Download Greenberg Employee Theft PDF

TitleGreenberg Employee Theft
File Size830.3 KB
Total Pages8
Document Text Contents
Page 1

Journal of Applied Psychology
1990, Vol. 75, No. 5,561-568

Copyright 1990 by the American Psychological Association, Inc.

Employee Theft as a Reaction to Underpayment Inequity:
The Hidden Cost of Pay Cuts

Jerald Greenberg
Faculty of Management and Human Resources

Ohio State University

Employee theft rates were measured in manufacturing plants during a period in which pay was
temporarily reduced by 15%. Compared with pre- or postreduction pay periods (or with control

groups whose pay was unchanged), groups whose pay was reduced had significantly higher theft

rates. When the basis for the pay cuts was thoroughly and sensitively explained to employees,

feelings of inequity were lessened, and the theft rate was reduced as well. The data support equity

theory's predictions regarding likely responses to underpayment and extend recently accumulated

evidence demonstrating the mitigating effects of adequate explanations on feelings of inequity.

Employee theft constitutes one of the most pervasive and
serious problems in the field of human resource management.
Although exact figures are difficult to come by, the American
Management Association (1977) has estimated that employee
theft cost American businesses from $5 billion to $10 billion in
1975, representing the single most expensive form of nonviolent
crime against businesses.

Traditionally, social scientists have considered several plausi-
ble explanations for employee theft. Among the most popular
are theories postulating that theft is the result of attempts to
ease financial pressure (Merton, 1938), moral laxity among a
younger workforce (Merriarn, 1977), available opportunities
(Astor, 1972), expressions of job dissatisfaction (Mangione &
Quinn, 1975), and the existence of norms tolerating theft
(Horning, 1970). More recently, Hollingerand Clark (1983) con-
ducted a large-scale survey and interview study designed to
explore these and other explanations of employee theft. Interest-
ingly, they found that the best predictor was employee attitudes:
"When employees felt exploited by the company . . . these
workers were more involved in acts against the organizations as
a mechanism to correct perceptions of inequity or injustice"
(Hollinger & Clark, 1983, p. 142).

Hollinger and Clark's (1983) suggestion that employee theft
is related to feelings of injustice is consistent with several
schools of sociological and anthropological thought. For exam-
ple, in studies of hotel dining room employees (Mars, 1973) and
maritime dock workers (Mars, 1974), Mars found that employ-
ees viewed theft not as inappropriate but "as a morally justified
addition to wages; indeed, as an entitlement due from exploit-
ing employers" (Mars, 1974, p. 224). Similarly, Kemper (1966)

A preliminary report of the research reported in this article was

presented at the annual meeting of the Academy of Management, San

Francisco, August 1990.

1 gratefully acknowledge the helpful comments of Robert J. Bies and

three anonymous reviewers on an earlier draft of this article.

Correspondence concerning this article should be addressed to Jer-

ald Greenberg, Faculty of Management and Human Resources, Ohio

State University, 1775 College Road, Columbus, Ohio 43210-1399.

argued that employee theft may be the result of "reciprocal

deviance," that is, employees' perceptions that their employers
defaulted on their obligations to them, thereby encouraging
them to respond with similar acts of deviance. Fisher and
Baron (1982) made a similar argument in presenting their eq-
uity-control model of vandalism. They claimed that vandalism
is a form of inequity reduction in that an individual vandal's
breaking the rules regarding property rights follows from his or
her feelings of mistreatment by authorities. Recent evidence in
support of this idea is found in a study by DeMore, Fisher, and

Baron (1988). In that study, university students claimed to en-
gage in more vandalism the less fairly they felt they had been
treated by their university and the less control they believed
they had over such treatment.

Such conceptualizations are in keeping with current theoreti-

cal positions in the field of organizational justice (Greenberg,
1987). These formulations allow more precise hypotheses to be
developed regarding when employee theft is likely to occur. For
example, consider equity theory's (Adams, 1965) claim that
workers who feel inequitably underpaid (i.e, those who believe
that the rewards they are receiving relative to the contributions
they are making are less than they should be) may respond by
attempting to raise their outcomes (i.e, raise the level of rewards
received). Although research has supported this claim (for a
review, see Greenberg, 1982), studies have been limited to situa-
tions in which persons paid on a piece-work basis produce

more goods of poorer quality to raise their outcomes without
effectively raising their inputs. Given earlier conceptual claims
and supporting evidence associating student vandalism with
inequitable treatment (DeMore et al, 1988), it may be reasoned
analogously that employee theft is a specific reaction to under-
payment inequity and constitutes an attempt to bring outcomes
into line with prevailing standards of fair pay.

Recent research in the area of procedural justice (Lind &
Tyler, 1988) has shown that perceptions of fair treatment and
outcomes depend not only on the relative level of one^ out-
comes but also on the explanations given for those outcomes
(for a review, see Folger & Bies, 1989). For example, researchers
have found that decision outcomes and procedures were better


Page 2


accepted when (a) people were assured that higher authorities

were sensitive to their viewpoints (Tyler, 1988), (b) the decision

was made without bias (Lind & Lissak, 1985), (c) the decision

was applied consistently (Greenberg, 1986), (d) the decision was

carefully justified on the basis of adequate information (Sha-

piro & Buttner, 1988), (e) the decisionmakers communicated

their ideas honestly (Hies, 1986), and (f) persons influenced by

the decision were treated in a courteous and civil manner (Bies

& Moag, 1986). Such findings suggest that interpersonal treat-

ment is an important determinant of reactions to potentially

unfair situations (Tyler & Bies, 1990).

It is an interesting idea that perceptions of inequity (and

corresponding attempts to redress inequities) may be reduced

when explanations meeting the criteria presented in the preced-

ing paragraph are offered to account for inequitable states. This

notion was tested in the present study by capitalizing on a

naturalistic manipulation—a temporary pay reduction for em-

ployees of selected manufacturing plants. Data were available

for 30 consecutive weeks: 10 weeks before a pay reduction oc-

curred, 10 weeks during the pay-reduction period, and 10 weeks

after normal pay was reinstated. Following from equity theory,

it was hypothesized that ratings of payment fairness would be

lower during the pay-reduction period than during periods of

normal payment (i.e, before and after the pay reduction). It was

similarly hypothesized that rates of employee theft would be

higher during the reduced-pay period than during periods of

normal payment. Such actions would be consistent with equity

theory's claim that one likely way of responding to underpay-

ment inequity is by attempting to raise the level of rewards

received. Although not previously studied in this connection,

employee theft is a plausible mechanism for redressing states of

inequity (Hollinger & Clark, 1983).

Additional hypotheses were derived from recent research

(e.g, Cropanzano & Folger, 1989; Folger & Martin, 1986; Sha-

piro & Buttner, 1988; Weiner, Amirkham, Folkes, & 'Varette,

1987) showing that explanations for negative outcomes mitigate

people's reactions to those outcomes (for a review, see Folger &

Bies, 1989; Tyler & Bies, 1990). Generally speaking, in these

studies the use of adequate explanations (i.e, ones that relied on

complete, accurate information presented in a socially sensitive

manner) tended to reduce the negative reactions that resulted

from such outcomes and facilitated acceptance of the out-

comes. From the perspective of Folger's (1986) referent cogni-

tions theory, adequate explanations help victimized parties

place their undercompensation in perspective by getting them

to understand that things could have been worse. As such, ade-

quate explanations were expected in the present study to lessen

the feelings of inequity that accompanied the pay cut. Thus, it

was reasoned that employees' feelingsof payment inequity, and

attempts to reduce that inequity (such as by pilfering), would be

reduced when adequate explanations were given to account for

the pay reduction. Specifically, it was hypothesized that the

magnitude of the expressed inequity—and the rate of employee

theft—would be lower when pay reductions were adequately

explained than when they were inadequately explained.



Participants in the study were nonunion employees working for 30
consecutive weeks in three manufacturing plants owned by the same

Table I
Distribution of Attrition and Turnover Across Conditions


Starting Missing Before During After Final

n data pay cut pay cut payout n

(Plant A)

(Plant B)

(Plant Q







1 1

1 12

1 0







parent company. The plants were located in different sections of the
midwestern United States and manufactured small mechanical parts
mostly for the aerospace and automotive industries. The employees'
average age (M =28.5 years), level of education (M =11.2 years), and
tenure with the company (M = 3.2 years) did not significantly differ

among the three plants, F < 1.00, in all cases. The local unemployment
rates in the communities surrounding the three plants were not signifi-

cantly different from each other (overall M = 6.4%), F < 1.00. It is

important to establish this equivalence of characteristics across re-
search sites because the assignment of individuals to conditions was

not random across sites, thereby precluding the assumption of equiva-

lence afforded by random assignment (Cook & Campbell, 1976).
As the study began, Plant A employed 64 workers in the following

jobs: 5 salaried low-level managerial employees (4 men, 1 woman); 47

hourly-wage semiskilled and unskilled production workers (38 men, 9
women); and 12 hourly-wage clerical workers (all women). Almost
identical proportions with respect to job type (and sex of employees
within job type) existed in Plant B (n = 53) and Plant C (n = 66).

Because some employees failed to complete questionnaires during
some weeks, and because some employees voluntarily left their jobs

during the study period, complete sets of questionnaires were available
from 55 employees of Plant A, 30 employees of Plant B, and 58 employ-
ees of Plant C This constituted a total sample of 143 employees, distrib-
uted to conditions as summarized in Table 1. The demographic charac-

teristics of the 40 workers who were not included in the study did not
differ significantly from the characteristics of the 143 who remained in
the study (in all cases, F < 2.00), minimizing the possibility that those

who remained in the study were a select group.


Because of the loss of two large manufacturing contracts, the host
company was forced to reduce its payroll by temporarily cutting wages
by 15% across the board in two of its manufacturing plants (Plants A
and B). This was done in lieu of laying off any employees. After this
decision was made, I was asked to help assess the impact of the wage

cuts in several key areas, including employee theft. Each of the pay-
ment-group manipulations was carried out in a separate plant. The
assignment of Plant A to one experimental condition and Plant B to
another experimental condition was determined at random. Assign-

ment to the control group was determined by the host company's deci-
sion that pay cuts were not necessary in Plant C.1

1 Admittedly, conducting the study in this manner meant that the
two randomly assigned groups may have been nonequivalent with re-
spect to some unknown variables that might have otherwise affected
the results (Cook & Campbell, 1976). However, some reassurance of

Page 3


The adequate explanation condition was created in Plant A. To effect

this, a meeting (lasting approximately 90 min) was called at the end of a

work week. At that meeting, all employees were told by the company

president that their pay was going to be reduced by 15%, effective the

following week, for a period expected to last 10 weeks. During this

meeting several types of explanations were provided. On the basis of

recent research (Folger & Bies, 1989; Tyler & Bies, 1990), I hypothe-

sized that these explanations would mitigate reactions to the pay cut.

The workers were told that company management seriously regretted

having to reduce their pay but that doing so would preclude the need

for any layoffs. They were further assured that all plant employees

would share in the pay cuts and that no favoritism would be shown.2 A

relevant verbatim passage follows:

Something we hate to do here at [company name] is lay off any of
our employees. But, as you probably know, we've lost our key
contracts with [company names], which will make things pretty
lean around here for a little while. As a result, we need to cut
somewhere, and we've come up with a plan that will get us through
these tough times. I've been working on h with [name of person]
in accounting, and we're sure it will work. The plan is simple:
Starting Monday; we will each get a 15% cut in pay This applies to
you, to me, to everyone who works here at [name of plant ]. If we do
h this way, there'll be no cut in benefits and no layoffs—just a 15%
pay reduction. So, either your hourly wages or your salary will be
reduced by 15%. Will it hurt? Of course! But, it will hurt us all
alike. We're all in it together. Let me just add that it really hurts me
to do this, and the decision didn't come easily. We considered all
possible avenues, but nothing was feasible. I think of you all as
family, and it hurts me to take away what you've worked so hard
for. But, for the next 10 weeks, we'll just have to tough it out.

In addition to these remarks, the basis for the decision was clearly

explained and justified by presenting charts and graphs detailing the

temporary effects of the lost contracts on cash-flow revenues. Projec-

tions verified that the cash-flow problem dictating the need for the pay

cuts was only temporary, and this was clearly explained. All employees

were assured that the pay cut was designed to last only 10 weeks.3

Specifically, the employees were told the following:

The reason I'm sharing all this information with you is that I want
you to understand what is happening here. It's just a temporary
problem we're facing, and one that I hope will never happen again.
At least the best course of action from our accounting department
is clear: The pay cuts will work, and they will not have to last
longer than 10 weeks. The new jobs we'll be picking up from
[name of company] will really help get us back on our feet. Hope-
fully, by then we'll be stronger than ever. Of course, I know we're
no stronger than our people, and I personally thank each and
every one of you for your strength.

The tone of the presentation was such that a great deal of respect was

shown for the workers, and all questions were answered with sensitiv-

ity. Approximately 1 hr was spent answering all questions. Each re-
sponse brought an expression of remorse at having to take such action

(e.g, "Again, I really wish this weren't necessary."). The good intent of

between-group similarity is provided by the demonstrated equivalence

between worker characteristics, economic conditions, and job duties

for both plants. Moreover, the deliberate assignment of Plant C to the
control condition raises the possibility that something besides the lack

of manipulation may have been responsible for the results (Cook &

Campbell, 1976). However, informal postexperiment interviews with

plant officials and employees confirmed that no unusual "local his-

tory" events occurred during the study period. Further assurance that

this was not a problem comes from the fact that, before and after the

pay cut, the control group's responses were identical to the other

groups' responses for all measures used in the study.

this message was reinforced by the fact that the president issued the
message in person.

Plant B was the site of the inadequate explanation condition. Here, a

meeting lasting approximately 15 min was called at the end of a work

week. All employees were told by a company vice president that their

pay was going to be reduced by 15%, effective the following week, for a

period expected to last 10 weeks. The only additional information that

was provided indicated that the lost contracts dictated the need for the

pay cut. No expressions of apology or remorse were shared, and the

basis for the decision was not clearly described. The following verba-

tim remarks characterize this condition:

It is inevitable in a business like ours that cost-cutting measures
are often necessary to make ends meet. Unfortunately, the time
has come for us to take such measures here at [company name]. I
know it wont be easy on anyone, but [name of company president ]
has decided that a 15% across-the-board pay cut will be instituted
effective Monday. This is largely the result of the fact that we've
lost our contracts with [name of companies]. However, soon we'll
be picking up jobs with [name of company], so we're sure the pay
cuts will last only 10 weeks. I realize this isn't easy, but such reduc-
tions are an unfortunate fact of life in the manufacturing business.
On behalf of [company president's name] and myself, we thank
you for bearing with us over these rough times. I'll answer one
or two questions, but then I have to catch a plane for another

Finally, because the parts manufactured at Plant C were unaffected

by the lost contracts, no pay cuts were mandated there. Plant C consti-
tuted the control condition for the study.


Two categories of dependent measures were used: actuarial data on

employee theft, and self-report measures tapping some of the pro-

cesses assumed to be underlying the theft behavior.

2 Before the meetings scheduled in each plant, the individuals in-

volved (i.e., company president in Plant A and a vice president in Plant

B) met with me to develop outlines of their presentations. Several care-

fully crafted sentences conveying salient aspects of the manipulation

were prepared for inclusion in the speaker's notes. Because local com-

pany norms dictated using informal meetings instead of formal presen-

tations, complete scripts for the entire sessions could not be prepared

in advance. As a result, it was necessary to establish that key differ-

ences in the manipulated variables were actually communicated in the

meetings. With this in mind, each session was videotaped, and the

videotapes were played back to a group of 112 undergraduate students

after all identifying information was deleted. The students were asked

to indicate in which of the two tapes (Tape A for Plant A; Tape B for

Plant B) the speaker (a) presented more information about the pay cuts

and (b) expressed greater remorse about the pay cuts. The order of

presentation of the tapes was randomized. Virtually all of the students

agreed that the speaker on Tape A presented more information and

expressed greater remorse. Taken together with my in-person confir-

mation that the manipulations were conducted as desired, these find-

ings suggest that differentially adequate explanations were given to the

two groups. Unfortunately, it was not possible to conduct further analy-

ses on these tapes because the host company insisted that they be

destroyed to prevent the unwanted dissemination of sensitive company

3 Because of the sensitive and privileged nature of the internal ac-

counting information, I was not permitted to divulge these data. In-

deed, although I helped company officials present this information in

understandable form, these charts and graphs were never made part of
my file.

Page 4


Employee theft rates. The measure of employee theft used for this
study was the company accounting department^ standard formula for

computing "shrinkage." The formula yielded the percentage of inven-

tory (e.g, tools, supplies, etc.) unaccounted for by known waste, sales,

use in the conduct of business, or normal depreciation. (For a discus-
sion of the difficulties attendant to deriving such measures, see Hoi-

linger & Clark, 1983.) These measures were obtained unobtrusively
(during nonwork hours) by representatives of the company's head-

quarters on a weekly basis during the study period. The persons taking

inventory were aware of any legitimate factors that contributed to ac-
counted-for changes in inventory levels (such as shipments received,
supplies used during projects, etc.) but were blind to the experimental

Because no single standard for computing shrinkage is uniformly
used (Hollinger & Clark, 1983), it was not possible to compare the base

rates of employee theft in the present sample to any industry-wide

average. However, evidence that the employee theft rate studied here
was not atypical was provided by showing that the mean theft rate for
the 10-week period before the pay cut was not significantly different

from the overall theft rate for all three plants for the prior year, F <

1.00. These data are important in that they provide some assurance
that the changes in theft rates observed were not simply deviations

from unusual patterns that later merely regressed to the mean.5

Questionnaire measures. Two types of questionnaire measures were
needed to establish the validity of the study and to facilitate interpreta-

tion of the theft data—one group of questions to verify differences in
familiarity with the basis for establishing pay (the manipulation
check), and another group of questions to establish differences in per-

ceived payment equity. The questionnaires were administered bi-
weekly (during odd-numbered weeks in the study period) at the plant
sites during nonworking hours. Because a larger, unrelated study had

been going on for several months, the workers were used to completing
questionnaires, making it unlikely that any suspicions were aroused by

the questions inserted for this study. Participants were assured of the

anonymity of their responses.

The "pay basis" measure was designed to provide a check on the

validity of the payment-group variable. Participants answered four

items on a 5-point scale ranging from not at all (1), to slightly (2), to

moderately (3), to highly (4) to extremely (5). The questions were (a)
"How adequate was your employer's explanation regarding the basis of
your current pay?" (b) "How familiar are you with the way your em-

ployer determines your pay?" (c) "How thoroughly did your employer
communicate the basis for your current pay to you?" and (d) "How
much concern did your employer show about your feelings when com-
municating your pay?" A high degree of internal consistency was

found for these items (coefficient alpha = .89).

The "pay equity" measure consisted of four items, three of which
were anchored with the same scale points as the pay basis items. Specif-
ically, participants responded to the following items: (a) "To what ex-

tent do you believe your current pay reflects your actual contributions
to the job?" (b) "How fairly paid do you feel you currently are on your
job?" and (c) "How satisfied are you with your current overall pay
level?" The fourth item asked, "Relative to what you feel you should be
paid, do you believe your current pay is: much too low, a

little too low, about right, a little too high, much too
high?" Because only the first 3 points of this bidirectional scale were
actually used, responses to this 3-point scale were combined with the
5-point unidirectional scales for the other items. Coefficient alpha was
high (.84), justifying combining the individual items. The option of
using existing standardized scales tapping reactions to pay (e.g., the
Pay Satisfaction Questionnaire; Heneman & Schwab, 1985) was re-
jected in favor of ad hoc measures because these were judged to be
much more sensitive to the measurement objectives of the present
study (cf. Heneman, 1985).


Preliminary Analyses

Prior to the principal data analyses, preliminary analyses

were conducted to determine whether to separate the 15 bi-

weekly questionnaire responses into three equal groups, re-

flecting responses before, during, and after the pay cut. The

five 2-week response periods were treated as a repeated mea-

sure in mixed-design analyses of variance (ANOVAS) in which the

payment group was the between-subjects factor (adequate expla-

nation, inadequate explanation, no pay cut). Separate analyses

were conducted for each of the three groups. Because no signifi-

cant main effects or interactions involving the response periods

were obtained in analyses for either questionnaire measure (all

Fs < 1.00), the decision was made to combine the observations

into three groups composed of more reliable observations (be-

fore, during, or after the pay cut).

Because only one employee-theft-rate figure was reported for

each week (the figure was aggregate, as opposed to individual,

data), it was not possible to conduct a parallel set of ANOVAS for

this measure. However, separate tests were performed within

each payment group to compare each week's theft rate to the

mean for all 10 weeks. This process was repeated separately for

each of the three response periods (i.e, before, during, and after

the pay cut). Because no significant effects emerged in any of

these analyses (all values of t < .50, df= 9), the decision was

made (paralleling that for the questionnaire measures) to group

the weekly scores into three 10-week response periods.

Employee Theft Rate

Analyses of theft rates were based on a 3 x 3 mixed-design

ANOVA in which payment group was the between-subjects vari-

able, response period was the within-subjects variable, and the

10 weekly theft rates within each cell constituted the data. A

significant Payment Period X Response Period interaction was

found, F(4, 56) = 9.66, p < .001. Figure 1 summarizes the

means contributing to this interaction.

For each payment group, simple effects tests were performed

to determine whether the means differed significantly across

response periods. Any significant effects were followed up with

the Tukey honestly significant difference (HSD) procedure

(with alpha set at .05). In addition, tests for quadratic trend

components were performed using orthogonal polynomials

(Hays, 1963). This analysis was performed to note trends in the

data over time in a situation in which the number of available

4 Although the theft-rate figures (i.e, percentage of inventory loss
unaccounted for) were used internally to compute dollar-loss figures,
data substantiating a specific dollar-loss amount caused by the thefts

were not made available to me. Again, this decision was prompted by
the company's desire to avoid potential embarrassment.

5 Unfortunately, week-by-week theft-rate data were not available

prior to the study period. As a result, it was impossible to compare the
weekly theft rates during the study to earlier weekly theft rates. Thus, it
was not possible to rule out the possibility raised by one reviewer that
the results may reflect some seasonal fluctuations in theft that coin-
cided with the manipulation period.

Page 5









2.0 -


0 Inadequate explanation

- • Adequate explanation

1 Control

Before During

Time Period Relative to Pay Cut


Figure 1. Mean percentage of employee theft as a function

of time relative to pay cut.

data points was too small to use time series analyses (Zuwaylif,

A simple effects test within the inadequate-explanation con-
dition was significant, F(2, 27) = 9.15, p < .001. Post hoc tests
revealed that significantly higher levels of theft were observed
during the pay reduction than before or after the pay reduction.
Consistent with this configuration, the quadratic trend was
highly significant, F(l, 27) = 12.18, p< .001.

Within the adequate-explanation condition, the overall sim-
ple effects test was weaker but still significant, F(2, 52) = 3.76,
p < .05. This effect was the result of a similar, though less
pronounced, pattern of means showing theft to be higher dur-
ing the pay cut than either before or after the pay cut. Tests for a
quadratic trend component tailed to reach conventional levels
of significance, F(2, 52) = 2.10, p< .15.

Finally, within the control group, simple effects tests revealed
that the means did not differ from each other significantly
across the three response periods, F(2, 55) < 1.00.

To establish pre- and postmanipulation equivalence, it was
useful to compare means between payment groups (adequate
explanation, inadequate explanation, no pay cut) within re-
sponse periods. Simple effects tests showed no significant sim-
ple main effects of payment group before or after the pay cut,
F < 1.00 in both cases. However, the effect of payment group
was highly significant during the pay cut, F(2, 27) = 10.71, p <
.001. Tukey HSD tests revealed that the three means were signif-
icantly different from each other. In other words, within the
pay-reduction period, the theft rate in the inadequate-explana-
tion condition (M = 8.9) was significantly higher than that in
the adequate-explanation condition (M = 5.7), which was in
turn higher than that in the control condition (M = 3.7).

Questionnaire Responses

Responses to the pay basis and pay equity questionnaires
were analyzed with ANOVA designs identical to that used for the

employee-theft measure. For these dependent variables, how-
ever, the data consisted of individual responses to the summed
items constituting each questionnaire within each cell. The two
questionnaire measures were not significantly correlated,

For the pay basis measure, a significant Payment Group X
Response Period interaction was obtained, F(4,280) = 256.10,
p < .0001. The corresponding means and standard deviations
are summarized at the top of Table 2. As shown, post hoc tests
revealed that employees in the adequate-explanation condition
demonstrated greater understanding of the basis for pay deter-
mination than employees in the other two conditions once the
explanation occurred (i.e, during and after the pay cut). The
adequate-explanation manipulation successfully enhanced em-
ployees' understanding of the basis for pay determination.

A significant interaction effect also was obtained for the pay
equity measure, F(4,280) = 29.05, p < .001. The corresponding
means and standard deviations are summarized at the bottom
of Table 2. As shown, post hoc tests revealed that during the pay
cut, employees in the inadequate-explanation condition ex-
pressed the greatest perceptions of pay inequity. Workers whose
pay reductions were adequately explained to them did not ex-
press heightened payment inequity while their pay was re-


A summary of missing data and data lost because of volun-
tary turnover appears in Table 1. Not surprisingly, the majority
of the turnover occurred among employees experiencing inade-
quately explained pay reductions (12 of the 52 workers, or 23.1 %
of those still on the job at that time). Resignations in other
conditions were uniformly 5% or less. Consistent with this, the
distribution of resignations over conditions during the pay-cut
period was highly significant, x2(2, JV= 13) = 20.48, p < .001—a
result of the fact that 12 of the 13 resignations occurred in the
inadequate-explanation group. By contrast, the distribution of
resignations across conditions was equal before the pay cut,
X2(2, N= 3) < 0.5, and after the pay cut, x2(2, N= 5) < 0.5.


The data support the hypothesis derived from equity theory
(Adams, 1965) that workers experiencing underpayment ineq-
uity would attempt to redress that inequity by raising their in-
puts—in the present case, by pilfering from their employer.
Indeed, while workers experienced a 15% pay reduction, they
reported feeling underpaid and stole over twice as much as they
did when they felt equitably paid. Two distinct interpretations
of these theft data may be offered, both of which are consistent
with equity theory (Adams, 1965). First, it is possible that the
pay reduction led to feelings of frustration and resentment,
which motivated the aggressive acts of theft. This possibility is
in keeping with recent research findings demonstrating that
pay cuts are associated with negative affective reactions to orga-
nizational authorities (Greenberg, 1989) and that increases in
vandalism correlate positively with perceptions of mistreat-
ment by authorities (DeMore et al, 1988). Such an interpreta-
tion follows from a reciprocal deviance orientation to inequity

Page 6


Table 2

Data Summaries for Questionnaire Measures

Response period


Measure/payment group

Pay basis*
Inadequate explanation
Adequate explanation

Pay equity1'
Inadequate explanation
Adequate explanation
























Note. Within each row and each column, means not sharing a common subscript are significantly different
from each other beyond the .05 level on the basis of the Tukey honestly significant difference technique
corrected for confounded comparisons with the Cicchetti (1972) approximation.
* Mean scores for the pay basis measure could range from 20 to 100. Higher scores reflect greater degrees
of familiarity with the basis for establishing pay. * Mean scores for the pay equity measure could range
from 20 to 90. Higher scores reflect greater degrees of perceived payment equity.

reduction, which suggests that employees' acts of deviance are

encouraged by their beliefs that their employers defaulted on

their obligations to them by reducing their pay (Kemper, 1966).

From this perspective, acts of theft may be understood as a

manifestation of feelings of mistreatment.

It is also possible to interpret the thefts as direct attempts to

correct underpayment inequity by adjusting the balance of val-

ued resources between the worker and the specific source of

that inequity. As such, acts of theft may be interpreted as unof-

ficial transfers of outcomes from the employer to the employee.

Because no direct evidence is available suggesting that the

stolen items had any positive valence to the employees, it is

impossible to claim unambiguously that the theft rates repre-

sented employees' attempts to increase their own outcomes.

Although such an interpretation is consistent with a consider-

able amount of evidence on the distribution of rewards and

resources (for reviews, see Freedman & Montanari, 1980; Lev-

enthal, 1976), it is also possible that disgruntled employees may

have been content to reduce the valued resources available to

the agent of their discontent. That is, they may have been moti-

vated to reduce the employer's worth whether or not doing so

directly benefited themselves. Unfortunately, the questionnaire

items that would have been necessary to provide more refined

interpretations of the present data might also have aroused sub-

jects' suspiciousness that theft was being studied, thereby

creating the potential for subject reactance (Webb, Campbell,

Schwartz, Sechrest, & Grove, 1981). As a result, no such self-re-

port data were collected. Nevertheless, the results are clearly in

keeping with equity theory.

The present data also reveal a critical moderator of the ten-

dency to pilfer to restore equity with one^ employer—namely,

the use of an adequate explanation for the pay cut. Pay cuts that

were explained in an honest and caring manner were not seen

by employees as being as unfair as pay cuts that were not ex-

plained carefully. Accordingly, reactions to carefully explained

underpayment also were less severe (t£, the pilferage rates were

lower). These findings add to a recently developing body of

research showing that the use of adequately reasoned explana-

tions offered with interpersonal sensitivity tends to mitigate the

negative effects associated with the information itself (for re-

views, see Folger & Bies, 1989; Tyler & Bies, 1990). The explana-

tions used in the present study were obviously quite successful

in reducing costs, both to employees (in terms of inequity dis-

tress) and employers (in terms of pilferage and turnover).

An interesting and important aspect of the present study is

that a sizeable portion of the participants in the inadequate-ex-

planation condition voluntarily left their jobs during the pay-

reduction period; in fact, a much larger proportion resigned

than did so in any other condition {or within the same condition

at other times). It is tempting to take this finding as support for

the idea that quitting one's job is an extreme form of reaction to

underpayment inequity (Finn & Lee, 1972) and that the volun-

tary turnover found here was another form of reaction to ineq-

uity. However, because of the nonrandom design of the study, it

is not possible to rule out factors other than the experimental

manipulation—a difficulty common to quasi-experimental

studies (Cook & Campbell, 1976). Despite this problem, several

facts lend support to the inequity interpretation. First, the find-

ing that the theft rate immediately before the manipulations

did not differ significantly from the previous year's theft rate

suggests that nothing out of the ordinary was happening that

may have been responsible for the results. Second, because the

theft rate was highest precisely under the only conditions in

which feelings of inequity were high (i.e, during the pay-cut

period following an inadequate explanation), feelings of ineq-

uity and theft rate probably are related, both resulting from the

manipulated variable exactly as predicted by equity theory

(Adams, 1965) and referent cognitions theory (Folger, 1986).

Because this interpretation is theoretically supported, its posi-

tion is strengthened relative to alternatives that may be raised in

the absence of random assignment.

Generalizing from the present findings, it appears that ade-

quately explaining inequitable conditions may be an effective

means of reducing potentially costly reactions to feelings of

Page 7


underpayment inequity To be effective, however, such explana-
tions must be perceived as honest, genuine, and not manipula-
tive (Tyler, 1988). Still, to the extent that underpayment condi-
tions are acknowledged and justified by employers (as opposed
to ignored or minimized by them), it appears that both workers
and their organizations may stand to benefit. Given the high
costs of employee theft (American Management Association,
1977), it appears that explaining the basis for inequities may be
a very effective (and totally free) mechanism for reducing the
costs of employee theft.

Practical implications notwithstanding, the present findings
raise some important questions for equity theory (Adams, 1965)
about the use of various modes of inequity reduction. Whereas
the focus of this study was on pilferage, turnover was another
type of response that occurred. Unfortunately, the nature of the
present data makes it impossible to determine the trade-offs
between various modes of inequity reduction. Did some em-
ployees resign in response to underpayment while others (per-
haps those with fewer options for alternative employment)
stayed on and expressed their negative feelings by stealing? Or
was it that the most aggrieved employees stole company prop-
erty before leaving, while others simply lowered their inputs?
Because the theft rates were aggregate, actuarial data and could
not be traced to particular employees, and because perfor-
mance data were not collected, it was not possible to determine
when and how different forms of inequity-reduction behavior
are likely to occur. As a result, serious questions remain regard-
ing how different inequity-resolution tactics may be used in
conjunction with each other.

Confidence in interpretations of the present findings is lim-
ited because actuarial-level dependent measures (theft and turn-
over) were collected in conjunction with an individual-level vari-
able (perceived payment equity), thereby making it impossible
to conduct mediational analyses of the results. Exacerbating
this problem is the fact that the use of a quasi-experimental
design does not allow the discounting of alternative explana-
tions (as noted earlier). Thus, although it is plausible that ineq-
uity leads to stealing unless mitigated by an adequate explana-
tion, it is impossible to statistically discount the alternative pos-
sibility that unknown preexisting differences between the
plants constituting the payment groups (eg, different norms
against stealing or differential acceptance of management's
promise that the pay cut would be temporary) may have been
responsible for the results. However, in support of the present
findings, it is important to note that such limitations are inher-
ent to some degree in all quasi-experimental research designs
(Cook & Campbell, 1976).

Although nonrandom assignment precludes the discounting
of alternative explanations, support for the present interpreta-
tion of the data may be derived from converging sources of
theoretically based data. In this case, several lines of analogous
research converge with my claim that adequate explanations
enhanced the acceptance of undesired outcomes. For example,
Folger and his associates (e.g, Folger & Martin, 1986; Folger,
Rosenfield, & Robinson, 1983) measured laboratory subjects'
feelings of discontent in reaction to procedural changes that
created unfavorable conditions for them. Consistent with refer-
ent cognitions theory (Folger, 1986), Folger and his colleagues
found that these feelings of discontent were reduced only when

the need to make procedural changes was adequately ex-
plained. Similarly, in another line of investigation, Wsiner et al.
(1987) found that persons victimized by another's harmdoing
expressed less anger toward the harmdoer when claims of miti-
gating circumstances were offered for the harmdoert actions.
Both lines of investigation show that negative affective reac-
tions are reduced by the presentation of adequate explanatory
information. As such, they provide good convergent evidence
for the my claim that adequately explained pay cuts mitigated
feelings of inequity and reactions to underpayment inequity.

Finally, an important question may be raised about the com-
pound nature of the explanation manipulation used in the pres-
ent study. Because the adequate-explanation condition and the
inadequate-explanation condition differed along several di-
mensions (postulated a priori to contribute to mitigation of the
effects of the inequity), it was not possible to determine the
individual effects of the various contributing factors. Specifi-
cally, the explanations differed in terms of several factors. Some
of these, such as the quality of the information and the interper-
sonal sincerity of its presentation, have been recognized as miti-
gating reactions to undesirable outcomes (Shapiro & Buttner,
1988). Other differences between conditions, such as possible
differences in the credibility of the source (the president versus
the vice president) have not yet been studied. Clearly, the
unique effects of these factors are prime candidates for future
research efforts.

To conclude, the results of the present study shed new light on
employee theft—one of the most important problems in the
field of human-resource management. The evidence confirms
that employee theft is a predictable response to underpayment
inequity and reveals that such reactions can be substantially
reduced by the inexpensive tactic of explaining the basis for the
inequity in clear, honest, and sensitive terms.


Adams, J. S. (1965). Inequity in social exchange. In L. Berkowitz (Ed.),

Advances in experimental social psychology (Vol. 2, pp. 267-299).

San Diego, CA: Academic Press.

American Management Association. (1977, March). Summary over-


and level of knowledge in three areas concerning crimes against busi-

ness: Draft report. Washington, DC: National Institute of Law En-

forcement and Criminal Justice, Law Enforcement Assistance Ad-

Astor, S. D. (1972, March). Twenty steps to preventing theft in business.

Management Review, 61(3), 34-35.

Bies, R. J. (1986, August). Identifying principles of interactional justice:

The case of corporate recruiting. Symposium conducted at the annual

meeting of the Academy of Management, Chicago, IL.

Bies, R. J, & Moag, 1. S. (1986). Interactional justice: Communication

criteria of fairness. In R. J. Lewicki, B. H. Sheppard, & B. H. Bazer-

man (Eds}, Research on negotiation in organizations (Vol. 1, pp. 43-

55). Greenwich, CT: JAI Press.
Cicchetti, D. Y(1972). Extension of multiple-range tests to interaction

tables in the analysis of variance: A rapid approximate solution.

Psychological Bulletin, 77, 405-408.

Cook, T. D, & Campbell, D. T. (1976). The design and conduct of

quasi-experiments and true experiments in field settings. In M. D.

Dunnette (Ed.), Handbook of industrial and organizational psychol-

ogy (pp. 223-326). Chicago: Rand McNally.

Similer Documents