the environment. However, given theimited number of charitable donations toward animal causes,his prediction is less clear. Further, how socially responsibleroducts that benefit both the environment in addition to anothereneficiary will compare to products benefitting the environmentlone is also unclear. One might expect that adding additionaleneficiaries will provide more motivation for consumers, buthis is speculative as there is no literature to support a directionalypothesis.

omain

Beyond the type of social responsibility, there is some evi-ence that differences in WTP for socially responsible productsary as a function of product type. If a retailer is consideringtocking a socially responsible product, research suggests that

hich type of product it is may affect consumers’ WTP. For

nstance, Akkucuk (2011) finds significant differences in howuch consumers are willing to pay for recycled furniture com-

ared to other products such as tires or cell phones. However,

pss

positive amount. Social norms may increase the proportion ofpeople willing to pay a premium.

his has not systematically been examined. Since there is cur-ently limited research looking at WTP across product domains,irectional differences are exploratory.

ocial norms

Even within product domains (e.g., wood products), differentocially responsible products may elicit differences in willing-ess to pay. Consumers were found to be willing to pay aower price premium over the base price for an environmentallyriendly new home compared to other environmentally friendlyood products such as a kitchen remodeling job or a ready-

o-assemble chair (Ozanne and Vlosky 1997). However, it isnknown what may be driving these differences.

It has been reported that frequently purchased products gar-er a greater WTP than less frequently purchased products (Caind Aguilar 2012; Teisl et al. 2002). Teisl et al. speculate thathis is because frequently purchased products are seen as having

greater impact on society. Yet, another possibility is that thereay be stronger social norms for some types of socially respon-

ible products compared to others. Products that are frequentlyurchased may have more visibility and thus greater social pres-ure to purchase products produced in a socially responsibleanner. If social norms, rather than frequency of purchase, can

xplain differences in WTP, it would be of particular interest toetailers since social norms are more malleable than changingroduct characteristics such as purchase frequency.

The evidence for social norms is currently mixed. Socialorms have been shown to predict WTP in some environmen-al contexts (Spash et al. 2009), but not consistently (Carlsson,arcia, and Löfgren 2010). However, past research on socialorms has primarily examined social norms within the contextf one product type or by manipulating social norms artificially.hus, in the current research, we investigate whether average

erceptions of social norms for different types of socially respon-ible products can predict WTP. We hypothesize that strongerocial norms will elicit greater WTP.

2 l of Retailing 90 (2, 2014) 255–274

C

ogscFqppisp

C

bifsstrtastdedOWch

iBatnts

C

Jrts(F

Table 2Hypotheses.

Independent variable Hypothesis

Beneficiary H1: Socially responsible products that benefit peoplewill have a greater proportion of participants that arewillingness to pay a premium compared to sociallyresponsible products that benefit the environment.H2: Socially responsible products that benefit peoplewill garner a greater average premium compared toproducts that benefit the environment.

Social Norms H3: Socially responsible products that have greatersocial responsibility norms will have a greaterproportion of participants that are willing to pay apremium.H4: Socially responsible products that have more socialresponsibility norms will garner a greater averagepremium compared to socially responsible products that

appodpU1

A

rAa

rgaotgonq

id(r

M


ertification

A product that has been certified has passed some formf quality assurance test or met some objective standards thatuarantees the product meets a minimum level of social respon-ibility. It typically costs money for companies to becomeertified so the cost of products that are certified is often higher.or example, Fair Trade certification requires the payment ofuarterly licensing fees (Fair Trade USA). Most companies willass this additional cost on to consumers often resulting in higherrices for fair traded products (Stecklow and White 2004). Thus,t is in a retailer’s interest to know whether certification of aocially responsible product increases willingness to pay for theroduct.

hanges in WTP over time

It is currently uncertain whether WTP for socially responsi-le products is changing over time, and if so, in which directiont is changing. There is both anecdotal and academic research inavor of increasing WTP over time. Trends show that corporateocial responsibility is becoming more prominent and that con-umers are increasingly looking at social responsibility reportshat companies furnish (White 2012). Recently, Nielsen (2011)eported a 5 percent increase in the last two years in willingnesso reward a company that gives back to society. Additionally,cademic research has shown a significant positive relation-hip between time and WTP (Cai and Aguilar 2013). However,here is also both anecdotal and academic support in favor ofecreasing WTP over time. In the last few years, sales of someco-friendly options, such as cleaning supplies, have been inecline (Clifford and Martin 2011; Kurutz 2011). Additionally,zanne and Vlosky (2003) find a negative relationship betweenTP for socially responsible products and time. Therefore, it is

urrently unclear whether WTP for socially responsible productsas systematically changed over time.

In addition to the year of data collection, other character-stics of the study design may be relevant moderators of WTP.roadly, these include whether the study is incentive compatiblend whether the study allows negative responses. These charac-eristics are discussed further in the WTP section below. In theext section, we provide some background on what willingnesso pay is conceptually and how we use it as a measure of effectize in our meta-analysis.

Our hypotheses are summarized in Table 2.

Willingness to pay

onceptual underpinnings

Definitions of customer value or WTP vary (see Jedidi andagpal 2009). A commonly used definition is the economic term,eservation price, or the maximum amount a customer is willing

o pay for a good, or, stated differently, the price at which a con-umer is indifferent between buying and not buying the productJedidi and Zhang 2002). However, there are other definitions.or example, Hauser and Urban (1986) define reservation price

tea

have less social responsibility norms.

s the minimum price that the consumer will not purchase theroduct. Varian (1992) indicates that reservation price is therice at which or below the consumer will purchase one unitf the product. As Jedidi and Jagpal (2009) note, these threeefinitions have slightly different interpretations in terms of therobability of purchase at the reservation price with Hauser andrban being zero, Jedidi and Zhang being .5, and Varian being.0.

pproaches to measuring WTP

There are many methods used to elicit WTP. Open-endedesponses allow a respondent to supply their own value of WTP.lternatively, multiple choice methods provide respondents with

range of options from which they can choose.Another approach, often used by economists to value natural

esources, is called contingent valuation. Two types of contin-ent valuation are possible. A within-subjects method provides

respondent with a starting price and continues to incrementr decrement the price from this point until a consumer revealsheir WTP. A between-subjects contingent valuation techniqueives different prices to each respondent and asks them whetherr not they would buy the product at the stated price in yes oro form. The sum of all the respondents’ answers to the singleuestion allows for the creation of a demand curve.

Conjoint analysis estimates WTP for an attribute by estimat-ng the indirect utility function, V, based on the experimentalesign and then estimating the marginal rate of substitutionMRS) between an attribute (m) and price (p) using the followingelationship:

RS = −∂V/∂m∂V/∂p

. (1)

Auctions, both those that are naturally occurring and those

hat are staged as part of an experiment, can also be used tostimate WTP (e.g., Ellis, McCracken, and Skuza 2012). In anuction, consumers bid on the right to buy the item up for auction.

l of R

atApt

sacsapBuhemreits

ittTicmsc

W

sydaWttualqsctc

o

tdapeasbab

U

avIsciibpswciwttTiw

P

mspttsiwpfaatma

S.M. Tully, R.S. Winer / Journa

Real purchase data can be used to estimate WTP. Hedonicnalyses typically use cross-sectional data where there is varia-ion in product prices for different levels of product attributes.lternatively, field experiments can be run varying prices androduct characteristics. The resulting sales can be used to inferhe demand and WTP for the manipulated product features.

While each method of eliciting WTP may result inmall differences, an underlying characteristic of all of thesepproaches is whether the method of elicitation is incentiveompatible (Miller et al. 2011). Incentive compatible mea-urement approaches attempt to ensure that respondents haven incentive to “tell the truth” about their WTP. For exam-le, Wertenbroch and Skiera (2002) apply the well-knownecker–DeGroot–Marschak (BDM) approach for measuring thetility of lotteries. Ding (2007) and Dost and Wilken (2012)ave also applied the BDM approach to conjoint analysis andstimating ranges of WTP rather than a point estimate. Someethods like auctions are incentive compatible by design. While

eal purchase data cannot capture a person’s absolute high-st willingness to pay, purchase data is also considered to bencentive-compatible (Wertenbroch and Skiera 2002) becausehe purchase decision has real consequences. Other methodsuch as hypothetical multiple choice scenarios are not.

Given the wide variety of approaches to estimating WTP, its not surprising that there are a number of studies that attempto compare them. The most comprehensive comparison study inhe marketing literature was conducted by Miller et al. (2011).he results showed that the incentive-based methods (BDM and

ncentive-based conjoint) had much lower bias against the “real”riterion than the “hypothetical” methods. Thus, we expect thatethods that are incentive compatible will elicit lower WTP for

ocially responsible products than methods that are not incentiveompatible.

TP as a measure of effect size

In most of the previous meta-analyses in marketing, the effectize or dependent variable studied is based on statistical anal-ses of time-series of other secondary data not on primaryata collection from human subjects. Examples include pricend advertising elasticities. As noted above, the definitions of

TP/reservation price have nuanced differences. In addition,he vast majority of papers included in our meta-analysis basedheir results on experiments or other data collection methodstilizing human subjects. Unfortunately, many of the papers wenalyzed did not disclose the specific instrument used to col-ect the data leaving the subjects’ interpretation of the WTPuestion or questions unclear. As a result, we make the neces-ary assumption that differences in interpretation of the WTPonstruct across studies do not have an effect on the parame-er estimates of our models and, by implication, our substantiveonclusions.

Dependent variables

In the current research, we consider two dependent variablesr effect sizes to examine consumers’ WTP. We considered both

C

t

etailing 90 (2, 2014) 255–274 259

he percentage change in WTP from a base price (the uncon-itional percent premium) for the socially responsible products well as the proportion of participants willing to pay a priceremium for a socially responsible product. These two depend-nt measures were chosen due to their frequency in reportings well as their ability to make meaningful comparisons acrosstudies. Further, these two dependent variables capture related,ut not conceptually overlapping aspects of consumers’ WTP,s evidenced by their moderate correlation in studies that reportoth (r = .54).

nconditional percent premium

The percentage difference from the base price that consumersre willing to pay allows us to quantify the increased monetaryalue that consumers place on the socially responsible element.t is essentially capturing the percent premium a socially respon-ible product can demand. If a socially responsible element isostly to implement, this measure can be useful in determin-ng whether the increased cost can be offset by the potentialncrease in WTP. However, the percent difference in WTP isoth a function of the number of people who are willing to pay aremium and the actual valuation placed on the socially respon-ible element. This dependent measure is capturing the averageillingness to pay across all participants (the unconditional per-

ent premium). For instance, if the average premium in WTPs 25 percent, this might be because the average respondent isilling to pay 25 percent more or it might be because half of

he respondents are willing to pay a 50 percent premium andhe other half of the respondents will not pay any premium.herefore, while the average premium across all respondents

s important, so is understanding the proportion of participantsilling to pay a premium.

roportion of participants willing to pay a premium

The proportion of participants who are willing to pay a pre-ium of any size can be seen as an indicator of potential market

ize. As this percentage gets larger, it suggests that a greater pro-ortion of the population value the socially responsible elemento some extent. However, this dependent measure cannot quan-ify the change in WTP. For instance, a study in which everyingle person is willing to pay a 1 percent premium and a studyn which every person is willing to pay a 30 percent premiumould both show 100 percent of the participants willing to pay aremium. However, clearly these outcomes have important dif-erences when determining both customer value and pricing for

socially responsible good. Because of each of their strengthsnd limitations, we feel both dependent measures are importanto be able to understand meaningful differences in WTP in a

eta-analysis. Further, differences in these dependent variablescross analyses are of theoretical and practical interest.

onditional percent premium

We also calculated the average percent premium among onlyhe participants who were willing to pay more for the socially

2 l of R

rtoptFiea

I“mclTaeW

p(larpwj

owr(SbcobafetAeutisucsso

ya(Dpamstwpa

C

mmfcuuuudesdvf

I

rmpihustoiFuctteus


esponsible product. This may in fact be the most accurate metrico understand WTP since it is a function of both the proportionf participants who are willing to pay a premium and the overallercent premium. Unfortunately, only 21 papers (48 observa-ions) provided enough information to calculate this measure.urther, these observations did not provide enough variance in

ndependent variables (e.g., 41 of 48 observations benefitted thenvironment). Thus, we did not use this dependent measure inny of our analyses.

Data collection

Data were collected using electronic databases (i.e., theSCWeb of Knowledge, Google Scholar, NYU Library CatalogsBobcat”) and reference lists from identified studies includingeta-analyses (farm animal welfare: Lagerkvist and Hess 2011;

ertified wood: Cai and Aguilar 2013) and review papers (eco-abels: Gallastegui 2002; fair trade: Anderfer and Liebe 2012).he literature search used the following keywords, individu-lly and in combination: animal rights, eco-label, environment,thical, fair trade, green, labor, social, willingness to pay, andTP.We included both published and unpublished literature since

ublication bias can skew the results of a meta-regressionStanley 2005). To further encourage the inclusion of unpub-ished papers on the topic, we also searched the websites ofuthors who have an established history of conducting socialesponsible WTP studies. Just under ten percent of the includedapers are working papers. In addition to including unpublishedorks, the final sample includes studies from a broad range of

ournals, totaling over fifty different outlets.In order for the study to be included, it had to meet a range

f criteria. In general, we examined whether consumers areilling to pay a premium for products produced using socially

esponsible production practices that do not offer private benefitse.g., improved health, economic advantages) for the consumer.pecifically, we decided to restrict our data to products having aenefit conferred solely on society so that any incremental WTPould be attributed to this societal benefit. As a result, studies onrganic foods, which are commonly believed to provide healthenefits, and studies on hybrid vehicles and energy reducingppliances, which provide economic incentives, were excludedrom our analysis. Without this requirement, personal differ-nces in valuations of the private benefits may artificially inflatehe valuations of the WTP for the socially responsible element.dditionally, the social responsibility element had to be inher-

ntly part of the product itself. For instance, the company couldse socially responsible practices in the types of materials used,he production methods employed, or the labor required in mak-ng or testing the product. Products from companies who doocially responsible things outside of the production of the prod-ct can be categorized as products from socially responsible

ompanies rather than being socially responsible products. Thus,tudies regarding other goodwill that the company generates out-ide of the sale of a product, such as donations to charity, areutside the scope of the current research.

eaT

etailing 90 (2, 2014) 255–274

In total, the full data set includes eighty-one usable papersielding 174 observations. Thirty-six papers (96 observations)re included in both dependent measures data sets. Nine papers11 observations) are unique to the proportion of participantsV set and thirty-five papers (67 observations) are unique to theercent difference from base price DV set. Given the increasingmount of research in the area of social responsibility and theany types of behavior that can be considered socially respon-

ible, we do not claim to provide a comprehensive review ofhe social responsibility WTP literature. However, we believee have collected a reasonably large and representative sam-le of studies on the WTP for socially responsible products fornalysis.

alculating the dependent variables

Many papers directly report one or both of the dependenteasures. For papers that did not directly report one of botheasures, we attempted to compute the dependent variables

rom empirical results provided in the paper. When multiplehoice responses offered ranges, the midpoint of the range wassed as the WTP for that option. When conjoint analysis wassed and WTP estimates were not provided by the authors, wesed expression (1) to calculate the marginal WTP based on thetility function estimates. In rare cases, social responsibility wasescribed as being negative (i.e., the product is harmful to thenvironment or employees in some way). For these studies, theign of the percentage change of WTP was reversed. If a studyid not provide sufficient information to calculate a dependentariable, that paper was not used as part of the observation setor that dependent variable.

ndependent variables

Socially responsible benefit. The beneficiaries of sociallyesponsible undertakings can be broadly categorized into threeain types: animals, the environment, or people. Products which

rovide positive social benefits to animals provide better liv-ng conditions for animals through free range practices, moreumane types of castration etc. Environmentally friendly prod-cts protect rainforests, reduce pollution, and preserve waterources, for example. We also categorized practices such as sus-ainable seafood as benefitting the environment since the usef these practices is more about eco-system sustainability thant is about providing benefits to the sea animals themselves.inally, people can be beneficiaries of socially responsible prod-cts when the product provides fair wages or good workingonditions to its employees. In some papers, there were mul-iple recipients of the socially responsible element. In all ofhese cases, the beneficiary was the environment in addition toither people or animals. These were coded as “multiple” andsed to understand if providing more than one beneficiary typeignificantly increases willingness to pay.

Domain. We broadly categorized the product type used inach study. There were five categories that were clearly differentnd that had a reasonable number of observations per group.hese were wood, electronics, food, clothing, and other. “Other”

l of R

ico

crasteaDip

arNtprtnbrtppbatorrso

mc

spotia3lc

ih

v

pa

pmt“wtttns

pPctl

t

eaWausTe

vshectmwararpcr


ncorporated any product that was not encompassed by the otherategories, but did not have enough observations to serve as itswn category (e.g., hotel rooms).1

Social norms. As a means of understanding consumers’ per-eptions about the social norms associated with different sociallyesponsible products, we polled an online sample and explicitlysked them to rate the social norms of the products used in thetudies included in our meta-analysis. Two hundred and six par-icipants from Mechanical Turk participated in the short study inxchange for a small sum of money. Twenty participants failedn instructional manipulation check (Oppenheimer, Meyvis, andividenko 2009) and were thus excluded from the analyses leav-

ng a final sample of 186 participants (mean age = 35.5, 60.2ercent male).

The study procedure was as follows. Participants were given definition of social responsibility and explained how socialesponsibility can benefit people, animals, or the environment.ext, they read a definition of social norms. We then explained

hat they would rate a series of products with socially responsibleractices that benefit either people, animals, or both the envi-onment in terms of how strong any social norms to buy theseypes of products are and provided an example product that wasot part of the data set (disposable diapers with environmentalenefits). Each participant rated 25 randomly selected sociallyesponsible products from the full list of 69 possible combina-ions. They were asked “How strong are the social norms tourchase [product] with socially responsible benefits for [peo-le/animals/the environment]” (1 = there are no social norms touy this, 4 = there are moderate social norms to buy this, 7 = therere strong social norms to buy this). On each page we providedhe respondent with a reminder definition and some examplesf what constituted socially responsible benefits for the envi-onment, for people, and for animals so they could be used as aeference. Means from this study were used as an explicit mea-ure of social norms.2 The study yielded a reasonable amountf variance with means ranging from 2.41 to 5.65.

Certification. Certification was coded as present if the paperentioned that the product was labeled or described as being

ertified.Imputed year. We examined whether willingness to pay for

ocially responsible products has changed over time. Someapers specified the time period of the data collection whilethers did not. To impute the data collection year for the studieshat did not specifically provide one, we averaged the differencen number of years between data collection and publication forll studies that supplied the dates. The average difference was.72 years. Thus, we subtracted 3.72 years from the date of pub-ication for the remaining studies to ascribe an approximate dataollection date.

Elicitation method. Studies were coded as incentive compat-ble if, as described in the WTP section, participants’ responsesad potential consequences for receiving or not receiving the

1 We thank a reviewer for the idea of including domain as an independentariable.2 Results of the social norm study are available upon request from the authors.

mfStoftO

etailing 90 (2, 2014) 255–274 261

roduct. This included studies using real purchase data as wells others such as auctions.

We also coded whether the method of elicitation allowedarticipants to respond with negative changes in WTP. Paperseasuring incremental WTP do not always permit respondents

o indicate negative values. For instance, a question that asks,How much more would you be willing to pay if the productere environmentally friendly?” does not allow a respondent

o provide a negative WTP even if the respondent is allowedo respond in an open-ended format. Such a question is likelyo skew WTP upwards. Thus, we expect that methods that doot allow negative WTP estimates will estimate higher WTP forocially responsible products compared to methods that do.

Peer-reviewed. This is a dummy variable for whether theaper was published in a peer-reviewed outlet (1) or not (0).ublication bias is expected to produce significant positiveoefficients on this variable indicating that peer-reviewed papersended to have higher values of WTP and thus result in higherikelihood of publication.

Papers used in the analyses and their main characteristics withhe dependent measures collected are available in Appendix A.

Data analysis

Many meta-analyses report Q-tests of homogeneity of theffect sizes where the null hypothesis is that all of the effects sizesre estimating the same population mean. Following Lipsey andilson (2001), we conducted Cochran’s Q-tests on the percent-

ge premium and the percentage willing to pay a premium andsing the appropriate Chi-square test rejected the null hypothe-is for both (χ2 = 1591.32 and 3508.74, respectively, p < .001).his indicates that it is necessary to model the differences inffect sizes using a set of design or independent variables.

We first ran a basic OLS regression on each of the dependentariables. However, it is well-known that meta-analysis modelsuffer from heteroscedasticity (Farley and Lehmann 1986). Theeteroscedasticity results from two sources. First, the effects arestimated in each individual study with varying amounts of pre-ision and different sample sizes. Second, multiple effects areaken from the same study using similar methodology. Since

any of the papers had multiple measures of the effects sizes,e followed the suggestion made by Bijmolt and Pieters (2001)

nd used the complete set of measurements from each studyather than a single measure representing all the results. Toccount for the sources of heteroscedasticity, we also used aandom effects estimation approach which was among the bestrocedures tested by Bijmolt and Pieters (2001) (the metaregommand from Stata; see Harbord and Higgins 2008). Theandom effects model is an appealing specification because itakes the assumption that the population effect sizes for dif-

erent studies are randomly drawn from a normal distribution.ince individual study variances were unavailable for most of

he observations, the weights used were the inverse of the logs

f the number of participants in the study. We used the log trans-ormation due to the highly varying number of participants inhe individual studies which ranged from 34 to over 100,000.f note, however, results from the OLS regression and from the

262 S.M. Tully, R.S. Winer / Journal of Retailing 90 (2, 2014) 255–274

Table 3Number of observations by independent variable for each dependent variable.

Variable Unconditional percent premium DV Proportion of participants DV Papers reporting both DVs# of observations # of observations # of observations

Willingness to pay 163 107 96Beneficiary

Environment 111 87 81People 22 12 7Animals 21 3 3Multiple (environment + other) 9 5 5

DomainWood 64 55 52Electronics 11 8 8Food 50 16 14Clothing 18 13 10Other 20 15 12

Certified 97 68 63Peer reviewed 145 93 83AI

ww

wttuadpamsf

tp1ttpstep

it

w

p

maaumr5p

a(cfrcsHa

i

S

r

llow negative 86 ncentive compatible 32

eighted random effects model are substantively equivalent soe only report the latter.3

To assess robustness of our findings, we ran the sameeighted random effects model using only the set of papers

hat report both dependent measures. While the sample size forhis model is considerably smaller, use of this model helps tonderstand whether discrepancies across dependent measuresre a function of differences in conceptualization across the twoependent measures or whether differences may be due to theapers included in the meta-regressions. Small discrepanciesrise when comparing results of the OLS and random effectsodels to the random effects regressions using the smaller data

et which reports both DVs. These discrepancies are discussedor the independent variables when applicable.

Empirical results

The unconditional percent premium across the 163 observa-ions reporting this measure ranged from −24.3 percent to 87.1ercent.4 The mean difference in WTP over the base price is6.8 percent, SD = 18.2 (t(163) = 11.78, p < .001). Thus, acrosshe 73 papers where we could get a measure of the uncondi-ional percent premium, people are willing to pay a substantialremium over the base price for a product with a socially respon-ible element. We note that about 2/3rds of the studies focus on

he environment as the beneficiary of the socially responsiblelement as seen in Table 3. This shows a lack of studies usingeople, animals, or any combinations of the three.5

3 In addition, we ran the models using alternative estimation proceduresncluding correcting for heteroscedasticity only. Again, the results were subs-antively equivalent across methods.

4 We eliminated four observations whose percentages differences in WTPere over five standard deviations above the mean.5 We will discuss the research implications of this lack of studies later in theaper.

gipasbfi

cct

31 284 2

Although the conditional percent premium (the percent pre-ium among participants who are willing to pay a positive

mount) did not provide enough observations to conduct a fullnalysis, we conducted an analysis of the difference between thenconditional percent premium and the conditional percent pre-ium among the available observations. Paired sample t tests

evealed that the conditional percent premium is on average.77 percent (SD = 3.69) greater than the unconditional percentremium, t(53) = 11.48, p < .001.

The mean proportion of participants who were willing to pay premium across the 107 observations is 59.7 percent, SD = 21.0t(106) = 29.43, p < .001). This proportion ranged from 9.8 per-ent to 92.6 percent. We note that about 80 percent of the studiesocus on the environment as the beneficiary of the sociallyesponsible element. Thus, as with studies measuring the per-ent difference dependent variable, there is a significant lack oftudies using people, animals, or any combinations of the three.istograms of the distribution of the two dependent variables

re shown in Fig. 1.The empirical results for the random effects models are shown

n Table 4.

ocially responsible benefit

Overall, results of the meta-analysis suggest that sociallyesponsible products which benefit the environment appear toarner a lower WTP than products benefitting other beneficiar-es. In support of H1 and H2, results demonstrate that a greaterroportion of consumers are willing to pay a premium and theverage percentage premium is greater for products where theocial responsibility benefits humans compared to products thatenefit the environment (the omitted base). This is a robustnding across all models and dependent measures.

The results suggest that people are willing to pay a signifi-antly greater percent premium for products benefitting animalsompared to products benefitting the environment. Whilehe results are directionally consistent for the proportion of

S.M. Tully, R.S. Winer / Journal of Retailing 90 (2, 2014) 255–274 263

A: Unconditional Percent Premium in Willingness to Pay

B: Proportion of Participants Willing to Pay A Premium

C: Bias Adjus ted Uncondition al Percent Prem ium in Wi lling ness to Pay

D: Bi as Adjus ted Proportion of Par ticipa nts Willi ng to Pay A Pr emium

0%

5%

10%

15%

20%

25%

30%

35%

≤ 0 0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100

% o

f Sam

ple

0%

5%

10%

15%

20%

25%

30%

0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-10 0

% o

f Sam

ple

0%

5%

10%

15%

20%

25%

30%

35%

≤ 0 0- 10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-10 0

% o

f Sam

ple

0%

5%

10%

15%

20%

25%

30%

0-50

% o

f Sam

ple

aw an

piftto

cnfmumm

F

ompdwbvmt

S

0-10 10-20 20-30 30-40 4

Fig. 1. Histograms of r

articipants dependent measure, there is no significant increasen the proportion of participants that are willing to pay a premiumor products benefitting animals compared to those benefittinghe environment. Of note, however, is that the number of observa-ions for animals is much smaller in the data set for the proportionf participants measure.

Products offering more than one socially responsible benefi-iary (i.e., the environment plus an additional beneficiary) showo significant increase in WTP in either of the models using theull data set, but demonstrate a significant increase in WTP in theodels using the observations that report both dependent meas-

res. The discrepancy in significance for WTP across dependenteasures is likely due to the small number of observations using

ultiple beneficiaries.The results of the beneficiary are interesting for two reasons.

irst, the biggest push for new products has been for those based

ciu

50-60 60-70 70-80 80-90 90-10 0

d bias adjusted results.

n environmental concerns such as the Clorox Green Works lineentioned in the introduction. The current results suggest that

roducts that offer human benefits such as good working con-itions may be able to obtain a greater price premium and haveider appeal than those focusing on animal or environmentalenefits. Second, as we noted from the results in Table 2, theast majority of research in this area has focused on environ-ental benefits. More research is needed on WTP for products

hat benefit humans, animals, or have multiple beneficiaries.

ocial norms

The social norm variable was created to provide more con-lusive evidence for whether social norms have a significantmpact on consumers’ WTP for socially responsible prod-cts. Unfortunately, the results of the meta-analysis again

264 S.M. Tully, R.S. Winer / Journal of Retailing 90 (2, 2014) 255–274

Table 4Results of weighted random effects regression.

Variable All observations Observations which report both dependent variables

Unconditional percentpremium (n = 163)

Proportion of participantsWTP a premium (n = 107)

Unconditional percentpremium (n = 96)

Proportion of participantsWTP a premium (n = 96)

Coefficient (SE) Coefficient (SE) Coefficient (SE) Coefficient (SE)

Constant −447.79 (696.54) −1309.53 (927.17) 31.13 (644.61) −832.65 (866.20)Social norms −1.81 (2.31) 6.67 (3.13)c −2.46 (2.17) 5.19 (2.92)dBenefiary: people 14.32 (4.55)a 13.49 (6.33)c 18.75 (5.93)b 25.00 (7.97)b

Benefiary: animal 12.12 (5.25)c 10.73 (12.25) 17.94 (8.35)c 17.52 (11.22)Benefiary: multiple 4.76 (6.17) 10.78 (9.62) 12.35 (6.71)d 16.25 (9.01)d

Domain: wood −0.55 (5.12) 2.33 (7.11) 5.15 (4.98) −0.93 (6.69)Domain: electronics −12.04 (6.22)d −0.45 (8.13) −1.38 (5.63) −0.42 (7.57)Domain: food 2.85 (4.79) 19.18 (7.37)c 5.86 (5.55) 11.09 (7.45)Domain: clothing 12.24 (5.51)c 5.31 (7.43) 28.46 (6.16)a 2.89 (8.28)Imputed year 0.24 (0.35) 0.68 (0.46) −0.00 (0.32) 0.44 (0.43)Certified 7.02 (3.24)c 7.70 (5.11) 6.64 (3.68)d 9.88 (4.95)c

Peer-reviewed −21.14 (6.11)b −18.04 (6.99)b −11.26 (4.66)c −22.60 (6.25)aAllows negatives −10.22 (2.97)a −17.54 (4.86)a −13.62 (3.62)a −26.11 (4.87)aIncentive compatible 17.22 (3.60)a −3.46 (9.58) 19.40 (9.43)c 34.71 (12.67)b

Adj R2 0.268 0.337 0.473 0.448

a p ≤ .001.b p ≤ .01.c p ≤ .05.

dnaeofthTpano

D

aosfb

C

ciatA

oib

E

nctrsl

nmtfdusscpt2i

d p ≤ .10.

emonstrate mixed findings. Against H3, there appears to beo effect of social norms on the percent premium participantsre willing to pay. However, in support of H3, there is somevidence that social norms positively influence the proportionf consumers who are willing to pay a premium. The coefficientor social norms is significant and positive for the model usinghe full data set of 106 observations. Using the smaller data set,owever, this finding drops to marginal significance (p = .08).hese results suggest that retailers may be able to increase theotential market size for socially responsible products if they areble to encourage greater social norms or perceptions of socialorms for socially responsible products through advertisementsr visibility of purchases.

omain

The results of the model suggest that clothing may elicit greater percent premium compared to “other” products (themitted base). None of the other categories were consistentlyignificantly different from the base category across both theull model and the model using only observations that reportoth dependent measures.

ertification

Certification has a positive effect on the percent premiumonsumers are willing to pay; the implication from the results

n Table 4 is that it increases the premium by about 7 percent-ge points. There is also some evidence that it may increasehe proportion of consumers willing to pay a positive amount.lthough certification was not significant in the model using all

tect

bservations, it did predict a larger proportion of consumers will-ng to pay a premium in the model among observations reportingoth dependent measures.

licitation method

As predicted, studies that allowed respondents to indicate aegative WTP for socially responsible products had a signifi-antly lower WTP for socially responsible products comparedo studies that did not allow negative values. This finding wasobust across all models and both dependent measures. Thus,tudies that only provide positive options for participants areikely biased upwards.

Papers using incentive compatible methods show a sig-ificantly higher WTP for the percent premium dependenteasure. This finding was robust in both models and con-

rary to expectations. Part of this finding may be due to theact that incentive-compatible studies include real purchaseata. In the real world, prices for socially responsible prod-cts are often much greater than products that do not haveocially responsible characteristics. Additionally, it has beenuggested that auctions can increase consumers’ WTP due toompetitiveness among participants or participants becomingsychologically committed once they have bid, even when auc-ions are incentive compatible. (Ku, Galinsky, and Murnighan006; Ku, Malhotra, and Murnighan 2005). We see no signif-cant differences for studies using incentive-compatibility in

he proportion of participants dependent measure in the mod-ls using all possible observations, but a positive and significantoefficient for incentive-compatibility on the proportion of par-icipants dependent measure in the regression using the smaller

l of R

dts

P

ppatttdpttHmp

A

geierapittnpcAtm

1moprr

nsoatw

bwt

rttmoetgprufe

mrittonsemntwdifoeasetp

aswastmcnor


ata set. However, there are a very small number of observa-ions that are incentive-compatible for this dependent measureo results should be interpreted with caution.

eer-reviewed

If there is publication bias, one would expect that publishedapers would have a higher percentage WTP and a greater pro-ortion of consumers willing to pay a premium than those thatre rejected or otherwise unpublished. However, results fromhe meta-analysis suggest otherwise. Papers that have not gonehrough a peer review process report higher WTP than thosehat have and report a significantly greater proportion of respon-ents willing to pay a positive amount for socially responsibleroducts. Thus, there does appear to be evidence of publica-ion bias, however it is in the unanticipated direction. Of note,hese results are in line with the findings of Lagerkvist andess (2011) who attribute this finding to the imposition ofore rigorous techniques of eliciting WTP in peer-reviewed

apers.

djusting for researcher method choices

Albers, Mantrala, and Sridhar (2010) and Tellis (1988) sug-est that mean effect sizes should be adjusted for the significantffects of methodological choices that researchers make that arendependent of the underlying theoretical explanations for theffect sizes. They call these “bias”-adjusted effect sizes. As aesult, we corrected the effect sizes in the following way. Wedjusted the unconditional percent difference in willingness toay by the parameters from Table 4 of incentive compatibil-ty and whether the study allowed for negative responses andhe proportion willing to pay a premium effect size only forhe latter. The bias-adjusted percentage difference in willing-ess to pay decreases from 16.8 percent to 14.8 percent and theroportion willing to pay a premium decreases from 59.7 per-ent to 51.6 percent (Fig. 1 shows the bias-adjusted histograms).lthough both are significantly different at the p < .01 level,

he bias adjusted numbers are still within the range of the raweans.

Conclusions

Our meta-analysis included studies from 80 papers with74 unique observations. It is, to our knowledge, the largesteta-analysis on WTP for socially responsible products and the

nly meta-analysis that looks at WTP across beneficiary androduct type. By including such a diverse set of papers, theesults of our meta-analysis provide important implications foretailers.

First, overall, people are willing to pay a positive and sig-ificant premium for socially responsible products. On average,ocially responsible products demand a 16.8 percent premium

ver products without socially responsible features. Further,lmost 60 percent of people indicate that they are willingo pay some type of premium for these products. Becausee excluded products that conferred any type of private

(w

f

etailing 90 (2, 2014) 255–274 265

enefit, the results suggest that the majority of consumers areilling to pay extra for benefits that do not directly benefit

hemselves.In terms of the relative value of socially responsible products,

esults from both dependent variables suggest that manufac-urers’ historical emphasis on the development of productshat demonstrate an environmental benefit may be subopti-

al for profitability as products with human beneficiariesbtain a greater price premium across both types of depend-nt measures. All else equal, it appears retailers may be ableo garner a higher price premium for products that promoteood labor practices and benefits such as fair trade com-ared to those promoting environmental benefits. Additionally,etailers may attract more consumers by advertising prod-cts that they carry which confer socially responsible benefitsor people opposed to products conferring benefits on thenvironment.

While the results of the social norm variable are somewhatixed, they are promising. The results suggests that to the extent

etailers can influence social norms via advertisements, visibil-ty of purchases, or other means, they may be able to increasehe potential market size for the socially responsible productshey carry. This finding is noteworthy in light of the limitationsf the social norm measure. The method used to elicit socialorms had many advantages including being a natural mea-ure of consumers’ beliefs of the strength of social norms forach specific product listed with the beneficiary (people, ani-als, environment or multiple). However, the measure of social

orms had multiple sources of noise which might have con-ributed to the relatively weak results. To elicit social norms,e used an online sample of U.S. respondents. This is likely aifferent population than the populations used in many of thendividual studies, possibly differing in important demographicactors. Additionally, the social norm variable was elicited atne specific time (in 2013). Social norms may vary over time,specially if social responsibility evolves in terms of awarenessnd popularity. Given these limitations, the significance of theocial norms variable on the proportion of participants WTP isncouraging. Future research would benefit to better understandhe role that social norms plays in WTP for socially responsibleroducts.

Our meta-analysis does have limitations. First, with the broadrea of socially responsible products, it is likely that not everytudy on the topic was included in our meta-analysis. However,e did have a relatively large sample size and feel that the papers

re broad and representative. Another limitation was the neces-ary assumption that all the methods of eliciting WTP acrosshe studies were measuring the same underlying construct. As

entioned in the WTP section above, there are multiple con-eptualizations of WTP. Many, if not most of the studies, didot provide enough information to ascertain which definitionf WTP they believed they were eliciting. Further, even ifesearchers believed they were getting one meaning of WTP

e.g., the indifference point), it is difficult to know if respondentsere interpreting the question the same way.Another limitation of this meta-analysis offers insights for

uture research. Our meta-analysis had a limited number of

2 l of R

oftToaLaicanfcstuwamn

ipnw

cpmHGfootlcimfiboie

oifWTP for socially responsible products.


bservations for some of the independent variables, particularlyor the proportion of participants dependent measure. This con-ributed to difficulty in interpreting discrepancies across models.he variables particularly affected by the small number ofbservations were the animal and multiple beneficiary typesnd the incentive-compatible dummy. As noted by Farley,ehmann, and Mann (1988), it is possible to use a meta-analysiss the basis for designing other studies to have maximummpact on knowledge generation in a field. The limitationsaused by the small number of observations for these vari-bles identify fruitful areas for future research. While we didot perform the analysis suggested by Farley et al., it is clearrom the data presented in Table 2 and from the discrepan-ies across models that there are understudied areas within theocial responsibility literature. This meta-analysis highlightshe need for research on WTP for socially responsible prod-cts that benefit animals, or a combination of beneficiaries asell as studies which compare WTP across different benefici-

ries. Additionally, more research using incentive-compatibleethods, especially when exploring potential market size, is

eeded.The results of this meta-analysis also suggest that research

s necessary to investigate the mechanisms underlying the inde-

endent variables affecting the variation in WTP. Research iseeded to understand the underlying reason that consumers areilling to pay more for products that benefit other humans

etailing 90 (2, 2014) 255–274

ompared to those that benefit the environment. A simple inter-retation may be that people view benefits to other people asore important and are therefore more willing to pay a premium.owever, alternative reasons are also plausible. For instance,leim et al. (2013) found that a major impediment of WTP

or environmentally friendly products was consumers’ belieff a lack of expertise. Perhaps consumers feel a greater sensef expertise in understanding labor conditions and fair wageshan they do about ways to help the environment. A simi-ar, but distinct explanation could be in regard to perceivedonsumer effectiveness (Balderjahn 1988; Roberts 1996). Its possible that consumers feel as though the difference they

ake when choosing socially responsible products that bene-t other people makes more of an impact than purchases thatenefit the environment. Research that can identify the causef this disparity in WTP could introduce useful interventions toncrease WTP for socially responsible products that benefit thenvironment.

This meta-analysis synthesizes the results of a large streamf literature conducted over the last few decades and providesmportant insights. We hope that this paper provides a roadmapor future research attempting to better understand consumers’

Appendix A. Summary of included papers and maincharacteristics

S.M.

Tully, R

.S. W

iner /

Journal of

Retailing

90 (2,

2014) 255–274

267

Authors Peerreviewed

Product Domain Socialnorm

Beneficiary Certified Year of datacollection

Incentivecompatible

AllowsnegativeWTP

Proportion ofparticipants DV(percent)

Unconditionalpercentpremium DV

Aguilar and Cai(2010)

Yes Bedside table Wood 3.30 Environment Yes 2006.3 No No 34.0 20.7Yes Bedside table Wood 3.30 Environment Yes 2006.3 No No 50.0 39.3

Aguilar and Vlosky(2007)

Yes Dining room set Wood 3.26 Environment Yes 1995 No No 61.3 16.8Yes Kitchen

remodelingWood 3.61 Environment Yes 1995 No No 56.9 10.9

Yes New home Wood 4.98 Environment Yes 1995 No No 65.5 23.0Yes Chair Wood 3.21 Environment Yes 1995 No No 63.3 13.2Yes Dining room set Wood 3.26 Environment Yes 2005 No No 68.0 12.7Yes Kitchen


Yes New home Wood 4.98 Environment Yes 2005 No No 71.4 27.1Yes Chair Wood 3.21 Environment Yes 2005 No No 63.0 15.3

Akkucuk (2011) Yes Auto part Electronics 3.33 Environment No 2007.3 No Yes 31.0 −24.3Yes Cellular phone Electronics 3.83 Environment No 2007.3 No Yes 23.0 −21.6Yes Furniture Other 3.60 Environment No 2007.3 No Yes 26.0 −13.9Yes Nonsanitary paper

productWood 4.61 Environment No 2007.3 No Yes 22.0 −20.4

Yes Printer fax Electronics 3.72 Environment No 2007.3 No Yes 31.0 −13.7Yes Sanitary paper

productWood 4.15 Environment No 2007.3 No Yes 23.0 −24.1

Yes Tire Other 4.17 Environment No 2007.3 No Yes 19.0 −11.9Anderson and Hansen (2003) No CD rack Wood 2.41 Environment Yes 1999.3 No Yes – 23.6Anderson and Hansen (2004) Yes Plywood Wood 3.57 Environment Yes 2002 Yes No 36.9 –

Basu and Hicks(2008)

Yes Coffee Food 4.57 People Yes 2004.3 No No – 47.0Yes Coffee Food 4.57 People Yes 2004.3 No No – 57.0

Bennett (1997) Yes Eggs Food 4.53 Animals No 1993.3 No No 86.0 30.7Bennett and Larson (1996) Yes Eggs Food 4.53 Animals No 1992.3 No Yes 68.0 18.0Berghoef and Dodds (2011) Yes Wine Food 3.02 Environment Yes 2009 No No 78.0 8.3

Bjørner, Hansen, andRussell (2004)

Yes Paper towels Wood 4.96 Environment Yes 2000.3 Yes Yes – 1.9Yes Toilet paper Wood 4.44 Environment Yes 2000.3 Yes Yes – 18.1

Brouhle and Khanna(2012)

Yes Paper towels Wood 4.96 Environment Yes 2008.3 Yes Yes – 39.0Yes Toilet paper Wood 4.44 Environment Yes 2008.3 Yes Yes – 35.2

Camacho-Cuena et al.(2004)

Yes Office table Wood 3.24 Environment No 2000.3 Yes Yes – 20.5Yes Office table Wood 3.24 Environment No 2000.3 Yes Yes – 21.3Yes Office table Wood 3.24 Environment No 2000.3 No Yes – 22.1Yes Office table Wood 3.24 Environment No 2000.3 No Yes – 23.3

Carlsson, Frykblom,and Lagerkvist(2007a)

Yes Chicken Food 4.99 Animals No 2003.3 No No −4.8Yes Minced beef Food 4.23 Animals No 2003.3 No No 66.0 9.0

Carlsson, Frykblom,and Lagerkvist(2007b)

Yes Eggs Food 4.53 Animals No 2003.3 No No – 20.0Yes Eggs Food 4.53 Animals No 2003.3 No No – 54.9

268

S.M.

Tully, R

.S. W

iner /

Journal of

Retailing

90 (2,

2014) 255–274




Incentivecompatible

AllowsnegativeWTP


Unconditionalpercent premiumDV

Carlsson, Garcia, andLöfgren (2010)

Yes Coffee Food 4.57 People Yes 2006.3 No Yes – 34.0Yes Coffee Food 4.25 Environment Yes 2006.3 No Yes – 34.9

Carter (2009) Yes Music CDs Other 3.23 People No 2005.3 No Yes 66.0 –Casadesus-Masanell et al. (2009) Yes Flannel shirt Clothing 3.39 Environment

and peopleYes 2005.3 Yes No – 10.6

Cha, Chun, andYeo-Chang (2009)

Yes Wood flooring Wood 3.87 Environment Yes 2008 No No 79.4 7.6Yes Copier paper Wood 4.87 Environment Yes 2008 No No 79.4 9.8Yes Wood dining table Wood 4.06 Environment Yes 2008 No No 79.4 6.8Yes Wood frame Wood 3.88 Environment Yes 2008 No No 79.4 11.6

Cranfield et al. (2010) Yes Coffee Food 4.57 People No 2006.3 No Yes – 7.4Yes Coffee Food 4.57 People Yes 2006.3 No Yes – 18.5

De Pelsmacker, Driesen, and Rayp (2005) Yes Coffee Food 4.57 People Yes 2001.3 No Yes – 10.2Dickson (2001) Yes Man’s dress shirt Clothing 3.39 People No 1997.3 No Yes – 36.0Didier and Lucie (2008) Yes Chocolate Food 3.69 People Yes 2004.3 Yes Yes – 87.1Drozdenko, Jensen, and Coelho (2011) Yes MP3 player Electronics 2.72 Environment No 2007.3 No No – 9.5Ellis, McCracken, and Skuza (2012) Yes Cotton T-shirt Clothing 3.97 Environment No 2008.3 Yes Yes 74.0 28.1Erwann (2009) Yes Fish Food 4.34 Environment No 2006 No No 80.8 10.9Forsyth, Haley, and Kozak (1999) Yes Wood products Wood 4.37 Environment No 1995.3 No No 67.3 5.4

Fussell (2011) No Products Other 4.82 People No 2009.5 No Yes 82.9 9.5No Products Other 4.82 People No 2009.5 No Yes 92.4 10.8No Products Other 4.82 People No 2009.5 No Yes 91.0 13.6No Products Other 4.82 People No 2009.5 No Yes 92.6 14.5

Galarraga and Markendya (2004) Yes Coffee Food 4.42 Environmentand people

Yes 2000.3 Yes Yes – 11.3

Grönroos and Bowyer(1999)

Yes Lumber Wood 3.70 Environment Yes 1997 No No 24.0Yes Lumber Wood 3.70 Environment Yes 1997 No No 36.0

Guagnano (2001) Yes Paper towels Wood 4.96 Environment No 1997.3 No No 86.0 32.1

Ha-Brookshire andNorum (2011)

Yes Cotton shirt Clothing 3.97 Environment No 2007.3 No No 57.0 17.3Yes Cotton shirt Clothing 3.97 Environment No 2007.3 No No 54.9 18.5Yes Cotton shirt Clothing 3.97 Environment No 2007.3 No No 55.1 18.6

Hertel, Scruggs, andHeidkamp (2009)

Yes Coffee Food 4.57 People No 2006 No No 75.0 –Yes Sweaters Clothing 3.81 People No 2006 No No 68.0 –

Howard and Allen (2008) Yes Strawberries Food 3.86 People No 2006 No No 87.4 68.0Hurley, Miller, and Kliebenstein (2006) Yes Pork loin chops Food 2.96 Environment No 2002.3 Yes Yes 62.0 22.4

Hustvedt, Peterson,and Chen (2008)

Yes Wool gloves Clothing 2.76 Environment Yes 2004.3 No Yes – 1.8Yes Wool gloves Clothing 3.41 Animals No 2004.3 No Yes – 2.8

Jensen et al. (2003) Yes Oak shelvingboard

Wood 3.57 Environment Yes 1999.3 No No 42.2 14.9

S.M.

Tully, R

.S. W

iner /

Journal of

Retailing

90 (2,

2014) 255–274

269




Incentivecompatible

AllowsnegativeWTP



Jensen et al. (2004) Yes Oak chair Wood 4.06 Environment Yes 2000.3 No No 31.0 8.0Yes Oak shelving

boardWood 3.57 Environment Yes 2000.3 No No 31.0 13.0

Yes Oak table Wood 4.06 Environment Yes 2000.3 No No 31.0 5.6

Kang et al. (2012) Yes Hotel rooms Other 3.33 Environment No 2008.3 No No 30.0 3.8Kruger (2010) No Printer paper Wood 4.87 Environment Yes 2006.3 No No – 37.0Kuminoff, Zhang, and Rudi (2010) Yes Hotel rooms Other 3.33 Environment Yes 2006.3 Yes Yes – 19.0Langen (2011) Yes Coffee Food 4.57 People Yes 2007.3 No Yes – 12.9Lee et al. (2007) Yes Wood products Wood 4.37 Environment Yes 2005 No Yes 62.0 6.8

Levinson (2010) No T-shirts Clothing 3.97 Environment No 2006.3 No No 64.0 46.0No T-shirts Clothing 3.97 Environment No 2006.3 No No 75.0 52.0No T-shirts Clothing 3.97 Environment No 2006.3 No No 79.0 54.0No T-shirts Clothing 3.97 Environment No 2006.3 No No 81.0 62.0

Liljenstolpe (2008) Yes Pork fillet Food 4.24 Animals No 2004.3 No Yes – 13.0Yes Pork fillet Food 4.24 Animals No 2004.3 No Yes – 15.0Yes Pork fillet Food 4.24 Animals No 2004.3 No Yes – 19.0Yes Pork fillet Food 4.24 Animals No 2004.3 No Yes – 20.0

Lin (2010) Yes Cotton apparel Clothing 4.38 Environment No 2007 No Yes 52.7 –Loureiro (2003) Yes Wine Food 3.02 Environment No 2001 No No 56.5 1.3

Loureiro and Lotade(2005)

Yes Coffee Food 4.57 People No 2002 No No 84.9 3.3Yes Coffee Food 4.42 Environment

and animalsNo 2003 No No 84.9 3.1

Loureiro, McCluskey, and Mittelhammer (2002) Yes Apples Food 3.94 Environment Yes 1998.3 No Yes – 5.0Mahé (2010) Yes Bananas Food 3.76 People Yes 2006 No No 86.0 –Manaktola and Jauhari (2007) Yes Hotel rooms Other 3.33 Environment No 2003.3 No No 15.0 –McVittie, Moran, and Nevison (2006) No Chicken Food 4.99 Animals No 2002.3 No No – 48.5

Michaud and Llerena(2011)

Yes Recycled singleuse cameras

Electronics 3.53 Environment No 2007.3 Yes Yes – 1.4

Yes Remanufacturedsingle use cameras

Electronics 3.20 Environment No 2007.3 Yes Yes – −16.4

Michaud, Llerena, andJoly (2012)

Yes Roses (flowers) Other 3.70 Environment Yes 2008.3 Yes Yes – 64.0Yes Roses (flowers) Other 3.70 Environment Yes 2008.3 Yes Yes – 84.0

Millar and Baloglu(2011)

Yes Hotel rooms Other 3.33 Environment Yes 2007.3 No Yes 9.8 0.2Yes Hotel rooms Other 3.33 Environment Yes 2007.3 No Yes 18.0 1.1

Mohamed and Ibrahim (2007) Yes Wood products Wood 4.37 Environment Yes 2003.3 No No 38.0 5.5Moon et al. (2002) Yes Agriculture Food 5.13 Environment No 1995 No No 83.0 18.0

270

S.M.

Tully, R

.S. W

iner /

Journal of

Retailing

90 (2,

2014) 255–274




Incentivecompatible

AllowsnegativeWTP



Moosmayer (2012) Yes Athletic shoes Clothing 3.91 Environmentand people

Yes 2008.3 No Yes 69.0 19.5

Yes Athletic shoes Clothing 3.91 Environmentand people

Yes 2008.3 No Yes 77.0 41.0

Yes Mobile phone Electronics 3.91 Environmentand people

Yes 2008.3 No Yes 47.0 11.3

Yes Mobile phone Electronics 3.91 Environmentand people

Yes 2008.3 No Yes 35.0 23.7

Muñoz and Rivera (2002) No Hotel rooms Other 3.33 Environment No 2002 No Yes 40.0 –

Norwood and Lusk(2011)

Yes Pork Food 4.24 Animals No 2007.3 Yes Yes – 9.5Yes Pork Food 4.24 Animals No 2007.3 Yes Yes – 12.2Yes Pork Food 4.24 Animals No 2007.3 Yes Yes – 25.2Yes Pork Food 4.24 Animals No 2007.3 Yes Yes – 27.9Yes Pork Food 4.24 Animals No 2007.3 Yes Yes – 30.6Yes Pork Food 4.24 Animals No 2007.3 Yes Yes – 55.8

Oleson et al. (2010) Yes Salmon Food 3.93 Animals Yes 2006.3 Yes Yes – 16.9

Ozanne, Bigsby, andVlosky (1999)

Yes Clear wood forshelving

Wood 3.57 Environment Yes 1997 No No 78.3 22.3

Yes Outdoor furnitureset


Yes Chair Wood 3.21 Environment Yes 1997 No No 73.9 17.6Yes Kitchen


Yes Wood in newhome


Ozanne and Vlosky(1997)



Yes New home Wood 4.98 Environment Yes 1995 No No 64.0 4.4Yes Chair Wood 3.21 Environment Yes 1995 No No 62.0 14.4Yes Studgrade stud Wood 2.89 Environment Yes 1995 No No 71.0 18.7

Ozanne and Vlosky(2003)



Yes New home Wood 4.98 Environment Yes 2000 No No 67.0 4.4Yes Chair Wood 3.21 Environment Yes 2000 No No 63.0 11.7Yes Studgrade stud Wood 2.89 Environment Yes 2000 No No 73.0 17.3

Pajari, Peck, andRametsteiner (1999)

No Wood furniture Wood 4.06 Environment Yes 1995.3 No Yes 50.9 1.4No Wood furniture Wood 4.06 Environment Yes 1995.3 No Yes 42.0 1.6No Wood furniture Wood 4.06 Environment Yes 1995.3 No Yes 58.1 2.4No Wood furniture Wood 4.06 Environment Yes 1995.3 No Yes 66.4 3.4No Wood furniture Wood 4.06 Environment Yes 1995.3 No Yes 62.3 4.9

S.M.

Tully, R

.S. W

iner /

Journal of

Retailing

90 (2,

2014) 255–274

271




Incentivecompatible

AllowsnegativeWTP



Prasad et al. (2004) Yes Athletic socks Clothing 3.29 People Yes 2002 Yes No 30.0 –

Rode, Hogarth, andLe Menestrel (2008)

Yes Units Other 3.50 People Yes 2004.3 Yes Yes – 24.4Yes Units Other 3.50 People Yes 2004.3 Yes Yes – 26.4Yes Units Other 3.50 People Yes 2004.3 Yes Yes – 64.8

Roheim, Asche, and Santos (2011) Yes Alaskan pollack Food 4.34 Environment Yes 2007.3 Yes Yes – 14.2

Rousu and Corrigan(2008)

Yes Bananas Food 3.76 People No 2004.3 Yes Yes – 9.3Yes Chocolate Food 3.69 People No 2004.3 Yes Yes – 19.2

Saphores et al. (2007) Yes Cell phones Electronics 3.83 Environment Yes 2004 No No 69.3 2.5Yes Personal

computersElectronics 3.76 Environment Yes 2004 No No 70.8 2.3

Schollenberg (2012) Yes Coffee Food 4.57 People Yes 2008.3 Yes Yes – 38.0

Shen (2012) Yes Battery Electronics 4.48 Environment Yes 2007 No No 80.9 9.5Yes Building material Other 4.85 Environment Yes 2007 No No 83.4 9.5Yes Furniture Other 3.60 Environment Yes 2007 No No 86.3 9.5Yes Glass tableware Other 2.78 Environment Yes 2007 No No 77.8 8.7Yes Recycled paper Wood 5.65 Environment Yes 2007 No No 79.5 8.8Yes Soft drink Food 2.99 Environment Yes 2007 No No 76.6 8.8

Thompson et al.(2010)

Yes Plywood Wood 3.57 Environment Yes 2006.3 No Yes – 16.9Yes Wood dining table Wood 4.60 Environment Yes 2006.3 No Yes – 0.4

Tonsor, Olynk, andWolf (2009)

Yes Pork Food 4.24 Animals No 2005.3 No Yes – 11.2Yes Pork Food 4.24 Animals No 2005.3 No Yes – 42.3

Trudel and Cotte(2009)

Yes Coffee Food 4.42 Environmentand people

Yes 2005.3 No Yes – 16.8

Yes Coffee Food 4.42 Environmentand people

Yes 2005.3 No Yes – 28.9

Yes Cotton T-shirt Clothing 3.97 Environment No 2005.3 No Yes – 2.0Yes Cotton T-shirt Clothing 3.97 Environment No 2005.3 No Yes – 3.4Yes Cotton T-shirt Clothing 3.97 Environment No 2005.3 No Yes – 5.8Yes Cotton T-shirt Clothing 3.97 Environment No 2005.3 No Yes – 13.5

Van Kempen et al. (2009) Yes Firewood Wood 3.86 Environment Yes 2005.3 No No 73.0 21.0

Veisten (2002) Yes Wood dining table Wood 4.06 Environment Yes 1998.3 No No 32.0 1.0Yes Wood dining table Wood 4.06 Environment Yes 1998.3 No No 39.0 1.6

Veisten (2007) Yes Wood dining table Wood 4.06 Environment Yes 1997 No No 31.2 2.0Yes Wood dining table Wood 4.06 Environment Yes 1997 No No 39.8 16.0Yes Wood dining table Wood 4.06 Environment Yes 1998 No No 38.1 6.0Yes Wood dining table Wood 4.06 Environment Yes 1998 No No 37.1 7.5

Xu et al. (2012) Yes Seafood Food 4.34 Environment No 2009 No No 67.0 4.9Yang et al. (2012) Yes Coffee Food 4.57 People Yes 2008 No No 89.0 22.7

Yip, Knowler, andHaider (2012)

No Salmon Food 4.34 Environment Yes 2008.3 No Yes – 3.9No Salmon Food 4.34 Environment Yes 2008.3 No Yes – 9.8

Yue, Hurley, andAnderson (2011)

Yes Plants Other 4.24 Environment No 2007.3 Yes Yes – 13.9Yes Plants Other 4.24 Environment No 2007.3 Yes Yes – 34.0

2 l of R

A

A

A

A

A

A

A

B

B

B

B

B

B

B

B

B

B

C

C

C

C

C

C

C

C

C

D

D

D

D

D

D

E

E

F

F

F

F

F

G


References

guilar, Francisco X. and Zhen Cai (2010), “Conjoint Effect of EnvironmentalLabeling, Disclosure of Forest of Origin and Price on Consumer Prefer-ences for Wood Products in the US and UK,” Ecological Economics, 70(December), 308–16.

guilar, Francisco X. and Richard P. Vlosky (2007), “Consumer Willingnessto Pay Price Premiums for Environmentally Certified Wood Products in theU.S.,” Forest Policy and Economics, 9 (May), 1100–12.

kkucuk, Ulas (2011), “Combining Purchase Probabilities and Willingness toPay Measures: A Case on Recycled Products,” European Journal of SocialSciences, 23 (3), 353–61.

lbers, Sönke, Murali K. Mantrala and Shrihari Sridhar (2010), “PersonalSelling Elasticities: A Meta-analysis,” Journal of Marketing Research, 47(October), 840–53.

nderfer, Veronika A. and Ulf Liebe (2012), “Research on Fair TradeConsumption—A Review,” Journal of Business Ethics, 106, 415–35.

nderson, Roy C. and Eric N. Hansen (2003), “Segmenting Consumers ofEcolabeled Forest Products,” in An Analysis of Consumer Response toEnvironmentally Certified, Ecolabeled Forest Products, dissertation pre-sented May 9, 2003, at Oregon State University, Corvallis, OR, (accessedJanuary 25, 2013), [http://ir.library.oregonstate.edu/xmlui/bitstream/handle/1957/7526/Anderson Roy C.pdf?sequence=1].

and Eric N. Hansen (2004), “Determining ConsumerPreferences for Ecolabeled Forest Products: An Experimental Approach,”Journal of Forestry, 102 (June), 28–32.

ssmus, Gert, John U. Farley and Donald R. Lehmann (1984), “How AdvertisingAffects Sales: Meta-analysis of Econometric Results,” Journal of MarketingResearch, 21 (February), 65–74.

alderjahn, Ingo (1988), “Personality Variables and Environmental Attitudes asPredictors of Ecologically Responsible Consumption Patterns,” Journal ofBusiness Research, 17 (1), 51–6.

asu, Arnab K. and Robert L. Hicks (2008), “Label Performance and theWillingness to Pay for Fair Trade Coffee: A Cross-national Perspective,”International Journal of Consumer Studies, 32 (September), 470–8.

ennett, Richard (1997), “Farm Animal Welfare and Food Policy,” Food Policy,22 (August), 281–8.

ennett, Richard and Douglas Larson (1996), “Contingent Valuation of thePerceived Benefits of Farm Animal Welfare Legislation: An ExploratorySurvey,” Journal of Agricultural Economics, 47 (2), 224–35.

erghoef, Naomi and Rachel Dodds (2011), “Potential for Sustainability Eco-labeling in Ontario’s Wine Industry,” International Journal of Wine BusinessResearch, 23 (4), 298–317.

ijmolt, Tammo H.A. and Rik G.M. Pieters (2001), “Meta-analysis in Market-ing When Studies Contain Multiple Measurements,” Marketing Letters, 12,157–69.

ijmolt, Tammo H.A., Harald van Heerde and Rik G.M. Pieters (2005), “NewEmpirical Generalizations on the Determinants of Price Elasticity,” Journalof Marketing Research, (May), 141–56.

jørner, Thomas Bue, Lars Gårn Hansen and Clifford S. Russell (2004), “Envi-ronmental Labeling and Consumers’ Choice—An Empirical Analysis ofthe Effect of the Nordic Swan,” Journal of Environmental Economics andManagement, 47 (May), 411–34.

rouhle, Keith and Madhu Khanna (2012), “Determinants of ParticipationVersus Consumption in the Nordic Swan Eco-labeled Market,” EcologicalEconomics, 73 (January), 142–51.

usinessWeek (2012), “Levi’s Has a New Color for Blue Jeans: Green,” (October22–28), 26–27.

ai, Zhen and Francisco X. Aguilar (2013), “Meta-analysis of Consumer’sWillingness-to-Pay Premiums for Certified Wood Products,” Journal of For-est Economics, 19 (January), 15–31.

amacho-Cuena, Eva, Aurora García-Gallego, Nikolaos Georgantzís and Ger-ardo Sabater-Grande (2004), “An Experimental Validation of HypotheticalWTP for a Recyclable Product,” Environmental and Resource Economics,27 (March), 313–35.

arlsson, Fredrik, Peter Frykblom and Carl Johan Lagerkvist (2007a), “FarmAnimal Welfare Testing for Market Failure,” Journal of Agriculture andApplied Economics, 39 (April), 61–73.

G

etailing 90 (2, 2014) 255–274

, and(2007b), “Consumer Willingness to Pay for Farm Animal Welfare: MobileAbattoirs Versus Transportation to Slaughter,” European Review of Agricul-tural Economics, 34 (September), 321–44.

arlsson, Fredrik, Jorge H. Garcia and ´̊Asa Löfgren (2010), “Conformity andthe Demand for Environmental Goods,” Environmental and Resource Eco-nomics, 47 (November), 407–21.

arter, Robert E. (2009), “Will Consumers Pay a Premium for Ethical Informa-tion?,” Social Responsibility Journal, 5 (4), 464–77.

asadesus-Masanell, Ramon, Michael Crooke, Forest Reinhardt and VishalVasishth (2009), “Households’ Willingness to Pay for ‘Green’ Goods: Evi-dence from Patagonia’s Introduction of Organic Cotton Sportswear,” Journalof Economics & Management Strategy, 18 (March), 203–33.

ha, Junhee, Jung-Nam Chun and Youn Yeo-Chang (2009), “Consumer Will-ingness to Pay Price Premium for Certified Wood Products in South Korea,”Journal of Korean Forest Society, 98 (2), 203–11.

lifford, Stephanie and Andrew Martin (April 21, 2011), “As Consumers CutSpending, ‘Green’ Products Lose Allure,” The New York Times, (retrievedDecember 15, 2012), [www.nytimes.com/2011/04/22/business/energy-environment/22green.html? r=1%3Famp;pagewanted=all].

ranfield, John, Spencer Henson, James Northey and Oliver Masakure (2010),“An Assessment of Consumer Preference for Fair Trade Coffee in Torontoand Vancouver,” Agribusiness, 26 (2), 307–25.

e Pelsmacker, Patrick, Liesbeth Driesen and Glenn Rayp (2005), “Do Con-sumers Care About Ethics? Willingness to Pay for Fair-Trade Coffee,”Journal of Consumer Affairs, 39 (2), 363–86.

ickson, Marsha A. (2001), “Utility of No Sweat Labels for Apparel Consumers:Profiling Label Users and Predicting Their Purchases,” Journal of ConsumerAffairs, 35 (1), 96–119.

idier, Tagbata and Serieix Lucie (2008), “Measuring Consumer’s Willing-ness to Pay for Organic and Fair Trade Products,” International Journal ofConsumer Studies, 32 (September), 479–90.

ing, Min (2007), “An Incentive-Aligned Mechanism for Conjoint Analysis,”Journal of Marketing Research, 44 (May), 214–23.

ost, Florian and Robert Wilken (2012), “Measuring Willingness to Pay asa Range: When Should We Care?,” International Journal of Research inMarketing, 29, 148–66.

rozdenko, Ronald, Marlene Jensen and Donna Coelho (2011), “Pricing ofGreen Products: Premiums Paid, Consumer Characteristics and Incentives,”International Journal of Business, Marketing, and Decision Sciences, 4 (1),106–16.

llis, Joan L., Vivki A. McCracken and Nathan Skuza (2012), “Insights into Will-ingness to Pay for Organic Cotton Apparel,” Journal of Fashion Marketingand Management, 16 (3), 290–305.

rwann, Charles (2009), “Eco-labelling: A New Deal for a More DurableFishery Management?,” Ocean & Coastal Management, 52 (May), 250–7.

air Trade USA (2012), (retrieved January 23, 2012), [http://fairtradeusa.org/certification].

arley, John U. and Donald R. Lehmann (1986), Meta-analysis in Marketing,Lexington, MA: Lexington Books.

arley, John U., Donald R. Lehmann and Lane H. Mann (1988), “Designingthe Next Study for Maximum Impact,” Journal of Marketing Research, 35(November), 496–501.

orsyth, Keith, David Haley and Robert Kozak (1999), “Will Consumers PayMore for Certified Wood Products?,” Journal of Forestry, 97 (February),18–22.

ussell, Rachel (2011), “Consumer’s Willingness to Pay a Price Pre-mium for Living Wage Products,” paper presented at UndergraduateResearch Symposium, May 2011, at Duke University, Durham, NC,(accessed January 25, 2013), [https://econ.duke.edu/uploads/media items/fussell-rachel-2011-duke-economic-

nyu stern school of business - the role of the beneficiary...

Documents