what will consumers pay for social product features?1022212816261.pdf · the importance of ethical...

24
ABSTRACT. The importance of ethical consumerism to many companies worldwide has increased dramat- ically in recent years. Ethical consumerism encom- passes the importance of non-traditional and social components of a company’s products and business process to strategic success – such as environmental protectionism, child labor practices and so on. The present paper utilizes a random utility theoretic exper- imental design to provide estimates of the relative value selected consumers place on the social features of products. KEY WORDS: animal rights, choice modeling, ethical consumerism, willingness-to-pay, workers’ rights Recent studies on ethical consumerism suggest that consumers increasingly care about the ethical components of products and business processes and that these concerns have financial implications for the businesses involved (CAFOD, 1998; Elliot and Freeman, 2001; Marymount University, 1999). However, the conclusions from most of these studies were derived from surveys in which respondents were asked to simply rank the importance of a list of ethical issues. Such questions do not require consumers to trade-off ethical features of products against traditional features nor do they seek to determine the degree to which consumers would sacrifice to make these tradeoffs. Hence, responses to these types of survey questions tend to overstate the impor- tance of ethical features because there are obvious responses that are more socially acceptable and there is little ability to estimate the magnitude of the importance of the stated opinion (what statisticians call “effect size”). Despite such methodological problems, recent developments suggest that there is a trend towards more consumer activism with respect to the “social behaviors” of organizations, especially large and well-known multinational corporations. For example, the number of protests directed at international organizations like the WTO and at global companies like Nike has increased dramatically in recent years. Demonstrators have often become the main focus of news reports during large-scale meetings such as the WTO, G7 and the World Bank; and the number of groups focusing their attention on the social behaviors of companies also seems to have increased in recent years. For example, Elliott and Freeman (2001) identified over forty anti- sweatshop organizations in the U.S. alone, a large number considering the focus of those organizations on a single issue. Even larger numbers of groups and members can be found in areas such as environmental protectionism, human rights, and animal rights with large and well-established organizations like Greenpeace, Amnesty International and the World Wildlife Fund. The purpose of this paper is to try to clarify the extent to which consumers “value” ethical product features when making purchases by utilizing a distinctive methodology – structured choice experiments (Louviere et al., 2000) – that What Will Consumers Pay for Social Product Features? Journal of Business Ethics 42: 281–304, 2003. © 2003 Kluwer Academic Publishers. Printed in the Netherlands. Pat Auger Paul Burke Timothy M. Devinney Jordan J. Louviere Pat Auger is Senior Fellow at the Melbourne Business School. Paul Burke is a PhD candidate. Timothy Devinney is Professor and Director of the Centre for Corporate Change at the Australian Graduate School of Management. Jordan Louviere is Professor of Marketing at the University of Technology, Sydney.

Upload: truongduong

Post on 21-Apr-2018

216 views

Category:

Documents


1 download

TRANSCRIPT

ABSTRACT. The importance of ethical consumerismto many companies worldwide has increased dramat-ically in recent years. Ethical consumerism encom-passes the importance of non-traditional and socialcomponents of a company’s products and businessprocess to strategic success – such as environmentalprotectionism, child labor practices and so on. Thepresent paper utilizes a random utility theoretic exper-imental design to provide estimates of the relativevalue selected consumers place on the social featuresof products.

KEY WORDS: animal rights, choice modeling,ethical consumerism, willingness-to-pay, workers’rights

Recent studies on ethical consumerism suggestthat consumers increasingly care about theethical components of products and businessprocesses and that these concerns have financialimplications for the businesses involved (CAFOD,1998; Elliot and Freeman, 2001; MarymountUniversity, 1999). However, the conclusions frommost of these studies were derived from surveysin which respondents were asked to simply rankthe importance of a list of ethical issues. Suchquestions do not require consumers to trade-offethical features of products against traditionalfeatures nor do they seek to determine the degree

to which consumers would sacrifice to makethese tradeoffs. Hence, responses to these typesof survey questions tend to overstate the impor-tance of ethical features because there are obviousresponses that are more socially acceptable andthere is little ability to estimate the magnitudeof the importance of the stated opinion (whatstatisticians call “effect size”).

Despite such methodological problems, recentdevelopments suggest that there is a trend towardsmore consumer activism with respect to the“social behaviors” of organizations, especiallylarge and well-known multinational corporations.For example, the number of protests directed atinternational organizations like the WTO andat global companies like Nike has increaseddramatically in recent years. Demonstrators haveoften become the main focus of news reportsduring large-scale meetings such as the WTO,G7 and the World Bank; and the number ofgroups focusing their attention on the socialbehaviors of companies also seems to haveincreased in recent years. For example, Elliottand Freeman (2001) identified over forty anti-sweatshop organizations in the U.S. alone, alarge number considering the focus of thoseorganizations on a single issue. Even largernumbers of groups and members can be foundin areas such as environmental protectionism,human rights, and animal rights with large andwell-established organizations like Greenpeace,Amnesty International and the World WildlifeFund.

The purpose of this paper is to try to clarifythe extent to which consumers “value” ethicalproduct features when making purchases byutilizing a distinctive methodology – structuredchoice experiments (Louviere et al., 2000) – that

What Will Consumers Pay for Social Product Features?

Journal of Business Ethics

42: 281–304, 2003.© 2003 Kluwer Academic Publishers. Printed in the Netherlands.

Pat AugerPaul Burke

Timothy M. DevinneyJordan J. Louviere

Pat Auger is Senior Fellow at the Melbourne BusinessSchool.

Paul Burke is a PhD candidate.Timothy Devinney is Professor and Director of the Centre

for Corporate Change at the Australian Graduate Schoolof Management.

Jordan Louviere is Professor of Marketing at the Universityof Technology, Sydney.

allows us to estimate the dollar value of specificproduct configurations without the standard“politically correct” bias that exists with tradi-tional survey methods. In addition, we areconcerned not just with whether there are indi-viduals in the community who value socialproduct features but whether or not we can accu-rately gauge who these people are and whetherinformation presentation might cause their pref-erences to vary.

Our purpose is not to definitively determinewhether ethical consumerism is or is not a vitalforce in the economy but to address the moresubtle issue associated with whether or not con-sumers are little more than “arm chair” ethicists.This has value both to policy makers andbusiness. In the case of the latter, it is importantfor political decision makers to judge the extentto which vocal advocacy for social causes isrepresentative of a more deep seeded and latentconcern upon which people are willing to act,if given the opportunity. In the case of theformer, business needs to have accurate infor-mation about what its consumers truly want asrepresented by what they will pay for and needto know whether what consumers state currentlyis representative of what they truly believe.

The importance of ethical consumerism

The rising importance of ethical managementdimensions is easy to observe in the popularpress. The travails of corporate giants such asNike and Adidas can be juxtaposed against veryvisible “corporate responsibility” stances of com-panies like The Body Shop, Levi’s and Reebok.Nike’s labor practice problems in Indonesia andVietnam so vividly highlighted by the CBSprogram 48 Hours a few years ago (17 October1996) revealed either seemingly poor manage-rial control or very poor public relations. Indeed,recent revelations about the intimidation of Nikefactory workers in Indonesia suggest that theseproblems persist in some locations (Luh, 2001).Similarly, Kathie Lee Gifford’s “squeaky clean”image suffered after repeated stories showing herclothing (retailed at K-Mart) being made in sweatshops (Miller, 1997).

A more cynical view on such issues wouldpoint out that much of what appears on thesurface as moral corporate behavior is reallysubtle marketing and/or attempts by organizedlabor groups in developed countries to ensurethat jobs do not shift to lower wage countriesoverseas (Elliott and Freeman, 2001). Forexample, it is difficult to distinguish a morally“responsible” stance, such as the labels onReebok’s soccer balls (“Guarantee: Manufacturedwithout child labor”, Miller, 1997), from asophisticated strategy aimed at differentiation,such as Reebok’s visible support of AmnestyInternational. The latter presumably plays againstNike (and its labor problems) and Adidas (whichhas been accused of using prison labor in China(Smith and Copetas, 1998)). In fact, althoughReebok’s Director of Human Rights, DougCahn, stated that “there’s a correlation betweenfactories producing good quality products andthose with good working conditions” (Click,1996), Reebok does not seem to have a cost orquality advantage over its less “responsible” com-petitors. Indeed, until recently Nike consistentlyoutperformed Reebok on both quality and finan-cial performance measures. Lerner and Fryxell(1988) show that the only consistent factorsrelated to measures of “corporate social perfor-mance” are the industry in which one operates,one’s size and one’s advertising intensity (thelatter two must be large).

Gordon Fairclough (1996) noted there is littlethat is “black and white” about the child labordilemma. Those who tout sanctions and boycottsto deal with the problem do little more than“assuage the consciences of Western consumers”because nothing replaces these children’s workexcept destitution. Thus, a guarantee that asoccer ball is not made by child labor does notnecessarily mean better welfare for children whocould have made the soccer ball if they areexcluded from the opportunity to make even alimited wage. Neil Hawkins of CARE Cambodiasummed up some of these difficult issues bysaying “I don’t believe that the people who buyGap and Nike want these girls to work inbrothels because they lost their jobs at thegarment factory” (AsiaWeek, 2001). Similarly,consumer boycotts and associated tactics create

282 Pat Auger et al.

their own moral dilemma in a society where freechoice is viewed as the most paramount ofhuman rights (Garrett, 1986). Furthermore,several researchers have shown that trade sanc-tions or import tariffs against countries that usechild labor do not always reduce the use of childlabor in those countries (e.g., Ranjan, 2001). Infact, Edmonds and Pavcnik (2002) recentlyshowed that globalization (or an elimination ofsanctions and/or tariffs) has reduced the use ofchild labor in Vietnam, especially for olderfemale children. An equally important findingwas that the percentage of children attendingschool significantly increased over the sameperiod, which suggests that children who had leftthe workforce were now attending schools.

The emotional level of such discussions clearlycan make rational discourse difficult, but in lightof the apparent importance of the issue, it is sur-prising how little substantive academic researchis available, and what is available about theimportance of ethical consumerism is both mixedand thin. At the most basic of levels two ques-tions need to be considered and are the subjectof this paper:

(1) When forced to make substantive trade-offs, do we find that consumers valueethical product features to any greatextent? And, if this is true,

(2) Can we make any statements regardingdifferences between observable groups ofconsumers in the extent to which theyhold such a disposition?

There is considerable evidence to suggest thatethical consumer segments indeed exist as dis-tinctive groups. In general, the ethics literaturehas shown that consumers indicate that theyvalue moral stances. For example, Fullerton et al.(1996) showed that consumers generally are quiteintolerant with regard to ethical abuses by bothretailers (e.g., cheating customers) and consumers(e.g., abusing the goodwill of retailers). Whalenet al. (1991) and Pitts et al. (1991) studied therelationship between situational variables andsubjects’ ethical predispositions and purchaseintentions and their evaluations of a store’smanagers. They base their work on the simplefinding that consumers will “avoid transactions

with dishonest, unethical sellers” (Abratt andSacks, 1988), and show that consumers arrive atthis decision differently depending on the situa-tion. In circumstances where consumers faced apersonal ethical transgression, their cognitiveschema were simpler and more likely to berelated to their ethical predisposition (althoughweakly). In vicarious situations in which ethicaltransgressions are more general and less direct,more complex cognitive schema exist withless direct linkage to consumers’ ethical predis-position.

More typical are attempts to understand ethicalconsumers using multiple item scales. Forexample, Roberts (1996) developed an 18-itemscale measuring “responsible consumer behavior”that included measures like “I have purchasedproducts because they cause less pollution” and“I do not buy products from companies that dis-criminate against minorities”. Three importantresults emerge from his work: (a) non-ethicalconsumers exist (39 percent of respondents indi-cated no ethical concerns); (b) very little of thevariance on his scale is related to demographicvariables; and (c) the relationship betweenexpressed environmental and social concern andactive consumer action is weak. This latter resultis significant for two reasons. First, it relatesdirectly to our primary research question, namelyare people willing to act, at a cost, based uponan ethical consumer disposition? Second, it is notentirely consistent with previous research in thefield of moral judgment, and more importantly,the link between moral judgment and observedbehavior (e.g., Rest, 1986; Rest et al., 1997;Thoma et al., 1986). More specifically, Rest(1983, 1984) proposed that moral judgment isone of four component internal processes thatdetermine moral behavior and that moralbehavior is codetermined by these four internalprocesses. Two implications result from thistheory: (1) there should be a significant linkbetween moral judgment and moral behavior,and (2) the link between moral judgment andmoral behavior may be weak since the otherthree components also codetermine moralbehavior. Empirical evidence to date seems tosupport this theory. For example, Thoma et al.(1986) found that 32 out of 47 analyses (from a

What Will Consumers Pay for Social Product Features? 283

number of different studies) showed a significantlink between moral judgment and behavior.More recently, Rest et al. (1999) cite over 60published articles that uncovered a relationshipbetween moral judgment and behavior.

Studies that examined the importance ofethical consumerism, our main question, fall intotwo categories: those identifying and measuringthe strength of consumer preferences and theanalysis of factors that influence adoption ofsocially responsible products. Virtually all suchresearch to date has been related to environ-mentalism or “green” consumers. For example,Sriram and Forman (1993) examined the strengthof consumer preferences using conjoint analysis.They studied milk, washing machines anddeodorant using Dutch and American respon-dents, and varied socially responsible featureslike recyclable packaging (milk), energy effi-ciency (washing machines) and animal testing(deodorant). They found that energy efficiencywas not important to either group, but Americanconsumers were more concerned about milk inrecyclable packages than Dutch consumers, whowanted cheap, low fat milk in non-plastic con-tainers. Both groups showed concern towardstesting of deodorants on animals. Schrum et al.(1995) investigated the associated advertisingstrategy to green consumers, who were morelikely to be opinion leaders, knowledgeableinformation seekers and careful shoppers. Theyfound that only consumers who were activeinformation seekers would switch from theircurrent brand to a less effective but environ-mentally safer brand.

More recent studies have attempted to dealwith the willingness of consumers to pay morefor products with “acceptable ethical features”,especially with respect to labor standards.For example, several studies conducted atMarymount University (1999) found that 75percent of consumers would avoid shopping ina store if they knew the goods were producedunder bad conditions. More importantly, thesesame consumers indicated that they would pay$1 more for a $20 item that was made undergood conditions. Similar results were obtainedfrom University of Maryland studies (2000) inwhich roughly 75 percent of consumers said they

would pay $5 more on a $20 item if they knewit was not fabricated in a sweatshop.

A more rigorous study by Elliott and Freeman(2001) produced some additional insights into thebehavior of consumers. First, the authors foundthat consumers were willing to pay more forproducts made under good conditions, but thatthe price of the products affected how muchmore. For example, their results indicate thatconsumers were willing to pay 28 percent morefor $10 items, but 15 percent more for $100items. Further investigations led to some inter-esting generalizations. The authors found rela-tively high elasticities of demand for productsmade under good conditions but low elasticitiesfor products made under bad conditions. Theseresults support the anecdotal evidence presentedearlier. That is, companies can potentially losefrom having their products identified as beingmade under bad conditions but have little to gainfrom marketing their products as being madeunder good conditions. These results are sup-ported by Folkes and Kamins (1999) who foundthat the attitudes of consumers towards anorganization were affected more by unethicalbehavior than by prosocial behavior; that is,prosocial behavior did not compensate for aninferior product, but unethical behavior had asignificant impact on attitudes even when theproduct had superior features.

Osterhus (1997) and Bhate and Lawler (1997)examined factors affecting adoption of sociallyresponsible products. Bhate and Lawler foundthat innovators were more likely to be environ-mentally friendly consumers than adapters.Osterhus studied the influence of seven key con-structs (personal cost, personal benefit, personalnorms, social norms, attribution of responsibility,awareness of consequences, and trust), and devel-oped a model that blended normative, structuraland economic factors to determine pro-socialconsumer behavior. His three major findingswere as follows: (1) normative influences do notautomatically translate into behavior (consistentwith Roberts, 1996), (2) people are stronglyinfluenced by personal costs and rewards, and (3)personal norms matter, but interact with respon-sibility and trust.

Our review of prior research uncovered a

284 Pat Auger et al.

number of gaps in the literature. First, the liter-ature suggests that there are consistent indicationsthat, on some criteria, segments of consumersexist where ethical attributes are valued; or atleast we cannot reject the hypothesis that theyexist. However, the characteristics of the con-sumers who make up those segments are notnearly as clear. In other words, we know thatethical consumers exist but we don’t know whothose consumers are. Moreover, the limitedresearch on the characteristics of ethical con-sumers suggests that demographics are not goodpredictors of ethical consumerism (e.g., Roberts,1996). Second, evidence for the importance ofethical or social components of products is lessclear. Recent studies (e.g., Elliott and Freeman,2001) have attempted to quantify the dollar valueof these consumer preferences but we know thatthe measurement of simple, unconstrained statedpreferences will overestimate the importance ofproduct features for which there are obvious,socially correct, responses. Indeed work inrandom utility theory indicates that this willalmost always be true since response variabilityis lower in surveys than would be exhibited inthe real world. Clearly more rigorous experi-mental methodology is required. Furthermore,these studies have usually concentrated on asingle ethical attribute (e.g., working conditionsfor Elliott and Freeman, 2001) whereas productsoften incorporate a number of ethical attributes(Crane, 2001). Hence, an expansion of thenumber of ethical attributes within a single study(and product category) and a determination oftheir relative importance to consumers usingrigorous experimental methodology fills impor-tant gaps in the literature on ethical con-sumerism.

Research approach

Our research approach involves designing andimplementing a two-stage experiment thatrequires two survey instruments. The two instru-ments are an ethical disposition survey (EDS) anda choice experiment (Louviere et al., 2000). TheEDS contained 20 Machiavellianism questionsfrom the MACH IV Machiavellianism scale of

Christie and Geis (1970). The choice experimentcontained eight experimental conditions plustwo control conditions. The eight experimentalconditions included all attributes in the hypo-thetical product profiles (i.e., both functionaland ethical attributes) and involved an overar-ching experiment in which subjects weresupplied with a professionally designed newsarticle about ethical factors related to the pro-duction of the products used in the choiceexperiment, bar soaps or athletic shoes (seeAppendix A). The control conditions did notinclude any of the ethical attributes in thechoice experiment profiles. Subjects in thecontrol conditions received either “full” infor-mation (control 1) about all ethical factors(described later), or no information (control 2)about ethical factors. Subjects were also randomlyassigned to one of three conditions whereby twogroups of subjects received both a choice exper-iment and an EDS survey, and the third groupgot the choice experiment only. For thosesubjects receiving both surveys half saw the EDSsurvey first and half received the choice experi-ment first.

The surveys were administered to three groupsof subjects: MBA students at an Australian uni-versity, undergraduate students at a Hong Konguniversity (see Calder et al., 1981, 1982 for a dis-cussion of the validity of the use of students inexperiments), and supporters of the human rightsorganization Amnesty International (in Australia).We chose those three groups deliberately in orderto maximize the effect size. That is, we wereinterested in trying to see how large the overallrange of effects of ethical attributes might berather that trying to estimate the effects in thepopulation per se. As an incentive to participate,the student subjects were given an opportunityto enter a lottery with total prizes of A$500 (inAustralia) and $HK4000 (in Hong Kong). TheAmnesty International supporters had A$5donated to Amnesty International for each setof surveys completed. The Hong Kong surveyswere translated into Chinese and back-translatedto ensure consistency. The sample is deliberatelynon-representative of the society but was chosento provide a widest range of attitudes towardethical products – from more conservative MBAs,

What Will Consumers Pay for Social Product Features? 285

to status conscious young students, to self-revealed supporters of social causes.

The choice experiment survey requiredrespondents to: (1) evaluate their most recentlypurchased brand, (2) decide whether to considerand purchase 32 hypothetical bath soap orathletic shoe products, and (3) answer a series ofsocio-demographic questions. Bar soaps andathletic shoes were used because of familiarityand relevance to specific ethical issues, namely

environmentalism, labor and animal rights, andto encompass high involvement and low involve-ment product purchases. Subjects were randomlyassigned either to bar soap or athletic shoesurveys. The functional and ethical productattributes for both products are shown in TableI. The functional product attributes were pre-tested to ensure their relevance to consumerpurchase decisions and price levels were consis-tent with current prices in both markets. We

286 Pat Auger et al.

TABLE IProduct features and ethical attributes used in the experiment

Athletic Shoes 00000000000000000000Soap

Basic product features:

Shock absorption/cushioning (LOW or HIGH) Shape (ROUNDED or SQUARE)

Weight (LIGHTER or HEAVIER) Natural ingredients (NO or YES)

Ankle support (LOW CUT or HIGH CUT) Scented (NO or YES)

Sole durability (SHORT or LONG) Artificial colors (NO or YES)

Breathability/ventilation (LOW or HIGH) Moisturizer (NO or YES)

Fabrication Materials (SYNTHETIC or LEATHER) Anti-bacterial protection (NO or YES)

Reflectivity at night (NO or YES) Will it clog your pores? (NO or YES)

Comfort/fit (LOW or HIGH) Will it worsen your acne? (NO or YES)

Brand of shoe (Nike, Adidas, Reebok, or Others: Brand name (MAJOR MULTI-NATIONAL orNew Balance, Converse, Brooks, Fila, LOCAL BRAND)Puma, Etonic, Asics, Saucony)

Price ($40, $70, $100, $130) – in Australia Price ($2.25, $1.65, $1.05, $0.45) – in Australia

Price ($300, $550, $800, $1,050) – in HK Price ($6, $9, $12, $15) – in HK

Ethical features:

Is child labour used in making the product? Biodegradable formulation? (NO or YES)(NO or YES)

Are workers paid above minimum wage? Tested on animals? (NO or YES)(NO or YES)

Are workers’ working conditions dangerous? Animal by-products used as ingredients? (NO or YES) (NO or YES)

Are workers’ living conditions at the factory acceptable? (NO or YES)

Note: Two level items coded as –1 and +1. The first item (e.g., NO) is coded –1 and the second (e.g., YES)is coded +1.

chose ethical product attributes on the basis oftheir pertinence to each product category andrepresentativeness of ethical concerns expressedby human rights, environmental and animalactivist groups and journalists.

Respondents were shown hypotheticalnewswire articles that highlighted the features ofthe product category (see Appendix A). Thesearticles were pre-tested for believability and com-parability across both product categories andthese tests indicated that they were realistic andeffectively similar in style and content. Allsubjects received a core part of the articledescribing functional features of their productcategory, but were randomly assigned to eightexperimental conditions that systematically variedthe presence or absence of the ethical factorsmentioned in the articles (see Table II).

Results

Simple sample results

In total, 1,253 people were surveyed: 396undergraduate subjects in Hong Kong, 357MBA students in Australia, and 500 AmnestyInternational supporters (also in Australia). 111surveys were completed and returned from theHong Kong students (28 percent), 162 from theAustralian MBAs (45 percent), and 172 (34percent) from the Amnesty International sup-porters. Table III contains summary statisticsthat characterize the respondents. Hong Kongsubjects are mostly undergraduates, hence areconsiderably younger than both the Australiangraduate students and the Amnesty International(AI) supporters. They are also less likely to havechildren, be married or have postgraduatedegrees.1 Considerably more women respondedin the Amnesty International supporter (67

What Will Consumers Pay for Social Product Features? 287

TABLE IIChoice experiment conditions

Experimental condition 00000000000000000000000Product category

Athletic shoes Soap

0 – Control (No ethical features in article profile) NNNN NNN1 – Control (All ethical features in article profile) MMMM MMM2 NNNN NNN3 NNMM NNM4 NMNM NMN5 NMMN NMM6 MNNM MNN7 MNMN MNM8 MMNN MMN9 MMMM MMM

N = Ethical feature is not mentioned in the article and included in surveyM = Ethical feature is mentioned in the article and included in surveyOrder of ethical features (e.g, MMM and MMMM implies that all are mentioned):

Athletic Shoes 00000000000000Soap

Child labour Biodegradable formulationWorkers paid above minimum wage Animal testingWorking conditions Animal by-products used as ingredientsLiving conditions

288 Pat Auger et al.

TABLE IIISample characteristics

Hong Kong Australian AI supporter university university sample

sample sample

Male (percent) 46.40 68.10 33.00

Age (percentages)

≤ 19 00.90 00.62 01.5020–29 93.80 61.11 16.9030–39 05.40 31.48 21.9040–49 00.00 03.70 27.3050+ 00.00 01.23 32.30

Education (highest degree)

High School 18.02 01.85 11.20Attended university 27.93 00.00 15.70University degree 45.05 54.94 54.30Post graduate degree 03.60 38.27 18.80

Family income

≤ $20 999 36.04 14.20 13.80$21 000–$35 999 33.33 54.32 19.50$36 000–$61 999 13.51 00.00 29.20$62 000–$77 999 05.41 24.69 07.70≥ $78 000 09.91 00.00 29.70

Lifestyle

Single 93.69 59.26 30.30Married 03.60 23.46 49.00Divorced 00.00 03.09 08.60Cohabiting 00.90 11.11 01.50

Children (percent having) 01.70 14.29 48.00

Ethnicity

White (European/American) 00.00 46.30 83.60Chinese 98.20 19.75 00.00North or South East Asian 00.00 04.94 15.90South Asian 00.00 09.88 00.50Black, Native or Arabic 00.00 03.09 00.00

Note: At the time of the surveys, A$1.00 = US$0.63 and US$1.00 = HK$7.73. Numbers may not add up to100 percent due to missing data.

percent) and Hong Kong (54 percent) samplesthan in the Australian university sample (32percent) due largely to a greater proportion offemales in these samples (the gender balance ineach response group did not differ significantlyfrom the sample characteristics). Although thesampling does not claim to be representative ofeither the populations of Australia or HongKong, these statistics do indicate that a wide anddiverse group of individuals took part in thesurvey and they are representative of a wide crosssection of society.

We undertook additional comparisons toidentify other differences in the samples. Theaverage MACHIV scale score was approximatelythe same in both groups of students but muchlower in the case of the AI supporters (theMACHIV scale has a neutral midpoint of 100and ranges between 60 and 140). Specifically,t-tests between the samples indicated that thestudents were significantly different on theiraverage MACHIV score from the AI supportersat 0.001 level (t values of 3.30 for Australia vsAI and 3.96 for Hong Kong vs AI) while the twostudent samples were not significantly different(t value of 0.20). We also found that womenscored lower on Machiavellianism (mean of 95.10versus 97.48 for males; t = 2.50 p < 0.01). Totest whether there were sample differences onthese measures other than what would be relatedto demographic and individual differences weregressed the MACHIV scores on a host ofdemographic and individual characteristics and adummy variable for the sample (with theAustralian MBA as the baseline). The results(available from authors) imply that there are nosample differences of note other than what isrelated to demographic and individual differ-ences.

Table IV summarizes the evaluations of con-sumers’ current brands. The figures in Table IVexclude subjects who indicated that they did notfeel confident enough to know the feature pos-sessed by their current brand. Table IV suggeststhat consumers are reasonably confident in theirknowledge about non-ethical features of currentproducts, but are quite poor at rememberingsome of the most basic ethical attributes of theproducts they purchase. For example, 90 percent

of Hong Kong subjects did not know the ethicalfeatures of their current bath soap and only 5percent knew ethical features of their athleticshoes; 80 percent of the Australians did not knowthe ethical features of their current soap and only10 percent knew the ethical features of athleticshoes. AI supporters were somewhat moreknowledgeable about the ethical features of theirbath soap but less knowledgeable about ethicalfeatures of their athletic shoes. Nike dominatedthe brands purchased by all consumers, particu-larly so in Hong Kong.

Conditional choice experiment results

Our next set of analyses consisted of a series ofbinary logit models that take into account dif-ferences in the EDS, demographics, and infor-mation on the nature and breadth of ethicalattributes. The analyses produced a mass of datadue the number of variables included in themodels. We included the following variables foreach product category: (1) the features of theproducts (traditional and ethical), (2) demo-graphic variables (including MACHIV score), (3)controls for missing demographic variables, (4)controls for experimental conditions (informa-tion supplied in articles), (5) 2-way interactionsbetween experimental conditions and demo-graphics (a total of 44 variables for soap and 55for athletic shoes), (6) 2-way interactionsbetween the ethical features, and demographicsand experimental conditions (33 variables forsoap and 44 for shoes), and (7) 3-way interac-tions between ethical features, experimental con-ditions, and demographic variables (30 variablesfor soap and 40 for shoes).

The large number of variables in each analysisprecludes us from including the details of theanalyses in the paper. Instead, we conductedseveral additional analyses to simplify the pre-sentation of our results: (1) an analysis of theimportance of specific variable groups; and (2) astylistic presentation of different consumersegments based on their valuation of ethicalfeatures that highlights the magnitude of theimportance of these features.

What Will Consumers Pay for Social Product Features? 289

290 Pat Auger et al.

TABLE IVCurrent product evaluations

HK university Australian university AI supporter sample sample sample

Soap (percent knowing)

Shape 087.03 097.24 92.55Ingredients 035.18 058.53 51.06Scent 096.30 095.12 91.49Artificial coloring 038.89 063.41 55.32Moisterizer 066.67 080.48 65.96Anti-bacterial 038.89 074.39 62.77Pore clogging 027.78 037.80 22.34Acne causing 025.29 040.24 27.66Price 064.81 090.24 78.72Biodegradable 007.41 023.17 20.21Animal testing 009.25 014.63 21.28Animal by-products 011.11 014.63 29.79

Soap brand (percent of last purchases)

Local brand 018.51 036.58 45.74Multinational brand 040.74 031.70 23.40Don’t remember 040.75 031.72 30.85

Athletic shoe (percent knowing)

Shock absorption 067.77 087.50 75.31Weight 080.70 090.00 71.60Suppleness (ankle support) 089.50 092.50 80.25Sole durability 070.18 075.00 55.56Breathability 063.15 081.25 46.91Fabric 061.40 097.50 83.95Reflectivity 094.73 096.25 83.95Fit 092.98 096.25 85.19Price 100.00 100.00 88.89Child labor 005.26 011.25 06.17Minimum wage 007.01 008.75 02.47Dangerous working conditions 005.26 010.00 02.47Acceptable living standards 003.51 010.00 02.47

Athletic shoe brand (percent of last purchases)

Nike 070.17 040.00 29.63Reebok 005.26 013.75 06.17Adidas 007.01 007.50 12.35Asics 000.00 008.75 04.94All others 017.56 030.00 46.91

What factors matter to consumer purchasing behavior?Because of the complexity of our estimatedmodel, we investigated the significance of groupsof variables by removing them from the analysesand comparing the consideration and purchaselikelihoods to the base model using a chi-squaretest. This is similar to performing an ANOVAin which the conditions are presence/absence ofthe variable(s) in question and the dependentvariable is the probability of consideration orpurchase. This allows us to state statisticallywhether the addition of a variable (such as childlabor) or a group (such as age and gender)significantly impacts respondents’ probability ofconsidering or purchasing a specific product/feature bundle. The results of these analyses arepresented in Tables V and VI for bath soaps andathletic shoes, respectively.

The results clearly show that ethical featureshave a substantial impact on the purchase inten-tions of the consumers in our samples. Theseeffects are demonstrated by the high levelsof significance when the ethical features areremoved from the models as a group and by thepredominance of large effect sizes for these vari-ables. This large effect sizes must, however, bekept within the context of our sampling proce-dure, which aimed to maximize effect sizes bycarefully selecting the groups of respondents (e.g.,Amnesty International volunteers versus businessschool students). Of the possible significanteffects, between 20.4 percent (for athletic shoes)and 18.5 percent (for bath soap) were related tothe ethical features.2 Further examination ofindividual ethical features indicates some subtledifferences. For bath soaps, animal testing ishighly significant but biodegradability is muchless important and use of animal-by-products isinsignificant. A similar pattern emerges from theathletic shoe analyses with the overall ethicalfeatures group highly significant, but only childlabor and dangerous working conditions indi-vidually significant. Child labor is clearly themost important ethical feature in purchase/con-sideration probabilities, dominating a majority ofthe functional features.

We found little support for a relationshipbetween ethical disposition and ethical decision-making. The Machiavellianism group of variables

was only significant in one model – the consid-eration model for athletic shoes – and we con-cluded that ethical disposition has little predictivepower with respect to ethical decision-making.These results are consistent with Roberts (1996),but slightly inconsistent with the moral judgmentliterature (e.g., Rest, 1986). We qualify thisinconsistency as “slight” since the theoreticallink between moral judgment and behavior isexpected to be modest (Rest et al., 1997) andwith nearly one third of empirical studies unableto show a statistically significant relationshipbetween the two (Thoma et al., 1986). On theother hand, demographic variables were signifi-cant in all models, especially age and gender,lifestyle (e.g., divorced, single, married andcohabitation), and ethnicity. The income andeducation group was highly significant in thebath soap models, but much less significant in theathletic shoes models (this appears to be drivenalmost entirely by income). Interestingly, wecould discern no immediately obvious pattern.For example, for bath soaps, older females wereless likely to purchase products using animaltesting and more likely to purchase biodegrad-able products; however, for athletic shoes,younger consumers were more affected by childlabor and older consumers by minimum wagepayments. Although all consumers respondednegatively to bad ethical features, Chinese con-sumers were less negative in their responses andWhite consumers exhibited more extremeresponses to “bad” features. It is worth notingthat this effect is not confounded with the threeseparate samples because the AI and MBAsamples include a reasonable proportion of non-white respondents.

An important issue is the extent to whichinformation on ethical features can affect onthe nature and breadth of purchase intentionsand valuation. Our results strongly suggest thatproviding information about ethical featuresincreases the size of that feature’s effect onconsumer purchase probabilities. For example,28.5 percent (athletic shoes) to 28.7 percent (bathsoap) of the possible significant effects wereinfluenced either directly or indirectly by men-tioning an ethical feature. This was particularlytrue for animal testing in the case of bath soaps

What Will Consumers Pay for Social Product Features? 291

292 Pat Auger et al.

TA

BL

E V

Eff

ect

of v

aria

ble

grou

ps o

n co

nsid

erat

ion

and

purc

hase

pro

babi

lity

equa

tion

s: b

ath

soap

s

Con

side

rati

on m

odel

Pur

chas

e m

odel

Log

Deg

rees

of

ρ2L

ogD

egre

es o

f ρ2

likel

ihoo

dfr

eedo

mlik

elih

ood

free

dom

Full

mod

el (

All

vari

able

s in

clud

ed)

1758

.86

143

0.22

214

20.2

514

30.

224

Eff

ect

of r

emov

ing

ethi

cal

prod

uct

feat

ures

All

ethi

cal

prod

uct

feat

ures

1381

.59*

**07

70.

174

1072

.01*

**07

70.

171

Bio

degr

adab

le f

orm

ulat

ion

1686

.89*

**12

10.

213

1350

.83*

**12

10.

215

Ani

mal

tes

ting

is

done

1497

.23*

**12

10.

189

1171

.09*

**12

10.

187

No

anim

al b

ypro

duct

s us

ed17

10.4

0***

121

0.21

613

60.1

6***

121

0.21

7

Eff

ect

of e

thic

al f

eatu

res

men

tion

ed i

n ar

ticl

e14

13.6

4***

061

0.17

910

92.1

1***

061

0.17

4

Eff

ect

of r

emov

ing

dem

ogra

phic

s

All

dem

ogra

phic

cha

ract

eris

tics

1222

.19*

**03

40.

135

0947

.08*

**03

40.

131

Age

and

gen

der

1624

.02*

**11

80.

205

1314

.00*

**11

80.

210

Life

styl

e an

d ch

ildre

n15

67.5

0***

100

0.19

812

06.3

1***

100

0.19

2In

com

e an

d ed

ucat

ion

1645

.45*

**12

80.

204

1267

.73*

**12

80.

203

Eth

nici

ty16

75.8

5***

118

0.21

213

00.5

0***

118

0.20

7

Eff

ect

of m

achi

avel

liani

sm17

00.4

1***

131

0.21

513

48.1

1***

131

0.21

5

Eff

ect

of t

he i

nter

acti

on b

etw

een

ethi

cal

feat

ures

, ar

ticl

e m

enti

ons

and

dem

ogra

phic

s17

05.2

1***

113

0.21

513

60.5

9***

113

0.21

7

Not

e: *

p<

0.1

0.

**p

< 0

.05.

**

*p

< 0

.01.

What Will Consumers Pay for Social Product Features? 293

TA

BL

E V

IE

ffec

t of

var

iabl

e gr

oups

on

cons

ider

atio

n an

d pu

rcha

se p

roba

bilit

y eq

uati

ons:

ath

leti

c sh

oes

Con

side

rati

on M

odel

Pur

chas

e M

odel

Log

D

egre

es o

f ρ2

Log

D

egre

es o

f ρ2

likel

ihoo

dfr

eedo

mlik

elih

ood

free

dom

Full

mod

el (

all

vari

able

s in

clud

ed)

1596

.54

180

0.19

711

94.3

018

00.

211

Eff

ect

of r

emov

ing

ethi

cal

prod

uct

feat

ures

All

ethi

cal

prod

uct

feat

ures

1265

.43*

**09

20.

156

0971

.56*

**09

20.

172

Chi

ld l

abor

is

used

14

36.1

1***

158

0.17

710

90.6

3***

158

0.19

3M

inim

um w

age

is p

aid

1552

.54*

**15

80.

191

1169

.20*

**15

80.

206

Dan

gero

us w

orki

ng c

ondi

tion

s15

01.2

4***

158

0.18

511

31.7

3***

158

0.20

0A

ccep

tabl

e liv

ing

cond

itio

ns15

49.1

5***

158

0.19

111

59.1

8***

158

0.20

5

Eff

ect

of e

thic

al f

eatu

res

men

tion

ed i

n ar

ticl

e10

60.6

6***

076

0.13

108

41.5

2***

076

0.14

9

Eff

ect

of r

emov

ing

dem

ogra

phic

s

All

dem

ogra

phic

cha

ract

eris

tics

0830

.29*

**04

20.

102

0592

.59*

**04

20.

105

Age

and

gen

der

1337

.82*

**14

80.

165

1001

.74*

**14

80.

177

Life

styl

e an

d ch

ildre

n14

17.9

6***

125

0.17

510

82.8

7***

125

0.19

1In

com

e an

d ed

ucat

ion

1498

.38*

**16

10.

185

1148

.65*

**16

10.

203

Eth

nici

ty14

65.0

2***

149

0.18

110

64.8

6***

149

0.18

8

Eff

ect

of m

achi

avel

liani

sm14

06.0

4***

165

0.17

311

40.2

2***

165

0.20

1

Eff

ect

of t

he i

nter

acti

on b

etw

een

ethi

cal

feat

ures

, ar

ticl

e m

enti

ons

and

dem

ogra

phic

s15

29.8

5***

140

0.18

911

48.3

5***

140

0.20

3

Not

e: *

p<

0.1

0.

**p

< 0

.05.

**

*p

< 0

.01.

and child labor, minimum wages and workingconditions in the case of athletic shoes. It also isinteresting to note that the size of the impactsof features that were mentioned was much largerfor bath soaps than athletic shoes. We mightspeculate as to why this might be, such as thepossibility that more people are knowledgeableabout the plight of shoe workers or the impacton low involvement products is relatively larger,but our data and results do not permit firmerinsights, so a better explanation must wait forfuture research.

What is the value of ethical product features? The choice experiment approach allows us toconvert the probability of consideration andpurchase directly into conditional dollar equiva-lents. By comparing the dollar value of specificbundles of product features one can estimate thedollar equivalent of the utility that a consumerderives from the presence/absence of specificfeatures. Details of how to calculate “willingness-to-pay” for features are discussed in Louviere etal. (2000), but briefly, the desired quantity issimply the price sensitivity adjusted differencein the expected maximum utilities of the dif-ferent product mixes. Hence if the bundle ofproduct attributes can be represented by vectorJ = {j1, j2, . . . , jk} where Jk represents J withone product feature (k) changed (e.g., twoproduct are identical in every way except thatone includes child labor), the dollar value dif-ference between J and Jk is [1/–βprice](EU(Jk) –EU(J)), where EU(•) is the expected value of themaximum utility of a set of product features and–βprice is the price coefficient from the binarylogit model.

To simplify the presentation, we provide twosets of results. The first includes the estimatedvalue of specific attributes independent of demo-graphic or personality characteristics (we onlyprovide results for the consideration modelbecause results for the buy model were similar),which allows us to look at the direct effects ofpresence/absence of each feature for the sampleas a whole. These values (or “willingness-to-pay”estimates) essentially measure the consumersurplus received from a feature. For bath soapsthe results are shown in Figure 1 and for athletic

shoes in Figure 2 (along with an indication ofwhich estimates where significantly different fromzero). What is striking about the estimates is therelatively large magnitude of the ethical features.In the case of bath soaps, animal testing has thelargest overall effect, while the other two attrib-utes have magnitudes similar to the moisturizersand natural ingredients. In the case of athleticshoes, child labor is valued more than any otherattribute other than the shoe’s “fit”. The otherfour ethical features are on par with the mostsignificant of the base features of the shoes.

The second analysis is more stylized and showsthe creation of a series of illustrative segmentsbased upon specific combinations of demo-graphic attributes. The models that we estimatedpotentially allow us to examine more than 800billion possible segment combinations; hence wereduced this to a manageable size by simply illus-trating a range of segments based on the per-centage of the total product value estimated tobe attributable to ethical product features.3,4 Tenstylized segments are presented in Table VII forsoaps and Table VIII for athletic shoes,5 whichshows that the range of possible impacts of theethical features can be considered in two ways:(1) by looking at the percentage of the value theconsumer puts on the product, or (2) the dollarvalue they place on an individual feature. In thecase of the former, this ranges from a low ofaround 16 percent to a high of 97 percent for theoverall impact of the ethical features on consumerchoices. However, these figures are a bit decep-tive because consumers who place relatively littlevalue on the features tend to obtain somewhathigher total value from the products. Thus, ifwe examine the range of dollar values we seethat biodegradability is valued fairly lowly($0.00–$0.16) but that some segments put quitehigh value on no animal testing ($0.06–$0.87)and the absence of animal byproducts ($0.00–$0.63). More interestingly, the segments exhibitdifferent patterns: e.g., segments A and B exhibitlittle interest in ethical features; segments C, G,H, I and J seem focused on animal testing andwill pay between $0.33 and $0.87 to purchaseproducts that do not feature testing; segmentF value biodegradability and avoid animalbyproducts and will pay a fair amount to avoid

294 Pat Auger et al.

animal byproducts; and segments D and E showdistinct preferences for animal testing and animalbyproducts.

Table VIII contains the results for athleticshoes and used the same algorithm to createsegments. Again, ethical features exhibit a widerange of impacts, but the percentages are largerwith a low of around 31 percent and a high of94 percent. However, unlike bath soaps the totalvalue of the product does not seem to be relatedto the value of the ethical features becausesegments that value the product highly also mayvalue the ethical features highly, as exhibitedby segments G and H. The results for the rangeof dollar values suggests that acceptable livingconditions are not highly valued ($0.47–$29.74), payment of minimum wages is moder-ately important ($0.32–$35.09), almost everyonedislikes child labor ($0.28–$84.73), and there isa wide range of values for dangerous workingconditions ($0.03–$121.44). Again, what is more

interesting is that different segments havedifferent patterns, albeit this is less clear-cut thanfor bath soaps. For example, segment A is uniquein not caring about child labor but was quiteconcerned about working conditions; segmentsB and E focused more on child labor andminimum wages; segments C, D and I were con-cerned about child labor; segments G and Hwere concerned about working conditions,minimum wages, and child labor, with segmentH also focused on living conditions; segmentsF and J were similar to G and H in so far asthey focused on all the issues but found childlabor and dangerous working conditions morevalued and wages less valuable; and segment Fwas somewhat concerned about acceptable livingconditions. In general, our results indicate strongethical value positions in all segments.

What Will Consumers Pay for Social Product Features? 295

Note: * p < 0.10. ** p < 0.05. *** p < 0.01.

Figure 1. Valuation of base and ethical attributes of bath soaps independent of demographics and all otherexperimentation manipulations (consideration model estimations only)

Discussion

General findings

Our results demonstrate that consumers have ageneral and fairly rational view of ethical issuesas they pertain to product purchases. Equallyimportantly, most consumers seem quite ignorantof the ethical features that comprise the productsthey consider and purchase. Overall and onaverage, our subjects displayed a relatively strongdislike of animal abuse and child labor and overallgave quite high valuations to the ethical com-ponents of products. In addition, we find rathersignificant effects for specific ethical features forcertain types of individuals but not for others.Although we found little association betweenpersonality type disposition (Machiavellianism)and the degree to which individuals wouldconsider ethical factors, there were strong asso-ciations with several demographic variables suchas age, gender, lifestyle, and ethnicity (among

others). The latter results allow one to obtainimportant insights into identifying consumergroups (segments) that are more likely to considerethical features, and we showed that our analyt-ical approach could be utilized to estimate thevalues that different groups place on specificbundles of ethical product features. This mayseem trivial but it challenges directly the limitedempirical work on environmentalism where thesearch for “green” consumers has come up empty(e.g., Manrai et al., 1997) and suggests thatreadily available and measurable demographicvariables may be useful for identifying consumerswho are more likely to respond to an ethicalposition by an organization.

Theoretical and practical implications

We began this study with the simple goal ofaddressing a much discussed but little understoodphenomenon, the willingness of consumers to

296 Pat Auger et al.

Note: * p < 0.10. ** p < 0.05. *** p < 0.01.

Figure 2. Valuation of base and ethical attributes of athletic shoes independent of demographics, brand and allother experimentation manipulations (consideration model estimations only)

What Will Consumers Pay for Social Product Features? 297

TA

BL

E V

IISt

yliz

ed s

egm

ent

exam

ples

bas

ed o

n va

luat

ions

fro

m c

onsi

dera

tion

est

imat

es:

bath

soa

ps

Styl

ized

seg

men

tsA

BC

DE

FG

HI

J

Perc

enta

ge o

f pr

oduc

t va

lue

in e

thic

al f

eatu

res

16%

25%

31%

34%

40%

42%

63%

64%

95%

97%

Val

ue o

f in

divi

dual

fea

ture

s:

Bio

degr

adab

ility

$0.0

9$0

.01

$0.0

5$0

.09

$0.1

0$0

.16

$0.0

2$0

.02

$0.0

0$0

.00

Ani

mal

tes

ting

$0.1

1$0

.06

$0.4

0$0

.20

$0.1

3$0

.10

$0.3

3$0

.43

$0.5

3$0

.87

Ani

mal

byp

rodu

cts

$0.0

5$0

.13

$0.1

1$0

.18

$0.2

7$0

.63

$0.0

1$0

.01

$0.0

0$0

.00

Tota

l va

lue

of p

rodu

ct$1

.55

$0.7

9$1

.83

$1.4

2$1

.26

$2.1

5$0

.56

$0.7

2$0

.55

$0.9

0

Dem

ogra

phic

char

acte

ristic

s:

Gen

der

Mal

eFe

mal

eM

ale

Mal

eM

ale

Mal

eFe

mal

eFe

mal

eM

ale

Mal

eM

arit

al s

tatu

sSi

ngle

Sing

leM

arri

edM

arri

edC

ohab

itin

gC

ohab

itin

gM

arri

edM

arri

edD

ivor

ced

Div

orce

dE

duca

tion

Post

gra

dU

nive

rsit

yPo

st g

rad

Post

gra

dH

igh

scho

olPo

st g

rad

Uni

vers

ity

Post

gra

dPo

st g

rad

Post

gra

dE

thni

city

Whi

teW

hite

Non

W/C

Whi

teW

hite

Whi

teN

on W

/CN

on W

/CC

hine

seN

on

W/C

Inco

me

(ran

ge i

n $0

00)

21–3

621

–36

21–3

621

–36

21–3

621

–36

21–3

636

–62

36–6

262

–80

Age

(ra

nge)

30–3

920

–29

30–3

920

–29

40–4

920

–29

40–4

940

–49

30–3

940

–49

Chi

ldre

nN

one

Two

Non

eN

one

Non

eN

one

Non

eN

one

Non

eN

one

Not

e: N

on W

/C i

ndic

ates

eth

nici

ty t

hat

is n

eith

er w

hite

nor

Chi

nese

. T

his

incl

udes

Nor

th o

r So

uth

Eas

t A

sian

, So

uth

Asi

an,

Bla

ck,

Nat

ive

orA

rabi

c.

298 Pat Auger et al.

TA

BL

E V

III

Styl

ized

seg

men

t ex

ampl

es b

ased

on

valu

atio

ns f

rom

con

side

rati

on e

stim

ates

: at

hlet

ic s

hoes

Styl

ized

seg

men

tsA

BC

DE

FG

HI

J

Perc

enta

ge o

f pr

oduc

t va

lue

in e

thic

al f

eatu

res

31%

32%

47%

55%

56%

59%

65%

74%

78%

94%

Val

ue o

f in

divi

dual

fea

ture

s:

No

child

lab

or00

$0.2

80$

23.8

60$

77.5

8$3

0.96

0$25

.02

0$79

.19

0$74

.40

0$85

.43

0$78

.87

0$79

.22

Min

imum

wag

e is

pai

d00

$8.2

20$

47.4

900

$6.4

10$

0.32

0$15

.22

0$12

.15

0$35

.09

0$34

.85

00$3

.01

00$2

.90

Wor

king

con

diti

ons

are

not

dang

erou

s0$

30.8

600

$5.9

300

$7.7

20$

0.03

00$1

.80

0$19

.66

0$59

.89

$121

.44

00$2

.77

0$34

.76

Liv

ing

cond

itio

ns

are

acce

ptab

le00

$1.8

500

$3.2

000

$0.7

40$

1.35

0$18

.34

0$26

.49

00$0

.47

0$29

.74

00$4

.44

00$3

.49

Tota

l va

lue

of p

rodu

ct$1

34.1

6$2

51.1

7$1

98.1

9$5

9.53

$108

.61

$233

.71

$262

.85

$366

.87

$114

.18

$128

.68

Dem

ogra

phic

char

acte

ristic

s:

Gen

der

Fem

ale

Fem

ale

Fem

ale

Mal

eM

ale

Fem

ale

Mal

eM

ale

Fem

ale

Fem

ale

Mar

ital

sta

tus

Mar

ried

Sing

leSi

ngle

Mar

ried

Sing

leC

ohab

itin

gM

arri

edD

ivor

ced

Coh

abit

ing

Mar

ried

Edu

cati

onH

igh

Hig

h U

nive

rsit

yH

igh

Hig

h Po

st g

rad

Hig

h U

nive

rsit

yU

nive

rsit

yU

nive

rsit

ysc

hool

scho

olsc

hool

scho

olsc

hool

Eth

nici

tyC

hine

seW

hite

Non

W/C

Non

W/C

Whi

teC

hine

seW

hite

Whi

teN

on W

/CW

hite

Inco

me

(ran

ge i

n $0

00)

62–8

021

–36

21–3

621

–36

21–3

662

–80

36–6

236

–62

62–8

062

–80

Age

(ra

nge)

40–4

940

–49

20–2

920

–29

40–4

920

–29

20–2

930

–39

20–2

930

–39

Chi

ldre

nTw

oTw

oN

one

Non

eN

one

Non

eTw

oTw

oN

one

Two

Not

e:N

on W

/C i

ndic

ates

eth

nici

ty t

hat

is n

eith

er w

hite

nor

Chi

nese

. T

his

incl

udes

Nor

th o

r So

uth

Eas

t A

sian

, So

uth

Asi

an,

Bla

ck,

Nat

ive

orA

rabi

c.

consider non-product ethical features in theirpurchase decisions. Although there has been con-siderable discussion about the role of non-com-mercial factors in business activity, there is littlesolid empirical scientific evidence to support anyone position. For example, although the SouthAfrica boycott is looked on by human rightsgroups as a model for how such activity canbe effective in changing a regime’s stance, andis being replicated in the case of Burma, thescientific evidence appears to indicate that thefinancial impact of the boycott was minor (Teohet al., 1999). However, this is only one side ofthe story because if publicity can be targeted ata significant and concerned group of consumers,it is conceivable that more than the pride ofsenior management can be damaged. Forexample, the alignment of anti-globalizationgroups with labor activists could be quite coun-terproductive. Edmonds and Pavcnick (2002)show that targeted trade sanctions on exportsfrom developing countries to eradicate childlabor lead to significantly worse living and socialconditions for children and women. What wecontribute is the fact that the average consumerin our sample is quite willing to pay a signifi-cant percentage of the value of the product forspecific ethical features. If this is a somewhatmore generalizable phenomenon it indicates thatperhaps the most significant motivator forbusiness to take ethical considerations seriouslyis not the stick of policy sanctions or rabid pro-testers but the fact that their primary stakeholders– their customers – may be more amenable topaying for such features than previously believed.

What is clear from the extensive work in thearea is that individuals, at least in developedcountries where all the research has been doneto date, have attitudes that are aligned with amore ethical stance around purchasing. However,as we have noted there is a paucity of evidencethat when we go beyond attitude to behavior.Luzar and Cossé (1998) apply the Azjen andFishbein (1980) model to show that the takinginto consideration beliefs, intention and attitudessignificantly increases the predictive validity ofcontingent valuation models, which is a variantof the methodology used here. The main findingof their work is that attitudes mattered when they

were most salient; meaning that they were atti-tudes quite clearly related to the issue at handrather than more general behavioral measures.

This is consistent with our findings althoughwe don’t measure specific attitudes towardthe ethical features used.6 What we do knowis that general beliefs, as measured by theMachiavellianism scale,7 do not relate to anyspecific level of intention to pay. However, whatwe do show is that salience matters and saliencecan be altered with information. For example,despite occasionally extensive media coverageafforded to ethical issues, our results show thatmost consumers do not understand the ethicaldimensions of the products that they purchase.However, it is equally obvious that some con-sumers could be convinced to alter their purchasepatterns if relevant ethical information is pre-sented in an adequate and effective way. Ouranalyses clearly show the significant impact ofinformation on ethical features and its relation toethical purchase intentions. This is consistentonce again with the Ajzen and Fishbein (1980)model as applied by Luzar and Cossé (1998)where they find that intention to act is moder-ated by opportunity to act in a specific way.

Hence, our results provide support for anec-dotal evidence of the effectiveness of the corpo-rate strategies of organizations like the BodyShop; i.e., satisfy consumers’ basic needs in theproducts (the functional features) while at thesame time meet their ethical requirements. Wewould speculate that companies could be under-estimating the power of their ethical stance as atleast a short-term differentiator. Consumers mayact very differently once the right sort of infor-mation is provided. Whether one wanted to goso far as to argue that such evidence implies thatproducts should require mandatory labeling thatprovided certain social guarantees (as is now donewith respect to certain products in Europe) isbeyond the scope of this work and represents amoral stance rather than an empirical one.However, we should caution that our work, aswell as that of others, shows that there is a sig-nificant group of people who do not value anethical product position. Such mandatory policyaction would be imposing costs on these indi-viduals that they are clearly unwilling to bear.

What Will Consumers Pay for Social Product Features? 299

Limitations and future research

The focus of our research was limited. Weexamined only two countries (Australia andHong Kong), and a limited sample within each(university and graduate students, and AmnestyInternational supporters), and two products(shoes and soap). We restricted ourselves to anarrow set of labor, environmental and animalrights issues. However, these issues span at leastfive of the areas covered by the UniversalDeclaration of Human Rights, the InternationalConvention on Civil and Political Rights and theInternational Covenant on Economic, Social andCultural Rights – working under favorable con-ditions, the right to rest and leisure, the right tofood, clothing and housing, children’s rights, andthe right of procreation. However, even withthese limitations we found some consistent andinteresting results that have potentially importantimplications for the way corporations’ view theircustomers.

Naturally, our research also has limitations. Forexample, a wider sample of consumers, productsand ethical features would seem to be bothdesirable and necessary. In particular, a broadersample of countries could provide insights aboutwhether our results were due to a more affluentsample of consumers. For example, when theUniversal Declaration of Human Rights wasbeing formulated, it was found that leadersaround the world considered the same set ofrights important but for very different reasons.Indeed, Jacques Maritain noted that “we agreeabout the rights so long as no one asks us why”(Glendon, 1998). Also, had we been able tocontrol for cultural orientation (e.g., Aaker andMaheswaran, 1997), we could have avoided grosscharacterizations of country cultures. Finally,better integration of work on brand personalityand equity (e.g., Aaker, 1997; Erdem and Swait,1998) could provide a useful addition to both theethical consumerism and branding literaturesbecause our results suggest that there are aspectsof brands not covered by this literature.

The growing importance of this topic can beseen in the current movement toward socialaccountability audits (Economist, 1999), and theincreasing amounts of money that corporations

spend to “fix” systems under pressure fromincreasingly sophisticated interest groups. Ourresearch is relevant to this issue because it showsthat although consumers do not understand theethical dimensions of the products they purchase,these dimensions can influence their purchasesif information about them is properly presented.From a managerial perspective, more work needsto be done on educating consumers to thehidden ethical nature of the products theypurchase. Corporations also need to understandwhich customers care about ethical dimensionsand what dimensions they care about. However,our results also suggest that consumers won’tsacrifice product performance for ethical con-siderations in spite of what activists might hope.

Acknowledgements

This research was supported by funding fromthe Faculty of Business at the City Universityof Hong Kong, the Centre for CorporateChange at the Australian Graduate School ofManagement and the Australian ResearchCouncil. We would like to thank AmnestyInternational for their support and participationin this research. The opinions expressed hereinare those of the authors only. The authors arelisted alphabetically and contributed equally tothe project.

Appendix A: Examples of primingarticles

Athletic Shoes – All Ethical Features Mentioned

A sneaker by any other name

By Sandra Berwyn

The product choice available to today’s athlete – pro-fessional, amateur or casual – is truly amazing. Alsowhat was once a product for use only by the athlet-ically inclined has become an everyday fashion item.The humble sneaker has come of age.

Today’s sophisticated athletic shoes are made formany different people and purposes. They differ basednot only on comfort and cushioning but have many

300 Pat Auger et al.

additional specialized characteristics. Shoes vary basedon their ability to ventilate your feet, whether theysupport your ankles, their weight and the durabilityof the soles. Reflective athletic shoes protect thenighttime athlete by increasing his/her visibility.Shoes are available in a variety of synthetic and naturalmaterials.

Most athletic shoes are made in developing nationswhere labour rates are lower and production lesscostly. This has raised a dilemma for shoe manufac-turers since the labour standards in these countriescan be quite lax. It is not uncommon to find thatproducts coming from these countries have beenmanufactured using child labour or produced insubstandard manufacturing facilities. Additional com-plaints are that many workers work for less than thelegally mandated minimum wage and their employersdo not provide adequate living conditions at thefactory.

It was once the case that when you purchased arunning shoe or basketball shoe your choice waslimited to a few standard options. However, thevariety available to today’s consumer is a blisteringarray that is meant to satisfy almost any consumer’sathletic requirements.

Source: GlobeNet News Service

Athletic Shoes – No Ethical Features Mentioned

A sneaker by any other name

By Sandra Berwyn

The product choice available to today’s athlete – pro-fessional, amateur or casual – is truly amazing. Alsowhat was once a product for use only by the athlet-ically inclined has become an everyday fashion item.The humble sneaker has come of age.

Today’s sophisticated athletic shoes are made formany different people and purposes. They differ basednot only on comfort and cushioning but have manyadditional specialized characteristics. Shoes vary basedon their ability to ventilate your feet, whether theysupport your ankles, their weight and the durabilityof the soles. Reflective athletic shoes protect thenighttime athlete by increasing his/her visibility.Shoes are available in a variety of synthetic and naturalmaterials.

It was once the case that when you purchased arunning shoe or basketball shoe your choice waslimited to a few standard options. However, thevariety available to today’s consumer is a blistering

array that is meant to satisfy almost any consumer’sathletic requirements.

Source: GlobeNet News Service

Bath Soaps – All Ethical Features Mentioned

It’s not just an ordinary bar of soap

By John Baldwin

Soap is one of the oldest and most basic commodi-ties known to mankind. It exists in a variety of formsand is used by billions of people everyday. Theordinary bath soap, that bar sitting in your bathroom,can be anything from the very simple formulationused, and perhaps made, by your grandmother to aquite complex mixture of ingredients.

Today’s sophisticated soaps are made for manydifferent people and purposes. They vary based notonly fragrance and moisturizing capacity but havemany additional medicinal characteristics. Non-come-dogenic soaps keep the bather’s pores open andunclogged while non-acnegenic soaps are speciallyformulated so as not to aggravate acne conditions.“All natural” soaps avoid the use of non-natural ingre-dients and artificial colouring and “anti-bacterial”soaps aim to stop the spread of germs.

Traditional soaps are made from animal by-products(e.g., from tallow, a rendering of beef fat). A concernfor animal rights and environmental protection hashad an effect on the lowly soap. Companies todaymarket products guaranteed not to be tested onanimals or use any animal by-products. Similarly,concern for the environment has led to the develop-ment of soaps with biodegradable ingredients.

It was once the case that when you purchased abar of soap your only choice was the rectangular baror the round bar. However, the variety available totoday’s consumer is a blistering array that is meant tosatisfy almost any consumer’s skin type and cleaningrequirements.

Source: World News Service

Bath Soaps – No Ethical Features Mentioned

It’s not just an ordinary bar of soap

By John Baldwin

Soap is one of the oldest and most basic commodi-ties known to mankind. It exists in a variety of forms

What Will Consumers Pay for Social Product Features? 301

and is used by billions of people everyday. Theordinary bath soap, that bar sitting in your bathroom,can be anything from the very simple formulationused, and perhaps made, by your grandmother to aquite complex mixture of ingredients.

Today’s sophisticated soaps are made for manydifferent people and purposes. They vary based notonly fragrance and moisturizing capacity but havemany additional medicinal characteristics. Non-come-dogenic soaps keep the bather’s pores open andunclogged while non-acnegenic soaps are speciallyformulated so as not to aggravate acne conditions.“All natural” soaps avoid the use of non-natural ingre-dients and artificial colouring and “anti-bacterial”soaps aim to stop the spread of germs.

It was once the case that when you purchased abar of soap your only choice was the rectangular baror the round bar. However, the variety available totoday’s consumer is a blistering array that is meant tosatisfy almost any consumer’s skin type and cleaningrequirements.

Source: World News Service

Notes

1 Somewhat wishfully, many of the Hong Kongrespondents answered the question about having auniversity degree as if they had already achieved thedegree.2 These numbers are determined by counting thenumber of times a variable appears (excluding price,brand and control conditions).3 This turns out to be a hugely complex task sinceevery valuation needs to be considered relative toalternative bases; that is, each segments value of eachattribute is conditional upon the level of all otherattributes. For simplicity, we assumed that (1) all the“valued” base product features were available and (2)that all changes in other characteristics, income, edu-cation and so on, were relative to the mean of thesample.4 We also limited out illustration by estimating thesegment values using the consideration model andkeeping the range of demographics restricted. So, forexample, we only looked at the range of incomes andages above and below the median (not the full rangeof the sample).5 It is important to not overly compare the estimatesfrom Figures 1 and 2 with those from Tables VII andVIII. The estimates in Figures 1 and 2 assume nodemographic heterogeneity exists – in essence zeroing

out all but direct effects and comparing the results toa product in which none of the attributes features arepresent – while those in Tables VII and VIII are basedupon the differences from a mean profile. Given theresults from Tables VII and VIII show that the demo-graphics matter quite a lot, the estimates in Figures 1and 2 should be utilized just to understand the mag-nitude of direct effects. This can be seen by the factthat Table VII shows that animal byproducts are notimportant but the direct effect shown in Figure 1 issignificant. This is due entirely to the negating effectof demographics on the impact of the direct effect.6 In the case of the Australian MBA sample and theHong Kong undergraduate sample we also includedquestions taken from the MORI poll used byCAFOD. These items asked general intent questionsand could arguably said to represent attitudes withregard to specific ethical positions (e.g., child labor,work practices, animal testing, etc.). There was norelationship between the response to these uncon-strained survey questions and intention to purchasespecific products containing certain ethical featuresthat were not being picked up by demographics. Itwas excluded from the Amnesty International samplefor simplicity of execution.7 Just for clarity it should be stated that we used anumber of different belief measures tapping moralrelativism and ethical idealism. None of thesemeasures had any impact either. The Machiavellianismscale was kept as evidence of this lack of relevance toconsumers’ valuations.

References

Aaker, J.: 1997, ‘Dimensions of Brand Personality’,Journal of Marketing Research 34(3), 347–356.

Aaker, J. and D. Mahaeswaran: 1997, ‘The Effect ofCultural Orientation on Persuasion’, Journal ofConsumer Research 24(4), 315–328.

Abratt, R. and D. Sacks: 1988, ‘The MarketingChallenge: Towards Being Profitable and SociallyResponsible’, Journal of Business Ethics 7, 497–508.

Ajzen, I. and M. Fishbein: 1980, UnderstandingAttitudes and Predicting Social Behavior (Prentice Hall,Englewood Cliffs, NY).

AsiaWeek: 2001, ‘Fair Pay’, 27(1).Bhate, S. and K. Lawler: 1997, ‘Environmentally

Friendly Products: Factors That Influence TheirAdoption’, Technovation 17(8), 457–465.

CAFOD: 1998, ‘Views from the South, ConferenceReport on Ethical Trade’, September, http://www.cafod.org/uk/policyviews.htm.

302 Pat Auger et al.

Calder, B., L. Phillips and A. Tybout: 1981,‘Designing Research for Application’, Journal ofConsumer Research 8(3), 197–207.

Calder, B.: 1982, ‘The Concept of External Validity’,Journal of Consumer Research 9(4), 240–244.

Christie, R. and F. L. Geis: 1970, Studies inMachiavellianism (Academic Press, New York).

Click, J.: 1996, ‘New Business Standards Focus onHuman Rights’, HRMagazine 41(6), 65–72.

Crane, A.: 2001, ‘Unpacking the Ethical Product’,Journal of Business Ethics 30, 361–373.

Economist: 1999, ‘Business Ethics: Sweatshop Wars’,27 February, 68–69.

Edmonds, E. and N. Pavcnik: 2002, ‘DoesGlobalization Increase Child Labor? Evidence fromVietnam’, Working paper No. 8760, NationalBureau of Economic Research, Cambridge, MA.

Elliott, K. A. and R. B. Freeman: 2001, ‘White Hatsor Don Quixotes? Human Rights Vigilantesin the Global Economy’, Working paperNo. 8102, National Bureau of Economic Research,Cambridge, MA.

Erdem, T. and J. Swait: 1998, ‘Brand Equity as aSignaling Phenomenon’, Journal of ConsumerPsychology 7(2), 131–157.

Fairclough, G.: 1996, ‘It Isn’t Black and White’, FarEastern Economic Review 159(10), 54–57.

Folkes, V. S. and M. A. Kamins: 1999, ‘Effects ofInformation About Firms’ Ethical and UnethicalActions on Consumers’ Attitudes’, Journal ofConsumer Psychology 8(3), 243–259.

Fullerton, S., K. B. Kerch and H. R. Dodge: 1996,‘Consumer Ethics: An Assessment of IndividualBehavior in the Market Place’, Journal of BusinessEthics 15(7), 805–814.

Garrett, D. E.: 1986, ‘Consumer Boycotts: Are theTargets Always the Bad Guys’ Business and SocietyReview (Summer), 17–21.

Glendon, M. A.: 1998, ‘Propter Honoris Respectum:Knowing the Universal Declaration of HumanRights’, Notre Dame Law Review 73(5), 1153–1190.

Lerner, L. D. and G. E. Fryxell: 1988, ‘An EmpiricalStudy of the Predictors of Corporate Performance:A Multi-Dimensional Analysis’, Journal of BusinessEthics 7(12), 951–959.

Louviere, J. J., D. A. Hensher and J. D. Swait: 2000,Stated Choice Methods: Analysis and Applications(Cambridge University Press, Cambridge, U.K.).

Luh, S. S.: 2001, ‘Nike Factories Accused ofIntimidating Workers’, Asian Wall Street Journal (17 May).

Luzar, E. J. and K. J. Cossé: 1998, ‘Willingness to Payor Intention to Pay: The Attitude-Behavior

Relationship in Contingent Valuation’, Journal ofSocio-Economics 27(3), 427–444.

Manrai, L. A., A. K. Manrai, D. N. Lascu and J. R.Ryans Jr.: 1997, ‘How Green-Claim Strength andCountry Disposition Affect Product Evaluation andCompany Image’, Psychology and Marketing 14(5),511–537.

Marymount University Center for Ethical Concerns:1999, ‘The Consumers and Sweatshops’,November, http://www.marymount.edu/news/garmentstudy/overview.html.

Miller, C.: 1997, ‘Marketers Weigh the Effects ofSweatshop Crackdown’, Marketing News 31(10), 1,19.

Osterhus, T. L.: 1997, ‘Pro-Social ConsumerInfluence Strategies: When and How Do TheyWork?’, Journal of Marketing 61(4), 16–29.

Pitts, R. E., J. K. Wong and D. J. Whalen: 1991,‘Consumers’ Evaluative Structures in Two EthicalSituations: A Means-End Approach’, Journal ofBusiness Research 22(2), 119–130.

Ranjan, P.: 2001, ‘Credit Constraints and thePhenomenon of Child Labor’, Journal ofDevelopment Economics 64, 81–102.

Rest, J.: 1983, ‘Morality’, in P. H. Mussein (seriesed.), and J. Flavell and E. Markman (vol. eds.),Handbook of Child Psychology: Vol. 3. CognitiveDevelopment, 4th Edition (Wiley, New York), pp.556–629.

Rest, J.: 1984, ‘The Major Components of Morality’,in W. Kurtines and J. Gerwitz (eds.), Morality, MoralBehavior, and Moral Development (Wiley, New York),pp. 24–40.

Rest, J.: 1986, Moral Development: Advances in Researchand Theory (Praeger, New York).

Rest, J., L. Edwards and S. Thoma: 1997, ‘Designingand Validating a Measure of Moral Judgment: StagePreference and Stage Consistency Approaches’,Journal of Educational Psychology 89(1), 5–28.

Rest, J., D. Narvaez, M. J. Bebeau and S. J. Thoma:1999, Postconventional Moral Thinking: A Neo-Kohlbergian Approach (Lawrence Erlbaum Press,Mahwah, NJ).

Roberts, J. A.: 1996, ‘Will the Real SociallyResponsible Consumer Please Step Forward?’,Business Horizons 39(1), 79–83.

Shrum, L. J., J. A. McCarty and T. M. Lowrey: 1995,‘Buyer Characteristics of the Green Consumer andTheir Implications for Advertising Strategy’, Journalof Advertising 24(2), 71–82.

Smith, C. S. and A. C. Copetas: 1998, ‘FormerChinese Prisoner Claims Adidas Balls Were Madein Prison’, Wall Street Journal (26 June).

What Will Consumers Pay for Social Product Features? 303

Sriram, V. and A. M. Forman: 1993, ‘The RelativeImportance of Products’ Environmental Attributes:A Cross-Cultural Comparison’, InternationalMarketing Review 10(3), 51–70.

Teoh, S. H., I. Welch and C. P. Wazzan: 1999, ‘TheEffect of Socially Activist Investment Policieson the Financial Markets: Evidence from TheSouth African Boycott’, Journal of Business 72(1),35–89.

Thoma, S. J., J. Rest and R. Barnett: 1986, ‘MoralJudgment, Behavior, Decision Making, andAttitudes, in J. Rest (ed.), Moral Development:Advances in Theory and Research (Praeger, NewYork), pp. 133–175.

University of Maryland, Program on InternationalPolicy Attitudes: 2000, ‘Americans onGlobalization: A Study of Public Attitudes’, March,http://www.pipa.org/onlinereports/globaliza-tion/global_rep.html.

Whalen, D. J., R. E. Pitts and J. K. Wong: 1991,‘Exploring the Structure of Ethical Attributions asa Component of the Consumer Decision Model:The Vicarious Versus Personal Perspective’, Journalof Business Ethics 10(4), 285–293.

Pat AugerMelbourne Business School,

200 Leicester Street,Carlton, Victoria 3053, Australia

E-mail: [email protected]

Paul BurkeTimothy M. Devinney (contact author)

Australian Graduate School of Management,Sydney NSW 2052, Australia

E-mail: [email protected]@agsm.edu.au

Jordan J. LouviereUniversity of Technology,

1–59 Quay Street, Haymarket,Sydney NSW 2007, Australia

E-mail: [email protected]

304 Pat Auger et al.