1995 AMJ HPWS Paper PDF

Academy of Management Journal
1995, Vol. 38, No. 3, 635872.
THE IMPACT OF HUMAN RESOURCE

MANAGEMENT PRACTICES ON TURNOVER,
PRODUCTIVITY, AND CORPORATE
FINANCIAL PERFORMANCE
MARK A. HUSELID
Rutgers University
This study comprehensively evaluated the links between systems of

High Performance Work Practices and firm performance. Results
based on a national sample of nearly one thousand firms indicate that
these practices have an economically and statistically significant im-
pact on both intermediate employee outcomes (turnover and produc-
tivity) and short- and long-term measures of corporate financial per-
formance. Support for predictions that the impact of High Perfor-
mance Work Practices on firm performance is in part contingent on
their interrelationships and links with competitive strategy was lim-
ited.
The impact of human resource management (HRM) policies and prac-

tices on firm performance is an important topic in the fields of human re-
source management, industrial relations, and industrial and organiza-
tional psychology (Boudreau, 1991; Jones & Wright, 1992; Kleiner, 1990).
An increasing body of work contains the argument that the use of High Per-
formance Work Practices, including comprehensive employee recruitment
and selection procedures, incentive compensation and performance man-
agement systems, and extensive employee involvement and training, can
improve the knowledge, skills, and abilities of a firms current and po-
tential employees, increase their motivation, reduce shirking, and en-
hance retention of quality employees while encouraging nonperformers to
leave the firm (Jones & Wright, 1992; U.S. Department of Labor, 1993).
I am very grateful to Brian Becker for his many helpful comments on this article and for
his direction and guidance on the dissertation on which it is based. I would also like to thank
James Begin, Peter Cappelli, James Chelius, John Delaney, Steve Director, Jeffrey Keefe, Mor-
ris Kleiner, Douglas Kruse, Casey Ichniowski, David Levine, George Milkovich, Barbara Rau,
Frank Schmidt, Randall Schuler, Anne Tsui, David Ulrich, seminar participants at Cornell
University and the University of Kansas, and this journals anonymous referees for their com-
ments on earlier versions. Any and all remaining errors are mine.
This study was partially funded by grants from the Human Resource Planning Society,
the Society for Human Resource Management Foundation, the Mark Diamond Research
Fund, and the SUNY-Buffalo School of Management. The interpretations, conclusions, and
recommendations, however, are mine and do not necessarily represent the positions of these
institutions.
635
636 Academy of Management Journal June
Arguments made in related research are that a firms current and po-
tential human resources are important considerations in the development
and execution of its strategic business plan. This literature, although large-
ly conceptual, concludes that human resource management practices can
help to create a source of sustained competitive advantage, especially
when they are aligned with a firms competitive strategy (Begin, 1991; But-
ler, Ferris, & Napier, 1991; Cappelli& Singh, 1992; Jackson& Schuler, 1995;
Porter, 1985; Schuler, 1992; Wright & McMahan, 1992).
In both this largely theoretical literature and the emerging conven-
tional wisdom among human resource professionals there is a growing con-
sensus that organizational human resource policies can, if properly con-
figured, provide a direct and economically significant contribution to firm
performance. The presumption is that more effective systems of HRM
practices, which simultaneously exploit the potential for complementar-
ities or synergies among such practices and help to implement a firms
competitive strategy, are sources of sustained competitive advantage. Un-
fortunately, very little empirical evidence supports such a belief. What em-
pirical work does exist has largely focused on individual HRM practices
to the exclusion of overall HRM systems.
This study departs from the previous human resources literature in
three ways. First, the level of analysis used to estimate the firm-level im-
pact of HRM practices is the system, and the perspective is strategic rather
than functional. This approach is supported by the development and val-
idation of an instrument that reflects the system of High Performance
Work Practices adopted by each firm studied. Second, the analytical fo-
cus is comprehensive. The dependent variables include both intermedi-
ate employment outcomes and firm-level measures of financial perfor-
mance, and the results are based on a national sample of firms drawn from
a wide range of industries. Moreover, the analyses explicitly address two
methodological problems confronting survey-based research on this top-
ic: the potential for simultaneity, or reverse causality, between High Per-
formance Work Practices and firm performance and survey response bias.
Third, this study also provides one of the first tests of the prediction that
the impact of High Performance Work Practices on firm performance is
contingent on both the degree of complementarity, or internal fit, among
these practices and the degree of alignment, or external fit, between a firms
system of such practices and its competitive strategy.
THEORETICAL BACKGROUND
The belief that individual employee performance has implications for

firm-level outcomes has been prevalent among academics and practition-
ers for many years. Interest in this area has recently intensified, however,
as scholars have begun to argue that, collectively, a firms employees can
also provide a unique source of competitive advantage that is difficult for
its competitors to replicate. For example, Wright and McMahan (1992),
1995 Huselid 637
drawing on Barneys (1991) resource-based theory of the firm, contended

that human resources can provide a source of sustained competitive ad-
vantage when four basic requirements are met. First, they must add val-
ue to the firms production processes: levels of individual performance
must matter. Second, the skills the firm seeks must be rare. Since human
performance is normally distributed, Wright and McMahan noted, all hu-
man resources meet both of these criteria. The third criterion is that the
combined human capital investments a firms employees represent cannot
be easily imitated. Although human resources are not subject to the same
degree of imitability as equipment or facilities, investments in firm-spe-
cific human capital can further decrease the probability of such imitation
by qualitatively differentiating a firms employees from those of its com-
petitors. Finally, a firms human resources must not be subject to re-
placement by technological advances or other substitutes if they are to pro-
vide a source of sustainable competitive advantage. Although labor-sav-
ing technologies may limit the returns for some forms of investment in
human capital, the continuing shift toward a service economy and the al-
ready high levels of automation in many industries make such forms of
substitution increasingly less probable.
Wright and McMahans work points to the importance of human re-
sources in the creation of firm-specific competitive advantage. At issue,
then, is whether, or how, firms can capitalize on this potential source of
profitability. Bailey (1993) contended that human resources are frequent-
ly underutilized because employees often perform below their maximum
potential and that organizational efforts to elicit discretionary effort from
employees are likely to provide returns in excess of any relevant costs. Bai-
ley argued that HRM practices can affect such discretionary effort through
their influence over employee skills and motivation and through organi-
zational structures that provide employees with the ability to control how
their roles are performed.
HRM practices influence employee skills through the acquisition and
development of a firms human capital. Recruiting procedures that provide
a large pool of qualified applicants, paired with a reliable and valid se-
lection regimen, will have a substantial influence over the quality and type
of skills new employees possess. Providing formal and informal training
experiences, such as basic skills training, on-the-job experience, coaching,
mentoring, and management development, can further influence employ-
ees development.
The effectiveness of even highly skilled employees will be limited if
they are not motivated to perform, however, and HRM practices can affect
employee motivation by encouraging them to work both harder and
smarter. Examples of firm efforts to direct and motivate employee behav-
ior include the use performance appraisals that assess individual or work
group performance, linking these appraisals tightly with incentive com-
pensation systems, the use of internal promotion systems that focus on em-
ployee merit, and other forms of incentives intended to align the interests
of employees with those of shareholders (e.g., ESOPS and profit- and gain-
sharing plans).
Finally, Bailey (1993) noted that the contribution of even a highly
skilled and motivated workforce will be limited if jobs are structured, or
programmed, in such a way that employees, who presumably know their
work better than anyone else, do not have the opportunity to use their
skills and abilities to design new and better ways of performing their
roles. Thus, HRM practices can also influence firm performance through
provision of organizational structures that encourage participation among
employees and allow them to improve how their jobs are performed.
Cross-functional teams, job rotation, and quality circles are all examples
of such structures.
Thus, the theoretical literature clearly suggests that the behavior of
employees within firms has important implications for organizational per-
formance and that human resource management practices can affect indi-
vidual employee performance through their influence over employees
skills and motivation and through organizational structures that allow em-
ployees to improve how their jobs are performed. If this is so, a firms HRM
practices should be related to at least two dimensions of its performance.
First, if superior HRM practices increase employees discretionary effort,
I would expect their use to directly affect intermediate outcomes, such as
turnover and productivity, over which employees have direct control.
Second, if the returns from investments in superior HRM practices exceed
their true costs, then lower employee turnover and greater productivity
should in turn enhance corporate financial performance. Therefore, in an-
ticipation of an estimation model that focuses on these dependent vari-
ables, my review of the empirical literature concentrates on prior work ex-
amining the influence of HRM practices on employee turnover, produc-
tivity, and corporate financial performance.
PRIOR EMPIRICAL WORK
Individual HRM Practices and Firm Performance
Turnover. Prior work has examined the determinants of both indi-

vidual employees departures and aggregate organizational turnover, al-
though most of the prior work has focused on the former. For example,
Arnold and Feldman (1982), Baysinger and Mobley (1983), and Cotton and
Tuttle (1986) concluded that perceptions of job security, the presence of
a union, compensation level, job satisfaction, organizational tenure, de-
mographic variables such as age, gender, education, and number of de-
pendents, organizational commitment, whether a job meets an individuals
expectations, and the expressed intention to search for another job were
all predictive of employees leaving, and Sheridan (1992) found that per-
ceptions of organizational culture influenced turnover. Thus, the theoret-
ical rationale for examining the effects of HRM practices on turnover lies
1995 Huselid 639
in their effects on these individual-level factors. Among the few empiri-

cal papers on the effects of specific HRM practices on aggregate turnover,
the work of McEvoy and Cascio (1985), who showed that job enrichment
interventions and realistic job previews were moderately effective in re-
ducing turnover, is notable.
Productivity. Research on the impact of HRM practices on organiza-
tional productivity is more extensive. Cutcher-Gershenfeld (1991) found
that firms adopting transformational labor relationsthose emphasizing
cooperation and dispute resolutionhad lower costs, less scrap, higher
productivity, and a greater return to direct labor hours than did firms us-
ing traditional adversarial labor relations practices. Katz, Kochan, and
Weber (1985) demonstrated that highly effective industrial relations sys-
tems, defined as those with fewer grievances and disciplinary actions and
lower absenteeism, increased product quality and direct labor efficiency,
and Katz, Kochan, and Keefe (1987) showed that a number of innovative
work practices improved productivity. Katz, Kochan, and Gobeille (1983)
and Schuster (1983) found that quality of work life (QWL), quality circles,
and labor-management teams increased productivity. Bartel (1994) estab-
lished a link between the adoption of training programs and productivity
growth, and Holzer (1987) showed that extensive recruiting efforts in-
creased productivity. Guzzo, Jette, and Katzells (1985) meta-analysis
demonstrated that training, goal setting, and sociotechnical systems design
had significant and positive effects on productivity. Links between in-
centive compensation systems and productivity have consistently been
found as well (Gerhart & Milkovich, 1992; Weitzman & Kruse, 1990). Fi-
nally, employee turnover also has an important influence on organizational
productivity (Brown & Medoff, 1978).
Corporate financial performance. A number of authors have explored
the links between individual HRM practices and corporate financial per-
formance. For example, Cascio (1991) and Flamholtz (1985) argued that the
financial returns associated with investments in progressive HRM practices
are generally substantial. Similarly, work in the field of utility analysis
(Boudreau, 1991; Schmidt, Hunter, MacKenzie, & Muldrow 1979) has con-
cluded that the value of a one-standard-deviation increase in employee per-
formance measured in dollars (SDy) is equivalent to 40 percent of salary
(per employee) and that the organizational implications of human re-
source management practices that can produce such an increase are con-
siderable. Although most of the empirical work on this topic has been con-
ducted in laboratories, Becker and Huselid (1992) presented field data sug-
gesting that SDy may in fact be well in excess of 40 percent of salary.
Similarly, Terpstra and Rozell (1993) found a significant and positive link
between the extensiveness of recruiting, selection test validation, and the
use of formal selection procedures and firm profits, and Russell, Terborg,
and Powers (1985) demonstrated a link between the adoption of employ-
ee training programs and financial performance. The use of performance
appraisals (Berman, 1991) and linking such appraisals and compensation
have also been consistently connected with increased firm profitability

(Gerhart & Milkovich, 1992).
Limitations of the Prior Empirical Work
In summary, prior empirical work has consistently found that use of
effective human resource management practices enhances firm perfor-
mance. Specifically, extensive recruitment, selection, and training proce-
dures; formal information sharing, attitude assessment, job design, griev-
ance procedures, and labor-management participation programs; and per-
formance appraisal, promotion, and incentive compensation systems that
recognize and reward employee merit have all been widely linked with val-
ued firm-level outcomes. These policies and procedures have been labeled
High Performance Work Practices (U.S. Department of Labor, 1993), a des-
ignation I adopt here.
However, if this line of research is to be advanced, several serious lim-
itations in the prior empirical work have to be addressed. Two are method-
ological, and one involves both conceptual and measurement issues. The
first issue concerns the potential simultaneity between High Performance
Work Practices and corporate financial performance, a problem exacer-
bated by the prevalence of cross-sectional data in this line of research. For
example, if higher-performing firms are systematically more likely to
adopt High Performance Work Practices, then contemporaneous estimates
of the impact of these practices on firm performance will be overstated.
Alternatively, it may be that otherwise lower-performing firms turn to
High Performance Work Practices as a remedy. If so, then such cross-sec-
tional estimates will understate the true effects of HRM practices on firm
performance. This form of simultaneous relationship is less probable in
the case of turnover and productivity, because these variables would be
unlikely to widely influence the selection of High Performance Work Prac-
tices. However, given the direct link between firm profits and the avail-
ability of slack resources for investment in such practices, it is easy to
imagine a firms financial performance having such an influence.
A second methodological problem is related to the widespread col-
lection of data via questionnaire. Because survey respondents generally
self-select into samples, selectivity or response bias may also affect results.
The most common form of selectivity bias occurs when the probability of
responding to a questionnaire is related both to a firms financial perfor-
mance and the presence of High Performance Work Practices. Without
knowing the direction of these relationships a priori, however, a researcher
cannot determine the effect on the impact of such practices on firm per-
formance. Despite a well-developed literature devoted to the statistical cor-
rection of selection bias (Heckman, 1979), such correction has rarely been
attempted in prior work.
Systems of HRM practices and the concept of fit. The third signifi-
cant limitation of prior work is its widespread conceptual focus on single
High Performance Work Practices, and the measurement problems inher-
1995 Huselid 641
ent in broadening the focus to a system of such practices. A focus on in-

dividual practices presents both theoretical and methodological dilemmas,
as both recent research (Arthur, 1992; MacDuffie, 1995; Osterman, 1987a,
1994) and conventional wisdom would predict that firms adopting High
Performance Work Practices in one area are more likely to use them in oth-
er areas as well. Therefore, to the extent that any single example reflects
a firms wider propensity to invest in High Performance Work Practices,
any estimates of the firm-level impact of the particular practice will be up-
wardly biased. This likely bias presents a significant limitation for a line
of research that attempts to estimate the firm-level impact of a firms en-
tire human resources function, as the sum of these individual estimates
may dramatically overstate their contribution to firm performance.
The potential for bias associated with a focus on individual policies
has not been lost on several scholars, who have recently linked data on
systems of High Performance Work Practices with valued firm-level out-
comes. For example, Delaney (in press) found the widespread use of pro-
gressive human resource management practices to have a strong and neg-
ative effect on organizational turnover in the manufacturing sector. Ich-
niowski, Shaw, and Prennushi (1993), using longitudinal data from 30 steel
plants, found the impact of cooperative and innovative HRM practices
to have a positive and significant effect on organizational productivity.
Similarly, Arthur (1994) found in 30 steel minimills that those with
commitment human resource systems, emphasizing the development of
employee commitment, had lower turnover and scrap rates and higher pro-
ductivity than firms with control systems, emphasizing efficiency and
the reduction of labor costs. Finally, MacDuffie (1995) found that bun-
dles of internally consistent HRM practices were associated with higher
productivity and quality in 62 automotive assembly plants.
Each of these studies has focused on the impact of systems of High
Performance Work Practices on employee turnover or productivity. Re-
search on the links between systems of work practices and corporate fi-
nancial performance is much more limited. Kravetz (1988) and Schuster
(1986) each matched data on global human resource management pro-
gressiveness with accounting indexes of firm profits. Although both au-
thors concluded that more progressive HRM practices were associated
with enhanced performance, the analyses in each study were restricted to
simple bivariate correlations and thus did not control for variables such
as firm size or industry. Ichniowski (1990) concluded that the use of pro-
gressive HRM practices was associated with both high productivity and
high financial performance in 65 business units, but owing to data limi-
tations, he too was unable to resolve the issue of simultaneity between
HRM practices and firm performance or provide results beyond a single
sector, manufacturing.
In short, although a growing empirical literature focuses generally on
the impact of High Performance Work Practices, prior work has been lim-
ited in terms of the range of practices evaluated, the dependent variables,
and the industry context. For example, a finding that systems of work prac-
tices affect turnover or productivity does not necessarily mean that these
practices have any effect on firm profits, and the discovery that systems
of High Performance Work Practices affect profitability begs the important
issue of the processes through which they influence firm financial per-
formance. Therefore, unlike prior work this study included the full range
of organizational human resource practices, examined those practices in
terms of their impact on both immediate employment outcomes and cor-
porate financial performance, and did so within the context of a broad
range of industries and firm sizes. My initial summary hypotheses can be
stated as follows:
Hypothesis 1a: Systems of High Performance Work Prac-
tices will diminish employee turnover and increase pro-
ductivity and corporate financial performance.
Hypothesis 1b: Employee turnover and productivity will
mediate the relationship between systems of High Per-
formance Work Practices and corporate financial per-
formance.
The second hypothesis will allow for one of the first empirical tests
of a diverse theoretical literature positing the importance to firm perfor-
mance of synergies and fit among human resource practices as well as be-
tween those practices and competitive strategy (Milgrom & Roberts, 1993).
Baird and Meshoulam (1988) described the first of these complementari-
ties as internal fit. Their primary proposition was that firm performance
will be enhanced to the degree that firms adopt human resource manage-
ment practices that complement and support each another. Similarly, Os-
terman (1987a) argued that there should be an underlying logic to a firms
system of HRM practices and that certain policies and practices fit together.
Osterman (1994) found that firms valuing employee commitment, for in-
stance, are less likely to use temporary employees and more likely to in-
vest in innovative work practices such as skills training and incentive com-
pensation. A tangible focus on employee commitment can be expected to
help produce a stable core of employees, thus increasing the probability
that a firm will reap the benefits associated with investments in training.
And a preference for committed employees and the use of incentive com-
pensation may also help attract high-performing employees, because, all
else being equal, employees in such firms will receive higher wages to
match their greater productivity. Similarly, the returns from the use of valid
selection procedures are likely to be greater when a firms performance ap-
praisal and incentive compensation systems can recognize and reward
good employee performance, and incentive compensation systems should
perform best when linked with high-quality performance appraisals. An
internal promotion system provides a strong incentive for employees to re-
main with a firm and, when combined with the appropriate incentive com-
pensation and performance appraisal systems, can magnify the returns
1995 Huselid 643
from investments in employee development activities. Finally, the effec-

tiveness of employee participation systems will be enhanced if employ-
ees know their efforts will be rewarded and will increase the probability
of their advancement. Thus,
Hypothesis 2: Complementarities or synergies among
High Performance Work Practices will diminish em-
ployee turnover and increase productivity and corporate
financial performance.
A second form of complementarity, Baird and Meshoulams (1988) ex-
ternal fit, occurs at the intersection of a firms system of HRM practices
and its competitive strategy. The notion that firm performance will be en-
hanced by alignment of HRM practices with competitive strategy has
gained considerable currency in recent years and in fact underlies much
of the recent scholarship in the field (Begin, 1991; Butler et al., 1991; Cap-
pelli & Singh, 1993; Jackson & Schuler, 1995; Schuler, 1992; Wright &
McMahan, 1992). Moreover, a developing literature suggests that firms do
indeed attempt to match HRM practices with competitive strategies. For
example, Jackson, Schuler, and Rivero (1989) found that firms pursuing a
strategy of innovation used HRM practices that were broadly consistent
with that approach, and Arthur (1992) found that steel minimills adopt-
ing a strategy of differentiation emphasized employee commitment. Sim-
ilarly, Snell and Dean (1992, 1994) that found human resource management
practices varied systematically with type of manufacturing system, indi-
vidual job characteristics, and firm environment. Although no empirical
work has suggested that firms with better external fit exhibit higher per-
formance, the expectation that they should provides my final hypothesis:
Hypothesis 3: Alignment of a firms system of High Per-
formance Work Practices with its competitive strategy
will diminish employee turnover and increase produc-
tivity and corporate financial performance.
Fit Versus Best Practices

The internal fit perspective suggests that the adoption of an internal-
ly consistent system of High Performance Work Practices will be reflect-
ed in better firm performance, ceteris paribus: It should be possible to iden-
tify the best HRM practices, those whose adoption generally leads to val-
ued firm-level outcomes. The external fit perspective raises the conceptual
issue of whether any particular human resources policy can be described
as a best practice, or whether, instead, the efficacy of any practices can on-
ly be determined in the context of a particular firms strategic and envi-
ronmental contingencies. Although prior work has yet to provide a direct
test of these competing hypotheses, recent research finding strong main
effects for the adoption of High Performance Work Practices lends credence
to the best practices viewpoint.
The argument that firm performance will be enhanced to the degree

a firms HRM practices are matched with its competitive strategy is, how-
ever, compelling. In fact, the internal and external fit hypotheses may not
be altogether inconsistent: All else being equal, the use of High Perfor-
mance Work Practices and good internal fit should lead to positive out-
comes for all types of firms. However, at the margin, firms that tailor their
work practices to their particular strategic and environmental contingen-
cies should be able to realize additional performance gains. For example,
most firms should benefit from the use of formal selection tests, although
the results of such tests could be used to select very different types of peo-
ple, with those differences perhaps depending on competitive strategy.
Likewise, the use of formal performance appraisal and incentive com-
pensation systems has been widely found to enhance firm performance.
However, each of these practices can be used to elicit very different types
of behaviors from employees. In short, the process of linking environ-
mental contingencies with HRM practices may vary across firms, but the
tools firms use to effectively manage such links are likely to be consistent.
The issue of whether internal, external, or both types of fit affect firm per-
formance is central, and later in this article I provide an explicit test of
these hypotheses.
METHODS
Sample and Data Collection
A study of this type presents a number of data collection challenges.

It requires as broad a sample as possible and at the same time requires that
each data point provide comprehensive information on both organiza-
tional human resource practices and strategies and firm-level performance.
Thus, my sample was drawn from Compact Disclosure, a database con-
taining comprehensive financial information from 10-K reports1 on near-
ly 12,000 publicly held U.S. firms. Firms were included in the sample if
they had more than a hundred employees and excluded if they were for-
eign-owned, holding companies, or publicly held divisions or business
units of larger firms. These criteria yielded 3,452 firms representing all ma-
jor industries.
Firm-level data on High Performance Work Practices were collected
with a questionnaire mailed to the senior human resources professional
in each firm. I pretested the survey items with a number of colleagues and
human resource professionals and conducted a pilot study using all sur-
vey materials. In the main study, representatives of 968 firms submitted
usable responses, for an overall response rate of 28 percent.
1
10-k reprints are informational documents filed with the Securities and Exchange
Commission.
1995 Huselid 645
Questions concerning each High Performance Work Practice (de-

scribed below) were asked separately for exempt and nonexempt em-
ployees, and respondents indicated the proportion of employees in each
category who were affected by each practice. I then weighted the response
to each item by the proportion of employees in each category: the value
for each question was the sum across categories. Prior work has frequent-
ly employed a dummy variable to indicate the presence or absence of each
practice; the specification used here is more sensitive to the breadth of im-
plementation of each practice throughout a firm.
Measurement of High Performance Work Practices
Scale development. Prior work on the measurement of High Perfor-
mance Work Practices is extremely limited. The only relevant study was
conducted by Delaney, Lewin, and Ichniowski (1989), who in 1986 sent
7,765 business units for which data were available on the COMPUSTAT
tapes a 29-page questionnaire concerning a wide variety of HRM practices.
From the responses of 495 firms (a 6.4 percent response rate), Delaney and
colleagues concluded that ten practices in the areas of personnel selection,
performance appraisal, incentive compensation, job design, grievance pro-
cedures, information sharing, attitude assessment, and labor-management
participation represented sophistication in human resource manage-
ment. In this study, I adopted those ten items because they are consistent
with the prior empirical work. However, to provide a more exhaustive list
of contemporary High Performance Work Practices, I added items assess-
ing three practices widely found to affect a firms performance: the in-
tensity of its recruiting efforts (selection ratio), the average number of hours
of training per employee per year, and its promotion criteria (seniority ver-
sus merit).
These 13 items broadly represent the domain of High Performance
Work Practices identified in prior work (U.S. Department of Labor, 1993).
These items also represent important choice variables on which many
firms differ significantly (Delaney et al., 1989). However, the substantial
conceptual and empirical overlap among these items and my desire to
adopt a systems perspective make determination of the independent con-
tribution of each practice to firm performance impractical. Therefore, to
uncover the underlying factor structure associated with these practices, I
factor-analyzed each items standard score, using principal component ex-
traction with varimax rotation. Two factors emerged from these analyses;
and I constructed a scale for each by averaging the questions loading un-
ambiguously at .30 or greater on a single factor. Table 1 shows these re-
sults and the questionnaire items.
Following Bailey (1993), I named the first factor employee skills and
organizational structures. This factor includes a broad range of practices
intended to enhance employees knowledge, skills, and abilities and there-
after provide a mechanism through which employees can use those at-
tributes in performing their roles. Specifically, a formal job design program
TABLE 1
Factor Structure of High Performance Work Practicesa
Questionnaire Item 1 2 Alpha
Employee skills and organizational structures .67
What is the proportion of the workforce who are
included in a formal information sharing program
(e.g., a newsletter)? .54 .02
What is the proportion of the workforce whose job
has been subjected to a formal job analysis? .53 .18
What proportion of nonentry level jobs have been
filled from within in recent years? .52 .36
What is the proportion of the workforce who are
administered attitude surveys on a regular basis? .52 .07
What is the proportion of the workforce who participate
in Quality of Work Life (QWL) programs, Quality
Circles (QC), and/or labor-management participation
teams? .50 .04
What is the proportion of the workforce who have
access to company incentive plans, profit-sharing
plans, and/or gain-sharing plans? .39 .17
What is the average number of hours of training received
by a typical employee over the last 12 months? .37 .07
What is the proportion of the workforce who have access
to a formal grievance procedure and/or complaint
resolution system? .36 .13
What proportion of the workforce is administered an
employment test prior to hiring? .32 .04
Employee motivation .66
What is the proportion of the workforce whose
performance appraisals are used to determine their
compensation? .17 .83
What proportion of the workforce receives formal
performance appraisals? .29 .80
Which of the following promotion decision rules do you
use most often? (a) merit or performance rating alone;
(b) seniority only if merit is equal; (c) seniority among
employees who meet a minimum merit requirement;
(d) seniority.b .07 .56
For the five positions that your firm hires most frequently,
how many qualified applicants do you have per position
(on average)? .15 .27
Eigenvalue 2.19 1.76
Proportion of variance accounted for 16.80 13.60
a
Bold type indicates that the associated question loads unambiguously at .30 or greater
on a single factor.
b
Item was reverse-coded.
and enhanced selectivity will help ensure employee-job fit, and provid-
ing formal training will enhance the knowledge, skills, and abilities of both
new and old employees. Quality of work life programs, quality circles, and
labor-management teams are all forms of participation that allow em-
1995 Huselid 647
ployees to have direct input into the production process. Likewise, infor-
mation-sharing programs, formal grievance procedures, and profit- and
gain-sharing plans help to increase the probability that employee partici-
pation efforts will be effective because such programs provide a formal
mechanism for employer-employee communication on work-related issues.
The Cronbach's alpha for this scale was .67.
The second factor, which I named employee motivation (Bailey,
1993), is composed of a more narrowly focused set of High Performance
Work Practices designed to recognize and reinforce desired employee be-
haviors. These practices include using formal performance appraisals,
linking those appraisals tightly with employee compensation, and focus-
ing on employee merit in promotion decisions. Conceptually, core com-
petencies among employees are developed through selection, training,
and the design of work (factor 1, employee skills and organizational struc-
tures) and are subsequently reinforced through the second factor, em-
ployee motivation. The Cronbach's alpha for the employee motivation
scale was .66.
Scale validation. Although the correspondence between these scales
and the prior conceptual work was encouraging, I also performed several
analyses to demonstrate their convergent validity. I began by identifying
two external measures of the degree to which firms valued their employ-
ees by investing in them. First, widespread investments in High Perfor-
mance Work Practices are likely to require additional human resources staff
to assist in their implementation. Thus, the ratio of human resources staff
to total employees is a proxy for the importance a firm places on its hu-
man resources. I found the simple correlation between both factors and this
ratio to be .19 (p < .001). Thus, as expected, firms with high levels of High
Performance Work Practices also vote with their dollars and invest in
human resources staff. However, those staff levels may also reflect a firms
level of bureaucracy or institutional conditions related to its industry, ar-
eas potentially unrelated to the importance it places on human resources.
As a test of this possibility, I also regressed the work practices scales on
the human resources staff ratio and controls for firm size and industry. The
human resources staff ratio remained positive and highly significant in
each of these equations.
Second, I assumed that if a firms senior managers saw human re-
sources as crucial to organizational performance, it would (1) communi-
cate this importance to external audiences and (2) invest in High Perfor-
mance Work Practices. Thus, following Keats and Hitt (1988), I took all
available presidents letters and managements discussions for each firm
from the annual reports contained in Compact Disclosure. These docu-
ments were subsequently content-analyzed for any reference to the im-
portance of human resources, human capital, or the like, or to the impor-
tance of personnel, people, employee, staff, or workforce. Firms that made
such comments were coded 1; others were coded 0. Of the 763 firms for
which annual reports were available, 310 mentioned the importance of hu-
man resources (41 percent), and 453 did not. The employee skills and or-
ganizational structures score of firms citing the importance of human re-
sources was significantly higher than that of those making no such com-
ments (t = 2.33, p < .01). This difference remained significant in a logis-
tic regression model with controls for firm size and industry. Although the
equivalent tests for the employee motivation scale had the expected sign,
they did not reach significance at conventional levels. These findings are
plausible given the nature of the items included in each scale, The items
included in the employee skills and organization structures factor reflect
widespread investments in High Performance Work Practices intended to
develop employee core competencies and thereafter provide a mecha-
nism through which employees can influence their roles. The items in-
cluded in the employee motivation factor, however, are much more nar-
row in that they are intended to recognize and compensate employees for
behaviors consistent with the interests of the firms shareholders. Thus, it
is perhaps unsurprising that they are not reflected in such a broad context
as the firms annual report.
Finally, using different samples and time periods, but similar mea-
sures of High Performance Work Practices, Delaney (in press) reported re-
sults for turnover, and Ichniowski (1990) and Ichniowski and colleagues
(1993) reported results for productivity that are highly similar to those pre-
sented below. In short, as an initial attempt to develop indexes of the adop-
tion of High Performance Work Practices that can be used to determine if
extensive use of these practices really is better, these scales demonstrate
encouraging levels of reliability and validity.
Measurement of Internal and External Fit
Despite prior work arguing that enhanced internal and external fit will
enhance firm performance, the relevant research has not specified the func-
tional form that fit can be expected to take. In the business strategy liter-
ature, however, Venkatraman (1989) concluded that fit is most common-
ly measured in terms of a moderated relationship, or interaction, between
two variables. For example, the relationship between a firms competitive
strategy and its performance could co-vary with the type of environment
in which it operates. A second category of fit that is relevant in this con-
text is the degree of match between two variables. Fit as matching differs
from fit as moderation in that an explicit external performance criterion
is lacking (Venkatraman, 1989). For example, one might argue that fit has
been achieved if a firms competitive strategy and its structure have been
aligned, based on an a priori theoretical prediction, regardless of the out-
come. In the following sections, I develop several alternative indexes to
assess degree of internal and external High Performance Work Practices fit,
using Venkatramans categories of fit as moderation and fit as matching.
Given the paucity of prior work in the area, however, these measures
should be considered highly exploratory and the results interpreted with
caution.
1995 Huselid 649
Internal fit as moderation. Internal fit among work practices could be

expected to take the form of complementarity or synergy both within and
between the employee skills and organizational structures and employee
motivation factors. An indication of complementarity within each factor
would be reflected in positive mixed partial derivatives among the High
Performance Work Practices (Milgrom & Roberts, 1993), a rough proxy of
which would involve interacting each practice within each factor with
every other practice. Unfortunately, the use of such a measure is highly
impractical, largely because of the generally high levels of multicollinearity
among High Performance Work Practices. Therefore, in this study I focused
on the development of measures of internal fit between factors. Concep-
tually, the potential for synergies among High Performance Work Practices
should increase when these practices have been consistently implement-
ed throughout a firm. Moreover, the degree of consistency in the imple-
mentation of practices should interact with their overall level in that con-
sistently applied high levels of High Performance Work Practices should
have the greatest impact on firm performance. Thus, the first measure of
internal fit I developed consists of the interaction between the degree of
human resources policy consistency and the respective factors. Human re-
sources policy consistency was assessed with this Likert-scale survey
item: How would you describe the consistency of your human resource
policies across any divisions or business units your firm may have? (em-
phasis in original). Unfortunately, this measure is less than ideal for two
reasons. First, it has restricted range, as firms that by definition do not
adopt human resource policies consistently, such as holding companies,
were excluded from the sample. Second, because the two factor scales were
based on the proportions of coverage of exempt and nonexempt employ-
ees throughout a firm, a firm with a high score on these variables must have
widely adopted each practice.
The second measure of internal fit as moderation I adopted consists
of the interaction between these two measures. Based on the assumption
that the returns from investments in employee skills and organizational
structures will be higher to the extent that firms have also devoted sig-
nificant resources to employee motivation, this measure provides a
straightforward test of the magnitude of any such returns. This scale is su-
perior to the first internal fit-as-moderation measure in that it does not ex-
hibit the psychometric problems outlined above.
Internal fit as matching. The second broad category of internal fit con-
sists of the degree of match between the two factor scales (Venkatraman,
1989). In the current context, internal fit as matching would occur if a firm
were consistently low, medium, or high on both factors. As a test of the
matching model of internal fit, I calculated the absolute value of the dif-
ference between a firms scores on the employee skills and organization-
al structures and employee motivation scales (Venkatraman, 1989).
External fit as moderation. My first measure of external fit as mod-
eration indicates the degree of correspondence between each firms com-
petitive strategy and its system of High Performance Work Practices. Porter
(1985) provided the dominant typology of competitive strategies in the
business policy literature; the types specified are cost leadership, differ-
entiation, and focus. To provide an estimate of a firms competitive strat-
egy, each respondent indicated the proportion of its annual sales derived
from each of those strategies. In view of prior work (Jackson et al., 1989;
Jackson & Schuler, 1995), I assumed that a predominantly differentiation
or focus strategy would require more intensive investments in High Per-
formance Work Practices than would a cost leadership strategy. Thus, to
test the external fit-as-moderation hypothesis, I interacted the proportion
of sales derived from either a differentiation or focus strategy with scores
on the employee skills and organizational structures and employee moti-
vation scales, respectively.2
My second measure of external fit as moderation is based on behav-
ioral indication of the emphasis each firm placed on aligning its human
resource management practices and competitive strategy. Specifically, re-
spondents indicated whether or not they attempted to implement each of
seven strategic human resource management activities for all employees
(the Appendix lists these activities). I then constructed an index by adding
the number of affirmative responses to each question ( = .69).3 To test
my expectation that the returns from investments in both factors will be
greater when firms explicitly attempt to link human resources and busi-
ness objectives, I interacted each firms score on the strategic HRM index
with each factor score.
2
I focused on the differentiation and focus strategies for two reasons. First, as noted, I
assumed that the use of a differentiation or focus strategy would require more intensive in-
vestments in High Performance Work Practices than would use of a cost leadership strate-
gy. Second, because survey respondents were asked to indicate the proportions of their firms
annual sales derived from each of these strategies, their responses were constrained to equal
100 percent. Thus, the proportion of sales derived from cost leadership equaled 1 (dif-
ferentiation + focus), and any model that included all the strategy variables and the inter-
actions between these variables and the practices scales would be collinear. Therefore, to
gauge the impact of each strategy separately, I estimated models for each type. In these analy-
ses, cost leadership and its interactions with the practices scales produced results very sim-
ilar to those for differentiation and focus (the results were generally nonsignificant). In ad-
dition, I created a dummy variable that equaled 1 if the combined value of differentiation
plus focus was greater than 67 percent (that is, the majority) and 0 otherwise, thereby in-
corporating all three competitive strategies in a single variable. These results were also con-
sistent with the results presented in the text.
3
This measure was adapted from Devanna, Fombrun, Tichy, and Warren (1982). One
might argue that, given prior theoretical work, these activities should also be considered High
Performance Work Practices and included in the measurement scales. However, as present-
ed in the questionnaire, these seven items represent broad human resources management
goals, and respondents were only asked to indicate whether they attempted to implement
them for all employees. In comparison, the 13 items included in the practices scales refer to
specific policies, and respondents were asked to indicate the current prevalence of each type
of activity by category of employee. Thus, the items included in the scales and the strategic
HRM index differ in both scale of measurement and level of analysis.
1995 Huselid 651
External fit as matching. Finally, I calculated the fit-as-matching vari-

able by taking the absolute value of the difference between the Z score of
the proportion of sales resulting from a differentiation or focus strategy and
the respective factor scores (Venkatraman, 1989). This variable indicates
the degree to which firms adopting differentiation or focus strategies also
employ high levels of High Performance Work Practices and vice versa.
My expectation was that each fit-as-moderation interaction would be
positive and significant for the financial performance dependent vari-
ables. Given that a lower score for the fit-as-matching variables indicates
greater fit, I expected each of these measures to be negative and signifi-
cant.
Dependent Variables
Turnover. The level of turnover within each firm was assessed with
a single questionnaire item, What is your average annual rate o f
turnover? (emphasis in original). This question was asked separately for
exempt and nonexempt employees, and the level of turnover for each
firm is therefore the weighted average across each of these categories. This
variable should be interpreted with caution, however. First, consistent with
most of the prior work in this area (Cotton & Tuttle, 1986), this measure
includes both voluntary employee departures (quitting) and involuntary
ones (firings). Therefore, to the extent that human resource management
practices affect voluntary but not involuntary separation, my estimates of
the impact of HRM practices on turnover maybe understated. The salience
of this issue is increased as my data were collected in a period of wide-
spread corporate downsizings (fiscal year 1991), which increase all forms
of turnover.
Second, economists typically view turnover as a choice variable for
firms, involving a trade-off between employee separations and wages,
benefits, and working conditions. Prior empirical work has substantiated
this view (Bluedorn, 1982; Osterman, 1987b). However, in a substantial
body of empirical research lower turnover has been associated with de-
sirable organizational outcomes (Baysinger & Mobley, 1983; Osterman,
1987b). Although recognizing that each firm may have an optimal rate of
turnover (Abelson & Baysinger, 1984), in this study I assumed that low
rates of turnover are preferred to high rates. Given that my model for
turnover controls for employee compensation, I believe this assumption
to be justified.
Productivity. The logarithm of sales per employee is a widely used
measure of organizational productivity and was adopted here to enhance
comparability with prior work (Ichniowski, 1990; Pritchard, 1992). The pri-
mary advantages of this measure are that it provides a single index that
can be used to compare firms productivity as well as to estimate the dol-
lar value of returns for investments in High Performance Work Practices.
It should be emphasized that productivity is not synonymous with prof-
itability, however; a firm can go bankrupt maximizing sales per employ-
ee while ignoring current costs. Models specifying productivity as the log-

arithm of net income per employee (an alternative, although less fre-
quently used measure) produced very similar results.
Corporate financial performance. Prior work on the measurement of
corporate financial performance is extensive. Perhaps the primary dis-
tinction to be made among the many alternative measures is between
measurements of accounting and economic profits (Becker & Olson, 1987;
Hirsch, 1991). Economic profits represent the net cash flows that accrue
to shareholders; these are represented by capital (stock) market returns. Ac-
counting profits can differ from economic profits as a result of timing is-
sues, adjustments for depreciation, choice of accounting method, and
measurement error. Additionally, economic profits are forward-looking and
reflect the markets perception of both potential and current profitability,
but accounting data reflect an historical perspective. Although there is
widespread agreement in the literature that capital market measures are
superior to accounting data, accounting data provide additional relevant
information (Hirschey & Wichern, 1984). Moreover, accounting data are of-
ten the focus of human resource managers who must allocate scarce re-
sources. Therefore, I used both a market-based measure (Tobin's q) and an
accounting measure (gross rate of return on capital, or GRATE) of corpo-
rate financial performance. Each is the best available measure of its type
(Hall, Cummins, Laderman, & Mundy, 1988; Hirsch, 1991; Hirschey &
Wichern, 1984).
The logarithm of Tobin's q was calculated by dividing the market val-
ue of a firm by the replacement cost of its assets (Hall et al., 1988; Hirsch,
1991). Conceptually, q is a measure of the value added by management. I
calculated the measure of accounting profits, gross rate of return on cap-
ital, by dividing cash flow by gross capital stock (Hall et al., 1988; Hirsch,
1991). GRATE is a better measure of accounting profits than the traditional
return on assets or return on equity because it is not as greatly affected by
depreciation or other noncash transactions (Brainard, Shoven, & Weiss,
1980; Hall et al., 1988). The calculations I used for q and GRATE were tak-
en from Hall and colleagues. Because some data were missing, I was un-
able to complete all the adjustments to firm capital structure those sources
recommend. However, I was able to estimate the sensitivity of my results
to the missing variables by substituting values for them across all reason-
able ranges into my calculations; the analyses indicated that the missing
data did not materially affect my estimates. As is described below, I em-
ployed both contemporaneous and subsequent (t + 1) years corporate fi-
nancial performance data as a partial control for the effects of simultane-
ity bias.
Research in the field of financial economics often omits firms in the
utility and banking industries because they are subject to governmental reg-
ulation. In this study, these industries accounted for 184 of the firms on
which I had complete data. Results of analyses omitting these firms were
consistent with the results presented below.
1995 Huselid 653
Control Variables
The estimation models were developed to provide unbiased estimates

of the impact of High Performance Work Practices on firm performance.
Thus, the selection of the control variables for each dependent variable was
based on a careful review of the prior empirical work (cf. Huselid, 1993),
focusing on those variables likely to be related to both the dependent vari-
ables and the use of High Performance Work Practices. The controls for
each dependent variable included firm size (total employment), capital in-
tensity, firm- and industry-levels of union coverage, industry concentra-
tion, recent (five-year) growth in sales, research and development inten-
sity, firm-specific risk (beta), industry levels of profitability, net sales, to-
tal assets, and 34 dummy variables representing 35 two-digit Standard
Industrial Classification (SIC) codes. Unfortunately, there was no straight-
forward measure of firms total wage bills available for inclusion in the
turnover model. However, selling, general, and administrative expenses is
a common income statement item that serves as a proxy for employee com-
pensation. This variable is measured with error because it includes a
number of items not directly related to wage expenses and excludes some
wages directly related to production (the latter are typically included in
the cost of goods sold). If selling, general, and administrative expenses is
an adequate proxy, however, the level of union coverage should have a pos-
itive and sizable effect on it, as it does on compensation (Lewis, 1986), if
the elements unrelated to compensation are invariant to union coverage.
As a test of this proposition, I regressed selling, general, and administra-
tive expenses on firm-level union coverage and a series of control variables.
Union coverage was significant and positive, and the magnitude of this ef-
fect was economically significant. Firms with an average level of union
coverage (11.34 percent) had 8.1 percent higher selling, general, and ad-
ministrative expenses than firms with no union coverage. Alternatively,
each one-standard-deviation increase in union coverage created a 19 per-
cent increase in these expenses. Finally, firms with 100 percent union cov-
erage had 125 percent higher selling, general, and administrative expens-
es than firms with no unions. These figures are broadly consistent with
the 11 to 14 percent union wage premium calculated by Lewis (1986) and
provide support for the assertion that these expenses are an adequate
proxy for employee compensation. Finally, turnover was also included as
a control variable in the productivity models because prior work has iden-
tified it as an important determinant of productivity with strong links to
High Performance Work Practices.
Financial data were taken primarily from Compact Disclosure. I took
considerable care to ensure that all data were matched to the same ac-
counting period (July 1, 1991, to June 30, 1992), Missing data were re-
trieved from Moodys Industrial Manual or the Standard & Poors Stock
Price Guide, where possible. Otherwise, missing data were eliminated
listwise for each dependent variable. Stock price data were gathered from
the Investment Statistics Laboratory Daily Stock Price Record for Decem-
ber 31. Stock dividend and stock split data were gathered from Standard
& Poors Stock Price Guide. Capital intensity was calculated as the loga-
rithm of the ratio of gross property, plant, and equipment over total em-
ployment. The five-year trend in sales growth and R&D intensity (the log-
arithm of the ratio of R&D expenditures to sales) and compensation lev-
els (proxied by selling, general, and administrative expenses) were
calculated directly from the accounting data. Firm-level union coverage
and total employment were taken from the questionnaire, and industry-
level unionization data were taken from Curme, Hirsch, and McPherson
(1990). Concentration ratios were calculated by dividing the sum of the
largest four firms' sales within each industry by the total sales for that in-
dustry. The systematic component in the variability of a firm's stock price
(systematic risk, or beta) was calculated using the Center for Research on
Stock Prices (CRSP) database and a 250-day period. Initially, betas were
only available for 543 firms. Using an auxiliary regression equation, I in-
puted data for the missing observations (R 2 = .40).
RESULTS
Table 2 presents means, standard deviations, and correlations. The em-
ployee skills and organizational structures and employee motivation scales
reflect an average of standard scores, so their means are very near zero.
Turnover averaged 18.36 percent per year, and the logarithm of the pro-
ductivity averaged 12.05, or annual sales of $171,099 per employee. The
mean q was .46, and the average annual gross rate of return was 5.10 per-
46
cent. This value for q (e . = 1.58) implies that the market value of the av-
erage firm was 58 percent greater than the current replacement cost of its
assets. This result indicates that managements were generally working in
the interest of the shareholders to increase the value of their equity. A
GRATE value of 5.10 implies that each dollar invested in capital stock gen-
erates five cents in annual cash flow. Each of these values is consistent
with the results of prior work (Becker & Olson, 1992; Hirsch, 1991). Av-
erage total employment was 4,413 (the logarithm of this variable was used
in all subsequent analyses); firm level unionization averaged 11.34 percent;
and industry-level unionization averaged 13.97 percent. Total employment
and union coverage were lower than in most prior work in this area, pri-
marily because previous research has focused on the manufacturing sec-
tor, which is more heavily unionized. Finally, as expected, the employee
skills and organizational structures scale was negatively related to turnover,
while both scales were positively related to productivity and corporate fi-
nancial performance.
Tables 3 through 6 present the regression analysis results for Hy-
potheses la and lb. The first equation in each table contains the first fac-
tor scale, employee skills and organizational structures, the second equa-
tion contains the employee motivation scale, and the third equation con-
tains both. These analyses provide some indication of the sensitivity of the
TABLE 2
Means, Standard Deviations, and Correlationsa
Variables Means s.d. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
1. Turnover 18.36 21.87
2. Productivity 12.05 0.99 .24
3. Tobin's q 0.46 1.64 .10 .07
4. Gross rate of
return on assets 5.10 23.00 .03 .15 .35
5. Employee skills and
organizational
structures 0.02 0.52 .08 .06 .09 .13
6, Employee motivation 0.00 0.78 .04 .03 .20 .01 .15
7. Total employment 4,412.80 18,967.45 .13 .22 .02 .12 .18 -.15
8. Capital intensity 3.96 1.32 .29 .48 .11 .02 .05 -.23 .01
9. Firm union coverage 11.34 24.28 .14 .05 .09 .02 .05 -.51 .21 .29
10. Industry union
coverage 13.97 13.55 .22 .11 .11 .00 .04 -.36 .19 .40 .36
11. Concentration ratio 0.38 0.25 .05 .15 .03 .14 .08 -.12 .17 .02 .08 .16
12. Sales growth 0.61 1.08 .06 .06 .13 .01 .03 .12 .02 .04 .16 .03 .10
13. R&D intensity 0.03 0.06 .09 .01 .10 .11 .01 .18 .14 .06 .12 .12 .03 .04
14. Systematic risk 1.06 0.32 .09 .08 .05 .05 .00 .19 .06 .23 .18 .20 .08 .09 .10
15. Selling, general &
administrative
expenses b 286.54 1,622.02 .02 .31 .09 .18 .23 .00 .77 ,21 .16 .13 .01 .00 .11 .01
16. HRM policy
consistency 4.54 1.10 .10 .04 .01 .04 .14 .23 .12 .06 .12 .07 .08 .00 .03 .03 .08
17. Differentiation/focus 0.01 1.02 .03 .10 .12 .02 .05 .18 .03 .13 .15 .15 .00 .01 .10 .06 .01 .05
18. Strategic HRM index 3.36 1.98 .02 .01 .00 .08 .33 .06 .25 .01 .08 .01 .04 .01 .04 .07 .24 .05 .05
a
N = 816. All correlations greater than or equal to .05 are significant at the .05 level; those > .07 are significant at the .01 level, and those > .10
are significant at the .001 level (one-tailed tests). Raw means are reported for total employment and selling, general, and administrative expenses to ease
interpretation. The logarithms for these variables are used in all subsequent analyses.
b
In millions of dollars.
findings to my specification and a very rough estimate of the degree of bias

associated with a focus on individual facets of High Performance Work
Practices. As a test of Hypothesis 1b, in the fourth equation in Tables 5 and
6 I added turnover and productivity to the models for Tobin's q and
GRATE.
Turnover
Table 3 shows the regression results for turnover. Each equation
reached significance at conventional levels, and the control variables gen-
erally had the expected signs and significance levels. Although employ-
ee skills and organizational structures was consistently negative and sig-
nificant, employee motivation was not significant in either model. This re-
sult is less surprising when it is recognized that the use of incentive
compensation systems may actually encourage employees who are per-
forming poorly to leave a firm.
I next estimated the practical significance of the impact of High Per-
formance Work Practices on turnover, from the results of the third equa-
tion shown in Table 3. With all other variables held at their means, firms
with employee skills and organizational structures and employee moti-
vation scores three standard deviations below the mean exhibited 21.48
percent turnover, but firms with scores three standard deviations above the
mean had 15.36 percent turnover. This 40 percent decrease of course
would be the maximum expected effect of high performance practices, be-
cause it implies that a firm has moved from the total absence of any ef-
fective human resource programs to complete participation across all di-
mensions. A more representative estimate can be made by calculating the
effect of a one-standard-deviation increase in each practice scale on
turnover. Each such increase reduced turnover 1.30 raw percentage points,
or 7.05 percent relative to the mean. Considering that this model controls
for firm size, the impact of unions, and employee compensation (selling,
general, and administrative expenses), this effect is practically as well as
statistically significant. In fact, this specification provides a highly re-
strictive test of the impact of High Performance Work Practices on turnover,
as the inclusion of selling, general, and administrative expenses controls
not only for employee wage levels but also for any direct costs associated
with the implementation of these practices. Removing this variable and
thus allowing the effect of High Performance Work Practices on compen-
sation to be reflected in the direct effect of such practices increased the
magnitude of their impact on turnover by more than 20 percent.
Productivity
Table 4 presents the regression results for productivity. As above,
each equation reached significance at conventional levels, and the control
variables generally had the expected signs and significance levels. When
employee skills and organizational structures and employee motivation
were entered individually (models 4 and 5), each was positive and sig-
1995 Huselid 657
TABLE 3
Results of Regression Analysis for Turnover
Model I Model 2 Model 3
Variables b s.e. b s.e. b s.e.
Constant 44.965*** 9.418 46.363*** 9.420 44.758*** 9.486

Logarithm of total
employment 2.656*** 0.772 2.507*** 0.778 2.637*** 0.783
Capital intensity 2.229*** 0.659 2.279*** 0.663 2.240*** 0.663
Firm union coverage 0.088*** 0.029 0.089*** 0.032 0 . 0 9 0 * * * 0.032
Industry union coverage 0.222*** 0.080 0.225*** 0.080 0,222*** 0.080
Concentration ratio 1.376 3.611 1.369 3.617 1.360 3.615
Sales growth 0.329 0.592 0.362 0.592 0.332 0.593
R&D/sales 3.509 11.298 3.211 11.409 3.278 11.403
Systematic risk 1.490 2.158 1.577 2.177 1.532 2.176
Selling, general, &
administrative expenses 2.168*** 0.749 2.175*** 0.763 2.145*** 0.763
Employee skills and
organizational structures 1.769* 1.245 1.743* 1.258
Employee motivation 0.359 1.036 0.162 1.045
R2 0.385*** 0.383*** 0.385***
R 2 0.002 a 0.120 a 0.002 a
F for R 2 2.017 0.730 1.020
N 855 855 855
a
These statistics reflect the incremental variance accounted for when employee skills
and organizational structures and employee motivation, respectively, are added to the com-
plete specification for each model. The impact of High Performance Work Practices on the
dependent variable is underestimated by this statistic because the assumptions that the in-
dependent variables are orthogonal and have been entered on the basis of a clear causal or-
dering are not appropriate in the current study. This caveat applies to all reported results.
*p < .10, one-tailed test
**p < .05, one-tailed test
***p < .01, one-tailed test
nificant at conventional levels. In model 6, which includes both employ-

ee skills and organizational structures and employee motivation, only the
coefficient for the later remained significant. This finding graphically
demonstrates the need to adopt a systems perspective in evaluating the
links between High Performance Work Practices and firm-level outcomes
and the way in which focusing on a subset of human resources manage-
ment practices may overstate their effects. In fact, these analyses are like-
ly to understate the biases associated with a focus on individual High Per-
formance Work Practices, as I focus here on the impact of omitting an en-
tire facet of these practices, rather than a single practice.
To estimate the practical significance of the impact of High Perfor-
mance Work Practices on productivity, I next calculated the impact of a
one-standard-deviation increase in each practices scale on the numerator
of productivity (net sales) while holding total employment and all other
variables at their means. These analyses were based on model 6 from
TABLE 4
Results of Regression Analysis for Productivity
Model 4 Model 5 Model 6
Variables b s.e. b s.e. b s.e.
Constant 10.919*** 0.227 10.899*** 0.225 10.899*** 0.225

Logarithm of total
employment 0.123*** 0.018 0.119*** 0.017 0.123*** 0.018
Capital intensity 0.399*** 0.025 0.404*** 0.025 0.403*** 0.024
Firm union coverage 0.000 0.001 0.001 0.001 0.001 0.001
Industry union coverage 0.001 0.003 0.001 0.003 0.000 0.003
Concentration ratio 0.240** 0.146 0.251** 0.145 0.251** 0.145
Sales growth 0.105*** 0.024 0.100*** 0.024 0.101*** 0.024
R&D/sales 0.771** 0.457 1.004*** 0.457 1.002** 0.457
Systematic risk 0.083 0.087 0.042 0.087 0.043 0.087
Turnover 0.003** 0.001 0.003** 0.001 0.003** 0.001
Employee skills and
organizational structures 0.073* 0.050 0.046 0.051
Employee motivation 0.160*** 0.041 0.154*** 0.041
R2 0.490*** 0.498*** 0.498***
R 2 0.001 a 0.010 a 0.010 a
F for R 2 2.100 15.448*** 8.136***
N 855 855 855
a
These statistics reflect the incremental variance accounted for when employee skills
and organizational structures and employee motivation, respectively, are added to the com-
plete specification for each model. The impact of High Performance Work Practices on the
dependent variable is underestimated by this statistic because the assumptions that the in-
dependent variables are orthogonal and have been entered on the basis of a clear causal or-
dering are not appropriate in the current study. This caveat applies to all reported results.
Table 4. The findings indicate that each one-standard-deviation increase

raises sales an average of $27,044 per employee. This substantial figure rep-
resents nearly 16 percent of the mean sales per employee ($171,099).
However, this is a single-period estimate, and spending on High Perfor-
mance Work Practices should be thought of as an investment that can rea-
sonably be assumed to produce gains for longer than a single year. If the
effects of increasing such practices are arbitrarily assumed to accrue for a
five-year period at an 8 percent discount rate, the present value increase
in sales will be $107,979 per employee. It should be noted that the as-
sumption underlying this specification is that High Performance Work
Practices increase sales for a fixed number of employees rather than in-
crease efficiency (lower employment) given a constant level of sales. Oth-
erwise identical specifications that modeled sales as a function of total em-
ployment produced very similar results.
1995 Huselid 659
Corporate Financial Performance
Table 5 presents the results for Tobin's q, and Table 6 shows the same
specifications for the gross rate of return on assets. Each equation reached
significance at conventional levels, and the control variables generally had
the expected signs and significance levels. For example, consistent with
Hirsch (1991), R&D expenditures were positively related to q but negatively
related to GRATE. Hirsch speculated that these relationships occur because
firms with high current R&D expenditures have lower reported profits but
higher expected future earnings. More centrally, the results for q showed
the employee skills and organizational structures and employee motiva-
tion scales to be significant in each equation. For GRATE, employee skills
and organizational structures was positive and significant in each model
but employee motivation was not. Although the diversity in these results
reinforces the importance of researchers' considering multiple outcomes
when evaluating the impact of human resources department activities
(Tsui, 1990), the structure of incentive systems in many firms may help
to explain them. Given the numerous problems associated with the use of
accounting measures of firm performance in incentive compensation sys-
tems (Gerhart & Milkovich, 1992), many firms have begun to explicitly link
employee compensation with capital market returns. This shift may help
to explain why employee motivation has a much stronger impact on the
market-based performance measure than on the accounting returnsbased
measure.
I next assessed the practical significance of the impact of High Per-
formance Work Practices on firm profits. To do so, I estimated the impact
of a one-standard-deviation increase on the numerator of both Tobin's q
and GRATE while holding their denominators and all other variables at
their means. These analyses were based on models 9 and 13 from Tables
5 and 6, respectively. In terms of market value, the per employee effect of
increasing such practices one standard deviation was $18,641 (relative to
q). Such an increase in market value is not likely to occur immediately,
however. A more likely scenario is that investments in High Performance
Work Practices create an asset that provides an annual return. If one as-
sumes (again, arbitrarily) that these returns accrue over a five-year period
at an 8 percent discount rate, then such an investment would provide an
annuity of $4,669 per employee per year.
Estimates of the practical effects of increasing use of these practices
can also be made on the basis of annual accounting profits. Relative to
GRATE, each one-standard-deviation increase in High Performance Work
Practices increased cash flow $3,814. These figures are remarkably close
to the five-year annuity values calculated above.
Summary of financial performance results. In short, although there
is strong support for the hypotheses predicting that High Performance
Work Practices will affect firm performance and important employment
TABLE 5
Results of Regression Analysis Results for Tobin's q
Model 7 Model 8 Model 9 Model 10
Variables b s.e. b s.e. b s.e. b s.e.
Constant 0.672* 0.505 0.515 0.495 0.642 0.502 2.166* 0.995
Log of total employment 0.065** 0.040 0.082** 0.039 0.067** 0.040 0.106*** 0.041
Capital intensity 0.125*** 0.054 0.115** 0.054 0.119** 0.054 0.251*** 0.063
Firm union coverage 0.000 0.002 0.004 0.003 0.004* 0.003 0.003 0.003
Industry union coverage 0.002 0.007 0.003 0.007 0.003 0.007 0.005 0.007
Concentration ratio 0.443* 0.326 0.469* 0.325 0.471* 0.324 0.400 0.321
Sales growth 0.205*** 0.053 0.195*** 0.053 0.198*** 0.054 0.172*** 0.054
R&D/sales 2.354*** 1.009 1.935*** 1.013 1.937** 1.013 2.198** 1.005
Systematic risk 0.039 0.194 0.115 0.194 0.112 0.194 0.099 0.192
Employee skills and
organizational structures 0.215* 0.113 0.165* 0.113 0.139 0.112
Employee motivation 0.297*** 0.090 0.277*** 0.091 0.227*** 0.091
Turnover 0.007***
Productivity 0.271***
R2 0.138*** 0.146*** 0.148*** 0.167***
AR2 0.004 a 0.012 a 0.014 a 0.033 a
F for AR2 3.635* 10.842*** 6.483*** 7.781***
N 826 826 826 826
a
These statistics reflect the incremental variance accounted for when employee skills and organizational structures, employee motivation,
turnover, and productivity, respectively, are added to the complete specification for each model. The impact of High Performance Work Prac-
tices on the dependent variable is underestimated by this statistic because the assumptions that the independent variables are orthogonal and
have been entered on the basis of a clear causal ordering are not appropriate in the current study.
TABLE 6
Results of Regression Analysis for Gross Rate of Return on Assets
Model 11 Model 12 Model 13 Model 14
Variables b s.e b s.e. b s.e. b s.e.
Constant 0,126** 0.072 0.159** 0.072 0.125*** 0.072 0.588*** 0.140
Log of total employment 0.019*** 0.006 0.023*** 0.006 0.019*** 0.006 0.025*** 0.006
Capital intensity 0.011* 0.007 0.012* 0.008 0.011* 0.008 0.009 0.009
Firm union coverage 0.000 0.000 0.000 0.000 0.000 0000 0.000 0.000
Industry union coverage 0.000 0.001 0.000 0.001 0.000 0.001 0.001 0.000
Concentration ration 0.077** 0.046 0.075** 0.046 0.076** 0.045 0.065* 0.046
Sales growth 0.008 0.007 0.008 0.007 0.008 0.007 0.004 0.008
R&D/sales 0.213* 0.144 0.202** 0.146 0.201** 0.145 0.153** 0.145
Systematic risk 0.050* 0.027 0.049* 0.028 0.048* 0.027 0.048 0.027
Employee skills and
organizational structures 0.041** 0.016 0.043** 0.016 0.040* 0.016
Employee motivation 0.003 0.013 0.008 0.013 0.015 0.013
Turnover 0.000 0.000
Productivity 0.044*** 0.011
R2 0.117*** 0.109*** 0.117*** 0.137***
R2 0.008 a 0.001 a 0.008 a 0.027 a
F for R2 6.649*** 0.680*** 3.356*** 6.157***
N 826 826 826 826
a
These statistics reflect the incremental variance accounted for when employee skills and organizational structures, employee motivation,
turnover, and productivity, respectively. are added to the complete specification for each model, The impact of High Performance Work Prac-
tices on the dependent variable is underestimated by this statistic because the assumptions that the independent variables are orthogonal and
have been entered on the basis of a clear causal ordering are not appropriate in the current study.
outcomes, the results are not completely unambiguous. Notably, the sig-
nificant effects found are also financially meaningful. Moreover, where
these effects are meaningful their magnitude is consistent across very dif-
ferent measures of financial performance. For example, a one-standard-de-
viation increase in High Performance Work Practices yields a $27,044 in-
crease in sales and a $3,814 increase in profits. The ratio of these variables
(cash flow to sales) at 14 percent is very near the sample mean of 10 per-
cent. And assuming that the market value of a firm reflects the discount-
ed net present value of all future cash flows, the present value of these cash
flows ($15,277 at 8 percent for five years) is remarkably close to the esti-
mated per employee impact on firm market value of $18,614. The point
of these analyses is to demonstrate that High Performance Work Practices
have impacts of similar magnitude on each dependent variable of inter-
est. In fact, these results show a remarkable level of internal consistency,
especially given the fact that they are based on measures of firm perfor-
mance that are only moderately intercorrelated.
Sources of the Gains from High Performance Work Practices
The next series of analyses examined the processes through which
High Performance Work Practices affect corporate financial performance.
Specifically, Hypothesis 1b states that employee turnover and productiv-
ity will mediate the relationship between systems of work practices and
corporate financial performance. Following Baron and Kenny (1986), I first
regressed the mediating variables (turnover and productivity) on the prac-
tices scales (see Tables 3 and 4). The next step was to regress each de-
pendent variable on those scales (see models 7, 8, and 9 in Table 5 and
models 11, 12, and 13 in Table 6). The significant effects shown in each
case are necessary but not sufficient conditions to establish that mediation
exists. Finally, as an estimate of the magnitude of any mediation effect, I
regressed the dependent variables on the work practices scales and the me-
diating variables. These results are shown in the final models in Tables 5
and 6. Here, the decrement in the coefficients for the employee skills and
organizational structures and employee motivation scales as turnover and
productivity are entered into the profitability equations provides an esti-
mate of the degree to which the effects of High Performance Work Prac-
tices on firm performance can be attributed to these factors.
As expected, the coefficient on each practices scale becomes smaller
once turnover and productivity have been entered into the models. The
magnitude of this effect can be shown by calculating the proportionate
change in the impact of High Performance Work Practices on corporate fi-
nancial performance that can be attributed to the inclusion of turnover and
productivity. Although, on the average, the coefficients on the two scales
fall by approximately 20 percent each when turnover and productivity are
entered into the models, the joint effect is to reduce the estimated finan-
cial impact of High Performance Work Practices on q by 74 percent and on
GRATE by 77 percent. This effect is sizable and suggests that a significant
1995 Huselid 663
proportion of the impact of High Performance Work Practices on corpo-

rate financial performance is attributable to either lower turnover or high-
er employee productivity, or to both. The fact that turnover and produc-
tivity are temporally antecedent to my measures of firm profits and that
the contemporaneous estimates of the profitability effects were highly
similar increases my confidence in these results.
Evidence of Complementarity
The final phase in the analyses was to evaluate the influence of in-
ternal and external fit on the dependent variables. Owing to space con-
straints, I focus here on firm profits, but the results for turnover and pro-
ductivity were similar. The results of Tobin's q and GRATE appear in Ta-
bles 7 and 8, respectively, where the internal and external fit measures I
developed were individually added to the complete specifications pre-
sented in Tables 5 and 6.
Internal fit as moderation. The first measure of fit I developed was
the interaction between the degree of human resources policy consisten-
cy and each of the scales measuring High Performance Work Practices.
These results were uniformly nonsignificant. Conversely, the second mea-
sure of internal fit, the interaction between the employee skills and orga-
nizational structures and employee motivation scales, was positive and sig-
nificant for both Tobin's q and GRATE.
Internal fit as matching. The internal fit-as-matching variable, which
assesses the degree to which a firm adopts the same level of High Perfor-
mance Work Practices throughout its operations, is presented in the final
column of Tables 7 and 8. These results were negative and significant for
GRATE but nonsignificant for q.
External fit as moderation. The first external fit-as-moderation vari-
ables reflect the interaction between the proportion of sales associated with
differentiation and focus strategies and the employee skills and organiza-
tional structures and employee motivation scales respectively. These re-
sults were uniformly nonsignificant.
The second measures of external fit as moderation reflects the inter-
action between firms' scores on the strategic HRM index and the practices
scales. With the exception of the interaction between this index and em-
ployee motivation for GRATE, these analyses were also uniformly non-
significant.
External fit as matching. Finally, the fit-as-matching variables for ex-
ternal fit show the coefficient for q to be positivein the unanticipated
directionand significant, but nonsignificant elsewhere.
In summary, most of the coefficients on the fit measures had the ex-
pected signs, and the interaction of employee skills and organizational
structures and employee motivation was consistently positive and signif-
icant. But despite the strong theoretical expectation that better internal and
external fit would be reflected in better financial performance, on the
whole the results did not support the contention that either type of fit has
TABLE 7
Estimates of the Impact of Internal and External Fit on Tobin's qa
Variables Model 15 Model 16 Model 17 Model 18 Model 19
Employee skills and 1.676* 0.165* 0.157* 0.182 0.136
organizational structures (0.121) (0.119) (0,114) (0.228) (0.121)
Employee motivation 0.287*** 0.299*** 0.283*** 0.295** 0.248**
(0.095) (0.094) (0.092) (0.145) (0.101)
Internal fit
HR policy consistency 0.080
(0.064)
HR policy consistency X
employee skills and 0.005
organization structures (0.117)
HR policy consistency X 0.040
employee motivation (0.072)
Employee skills and
organizational structures X 0.192*
Match: Employee skills and
organizational structures 0.084
and employee motivation (0.129)
External fit
Differentiation/focus 0.063
(0.057)
Differentiation/focus X
organizational structures (0.105)
Differentiation/focus X 0.048
Strategic HR index 0.061**
(0.032)
Strategic HR index X
Strategic HR index X 0.002
Match: Differentiation/focus
and employee skills and 0.145*
Match: Differentiation/focus 0.034
R2 0.152*** 0.153*** 0.151*** 0.155*** 0.153***
R2 0.002 0.002 0.003 0.004 0.003
F for R2 0.658 1.924 1.003 1.185 0.858
N 826 826 826 826 826
a Standard errors are in parentheses.

1995 Huselid 665
TABLE 8
Estimates of the Impact of Internal and External Fit on Gross Rate of
Return on Assetsa
Variables Model 20 Model 21 Model 22 Model 23 Model 24
Employee skills and 0.045*** 0.054*** 0.044*** 0.070** 0.050***
organizational structures (0.017) (0.017) (0.016) (0.032) (0.017)
Employee motivation 0.011 0.009 0.009 0.045** 0.025*
(0.013) (0.013) (0.013) (0.020) (0.015)
Internal fit
HR policy consistency 0.011
(0.008)
HR policy consistency X
organization structures (0.016)
HR policy consistency X 0.007
Employee skills and
organizational structures X 0.035**
Match: Employee skills and
organizational structures 0.040**
External fit
Differentiation/focus 0.003
(0.015)
Differentiation/focus x
Differentiation/focus x 0.003
Strategic HR index 0.000
(0.005)
Strategic HR index X
Strategic HR index X 0.012**
Match: Differentiation/focus
and employee skills and 0.004
Match: Differentiation/focus 0.002
R2 0.132*** 0.131*** 0.127*** 0.132*** 0.133***
R2 0.003 0.004 0.001 0.005 0.006
F for R2 0.438 3.299 0.121 1.430 1.659
N 826 826 826 826 826
a
Standard errors are in parentheses.
any incremental value over the main effects associated with the use of High
Performance Work Practices.
Empirical Estimation Issues
The strength and magnitude of these results must be interpreted in
light of several potential confounds inherent in the design of this study.
Perhaps the primary threat to the validity of this studys findings is the po-
tential for endogeneity or simultaneity between corporate financial per-
formance and High Performance Work Practices. I dealt with this issue in
two ways. First, because a firms current work practices can be expected
to affect both present and future profitability, I used both contemporane-
ous and subsequent (t + 1 year) measures of corporate financial perfor-
mance. Although the results for the subsequent years' profits were slight-
ly weaker than the contemporaneous results, they were highly consistent.
Thus, I present them here because they are more conservative.
Second, to assess the magnitude of any simultaneity between High
Performance Work Practices and firm profits, I used Hausman's (1978) test
to evaluate the ordinary-least-squares (OLS) regression assumption that the
High Performance Work Practices scales are exogenous in the profitabili-
ty models. In analyses whose results are not shown, I generated a predicted
value for the employee skills and organizational structures and employee
motivation scales using a reduced-form model. Then I included each scale
and its predicted value in an OLS model for Tobin's q and GRATE. A sig-
nificant coefficient on the predicted value for each scale would indicate
it is endogenous in the model being estimated (Hausman, 1978). Although
these results showed that the High Performance Work Practices scales were
not in fact endogenous in the profitability models, the general controver-
sy surrounding the use of this test (Addison & Portugal, 1989) led me to
estimate two-stage least-squares models for each dependent variable as a
formal correction for simultaneity. Not only were these results consistent
with the OLS results, but they were in each case somewhat larger than the
OLS results presented here.
Survey response bias was also considered directly. The presence of re-
sponse bias implies that unobserved determinants of the decision to re-
spond to this studys survey are related to both firm performance and High
Performance Work Practices. Given the extensive control variables in-
cluded in my models, such bias is unlikely. However, to formally test this
possibility, I used Heckman's (1979) procedure, which generates an inverse
Mills' ratio that I then included in the OLS and two-stage least-squares re-
gression models for each dependent variable to control for selectivity
bias. In each case, the relationship between the work practice measures and
the dependent variables remained consistent with the results presented
above, and in no case would these corrections have altered my conclu-
sions. In fact, most of the corrections for simultaneity and selectivity bias
produced estimates of the impact of High Performance Work Practices larg-
1995 Huselid 667
er than my OLS regression results, which implies that the specifications

relied on here are conservative.
DISCUSSION
Prior work in both the academic and popular press has argued that the
use of High Performance Work Practices will be reflected in better firm per-
formance. This study provides broad evidence in support of these asser-
tions. Across a wide range of industries and firm sizes, I found consider-
able support for the hypothesis that investments in such practices are as-
sociated with lower employee turnover and greater productivity and
corporate financial performance. That my results were consistent across
diverse measures of firm performance and corrections for selectivity and
simultaneity biases lends considerable confidence to these conclusions.
The magnitude of the returns for investments in High Performance
Work Practices is substantial. A one-standard-deviation increase in such
practices is associated with a relative 7.05 percent decrease in turnover
and, on a per employee basis, $27,044 more in sales and $18,641 and
$3,814 more in market value and profits, respectively. These internally
consistent and economically and statistically significant values suggest that
firms can indeed obtain substantial financial benefits from investing in the
practices studied here. In addition, these estimates imply a constant lev-
el of investment in such practices each year. If an increase requires only
a one-time expense (as perhaps could be the case with recruiting or se-
lection costs), these values will be underestimates of the impact of High
Performance Work Practices on firm performance. Moreover, these calcu-
lations only include a firm's portion of the gains from increasing use of
these practices. Presumably, some of the value created by adopting more
effective HRM practices will accrue to employees, in the form of higher
wages and benefits (Becker & Olson, 1987). Since higher levels of High Per-
formance Work Practices lead to lower turnover, and presumably greater
employment security, there appears to be considerable justification for en-
couraging firms to make such investments from a public policy perspec-
tive.
The impact of High Performance Work Practices on corporate finan-
cial performance is in part due to their influence on employee turnover
and productivity. The identification of some of the processes through
which these practices affect firm profits helps to establish the plausibili-
ty of a link with corporate financial performance. However, some of their
influence on firm profits remains unaccounted for, and the source of these
remaining gains is an important topic for future research.
But despite the compelling theoretical argument that better internal
and external fit will increase firm performance, I found only modest evi-
dence of such an effect for internal fit and little evidence for external fit.
These findings are in fact consistent with recent attempts to model fit in
the organizational strategy literature (Venkatraman, 1989), and they are per-
haps unsurprising given the preliminary nature of the measures of fit I de-
veloped. And given the substantial main effects associated with systems
of High Performance Work Practices, one might conclude that the simple
adoption of such practices is more important than any efforts to ensure
these policies are internally consistent or aligned with firm competitive
strategy. However, the theoretical arguments for internal and external fit
remain compelling, and research based on refined theoretical and psy-
chometric development of these constructs is clearly required before such
a conclusion can be accepted with any confidence. The very large theo-
retical literature in the fields of human resources management based on
the premise that fit makes a difference cries out for more work in this area,
and the primary import of the current findings may in fact be to call at-
tention to this important line of research.
Finally, the reader is cautioned to recognize the limitations associat-
ed with the use of cross-sectional data when an attempt to draw conclu-
sions about causality is made. Although the use in this work of simulta-
neous equations, corrections for response bias, measures of current and
subsequent years' profits, extensive control variables, and a large and di-
verse sample mitigate many of the traditional methodological concerns,
longitudinal data on both High Performance Work Practices and firm per-
formance are needed to conclusively replicate the findings presented here.
But such data are extremely costly to generate and are as yet unavailable.
This caveat is not intended to obviate the central conclusions of this
study, however. Although traditional economic theory would suggest that
the gains associated with the adoption of High Performance Work Practices
cannot survive into perpetuity (because the returns from these invest-
ments will be driven toward equilibrium as more and more firms make
them), the substantial variance in the HRM practices adopted by domes-
tic firms and the expectation that investments in such practices help to cre-
ate firm-specific human capital that is difficult to imitate suggest that, at
least in the near term, such returns are available for the taking.
REFERENCES
Abelson, M. A., & Baysinger, B. D. 1984. Optimal and dysfunctional turnover: Toward an or-
ganizational-level model. Academy of Management Review, 9: 331341.
Addison, J. T., & Portugal, P. 1989. The endogeneity of union status and the application of
the Hausman test. Journal of Labor Research, 10: 437441.
Arnold, H. J., & Feldman, D. C. 1982. A multivariate analysis of the determinants of turnover.
Journal of Applied Psychology, 67: 350360.
Arthur, J. B. 1992. The link between business strategy and industrial relations systems in
American steel minimills. Industrial and Labor Relations Review, 45: 488506.
Arthur, J. B. 1994. Effects of human resource systems on manufacturing performance and
turnover. Academy of Management Journal, 37: 670687.
Bailey, T. 1993. Discretionary effort and the organization of work: Employee participa-
tion and work reform since Hawthorne. Working paper, Columbia University, New York.
1995 Huselid 669
Baird, L., & Meshoulam, I. 1988. Managing two fits of strategic human resource management.
Academy of Management Review, 13: 116128.
Barney, J. 1991. Firm resources and sustained competitive advantage. Journal of Manage-
ment, 17: 99120.
Baron, R, M., & Kenny, D. A. 1986. The moderator-mediator variable distinction in social psy-
chological research: Conceptual, strategic, and statistical considerations. Journal of Per-
sonality and Social Psychology 51: 11731182.
Bartel, A. P. 1994. Productivity gains from the implementation of employee training programs.
Industrial Relations, 33 411425.
Baysinger, B. D., & Mobley, W. H. 1983, Employee turnover: Individual and organizational
analysis. In K. M. Rowland & G. R. Ferris (Eds. ), Research in personnel and human re-
source management, vol 1: 269319. Greenwich, CT JAI Press.
Becker, B. E., & Huselid, M. A. 1992. Direct estimates of SDy and the implications for utili-
ty analysis. Journal of Applied Psychology, 77: 227233.
Becker, B. E., & Olson, C. A. 1987. Labor relations and firm performance. In M. M. Kleiner,
R. N. Block, M. Roomkin, & S. W. Salsburg (Eds.), Human resources and the performance
of the firm: 4385. Washington, DC: BNA Press.
Becker, B. E., & Olson, C. A. 1992. Unions and firm profits, Industrial Relations, 31: 395415.
Begin, J. P. 1991. Strategic employment policy: An organizational systems perspective. En-
glewood Cliffs, NJ: Prentice-Hall.
Bluedorn, A. 1982. The theories of turnover: Causes, effects, meanings. In S. Bacharach (Ed.),
Research in the sociology of organizations: 75128. Greenwich, CT JAI Press.
Borman, W. C. 1991. Job behavior, performance, and effectiveness. In M. D. Dunnette &L. M.
Hough (Eds.), Handbook of industrial and organizational psychology (2d ed.), vol. 2:
271326. Palo Alto, CA: Consulting Psychologists Press.
Boudreau, J. W. 1991. Utility analysis in human resource management decisions. In M. D.
Dunnette & L. M. Hough (Eds.), Handbook of industrial and organizational psycholo-
gy (2d ed.), VO l. 2: 621745. Palo Alto, CA: Consulting Psychologists Press.
Brainard, W., Shoven, J., & Weiss, L. 1980. The financial valuation of the return to capi-
tal. Washington, DC: Brookings Institution.
Brown, C., Medoff, J. 1978. Trade unions in the production process. Journal of Political
Economy, 86: 355378.
Butler, J. E., Ferris, G. R., & Napier, N. K. 1991. Strategy and human resources management.
Cincinnati: South-Western.
Cappelli, P., & Singh, H. 1992. Integrating strategic human resources and strategic manage-
ment. In D. Lewin, O. S, Mitchell, & P. Sherer (Eds.), Research frontiers in industrial
relations and human resources: 165192, Madison, WI: Industrial Relations Research
Association.
Cascio, W. F. 1991. Costing human resources: The financial impact of behavior in orga-
nizations (3d ed.). Boston: PWS-Kent.
Cotton, J, L., & Tuttle, J. M, 1986. Employee turnover: A meta-analysis and review with im-
plications for research, Academy of Management Review, 11: 5570.
Curme, M. A., Hirsch, B. T., & McPherson, D, A. 1990. Union membership and contract cov-
erage in the United States. Industrial and Labor Relations Review, 44: 533.
Cutcher-Gershenfeld, J. 1991. The impact on economic performance of a transformation in
industrial relations. Industrial and Labor Relations Review, 44: 241260.
Delaney, J. T. In Press. Unions, human resource innovations, and organizational outcomes.

Advances in Industrial and Labor Relations.
Delaney, J. T., Lewin, D., & Ichniowski, C. 1989. Human resource policies and practices in
American firms. Washington, DC: U.S. Government Printing Office.
Devanna, M. A., Fombrun, C., Tichy, N., & Warren, L. 1982. Strategic planning and human
resource management. Human Resource Management, 22: 111 7.
Flamholtz, E. G. 1985. Human resource accounting (2d ed.). San Francisco: Jossey-Bass.
Gerhart, B., & Milkovich, G. T. 1992. Employee compensation: Research and practice. In M. D.
Dunnette & L. M. Hough (Eds.), Handbook of industrial and organizational psycholo-
gy, vol. 3: 481569. Palo Alto, CA: Consulting Psychologists Press.
Guzzo, R. A., Jette, R. D., & Katzell, R. A. 1985. The effect of psychologically based inter-
vention programs in worker productivity: A meta analysis. Personnel Psychology, 38:
275291.
Hall, B. H., Cummins, C., Laderman, E. S., & Mundy, J. 1988. The R&D master file docu-
mentation. Technical working paper no. 72, National Bureau of Economic Research,
Boston.
Hausman, J. A. 1978. Specification tests in econometrics. Econometrics, 46: 12511271.
Heckman, J. J. 1979. Sample selection bias as a specification error. Econometrics, 47:
153161.
Hirsch, B. T. 1991. Labor unions and the economic performance of firms. Kalamazoo, MI:
W. E. Upjohn Institute for Employment Research.
Hirschey, M., & Wichern, D. W. 1984. Accounting and market-value measures of profitabil-
ity: Consistency, determinants, and uses. Journal of Business and Economic Statistics,
2: 375383.
Holzer, H. J. 1987. Hiring procedures in the firm: Their economic determinants and outcomes.
In M. M. Kleiner, R. N. Block, M. Roomkin, & S. W. Salsburg (Eds.), Human resources
and the performance of the firm: Washington, DC: BNA Press.
Huselid, M. A. 1993. Essays on human resource management practices, turnover, pro-
ductivity, and firm performance. Unpublished Doctoral Dissertation, State University
of New York at Buffalo.
Ichniowski, C. 1990. Manufacturing businesses. NBER working paper series no. 3449, Na-
tional Bureau of Economic Research. Cambridge, MA.
Ichniowski, C., Shaw, K., & Prennushi, G. 1993. The effects of human resource manage-
ment practices on productivity. Working paper, Columbia University.
Jackson, S. E., & Schuler, R. S. 1995. Understanding human resource management in the con-
text of organizations and their environments. In J. T. Spence, J. M. Darley, & D. J. Foss
(Eds.), Annual review of psychology, vol. 46: 237264. Palo Alto, CA: Annual Reviews,
Inc.
Jackson, S. E., Schuler, R. S., & Rivero, J. C. 1989. Organizational characteristics as predic-
tors of personnel practices. Personnel Psychology, 42: 727786.
Jones, G. R., & Wright, P.M. 1992. An economic approach to conceptualizing the utility of
human resource management practices. In K. Rowland & G. Ferris (Eds.), Research in
personnel and human resources management, vol. 10: 271299. Greenwich, CT JAI
Press.
Katz, H. C., Kochan, T. A., & Gobeille, K. R. 1983. Industrial relations performance, economic
performance, and QWL programs: An interplant analysis. Industrial and Labor Relations
Review, 37: 317.
1995 Huselid 671
Katz, H. C., Kochan, T. A., & Keefe, J. H. 1987. Industrial relations and productivity in the
U.S. automobile industry. Washington, DC: Brookings Institution.
Katz, H. C., Kochan, T. A., & Weber, M. R. 1985. Assessing the effects of industrial relations
systems and efforts to improve the quality of working life on organizational effectiveness.
Academy of Management Journal, 28: 509526.
Keats, B. W., & Hitt, M. A. 1988. A causal model of linkages among environmental dimen-
sions, macro organizational characteristics, and performance. Academy of Manage-
ment ]ournal, 31: 570598.
Kleiner, M. M. 1990. The role of industrial relations in firm performance. In J. A. Fossum &
J. Mattson (Eds.), Employee and labor relations: 4.234.43. Washington, DC: BNA Press.
Kravetz, D. J. 1988. The human resource revolution: Implementing progressive manage-
ment practices for bottom-line success. San Francisco: Jossey-Bass.
Lewis, H. G. 1986. Union relative wage effects. Chicago: University of Chicago Press.
MacDuffie, J. P. 1995. Human resource bundles and manufacturing performance: Organiza-
tional logic and flexible production systems in the world auto industry. Industrial and
Labor Relations Review, 48: 197221.
McEvoy, G. M., & Cascio, W. F. 1985. Strategies for reducing employee turnover: A meta-analy -
sis. Journal of Applied Psychology, 70: 342353.
Milgrom, P., & Roberts, J. 1993. Complementarities and fit: Strategy, structure, and orga-
nizational change. Working paper, Stanford University, Stanford, CA.
Osterman, P. 1987a. Choice of employment systems in internal labor markets. Industrial Re-
lations, 26: 4657.
Osterman, P. 1987b. Turnover, employment security, and the performance of the firm. In M. M.
Kleiner, R. N. Block, M, Roomkin, & S. W. Salsburg (Eds.), Human resources and the per-
formance of the firm: 275-317, Washington, DC: BNA Press.
Osterman, P. 1994. How common is workplace transformation and how can we explain who
adopts it? Results from a national survey. Industrial and Labor Relations Review, 47:
173188.
Porter, M. E. 1985. Competitive advantage: Creating and sustaining superior performance.
New York: Free Press.
Pritchard, R. D. 1992. Organizational productivity. In M. D. Dunnette & L. M. Hough (Eds.),
Handbook of industrial and organizational psychology (2d ed.), vol. 3: 443471. Palo
Alto, CA: Consulting Psychologists Press.
Russell, J. S., Terborg, J. R., & Powers, M. L. 1985. Organizational performances and organi-
zational level training and support. Personnel Psychology, 38: 849863.
Schmidt, F. L., Hunter, J. E., McKenzie, R. C., & Muldrow, T. W. 1979. Impact of valid selec-
tion procedures on work-force productivity. JournaI of Applied Psychology, 64: 609626.
Schuler, R. S. 1992. Strategic human resource management: Linking people with the needs
of the business. Organizational Dynamics, 20: 1932.
Schuster, F. E. 1986. The Schuster report: The proven connection between people and prof-
its. New York: Wiley.
Schuster, M. 1983. The impact of union-management cooperation on productivity and em-
ployment. Industrial and Labor Relations Review, 36: 415430.
Sheridan, J. E. 1992. Organizational culture and employee retention, Academy of Manage-
ment Journal, 35: 10361056.
Snell, S. A., & Dean, J. W., Jr. 1992. Integrated manufacturing and human resource manage-
ment: A human capital perspective. Academy of Management Journal, 35: 467504.
Snell, S. A., & Dean, J. W., Jr. 1994. Strategic compensation for integrated manufacturing: The
moderating effects of job and organizational inertia. Academy of Management Journal,
37: 11091140.
Terpstra, D. E., & Rozell, E. J. 1993. The relationship of staffing practices to organizational
level measures of performance. Personnel Psychology, 46: 2748.
Tsui, A. S. 1990. A multiple-constituency model of effectiveness: An empirical examination
at the human resource subunit level. Administrative Science Quarterly, 35: 458483.
U.S. Department of Labor. 1993. High performance work practices and firm performance.
Washington, DC: U.S. Government Printing Office.
Venkatraman, N. 1989. The concept of fit in strategy research: Toward a verbal and statisti-
cal correspondence. Academy of Management Review, 14: 423444.
Weitzman, M. L., & Kruse, D. L. 1990. Profit sharing and productivity, In A. S. Blinder (Ed.),
Paying for productivity: 95141. Washington, DC: Brookings Institution.
Wright, P. M., & McMahan, G. C. 1992. Theoretical perspectives for strategic human resource
management. Journal of Management, 18(2): 295320.
APPENDIX
Components of the Strategic Human
Resources Management Indexa
1. Match the characteristics of managers to the strategic plan of the firm.
2. Identify managerial characteristics necessary to run the firm in the long term.
3. Modify the compensation system to encourage managers to achieve long-term
strategic objectives.
4. Change staffing patterns to help implement business or corporate strategies.
5. Evaluate key personnel based on their potential for carrying out strategic goals.
6. Conduct job analyses based on what the job may entail in the future.
7. Conduct development programs designed to support strategic changes.
a
Adapted from Devanna, Fombrun, Tichy, and Warren (1982).
Mark A. Huselid is an assistant professor in the School of Management and Labor

Relations at Rutgers University. He holds a Ph.D. degree in human resource man-
agement from the State University of New York at Buffalo. His current research fo-
cuses on the linkages between human resource management systems, corporate strat-
egy, and firm performance.

1995 AMJ HPWS Paper PDF

Uploaded by

Copyright:

Available Formats

1995 AMJ HPWS Paper PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1995 AMJ HPWS Paper PDF

Uploaded by

Copyright:

Available Formats

Academy of Management Journal

1995, Vol. 38, No. 3, 635872.

THE IMPACT OF HUMAN RESOURCE

This study comprehensively evaluated the links between systems of

The impact of human resource management (HRM) policies and prac-

The belief that individual employee performance has implications for

drawing on Barneys (1991) resource-based theory of the firm, contended

PRIOR EMPIRICAL WORK

Individual HRM Practices and Firm Performance

Turnover. Prior work has examined the determinants of both indi-

in their effects on these individual-level factors. Among the few empiri-

have also been consistently connected with increased firm profitability

ent in broadening the focus to a system of such practices. A focus on in-

from investments in employee development activities. Finally, the effec-

Fit Versus Best Practices

The argument that firm performance will be enhanced to the degree

Sample and Data Collection

A study of this type presents a number of data collection challenges.

Questions concerning each High Performance Work Practice (de-

Internal fit as moderation. Internal fit among work practices could be

External fit as matching. Finally, I calculated the fit-as-matching vari-

ee while ignoring current costs. Models specifying productivity as the log-

The estimation models were developed to provide unbiased estimates

findings to my specification and a very rough estimate of the degree of bias

Constant 44.965*** 9.418 46.363*** 9.420 44.758*** 9.486

nificant at conventional levels. In model 6, which includes both employ-

Constant 10.919*** 0.227 10.899*** 0.225 10.899*** 0.225

Table 4. The findings indicate that each one-standard-deviation increase

Corporate Financial Performance

proportion of the impact of High Performance Work Practices on corpo-

a Standard errors are in parentheses.

er than my OLS regression results, which implies that the specifications

Delaney, J. T. In Press. Unions, human resource innovations, and organizational outcomes.

Mark A. Huselid is an assistant professor in the School of Management and Labor

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Constant 44.965* 9.418 46.363* 9.420 44.758*** 9.486

Constant 10.919* 0.227 10.899* 0.225 10.899*** 0.225