See discussions, stats, and author profiles for this publication at: [Link]

net/publication/269338278

Return Shipping Policies of Online Retailers: Normative Assumptions and the

Long-Term Consequences of Fee and Free Returns

Article in Journal of Marketing · September 2012

DOI: 10.1509/jm.10.0419

CITATIONS READS

213 4,908

1 author:

James G Maxham
University of Virginia
27 PUBLICATIONS 6,618 CITATIONS

SEE PROFILE

All content following this page was uploaded by James G Maxham on 30 September 2015.

The user has requested enhancement of the downloaded file.

 Amanda B. Bower & James G. Maxham III

Return Shipping Policies of Online

Retailers: Normative Assumptions
and the Long-Term Consequences of
Fee and Free Returns
To limit costs associated with product returns, some online retailers have instituted equity-based return shipping
policies, requiring customers to pay to return products when retailers determine that customers are at fault. The
authors compare the normative assumptions about customers that underlie equity-based return shipping policies
with the more realistic, positivist expectations as predicted by attribution, equity, and regret theories. Two
longitudinal field studies over four years using two surveys and actual customer spending data indicate that retailer
confidence in those normative assumptions is unjustified. Contrary to retailer assumptions, neither the positive
consequences of free returns nor the negative consequences of fee returns were reversed when customer
perceptions of fairness were taken into account. Depending on the locus and extent of blame, customers who paid
for their own return decreased their postreturn spending at that retailer 75%–100% by the end of two years. In
contrast, returns that were free to the consumer resulted in postreturn customer spending that was 158%–457% of
prereturn spending. The findings suggest that online retailers should either institute a policy of free product returns
or, at a minimum, examine their customer data to determine their customers’ responses to fee returns.

P
Keywords: product returns, online retailing, regret, equity, customer spending

roduct returns are a widespread and expensive prob- based on a consideration of inputs and outcomes, and prior
lem. For example, product returns of consumer elec- research has associated fairness perceptions with a positive
tronics cost retailers and manufacturers almost $17 effect on important postexchange customer reactions such
billion in 2011, representing a 21% increase in returns since as satisfaction, word of mouth, trust, commitment, and
2007 (Wolf 2012). Thus, many retailers have established repurchase intentions (e.g., Maxham and Netemeyer 2003;
return shipping policies intended to limit their own costs Oliver and Swan 1989a, b; Swan and Oliver 1991; Tax,
(e.g., Kandra 2000; Meyer 1999). A policy commonly insti- Brown, and Chandrashekaran 1998). We define “return
tuted by distant retailers (e.g., [Link]) is an equity- shipping policy cost fairness” (cost fairness) as the extent to
based return shipping policy: If the retailer determines that which customers believe the return shipping policy out-
it is to blame for the return, the retailer absorbs the return’s come (whether fee or free) is fair. Consistent with both
cost; otherwise, customers must pay those costs. While these prior work and the assumptions of equity-based return ship-
retailers appear to assume that consumers’ equity assess- ping policies, we expect that perceptions of return cost fair-
ments are the only relevant reaction to return shipping costs ness are positively related to postreturn repurchases (see
(for a model of retailers’ assumptions, see Figure 1), distant Figure 2).
retailers’ concern with the fairness is reasonable. “Fairness” Research has yet to investigate how these return ship-
refers to “rightness or deservingness” (Oliver 1997, p. 194) ping policies and associated costs can influence customer
evaluations and subsequent postreturn spending. In the pre-
sent research, we identify the apparent, normative assump-
tions underlying the equity-based return shipping policies
of free return (i.e., the retailer absorbs the return shipping
Amanda B. Bower is Professor of Business Administration/Marketing &

fee) versus a fee return (i.e., the customer pays the return
Advertising, Williams School of Commerce, Economics, & Politics, Was-

shipping fee) and compare those assumptions with a posi-

hington and Lee University (e-mail: bowera@[Link]). James G. Maxham

University of Virginia (e-mail: maxham@[Link]). This research was

tivist perspective on consumers’ psychological reactions
III is Chesapeake & Potomac Telephone Company Professor of Commerce,

and postreturn spending. Working with two leading online

funded in part by the Bernard A. Morin Fund for Marketing Excellence at

retailers, we coupled responses from two online surveys at

the McIntire School of Commerce. The authors thank Bill Ross, Ruth

key points over the course of customers’ return experiences

Bolton, Rick Netemeyer, David Mick, Amar Cheema, and research semi-

with the customers’ 24-month prereturn and 24-month

nar participants at the University of Virginia, Penn State University,

postreturn purchase histories. We find that consumer assess-

Georgetown University, and Peking University for helpful comments on

ments of fairness and attributions are inconsistent with the

earlier versions of this article. Robert Leone served as area editor for this
article.

© 2012, American Marketing Association Journal of Marketing

ISSN: 0022-2429 (print), 1547-7185 (electronic) 110 Volume 76 (September 2012), 110–124
#

FIGURE 1
Model of the Normative Assumptions of the Product Returns Process Underlying Equity-Based Return
Shipping Policies

At
Attributed
tributed to
to Fr ee Return
Free Returnn
Retailer
Retailer

2%
2%)#'"%+3(&#((%(($%0)&
)#'"%+33((&#((%(($%0)& 6,&$'(#77"'/#)',0&,-&
6,&$'(#77"'/#)',0&,-&
,-&4"#$%&'(&/,0('()%0)&
,-&4"#$%&'(&/,0('()%0)& +%)1+0&(*'77'08&7,"'/9&
+%) 1+0&(*'77'08&7,"'/9&
5' )*&)*%&/,0(1$%+3(&
5')*&)*%&/,0(1$%+3(& Equity
Eq u ity Postreturn
Po s tre tu r n
Pr oduct Fa
Product ilu re
Failure
#((%((($%0)&
#((%(($%0)& Accomplished
Ac c o m p lis h e d Spending
Sp e n d in g

!"
!"#$%&'(&%')*%+&),&
# $ % &'( &% ') * % + &) , & :,0(1$%+(&5'""&
:,0(1$%++(&5'""&
+%)
+%)#'"%+&,+&),&
#'"%+&,+&),&
At
Attributed
tributed to to Fe
Feee R e tu rn
Return 7%+/%';%&#&+%)1+0&
7%+/%';%&#&+%)1+0& =>1')9&'(&)*%&,0"9&
=>1')9&'(&)*%&,0"9&
Self/Consumer
Se lf /C o n s u m e r (*'77'08&7,"'/9&#(&-#'+&
(*'77'08&7,"'/9&#(&-#'+& +%(
+%(7,0(%&),&#&
7,0(%&),&#&
(% "-../,0(1$%+&
(%"-./,0(1$%+&
'-
-&)*%&,0%&),&4"#$%&
'-&)*%&,0%&),&4"#$%& +%)
+%)1+0&(*'77'08&
1+0&(*'77'08&
7#9 (&-,+&)*%&+%)1+0<&
7#9(&-,+&)*%&+%)1+0<& 7," '/9&)*#)&#---%/)(&
7,"'/9&)*#)&#--%/)(&
%;%0 &'-&)*%&/,0(1$%+&
%;%0&'-&)*%&/,0(1$%+& 7,()+%)1+0&
7,()+%)1+0&
7#9 (&-,+&')&
7#9(&-,+&')& (7%0?'08&
(7%0?'08&

Notes: Explanations of assumptions underlying process model structure are in italics.

FIGURE 2
Conceptual Model of Consumer Responses to Product Return Shipping Policies

Retailer
Attribution for
Product Returns

Return Shipping H1
Customer

H2 H4 H4
Policy Perceptions of Customer Regret Customer Spending
(Free/Fee) Cost Fairness

Self-Attribution
for Product
# Returns

H3

normative (and self-serving) assumptions of retailers. Not policies, web interface) could directly influence consumer
only do retailers overestimate the ameliorating (moderat- loyalty intentions in the present product return context.
ing) effects of attributions on fee returns, but they also Consistent with the present research, there is some
ignore consumers’ affect stemming simply from return fees. research indicating that return policies instituted with the
In addition, free returns resulted in increases in postreturn short-term gain in mind may have long-term negative con-
spending (from prereturn levels), and fee returns resulted in sequences for the retailer. Despite retailer desire to control
decreases in postreturn spending (from prereturn levels), all for “inappropriate” or “opportunistic” product returns with
regardless of blame attributions. stricter return policies (Davis, Hagerty, and Gerstner 1998;
Research on return policies is still developing. Both Hess, Chu, and Gerstner 1996), Wood (2001) finds that
Pasternack (2008) and Padmanabhan and Png (1997) exam- lenient policies (manipulated in two of her studies as
ine manufacturer return shipping policies offered to retail- including free shipping) were associated with increased
ers. Other research has assessed actual consumer responses probability of ordering from the retailer, heightened ratings
to return policies, suggesting the benefits to retailers of easy of product quality, and a reduction in overall purchase deci-
return policies. Anderson, Hansen, and Simester (2009) sion conflict. Viewing the return process as part of a cycle,
examine the value to consumers of the simple presence (vs. Petersen and Kumar (2009) find that while an increase in
absence) of a return option and suggest a model retailers product returns results in a decrease in marketing communi-
could use to optimize return policies. Mollenkopf et al. cations from the marketer toward that consumer, that same
(2007) find that previous service experiences (e.g., return increase in returns will result in an increase in future cus-

Return Shipping Policies of Online Retailers / 111

tomer repurchases (up to a threshold). This research sug- sion” (White 1991, p. 266). Solomon (1978) reviews
gests the value in comparing the return policy assumptions research in which causal agents were measured separately
that retailers make with actual consumer reactions. (vs. on opposite ends of a single scale), concluding that the
hydraulic assumption is “untenable.” Similarly, Taylor and
Koivumaki (1976) find when measuring the two separately
The Assumptions of Equity-Based that the correlation was –.14 (not significant [n.s.]).
Return Shipping Policies Therefore, in contrast to the assumptions underlying
The implementation of an equity-based return shipping pol- equity-based return shipping policies, a stronger attribution
icy is predicated on a variety of apparently implicit and nor- to the consumer may not necessarily result in a weaker attri-
mative assumptions that lead distant retailers to believe bution to the retailer (e.g., Johnson, Mullick, and Mulford
such a policy would be both cost-effective and reasonable 2002; Miller, Smith, and Uleman 1981). Although cus-
to customers. We consider those assumptions here and com- tomers may attribute some product failures exclusively to
pare them with the customer reactions suggested by prior the retailer, customers may instead attribute failure to them-
research. selves, to neither party, or perhaps to both (e.g., Folkes
1984; Kelley, Hoffman, and Davis 1993; Oliver 1997;
Assumption of Proportional Equity Weiner 2000; White 1991). In other words, we would expect
An assumption of equity-based return shipping policies is that retailer and consumer self-attributions are independent,
that consumers will perceive an exchange as fair if the out- without a near-perfect negative relationship. Figure 2 pre-
comes they receive are proportional to the inputs they con- sents the separate conceptualization and relationships.
tribute, which translates into the “one to blame is the one to
Consequences of Inaccurate Assumptions in
pay” philosophy (see Figure 1). Although this assumption is
Attribution and Equity Assessments
consistent with more traditional equity theories (e.g.,
Homans 1961), subsequent research suggests that people As a result of these assumptions, equity-based return ship-
prefer advantageous or positive inequity (i.e., the equitable ping policies allow for only two possible pairs between
behavior that results in the maximization of one’s own out- attributions and applied return shipping policy: Retailers
comes; e.g., Lapidus and Pinkerton 1995; Oliver 1997; only pay when the return is their own fault, and consumers
Oliver and Swan 1989a). The customer view of advanta- only pay when it is their own fault (see Figure 1). However,
geous inequity as “fairer” is prevalent in customer–retailer these policies do not take into account the reactions of con-
relationships (versus interpersonal). Customers may not see sumers who are required to pay for a return for which they
themselves as having equal responsibilities to the retailer in blame the retailer, nor do they allow for the possible bene-
the exchange and may have more substantial expectations fits that might accrue when a consumer receives a free
that the retailer bear much of the burden of the exchange return when there is a stronger self-attribution. Put differ-
(e.g., Berger, Conner, and Fisek 1974; Lapidus and Pinker- ently, what happens when a consumer is “miscategorized”
ton 1995; Oliver 1997; Oliver and Swan 1989a, b). Consis- and disagrees with the retailer’s assessment?
tent with prior research, we expect that customers receiving There is a strong likelihood that consumers will dis-
free returns will report significantly higher levels of cost agree with retailer assignment of responsibility to the con-
fairness and have greater relative postreturn repurchases sumer (Oliver 1997). Consumers have a tendency to take
than customers receiving fee returns, regardless of level of more credit for positive outcomes and less blame for fail-
blame attribution (though we do expect blame attribution to ures, particularly in a marketing relationship (e.g., Oliver
moderate the extent of the effect, as discussed subsequently). 1997; Valle and Wallendorf 1977). Given their preference
for positive inequity, consumers tend to put particular
Assumptions of Causal Attribution Dependence emphasis on consumer outcomes and retailer inputs, result-
Retailers employing equity-based return shipping policies ing in a disproportionate reaction to negative inequity (e.g.,
appear to assume that consumers’ attribution of responsibil- Oliver 1997; Walster, Berscheid, and Walster 1973). There-
ity for the return will be consistent with the retailers’ (see fore, the damage done to equity perceptions and postreturn
Figure 1). Furthermore, retailers assume that these attribu- repurchases by a fee (vs. free; consumer outcomes) return
tions are negatively related so that as responsibility will be disproportionately greater when consumers make
assigned to the consumer goes up, assignment to the retailer stronger retailer attributions (vs. weaker; retailer inputs).
necessarily goes down. Called the “hydraulic assumption” Thus:
in attribution theory, support for it means that causal agents H1: Return shipping policy and retailer attributions interact
for a given outcome should have “near perfect negative cor- such that customers who experience a fee (vs. free) return
relation between these judgments” (Bassili and Racine report disproportionately lower cost fairness and decrease
1990, p. 882), “as if causal candidates competed with one spending when they indicate stronger retailer attributions
than when they indicate weaker retailer attributions.
another in a zero-sum game” (Nisbett and Ross 1980, p.
128). However, the hydraulic assumption of attributions has There may also be positive effects of a “miscatego-
been largely disproven (e.g., Bassili and Racine 1990; Krull rized” free return: a free return for which consumers
2001; Miller, Smith, and Uleman 1981; Nisbett and Ross strongly attribute the return to themselves. Thus, the pre-
1980; Taylor and Koivumaki 1976) because “internal and ferred state of positive inequity (e.g., Lapidus and Pinkerton
external [attributions] are not opposites on a single dimen- 1995; Oliver 1997; Oliver and Swan 1989a) would be

112 / Journal of Marketing, September 2012

heightened when the customer receives a free return when return the product (e.g., Oliver 1997). Of particular interest
there are greater levels of self-attribution for the need for here is the effect that return shipping costs may have on that
the product return. Furthermore, under complaint condi- baseline level of regret. Customers facing a fee return will
tions, Lapidus and Pinkerton (1995) find no evidence to have an unrecoverable monetary cost due to return shipping
support their hypothesis that consumers feel guilt or other fees, in contrast to a nonpurchase from that retailer (Gilly
unpleasant emotional states as a result of this type of posi- and Gelb 1982). Comparison of this actual monetary loss to
tive inequity. When assessing resentment stemming from a the nonpurchase alternative may heighten feelings of regret
high/low outcome in an equitable/inequitable situation, they and, in particular, concerns about future return fees stem-
find that an interaction resulted largely from the dispropor- ming from future purchases. Consistent with prior research,
tionately low levels of resentment when participants experi- we expect that customers whose regret is further heightened
enced a positively inequitable situation. In other words, it is by a fee return will prevent the experience of future regret
unlikely that there would be any negative emotional reac- by reducing their purchases from the present distant retailer.
tions (e.g., guilt) to negate or neutralize the positive reac- Conversely, customers whose regret is lowered (i.e., greater
tions resulting from a free but “undeserved” return. Thus: levels of rejoicing) as a result of a free return may increase
postreturn spending, willingly making riskier purchases.
H2: Return shipping policy and self-attributions interact such
that customers who experience a free (vs. fee) return
Thus (see Figure 2):
report disproportionately higher cost fairness and increase H3: Compared with a baseline of regret stemming from the
spending when consumers make stronger self-attributions need to return the product, customers receiving free
than when they make weaker self-attributions. returns report significant decreases in that experienced
regret, whereas customers receiving fee returns report sig-
The Central Role of Regret on Consumer nificant increases in that experienced regret.
Responses While fairness and regret have appeared as constructs in
Retailers employing equity-based return shipping policies the same study (e.g., Verhoef, Franses, and Hoekstra 2001;
clearly assume that fairness is the key response to the Vorhees, Brady, and Horowitz 2006), the relationship
return, expecting that postreturn spending will be unaf- between the two remains to be addressed. As O’Shaugh-
fected by return shipping costs if the return was “fair.” nessy and O’Shaughnessy (2005) indicate, regret theory has
However, consumers may have other, more dominant reac- implications in equity considerations. We suggest that in
tions to a fee or free return shipping policy beyond the addition to the regret heightened by return costs, consumers
deservedness or fairness of return costs. Specifically, regret might also experience heightened regret as a result of being
refers to a negative feeling or “sense of sorrow” (Simonson treated in a manner they perceive as unfair. Xia, Monroe,
1992, p. 105) experienced in response to a negative out- and Cox (2004, p. 7) argue (but do not demonstrate) that if
come when a person compares his or her own actions to consumers believe a price to be unfair, they may choose to
alternative behaviors and preferable outcomes (i.e., counter- “leave the relationship, depending on their assessment of
factuals) that might have occurred instead (e.g., Zeelenberg, which action is most likely to restore equity” (for similar
Van Dijk, and Manstead 1998). The opposite of regret is logic, see O’Shaughnessy and O’Shaughnessy 2005). Thus
rejoicing or elation (e.g., Greenleaf 2004; Inman, Dyer, and (see Figure 2):
Jia 1997; Landman 1987), which occurs when a person’s H4: Cost fairness is negatively related to regret, with regret
choices lead to an outcome that is better than if other partially or wholly mediating the relationship between
choices were made. cost fairness and postreturn spending.
Consumers are strongly motivated to avoid the emo-
tional experience of regret, leading them to protect them-
selves against it (e.g., Cooke, Meyvis, and Schwartz 2001; Methods
Greenleaf 2004; Inman and McAlister 1994). The simple
anticipation of regret with regard to a future decision may Study 1: Equity-Based Return Shipping Policies
result in inaction (i.e., nonpurchase; e.g., Landman 1987; We conducted a longitudinal event field study over four years
Lemon, White, and Winer 2002; Simonson 1992; Tsiros and with a panel of online customers (average of 8.4 orders per
Mittal 2000). In contrast, an experience with rejoicing can year) who returned products to a leading e-commerce
lead people to make decisions that may involve riskier—but retailer of frequently purchased home, garden, and personal
the hope of better—outcomes. For example, Greenleaf items. To qualify for the panel, customers needed at least 24
(2004) demonstrates that auction sellers experienced rejoic- months of prereturn spending data. We gathered data at the
ing because the winning price of an auction was higher due following six time periods: (1) 24 months before the return
to the reserve price. These sellers subsequently set an even (i.e., 24 months prereturn [T0]), (2) 12 months leading up to
higher reserve price in a second auction, even though those the return (i.e., 12 months prereturn [T1]), (3) time of return
higher reserve prices might decrease the chances of a suc- (i.e., return [T2]), (4) soon after the retailer handled the
cessful second auction. Therefore, consistent with previous return (i.e., postreturn [T3]), (5) 12 months after the return
work, we expect regret to be negatively related to postreturn (i.e., 12 months postreturn [T4]), and (6) 24 months after
repurchases (see Figure 2). the return (i.e., 24 months postreturn [T5]). The T2 data
Consumers may already experience a baseline level of were collected during approximately the same month. Thus,
regret stemming from the product failure and the need to all customers shared approximately the same T0–T5 period,

Return Shipping Policies of Online Retailers / 113

and we had data from 24 months before and after the return tial return, and the postreturn customer spending variables
for each respondent. exclude the monetary value of the focal product return as
12 and 24 months prereturn (T0, T1). We collected well as the $25 gift certificate value.
yearly prereturn purchasing history for the 24 months Return shipping policy outcome. Fifty-three percent of
before the return for the 334 respondents who completed respondents received a free return. Consistent with an
the T2 and T3 surveys. These data included the number of equity-based return shipping policy, a return manager used
orders placed, the dollar value of the orders, and the product customer self-report as an input in making “fair” judgments
descriptions. We accounted for inflation in dollar variables regarding blame and allocation of return shipping costs. In
using the seasonally adjusted Consumer Price Indexes. general, the retailer in Study 1 assigned a fee return when
Return (T2). At the time of return, 500 customers either customers indicated one of the following reasons for the
telephoned the retailer or initiated a return using the form return: (1) the item did not fit, (2) the item was too expen-
enclosed in their order, triggering the T2 online question- sive, (3) the color did not match, (4) gift recipients did not
naire link to be e-mailed. Customers were offered a $25 gift want/need, (5) the item did not fit with other components,
certificate to complete the two surveys (T2 and T3). (Only or (6) the customers changed their minds. The retailer
offered a free return when customers returned an item for
39% of respondents in Study 1 redeemed the gift certificate,
the following reasons: (1) the item was damaged in transit,
and there were no significant differences in redemption
(2) the item was defective, or (3) the company shipped the
rates across fee and free conditions [p > .50].) Of the 500
wrong item.
surveys sent, 351 customers completed usable surveys, rep-
resenting a 70% response rate. The T2 survey first asked Contextual variables. We gathered potential covariates
customers to indicate the details of their return (i.e., product both from the surveys and the retailer database. These
numbers, whether the items were purchased as gifts, reason covariates include product involvement, the dollar amount
for returning, prepurchase awareness of the return shipping of shipping costs to return the product (regardless of fee or
policy, and whether they wanted a refund or exchange). free policy), the number of days to resolve the return, the
Customers completed questions regarding situational pur- number of days that passed after receiving the product
chase involvement, regret, attributions toward the retailer, before the return was initiated, the length of the customers’
and self-attributions, and all items were measured on a relationship with the retailer (measured at 24 months pre-
seven-point scale. We adapted to this study a three-item return), the dollar amount of the order, and the dollar amount
semantic differential involvement measure from prior of the returned items.
research (Ratchford 1987) to measure whether the purchase
of the product was highly involving. A three-item retailer Study 2: Generalizing Beyond Equity-Based
attribution measure asked respondents to indicate the extent Return Shipping Policies
to which the retailer was responsible for the return, while a To rule out that cost fairness and regret reactions are due to
separate three-item self-attribution measure assessed self- the type of return shipping policy (i.e., equity based), we
attribution. We adapted a measure from Tsiros and Mittal conducted a second longitudinal field study over the same
(2000) to measure customer perceptions of regret. Finally, 49-month period with an electronics retailer that used a dif-
respondents provided demographic and buyer profile infor- ferent return shipping policy. The retailer categorized prod-
mation (see the Appendix). ucts as qualifying for free or fee returns according to the
Postreturn (T3). After the completion of the return gross margins, warning consumers before purchase (even
requiring them to click a box noting their understanding)
process (refund or receipt of a product exchange), a second
whether a returned product would be subject to shipping
survey link was e-mailed to the 351 respondents who com-
charges regardless of blame. Customers who were reim-
pleted the T2 survey. Of those, 334 customers completed
bursed were categorized as “free” (n = 682, 53), and those
usable surveys, representing a 95% response rate for T3 and
who were not reimbursed were categorized as “fee” (n =
an overall 67% response rate. The sample had the following
614). Thirty-six percent of the returns were because cus-
demographic characteristics: 58% of the respondents were
tomers changed their minds; 27% were due to problems
female, 66% were 36–55 years of age, 75% held college
with item descriptions, installation, or instructions; and
degrees, and 89% reported that they return less than 20% of
37% were due to quality problems. Customers place an
their online purchases. In addition, this was the first product
average of 12.6 orders per year with the retailer. The data
return for all customers to this retailer, creating a baseline
collection procedures and measures for the electronics sam-
for accurately tracking customer perceptions regarding their
ple mirrored those employed in the first study. Of the 2750
first return experience with the focal retailer. The T3 survey
surveys sent at the time of return (T2), 1623 customers
assessed customer perceptions of regret and cost fairness completed usable surveys, representing a 59% response
measures adapted from prior research (Smith, Bolton, and rate. After the retailer handled the return, 1296 customers
Wagner 1999; Tax, Brown, and Chandrashekaran 1998). completed and submitted a usable T3 survey (an 80%
12 and 24 months postreturn (T4, T5). We collected two response rate for T3). We collected 24 months of pre- and
years of postreturn purchasing history for the 334 respon- postreturn purchasing history for each of our 1296 respon-
dents, including the number of orders placed, dollar value dents and queried the retailer’s database to collect the same
of the orders, and product descriptions. None of our respon- contextual variables collected in the first study. Our
dents returned the purchases made 24 months after their ini- response rate from the initial mailing of the first T2 ques-

114 / Journal of Marketing, September 2012

tionnaire to the completion of the T3 questionnaire was Results
47%. Customers were offered a $25 gift certificate to com-
plete the T2 and T3 surveys. (Only 23% of respondents in The Role of Attributions in Cost Fairness and
Study 2 redeemed the gift certificate, and there were no sig- Customer Spending
nificant differences in redemption rates across fee and free We argue that the postreturn spending among customers
conditions [p > .10].) The sample exhibited the following receiving a free return significantly increases from prereturn
demographic characteristics: 48% were female, 36% were spending, while the postreturn spending among customers
36–55 years of age, and 62% held college degrees. In addi- paying a fee return significantly decreases from prereturn
tion, the product return in this study represented the first spending levels. Instead of simply examining the effects of
return recorded by the retailer for each respondent, allowing return shipping policies on changes in spending at the end
for accurate tracking of customer perceptions regarding of the 24-month postreturn period, we examined the effects
their first return experience with the focal retailer. at both the 12-month and 24-month postreturn points. Our
Across both studies at both the prereturn 12- and 24- longitudinal research design enables us to determine whether
month marks, we found no significant differences among any changes in spending results in shorter-term effects (e.g.,
the free and fee groups in prereturn purchase rates, order limited to 12 months but rebounding to prereturn levels by
values, or value of returned products, nor were there signifi- 24) or longer-term trends of postreturn spending.
cant differences in the dollar amount of return shipping Initial analyses contradicted some of the retailer
across levels of attributions to retailer (all p > .10). Confir- assumptions underlying equity-based return shipping poli-
matory factor models indicated that our measures are psy- cies. The correlations between retailer attributions and self-
chometrically sound in both studies regarding model fit, attributions, though significant in both studies, are not “per-
discriminant validity, and internal consistency (see the fect” (Study 1:  = –.17; Study 2:  = –.14). This indicates
Appendix). that attributions are empirically distinct (Fornell and Lar-
cker 1981) and should be measured separately. Evidence
Checks for Respondent and Measure Bias also contradicts the assumption that retailer assignments of
To check for sample and nonresponse biases in each sample responsibility are consistent with consumer assignments.
using customer profile information in each of the retailers’ Considering only Study 1’s results because of the retailer’s
databases, we compared the demographic and buying pro- equity-based return shipping policy, customers making
files in our samples with three other customer groups: (1) stronger attributions to the retailer were required to pay
customers who returned products during our studies but did return shipping fees (n = 73; 46%) almost as frequently as
not participate in the studies (i.e., nonparticipants; Study 1: those who received a free return (n = 86; 54%), regardless
n = 285; Study 2: n = 1545), (2) customers who returned of self-attributions. Taking into account self-attributions,
products before our studies and did not receive our survey among those customers who would meet the retailer’s own
(i.e., nonsurveyed returners; Study 1: n = 567; Study 2: n = standards for a free return (i.e., stronger retailer attributions/
1780), and (3) customers who have never returned products weaker self-attributions; n = 81), 43% were required to pay
to the focal retailers (i.e., nonreturners; Study 1: n = 462; a fee. Similarly, among those customers who would meet
Study 2: n = 1378). There were no significant differences the retailer’s standards for a fee return (i.e., weaker retailer
attributions/stronger self-attributions; n = 105), 50% received
regarding the length of relationship with the retailers, age,
a free return. Similar proportions exist in Study 2, in which
total number of purchases, or average order value between
return fee responsibility is unrelated to equity decisions and
the three other customer groups and our samples (p > .10),
instead is determined by the type of product purchased. In
and they were similar across gender, income, and education.
other words, equity-based determinations of responsibility
Likewise, the reasons for returning and the retailer’s prod- were as consistent with customer judgments as determina-
uct return strategies were similar across groups. Other data tions entirely unrelated to assessments of equity decisions.
collection and analysis indicated that the three control
To test H1 and H2, we estimated a repeated measures
groups in each sample did not differ significantly from our general linear model with one categorical between-subjects
respondents’ in customer spending (p > .10). In addition, factor (return shipping policy outcome: free and fee), two
nonparticipants and nonsurveyed returners who did not pay continuous between-subjects factors (retailer attributions
for return shipping significantly increased their customer and self-attributions), one between-subjects dependent
spending over the next two years, while nonparticipants and variable (cost fairness), and one within-subject dependent
nonsurveyed returners who paid for return shipping costs variable captured across four time intervals (customer
significantly decreased their customer spending over the spending: T0, T1, T4, and T5). We also modeled six covari-
next two years (i.e., a negative in customer spending; p < ates: involvement, the dollar amount of return shipping
.01). Nonparticipants and nonsurveyed returners with free costs, the number of days to resolve the product return, the
returns repurchased at significantly higher rates than nonre- number of days that passed after receiving the product
turners, and nonparticipants and nonsurveyed returners with before the return was initiated, the length of the customers’
fee returns repurchased at significantly lower rates than relationship with the retailer, and the order dollar amount.
nonreturners (p < .01). Overall, these data checks suggest Last, we included consumers’ prepurchase awareness of
that potential response and nonresponse biases in ratings return shipping policy as a two-level blocking factor (i.e.,
are minimal. yes or no).

Return Shipping Policies of Online Retailers / 115

 Return shipping policy awareness was significantly .10; Study 2: T5 = 3026.40, t(1,287) = 11.96, 2 = .10, p <
related to cost fairness (Study 1: F(1, 319) = 22.96, p < .01, .01).
2 = .07; Study 2: F(1, 1281) = 57.62, p < .01, 2 = .04) and To explore H1, we examined the slopes of retailer attri-
customer spending (Study 1: F(1, 319) = 38.78, p < .01, 2 = bution across fee and free return shipping policies. Next, we
.11; Study 2: F(1, 1281) = 10.41, p < .01, 2 = .01), and conducted a spotlight analysis (Fitzsimmons 2008; Irwin and
therefore we retained it in the model (all other covariates McClelland 2001) at one standard deviation above the mean
were nonsignificant and thus were eliminated). Across both of retailer attribution (i.e., stronger retailer attributions) and
studies, the two-way interaction (shipping policy ¥ retailer one standard deviation below the mean of retailer attribu-
attribution) was significant for cost fairness (Study 1:  = tion (i.e., weaker retailer attributions) to explore the details
–.53, t(325) = 2.95, p < .01, 2 = .03; Study 2:  = –.623, of the interaction. As we hypothesize, and as we show in
t(1,287) = 4.60, p < .01, 2 = .02). Likewise, the shipping Figure 3, the drop in cost fairness from a free to a fee return
policy ¥ retailer attribution was significant for customer was greater when customers more strongly blamed the firm
spending at both T4 (Study 1: T4 = 416.50, t(325) = 4.38, than when they expressed weaker retailer attributions
2 = .06; Study 2: T4 = 1519.97, t(1,287) = 8.54, 2 = .05, (Study 1:  = .76, t(330) = 8.26; Study 2:  = .82, t(1,292) =
p < .01) and T5 (Study 1: T5 = 765.39, t(325) = 6.00, 2 = 9.36, p < .01). Yet the drop in customer spending from a

FIGURE 3
Effects of Return Shipping Policy
A: Effects of Return Shipping Policy and Retailer Attributions on Cost Fairness
& &
6.00 6.00 !"#$
!"#$
5.00 5.00
(")$ '"(%

4.00 4.00 %"#&

Cost Fairness

Cost Fairness

3.00 3.00
%"&'
%"&' - -

#"%) )"'#

2.00 2.00
-
-

- -

1.00
& 1.00
&

0 0
Free Return Fee Return Free Return Fee Return
Shipping Shipping Shipping Shipping
- - - - - - - -

Study 1 Weaker retailer attribution Study 2

Stronger retailer attribution -
-

B: Effects of Return Shipping Policy and Self-Attributions on Cost -Fairness

-

& - &
6.00 6.00
'"()
'"() $"'(

5.00 5.00
!"#$
!"#$
4.00 4.00
- - - -
Cost Fairness

Cost Fairness

- - - -
#"%!
3.00 - 3.00 %"!&
%"$$ -
%"#&
%"#&
2.00 2.00
-
-

- -

1.00
&
1.00
&

0 0
Free Return Fee Return Free Return Fee Return
Shipping Shipping Shipping Shipping
- - - - - - - -

Study 1 Weaker self-attribution Study 2

Stronger self-attribution -
-
Notes: Means for retailer attributions and self-attributions occur at one standard deviation below the grand - mean (weaker) and one standard
-

deviation above the grand mean (stronger).

-

116 / Journal of Marketing, September 2012

- - - -
- - - -
free to a fee return was more precipitous when customers Regarding H2, the two-way interaction (shipping policy ¥
expressed weaker retailer attributions than when they more self-attribution) was significant in both studies for cost fair-
strongly blamed the firm (Study 1:  = 734.93, t(330) = ness (Study 1:  = –.392, t(325) = –2.19, p < .01, 2 = .02;
10.03, p < .01; Study 2:  = 954.83, t(1,292) = 12.44, p < Study 2:  = –.615, t(1,287) = 4.60, p < .01, 2 = .02). In
.01). As such, H1 is partially supported (see Figure 4). Study 1, the return shipping policy ¥ self-attribution inter-

FIGURE 4
Retailer Attributions and Changes in Customer Spending
A: Study 1

$1,400.00
!%'(&"$&)

$1,200.00

$1,000.00
!"#"$%&
!"#"$%&

$800.00 .Free . and weaker

. . . retailer attributions
Fee. and. weaker retailer attributions
.
!)-#$,)
!)-#$,) .
Free . and. stronger retailer attributions
. . .

$600.00
. .

. Fee. and stronger . retailer attributions

. . . .
. . . .
!*(%$-(
!*(%$-( .

$400.00

$200.00
!(#&$#*
!(#&$#*

!%))$-#
!%))$-# !%+,$&-
!%+,$&-

0
!&,$&(

24 Months 12 Months 12 Months 24 Months

. . . . . . . .

Prereturn Prereturn Postreturn Postreturn

. . . . . . . .

B: Study 2

$6,000.00

$5,000.00
!(#)*"%)(

$4,000.00

Free
. and
. weaker retailer attributions
5 - - - - 3

$3,000.00
!"#"$"%&' .
Fee. and
. weaker retailer attributions
- - - -
.
Free
. and
. stronger retailer attributions
- - - - 3
. .

Fee. and
. stronger retailer attributions
!,#+&$%*"

$2,000.00
- - - -
. .

!*#&$+%)+

$1,000.00
!++'%("
!++'%("

%0
!''%"+
!*"'%+,
!*"'%+, !)%))
- 24 Months 12 Months 12 Months 24 Months
Prereturn Prereturn Postreturn Postreturn
- - - 5 - - - -
. . . . . . . .

Notes: The means for weaker attributions are one standard deviation below the grand mean, and the means for stronger attributions are one
standard deviation above the grand mean.

Return Shipping Policies of Online Retailers / 117

action was not significant for customer spending at both T4 tations) and customer spending, as well as the negative rela-
(T4 = 147.46, t(325) = 1.52, n.s.) and T5 (T5 = 44.10, tionship between regret and customer spending. To assess
t(325) = .34, n.s.). Yet the interaction was significant in Study these assumptions, we first estimated longitudinal structural
2 at both T4 (T4 = 497.49, t(1,287) = 2.70, 2 = .01) and T5 models to assess the relationships of cost fairness and regret
(T5 = 679.24, t(1,287) = 2.59, 2 = .01, p < .01). We con- on customer spending over time, as well as to note the
ducted a spotlight analysis at one standard deviation above amount of variance explained in customer spending over
the mean of self-attribution (i.e., stronger self-attributions) time.
and one standard deviation below the mean of self-attribution To test H4, we examined whether regret (T3) mediates
(i.e., weaker self-attributions) to explore the details of the the relationship between cost fairness (T2) and customer
interaction. As we hypothesized, the drop in cost fairness spending (T0, T1, T4, T5) in a manner consistent with
from a free to a fee return was more precipitous when cus- Baron and Kenny (1986). We examined four conditions for
tomers more strongly blamed themselves than when they mediation using structural equation modeling. The first con-
expressed weaker self-attributions (Study 1:  = .182, dition is satisfied if cost fairness affects the mediator
t(330) = 7.43, p < .01; Study 2:  = .741, t(1,292) = 9.34, p < (regret). The second condition is satisfied if regret affects
.01; see Figure 3). Similarly, in Study 2, the drop in cus- the dependent variable (customer spending). We estimated a
tomer spending from a free to a fee return was more precip- mediated structural equation model testing the direct paths
itous when customers more strongly blamed themselves from cost fairness Æ regret Æ customer spending. Both
than when they expressed weaker self-attributions ( = these conditions were met, as this model yielded marginal fit
.985, t(1,292) = 11.93, p < .01). Yet the spotlight analysis (Study 1: 2 = 120.54, p < .01; comparative fit index (CFI) =
was not significant in Study 1. Consistent with H2, cus- .97; Tucker–Lewis index (TLI) = .95; and root mean square
tomers in both studies who experienced a free return error of approximation (RMSEA) = .15; Study 2: 2 =
reported disproportionately higher cost fairness when they 371.18, p < .01; CFI = .96; TLI = .94; and RMSEA = .14).
made stronger self-attributions than when they made Moreover, the completely standardized exogenous path
weaker self-attributions (see Figure 3). In addition, cus- from cost fairness to regret (Study 1:  = –.73; Study 2:  =
tomers in Study 2 who experienced a free return reported –.65) and the endogenous path from regret to customer
disproportionately higher increased spending when they spending (Study 1:  = .70; Study 2:  = .74) were both sig-
nificant (p < .01).
made stronger self-attributions than when they made
The third condition is satisfied if cost fairness has a
weaker self-attributions (see Figure 5). Thus, H2 is sup-
direct effect on customer spending. Thus, we estimated a
ported in Study 2 and partially supported in Study 1.
direct model with only one direct path from cost fairness to
Regret customer spending. The model fit the data well (Study 1: 2 =
.56, p = .75; CFI = .99; TLI = .99; and RMSEA = .01; Study
To test H3, we estimated another repeated measures general
2: 2 = 14.63, p < .01; CFI = .99; TLI = .99; and RMSEA =
linear model with one within-subject factor (time: T2 and
.07), and the completely standardized path was significant
T3), one between-subjects factor (return shipping policy (Study 1:  = .60; Study 2:  = .51, p < .01), satisfying the
outcome: free and fee), one dependent variable (regret), the third mediating condition.
six previously used covariates, and two additional covariates The fourth mediating condition is satisfied if the direct
(retailer attributions and self-attributions). In Study 1, retailer path from cost fairness to customer spending becomes non-
attributions, self-attributions, and return shipping policy significant (i.e., full mediation) or reduced (partial media-
awareness were significantly related to regret (retailer: F(1, tion) when we included the mediated paths from cost fair-
323) = 70.63, p < .01, 2 = .18; self: F(1, 323) = 5.00, p < ness Æ regret Æ customer spending in a full model (i.e.,
.03, 2 = .02; awareness: F(1, 323) = 46.92, p < .01, 2 = the mediated model). The fit of the mediated model was
.13). Likewise, in Study 2, retailer attributions, self-attribu- better than the fit of the full model with the added exoge-
tions, and return shipping policy awareness were signifi- nous path from cost fairness to customer spending (Study 1:
cantly related to regret (retailer: F(1, 1285) = 339.28, p < 2diff = 107.33; Study 2: 2diff = 333.51; d.f. = 1, p < .01).
.01, 2 = .21; self: F(1, 1285) = 4.68, p < .03, 2 = .01; Moreover, the completely standardized path estimate
awareness: F(1, 1285) = 167.46, p < .01, 2 = .12; all other between cost fairness and customer spending became non-
covariates were nonsignificant and eliminated). Consistent significant (Study 1:  = .02; Study 2:  = .04, p > .10), indi-
with H3, customers receiving free returns reported signifi- cating that regret fully mediates the effect of cost fairness
cant decreases in postreturn regret from initial return levels, on customer spending. Moreover, the amount of variance
whereas customers receiving fee returns reported signifi- explained in customer spending was greater for the medi-
cant increases in postreturn regret from initial return levels ated model (Study 1: R2 = .78; Study 2: R2 = .74) than for
(Study 1: F(1, 329) = 206.16, p < .01, 2 = .39; Study 2: the full (Study 1: R2 = .63; Study 2: R2 = .61) or direct
F(1, 1,291) = 667.39, p < .01, 2 = .34; see Figure 6.) (Study 1: R2 = .45; Study 2: R2 = .34) models, suggesting
that cost fairness is a better predictor of customer spending
Effects of Fairness and Regret on Long-Term when modeled as an indirect effect through regret. In sum-
Customer Spending mary, regret mediates the effect of cost fairness on customer
One of the assumptions underlying equity-based return spending, in support of H4 in both studies.
shipping policies and/or our expectations is the positive To provide context to our findings, we conducted sev-
relationship between fairness (as per retailer and our expec- eral multigroup nested models in accordance with Neff

118 / Journal of Marketing, September 2012

 FIGURE 5
Self-Attributions and Changes in Customer Spending
A: Study 1

$1,200.00
"
!*,*-'$'"

$1,000.00

$800.00
!')+$'"
!')+$'" !%'(
!%'($')
$')

. Free
. . and
. .weaker
. self-attributions

$600.00 . Fee
. . and
. .weaker self-attributions
!""#$%&
!""#$%&
.

. Free
. . and
. stronger
. . self-attributions
.

. Fee
. . and
. stronger
. . self-attributions
.

$400.00

!-*'$-+
!-*'$-+

$200.00
!*)+$+(
!*)+$+(

!%"$(&

0
!%-$""

24 Months 12 Months 12 Months 24 .Months

Prereturn Prereturn Postreturn Postreturn
. .. . . .. .2 . . . . . . .

B: Study 2

$4,000.00
!&#$)$'**
!&#$"-',&

$3,500.00

$3,000.00
!"#$%&'()

$2,500.00
!"#,"+'%&

Free
. and
. weaker
. self-attributions
Fee. and
. weaker self-attributions
; . . .

$2,000.00
.
Free
. and
. stronger self-attributions
. . .
. .

Fee. and
. stronger self-attributions
. . .

$1,500.00
. .
. . .

$1,000.00

$500.00 !"()'$(
!"()'$(
!"),'-$
!"),'-$

0
!"*'+"
!+*'*)

24 Months . 12 Months 12 Months . 24 Months

! . . . . . . . .

Prereturn Prereturn Postreturn Postreturn

. . . . . .

Notes: The means for weaker attributions are one standard deviation below the grand mean, and the means for stronger attributions are one
standard deviation above the grand mean.

(1985) to examine whether the modeled parameter esti- (self-attributions: weaker and stronger) ¥ 2 (return shipping
mates varied significantly across eight relevant customer policy: free and fee). The chi-square tests across all nested
groups: 2 (retailer attributions: weaker and stronger) ¥ 2 models indicated that the parameter estimates were stable

Return Shipping Policies of Online Retailers / 119

 FIGURE 6 versally increase their repurchases. In other words, the pri-
Changes in Regret over Time mary conclusion for retailers from the present research is
that in the interest of increased sales, it is beneficial to insti-
A: Study 1
tute a free return shipping policy. At the very least, our
7 work is a call to online retailers to consult their own propri-
&

etary customer data to determine any effects of return ship-

6
ping costs on customer relationships and purchases.
Experienced Regret

5
&"$'
&"$'

4 The Dangers of Fee Return

#"&!
#"&!

This recommendation has the potential to elicit concerns

/

3
from retailers. Retailers have short-term motivations for
!"#$
!"#$

2
!"%!
!
!"%!

1 controlling return costs. As such, they may require cus-

0 tomers to absorb return shipping policies, so they can avoid
Time of/ Return/ (T2) Postreturn (T3) those costs themselves, or even induce consumers to keep
products they might otherwise return to maintain the profits
/

from the sale. Retailers may also be concerned with limiting

Free
/ return
/ shipping Fee
/ return
/ shipping

abusive returns. However, while retailers may be in control

B: Study 2 of determining who pays for the return shipping costs, our
findings remind retailers that customers will have their own
6 independent perceptions of blame, affective reactions to
&

5 return fees, and, most importantly, ability to decide whether

'"()
'"()
Experienced Regret

4 they will repurchase from the retailer. Depending on attri-

bution condition, fee returns universally resulted in a
#"''
#"''

3 decrease in spending, ranging from 74.84% to 100% (see

/ !"#$
!"#$

Figures 4 and 5).

%"&%
%
%"&%

2
1 Our findings strongly contradict the assumptions made
by retailers that attempt to control or limit their own costs
0
Time of/ Return/ (T2) Postreturn (T3) by instituting equity-based return shipping policies. First,
retailers are particularly ineffective at categorizing blame in
/

Free
/ return
/ shipping Fee
/ return
/ shipping a manner consistent with consumer perceptions. While a
properly executed equity-based return shipping policy
Notes: Means for retailer attributions and self-attributions occur at should have all retailer-blaming customers receiving free
one standard deviation below the grand mean (weaker) and returns, the retailer in Study 1 (which used such a policy)
assigned those customers free and fee returns in approxi-
one standard deviation above the grand mean (stronger).

(i.e., not significantly different) across the eight subgroups mately equal proportions. Similarly, we found that the
(p > .10), enhancing the predictive validity of the overall retailer in Study 1 (which assigned fee returns using an
equity-based return shipping policy) was approximately as
model.
consistent in assigning responsibility for the return to con-
sumers who held themselves and not the retailer responsible
Discussion as the retailer in Study 2, which used an entirely different
Contrary to economic research suggesting that retailers policy for assigning return fees.
should toughen online return shipping policies, our studies Even if retailers made attributions of blame consistent
suggest that such strategies might be shortsighted and that with customer perceptions, the consequences of fee returns
retailers should carefully consider how return shipping poli- for retailers are still negative and profound. In a “perfect”
fee condition, in which consumers strongly blame them-
cies affect revenues. We conducted two event field studies
selves and hold weak attributions to the retailer, consumers
simultaneously over approximately 49 months to assess the
in Study 1 (in which the retailer used an equity-based return
psychological and behavioral reactions of customers to
shipping policy) still decreased their spending by 88%, and
equity-based return shipping policies. Our expectations, as those in Study 2 decreased their spending by 93%. We
reflected in Figure 2, were supported, indicating that retail- found that customers appear to prefer advantageous or posi-
ers’ normative expectations (reflected in Figure 1) are tive inequity, perceiving free returns to be fairer than fee
largely inconsistent with consumer responses. Contrary to returns. In sharp contrast to the expectations of retailers, the
retailer assumptions, the actual return shipping policy cus- dominant effect of the valence of the return shipping policy
tomers received (whether free or fee) largely determined (fee/free) is not overcome by any combination of attribution
their postreturn spending regardless of attributions and cost conditions.
fairness. Both studies suggest that customers paying for One key reason for this result is that, contrary to the
their own product returns will universally decrease their expectations of retailers, the dominant response to product
repurchases and that those receiving free returns will uni- return shipping policies is not equity but rather regret. Cus-

120 / Journal of Marketing, September 2012

tomers’ negative emotions and sorrow related to return the importance of free returns to get people to take a chance
costs are the primary driver of postreturn shipping and, on purchases (Raphel 2004). In a related and developing
indeed, entirely explain equity’s effect on postreturn repur- issue, online retailers are increasingly willing to offer free
chases. Therefore, while equity played a role in postreturn (initial) shipping to customers to heighten the possibility of
spending, we found that it was only through customers’ an initial purchase, recognizing that the increase in sales
feelings of regret. more than make up for the increase in costs (Zimmerman
Retailers may not be able to rely on the type of return and Mattioli 2011). Our findings indicate that beyond
shipping policies to cue consumer reactions to the policies. avoiding the negative consequences of fee returns, there are
Our research suggests that the type of return shipping policy the substantial advantages to retailers of free returns.
heuristic used to determine the policy application (whether While our studies were conducted in online settings, the
equity-based or dependent on the type of product being implications of our findings may generalize to other retailer
returned) is largely irrelevant to how customers might react settings. These findings may also have implications for
to a fee or free outcome. Even though the two retailers in our brick-and-mortar retailers, when the return of the product
studies had two different metrics for determining whether entails a cost. Although restocking fees are often intended
the customer received the fee or free return shipping out- to get customers to “think twice” about returns (Meyer
come, one equity-based and one product-based, the conclu- 1999), they may actually limit future customer spending for
sions across both studies were (except for self-attribution fear of future restocking fees. Finally, the products in Study
findings) similar. This finding suggests that cost fairness 1 represented a wide cross-section of products from apparel
considerations play a smaller role than regret in shaping to housewares to decorative items (representing more than
consumers’ future repurchase decisions. 200 stockkeeping units), whereas the products in the second
Retailers must also realize that consumers may not warn study were a wide variety of consumer electronics and
a retailer if a fee return will result in a decrease in future accessories (representing more than 800 stockkeeping
repurchases. The retailers in both our studies received no units). This diversity of the product categories represented
formal complaints from fee returns, with these customers across both studies suggests that the implications of these
quietly decreasing repurchases (and, in some cases, findings may generalize to retailers carrying a variety of
sharply). While retailers implementing an equity-based product types. Further research should determine whether
return shipping policy may perceive the dearth of com- these findings generalize to retailers carrying a limited
plaints among fee returners as support for such a policy, depth or breadth of product line, particularly with regard to
analysis of the longer-term consequences of fee returns sug- the benefits of free returns. If a retailer has a limited variety
gests that a preferable option from a customer loyalty per- or depth of product, especially if it is a product that need
spective is to simply offer all customers a free return. not be purchased frequently, rejoicing customers may only
be able to purchase so much, regardless of their lack of
The Benefits of Free Returns anticipated regret. However, the advantages to the retailer
Offering free returns to consumers does not just help retail- of a free return may accrue to the retailer in other ways,
ers avoid the negative consequences of fee returns. Depend- such as word of mouth (e.g., Oliver 1997).
ing on the attribution condition, if customers received free
returns, postreturn spending at that retailer was 158%– The Original and Important Role of Regret
457% of prereturn spending by the end of two years (see Consistent with the expectations of equity-based return
Figures 4 and 5). This is one of a few articles suggesting shipping policies, perceptions of fairness are positively
that product returns and their associated frustrations and related to postreturn spending. However, the importance of
costs for retailers are not “necessary evils” (to use Petersen fairness is not consistent with retailer expectations. We
and Kumar’s [2009] term). Wood (2001) suggests the value found that those perceptions of fairness are mediated by a
to retailers of lenient return policies, supporting the expec- reaction unanticipated by retailers: regret. Consumers regret
tation that after customers have taken possession of the purchasing from a company that has treated them unfairly,
product they are also more likely to keep it. Anderson, which leads to a decrease in postreturn repurchases. Consis-
Hansen, and Simester (2009) assess the value to consumers tent with the ideas of Xia, Monroe, and Cox (2004), this
(and ultimately to retailers) of offering the return option. may be due to the consumer desire either to prevent future
Furthermore, up to a given threshold, more returns result in inequities or possibly to balance the past inequity by not
an increase in repurchases (Petersen and Kumar 2009). The repurchasing.
present research supports the assertion that reducing con- Beyond its mediating relationship with fairness is
sumer costs and decreasing the hurdles associated with regret’s direct relationship with product return shipping
returns can increase the repurchases to retailers and result in policies. Hess, Chu, and Gerstner (1996) normatively
long-term benefits. assume that a rational consumer will judge nonrefundable
Some online retailers selling products with relatively charges such as return shipping costs to be sunk costs,
high return rates, such as shoes (e.g., [Link]), fashion which should not play a role in subsequent decision mak-
(e.g., [Link]), and luggage (e.g., [Link]), have ing. However, consistent with previous research (e.g.,
already adopted free return shipping policies (Spencer Simonson 1992), a past fee return serves to increase con-
2003). Indeed, the previous owner of [Link] indicated cerns regarding future fees stemming from a future pur-

Return Shipping Policies of Online Retailers / 121

chase and failed product, and this anticipation serves as a Appendix
salient issue to customers in deciding whether they will pur-
chase again from a retailer (Petersen and Kumar 2009). The
Measures1
dampening effect that these fees have on regret and ulti- Experienced Regret (T2 = .95, T3 = .96)
mately on postreturn shipping are impressive. For example, 1. I regret purchasing this product from (retailer name).
24.8% of fee respondents in Study 1 and 32.4% in Study 2 2. I am feeling rejoiceful about buying this product from
had dropped to zero revenue by the end of two years post- (retailer name). (reverse-coded)
return, compared with 12% and 15% in the free return 3. I should not have purchased this product from (retailer name).
group.
Retailers appear to be underestimating the long-term Cost Fairness ( = .96)
benefit of a free return to the retailer itself. The sense of 1. With respect to the return shipping policy outcome, (retailer
rejoicing resulting from a free return resulted in significant name) handled the return in a fair manner.
postreturn repurchases. Similar to the saying “What would 2. I believe (retailer name) applies return shipping policies
you do if you know you couldn’t fail?” our respondents fairly when handling returns.
appeared to have the philosophy “What would you buy if 3. The final return shipping policy outcome I received from
you knew you wouldn’t have to pay to return it?” Sales (retailer name) was unfair. (reverse-coded)
increases were impressive in both studies by the end of two
years postreturn (Study 1: $620.80; Study 2: $2,552.68). Pre- and Postreturn Customer Spending
Our partial support for H1 may be a further indication of 1. Year 2 prereturn customer spending (T0): U.S. dollar
amount of annual purchases with (retailer name) from the
the significance that regret plays in consumer reactions.
24 months prereturn to 12 months prereturn.
Equity theory (and our hypothesis) would predict that a fee
2. Year 1 prereturn customer spending (T1): U.S. dollar
condition under strong retailer attribution conditions would amount of annual purchases with (retailer name) from the
result in disproportionately lower postreturn repurchase. 12 months prereturn to the product return event.
Although the interaction was significant, it was due to a dis- 3. Year 1 postreturn customer spending (T4): U.S. dollar
proportionate increase in postreturn spending resulting from amount of annual purchases with (retailer name) from the
free returns received when consumers had weaker retailer return event to 12 months postreturn.
blame. Interpreted in light of regret theory, these consumers 4. Year 2 postreturn customer spending (T5): U.S. dollar
may have rejoiced when they got a free return from a amount of annual purchases with (retailer name) from 12
blameless retailer. months postreturn to 24 months postreturn.
Aside from the practical managerial contributions of the
present research, there is also a significant theoretical con- Retailer Attributions ( = .99)
tribution. Both regret and equity are frequently discussed as 1. (Retailer name) is responsible for my need to return this
antecedents of satisfaction and future behavior (e.g., Cooke, product.
Meyvis, and Schwartz 2001; Lapidus and Pinkerton 1995; 2. To what extent was (retailer name) responsible for the
return that you experienced? (1 = “not at all responsible,”
Oliver 1997). However, there is limited research that and 7 = “totally responsible”)
includes both constructs in the same discussion or analysis, 3. To what extent do you blame (retailer name) for this return?
and the relationship between regret and equity is infre- (1 = “not at all,” and 7 = “completely”)
quently addressed. Chatterjee (2007) puts forth unsupported
expectations suggesting that regret might serve as an under- Self-Attributions ( = .99)
lying mechanism in the relationship between next-purchase
1. I am responsible for my need to return this product.
coupons and perceptions of retailer fairness. In their devel-
2. The return that I experienced was my fault.
opment of untested research propositions regarding con-
sumer reactions to unfair prices, Xia, Monroe, and Cox 3. To what extent do you blame yourself for this return? (1 =
“not at all,” and 7 = “completely”)
(2004) argue that negative emotions, regret included, serve
as the primary driver of future action. To the best of our
1Unless noted, items were anchored by 1 = “strongly disagree”
knowledge, the present research represents the first tested
and 7 = “strongly agree.” Confirmatory factor measurement mod-
hypotheses of the relationships between fairness, regret, and els across both studies indicate strong internal consistency. The
postpurchase behavior and, in particular, the first demon- average variance extracted between each pair of constructs is
stration that regret mediates the effects of fairness on post- greater than 2 (i.e., the squared correlation between two con-
purchase customer behavior. This finding suggests that cus- structs [Fornell and Larcker 1981]), indicating strong discriminant
tomers do not simply have negative affect as a result of validity. Measurement model (Anderson and Gerbing 1988):
Study 1: 2 = 830.19, d.f. = 294, CFI = .96, TLI = .95, and
being treated unfairly but actually regret being treated RMSEA = .07; Study 2: 2 = 3375.64, d.f. = 294, CFI = .95, TLI =
unfairly and are motivated to avoid inequitable treatment in .92, and RMSEA = .08.  = average composite alpha reliability
the future. estimate across both studies.

122 / Journal of Marketing, September 2012

Involvement ( = .85) 3. The purchase of this product (1 = “required very little
thought,” and 7 = “required a lot of thought”).
1. The purchase of this product was (1 = “very unimportant,”
and 7 = “very important”).
2. With regard to the purchase of this product, how concerned
Return Shipping Policy Awareness
were you about the outcome? (1 = “very unconcerned,” and 1. Were you aware of (retailer name)’s return shipping policy
7 = “very concerned”) before completing your order? (1 = “no,” and 2 = “yes”)

REFERENCES
Anderson, Eric T., Karsten Hansen, and Duncan Simester (2009), Irwin, Julie R. and Gary H. McClelland (2001), “Misleading
“The Option Value of Returns: Theory and Empirical Evi- Heuristics and Moderated Multiple Regression Models,” Jour-
dence,” Marketing Science, 28 (3), 405–423. nal of Marketing Research, 38 (February), 100–109.
Anderson, James C. and David W. Gerbing (1988), “Structural Johnson, Lee M., Rehan Mullick, and Charles L. Mulford (2002),
Equation Modeling in Practice: A Review and Recommended “General Versus Specific Victim Blaming,” Journal of Social
Two-Step Approach,” Psychological Bulletin, 103 (3), 411–23. Psychology, 142 (2), 249–63.
Baron, Reuben M. and David A. Kenny (1986), “Moderator– Kandra, Anne (2000), “Shipping Charges: Who Foots the Bill?”
Mediator Variables Distinction in Social Psychological PC World, 18 (9), 39.
Research: Conceptual, Strategic, and Statistical Considera- Kelley, Scott W., K. Douglas Hoffman, and Mark A. Davis (1993),
tions,” Journal of Personality and Social Psychology, 51 (6), “A Typology of Retail Failures and Recoveries,” Journal of
1173–82. Retailing, 69 (4), 429–52.
Bassili, John N. and John P. Racine (1990), “On the Process Rela- Krull, Douglas S. (2001), “On Partitioning the Fundamental Attri-
tionship Between Person and Situation Judgments in Attribu- bution Error: Dispositionalism and Correspondence Bias,” in
tion,” Journal of Personality and Social Psychology, 59 (5), Cognitive Social Psychology: The Princeton Symposium on the
881–90. Legacy and Future of Social Psychology, Gordon B.
Berger, Joseph, Thomas L. Conner, and M. Hamit Fisek (1974), Moskowitz, ed. Hillsdale, NJ: Lawrence Erlbaum Associates,
Expectation States Theory: A Theoretical Research Program. 211–27.
Cambridge, MA: Winthrop Publishers. Landman, Janet (1987), “Regret and Elation Following Action and
Chatterjee, Patrali (2007), “Advertising Versus Unexpected Next Inaction,” Personality and Social Psychology Bulletin, 13
Purchase Coupons: Consumer Satisfaction, Perceptions of (December), 524–36.
Value, and Fairness,” Journal of Product & Brand Manage- Lapidus, Richard S. and Lori Pinkerton (1995), “Customer Com-
ment, 16 (1), 59–69. plaint Situations: An Equity Theory Perspective,” Psychology
Cooke, Alan D.J., Tom Meyvis, and Alan Schwartz (2001), & Marketing, 12 (2), 105–122.
“Avoiding Future Regret in Purchase-Timing Decisions,” Jour-
Lemon, Katherine N., Tiffany Barnett White, and Russell S. Winer
nal of Consumer Research, 27 (March), 447–59.
(2002), “Dynamic Customer Relationship Management: Incor-
Davis, Scott, Michael Hagerty, and Eitan Gerstner (1998), “Return
porating Future Consideration into the Service Retention Deci-
Policies and the Optimal Level of ‘Hassle,’” Journal of Eco-
sion,” Journal of Marketing, 66 (January), 11–14.
nomics and Business, 50 (5), 445–60.
Maxham, James G., III, and Richard G. Netemeyer (2003), “Firms
Fitzsimmons, Gavan J. (2008), “Death to Dichotomizing,” Journal
Reap What They Sow: The Effects of Shared Values and Per-
of Consumer Research, 35 (1), 5–8.
ceived Organizational Justice on Customer Evaluations of
Folkes, Valerie S. (1984), “Consumer Reactions to Product Fail-
Complaint Handling,” Journal of Marketing, 67 (January),
ure: An Attributional Approach,” Journal of Consumer
46–62.
Research, 10 (March), 398–409.
Fornell, Claes and David F. Larcker (1981), “Evaluating Structural Meyer, Harvey (1999), “Many Happy Returns,” Journal of Busi-
Equation Models with Unobservable Variables and Measure- ness Strategy, 20 (4), 27–31.
ment Errors,” Journal of Marketing Research, 18 (February), Miller, Frederick D., Eliot R. Smith, and James Uleman (1981),
39–50. “Measurement and Interpretation of Situational and Disposi-
Gilly, Mary C. and Betsy D. Gelb (1982), “Post-Purchase Con- tional Attributes,” Journal of Experimental Social Psychology,
sumer Processes and the Complaining Consumer,” Journal of 17 (1), 80–95.
Consumer Research, 9 (December), 323–28. Mollenkopf, Diana A., Elliot Rabinovich, Timothy M. Laseter, and
Greenleaf, Eric A. (2004), “Reserves, Regret, and Rejoicing in Kenneth K. Boyer (2007), “Managing Internet Product
Open English Auctions,” Journal of Consumer Research, 31 Returns: A Focus on Effective Service Operations,” Decision
(September), 264–73. Sciences, 38 (2), 215–50.
Hess, James D., Wujin Chu, and Eitan Gerstner, (1996), “Control- Neff, James A. (1985), “Race and Vulnerability to Stress: An
ling Product Returns in Direct Marketing,” Marketing Letters, Examination of Differential Vulnerability,” Journal of Person-
7 (4), 307–317. ality and Social Psychology, 49 (2), 481–91.
Homans, George Caspar (1961), Social Behavior: Its Elementary Nisbett, Richard E. and Lee Ross (1980), Human Inference:
Forms. New York: Harcourt, Brace & World. Strategies and Shortcomings of Social Judgment. Englewood
Inman, J. Jeffrey, James S. Dyer, and Jianmin Jia (1997), “A Gen- Cliffs, NJ: Prentice-Hall.
eralized Utility Model of Disappointment and Regret Effects Oliver, Richard L. (1997), Satisfaction: A Behavioral Perspective
on Post-Choice Valuation,” Marketing Science, 16 (2), 97–111. on the Consumer. Boston: Richard D. Irwin/McGraw-Hill.
——— and Leigh McAlister (1994), “Do Coupon Expiration Dates ——— and John E. Swan (1989a), “Consumer Perceptions of
Affect Consumer Behavior?” Journal of Marketing Research, Interpersonal Equity and Satisfaction in Transactions: A Field
31 (August), 423–28. Survey Approach,” Journal of Marketing, 53 (April), 21–35.

Return Shipping Policies of Online Retailers / 123

 ——— and ——— (1989b), “Equity and Disconfirmation Percep- Taylor, Shelley E. and Judith H. Koivumaki (1976), “The Percep-
tions as Influences on Merchant and Product Satisfaction,” tion of Self and Others: Acquaintanceship, Affect, and Actor–
Journal of Consumer Research, 16 (December), 372–83. Observer Differences,” Journal of Personality and Social Psy-
O’Shaughnessy, John and Nicholas Jackson O’Shaughnessy chology, 33 (4), 403–408.
(2005), “Considerations of Equity in Marketing and Nozick’s Tsiros, Michael and Vikas Mittal (2000), “Regret: A Model of Its
Decision-Value Model,” Academy of Marketing Science Antecedents and Consequences in Consumer Decision Mak-
Review, 10, (published electronically), [available at http:// ing,” Journal of Consumer Research, 26 (March), 401–417.
[Link]/articles/[Link]]. Valle, Valerie and Melanie Wallendorf (1977), “Consumers’ Attri-
Padmanabhan, V. and I.P.L. Png (1997), “Manufacturer’s Returns butions of the Cause of Their Product Satisfaction and Dissat-
Policies and Retail Competition,” Marketing Science, 16 (1), isfaction,” in Consumer Satisfaction, Dissatisfaction and Com-
81–94. plaining Behavior, Ralph L. Day, ed. Bloomington: Indiana
Pasternack, Barry Alan (2008), “Optimal Pricing and Return Poli- University, 26–30.
cies for Perishable Commodities,” Marketing Science, 27 (1), Verhoef, Peter C., Philip Hans Franses, and Janny C. Hoekstra
133–40. (2001), “The Impact of Satisfaction and Payment Equity on
Petersen, J. Andrew and V. Kumar (2009), “Are Product Returns a Cross-Buying: A Dynamic Model for a Multi-Service Provider,”
Necessary Evil? Antecedents and Consequences,” Journal of Journal of Retailing, 77 (3), 359–78.
Marketing, 73 (May), 35–51. Vorhees, Clay M., Michael K. Brady, and David M. Horowitz
Raphel, Murray (2004), “And Many Happy Returns,” Art Business (2006), “A Voice from the Silent Masses: An Exploratory and
News, 31 (1), 46. Comparative Analysis of Noncomplainers,” Journal of the
Ratchford, Brian (1987), “New Insights About the FCB Grid,” Academy of Marketing Science, 34 (4), 514–27.
Journal of Advertising Research, 27 (4), 24–39. Walster, Elaine, Ellen Berscheid, and G. William Walster (1973),
Simonson, Itamar (1992), “The Influence of Anticipating Regret “New Directions in Equity Research,” Journal of Personality
and Responsibility on Purchase Decisions,” Journal of Con- and Social Psychology, 25 (February), 151–76.
sumer Behavior, 19 (June), 105–118. Weiner, Bernard (2000), “Attributional Thoughts About Consumer
Smith, Amy K., Ruth N. Bolton, and Janet Wagner (1999), “A Behavior,” Journal of Consumer Research, 27 (December),
Model of Customer Satisfaction with Service Encounters 382–87.
Involving Failure and Recovery,” Journal of Marketing White, Peter (1991), “Ambiguity in the Internal/External Distinc-
Research, 36 (August), 356–73. tion in Causal Attribution,” Journal of Experimental Social
Solomon, Sheldon (1978), “Measuring Dispositional and Situa- Psychology, 27 (3), 259–70.
tional Attributions,” Personality and Social Psychology Bul- Wolf, Alan (2012), “CE Returns Cost Industry $17B in 2011:
letin, 4 (4), 589–94. Report,” Twice, 27 (1), 82.
Spencer, Jane (2003), “‘I Ordered That?’ Web Retailers Make It Wood, Stacy L. (2001), “Remote Purchase Environments: The
Easier to Return Goods,” The Wall Street Journal, (September Influence of Return Policy Leniency on Two-Stage Decision
4), D1–D2. Processes,” Journal of Marketing Research, 38 (May), 157–69.
Swan, John E. and Richard L. Oliver (1991), “An Applied Analy- Xia, Lan, Kent B. Monroe, and Jennifer L. Cox (2004), “The Price
sis of Buyer Equity Perceptions and Satisfaction with Automo- Is Unfair! A Conceptual Framework of Price Fairness Percep-
bile Salespeople,” Journal of Personal Selling & Sales Man- tions,” Journal of Marketing, 68 (October), 1–15.
agement, 11 (2), 15–26. Zeelenberg, Marcel, Wilco W. van Dijk, and Antony S.R.
Tax, Stephen S., Stephen W. Brown, and Murali Chandrashekaran Manstead (1998), “Reconsidering the Relation Between Regret
(1998), “Customer Evaluations of Service Complaint Experi- and Responsibility,” Organizational Behavior and Human
ences: Implications for Relationship Marketing,” Journal of Decision Processes, 74 (3), 254–72.
Marketing, 62 (April), 60–76. Zimmerman, Ann and Dana Mattioli (2011), “Ship Free or Lose
Out: More Retailers Absorb Cost of Sending Packages to Vie
with Web-Only Rivals,” The Wall Street Journal, (November
25), B1.

124 / Journal of Marketing, September 2012

View publication stats