娀 Academy of Management Journal

2005, Vol. 48, No. 4, 661–676.

SLACK RESOURCES AND THE PERFORMANCE OF

PRIVATELY HELD FIRMS
GERARD GEORGE
University of Wisconsin—Madison

Empirical findings from publicly traded firms and behavioral arguments suggest a
positive influence of resource slack on financial performance. While this area has
remained unexplored in privately held firms, conceptual arguments indicate that
resource constraints may enhance performance. Longitudinal data on 900 privately
held firms confirm the differing influences of forms of slack on performance. Results
indicate that a combination of behavioral and resource constraints arguments are
necessary to explain the slack-performance relationship in privately held firms. The
implications of these findings for theories of resources and entrepreneurship are
discussed.

Organizations are continually challenged to fos- risk taking (Wiseman & Bromiley, 1996), innova-
ter growth and improve performance while endur- tion (Nohria & Gulati, 1996), and performance (Bro-
ing strong exogenous pressures and endogenous miley, 1991; Tan & Peng, 2003). Some researchers
constraints. Management scholars have offered have noted that explanations for the influence of
strategic and behavioral explanations of factors that different forms of slack on performance do not ar-
induce or impel organizations to compete and excel ticulate a priori arguments but tend to present post
in these evolving competitive landscapes. An emer- hoc rationalizations (Deephouse & Wiseman, 2000;
gent dialog within this paradigm is the role of re- Greve, 2003). Additionally, studies have focused
sources and their influence on managers’ aspira- on publicly traded firms and have largely ignored
tions. Resources act as inducements to experiment, privately held firms, leaving a gap in scholars’ un-
take risks, and make proactive strategic choices. derstanding of how financial slack may influence
Resources are also deployed to build capabilities performance in these firms. Privately held firms
that make firms competitive, maintain coalitions tend to be undercapitalized (Holtz-Eakin, Joulfaian,
that ensure the convergence of personal and organ- & Rosen, 1994a, 1994b) and the mechanisms by
izational goals, and act as buffers in periods of which slack influences performance may vary, as
economic duress. Given this critical role, the pres- managers of privately held firms differ in their de-
ence or absence of excess resources and their im- cision-making process from their counterparts in
pact on performance carries substantive implica-
public firms (Busenitz & Barney, 1997). In this ar-
tions for scholarship in organizational theory and
ticle, I articulate differing relationships between
the practice of management.
the forms of financial slack and the performance of
Slack is potentially utilizable resources that can
privately held firms.
be diverted or redeployed for the achievement of
Two theoretical streams help explain the role of
organizational goals. These resources vary in type
slack in privately held firms: research on the be-
(e.g., social or financial capital) and form (e.g., dis-
cretionary or nondiscretionary). Studies have used havioral theory of the firm (Cyert & March, 1963;
financial slack in different forms as a predictor of March, 1994), and the resource constraints litera-
ture (Baker & Nelson, 2005; Mosakowski, 2002).
Behavioral theory treats firms as coalitions of ac-
The author thanks Don Bergh and the three anony- tors, and its proponents argue that slack provides
mous reviewers of AMJ for their constructive reviews. opportunities for managers to appease their politi-
The comments of participants at the UW Management cal coalitions by allowing parties to pursue their
and Human Resources Seminar Series, Phil Bromiley, own agendas (Cyert & March, 1963). Researchers
Frederic Delmar, Larry Hunter, Andrew King, Anne
have amplified this concept by arguing that slack
Miner, Don Schwab, Ann Terlaak, Charlie Trevor, Johan
relaxes internal controls and creates funds that can
Wiklund, and Shaker Zahra helped improve the quality
of this work. The data support from Boris Nenide, Chris be redirected toward projects with uncertain out-
Lang, and Yanfeng Zheng is much appreciated. An ear- comes, fostering an environment for innovation.
lier version was published in the Best Paper Proceedings Along these lines, evidence indicates a positive
2003 of the Academy of Management. effect of slack on the innovation and performance
661
662 Academy of Management Journal August

of public firms (Bromiley, 1991; Damanpour, 1987; with substantial resource endowments, while oth-
Greve, 2003). ers are disadvantaged at founding and are born out
Alternatively, according to studies that can col- of scarcity. Young firms may be particularly vul-
lectively be termed the resource constraints litera- nerable to the quantity of their resource endow-
ture, firms with fewer resources are likely to lever- ments. Excess resources permit investments in de-
age them more efficiently (Baker & Nelson, 2005; veloping capabilities to overcome the liabilities of
Starr & Macmillan, 1990). The claim is that re- newness in start-ups, thereby increasing the prob-
source constraints alter the behavior by which re- ability of survival (Hannan, 1998). As firms evolve
sources are garnered and expended, forcing manag- with age, they may require resources to renew ca-
ers to improve allocative efficiency. While the pabilities or build new competencies, whereas
resource constraints literature has tended to focus other established firms may have fewer resource
on entrepreneurial firms, I believe that the theoret- demands. Therefore, studying the impact of age
ical assumptions of resource constraints, private provides insights into the effectiveness with which
ownership, and private ownership’s implications young and old firms deploy their slack resources to
for managerial behavior are similar for entrepre- improve performance.
neurial and privately held firms. Most entrepre- Slack resources allow firms to adapt to complex
neurial firms are privately held, but entrepreneur- competitive landscapes (Levinthal, 1997), thereby
ial firms typically have higher growth rates than impacting firm performance. Industry complexity
other private firms. For example, studies using pri- refers to the competition in an industry that stems
vately held firms and entrepreneurial firm samples from concentration, or the market share dominance
concur that resources are sparse in both types of of one or more firms (Dess & Beard, 1984). The
firms and elicit “bootstrapping” modes of manage- presence of adequate or excess resource endow-
rial responses (Baker, Pricer, & Nenide, 2000; ments provides the flexibility for a firm to decide
Holtz-Eakin et al., 1994a, 1994b). Similarly, entre- on a course of action when trying to adapt to its
preneurial new ventures and privately held firms environment (Thompson, 1967). Slack resources
exhibit comparable managerial responses to re- also help buffer firms from environmental shocks
source utilization, as ownership structures in these and influence the enactment of strategies. Firms
firms are homogeneous and foster risk taking may adapt to industry complexity by reconfiguring
(George, Zahra, & Wiklund, 2005). Accordingly, I their resource endowments to form new capabili-
develop arguments using both behavioral theory ties or by shifting resources into or away from their
and the resource constraints literature. existing markets. Firms with larger slack resource
Some studies have used economic arguments to endowments are more likely to have freedom in
articulate the influence of slack on firm perfor- their responses to competitor strategies, thereby in-
mance (e.g., Deephouse & Wiseman, 2000). Popular fluencing performance. This study extends existing
among them are agency theory, according to which literature by capturing the influence of age and
the interests of firms’ owners (principals) and man- industry complexity as moderators of the slack-
agers (agents) may diverge (Jensen & Meckling, performance relationship.
1976), and X-efficiency theory, according to which The significance of this research is threefold. The
firms are necessarily inefficient in the allocation of impact of resource deployment on the achievement
resources (Leibenstein, 1980). Agency theory may of organizational goals is fundamental to theories of
not be as applicable in privately held firms as in how organizations act, evolve, and perform. A ma-
large public firms because in private firms, princi- jor contribution of this study is to distinguish
pals and agents are more likely to be the same among forms of slack and to articulate the pattern
individuals (Fama & Jensen, 1983). Though there of their relationships with performance. Second, it
are recent conceptual arguments that some agency is the first longitudinal study of slack in privately
relationships may exist in family-owned private held firms. Even though there are more than 25
businesses, empirical validation is limited. There- million such firms, generating over 70 percent of
fore, I build my arguments using organizational U.S. employment (Small Business Administration
theories on slack. In some instances, I reconcile [SBA], 2003), previous studies of slack have not
these arguments with X-efficiency theory. focused on the relevance and patterns of slack in
Scholars have articulated two critical influences privately held firms. There are substantial difficul-
on an organization’s ability to utilize resources ef- ties with collecting reliable data for privately held
fectively, namely, age and environment (Stinch- firms. Their owners tend to be more reluctant to
combe, 1965; Thompson, 1967). Age is an impor- share financial and other information than the
tant moderator of the effectiveness with which managers of publicly traded firms, creating barriers
firms deploy resources. Some firms are created that could explain the low number of large-sample
2005 George 663

longitudinal studies. Given a void in behavioral thors (1994a) found that when extra resources were
explanations of privately held firm actions, some introduced through windfall gains such as inheri-
authors have called for a systematic investigation of tance, these businesses tended to have higher sur-
this topic (Baum, Locke, & Smith, 2001). Finally, I vival rates and to disclose earnings increases of up
examine the effects of age and environment on the to 20 percent in subsequent years. These authors
slack-performance relationship. Investigating the attributed this increase in earnings to productive
nexus of organizational and environmental influ- investments, in capital equipment, for example,
ences on the accumulation and deployment of suggesting that private firms are undercapitalized
slack resources strengthens the contributions of and resource infusion enhances performance. Al-
this study to management theory and practice. though researchers have found that in public firms,
low- and high-discretion resources vary in their
impacts on performance, it is unlikely that similar
THEORY DEVELOPMENT
relationships hold in privately held firms. Tan and
Slack is used to stabilize a firm’s operations by Peng (2003) noted that organizational theory was
absorbing excess resources during periods of useful to explain a positive relationship between
growth and by allowing firms to maintain their high-discretion slack and performance, while
aspirations and internal commitments during peri- agency theory could better explain the negative
ods of distress (Cyert & March, 1963; Levinthal & performance impact of low-discretion slack. These
March, 1981; Meyer, 1982). Slack provides that authors argued that managers of public firms
cushion of actual or potential resources that allows tended to exhibit agency problems when low-dis-
an organization to adapt successfully to internal cretion slack was available, perhaps by implement-
pressures for change in policy as well as to initiate ing excessive diversification. However, as agency
changes in strategy (Bourgeois, 1981). Through this issues are minimized in privately held firms be-
dual internal and external role, slack influences cause owners tend to be managers as well, diver-
performance. gence in the effects of high- and low-discretion
Only three conceptual studies have articulated a resources is unlikely. Therefore, I expect both low-
rationale for the existence of forms of slack (Bour- and high-discretion slack to be positively related to
geois, 1981; Bourgeois & Singh, 1983; Sharfman, performance, as the presence of slack implies that
Wolf, Chase, & Tansik, 1988). These studies have these firms are adequately capitalized rather than
forwarded classifications of slack based on mana- undercapitalized.
gerial discretion in the deployment of resources. Behavioral arguments suggest that slack en-
Sharfman and colleagues suggested that slack re- hances experimentation and risk taking, which in-
sources are anchored along a continuum of mana- fluence the innovativeness (Nohria & Gulati, 1996)
gerial discretion, wherein “low-discretion slack” and performance (Bromiley, 1991; Singh, 1986) of
provides less flexibility to managers and their stra- large firms. Therefore, I see two reasons for the
tegic options. These authors noted that absorbed positive effect of both high- and low-discretion
slack, or excess costs in specialized assets, is low slack on the performance of privately held firms.
discretion, while unabsorbed slack is high discre- First, slack eases capital restrictions and improves
tion. Though the discretionary dimension is con- the strategic choices of managers for investments
tinuous, these authors argued that low- and high- with positive performance implications. Second, it
discretion slack coexist and often are employed allows experimentation and risk taking, which may
together, implying nonexclusivity in forms of slack also have positive performance consequences.
resources. Examples of high-discretion financial re- However, performance declines are likely to be
sources are cash and receivables, and low-discre- seen at higher levels of slack. Slack may insulate a
tion resources may include debt, fixed assets, and firm from exogenous shocks (Thompson, 1967), en-
excess capacity. It is important to recognize that gendering managerial complacency or irrational
both low- and high-discretion slack resources offer optimism. Firms with large resource reserves might
implicit benchmarks for managers of their own be less impelled to undertake initiatives through
firm’s resources as compared to their competitors’. experimentation, which may decrease performance
This comparison of absolute levels of resources because new entrepreneurial opportunities are not
may motivate managers’ strategic actions and com- exploited. Another possibility is that managers may
petitive behaviors. become overly optimistic and implement inappro-
In privately held firms, the allocation and utili- priate strategic actions (Cooper, Dunkelberg, &
zation of resources are likely to dominate manage- Woo, 1988; de Meza and Southey, 1996) that de-
rial decisions. Using the personal tax returns of crease performance (Bateman & Zeithaml, 1989).
private business owners, Holtz-Eakin and coau- Private managers may be susceptible to biases in
664 Academy of Management Journal August

decision making such as the planning fallacy but having few resources, like many growth-ori-
(Kahneman & Lovallo, 1994) and escalation of com- ented firms, may enact materially different strate-
mitment (Ross & Staw, 1993). The planning fallacy gies from firms with greater resource availability
consists of overconfidence about the duration and but lower demands. This comparison of availability
viability of projects, and escalation of commitment and demand is pertinent to studying privately held
involves defending and continuing a course of ac- firms as well as new ventures where entrepreneurs
tion in spite of negative outcomes. Both biases are cobble together resources to satisfy needs (Baker &
likely to occur at high levels of slack. If managers Nelson, 2005). Therefore, I develop arguments for
perceive that their absolute levels of both high- and transient slack, as it provides valuable and relevant
low-discretion slack far exceed those of competi- insights into the impact of resources on perfor-
tors, they are likely to be more optimistic about mance, especially in entrepreneurial firms.
courses of action and may inadvertently implement Behavioral theory and resource constraints argu-
strategic actions that decrease performance. There- ments appear to diverge with regard to predictions
fore, for transient slack. Behavioral arguments suggest
that resource demand and its availability are inter-
Hypothesis 1a. In privately held firms, finan- twined. That is, slack not only reflects absolute
cial performance will increase and then de- levels across peer firms but also is relative to inter-
crease with increases in high-discretion slack. nal needs such as maintaining coalitions (Cyert &
Hypothesis 1b. In privately held firms, finan- March, 1963; March, 1994). Both the absolute and
cial performance will increase and then de- relative levels of slack influence managerial choice,
crease with increases in low-discretion slack. engendering experimentation and risk taking that
have positive performance consequences. Here, be-
Dynamism in the generation and deployment of havioral arguments lead to a positive relationship
slack resources is important to the evolution of between transient slack and performance.
managerial behavior and firm strategy (Greve, 2003; However, the literature on resource constraints
Levinthal & March, 1981). Prior studies have fo- suggests that firms with fewer resources than their
cused on “absolute” levels of slack rather than operational demands require are likely to be more
slack “relative to demand.” Although organiza- efficient as they find ways to leverage and stretch
tions’ aspiration levels change with a concomitant their available resources. Supporting the resource
need to generate slack (March, 1994), researchers constraints argument, some authors have hypothe-
have not explored the implications of resource de- sized that firms bootstrap their limited resources to
mand on the slack-performance relationship. Even achieve goals (Levi-Strauss, 1966; Starr & MacMil-
though this availability-demand concept has been lan, 1990). In a cross-sectional sample of 177 pri-
theorized (Bourgeois & Singh, 1983; Sharfman et vately held firms, Baker, Pricer, and Nenide (2000)
al., 1988), the tendency in empirical analyses has found that undercapitalized firms “outperformed”
been to emphasize the excess resources of a firm those with excess capital. These authors inferred
relative to its industry or comparison group, rather that resource constraints bridled the optimism of
than excess resources relative to the focal firm’s managers. Economic arguments of X-efficiency
operational demands. (Leibenstein, 1980) are also consistent with the re-
Here, I introduce transient slack, defining it as source constraints literature. Leibenstein suggested
excess resources available after resource demands that firms are inefficient in the deployment of re-
for operations have been met. The concept of tran- sources. Therefore, if demands for resources exceed
sient slack emphasizes the ephemeral nature of their availability, firms are likely to be more effi-
slack, whereas both high- and low-discretion slack cient in the deployment of resources to enhance
capture absolute levels (as compared to peer firms’ performance.
resources). This distinction is theoretically mean- My view is that behavioral arguments and re-
ingful and important for two reasons. First, tran- source constraints arguments are not necessarily
sient slack separates resource availability from the incongruent but operate at different slack levels.
resource demands placed on a system. By doing so, When transient slack is negative (demand substan-
it emphasizes the temporal patterns of an organiza- tially exceeds availability), privately held firms
tion’s resource generation and deployment profiles, may bootstrap and find more efficient and effective
as the goals and the needs of an organization evolve uses for limited capital. However, at higher levels
in a dynamic competitive landscape. Second, one of transient slack (availability substantially exceeds
can uniquely identify the performance effects of demand), firms may begin to experiment and be-
resource availability and resource demand using come proactive in their strategic choices. At zero or
transient slack. That is, firms facing high demands marginally positive levels of transient slack (avail-
2005 George 665

ability equals demand), the incentives to experi- had the opportunity to experiment with different
ment or bootstrap are minimal. Combining these types of resources and select the ones that best fit
arguments, I expect a curvilinear relationship in their demands. These firms, owing to their matu-
which performance is high when transient slack is rity, have more predictable needs and can better
substantially positive or substantially negative but envision slack deployment to improve perfor-
in which performance is low when transient slack mance. In sum, older firms are more likely to pos-
is near zero. Given that transient slack is a function sess large amounts of slack than younger firms and
of resource availability and demand, I hypothesize to be aware of their resource demands, conditions
separate relationships: that help the older firms generate and deploy slack
effectively. Therefore, I posit the following moder-
Hypothesis 2a. In privately held firms, resource
ated effect:
availability will be related in a curvilinear
manner with financial performance. Specifi- Hypothesis 3. In privately held firms, the im-
cally, performance will decrease and then in- pact of slack on performance will be more pos-
crease with increases in resource availability. itive in older firms than in younger firms.
Hypothesis 2b. In privately held firms, resource
demand will be related in a curvilinear manner
The Influence of Industry Complexity
with financial performance. Specifically, per-
formance will increase and then decrease with The competitive environment in which firms in-
increases in resource demand. teract plays a critical role in their survival and
performance (Levinthal, 1997). Publicly held and
privately held firms compete in overlapping prod-
The Influence of Age
uct-markets and may use similar distribution chan-
A unique moderator of the slack-performance re- nels and suppliers. Therefore, it becomes critical
lationship is likely to be the age of a firm. Younger for firms to adapt their strategies to their competi-
firms may behave differently in evolving land- tive environments. Industry complexity signifi-
scapes than older, more established firms. Slack is cantly influences the enactment of strategic re-
also time-dependent in both its accumulation and sponses. As noted above, industry complexity
its deployment (Cyert & March, 1963; Sharfman et refers to the competition arising from high concen-
al., 1988; Thompson, 1967). Accordingly, age and tration of market share among a few dominant play-
slack levels are likely to be positively related. Be- ers (Dess & Beard, 1984). For example, large firms
havioral arguments suggest that as firms grow control the market and restrict smaller players’ stra-
older, the number of their internal political coali- tegic choices, including pricing power, branding,
tions also increases. Managers anticipate and de- and leverage in distribution and logistics. In com-
ploy slack as a means to appease these coalitions. plex industries, slack protects an organization from
For example, managers induce experimentation environmental pressures. By being insulated, the
and risk taking by funding multiple projects that managers of privately held firms are not compelled
have profit potential and by searching for alterna- to enact responses to the strategic moves of com-
tive competencies (Levinthal & March, 1981; petitors that might decrease financial performance.
March, 1994). By increasing experimentation, these In complex environments, the managers of pri-
older firms are likely to enjoy higher performance vately held firms are also likely to be constrained in
than younger firms that possess fewer resources their strategic choices. When slack is present, these
with which they can experiment. managers perceive it as some guarantee of firm
Scholars have suggested that resources assume survival and execute “satisficing” rather than per-
greater importance in young firms (e.g., Stinch- formance-maximizing strategies (March, 1994;
combe, 1965). Some have argued that substantial Sharfman et al., 1988). Smith, Grimm, Gannon, and
resource endowments at a younger age may provide Chen (1991) found a negative relationship between
for greater investments in capability development, slack and the probability of reacting to competitor
enhancing a firm’s prospects for survival (Hannan, actions: managerial responses were attenuated be-
1998). However, young firms tend to be resource- cause of the buffer of extra resources. When slack is
constrained and to suffer from the “liability of new- high, managers may deny that solutions exist and
ness,” and they are thus less likely to have the become rigid in their strategies (Staw, Sandelands,
experience to predict their resource needs. There- & Dutton, 1981), stances that negatively impact per-
fore, young firms are likely to make less efficient formance. This behavioral reasoning is consistent
use of resource slack than older firms. Sharfman with economic arguments that firms are inefficient
and coauthors (1988) argued that older firms have in their deployment of resources (Leibenstein,
666 Academy of Management Journal August

TABLE 1
Descriptive Statistics and Correlationsa
Variablesa Mean s.d. 1 2 3 4 5 6 7 8 9 10 11 12

1. Firm size 16.54 2.21

2. Industry profitability 0.06 0.04 ⫺.02
3. Number of competitors 185.10 98.50 .15 .12
4. Size of competitors 3,787.50 2,622.00 .11 ⫺.25 ⫺.25
5. Industry complexity 0.21 0.23 ⫺.13 .23 ⫺.67 ⫺.09
6. Number of plants 0.50 0.50 .60 .02 .19 .02 ⫺.11
7. Firm age 19.30 19.40 .31 ⫺.03 ⫺.12 .08 .00 .13
8. Family management 0.32 0.46 ⫺.30 .05 ⫺.35 .11 .29 ⫺.23 .07
9. High-discretion slackb, c 41.10 305.30 .34 .00 .09 .02 ⫺.05 .12 .21 ⫺.09
10. Low-discretion slackd 0.00 24.29 ⫺.02 .00 ⫺.00 ⫺.00 .00 ⫺.01 ⫺.00 .02 ⫺.03
11. Resource availabilityc ⫺20.10 781.80 ⫺.09 ⫺.01 .01 ⫺.03 .00 ⫺.03 ⫺.22 .02 ⫺.40 .02
12. Resource demandc 104.80 768.20 .37 ⫺.01 .05 .04 ⫺.05 .13 .27 ⫺.08 .84 .01 ⫺.61
13. Performancec 147.08 707.70 .53 .01 .09 .03 ⫺.07 .19 .39 ⫺.12 .67 .01 ⫺.27 .71

a
Number of observations ⫽ 3,598. Correlations above .03 are significant at p ⬍ .05. Year and industry dummies are not reported.
b
Cash.
c
Millions of dollars.
d
Debt-to-equity ratio.

1980), and an industry environment where profit- We used financial ratios for calculating slack, as
ability is low is likely to exacerbate this perfor- has been done in other studies (Bromiley, 1991;
mance effect. Therefore, I posit a second moderated Deephouse & Wiseman, 2000; Singh, 1986). The
effect: use of financial ratios follows from arguments that
financial data provide verifiable indicators of man-
Hypothesis 4. In privately held firms, the im-
agerial behavior (Bourgeois, 1981; Tan & Peng,
pact of slack on performance becomes more
negative in high-complexity industries than in 2003). Though slack can exist in organizational ca-
low-complexity industries. pabilities and other assets, it is difficult to capture
such slack in a longitudinal sample of private
firms. While surveys might provide finer-grained
DATA AND MEASURES measures, sampling and temporal variability would
A sample with diverse industry structures and raise concerns of measurement error and validity. I
competitive conditions was developed through se- calculated the financial measures of slack for each
lection of five technology-intensive industries (Na- of the four years.
tional Science Foundation, 2000) and five non- For high-discretion slack, I measured the level of
technology-intensive industries, and 43 subsectors; cash reserves in a given year. Cash is the most
industry was defined at the three-digit SIC level, easily deployed resource and provides managers
and subsector, at the four-digit level. The industries the greatest degree of freedom in allocating it to
included were meat-packing products, textiles, pa- alternate uses. For low-discretion slack, I measured
per and packaging, wood products, metal products, the debt-to-equity ratio. In private firms, debt ca-
computers and equipment, pharmaceuticals, elec- pacity is typically lower than in publicly traded
tronics, instrumentation, and telecommunications. firms because the private firms tend to be under-
By including multiple industries, the sample incor- capitalized (i.e., have less equity in the business).
porates high variation in industry and firm growth With higher debt levels, the freedom to reallocate
rates (Dess & Beard, 1984), which may be material resources or raise additional debt to meet expedient
to the accumulation and deployment of slack. The needs becomes restricted. Private firms also tend to
data were drawn from the Dun & Bradstreet data- raise debt and invest these funds in fixed assets;
base and supplemented with data from Ward’s debt thereby serves to temper managerial overcon-
Business Directory of Privately Held Firms. We se- fidence (deMeza & Southey, 1996). The measures
lected a four-year window (1994 –97) because slack are consistent with those adopted in other studies
may be accumulated and deployed over time. This of publicly traded firms (Bromiley, 1991; Deep-
window also maximized the number of firms re- house & Wiseman, 2000).
porting financial data for contiguous years, yield- For transient slack, I developed two measures to
ing a sample of 900 firms (3,598 observations). capture the ephemeral nature of slack. Resource
2005 George 667

availability was the permanent capital of a firm the average return on assets (ROA) of publicly
(i.e., owner’s equity plus debt) less fixed assets and traded firms operating in the same industry as a
other noncurrent assets. Resource demand was an sample firm. A high ROA would suggest that the
estimate of cash required (five days worth of sales) industry was profitable and had the potential to
plus accounts receivable and inventory less ac- generate slack. Number of competitors was the
counts payable. Though these measures are margin- number of public firms competing in the same in-
ally complex, they have advantages over other mea- dustry as the sample firm, and size of competitors
sures used in previous research. First, measures was the average number of employees of those pub-
such as the current ratio reflect short-term assets lic firms. Finally, I included industry dummy vari-
relative to short-term liabilities and do not accu- ables for each three-digit SIC code represented in
rately reflect a firm’s resource requirements. Sec- the sample. Data for variables on publicly traded
ond, separating availability from demand helps to companies were drawn from Standard & Poor’s
differentiate among patterns of resource deploy- COMPUSTAT.
ment and allows testing if higher performance re- The firm effects for which I controlled were firm
sults from resource constraint—that is, demand is size, number of plants, and type of management.
greater than availability. This measure is also used Because slack is likely to be size dependent, con-
in financial software packages for loan default pre- trolling for size effects is important. Transforma-
dictions (e.g., Fintel). tions such as the logarithmic value of slack vari-
Given the variety in industry context in the sam- ables are ineffective in controlling for size owing to
ple, it is likely that slack also differs across indus- attrition in sample size when the mean-centered
tries. Since slack is operationally defined as excess slack variables are transformed. Also, dividing
absolute levels of resources (Nohria & Gulati, 1996), slack variables by assets would have confounded
I calculated slack as the deviation from the mean of analyses here through common denominators in
each of the 43 industry subsectors. This procedure regression equations, leading to biased estimates.
provided a close estimate of excess resources com- Therefore, an independent control for firm size, the
pared to industry norms. The recalculated data logarithm of the value of sales in a given year, was
were used for the analyses. used here. For number of plants, I included a cat-
Firm age was years since incorporation. Industry egorical variable if a firm had multiple plant loca-
complexity was Herfindahl’s index of homogeneity tions (single plant ⫽ 0). This measure partially
in industry competition and concentration of re- controls for firm size, as larger firms are also likely
sources, measured as the sum of the squared market to have multiple locations. Finally, a categorical
shares of publicly traded firms in a sector (four- variable for family management was included,
digit SIC). This index assumes a value of 0 to 1, coded 1 if a firm was categorized as family-owned
where 0 reflects perfect competition among numer- by Ward’s Directory.
ous incumbents with infinitesimal market shares.
A high value represents large players and a dispro-
METHODS AND RESULTS
portionate concentration of sales that restricts
smaller firms’ ability to enhance performance by Using ordinary least squares to estimate panel
limiting their strategic choices (Dess & Beard, data can result in biased estimates because of un-
1984). High concentration is associated with high observed heterogeneity. Although a random- or
industry complexity. fixed-effects model can capture these relationships,
Performance was gross profit, calculated as rev- such a model would not account for autocorrela-
enues less the cost of goods sold. Gross profit, a tion and heteroskedasticity (unequal error variance
good indicator of profitability, was correlated with in the regression errors) in time series data. I
net income at .94 (p ⬍ .0001) and with net worth at adopted a cross-sectional time series feasible gen-
.96 (p ⬍ .0001). We refrained from using net eralized least squares (FGLS) regression model be-
income or net worth because possible variability cause it provided reliable estimates in the presence
in the tax treatment of income in private firms of heteroskedasticity and autocorrelation (Wool-
might undermine the reliability of estimates of dridge, 2002). Table 1 reports the descriptive sta-
performance. tistics and correlations between the variables. The
Control variables for year, industry, and firm ef- inclusion of slack variables and moderating effects
fects were also used. For year effects, I dummy- increased model fit and explanatory power signifi-
coded each year as a categorical variable to help cantly; Table 2 gives the results of the FGLS regres-
control for the performance effects of any general sion analyses. The “log-likelihood” increased from
economic event or trend. For industry effects, I –18,203.7 for the base model to –12,653.7 for the
included five measures. Industry profitability was full model.
668 Academy of Management Journal August

TABLE 2
Cross-Sectional Time-Series FGLS Estimates of Slack as a Predictor of Performancea

Variables Model 1 Model 2 Model 3

Firm size 26.00*** ( 2.21) 3.45*** (0.62) 2.85*** (0.69)

Industry profitability ⫺18.85 (25.38) ⫺6.49 (6.92) ⫺7.92 (6.78)
Number of competitors 0.03 ( 0.02) 0.01 (0.01) 0.007 (0.01)
Size of competitors ⫺.0007 ( 0.001) ⫺.0003† (0.0001) ⫺0.0002 (0.0001)
Industry complexity ⫺2.31 ( 9.69) 0.13 (2.84) 0.72 (2.81)
Number of plants ⫺17.30** ( 5.96) ⫺3.03* (1.48) ⫺2.36 (1.58)
Firm age 1.23*** ( 0.24) 0.18** (0.06) 0.08 (0.06)
Family management ⫺4.39 ( 5.75) 1.55 (1.44) 1.04 (1.46)
High-discretion slack 0.92*** (0.04) 0.32** (0.11)
High-discretion slack squared ⫺.00001 (0.00001) 0.00007 (0.0001)
Low-discretion slack 0.07*** (0.02) 0.01 (0.01)
Low-discretion slack squared ⫺0.0003*** (0.0001) ⫺0.00006† (0.00003)
Resource availability ⫺0.15*** (0.01) 0.01 (0.02)
Resource availability squared ⫺.00001*** ( .000003) ⫺0.000004 (0.000003)
Resource demand 1.22*** (0.03) 1.20*** (0.03)
Resource demand squared ⫺.0001*** ( .000004) ⫺0.00006*** (0.00001)
High-discretion slack ⫻ firm age 0.01*** (0.001)
High-discretion slack ⫻ age ⫺0.000001* (0.000001)
High-discretion slack ⫻ complexity ⫺1.79* (0.73)
High-discretion slack squared ⫻ complexity ⫺0.0001 (0.001)
Low-discretion slack ⫻ age
Low-discretion slack squared ⫻ age
Low-discretion slack ⫻ complexity
Low-discretion slack squared ⫻ complexity
Resource availability ⫻ age
Resource availability squared ⫻ age
Resource availability ⫻ complexity
Resource availability squared ⫻ complexity
Resource demand ⫻ age
Resource demand squared ⫻ age
Resource demand ⫻ complexity
Resource demand squared ⫻ complexity

Constant ⫺412.30*** ⫺56.71*** ⫺45.13***

Log-likelihood ⫺18,203.70 ⫺14,424.40 ⫺14,133.60
Wald ␹2 161.17*** 3,328.01*** 2,596.70***
Incremental change (␹2) 2,447.70*** 102.83***

a
Number of observations ⫽ 3,598; number of firms ⫽ 900. Year and industry controls are included but not reported. Unstandardized
†
p ⬍ .10
* p ⬍ .05
** p ⬍ .01
*** p ⬍ .001

To check for robustness across different model direction reported in Table 2, but the main effect
specifications, I conducted a time series fixed-ef- became statistically insignificant. I also analyzed
fects analysis, finding significant change in the lagged ordinary least squares (OLS) and time series
amounts of variance explained (baseline model, Arellano-Bond models with one- and two-year lags.
R2 ⫽ .30 [F ⫽ 21.16, p ⬍ .0001]; model with “main When lagged models were included, the number of
effects” added, R2 ⫽ .41 [F ⫽129.23, p ⬍ .0001]; full observations was reduced by half (data had to be
model, including interaction effects, R2 ⫽ .56 [F ⫽ available for four contiguous years). As heteroske-
184.20, p ⬍ .0001]). Also, to check if the results dasticity (as determined by a Sargan test) was sig-
were sensitive to firm size beyond the control vari- nificant, the Arellano-Bond models required two-
ables used, I included the product terms of the step estimators that did not provide reliable
slack variables and total firm assets as predictors. coefficient estimates. Since heteroskedasticity is a
The pattern and significance of the results did not concern, I report the FGLS estimates controlling for
change, with the exception of low-discretion slack, heteroskedasticity and autocorrelation across panels
where the squared term remained significant in the using STATA statistical software.
2005 George 669

TABLE 2
Continued

Model 4 Model 5 Model 6 Model 7

3.34*** (0.61) 1.41*** (0.42) 2.04*** (0.59) 2.39*** (0.45)

⫺6.67 (6.91) ⫺6.09 (3.96) ⫺8.02 (6.65) ⫺20.74*** (4.14)
0.01† (0.01) 0.003 (0.01) 0.01 (0.01) 0.003 (0.004)
0.0003† (0.0001) ⫺0.0001 (0.0001) ⫺0.0002 (0.0001) ⫺0.0003*** (0.0001)
0.42 (2.81) ⫺0.40 (1.94) 3.85 (3.06) 0.91 (1.82)
⫺2.91* (1.45) ⫺0.78 (0.98) ⫺1.15 (1.41) ⫺1.19 (1.03)
0.17** (0.06) 0.06 (0.04) 0.15* (0.06) 0.04 (0.03)
1.55 (1.42) 1.32 (0.90) 1.23 (1.40) 0.70 (0.83)
0.93*** (0.03) 0.43*** (0.04) 0.63*** (0.04) ⫺0.49*** (0.10)
⫺0.00001 (0.00001) ⫺0.00004** (0.00001) 0.00001 (0.00001) 0.0002** (0.0001)
0.07 (0.04) 0.02 (0.01) 0.04* (0.02) 0.01 (0.02)
⫺0.0004** (0.0001) ⫺0.0001*** (0.00003) ⫺0.0002*** (0.0001) 0.0001* (0.0001)
⫺0.15*** (0.02) 0.08*** (0.02) 0.06 (0.04) 0.34*** (0.04)
⫺0.00001*** (0.00001) 0.0001*** (0.00001) ⫺0.00005* (0.00002) 0.0001* (0.00003)
1.22*** (0.03) 1.26*** (0.03) 1.35*** (0.04) 1.50*** (0.05)
⫺0.0001*** (0.000004) ⫺0.0001*** (0.00002) ⫺0.00005* (0.00002) ⫺0.0002*** (0.00002)
0.01*** (0.002)
⫺0.000002* (0.000001)
2.50*** (0.66)
⫺0.003** (0.001)
⫺0.003 (0.003) 0.0001 (0.002)
0.00001 (0.00001) ⫺0.000009 (0.00001)
⫺0.02 (0.12) ⫺0.01 (0.08)
0.0003 (0.001) ⫺0.0001 (0.001)
⫺0.002*** (0.001) ⫺0.004*** (0.01)
⫺0.000001*** (0.0000001) ⫺0.000001*** (0.0000002)
⫺0.94*** (0.27) ⫺0.77*** (0.26)
0.0004* (0.0002) 0.0001 (0.0001)
0.002** (0.001) ⫺0.0001 (0.001)
0.000001*** (0.0000002) 0.000001*** (0.0000002)
⫺0.95*** (0.29) ⫺1.58*** (0.28)
⫺0.00005 (0.0002) 0.00009 (0.0002)

⫺55.33*** ⫺22.96*** ⫺35.45*** ⫺35.57***

⫺14,427.10 ⫺13,052.10 ⫺14,355.40 ⫺12,653.70
3,400.40*** 4,239.60*** 2,844.06*** 7,870.50***
2.39 104.73*** 36.01*** 419.38***

coefficients are reported; standard errors are in parentheses.

As for the individual hypotheses, the nonlinear source availability and performance was signifi-
relationship between high-discretion slack and per- cant. Interestingly, both the linear and squared
formance was not supported (Hypothesis 1a). The terms were negative and significant (p ⬍ .001),
linear positive term was significant (p ⬍ .001), but findings that were contrary to the hypothesized
not the squared term (model 2, Table 2). Hypothesis increase in performance at high levels of resource
1b, stating an increase and subsequent decrease in availability (Hypothesis 2a). Hypothesis 2b, which
performance with increases in low-discretion states that there will be a curvilinear relationship
slack, was supported (p ⬍ .001). Since transient between resource demand and performance with a
slack was composed of resource availability and positive linear coefficient (p ⬍ .001) and a negative
resource demand, testing their individual effects squared term, was supported (p ⬍ .001).
(rather than deriving a difference score measure) To test the interaction effects of Hypotheses 3
also allowed the parameter estimates to adjust in- and 4, I included the product terms of age and
dependently rather than forcing the estimates to be complexity with the different forms of slack. Be-
equal. The curvilinear relationship between re- cause I hypothesized curvilinear relationships for
670 Academy of Management Journal August

FIGURE 1a
Moderating Effects of Age on the Relationship between High-Discretion Slack and Performancea

FIGURE 1b
Moderating Effects of Age on the Relationship between Resource Availability and Performancea

a
Performance and slack were measured in millions of dollars.

the main effects, I included the product terms of ing effects of industry complexity on the slack-
moderators for the linear as well as the squared performance relationship (Hypothesis 4) were sup-
terms of the slack variables. By doing so, I was able ported for both high-discretion and transient slack
to identify positive and negative concave relation- but not for low-discretion slack. As seen in Figures
ships (Cohen & Cohen, 1983). The squared terms 2a–2c, the slack-performance relationship becomes
provide a more complete explanation by prevent- negative (has a steeper negative slope) in more
ing misinterpretation of effects due to linearity and complex than in less complex industry environ-
additivity in correlated variables (Cortina, 1993). ments. Unlike age, for which the squared product
The moderating effect of age on the slack-perfor- terms were significant, industry complexity had
mance relationship (Hypothesis 3) was supported significant squared product terms only for resource
in both high-discretion (model 3) and transient availability, and not for the other forms of slack.
slack measures (resource availability and resource The implications are discussed next.
demand, model 5) but not in low-discretion slack
measures (model 4). As illustrated in Figure 1a, the
DISCUSSION AND IMPLICATIONS
relationship between high-discretion slack and per-
formance is stronger (has a steeper positive slope) This study provides valuable insights into the
in older firms than in younger firms. The moderat- accumulation and deployment of resources to
2005 George 671

FIGURE 2a
Moderating Effects of Industry Complexity on the Relationship between
High-Discretion Slack and Performance

FIGURE 2b
Moderating Effects of Industry Complexity on the Relationship between
Resource Availability and Performance

a
Measured in millions of dollars.

achieve performance in privately held firms and firms using longitudinal data. The finding that
enriches the ongoing dialog on the importance of when resource demands exceed availability, per-
resources. The study makes three important con- formance is likely to be higher in privately held
tributions to organizational theory. First, a major firms is meaningful and valuable for scholars,
contribution is to develop causal logic for the entrepreneurs, and managers of private firms.
influence of different forms of slack on firm per- Third, this study explores the implications of age
formance. Though included in studies examining and industry complexity for the slack-perfor-
the relationship between risk and return, a ration- mance relationship. The influence of age on slack
ale for the influence of slack has hitherto not informs the resource constraints and entrepre-
been the focus of much theory building and em- neurship literatures. Also, most studies have
pirical testing (Deephouse & Wiseman, 2000). tested the implications of slack and environment
Also within this domain, I have introduced the for strategic choices (Meyer, 1982; Smith et al.,
concept of transient slack, which captures slack 1991) but not for financial performance. By doing
availability relative to a focal firm’s resource de- so, this study makes strong theoretical and
mands. Second, this is the first study to test and empirical contributions by informing both the
find support for the impact of resource con- behavioral theory and the resource constraints
straints on the performance of privately held literatures.
672 Academy of Management Journal August

FIGURE 2c
Moderating Effects of Industry Complexity on the Relationship between
Resource Demand and Performance

a
Measured in millions of dollars.

Behavioral and Resource Constraints Roles in higher levels of performance. Resource availability
Entrepreneurship exhibits a slightly concave relationship with per-
formance: substantially negative (⫺1 standard de-
Is more better? Is less more? Proponents of be-
viation [s.d.] from the mean) resource availability
havioral theory have argued for the positive bene-
was associated with high performance ($240.1 mil-
fits of slack (Bourgeois, 1981; Cyert & March, 1963),
lion), and substantially positive (⫹1 s.d.) resource
while others have argued that having fewer re-
sources may improve performance (Baker & Nel- availability was associated with low performance
son, 2005; Mosakowski, 2002). My arguments here ($5.57 million), while the mean level of resource
combine both behavioral and resource constraints availability was associated with moderate perfor-
views. The results provoke a reframing of the the- mance ($131.9 million). Resource demand had a
oretical question for future research from the sim- greater impact on performance ($⫺788.1 to 1,036.8
ple question of whether slack is good for perfor- million at ⫺1/⫹1 s.d.) than resource availability.
mance to a more complex set of questions: How This finding is theoretically important because it
much of what form of slack is good for perfor- underscores that resource availability in excess of
mance? and When is slack good for performance? resource demand is unlikely to enhance perfor-
Two distinct findings from this study inform a mance in privately held firms. These results
behavioral perspective on resource slack. First, eco- encourage researchers to consider a distinction be-
nomic performance increases and then decreases tween “absolute-to-external peers” and “relative-
with an increase in low-discretion slack, but per- to-internal demands” forms of slack. Although
formance is positively related in linear manner to these results support the idea of resource con-
increases in high-discretion slack. Behavioral argu- straints having a positive performance effect, fur-
ments emphasize a linear positive relationship, em- ther exploration using samples of entrepreneurial
phasizing the principle that more slack is better for high-growth firms is necessary before conclusions
appeasing coalitions, experimentation, and risk about the nature and dynamism of this relationship
taking. My finding suggests that specific forms of can be drawn. The present results should be sug-
slack may have decreasing returns, depending on gestive enough to persuade scholars to pay close
the level of discretion that the slack provides man- attention to the research question and its sensitivity
agers, and that redeployment of large amounts of to the different forms of slack. In sum, behavioral
low-discretion slack can be counterproductive. arguments that suggest linear positive relationships
Second, when the resource demand and resource between all forms of slack and performance do not
availability components of transient slack are ana- receive unconditional support in privately held
lyzed separately, the results are intriguing and in- firms. The results suggest possible boundaries to
formative. Resource demand has a strong, positive the assumptions of behavioral theory arguments,
relationship with performance; expressed graphi- particularly the argument that managerial discre-
cally, it is slightly concave rather than inverted tion has positive performance consequences in all
U-shaped. Greater resource demand is related to contexts.
2005 George 673

Age and Industry Effects adds to the value of my findings. The magnitude of
effect for resource demand on performance appears
There are interesting organizational and environ-
to be higher than the impact of resource availability
mental findings in this study. The interaction be-
on performance. In low-complexity industries, per-
tween age and slack was significant in most cases,
formance changes from $⫺1,049.9 million to
with the exception of low-discretion slack. High-
$709.5 million at ⫺1/⫹1 s.d. of resource demand
discretion slack appeared to be of greater impor-
(Figure 2c). In high-complexity industries, perfor-
tance for older firms that have more predictable
mance changes from $⫺861.8 million to $461.1
resource needs. Age modified the slope of the rela-
million at ⫺1/⫹1 s.d. of resource demand. Here, as
tionship between slack and performance (Figures
resource demand increases, performance increases
1a and b). A shortfall (⫺1 s.d.) of high-discretion in tandem, but it does so at a slower rate when
slack had a greater negative consequence ($–106.2 complexity is higher. Taken together, findings on
million) for the performance of older firms than for resource availability and demand have competing
the performance of younger firms ($–7.91 million). influences on performance. In a broader pattern,
An excess (⫹1 s.d.) of high-discretion slack had a higher resource demand is associated with higher
greater positive effect on performance in older levels of performance, while moderate negative lev-
firms ($281.5 million) than in younger firms els of resource availability are associated with pos-
($213.7 million). Younger firms tended to exhibit itive performance levels. These moderated effects
sharper increases in performance when resource suggest that resource constraints mechanisms are
availability was high ($179.8 million) than when more complex and dependent on organizational
availability was low ($67.8 million). Even when and environmental factors than theorized in this
availability was negative relative to the industry paper and in other studies. Both age and complex-
average, younger firms tended to show positive per- ity interaction effects imply that resource con-
formance levels (Figure 1b). The impact of age on straints influence performance, but further research
resource availability in older firms was more mod- is required to fully explore the intricacies of this
est; there were marginal increases in performance argument.
from $102.3 million at low availability (⫺1 s.d.) to
$118.3 million at high resource availability (⫹1
s.d.). This finding suggests that the impact of slack Publicly Held versus Privately Held Firms
varies with age of a firm and the type of resource. Some studies of slack have combined behavioral
Interesting results in the moderating effects for and agency theory explanations of risk taking in
industry complexity emerged as well. Findings publicly held firms (Deephouse & Wiseman, 2000).
supported my hypothesis that an increase in indus- Although agency explanations may be appropriate
try complexity makes the relationship between in large, publicly held firms, the agency relation-
slack and performance more negative (Hypothesis ships in privately held firms may be different or
4). Further analysis revealed that the pattern of absent (Fama & Jensen, 1983). The theoretical and
influence on resource availability requires a more practical implications of slack levels in firms raise
complicated explanation (Figure 2b). In low-com- important and interesting questions. However, the
plexity industries, performance changes from lack of sampling and measurement consistency
$208.7 million to $60.9 million at ⫺1/⫹1 s.d. of across studies of slack forecloses comparison of
resource availability. However, in high-complexity public and private firms. It is likely that slack is
industries, performance changes from $521.1 mil- important for different reasons in public and in
lion to $⫺84.1 million at ⫺1/⫹1 s.d. of resource private firms. Future research will need to recon-
availability. We find that when resource availabil- cile the theoretical differences and the pattern of
ity is constrained (⫺1 s.d.), performance is substan- relationships between privately and publicly held
tially higher in high-complexity industries than in firms in a longitudinal setting.
low-complexity industries. When resource avail- My study focused on privately held firms, but
ability is high (⫹1 s.d.), performance is lower (more fine-tuning of the current sample could expand
negative) for high-complexity industries. The effects researchers’ limited knowledge of slack and its role
of complexity appear to be more pronounced when in organizational processes. For example, the dis-
resource constraints exist. This finding directly ad- tinction between nascent ventures, high-growth
dresses the managerial and theoretical relevance of firms (prior to initial public offering) and high-
resource constraints by articulating the impact of en- growth (post-IPO) firms may increase the theoreti-
vironment on slack deployment. cal richness of the slack construct. Similarly, the
The moderating effect of complexity on the rela- motivations and behaviors of managers may differ
tionship of resource demand and performance also for pre-IPO and post-IPO firms as the scrutiny of
674 Academy of Management Journal August

public markets and the pressing need to deliver at certain stages of growth. For instance, slack may
positive results every fiscal quarter is minimized in be critical during growth stages but may be absent
pre-IPO firms. Further research may help catego- then. The implications of slack for growth in entre-
rize the behavioral differences among these firms preneurial firms is an important issue that needs to
and may help further refine the logic for the slack- be addressed methodically.
performance relationship. Limitations aside, this is the first study to theo-
rize and test causal relationships between slack and
performance in a longitudinal sample of privately
Limitations and Conclusions
held firms. I found that the pattern of relationships
Despite the care taken with the analysis and the differs with the form of slack (high-discretion, low-
benefits of longitudinal study of privately held discretion, transient), and this finding contributes
firms, the present research has some limitations. to organizational theory that has, hitherto, ad-
First, though the classification using the discretion- dressed positive linear relationships between slack
ary nature of assets is appropriate for diverse forms and performance. The results support the resource
of slack, operational definition of slack using finan- constraints argument in privately held firms. Sup-
cial data does not account for other types of slack. port for this argument is seen especially when or-
For example, resource endowments in social capi- ganizational age and industry complexity are con-
tal, human capital, or organizational capabilities sidered as moderators. Taken together, the findings
can be extended for the influence of slack within offer preliminary evidence of the compatibility of
these types of resources and its impact on entrepre- behavioral and resource constraints arguments in
neurial outcomes such as new market entry. Be- explaining the dynamic of the slack-performance
cause firms need to invest resources to build organ- relationship in the context of privately held firms.
izational capabilities and, therefore, incur costs Although I concentrated on privately held firms,
before any payoffs are imminent (e.g., George, my findings on the pattern of resource demand and
2005), slack resource endowments may play a par- availability might also apply in public firms. Thus,
ticularly important role in the enactment of strate- this study contributes to pertinent discussion on
gies and the evolution of firms. Further investiga- the leveraging of organizational resources to en-
tion of the multiple forms of slack resources and hance performance through the adaptation and en-
entrepreneurial firms’ ability to leverage and de- actment of strategies in competitive environments.
ploy slack across potential alternate applications Future studies can leverage these findings to pro-
for a specific resource are interesting and fruitful vide comprehensive behavioral explanations for
avenues for scholarly inquiry. the patterns and temporal dynamism in the gener-
Second, the longitudinal design provides some ation and deployment of slack resources to achieve
temporal stability to these results, but these data organizational goals.
are restricted to a single time frame (1994 –97).
Though I controlled for year effects, an extension to
REFERENCES
this study might use panel data to account for dif-
ferent economic scenarios. Also, a longer time Baker, T., & Nelson, R. 2005. Making do with what’s at
frame could capture the dynamism in slack over hand: Entrepreneurial bricolage. Administrative
time. It is likely that slack levels change over time, Science Quarterly: In press.
and tracking this change might allow for a richer Baker, T., Pricer, R., & Nenide, B. 2000. When less is
theoretical argument than presented in this article. more: Undercapitalization as a predictor of firm suc-
Third, since slack and size are correlated, there are cess. In P. Reynolds, E. Autio, C. Brush, W. Bygrave,
likely to be variations across industries in the im- S. Manigart, H. Sapienza, & K. Shaver (Eds.), Fron-
tiers of entrepreneurship research: CD version.
pact of slack on performance. Though I controlled
Babson College: Babson Park, MA.
for industry effects, further studies might examine
multiple forms of slack in a single industry. For Bateman, T., & Zeithaml, C. 1989. The psychological
example, augmenting work on slack in network context of strategic decisions. Strategic Manage-
ment Journal, 10: 587–592.
resources in the biotech industry and its impact on
innovation (e.g., George, Zahra, & Wood, 2002), Baum, R., Locke, E., & Smith, K. 2001. A multidimen-
inclusion of financial and other forms of slack such sional model of venture growth. Academy of Man-
as human and social capital could provide a holis- agement Journal, 44: 292–303.
tic picture of an organization’s use of multiple re- Bourgeois, L. J. 1981. On the measurement of organiza-
source endowments. Finally, the influence of age tional slack. Academy of Management Review, 6:
on slack deployment is an understudied area. Slack 29 –39.
may assume greater levels of strategic importance Bourgeois, L. J., & Singh, J. 1983. Organizational slack
2005 George 675

and political behavior within top management izational mortality: Logical formalizations. Ameri-
groups. Academy of Management Proceedings: can Journal of Sociology, 104: 126 –164.
43– 49. Holtz-Eakin, D., Joulfaian, D., & Rosen, H. 1994a. Sticking
Bromiley, P. 1991. Testing a causal model of corporate it out: Entrepreneurial survival and liquidity con-
risk taking and performance. Academy of Manage- straints. Journal of Political Economy, 102: 53–75.
ment Journal, 34: 37–59.
Holtz-Eakin, D., Joulfaian, D., & Rosen, H. 1994b. Entre-
Busenitz, L., & Barney, J. 1997. Differences between en- preneurial decisions and liquidity constraints. Rand
trepreneurs and managers in large organizations: Bi- Journal of Economics, 25: 334 –347.
ases and heuristics in strategic decision making.
Jensen, M. C., & Meckling, W. H. 1976. Theory of the
Journal of Business Venturing, 12: 9 –30.
firm: Managerial behavior, agency costs, and owner-
Cohen, J., & Cohen, P. 1983. Applied multiple regres- ship structure. Journal of Financial Economics, 3:
sion/correlation analyses for the behavioral sci- 305–360.
ences. Hillsdale, NJ: Erlbaum.
Kahneman, D., & Lovallo, D. 1994. Timid choices and
Cooper, A. C., Dunkelberg, W. C., & Woo, C. Y. 1988. bold forecasts. In R. Rumelt, D. Schendel, & D. Teece
Entrepreneurs’ perceived chances for success. Jour- (Eds.), Fundamental issues in strategy: A research
nal of Business Venturing, 3: 97–108. agenda: 71–76. Boston: HBS Press.
Cortina, J. 1993. Interaction, nonlinearity and multicol- Leibenstein, H. 1980. Beyond economic man. Cam-
linearity: Implications for multiple regressions. Jour- bridge, MA: Harvard University Press.
nal of Management, 19: 915–922.
Levi-Strauss, C. 1966. The savage mind. Chicago: Uni-
Cyert, R., & March, J. 1963. A behavioral theory of the versity of Chicago Press.
firm. Englewood Cliffs, NJ: Prentice- Hall.
Levinthal D. A. 1997. Adaptation on rugged landscapes.
Damanpour, F. 1987. The adoption of technological, ad- Management Science, 43: 934 –950
ministrative, and ancillary innovations: Impact of
organizational factors. Journal of Management, 13: Levinthal, D., & March, J. 1981. A model of adaptive
675– 688. organizational search. Journal of Economic Behav-
ior and Organization, 2: 307–333.
deMeza, D., & Southey, C. 1996. The borrower’s curse:
Optimism, finance and entrepreneurship. Economic March, J. 1994. A primer on decision-making. New
Journal, 106: 375–386. York: Free Press.

Deephouse, D., & Wiseman, R. 2000. Comparing alterna- Meyer, A. 1982. Adapting to environmental jolts. Admin-
tive explanations for accounting risk-return rela- istrative Science Quarterly, 27: 515–537.
tions. Journal of Economic Behavior & Organiza- Mosakowski, E. 2002. Overcoming resource disadvan-
tion, 42: 463– 482. tages in entrepreneurial firms: When less is more. In
Dess, G., & Beard, D. 1984. Dimensions of organizational M. Hitt, D. Ireland, D. Sexton, & M. Camp (Eds.),
task environments. Administrative Science Quar- Strategic entrepreneurship: Creating an inte-
terly, 29: 52–73. grated mindset: 106 –126 (strategic management se-
ries). Oxford, England: Blackwell.
Fama, E. F., & Jensen, M. C. 1983. Separation of owner-
ship and control. Journal of Law and Economics, National Science Foundation. 2000. Industry, technol-
26: 301–325. ogy, and the global marketplace highlights, Indi-
cators 2000 (chapter 7): http://www.nsf.gov/sbe/srs/
George, G. 2005. Learning to be capable: Patenting and
seind00/access/c7/c7h.htm.
licensing at the Wisconsin Alumni Research Foun-
dation 1925–2002. Industrial & Corporate Change, Nohria, N., & Gulati, R. 1996. Is slack good or bad for
14: 119 –151. innovation? Academy of Management Journal, 39:
1245–1264.
George, G., Zahra, S., & Wiklund, J. 2005. Ownership and
the internationalization of small firms. Journal of Ross, J., & Staw, B. 1993. Organizational escalation and
Management, 31: 210 –233. exit: Lessons from the Shoreham nuclear plant.
Academy of Management Journal, 36: 701–732.
George, G., Zahra, S., & Wood, D. 2002. The effects of
business-university alliances on the innovative out- Sharfman, M., Wolf, G., Chase, R., & Tansik, D. 1988.
put and financial performance: A study of publicly Antecedents of organizational slack. Academy of
traded biotechnology companies, Journal of Busi- Management Review, 13: 601– 614.
ness Venturing, 17: 577– 608. Singh, J. 1986. Performance, slack, and risk taking in
Greve, H. 2003. Organizational learning from perfor- organizational decision making. Academy of Man-
mance feedback: A behavioral perspective on in- agement Journal, 29: 562–585.
novation and change. Cambridge, England: Cam- Small Business Administration. 2003. Administration’s
bridge University Press. economic plan (release # 03-02). Washington, DC:
Hannan, M. 1998. Rethinking age dependence in organ- SBA.
676 Academy of Management Journal August

Smith, K., Grimm, C., Gannon, M., & Chen, M. 1991. Orga- Wiseman, R., & Bromiley, P. 1996. Towards a model of
nizational information processing: Competitive re- risk in declining organizations: An empirical exam-
sponses and performance in the U.S. domestic airline ination of risk, performance and decline. Organiza-
industry. Academy of Management Journal, 34: 60–85. tion Science, 7: 524 –543.
Starr, J., & MacMillan, I. 1990. Resource cooptation via Wooldridge, J. 2002. Econometric analysis of cross-sec-
social contracting: Resource acquisition strategies tion and panel data. Cambridge, MA: MIT Press.
for new ventures. Strategic Management Journal,
11: 79 –93.
Staw, B., Sandelands, L., & Dutton, J. 1981. Threat rigid-
ity effects in organizational behavior—A multilevel
analysis. Administrative Science Quarterly, 26: Gerard (Gerry) George (ggeorge@bus.wisc.edu) is an as-
501–524. sistant professor of entrepreneurship at the University of
Stinchcombe, A. L. 1965. Social structure and organiza- Wisconsin—Madison. His research interests include ca-
tions. In J. G. March (Ed.), Handbook of organiza- pability development in organizations, international en-
tions: 142–193. Chicago: Rand McNally. trepreneurship, university science and technology trans-
fer, and the innovation and performance of start-ups and
Tan, J., & Peng, M. 2003. Organizational slack and firm
performance during economic transitions: Two stud- privately held firms. He serves as director of the applied
ies from an emerging economy. Strategic Manage- ventures program at the Weinert Center for Entrepreneur-
ment Journal, 24: 1249 –1264. ship at the UW School of Business.

Thompson, J. 1967. Organizations in action. New York:

McGraw-Hill.