How Does Enforcement Deter Gray Market Incidence?

Abstract

Gray market activity has become increasingly prevalent. The prevailing wisdom in marketing is to use more severe enforcement to deter gray marketing. However, the certainty and speed of enforcement may also have a bearing on the incidence of violations. This article examines whether and how enforcement deters gray marketing. The results from a field survey of manufacturers and an experimental design suggest that, by itself, enforcement severity has no impact. Deterrence results only when the multiple facets of enforcement are used in combination.

Full text

92
Journal of Marketing
Vol. 70 (January 2006), 92–106
©2006, American Marketing Association
ISSN: 0022-2429 (print), 1547-7185 (electronic)
Kersi D. Antia, Mark E. Bergen, Shantanu Dutta, & Robert J. Fisher
How Does Enforcement Deter Gray
Market Incidence?
Gray market activity has become increasingly prevalent. The prevailing wisdom in marketing is to use more severe
enforcement to deter gray marketing. However, the certainty and speed of enforcement may also have a bearing
on the incidence of violations. This article examines whether and how enforcement deters gray marketing. The
results from a field survey of manufacturers and an experimental design suggest that, by itself, enforcement sever-
ity has no impact. Deterrence results only when the multiple facets of enforcement are used in combination.
Kersi D. Antia is an assistant professor (e-mail: kantia@ivey.uwo.ca), and
Robert J. Fisher is R.A. Barford Professor of Communications (e-mail:
rfisher@ivey.uwo.ca), Richard Ivey School of Business, University of
Western Ontario. Mark E. Bergen is Carlson Professor of Marketing,
Department of Marketing and Logistics Management, Carlson School of
Management, University of Minnesota (e-mail: mbergen@csom.umn.
edu). Shantanu Dutta is Tappan Professor of Business-to-Business Mar-
keting, Department of Marketing, Marshall School of Business, University
of Southern California (e-mail: sdutta@marshall.usc.edu). This research
was supported by a Social Sciences and Humanities Research Council of
Canada grant (No. 410-2002-0605) at the University of Western Ontario;
additional funding was provided by the Zumberge Foundation, University
of Southern California, and the John M. Olin School of Business, Wash-
ington University, St. Louis. The authors gratefully acknowledge the assis-
tance of the Beauty and Barber Supply Institute in the data collection
efforts and the comments on previous drafts of this article by Rajesh
Chandy, John Hulland, Jaideep Prabhu, and Sourav Ray.
In today’s competitive markets, manufacturers increas-
ingly rely on their authorized distributors to perform
value-adding functions (Coughlan et al. 2001). To safe-
guard distributor incentives to perform these services,
manufacturers typically deploy resale restrictions through
explicit contracts or implicit agreements. By circumscribing
to whom distributors may sell, resale restrictions limit intra-
brand competition and maintain distributor margins. Gray
market activity—that is, the sale of genuine trademarked
products through distribution channels unauthorized by the
manufacturer or brand owner—poses a direct and signifi-
cant threat to manufacturers’ deployed resale restrictions.
Given free reign, gray markets create free-riding problems
across distributors that provide customer service, make a
selective distribution system more intensive, and harm dis-
tributors that have made specific investments in the channel
of distribution. Gray marketing is problematic for manufac-
turers because it can have a negative impact on distributor
relations and the manufacturer’s brand equity, ultimately
undermining the integrity of the distribution channel
(Corey, Cespedes, and Rangan 1989; Myers and Griffith
1999).
Gray markets are endemic across, though not limited to,
a wide variety of industries, ranging from heavy construc-
tion equipment, personal computers, and cellular phones to
perfumes, watches, personal care, and other consumer prod-
ucts. Estimates of gray market activity range from $10 bil-
lion in economy-wide annual gray market sales (Cespedes,
Corey, and Rangan 1988) to $20 billion in the information
technology sector alone (estimate by the Alliance for Gray
Market and Counterfeit Abatement; see www.agmaglobal.
org). Surveys conducted by Michael (1998) and Myers
(1999) confirm the increasing incidence and scope of gray
market activity.
Most firms are keenly aware of the costs of allowing
gray market activity to go unchecked (Bucklin 1993;
Hwang 1999; Nelson 1999) and allocate significant
resources to limiting the incidence of violations. In general,
the prevailing advice from the academic literature is that
increasing enforcement severity (i.e., raising the magnitude
of the punitive response undertaken) lowers the incidence of
gray market activity. For example, the enforcement litera-
ture in marketing (Banerji 1990; Dutta, Bergen, and John
1994) exclusively focuses on severity as the dimension of
enforcement that firms should manage. Such a focus
reflects the implicit assumption that the critical manage-
ment variable for manufacturers seeking effective gray mar-
ket deterrence is “the severity of the principal’s disciplinary
response to an agent’s violation of a contractual obligation”
(Antia and Frazier 2001, p. 67).
Although such advice may seem intuitively appealing
and finds favor with those urging manufacturers to “get
down to business” against gray market activity (Nelson
1999), to the best of our knowledge, there is no evidence of
the efficacy of severe enforcement in the context of gray
markets. A well-established literature spanning economics,
law, sociology, and social psychology, which has come to
be known as the deterrence doctrine, proposes that severe
enforcement alone may not be enough to curb the incidence
of violations. Instead, the doctrine advocates a broader
notion of enforcement, comprising (in addition to severity)
the certainty and speed of the response (Howe and Brandau
1988; Manson 2001; Posner 1985).
Understanding how these dimensions of enforcement
work together is essential to managing gray market inci-
dence more effectively. Yet to the best of our knowledge, no
attempt has been made to incorporate the possible role of
severity, certainty, and speed of enforcement together in
curbing gray market incidence; severity remains the sole

How Does Enforcement Deter Gray Market Incidence? / 93
focus of any discussion of enforcement, both in the trade
press (Hwang 1999; Nelson 1999) and in the relevant mar-
keting literature (Antia and Frazier 2001; Bergen, Heide,
and Dutta 1998). Accordingly, current enforcement-related
advice to managers is limited.
In this article, we attempt to broaden our conceptualiza-
tion of enforcement and explore the incidence of gray mar-
ket violations as a function of the severity, certainty, and
speed of the manufacturer’s general enforcement behavior.
In the interests of appropriate model specification, we also
incorporate several context-specific determinants of inci-
dence gained from multiple in-depth interviews and a
review of the rich literature on gray markets (Ahmadi and
Yang 2000; Cavusgil and Sikora 1988; Corey, Cespedes,
and Rangan 1989; Coughlan and Soberman 1998).
We first test our hypotheses with data collected from a
field survey of 104 manufacturers in the personal care prod-
ucts industry. The difficulty of obtaining such data is well
acknowledged, given the sensitive nature of the phenome-
non and its implications (Cavusgil and Sikora 1988; Myers
1999). Ours is the first study to bring evidence to bear on
the relationship between enforcement and gray market inci-
dence. We then complement the insights gained from the
field study with an experiment that we designed to (1) pro-
vide the perspective of the target rather than the source of
enforcement and (2) manipulate the facets of enforcement
in a controlled setting. Together, both studies yield rich
insights into the impact of enforcement on deterrence.
In the sections that follow, we introduce the deterrence
doctrine as the lens through which we view the relationship
between gray market incidence and enforcement. We then
discuss hypotheses that link gray market incidence to the
three facets of enforcement. We follow this with descrip-
tions of the research method and a discussion of the results
and implications of our studies. We conclude with the limi-
tations of our study and provide avenues for further
research.
Conceptual Framework
Gray Market Incidence, Enforcement, and the
Deterrence Doctrine
The notion of deterrence has its roots in the seminal works
of Bentham (1962) and Beccaria (1963) and refers to the
preventive effect of actual or threatened punishment on
potential offenders (Ball 1955). In the current context, the
aim of deterrence is to minimize or preclude violations of
deployed resale restrictions not only by the perpetrator but
also by other potential offenders. Thus, we relate the inci-
dence of gray market violations to the manufacturer’s
enforcement behavior in general rather than the punitive
response to a particular violation (cf. Antia and Frazier
2001; Bergen, Heide, and Dutta 1998).
Within a gray market context, enforcement refers to the
means by which the manufacturer ensures distributor com-
pliance with resale restrictions, including fines, litigation,
social ostracism, and termination. The manufacturer’s abil-
ity to enforce stems from the legitimate authority bestowed
by explicit or implicit agreements with its intermediaries
(see French and Raven 1959), but this is not the only way to
deter gray market activities or achieve dealer compliance.
In addition to enforcement, manufacturers may also with-
hold ongoing benefits, such as promotional discounts,
advertising support, and training, from noncompliant down-
stream intermediaries. Such influence attempts are related
to, though distinct from, enforcement as we define it.
Severity is fundamental to the notion of enforcement
and can be defined as the strength or intensity of corrective
actions across detected violations (Gibbs 1975). The costs
imposed on the violating party are a direct function of the
severity of the punishment that is received. The magnitude
of a fine affects the offending distributor’s net payoff from a
gray market violation (in this case, the profit captured by
the violation less the fine). Other nonfinancial costs may
also be important. The costs associated with litigation and
termination include the time spent in legal consultation and/
or in seeking alternative sources of supply. Gray market
violations are also likely to have a negative effect on the
relationship between the manufacturer and the supplier,
thus increasing monitoring and surveillance costs. Overall,
the potential gains from violating resale restrictions are
eroded by nontrivial costs, which reduces the incentive to
commit the violation (Becker 1968; Stigler 1970).
The marketing literature treats enforcement and its
severity as essentially synonymous. Dutta, Bergen, and
John (1994) and Bergen, Heide, and Dutta (1998) conceptu-
alize enforcement in terms of termination. Antia and Frazier
(2001) explicitly define enforcement in terms of severity.
Moreover, as it currently stands, the preceding literature
could be characterized as advocating (even taking for
granted) a single-minded focus on severity. Such a focus
rests on the assumption that the rigor of enforcement has a
significant inverse relationship to the incidence of viola-
tions, notwithstanding variations in certainty or speed.
Thus:
H1: The greater the severity of enforcement, the lesser is the
likelihood of gray market incidence.
Interaction with Certainty
In contrast to the single-minded focus on severity as a deter-
minant of deterrence in the marketing literature, work in
other disciplines emphasizes the need to consider enforce-
ment certainty (the likelihood that a violator will be pun-
ished) and speed (the time taken by the manufacturer to
undertake corrective action). As originally conceptualized
and specified, the deterrence doctrine emphasized the facets
of severity and certainty (Andenaes 1971; Bailey and Smith
1972; Bentham 1962; Gibbs 1968). Subsequent augmenta-
tions acknowledge the potential role of speed and incorpo-
rate its impact on deterrence (Gray et al. 1982). Mirroring
the development path of the literature, we first focus on the
potential deterrent impact of severity and certainty and then
incorporate the role of speed in our hypotheses.
Research in sociology conceptualizes deterrence as a
joint function of severity and certainty (Antunes and Hunt
1973; Gray and Martin 1969). The severity of enforcement
is posited to have significant deterrent value but only when
it is matched with a high likelihood of action being taken.

94 / Journal of Marketing, January 2006
The notion of matching facets of enforcement implies the
need for high levels of each facet of enforcement in order to
achieve deterrence.
By certainty, we mean the manufacturer’s general
propensity to undertake enforcement in response to viola-
tions (Gibbs 1975). To take corrective action, the manufac-
turer must be able to obtain information about violations
once they occur (Dutta, Heide, and Bergen 1999) and have
the motivation to enforce (Antia and Frazier 2001). This is
our point of departure from the deterrence doctrine, as rep-
resented in the sociology literature. In the latter case, the
motivation to eradicate or at least minimize crime is a
given. In contrast, manufacturers that detect gray market
violations may undertake selective enforcement for a vari-
ety of reasons (Banerji 1990; Bergen, Heide, and Dutta
1998; Coughlan and Soberman 1998). In the current con-
text, therefore, certainty of enforcement is a function of the
manufacturer’s ability and motivation. The former is
reflected in the manufacturer’s detection ability because
enforcement hinges on first knowing that a violation has
occurred (Antia and Frazier 2001).
Unfortunately, gray markets raise three well-known
problems with regard to inferring motivation. First, in con-
trast to sociology, the certainty of enforcement cannot be
estimated conveniently from archival information on
reported crime statistics and conviction rates (Gibbs 1975),
because no such information exists for gray market viola-
tions. In the absence of archival information on the inci-
dence of violations and manufacturers’ enforcement
responses, it is necessary to rely on microlevel, survey-
based data to gain insights into this highly sensitive issue.
Second, manufacturers do not know of all instances of vio-
lations (Wathne and Heide 2000), because detection is
costly (Stigler 1970) and imperfect (Dutta, Heide, and
Bergen 1999). Third, and perhaps equally important, no
manufacturer would admit to selective enforcement, thus
introducing social desirability bias (Fisher 1993).
Recognizing this potential for bias, survey-based studies
on enforcement have not solicited informant evaluations of
their own likelihood of taking action. Instead, motivation is
inferred from the reported enforcement response to particu-
lar instances of violations (for recent examples of this
approach, see Antia and Frazier 2001; Bergen, Heide, and
Dutta 1998). Moreover, previous studies have hypothesized
and found evidence of a direct and strong relationship
between detection ability and likelihood of enforcement
(Antia and Frazier 2001; Dutta, Bergen, and John 1994;
Dutta, Heide, and Bergen 1999; Ghosh and John 1999).
Given the preceding difficulty and in light of the strong
relationship between the two constructs, we represent cer-
tainty in terms of detection ability. Our approach is consis-
tent with and builds on the existing marketing literature on
enforcement.
The rationale for matched and high levels of severity
and certainty suggests that when perpetrators decide
whether to engage in gray market activities, they consider
not only the previously discussed costs imposed by severe
enforcement but also the likelihood of severe enforcement.
If detection ability and, consequently, the certainty of an
enforcement response are low, potential miscreants are
likely to discount the costs imposed by severe enforcement
(Luckenbill 1982). Severity poses no threat to the offender
if the offence remains undetected by the manufacturer.
Similar discounting of negative outcomes occurs if, for
example, high detection capabilities are accompanied by
less severe enforcement behavior. In this case, potential vio-
lators may reason that though the likelihood of apprehen-
sion is high (as a result of well-honed detection capabili-
ties), their offenses are likely to incur nothing more serious
than a “slap on the wrist” (Stafford et al. 1982). Proponents
of this perspective argue that the credibility of sanctions
hinges on both detection ability and severity (Gray et al.
1982; Hollinger and Clark 1983). The interaction perspec-
tive is also known as the “credibility of severe sanctions
hypothesis” (Grasmick and McLaughlin 1978) and suggests
the following:
H2: Severe enforcement deters gray market incidence when
the certainty of enforcement is high.
The Augmented Deterrence Doctrine
Early empirical work on deterrence (Silberman 1976)
focused exclusively on the certainty and severity of enforce-
ment. Only in later studies (Gray et al. 1982) did scholars
acknowledge and incorporate a third, additional dimension
of enforcement: speed. The speed of enforcement refers to
the time elapsed between the detection of violations and the
corrective actions taken in response. Similar to severity and
detection ability, our notion of speed is the time taken to
respond to detected violations.
Although the relationship between enforcement speed
and deterrence is not as well specified as it is for detection
ability and severity, our interdisciplinary literature review
yields some insights into the effects of variation in enforce-
ment speed. First, swift (slow) corrective action reduces
(increases) the length of time over which the party commit-
ting the violation may enjoy its payoff (Tedeschi 1976).
Second, delay in applying sanctions allows the violating
party “room to maneuver,” that is, to undertake actions to
avoid (or at least minimize) bearing the brunt of the costs
imposed by corrective action (Hufbauer, Schott, and Elliott
1990). Third, research in psychology (Diver-Stamnes and
Thomas 1995) and law (Manson 2001) suggests that tempo-
ral proximity of the violation and the consequent corrective
enforcement action reinforce the punitive consequences to
the perpetrator, as well as the group at large.
In contrast to the well-documented interaction between
severity and detection ability, there has been little, if any,
attention focused on explicating the interactions involving
speed and severity (for a notable exception, see Gray et al.
1982). The logic underlying this interaction mirrors that
which we proposed in H2, that the deterrent impact of
enforcement is deemed to reside in high levels of speed
combined with severity. It is plausible that severe enforce-
ment may do little to deter violations if the enforcement
response is inordinately delayed. Because a slow response
allows the perpetrator time to enjoy a longer payoff from
the violation, gives the perpetrator room to maneuver to
avoid bearing the full brunt of enforcement, and/or weakens
the causal link between violation and enforcement, delayed

How Does Enforcement Deter Gray Market Incidence? / 95
enforcement could dampen the deterrent impact of severity.
By extension, severity should have a significant effect on
deterrence when both the likelihood of detection and the
speed of enforcement are high. An enforcement strategy
that involves high levels of all three components serves to
emphasize the manufacturer’s commitment to maintaining
system integrity by presenting a unified, unambiguous
deterrent. Thus:
H3: Severe enforcement deters gray market incidence when
the speed of enforcement is high.
H4: Severe enforcement deters gray market incidence when
both detection ability and the speed of enforcement are
high.
Context-Specific Determinants of Gray Market
Incidence
Scholars of deterrence agree that any examination of deter-
rence must account for contextual variables that have a
bearing on the incidence of violations, independent of
enforcement (Gibbs 1975; Meier and Johnson 1977). Our
review of the academic literature on gray markets yielded
five factors that emerged most frequently: price differen-
tials, premium brand positioning, product scarcity, free-
riding potential, and customer heterogeneity on services
demanded.
According to Onkvisit and Shaw (1989, p. 205), “price
differential is the only true reason for the gray market to
exist.” In attempting to price optimally for local conditions
(Bucklin 1993; Duhan and Sheffet 1988), manufacturers
facilitate arbitrage, whereby the diverter may source prod-
uct from the low-priced market and sell it in the high-priced
market without the manufacturer’s authorization (Assmus
and Wiese 1995). A related driver of gray market violations
is the extent to which the product has a unique and differen-
tiated brand name (Bucklin 1993). The ability to obtain a
product with brand name cachet at significantly lower
prices is a powerful motivator of consumer purchase (Ass-
mus and Wiese 1995), which in turn drives the occurrence
of gray marketing. Gray market violations are also likely to
be high when the authorized distributors in a market are
unable to satisfy demand for the product. Pent-up demand
for a product creates the incentive necessary for arbitrage to
occur, and goods are diverted from markets in which the
product is readily available (and usually lower priced) to the
markets in which the product is in short supply (Banerji
1990).
The potential for some distributors or unauthorized third
parties to free ride off full-service distributors’ market
development efforts (Mathewson and Winter 1984; Rubin
1990; Telser 1960) also facilitates the development of a
gray market (Cross, Stephan, and Benjamin 1990).
Although services such as advertising, pre- and postsales
service, and warranty servicing (Coughlan et al. 2001) no
doubt add value to the product, the revenue streams accru-
ing to authorized distributors from their marketing efforts
are vulnerable to appropriation (Anderson and Weitz 1986).
Customer heterogeneity on services also is likely to foster
gray markets. Some customers may value and even require
services, whereas others may not be willing to pay for per-
ceived “frills.” Enterprising diverters would be quick to pro-
vide the actual product with little or no augmentation to
those unwilling to pay the high price for the “bundled”
product or service offering, thus realizing incremental
(albeit unauthorized) sales to an additional, hitherto ignored
customer segment (Ahmadi and Yang 2000; Yang, Ahmadi,
and Monroe 1998). The preceding discussion leads to the
following hypothesis:
H5: The greater the (a) price differential, (b) premium posi-
tioning, (c) product scarcity, (d) free-riding potential, and
(e) customer heterogeneity on services in a market, the
greater is the likelihood of gray market incidence.
Research Method
Study 1: Field Survey of Gray Marketing of
Personal Care Products
Our sampling frame comprised U.S.-based manufacturers
of branded personal care products (e.g., skin toners, hair
and nail care products, medicated soap) that distribute
through wholesaler distributors in the professional salon
industry. The reliance on external (non-company-owned)
distribution intermediaries and the high value-to-weight
ratio of the products in this study make this a particularly
appropriate context in which to test our hypotheses.
According to the Beauty and Barber Supply Institute, the
association representing personal care product manufac-
turers and distributors, more than half of its members had
experienced gray marketing of their products in some form
or other. Thus, the phenomenon of gray markets is well
known to most industry participants.
We defined the term “gray marketing” for informants,
and then we asked them to focus on a geographic market in
which they had experienced, or were likely to experience,
gray marketing. After they identified such a market, we
asked them to report their enforcement behavior (i.e., sever-
ity, certainty, and speed) in general rather than in response
to a particular violation. We did this to assess deterrence
attributable to a general propensity for enforcement rather
than a particular response. Consistent with previous
channel-related research in marketing (Anderson, Hakans-
son, and Johanson 1994; Anderson and Weitz 1992; Antia
and Frazier 2001; Cavusgil and Sikora 1988; Dutta, Heide,
and Bergen 1999; Dwyer, Schurr, and Oh 1987), we
solicited manufacturers’ evaluations of their own severity,
detection ability, and speed of enforcement behavior. We
developed and refined the survey instrument on the basis of
a literature review, interviews with channel management
personnel from the personal care products industry, and a
mailed pretest.
Measures. Our dependent variable of interest, incidence
(INCID), is a dichotomy that describes whether gray mar-
keting occurred during the previous two years. We coded
INCID as 1 if the informants indicated that they were aware
of diversion of any product from their product line within
the specified time period and 0 if otherwise.
We measured the independent variables with multi-item
summated scales. The final item sets, response formats,
individual item loadings, composite reliability, and average

96 / Journal of Marketing, January 2006
variance explained (AVE) for each scale appear in Appen-
dix A. The preceding metrics provide evidence that the psy-
chometric properties of our measures are satisfactory. Table
1displays the correlation matrix and descriptive statistics
for the variable set. In the following sections, we describe
the measurement approach for each variable.
The manufacturer’s enforcement severity (SEV) con-
sists of a four-item Likert scale. We derived the items that
constitute the scale from the work of Antia and Frazier
(2001), but we modified them to ground the measure in our
research context. Detection ability (DETECT) reflects the
extent to which the manufacturer is able to evaluate the
extent and nature of instances of product diversion. We
derived the scale for this construct from the work of Dutta,
Heide, and Bergen (1999) and, again, adapted it to our
research context. The final scale is composed of four
reverse-coded Likert items. Enforcement speed (SPEED)
refers to the time elapsed between detection of the violation
and the manufacturer’s undertaking of corrective action. We
developed the four specific items that constitute the scale
using prior research in sociology and social psychology
(Erickson, Gibbs, and Jensen 1977; Howe and Brandau
1988; Howe, Brandau, and Loftus 1996) and modified them
on the basis of field interviews.
Price differential (PRDIFF) refers to the extent to which
the manufacturer charges different prices for the product
across markets. Our scale for this variable consists of a
three-item Likert measure. Premium positioning (PREM)
describes the extent to which the manufacturer positions the
specific product as a high-end “premium” brand. On the
basis of our literature search and field interviews, we devel-
oped a four-item Likert scale for this variable. Product
scarcity (SCARCE) reflects the manufacturer’s inability to
supply the relevant market with enough product to satisfy
demand. We measured the extent of product scarcity with a
four-item Likert scale.
Potential for free riding (FR) reflects the extent to
which the manufacturer believes that other authorized dis-
tributors or unauthorized third parties can gain from the
sales efforts of its authorized distributors in that market.
During the interviews we conducted, it became clear that
we needed to capture the risk of both types of free riding.
This corresponds to a second-order confirmatory factor
model, in which the observed six items (three items reflect-
ing each first-order factor) are hypothesized to originate
from the two first-order factors (AUTHFR and UNAU-
THFR), and in turn, the first-order factors originate from a
second-order factor.
Customer heterogeneity (HETER) reflects the extent to
which the specific market is characterized by distinct cus-
tomer segments, in which each segment values services dif-
ferently. We measured this variable with a three-item Likert
scale, which we developed and refined on the basis of
pretests.
Data collection. The Beauty and Barber Supply Institute
agreed to send out a letter on our behalf that explained the
purpose of the study and informed each of its members to
expect further communication from us. A week later, we
contacted each firm by telephone to introduce ourselves,
ascertain the willingness to participate, and identify suitably
qualified key informants. The modal designation of the
informants was general manager; sales and channels man-
agers were also well represented. Of the 581 firms called,
we were unable to contact 21, 26 expressed their unwilling-
ness or inability to participate, and another 41 had open
lines or used company-owned distribution channels, leaving
us with a final sample of 493.
We then sent out a survey packet by mail to the identi-
fied key informants at the remaining 493 manufacturer
firms. We dropped an additional 20 informants from the
study because they indicated that the study did not apply to
their current business context. Two further rounds of call-
backs and two subsequent mailings yielded a total of 104
usable questionnaires, representing a 22% response rate. Of
the 104 responding firms, 40 reported experiencing at least
one incidence of gray marketing, and 64 reported otherwise.
Nonsignificant differences between early and late respon-
dents (Armstrong and Overton 1977) on the focal variables
of the study suggest that nonresponse bias is unlikely. Post
hoc checks indicate acceptable levels of involvement (M =
3.92 on a five-point scale; SD = 1.37) and knowledge (M =
4.34; SD = .97) with respect to their firms’ dealings in the
specific product-market.
Measure validation. Given our sample size and the
number of constructs, we conducted confirmatory factor
analyses (CFAs) on groups of maximally similar constructs
(see Moorman and Miner 1997), namely, the enforcement
characteristics (severity, speed, and detection ability) and
the contextual variables (price differences, customer hetero-
geneity, premium positioning, and supply shortage). The
higher-order construct conceptualization of free-riding
potential necessitated the estimation of 14 additional para-
meters, thus requiring us to conduct a separate CFA for this
construct to meet sample size requirements. The CFA
model diagnostics (see Table 2) suggest unidimensionality
of the reflective scales. All items loaded on their prespeci-
fied constructs and had t-values significant at .05, providing
evidence of convergent validity. Discriminant validity of the
scales is further supported by the Lagrange-multiplier tests;
none of the possible cross-loadings exceeds the critical
value of the chi-square with one degree of freedom (Speier
and Venkatesh 2002).
Given the significant correlation between enforcement
severity and its speed, we undertook a rigorous test of the
discriminant validity of these constructs in particular. We
asked three academics with expertise in the area of enforce-
ment to sort the eight items that purported to measure sever-
ity and speed. All three experts were able to perform the
item-sorting task with full accuracy. Interitem correlations
among the items measuring severity (average ρseverity = .68,
p< .0001) and speed (average ρspeed = .73, p< .0001) are at
a significantly higher level than cross-construct item corre-
lations (average ρspeed,severity = .53, p< .0001); t-tests of dif-
ferences further suggest that the differences in correlations
are significant. We also found that the AVE for each con-
struct was greater than the squared correlation between the
constructs (Fornell and Larcker 1981), thus providing fur-
ther evidence of discriminant validity between severity and
speed of enforcement. In all, the preceding tests suggest

How Does Enforcement Deter Gray Market Incidence? / 97
TABLE 1
Study 1: Descriptive Statistics
Correlation Matrix
12345678910111213
1. Incidence (INCID) —
2. Enforcement severity (SEV) 0.08*.89
3. Enforcement speed (SPEED) 0.07*.68* .92
4. Detection ability (DETECT) –.14*.28* .29* .85
5. Price differential (PRDIFF) 0.08*–.07 –.15 –.09 .79
6. Free-riding potential (FR) 0.38* .15 .11 –.31* –.11 —
7. Product scarcity (SCARCE) 0.08*–.04 –.24* –.01 –.02 00.08*.87
8. Premium positioning (PREM) 0.15*.13 .11 –.00 –.01 00.26* –.02 .89
9. Customer heterogeneity (HETER) 0.26* .26* .20* –.02 –.06 00.22* –.10 .12 .77
10. SEV ×DETECT –.28* –.01 –.07 .23* –.02 0–.28* –.02 –.05 –.04 —
11. SEV ×SPEED –.10*–.33* –.47* –.05 –.09 0–.14*.09 .07 –.09 .31* —
12. SPEED ×DETECT –.19*–.07 –.10 .31* –.06 0–.23* .01 .08 –.05 .80* 0.27* —
13. SEV ×SPEED ×DETECT –.12*.38* .36* .59* –.06 0–.14*.09 .11 –.11 –.03*–.12*–.07 —
Number of items 1.00* 444306.00* 443————
Minimum 0.00* 444306.00* 443————
Maximum 1.00* 20 20 20 14 30.00* 20 20 15 ————
M0.38*13.46 13.57 9.67 5.78 20.45*7.66 14.27 9.34 .68*0.28*0.29 –.07
SD 0.49*4.27 4.31 4.14 2.63 05.02*3.37 4.24 2.7 1.29*1.02*1.02 1.54
*r is significant at p< .05.
Notes: Diagonal entries represent alpha reliability values, where applicable.

98 / Journal of Marketing, January 2006
TABLE 2
Study 1: CFA Results for Reflective Measures
Average
Off-Diagonal Compara- Incremental Tucker– Lowest
Squared tive Fit Fit Lewis t-Value of
Construct Groups χ2pValue Residual Index Index Index Loading
SEV, SPEED, DETECT χ2
51 = 90.03 .001 .07 .96 .96 .94 5.81
FR χ2
8= 16.61 .030.10 .98 .98 .96 7.36
PRDIFF, SCARCE,
PREMIUM, HETER χ2
71 = 89.9 .070.05 .97 .97 .96 5.26
1Main effects in this case represent the conditional effects of
each enforcement characteristic on incidence at moderate (mean)
that severity and speed of enforcement have adequate dis-
criminant validity.
Model specification. We test our hypotheses using a
maximum likelihood estimation–based logistic regression
model, first specifying the main effects only (Equation 1)
and then adding the two- and three-way interactions among
the enforcement characteristics (Equation 2) as follows:
where INCIDi = 1 if firm i reports at least one gray market
violation and 0 if otherwise, and
Xi1 = SEV,
Xi2 = DETECT,
Xi3 = SPEED,
Xi4 = PRDIFF,
Xi5 = FR,
Xi6= SCARCE,
Xi7= PREM,
Xi8= HETER,
Xi9 = SEV ×DETECT,
Xi10 = SEV ×SPEED,
Xi11 = SPEED ×DETECT, and
Xi12 = SEV ×SPEED ×DETECT.
To reduce the potential multicollinearity arising from
multiplicative interaction terms (Aiken and West 1991) and
to yield coefficient estimates within the observable range of
the independent variables (Friedrich 1982; Jaccard, Turrisi,
and Wan1990), we mean centered all the explanatory varia-
bles and interactions created as the product of the mean-
centered components (see Brown, Dev, and Lee 2000; Jac-
card, Turrisi, and Wan 1990; Rokkan, Heide, and Wathne
2003).1
(1) P(INCID 1)
exp X
exp
i
0jij
==
+∑








+
∑
ββ
1
8
1βββ0jij
i
X
and
P(INCID 1)
ex
+∑








==
∑
1
8
2
,
()
ppX
exp X
0jij
0jij
ββ
ββ
+∑








++∑


∑
∑
1
12
1
12
1





,
levels of the other two facets. This interpretation is distinct from
the “constant effect of one variable over all values of another
variable” (Aiken and West 1991, p. 38) interpretation, as is the
case with uncentered (raw) data. However, both approaches are
consistent in that estimates of the latter may be recovered as a spe-
cial case of the mean-centered approach, if required (Freidrich
1982).
2Table 1 shows evidence of considerable multicolinearity
between the interaction variables and their components; such mul-
ticolinearity is known to inflate the standard errors of the param-
eter estimates, though the estimates themselves remain unbiased
(Jaccard, Turrisi, and Wan 1990). The statistical significance of the
higher-order interactions in the presence of the lower-order terms
indicates that the imprecision due to multicolinearity does not
pose a threat to validity (Buvik and John 2000). Parameter esti-
mates and significance levels remain robust across both model
specifications (the main effects–only and the fully specified inter-
action model), thus lending further confidence in our findings.
Table 3 displays the estimation results for both equa-
tions. The likelihood ratio for the main effects vector of
coefficients is not significantly different from zero (χ2(3) =
1.58, p= .66). In contrast, the likelihood ratio is signifi-
cantly different from zero for the two-way (χ2(3) = 6.69, p=
.08) and three-way (χ2(1) = 2.82, p= .09) interactions. Fur-
thermore, the interaction model (χ2= 35.71, p= .0004)
gives us confidence that the null hypothesis of all the coef-
ficients being zero can be rejected. The model correctly
classifies 80% of the observations, suggesting that the
inclusion of the interaction terms is warranted. Therefore,
we discuss the findings of this study as inferred from the
estimation of the interaction model.2
Study 1 results. None of the three enforcement charac-
teristics (severity, detection ability, and speed) has a signifi-
cant impact on deterrence (incidence) in isolation (b1= .11,
b2= –.04, b4= .01). However, our examination of the full
model coefficients requires us to separate the main effects
from the interaction terms (for a full explanation, see Aiken
and West 1991, p. 38; for a recent application in a channels
context, see Brown, Dev, and Lee 2000). We do this by dif-
ferentiating Equation 2 with respect to each of the three
enforcement characteristics. Equation 3 shows how INCID
varies as a function of SEV (for derivation and details on
obtaining simple slopes, standard errors, and t-values, see
Appendix B; for the results, see Table 4).
()323 5 7
∂
∂=+ + +
×
INCID
SEV bbDETECT b SPEED b SPEED
DEETECT.

How Does Enforcement Deter Gray Market Incidence? / 99
TABLE 3
Study 1: Parameter Estimates
Full Model
Independent Variable
Constant
DETECT (b1)
SEV (b2)
SEV ×DETECT (b3)
SPEED (b4)
SEV ×SPEED (b5)
SPEED ×DETECT (b6)
SEV ×SPEED ×DETECT (b7)
PRDIFF (b8)
FR (b9)
PREM (b10)
SCARCE (b11)
HETER (b12)
Coefficient
–.661
–.012
–.042
0.065
0.147
0.190
0.035
0.094
0.201
t-Value
–2.73***
0–.19***
0–.53***
00.81***
01.60***
03.01***
00.58***
01.31***
02.08***
Coefficient
(Interactions
Included)
–.519
0.109
–.041
–.055
0.010
–.009
–.008
–.007
0.145
0.206
0.056
0.121
0.194
t-Value
–1.8**0*
01.20***
0–.49***
–2.04***
00.11***
0–.59***
0–.30***
–1.68***
01.51***
02.89***
00.84***
01.45***
01.82***
Hypothesis
Supported?a
H1(No)
H2(Yes)
H3(No)
H4(Yes)
H5a (Yes)
H5b (Yes)
H5c (No)
H5d (Yes)
H5e (Yes)
χ2
8= 25.44; p= .0013 χ2
12 = 35.71; p= .0004
*p< .10 (one-tailed test).
**p< .05 (one-tailed test).
***p< .01 (one-tailed test).
aBased on interaction model estimates.
Main Effects Only
TABLE 4
Study 1: Impact of Enforcement Characteristics on Incidence
Enforcement Characteristic
Main Effects
SEV
DETECT
SPEED
Interaction Effects
Impact of SEV on INCID at various levels of DETECT (H2)
DETECTlow
DETECTmoderate
DETECThigh
Impact of SEV on INCID at various levels of SPEED (H3)
SPEEDlow
SPEEDmoderate
SPEEDhigh
Impact of SEV on Incidence at various levels of DETECT
and SPEED (H4)
DETECThigh and SPEEDhigh
DETECThigh and SPEEDlow
DETECTlow and SPEEDhigh
DETECTlow and SPEEDlow
Estimated Impact
on Incidence
(Simple Slope)
0.098
0.246
–.046
0.187
–.041
–.269
–.028
0.010
0.048
–.430
–.110
0.270
0.100
t-Value
00.670*
01.610*
0–.360*
01.350*
0–.490*
–1.96*0
0–.224*
00.110*
00.556*
–2.85*0
0–.700*
01.800*
00.650*
SE
.145
.152
.127
.137
.084
.137
.127
.091
.087
.157
.157
.157
.157
*Significant at the .05 level (two-tailed test).
The t-value of .098 is not significant at the .05 level.
Therefore, we are unable to reject the null hypothesis for
SEV, which in turn enables us to reject the main effect we
hypothesized in H1. We evaluate the main effects of
DETECT (∂INCID/∂DETECT = .246)and SPEED
(∂INCID/∂SPEED = –.046)similarly (for estimates, see
Table 4); in each case, the main effect is not significantly
different from zero. We now proceed to examine the inter-
action coefficients.
We find that just one two-way interaction is statistically
significant: SEV ×DETECT (b3= –.055, p< .05). On fur-
ther probing (see Table 4), we find that strict enforcement
behavior has a significant inverse relationship to gray mar-
ket incidence when it is combined with high detection abil-

100 / Journal of Marketing, January 2006
ity (∂INCID/∂SEV = –.269, p< .05). Note that no such
deterrent impact occurs under low (∂INCID/∂SEV = .187,
p> .10) or even moderate (∂INCID/∂SEV = –.041, p> .10)
levels of detection ability. The preceding result underscores
the importance of matched and high levels of severity and
detection ability, in support of the interaction we hypothe-
sized in H2. We fail to find support for the exploratory
hypotheses with respect to SPEED’s pairwise interaction
with SEV (b5= –.009). That is, SPEED does not seem to
confer any deterrent impact in combination with SEV or
with DETECT.
In Table 3, we find that the three-way interaction among
SEV, SPEED, and DETECT is statistically significant (b7=
–.007, p< .05), in support of H4, suggesting that the combi-
nation of all three facets of enforcement has a deterrent
impact. Post hoc probing of this significant interaction
reveals that increasing SEV results in deterrence only when
SPEED and DETECT both have high levels (the simple
slope of SEV on INCID is negative and significant). As can
be seen in Table 4, if either SPEED or DETECT is low, the
simple slope of SEV on INCID is not significant.
The significant and inverse relationship between SEV ×
DETECT and INCID suggests that the combination of high
levels of SEV and DETECT is sufficient to exert a deterrent
impact on gray market activity. Furthermore, the significant
three-way interaction suggests that when SPEED is added
to the combination of SEV and DETECT, deterrence is
enhanced. Our inability to find similar effects for the pair-
wise combinations involving SPEED or for a main effect of
SPEED suggests that speed must be accompanied by severe
action and high detection ability to achieve incremental
deterrence. Swift enforcement provides an additional deter-
rent “boost,” but not by itself.
Our hypothesis with respect to the gray market–specific
determinants of incidence finds significant support for four
of the five contextual variables. The presence of price dif-
ferentials among markets has a significant, though mar-
ginal, effect on gray market incidence (b8= .145, p< .10).
Free-riding potential has a strong, significant, direct rela-
tionship to gray market activity (b9= .206, p< .01). We do
not find any support for the hypothesized relationship
between premium positioning and gray market incidence
(b10 = .056). However, we find that product scarcity is posi-
tively related to the likelihood of gray market violations,
though marginally so (b11 = .121, p< .10). We also find a
strong positive relationship between customer heterogeneity
on services and gray market incidence (b12 = .194, p< .05).
To summarize, our examination of the main and interac-
tions effects suggests that none of the characteristics of
enforcement has any significant deterrent impact in isola-
tion. Matched and high levels of detection ability and sever-
ity are prerequisites for achieving deterrence, and swift
enforcement boosts the preceding interaction.
Study 2: Experiment
We conducted Study 2 to replicate and extend the results of
Study 1 with an alternative methodology, context, and mea-
sures. First, we conducted Study 2 as an experiment
because the dimensions of enforcement severity, detection
ability, and speed are likely to be naturally associated in
field contexts. Manufacturers that are concerned with gray
market activities tend to design detection systems that
enable them to identify problems accurately and then react
both quickly and severely against the offending firm. In
contrast, manufacturers that are not affected by gray market
activities are likely to be relatively lax on all three enforce-
ment facets. Given this logic, some combinations of
enforcement strategies are unlikely to occur naturally in
field settings, such as high severity but low speed and cer-
tainty of detection or low severity and low speed but high
certainty. The coincident use of these enforcement charac-
teristics is evident in our first study with significant correla-
tions between severity and speed (r = .68, p< .05), severity
and detection (r = .28, p< .05), and detection and speed (r =
.29, p< .05). Therefore, Study 2 uses an experimental
design to examine the full range of possible enforcement
strategies and to use orthogonal manipulations that enable
us to assess the independent effects of severity, detection,
and speed on deterrence.
Second, Study 2 assesses the robustness of the results of
Study 1 by examining deterrence from the perspective of
the dealer rather than the manufacturer. Given that it is deal-
ers’ evaluations of the effectiveness of the enforcement
facets that affect their willingness to engage in gray market
activities, it is important to understand their perceptions.
Study 2 also differs from Study 1 in its use of a continuous
rather than dichotomous measure of deterrence, thus
enabling us to estimate the psychometric properties of the
dependent variable.
Procedure. Participants were 112 MBA students who
were recruited by student representatives of an on-campus
exchange program at a major university. A donation of $10
was made toward the program on behalf of each participant
who completed the study. Sixty-four percent of the partici-
pants were male, their average age was 29 years, and they
had an average of three years of management experience.
Seventy-seven percent of participants had at least one year
of experience as “a supplier or dealing with suppliers in a
business situation.”
After a brief introduction to a study on manufacturer–
distributor relationships, participants were randomly
assigned to one condition in a 2 (severity: high versus
low) ×2 (certainty: high versus low) ×2 (speed: fast versus
slow) full-factorial between-subjects design. Participants
read a scenario that explained that an electronics compo-
nents dealer had violated its agreement with a manufacturer
by selling $100,000 of integrated circuits outside its terri-
tory. The scenario included manipulations of the severity,
certainty, and speed of the manufacturer’s response. The
method assumes that participants project themselves into
the hypothetical situation and provide answers that reflect
how dealers would actually respond to the situation outlined
in the scenario. Evidence suggests that projective methods
can accurately represent participants’ underlying attitudes
and behaviors (e.g., Fisher 1993) and that the judgments of
individual managers can provide important insights into
organization-level strategies (e.g., Chandy, Prabhu, and
Antia 2003). The scenario and manipulations appear in
Appendix C. Participants then responded to a series of
questions.

How Does Enforcement Deter Gray Market Incidence? / 101
The dependent variable was a three-item measure that
reflected the degree to which the offending dealer would be
deterred from engaging in future gray market activities by
the manufacturer’s enforcement activities. The items
included the following: “In future, the offending dealer
would definitely comply with the manufacturer’s regula-
tions governing unauthorized sales,” “In future, the offend-
ing dealer would not dare [to] disregard the resale restric-
tions of the manufacturer,” and “In future, the offending
dealer would be deterred from selling the manufacturer’s
products in an unauthorized manner.” The items were mea-
sured on a five-point Likert scale, anchored by “very
unlikely” and “very likely.” The mean of the summated
scale items was 2.74, with a standard deviation of .97 and a
coefficient alpha of .86.
Analysis. We used a full-factorial analysis of variance to
examine the manipulation checks and to test the hypothe-
ses. We found a significant effect of the detection manipula-
tion on the item “The manufacturer is very likely to catch
dealers who undertake gray market activities” (F(1, 111) =
131.51, p< .01), a significant effect of the severity manipu-
lation on the item “The manufacturer’s response to gray
market transactions is severe” (F(1, 111) = 76.10, p< .01),
and a significant effect of the speed manipulation on the
item “The manufacturer punishes gray market activity
immediately” (F(1, 111) = 122.77, p< .01). The manipula-
tions had no unintended main or interaction effects.
In terms of the hypotheses, we found significant main
effects of certainty (F(1, 111) = 5.20, p< .05) and severity
(F(1, 111) = 22.81, p< .01) on deterrence. In terms of inter-
actions, the two-way interaction between certainty and
severity was significant (F(1, 111) = 5.23, p< .05), as was
the three-way interaction effect (F(1, 109) = 4.17, p< .05).
Examination of the simple effects reveals that severity has a
positive effect on deterrence in three of four possible
instances: high certainty/high speed (F(1, 27) = 10.32;
Mlow severity = 2.43 > Mhigh severity = 3.29, p< .01), low cer-
tainty/high speed (F(1, 27) = 3.48; Mlow severity = 2.38 >
Mhigh severity = 3.15, p< .10), and high certainty/low speed
(F(1, 27) = 19.97; Mlow severity = 2.31 > Mhigh severity = 3.62,
p< .001). As we predicted, severity has no effect on deter-
rence when both certainty and speed are low (p> .90). Note
that when certainty and severity are high, speed has no
effect on deterrence (p> .30). We included participants’
gender and years of supervisory experience as covariates,
but they were not significant (p> .20). The analysis of vari-
ance interaction results appear in Figure 1.
Study 2 results. The results are consistent with those of
Study 1 in three of four possible effects of severity on deter-
rence. Consistent with Study 1, the results of Study 2 sup-
port the view that severity of response alone is insufficient
to reduce gray market activities. Increasing the severity of
the punishment for gray market violations had no effect
when dealer violations were unlikely to be detected and
when violators could expect a long period of time before
they would be punished. It seems logical that the threat of
even the most severe punishment will be ineffective if it has
a low likelihood of imposition and if it will only take place
in the distant future.
Increasing the severity of the manufacturer’s response to
gray market activity was effective when it was used in con-
cert with systems that ensured a high likelihood of detec-
tion, immediate responses to violations, or both. Consistent
with Study 1, participants indicated that future gray market
activities would be reduced when the likelihood of detec-
tion was high, regardless of whether enforcement speed was
fast or slow. The results support the view that deterrence is
a function of both the severity and the likelihood of punish-
FIGURE 1
Study 2: Certainty ×Speed ×Severity Interaction Effect
Low High
Severity
Low High
Severity
Slow speed
Fast speed
Deterrence Level
3.20
3.00
2.80
2.60
2.40
2.20
Deterrence Level
3.75
3.50
3.25
3.00
2.75
2.50
2.25
A: Low Certainty of Detection B: High Certainty of Detection
Slow speed
Fast speed

102 / Journal of Marketing, January 2006
ment. In contrast to the Study 1, Study 2 found that severity
had a deterrence effect when speed was high but the likeli-
hood of detection was low. By implication, speed appears to
compensate for detection certainty in some conditions.
Despite a low probability of detection and, thus, punish-
ment, participants believed that dealers would be less likely
to reoffend if severe punishments were meted out quickly.
Discussion
The results of the two studies are consistent in their support
of the deterrence doctrine, despite varying dramatically on
various dimensions, including methodology (field survey
versus experiment), perspective (manufacturer versus dis-
tributor), and measurement of the dependent variable
(dichotomous measure of actual incidence versus multi-
item measure of future recidivism). Both studies indicate
that the severity, certainty, and speed of enforcement inter-
act in their effects on deterrence. The results have important
implications for both research and practice in the area.
Research Implications
Prior research in marketing has implicitly taken the rela-
tionship between enforcement severity and deterrence to be
an article of faith. So well entrenched is this assumption
that there has been no attempt to test the efficacy of
enforcement in a marketing context. Our central finding is
that, by itself, enforcement severity does not deter gray
market incidence. As a result, an expanded conceptualiza-
tion of enforcement is necessary. Our study also sheds light
on how the facets of enforcement work together to mini-
mize gray market violations. Across both studies, none of
the three characteristics of enforcement (i.e., severity, cer-
tainty, and speed) has deterrent value alone. Rather, deter-
rence is most likely to occur when the penalties for gray
market violations are severe, when manufacturers are able
to detect violations or mete out punishments in a timely
fashion, or both.
The introduction of speed of response further extends
knowledge of the subtleties of enforcement. Specifically,
we find that swift enforcement bolsters the deterrent impact
of high levels of detection ability and severity. The ability to
increase the effectiveness of the latter combination is partic-
ularly interesting considering that, by itself, enforcement
speed has little to contribute to deterrence objectives. This
finding is of great importance to the marketing literature on
enforcement; it emphasizes the dual imperative for an
expanded conceptualization of enforcement, and it forces us
to examine the deterrent impact of enforcement rather than
take it for granted as research has done to date.
In examining the deterrent impact of enforcement, this
study also links the evolving marketing literature on oppor-
tunism (Brown, Dev, and Lee 2000; Dahlstrom and
Nygaard 1999; John 1984; Wathne and Heide 2000) and
enforcement (Antia and Frazier 2001; Bergen, Heide, and
Dutta 1998; Dutta, Bergen, and John 1994). The preceding
literature bases have tended to develop along distinct paths,
despite the obvious linkages. To the best of our knowledge,
ours is the first study to test the assumed deterrent efficacy
of enforcement. Our findings set the stage for a synthesis of
these still disparate research streams.
Our choice of empirical context facilitates an additional
important contribution. To date, the literature on gray mar-
kets has tended to develop along two lines: analytical mod-
els (Ahmadi and Yang 2000; Banerji 1990; Bucklin 1993;
Coughlan and Soberman 1998) and descriptive case studies
(Antia and Everatt 2000; Cespedes, Corey, and Rangan
1988). By their very nature and design, the preceding
efforts emphasize deeper understanding of a select few
issues relevant to gray markets. What is needed is an inte-
gration and subsequent large-sample validation of the com-
monly suggested drivers of gray market activity. We address
this crucial gap in the understanding of gray markets by
developing an integrative model of gray market incidence
and by validating this model with both field data and an
experiment. Our study represents an attempt to “take stock”
of received thought on gray market incidence.
Managerial Implications
Until now, the vocabulary and emphasis of the business and
academic press have been limited to enforcement severity.
Yet our results suggest that current prescriptions calling for
stricter measures against gray market participants are likely
to be ineffective. Managers who want to make progress
against gray market incursions must invest in systems that
increase the ability to both detect and quickly punish viola-
tors. It is only when severe enforcement behavior is com-
bined with certainty and/or speed that gray market inci-
dence is significantly curbed.
Our work points out the need to use an integrated
approach to the management of distribution channel rela-
tions. In addition to devoting attention to enforcement,
manufacturers should provide adequate incentives to autho-
rized distributors for the continued provision of valued ser-
vices in the right mix to the right targeted segment. Due
consideration should also be given to pricing and supply
issues, in acknowledgment of their potential impact on gray
market activity. The combination of all the preceding mea-
sures is more likely to lead to desired deterrence objectives.
Limitations and Future Research
Directions
Although the two studies we reported herein provide several
contributions to academics and managers alike, much work
remains to be done. The term “gray marketing” encapsu-
lates a variety of unauthorized sales, including cross-border
parallel importation, diversion, and domestic bootlegging,
to name but a few. Although cognizant of these subtle dis-
tinctions, we were unable to control for the location and
specific type of gray market activity because of nonre-
sponse on the particular items in our survey. We have taken
initial steps toward understanding the degree to which the
deterrence doctrine is generalizable by testing it from both
the manufacturer’s (Study 1) and the dealer’s (Study 2) per-
spective, but a variety of other industry and violation con-
texts remain unexamined.

How Does Enforcement Deter Gray Market Incidence? / 103
Our empirical context imposed certain constraints on
our construct operationalizations. First, the absence of
archival data on gray market incidence and the steps taken
by manufacturers in response forced us to restate our con-
ceptual framework in terms of detection ability rather than
the notion of certainty, as originally conceptualized by the
deterrence doctrine. Second, in Study 1, we measured gray
market incidence as a binary response in an attempt to
strike a balance between examining gray markets in a
meaningful manner and minimizing the potential for nonre-
sponse attributable to what could be perceived as excessive
probing of a “hot-button” channel issue. In Study 2, we
were able to replicate the findings of Study 1 using a con-
tinuous measure of deterrence in an experimental context.
We consider our studies a stimulus for more sophisticated
attempts to capture the intricacies of channel violations and
manufacturers’ subsequent responses.
Consistent with a well-established body of channels
research, we asked manufacturer informants to report their
own behavior. In Study 2, we considered the dealer’s per-
spective and found similar results. It would be useful for
research to incorporate the simultaneous views of both
manufacturers and dealers to understand this issue further.
Enforcement Severity (Composite Reliability = .90;
AVE = .69)
We have a policy of full enforcement of our sales
agreements. .75
We are well known for our strict policy on product
diversion. .75
We have a tough stance on product diversion. .95
We (would) take strict action against unauthorized
sales. .85
Enforcement Speed (Composite Reliability = .92;
AVE = .73)
Our response to violations is (would be)
instantaneous. .88
We (would) take immediate action against violations. .83
Very little time (would) elapse between detection of
violations and our response to them. .87
Our enforcement response process is (would be)
very timely. .85
Detection Ability (Composite Reliability = .87;
AVE = .63)
At a given time, it would be difficult to evaluate the
extent of product diversion. (R) .82
Determining compliance with resale restraints
requires a great amount of effort on our part. (R) .56
It would be difficult for us to evaluate the extent to
which our product is diverted. (R) .95
Our evaluation of the extent of unauthorized sales is
based on very “fuzzy” information. (R) .78
Price Differential (Composite Reliability = .80;
AVE = .58)
We maintain a uniform pricing policy between
markets. (R) .88
In general, we try to keep price differences between
markets to a minimum. (R) .80
The price we charge for the product in each market
varies considerably. .57
Free-Riding Potential
Unauthorized Free Riding (Composite Reliability = .85;
AVE = .65)
Customers may learn about our product from our
distributors and purchase it from unauthorized
sources. .91
Our distributors’ presales services might stimulate
unauthorized sales of the product. .80
Unauthorized distributors could benefit from the
market development efforts of our authorized
distributors. .70
Authorized Distributor Free Riding (Composite
Reliability = .92; AVE = .80)
Our authorized distributors’ sales efforts increase
sales of other authorized distributors of this
product. .84
The services provided by one authorized
distributor may help other authorized
distributors’ sales. .92
Authorized distributors could benefit from the
market development efforts of other authorized
distributors. .92
Product Scarcity (Composite Reliability = .88;
AVE = .64)
We have trouble keeping up with demand for this
product. .61
Order fulfillment for this product is frequently
delayed due to production constraints. .82
We frequently experience stock-outs of this product. .93
We do not produce enough product to satisfy
demand. .82
Premium Positioning (Composite Reliability = .89;
AVE = .67)
Our brand name commands a significant price
premium in this market. .74
Our company is considered a market leader in this
product category. .86
We are able to leverage our strong brand name
to a great extent in this market. .87
Our products have a strong reputation in this
market. .80
Customer Heterogeneity (Composite Reliability =
.78; AVE = .54)
Customers in this market require very different
service levels. .61
Our customers have very dissimilar product
preferences. .71
Customers in this market differ considerably in
their preference for service. .60
APPENDIX A
Measures
Notes: All items are measured on a five-point Likert scale, anchored by “strongly agree” and “strongly disagree.” (R) = reverse-scored items.

104 / Journal of Marketing, January 2006
Appendix B
Post Hoc Probing of Interactions
Two-Way Interaction Between SEV and DETECT
The objective of this post hoc probing is to ascertain the
impact of severity on incidence for different levels of detec-
tion ability. Accordingly, we rearrange the terms of Equa-
tion 2 to obtain the simple regression of INCID on SEV at
plus and minus one standard deviation of DETECT, respec-
tively (Aiken and West 1991, p. 13). This yields two simple
slope coefficients, indicating the impact of SEV on INCID
at each level of DETECT.
From Table 3, Equation 2,
INCID = –.519 – .041SEV + .109DETECT – .055SEV
×DETECT.
If we rearrange the terms,
INCID = (–.041 – .055DETECT)SEV + (–.519 + .109DETECT).
At DETECTL(DETECT = –4.1447),
INCID = (–.041 – .055 ×–4.1447)SEV
+ (–.519 + .109 ×–4.1447),
or
INCID = .187SEV – .971.
We obtain the simple slope coefficients similarly for
each two-way interaction (see Table 4). Using the variance–
covariance matrix of the beta estimators, we then obtain the
standard error and t-value for each simple slope.
Three-Way Interaction Among SEV, SPEED, and
DETECT
We rearrange the terms of Equation 2 yet again, this time
replacing SPEED and DETECT with values plus and minus
one standard deviation to represent high and low values of
SPEED and DETECT, respectively.
From Table 2, Equation 2,
INCID = –.519 + –.041SEV + .010SPEED + .109DETECT
– .055SEV ×DETECT – .009SEV ×SPEED
– .008SPEED ×DETECT
– .007SEV ×SPEED ×DETECT.
If we rearrange the terms,
INCID = (–.041 – .055DETECT – .009SPEED
– .007SPEED ×DETECT)SEV
+ (–.519 + .010SPEED + .1095DETECT
– .008SPEED ×DETECT).
We then obtain the simple slopes of SEV for the four
combinations of SPEED and DETECT:
At SPEEDHand DETECTH(SPEED = 4.3051, DETECT =
4.1447), INCID = –.355SEV – .167;
At SPEEDLand DETECTH(SPEED = –4.3051, DETECT =
4.1447), INCID = –.183SEV + .032;
At SPEEDHand DETECTL(SPEED = 4.3051, DETECT =
–4.1447), INCID = .351SEV – .785; and
At SPEEDLand DETECTL(SPEED = –4.3051, DETECT =
–4.1447), INCID = .023SEV – 1.157.
Using the variance–covariance matrix of the beta esti-
mators, we obtain the standard error and t-value for each
simple slope (see Table 4).
Appendix C
Study 2 Scenario and Experimental
Manipulations
Avon is a company that manufactures integrated chips (ICs)
for use in laptops and personal computers. Avon dealers are
expressly prohibited from selling their ICs outside their ter-
ritory (called gray market activity). Nevertheless, one of
Avon’s dealers recently sold a large order of Avon ICs
(approximately $100,000 in sales) outside its authorized
territory to get rid of excess inventory. Avon has the follow-
ing procedures and policies related to gray market activities
by its dealers:
Detection Manipulation
High: The dealer has a high probability of being caught.
Avon is able to identify unauthorized sales over 90%
of the time because of its sophisticated inventory
management system.
Low: The dealer has a low probability of being caught.
Avon is able to identify unauthorized sales only 10%
of the time because of an unsophisticated inventory
management system.
Severity Manipulation
High: The penalties are significant. If Avon identifies an
unauthorized sale, it levies a fine equal to twice the
value of the sale (in this case $200,000). If caught,
dealers lose a significant amount of money on the
gray market transaction.
Low: The penalties are not very significant. If it identifies
an unauthorized sale, it levies a fine equal to 15% of
the contract’s value. If caught, dealers still make a
small profit on the gray market transaction.
Speed Manipulation
Fast: Avon’s inventory management system is able to detect
gray market activities immediately. After a violation is
identified, the dealer is notified that [it is] being fined
within 24 hours. Consequently, dealers are notified
and punished before they make any additional gray
market sales.
Slow: Avon requires a dealer audit before it can confirm a
suspected gray market activity, so contract violations
are detected six months after the fact. After a viola-
tion is identified, the dealer is notified that [it is]
being fined within 24 hours. Consequently, many
dealers are notified and punished after they have
made additional gray market sales.

How Does Enforcement Deter Gray Market Incidence? / 105
REFERENCES
Ahmadi, Reza H. and B. Rachel Yang (2000), “Parallel Imports:
Challenges from Unauthorized Distribution Channels,” Mar-
keting Science, 19 (Summer), 279–94.
Aiken, Leona and Stephen West (1991), Multiple Regression: Test-
ing and Interpreting Interactions. Newbury Park, CA: Sage
Publications.
Andenaes, Johannes (1971), “The Moral or Educative Influence of
Criminal Law,” Journal of Social Issues, 27 (2), 17–31.
Anderson, Erin and Barton A. Weitz (1986), “Make-or-Buy Deci-
sions: Vertical Integration and Marketing Productivity,” Sloan
Management Review, 27 (Spring), 3–19.
——— and ——— (1992), “The Use of Pledges to Build and
Sustain Commitment in Distribution Channels,” Journal of
Marketing Research, 24 (February), 18–34.
Anderson, James, Hakan Hakansson, and Jan Johanson (1994),
“Dyadic Business Relationships Within a Business Network
Context,” Journal of Marketing, 58 (October), 1–16.
Antia, Kersi D. and Donna Everatt (2000), Looks.com(A): A Grey
Issue. London, Ontario: Ivey Publishing.
——— and Gary L. Frazier (2001), “The Severity of Contract
Enforcement in Interfirm Channel Relationships,” Journal of
Marketing, 65 (October), 67–81.
Antunes, George and A. Lee Hunt (1973), “The Impact of Cer-
tainty and Severity of Punishment on Levels of Crime in Amer-
ican States: An Extended Analysis,” Journal of Criminal Law
and Criminology, 64 (4), 486–93.
Armstrong, Scott J. and Terry S. Overton (1977), “Estimating
Nonresponse Bias in Mail Surveys,” Journal of Marketing
Research, 14 (August), 396–402.
Assmus, Gert and Carsten Wiese (1995), “How to Address the
Gray Market Threat Using Price Coordination,” Sloan Man-
agement Review, 36 (Spring), 31–41.
Bailey, William C. and Ronald W. Smith (1972), “Punishment: Its
Severity and Certainty,” Journal of Criminal Law, Criminology,
and Police Science, 63 (4), 530–39.
Ball, John C. (1955), “The Deterrence Concept in Criminology
and Law,” Journal of Criminal Law, Criminology, and Political
Science, 46 (September–October), 347–54.
Banerji, Shumeet (1990), A Theory of Gray Markets: The Case of
the Personal Computer Industry. Evanston, IL: Northwestern
University.
Beccaria, Cesare (1963), On Crimes and Punishments. Indianapo-
lis: Bobbs-Merrill.
Becker, Gary S. (1968), “Crime and Punishment: An Economic
Approach,” Journal of Political Economy, 76 (March–April),
169–217.
Bentham, Jeremy (1962), The Works of Jeremy Bentham. New
York: Russell and Russell.
Bergen, Mark, Jan Heide, and Shantanu Dutta (1998), “Managing
Grey Markets Through Tolerance of Violations,” Managerial
and Decision Economics, 19 (May), 157–66.
Brown, James R., Chekitan S. Dev, and Dong-Jin Lee (2000),
“Managing Marketing Channel Opportunism: The Efficacy of
Alternative Governance Mechanisms,” Journal of Marketing,
64 (April), 51–65.
Bucklin, Louis P. (1993), “Modeling the International Gray Mar-
ket for Public Policy Decisions,” International Journal of
Research in Marketing, 10 (4), 387–405.
Buvik, Arnt and George John (2000), “When Does Vertical Coor-
dination Improve Industrial Purchasing Relationships?” Jour-
nal of Marketing, 64 (October), 52–64.
Cavusgil, S. Tamer and E. Sikora (1988), “How Multinationals
Can Cope with Gray Market Imports,” Columbia Journal of
World Business, 23 (4), 75–85.
Cespedes, Frank V., Raymond E. Corey, and Kasturi V. Rangan
(1988), “Gray Markets: Causes and Cures,” Harvard Business
Review, 66 (4), 75–83.
Chandy, Rajesh K., Jaideep C. Prabhu, and Kersi D. Antia (2003),
“What Will the Future Bring? Technology Expectations, Dom-
inance, and Radical Product Innovation,” Journal of Marketing,
67 (July), 1–18.
Corey, Raymond E., Frank V. Cespedes, and Kasturi V. Rangan
(1989), Going To Market. Boston: Harvard Business School
Press.
Coughlan, Ann T., Erin Anderson, Louis W. Stern, and Adel I. El-
Ansary (2001), Marketing Channels. Upper Saddle River, NJ:
Prentice Hall.
——— and David A. Soberman (1998), “When Is the Best Ship a
Leaky One? Segmentation, Competition, and Gray Markets,”
working paper, INSEAD.
Cross, James, James Stephan, and Robert E. Benjamin (1990),
“Gray Markets: A Legal Review and Public Perspectives,”
Journal of Public Policy & Marketing, 9, 183–94.
Dahlstrom, Robert and Arne Nygaard (1999), “An Empirical
Investigation of Ex Post Transaction Costs in Franchised Dis-
tribution Channels,” Journal of Marketing Research, 36 (May),
160–70.
Diver-Stamnes, Ann C. and R. Murray Thomas (1995), Prevent,
Repent, Reform, Revenge: A Study in Adolescent Moral Devel-
opment. Westport, CT: Greenwood Press.
Duhan, Dale F. and Jane Mary Sheffet (1988), “Gray Markets and
the Legal Status of Parallel Importation,” Journal of Marketing,
52 (July), 75–83.
Dutta, Shantanu, Mark Bergen, and George John (1994), “The
Governance of Exclusive Territories When Dealers Can Boot-
leg,” Marketing Science, 13 (Winter), 83–99.
———, Jan Heide, and Mark Bergen (1999), “Vertical Territorial
Restrictions and Public Policy: Theories and Industry Evi-
dence,” Journal of Marketing, 63 (October), 121–34.
Dwyer, Robert F., Paul Schurr, and Sejo Oh (1987), “Developing
Buyer–Seller Relationships,” Journal of Marketing, 51 (April),
11–27.
Erickson, Maynard L., Jack P. Gibbs, and Gary F. Jensen (1977),
“The Deterrence Doctrine and the Perceived Certainty of Legal
Punishment,” American Sociological Review, 42 (April),
305–317.
Fisher, Robert J. (1993), “Social Desirability Bias and the Validity
of Indirect Questioning,” Journal of Consumer Research, 20
(September), 303–315.
Fornell, Claes and David F. Larcker (1981), “Evaluating Structural
Equation Models with Unobservable Variables and Measure-
ment Error,” Journal of Marketing Research, 18 (February),
39–50.
Freidrich, R.J. (1982), “In Defense of Multiplicative Terms in
Regression Equations,” American Journal of Political Science,
26 (November), 797–833.
French, John R. and Bertram Raven, eds. (1959), The Bases of
Social Power. Ann Arbor: University of Michigan Institute for
Social Research.
Ghosh, Mrinal and George John (1999), “Governance Value
Analysis and Marketing Strategy,” Journal of Marketing, 63
(Special Issue), 131–45.
Gibbs, Jack P. (1968), “Crime, Punishment, and Deterrence,”
Social Science Quarterly, 48 (March), 515–30.
——— (1975), Crime, Punishment, and Deterrence. New York:
Elsevier.
Grasmick, Harold G. and Steven D. McLaughlin (1978), “Deter-
rence and Social Control,” American Sociological Review, 43
(April), 272–78.

106 / Journal of Marketing, January 2006
Gray, Louis N. and J. David Martin (1969), “Punishment and
Deterrence: Another Analysis,” Social Science Quarterly, 50
(September), 389–95.
———, Alfred C. Miranne, David A. Ward, and Ben Menke
(1982), “A Game Theoretical Analysis of the Components of
Punishments,” Social Psychology Quarterly, 45 (4), 206–212.
Hollinger, Richard C. and John P. Clark (1983), “Deterrence in the
Workplace: Perceived Certainty, Perceived Severity, and
Employee Theft,” Social Forces, 62 (2), 398–418.
Howe, Edmond S. and Cynthia J. Brandau (1988), “Addictive
Effects of Certainty, Severity, and Celerity of Punishment
Judgments of Crime Deterrence Scale Value,” Journal of
Applied Social Psychology, 18 (9), 796–812.
———, ———, and Thomas C. Loftus (1996), “Integration of
Certainty, Severity, and Celerity Information in Judged Deter-
rence Value: Further Evidence and Methodological Equiva-
lence,” Journal of Applied Social Psychology, 26 (3), 226–32.
Hufbauer, Clyde Gary, Jeffrey J. Schott, and Kimberly Ann Elliott
(1990), Economic Sanctions Reconsidered: History and Cur-
rent Policy, 2d ed. Washington, DC: Institute for International
Economics.
Hwang, Suein L. (1999), “Tobacco: As Cigarette Prices Soar, a
Gray Market Booms,” TheWall Street Journal,(January 28), B1.
Jaccard, James, Robert Turrisi, and Choi K. Wan (1990), Interac-
tion Effects in Multiple Regression. Newbury Park, CA: Sage
Publications.
John, George (1984), “An Empirical Investigation of Some
Antecedents of Opportunism in a Marketing Channel,” Journal
of Marketing Research, 21 (August), 278–89.
Luckenbill, David F. (1982), “Compliance Under Threat of Severe
Punishment,” Social Forces, 60 (3), 811–25.
Manson, Allan (2001), The Law of Sentencing. Toronto: Irwin
Law.
Mathewson, George and Ralph Winter (1984), “An Economic
Theory of Vertical Restraints,” Rand Journal of Economics, 12
(Spring), 27–38.
Meier, Robert F. and Weldon T. Johnson (1977), “Deterrence as
Social Control: The Legal and Extralegal Production of Con-
formity,” American Sociological Review, 42 (April), 292–304.
Michael, James (1998), “A Supplemental Distribution Channel?
The Case of U.S. Parallel Export Channels,” Multinational
Business Review, 6 (Spring), 24–35.
Moorman, Christine and Anne S. Miner (1997), “The Impact of
Organizational Memory on New Product Performance and Cre-
ativity,” Journal of Marketing Research, 34 (February),
91–107.
Myers, Matthew B. (1999), “Incidents of Gray Market Activity
Among U.S. Exporters: Occurrences, Characteristics and Con-
sequences,” Journal of International Business Studies, 30 (1),
105–126.
——— and David A. Griffith (1999), “Strategies for Combating
Gray Market Activity,” Business Horizons, 42 (November–
December), 2–8.
Nelson, Emily (1999), “On the Rack: Why Tommy Hilfiger Just
Can’t Seem to Stay Out of Local Wal-Mart,” The Wall Street
Journal, (January 28), A1.
Onkvisit, S. and J.J. Shaw (1989), International Marketing: Analy-
sis and Strategy. Columbus, OH: Merrill Publishing Company.
Posner, Richard (1985), “An Economic Theory of the Criminal
Law,” Columbia Law Review, 85 (October), 1193–1231.
Rokkan, Aksel I., Jan B. Heide, and Kenneth H. Wathne (2003),
“Specific Investments in Marketing Relationships: Expropria-
tion and Bonding Effects,” Journal of Marketing Research, 40
(May), 210–24.
Rubin, Paul (1990), Managing Business Transactions. New York:
The Free Press.
Silberman, Mathew (1976), “Toward a Theory of Criminal Deter-
rence,” American Sociological Review, 41 (June), 442–61.
Speier, Cheri and Viswanath Venkatesh (2002), “The Hidden
Minefields in the Adoption of Sales Force Automation Tech-
nologies,” Journal of Marketing, 66 (July), 98–111.
Stafford, Mark C., Louis N. Gray, Ben A. Menke, and David A.
Ward (1982), “Modeling the Deterrent Effects of Punishment,”
Social Psychology Quarterly, 49 (December), 338–47.
Stigler, George (1970), “The Optimal Enforcement of Laws,”
Journal of Political Economy, 78 (May–June), 526–36.
Tedeschi, James T. (1976), Social Psychology: Interdependence,
Interaction and Influence. New York: John Wiley & Sons.
Telser, Lester G. (1960), “Why Should Manufacturers Want Fair
Trade?” Journal of Law and Economics, 3 (October), 86–105.
Wathne, Kenneth H. and Jan B. Heide (2000), “Opportunism in
Interfirm Relationships: Forms, Outcomes, and Solutions,”
Journal of Marketing, 64 (October), 36–51.
Yang, B. Rachel, Reza H. Ahmadi, and Kent B. Monroe (1998),
“Pricing in Separable Channels: The Case of Parallel Imports,”
Journal of Product and Brand Management, 7 (5), 433–40.