Reassessing the Entrepreneurial Spinoff Performance Advantage: A Natural Experiment Involving a Complete Population

Abstract

Through the discovery and exploitation of a natural experiment comprised of a complete industry population, this paper presents empirical evidence challenging widely held beliefs related to intra-industry entrepreneurial spinoffs. Extant spinoff theory holds that knowledge and capabilities are transferred from parent-firms to spinoffs in hereditary fashion, endowing spinoffs with a performance advantage over de novo entrants. Our analysis of all 612 industry entrants, including 448 spinoffs, paints a dramatically different picture. In the context of a complete population, we find that: (a) de novo entrants actually outperform spinoffs; (b) parent-firm quality exerts no discernible influence on spinoff quality; and, (c) founder-specific experience, not parental lineage, is the primary driver of spinoff performance heterogeneity.

Full text

Frontiers of Entrepreneurship Research
Volume 32 |Issue 12
CHAPTER XII. STTEGY Article 2
6-9-2012
REASSESSING THE ENTREPRENEURIAL
SPINOFF PERFORMANCE ADVANTAGE: A
NATUL EXPERIMENT INVOLVING A
COMPLETE POPULATION
Richard A. Hunt
University of Colorado, richard.hunt@colorado.edu
Daniel A. Lerner
University of Colorado
is Paper is brought to you for free and open access by the Entrepreneurship at Babson at Digital Knowledge at Babson. It has been accepted for
inclusion in Frontiers of Entrepreneurship Research by an authorized administrator of Digital Knowledge at Babson. For more information, please
contact digitalknowledge@babson.edu.
Recommended Citation
Hunt, Richard A. and Lerner, Daniel A. (2012) "REASSESSING THE ENTREPRENEURIAL SPINOFF PERFORMANCE
ADVANTAGE: A NATUL EXPERIMENT INVOLVING A COMPLETE POPULATION," Frontiers of Entrepreneurship
Research: Vol. 32: Iss. 12, Article 2.
Available at: hp://digitalknowledge.babson.edu/fer/vol32/iss12/2

FRONTIERS OF ENTREPRENEURSHIP RESEARCH 2012
REASSESSING THE ENTREPRENEURIAL SPINOFF
PERFORMANCE ADVANTAGE: A NATURAL EXPERIMENT
INVOLVING A COMPLETE POPULATION

Richard A. Hunt, University of Colorado at Boulder, USA
Daniel A. Lerner, University of Colorado at Boulder, USA
Abstr Act
Through the discovery and exploitation of a natural experiment comprised of a complete industry
population, this paper presents empirical evidence challenging widely held beliefs related to intra-
industry entrepreneurial spinoffs. Extant spinoff theory holds that knowledge and capabilities are
transferred from parent-firms to spinoffs in hereditary fashion, endowing spinoffs with a perfor-
mance advantage over de novo entrants. Our analysis of all 612 industry entrants, including 448
spinoffs, paints a dramatically different picture. In the context of a complete population, we find
that: (a) de novo entrants actually outperform spinoffs; (b) parent-firm quality exerts no discern-
ible influence on spinoff quality; and, (c) founder-specific experience, not parental lineage, is the
primary driver of spinoff performance heterogeneity.
IntroductIon
Intra-industry entrepreneurial spinoffs (sometimes called “spinouts”) occur when employees
leave a corporate parent to start a new, completely independent company as an entry vehicle into
the same industry as their former employer (Klepper, 2001), without support or sponsorship from
the parent-firm. Entrepreneurial spinoffs are richly in evidence during periods of new industry
formation, appearing as both a driver of new industry opportunities and a primary beneficiary of
emerging industry opportunities (Garvin, 1983). It has become popular in recent years for scholars
studying intra-industry business spinoffs to invoke the language of procreation and genetics as an
explanatory model for spinoff creation and performance. An expanding set of studies supporting
a progeny model (Phillips, 2002) variously refer to the parent-child ties (Klepper, 2001) as
“spawning” (Gompers et al., 2005; Chatterji, 2009), “inheritance” (Nelson, 1991; Agarwal et al.,
2004), “organizational births” “children” and “offspring” (Dyck, 1997), “parenting” (Klepper &
Sleeper, 2005), and “heredity” (Dick et al., 2011). The use of proto-biological speak is evidence
that spinoff scholars believe they have secured sufficient empirical support to advance a widening
set of “stylized facts” (Klepper 2009) that provide a theoretical foundation for the subject, namely:
that spinoff founders learn lessons from their parents that are advantageously deployed towards
an improved likelihood of survival and the achievement of superior performance. As Klepper and
Sleeper asserted, “Firms can be thought of as giving birth to spinoffs, so that spinoffs have parents
from whom they inherit specific traits, in this case knowledge” (2005: 1303).
Central among these “stylized facts” is the widely held belief that entrepreneurial spinoffs
live longer and perform better than de novo entrants (Brittain & Freeman, 1986; Christensen,
1993; Stuart & Sorenson, 2003; Agarwal, Echambadi, Franco & Sarkar, 2004; Klepper & Sleeper,
2005; Chatterji, 2009). “Spin-outs have a survival edge in the market over other entrants as the
1
Hunt and Lerner: REASSESSING THE ENTREPRENEURIAL SPINOFF PERFORMANCE ADVANTAGE
Frontiers of Entrepreneurship Research 2012

STRATEGY
result of a combination of entrepreneurial flexibility and inherited knowledge,” noted Agarwal
and colleagues (2004: 519). It is also widely propounded that high-performing parent-firms serve
as a wellspring for high-performing spinoffs (Gompers, Lerner & Sharfstein, 2005; Eriksson &
Kuhn, 2006; Klepper, 2009; Elfenbein, Hamilton & Zenger, 2010). As Klepper and Sleeper asserted,
“Spinoffs inherit technical and market-based knowledge from their parents that shapes their
nature at birth” (2005:1303). Simply put, the common wisdom has become: “Better-performing
firms have better-performing intra-industry spinoffs” (Klepper & Thompson, 2010: 5).
On the face of it, these increasingly formalized assertions present a formidable case in
support of a hereditary theory of entrepreneurial spinoffs. And yet, this general acclamation of
spinoff superiority is vexed by a very simple, very perplexing fact: most spinoffs fail to become
substantively operational. In fact, spinoff failure rates often exceed those exhibited by the general
population of firms (Garvin, 1983; Klepper, 2002; 2007). The sheer number of entrepreneurial
spinoff failures raises several important research questions. First, if spinoff performance can be
most aptly portrayed as involving a hereditary linkage between parent-firms and their spawn, then
why do so many spinoffs fail? Second, if high-performing parent-firms are generally thought to
produce high-performing spinoffs, then why do many low-performing spinoffs come from high-
performing parents and many high-performing spinoffs come from low-performing parents?
The extreme heterogeneity in spinoff performance constitutes a theoretical fissure that can only
be resolved through the use of fine-grained, comprehensive data that allows close examination of
spinoff successes and failures. Using data from a natural experiment that arose through legislative
action related to the regulation of asbestos removal and disposal in Colorado, we directly address
the performance heterogeneity complications by employing a complete industry population.
In this fashion, our study breaks new ground in several important respects. First, we contribute
to entrepreneurial spinoff theory by providing the first substantive head-to-head comparison
between a complete population of spinoff entrants and de novo entrants. Our study provides
the circumstances and data necessary to test the foundational assumption that market entrants
who possess high-quality parental lineages outperform those that do not. Through this empirical
breakthrough, we offer a corrective course for the development of a more robust and encompassing
spinoff theory. In the absence of a clear spinoff performance advantage, existing theories are
unable to explain the high failure rates, especially failed spinoffs spawned from better-performing
parent-firms (Eriksson & Kuhn, 2006; Klepper, 2009). Our analysis materially enriches spinoff
theory by providing explanatory bases for failures as well as successes.
Secondly, we contribute to the knowledge-transfer literature (e.g. Kogut & Zander, 1992) by
addressing the effect-conflation issues that have bedeviled prior studies of nascent industries
(Aldrich & Fiol, 1994), including those examining spinoffs. By dissecting the drivers of spinoff
performance heterogeneity, we offer a more detailed evaluation of parent knowledge transfers to
spinoffs. Prior studies have used aggregated spinoff data, thereby failing to capture performance
heterogeneity. This has led to strong, but untested assumptions linking parent performance to
spinoff performance. By comparing firm-level heterogeneity to overall heterogeneity, we offer
the first meaningful test of whether or not transferrable knowledge is instrumental to spinoff
outcomes. Finally, we provide important insights for entrepreneurship research methods through
our sensible explication of unobservable non-linear relationships (Daniels and Hogan, 2008) by
using a natural experiment, rather than simulated data. Much of the prior literature in this realm
has come to rely upon a patchwork of explanations that are conceptually sound but functionally
disengaged from the actual challenge of assessing truncation effects. Through the innovative use
2
Frontiers of Entrepreneurship Research, Vol. 32 [2012], Iss. 12, Art. 2
Posted at Digital Knowledge at Babson
http://digitalknowledge.babson.edu/fer/vol32/iss12/2

FRONTIERS OF ENTREPRENEURSHIP RESEARCH 2012
of a natural experiment, we address the persistent dilemma of data truncation, which inherently
understates spinoff occurrence and overstates spinoff survival and performance.
In the following section, we provide additional context regarding spinoff theory and develop
a set of four hypotheses that are tested using OLS regression models. After detailing the database
and methods used in the study, we present the results of the data analysis associated with this
natural experiment. We conclude by reflecting on the implications for future study.
theoretIcAl dev elopm ent And hypothe ses
Spinoff Theory and Data Truncation
Spinoff scholars have drawn strong connections to a wide range of seminal management theories
for the purpose of advancing spinoff studies, including: evolutionary theory (Nelson & Winter,
1982; 1987), organizational learning (Cyert & March, 1963; Fiol & Lyles, 1985; Leavitt & March
1988), tacit and explicit knowledge transfer (Kogut & Zander, 1992; Teece & Pisano, 1994; Franco
& Filson, 2006), and a variety of economic-based (Geroski 1995) and sociology-based (Hannan
& Freeman, 1977; 1989; Aldrich & Fiol, 1994) explanations for market entry. Credible linkage to
these theoretical frameworks requires scholars to demonstrate the soundness of a hereditary-based
theory of spinoff performance. Unfortunately, comprehensive spinoff data has been notoriously
elusive. For the data that has been gathered to-date, analysis has been confounded by left-side data
truncation, meaning that observations on both the dependent variable and regressors are missing
(Cameron & Trivedi, 2005). Figure 1 displays a stylized representation of the truncation problem.
The individual data points represent the complete population of firms entering a hypothetical
market. The truncation threshold represents the beginning of the observation window. This
means studies that define the observation window through the use of trade catalogues (Klepper,
2001), venture capital financing (Chatterji, 2008) or industry journals (Agarwal et al., 2004) may
exclude early spinoff failures from the analysis.
In Figure 1, the mean performance (YO), comprised of only the observed values of y (yo),
overstates the operational performance and understates the entry rate of the complete population.
This is because there exists an unobservable non-linear relationship in the complete population
of observations. Since the unobserved values of y (yu) do not share a linear relationship with the
population of values comprising YO, conventional efforts to relate YO and YU will be impaired by the
inability to apply the distributional assumptions underlying parametric correction tools, such as
the Tobit Model. Corrections that relax distributional assumptions such as maximum-likelihood
estimation and the Heckman two-step estimator are marginally more robust, but suffer from
the notable liability of being “fragile to even very minor misspecification of error distributions”
(Paarsch, 1982; Cameron & Trivedi, 2005). When the boundaries of possible solutions are so
broad as to include the parametric estimators that are assuredly wrong and the non-parametric
estimators that are assuredly uninformative, then these features render the estimators difficult to
interpret (Tsiatis, 2006) and unsatisfying to apply to management research.
Given the formidable methodological challenges associated with any examination of spinoffs,
a natural experiment involving a complete population is perhaps the only tool through which
this complex array of issues can be comprehensively addressed. Due to of the manner in which
Colorado chose to implement the federal requirements governing asbestos abatement, there exists
3
Hunt and Lerner: REASSESSING THE ENTREPRENEURIAL SPINOFF PERFORMANCE ADVANTAGE
Frontiers of Entrepreneurship Research 2012

STRATEGY
a well-defined database, extending back 25 years to the industry’s inception. Equipped with the
data from this uncensored population, we advance two arguments (see Figure 2):
H1a: De novo entrants live as long as or longer than spinoff entrants.
H1b: De novo entrants perform as well as or better than spinoff entrants.
Spinoff Performance Heterogeneity
Prior studies have largely attributed the heterogeneity of performance among entrepreneurial
spinoffs to hereditary linkages between parent-firms and their respective spawn, saying in essence
that good parents produce good kids and bad parents produce bad kids. “Spinoffs will initially
have the same expected profits and survival prospects as their parents, thus more innovative and
long-lived parents will have more innovative and long-lived spinoffs” (Klepper, 2001:646; Franco
& Filson, 2006). Given the effects of left-side truncation, in which a great preponderance of failed
firms are never included in the analysis (Figure 1), this assertion is based more on speculation
than empirical observation. Absent from all previous spinoff studies is an analysis of the variance
within the group of spinoffs spawned from a shared parent-firm. If the variance is greater within
the cohort of spinoffs for a parent-firm than for the population of spinoffs, then this would cast
doubt upon the conclusion that better-performing parents spawn better-performing spinoffs.
Data limitations have hindered scholars’ ability to address this critical question. Meanwhile,
the methods used in prior studies have identified successful spinoffs across multiple industries,
but these same methods have produced a spinoff theory that is unable to explain the high failure
rate. Because of this, there is reason to doubt the extent to which existing research has accurately
captured the performance outcomes of spinoffs. Instead, by linking the spinoff story primarily to
technological know-how (Agarwal et al., 2004) and knowledge appropriation (Eriksson & Kuhn
2006; Klepper & Thompson, 2010), existing theory explains the few success stories, while failing to
account for the far-larger legion of failed firms.
The only meaningful way to directly test the purported linkages between parent-firm quality
and spinoff quality is by examining the extent to which high-quality parents produce high-quality
spinoffs, and low-quality parents produce low-quality spinoffs. Doing so requires the availability
of a complete, non-truncated population of industry actors. The natural experiment involving
the Colorado asbestos abatement industry provides such data. From its inception in 1986, through
2010, 100 parent-firms spawned 448 spinoffs. Among these, 35 parent-firms produced five or
more spinoffs, and thirteen of those firms produced ten or more spinoffs. The presence of several
dozen parent-firms, each producing statistically a significant spinoff cohort group, provides an
unprecedented opportunity to scrutinize spinoff performance heterogeneity. If the transfer of
knowledge and capabilities from parent to spinoffs is a critical source of performance advantage,
then the variance in performance for the cohort of spinoffs spawned by any given parent should
be less than the overall performance variance. In directly challenging the dominant association
between parent quality and spinoff quality, we predict that cohort group heterogeneity will exhibit
variance exceeding the variance for the entire population of spinoffs (see Figure 2):
H2: Variation in performance within a parent-firm’s cohort of spinoffs will, on average, exceed
the variation in performance for the population of all spinoffs, regardless of parent-firm quality.
4
Frontiers of Entrepreneurship Research, Vol. 32 [2012], Iss. 12, Art. 2
Posted at Digital Knowledge at Babson
http://digitalknowledge.babson.edu/fer/vol32/iss12/2

FRONTIERS OF ENTREPRENEURSHIP RESEARCH 2012
Founder-Specific Experience
If it can be demonstrated that spinoffs exhibit extreme heterogeneity, then one must ask: What
is the driver of this variance, particularly in light of the hereditary-based theory of spinoffs? Prior
research has clearly focused on parent-firm quality: “Apparently the key to the performance of the
spinoffs is the quality of the environment in which founders worked and not the positions held
by the founders,” wrote Klepper (2002: 660). Others have agreed, noting that “prior employment
affiliations may influence not only new venture formation, but also product-market strategies and
firm survival” (Agarwal et al., 2004: 501). Knowledge creation, replication and transfer (Kogut &
Zander, 1992; Connor & Prahalad, 1996) are basic to the belief that parent-firms with large stocks
of knowledge will be a wellspring of successful spinoffs (Agarwal et al., 2004; Klepper, 2009).
“Pre-entry experience,” argued Klepper, “including experience in incumbent firms, impart(s) an
enduring advantage” (2002: 662). And yet, Agarwal and colleagues caution that “past authors have
assumed an underlying process of knowledge inheritance by a progeny firm, without explicitly
testing whether inheritance from an incumbent parent actually occurs” (2004: 502). In looking at
the disk drive industry, the authors nonetheless conclude that “knowledge is in fact inherited, and
a firm’s founder is a potentially more effective agent of transfer than a hired employee” (Agarwal
et al., 2004: 519).
Ultimately, however, the presence of extreme heterogeneity erodes the credibility of assertions
that lineage influences performance outcomes. Absent a clear linkage between parent-firm
performance and spinoff performance, founder-specific experience takes center stage. Extending
the findings of Chatterji’s study of the medical device industry (2008), we predict that spinoffs
founded by technical experts will perform demonstrably worse than spinoffs founded by non-
technical experts, who will have a higher success rate in obtaining market-based repricing of their
knowledge and capabilities. By dissecting the particulars of spinoff foundings, we predict that
founder experience will be a significant moderator of spinoff performance:
H3a: Spinoffs led by founders possessing only technical knowledge will exhibit lower survival
rates and performance levels than spinoffs led by founders possessing non-technical, general
market knowledge.
H3b: Spinoffs led by founders possessing non-technical, general market knowledge will
exhibit survival rates and performance levels comparable to de novo firms.
H3c: Spinoffs led by founders possessing both technical and non-technical, general market
knowledge will exhibit survival rates and performance levels higher than spinoffs led by
founders possessing only non-technical, general market knowledge.
metho ds An d dAtA
This empirical analysis of the Colorado asbestos abatement industry involves a retrospective
research design with archival data comprised of the complete population of firms having ever
entered or exited the market. The methodology employed in this study is an event-history analysis
(Tuma, Hannan & Groeneveld, 1979) of a comprehensive database hand-collected from more
than 1 million records at the Colorado Department of Public Health & Environment, covering
the period from industry inception in 1986 to the end of 2010. This 25-year period witnessed
5
Hunt and Lerner: REASSESSING THE ENTREPRENEURIAL SPINOFF PERFORMANCE ADVANTAGE
Frontiers of Entrepreneurship Research 2012

STRATEGY
the entry of 612 firms, objectively documented through licensing data. At the project level, 56,465
permits were issued towards for the removal of 21 million lineal feet and 234 million square feet
of asbestos-containing material (ACM), for revenue totaling $1.8 billion. Information is likewise
available for annual supervisor and worker certificates issued by the state.
As a consequence of the strict monitoring and reporting requirements associated with the
removal and disposal of ACM, an unusual level of detail is available. By law, companies must
obtain (and annually renew) a State-issued license prior to commencing any abatement work.
There is no reciprocity with other states for individual certificates or company licenses. Therefore,
firms seeking to perform abatement must possess a Colorado state license. This allows for
comprehensive tracking of every firm into and out of the industry-population. It also allows the
capture of nascent-stage firms that fail to complete even one project or survive beyond their first
annual license. This marks perhaps the first time that a dataset includes a statistically significant
population of organizational forms that fail prior to becoming substantively operational.
Spinoffs are by far the predominant mode of entry for firms in the asbestos abatement
industry, constituting 73% of the total market entrants. The widespread occurrence of spinoffs in
the context of highly granular data makes it an ideal platform test the hereditary theory of spinoffs.
The relatively small number of diversifying incumbents (just 9%) is also fortuitous because it
allows for a more direct comparison of spinoffs and de novo entrants.
Dependent Variables
Three separate dependent variables were used to test the performance advantage that is posited
by existing spinoff theory: Lifespan, Operational Performance and Performance Variance. Firm
Lifespan refers to the total duration of operational existence measured in years. Operational
Performance refers to the average projects per firm-year for each market entrant. Spinoff
Population Performance Variance refers to the spinoff population performance standard deviation,
recalculated for the exclusion of each parent-firm’s finite population of spinoffs.
Independent Variables
Entry Mode – This is a categorical variable. “1” indicates spinoffs and “0” otherwise.
Spinoff Founder Experience - This variable is derived from a set of orthogonal codes in which
a founder’s Colorado asbestos certifications are used to determine whether a spinoff founder has
technical experience, non-technical experience, or both.
Parent-Firm Spinoff Performance Variance - The variance in performance of spinoffs emanating
from the same parent. This is represented by the standard deviation of cohort (i.e. off-spring)
performance and is calculated separately for each parent-firm with five or more spinoffs.
Differences in Variation - The difference between the standard deviation in the performance of
all spinoffs and the standard deviation in performance of each parent-firm’s cohort of spinoffs.
Weighted Average Parent-Firm Spinoff Variation – This is a sum of the variance in performance
for each parent-firm’s finite population of spinoffs (“cohort”), divided by N firms. The parent-
6
Frontiers of Entrepreneurship Research, Vol. 32 [2012], Iss. 12, Art. 2
Posted at Digital Knowledge at Babson
http://digitalknowledge.babson.edu/fer/vol32/iss12/2

FRONTIERS OF ENTREPRENEURSHIP RESEARCH 2012
firm variances are weighted based on the total number of spinoffs spawned by each parent, so that
the spinoff cohort performance variance is proportional to each cohort’s size.
Controls & Model Analysis
Controls: Two separate vectors were developed model to control for macroeconomic and
industry-specific variables at entry. The macroeconomic vector (CONmacro) contains Colorado-
specific measures for construction, unemployment and economic activity. The Industry-specific
measures (CONind) are comprised of the number of firms that entered the industry in a given year,
the industry population for each year, and the entry group size relative to the population (Hannan
& Carroll 1992). Dummy codes were used to control for unobservable year-specific effects.
Model Specifications: OLS regression analysis and significant mean differences are employed to
derive and explicate the focal effects. Prior studies in support of a spinoff performance advantage
primarily have used lifespan as the determinant. Because of the uniquely detailed dataset we have
discovered, for the sake of robustness, both Firm Lifespan and Operational Performance were used
to compare spinoffs and de novo firms in testing Hypotheses 1a and 1b. The generalized OLS
equation for the population is represented by:
Firm Lifespanpop or Oper Perfpop =
b
0 +
b
1CONind +
b
2CONmacro +
b
3YEAR + b4ENTRYMODE (1)
Hypothesis 2 predicts that the average variance in spinoff performance for each cohort of
sibling firms spawned from the same parent will exceed the performance variance for the entire
population of spinoffs in this industry. Each parent-firm cohort performance variance was
subtracted from spinoff population variance through which we derived a function predicting that
the resultant difference from the population variance will be greater than zero (Figure 3):
Hnull: VARavg = VARpop (2)
H2: VARavg > VARpop
Hypotheses 3a, 3b and 3c predict that Spinoff Founder Experience will be a significant predictor
of Firm Lifespan and Operational Performance (Figure 2). The hypothesized model proposes that:
Firm Lifespanspin or Oper Perfspin =
b
0 +
b
1CONind +
b
2CONmacro +
b
3YEAR +
b
4FOUNDER (3)
The generalized relationships are formulated as:
PerformanceTech+GeneralExperience > PerformanceGeneralExperience > PerformanceTechExperience (4)
results
In light of extant theory, our analysis of the natural experiment data produced findings that are
surprising and significant, with noteworthy effect sizes and a high degree of confidence. Bivariate
correlations and descriptive statistics are provided in Tables 1 and 2. In support of Hypothesis
1a and 1b, we found that after accounting for a complete, non-truncated population of industry
entrants, there is no spinoff performance advantage. On the contrary, a head-to-head comparison
between spinoffs and de novo entrants (Table 3) shows that the average lifespan for entrepreneurial
7
Hunt and Lerner: REASSESSING THE ENTREPRENEURIAL SPINOFF PERFORMANCE ADVANTAGE
Frontiers of Entrepreneurship Research 2012

STRATEGY
spinoffs was less than half that of de novo entrants (t1,558= 15.03, p < 0.001). On average, spinoffs
annually completed fewer than half the number projects as de novo firms (t1,558, 558= 9.66, p < 0.001),
a fact that certainly contributed to the lower lifespan. Consistent with these mean differences, the
OLS regression model predicted that a de novo firm would live 1.2 years longer than a spinoff
(p < 0.001), as shown in Table 4. De alio firms, though not the focus of the analysis, similarly
underperformed de novo firms by a substantial margin (Table 3).
Hypothesis 2 predicted that the average performance variance for spinoff cohorts (i.e. siblings
sharing the same parent-firm) would exceed the performance variance for the complete popula-
tion of all spinoffs. If correct, this prediction would suggest that both low-achieving and high-
achieving parent-firms produce spinoffs of varying quality. The data in Table 6 shows that spinoff
performance is highly heterogeneous. The standard deviation for projects annually completed
by the entire population of spinoffs is 17.5. This is significantly lower than the weighted aver-
age standard deviation for all spinoff cohorts, which is 22.8 (t1,448= 9.25, p < 0.001). The thirteen
parent-firms that spawned ten or more spinoffs are listed in Table 6, as well. The weighted average
standard deviation for cohorts from this group of highly prolific parents is 24.26, also exceeding
the population variance (t1,168= 7.48, p < 0.001). Therefore, Hypothesis 2 finds strong support.
Given the finding that performance variance within spinoff cohorts is greater than the variance
between spinoff cohorts, the question arises: What is driving this variance? Hypotheses 3a, 3b and
3c examined this question through the lens of founder-specific experience. Mean comparisons
indicate that spinoffs founded by non-technical managers have double the lifespan of spinoffs
founded by technical managers. The mean difference of three years is highly significant (t1,448=
18.82, p < 0.001), as is the mean difference for firm performance, measured by completed projects
per firm-year, which is nearly 400% higher for firms with non-technical founders (t1,448= 8.99, p
< 0.001). These findings provide strong support for Hypotheses 3a and 3b. On the other hand,
it appears that founders possessing both technical and non-technical experience performed
equivalently to founders possessing only non-technical management experience. This result
requires the rejection of Hypothesis 3c, but confirms Chatterji’s (2009) finding that general
business acumen is the decisive component in spinoff performance, not technical knowledge.
conclu sIons
The cornerstone of dominant spinoff theory is that spawned firms live longer and perform
better than other entry modes due to knowledge acquired from parent-firms (e.g. Klepper 2009).
Further, top-quality parents, possessing larger stocks of capabilities, are presumed to spawn
more and better spinoffs than low-quality parents (Brittain & Freeman, 1986; Christensen, 1993;
Agarwal et al., 2004; Klepper & Sleeper, 2005). This examination of a complete population of
firms contradicts each of these theoretical assumptions. In fact, the only way to find support for
the tenets of spinoff theory through this natural experiment is to truncate the vast preponderance
of spinoff failures (Table 4, Models 2a, 2b & 2c). Contrary to the stylized facts that predominate in
spinoff research (Klepper & Thompson, 2010), our data provides evidence that hereditary-based
conceptions of entrepreneurial spinoffs significantly overstate the relationship between parent-
firm performance and spinoff performance. Taking into account a complete, non-truncated
population of firms, average spinoff performance is less than half that of de novo’s. Specifically, the
average lifespan for entrepreneurial spinoffs is 3.1 years versus 6.6 years for de novo firms; also on
average, spinoffs completed 18.2 projects per firm-year versus 37.3 projects per firm-year for de
8
Frontiers of Entrepreneurship Research, Vol. 32 [2012], Iss. 12, Art. 2
Posted at Digital Knowledge at Babson
http://digitalknowledge.babson.edu/fer/vol32/iss12/2

FRONTIERS OF ENTREPRENEURSHIP RESEARCH 2012
novo entrants. Of the 448 spinoffs that entered the industry, 178 exited within one year and 116
exited without ever performing a single project.
This evidence shows that spinoff performance is highly heterogeneous, even among spinoffs
emanating from the same parent-firm. In sharp contrast to prior studies using truncated data, we
find that this heterogeneity is uncorrelated with parent-firm quality. If hereditary endowments
were sources of a performance advantage, then one would largely expect to see high-performing
parents spawning high-performing spinoffs and low-performing parents spawning low-perform-
ing spinoffs. In fact, however, there is no discernible relationship. Rather, the performance variance
within cohorts from shared parent-firms is significantly larger than the population variance, indi-
cating that high-performing parents spawn many low-performing spinoffs and low-performing
parent-firms produce many high-performing spinoffs. If hereditary knowledge had been a vital
source of performance advantage, then there would be relatively little variation in the performance
of spawned entities emanating from the same parent. In fact, however, such variation is rampant.
Disaggregation of the spinoff data reveals a fuller story. Spinoff performance is clearly
bifurcated between technical and non-technical founders. The average operational existence
for firms founded by technical supervisors is less than half that of non-technical founders. On
average, technical founders completed 6.8 projects per firm-year, versus 32.1 projects per firm-
year for non-technical founders. These results suggest that contrary to extant theory, the key driver
of spinoff performance is less a function of parental knowledge transfers (Klepper, 2001; Agarwal
et al., 2004; Gompers et al., 2005; Klepper & Thompson, 2010) and more a function of differential
outcomes based on founder-specific experience.
In addition to the empirical and theoretical contributions derived from this study, we also
provide a valuable portal to the phenomenon of unobserved non-linear relationships. The study of
nascent-stage organizations inherently confronts the fact that surviving firm data is readily avail-
able while failing firm data evaporates. Applying the techniques commonly used in prior spinoff
studies, it is apparent that data truncation would exclude hundreds of early failures. An examina-
tion of the OLS results (Table 4) reveals the extent to which this is true. In our non-truncated
model consisting of a complete population (Model 1c), Entry Mode is a significant predictor with
by far the largest effect size in the model, indicating that a spinoff will, on average, survive 1.2 years
less than a de novo firm. Meanwhile, a model reflecting the event truncation that is typical of prior
spinoff studies (Model 2c), reverses the sign, indicating that a spinoff will survive 0.4 year longer
than a de novo firm. Such are the pronounced effects of data truncation.
Overall, these results pose significant challenges to the dominant, hereditary-focused concep-
tions of intra-industry spinoffs. Through the lens of this natural experiment, our evidence sug-
gests that the purported spinoff performance advantage requires reassessment.
CONTACT: Richard A. Hunt; University of Colorado – Boulder; Leeds School of Business, 419
UCB, Boulder, Colorado 80309-0419; richard.hunt@colorado.edu.
9
Hunt and Lerner: REASSESSING THE ENTREPRENEURIAL SPINOFF PERFORMANCE ADVANTAGE
Frontiers of Entrepreneurship Research 2012

STRATEGY
refer ence s
Agarwal, R., R. Echambadi, A. Franco & M. Sarkar. (2004). Knowledge transfer through inheritance:
Spinout generation, development, and survival. Academy of Management Journal, 47:
501-522.
Aldrich, H. E., & Fiol, C. M. (1994). Fools rush in? The institutional context of industry creation.
Academy of Management Review, 19: 645-670.
Brittain, J. & J. Freeman. (1986). Entrepreneurship in the semiconductor industry. Working paper.
Cabral, L. & Z. Wang. (2008). Spin-offs: Theory and evidence. Federal Reserve Bank of
Kansas City Working Paper, RWP08-15.
Cameron, A. & P. Trivedi. (2005). Microeconometrics: Methods & Application. Cambridge Press.
Chatterji, A. K. (2009). Spawned with a silver spoon? Entrepreneurial performance and innovation
in the medical device industry. Strategic Management Journal, 30: 185-206.
Christensen, C. M., (1993). The rigid disk drive industry: A history of commercial and technological
turbulence. Business History Review, 67: 531-588.
Conner, K. R., & C.K. Prahalad. (1996). A resource-based theory of the firm: Knowledge versus
opportunism. Organization Science, 7: 477-501.
Cyert, R. & J. March. (1963). A behavioral theory of the firm. Englewood, NJ: Prentice-Hall
Dick, J., K. Hussinger, B. Blumberg & J. Hagedoorn. (2011). Is success hereditary? Evidence on the
performance of spawned ventures. Maastricht University Meteor working paper.
Dyck, B. (1997). Exploring organizational family trees: A multigenerational approach for the
studying organizational births. Journal of Management Inquiry, 6: 222-234.
Elfenbein, D., B. Hamilton & T. Zenger. (2010). The small firm effect and the entrepreneurial
spawning of scientists and engineers. Management Science, 56(4): 659-681
Eriksson, T. & J. M. Kuhn. (2006). Firm spin-offs in Denmark, 1981-2000 - Patterns of entry and
exit. International Journal of Industrial Organization, 24, pp. 1021-1040.
Fiol C. & M. Lyles. (1985). Organizational learning. Academy of Management Review,10:803- 13
Franco, A., & D. Filson. (2006). Knowledge diffusion through employee mobility. RAND Journal
of Economics, 37: 841-860.
Garvin, D. (1983). Spin-offs and the new firm formation process. California Management Review,
25: 3-20.
Geroski, P.A., 1995, What do we know about entry? International Journal of Industrial Organization,
13:421-440.
Gompers, P., J. Lerner & D. Scharfstein. (2005). Entrepreneurial spawning: Public corporations
and the genesis of new ventures 1986-1999. Journal of Finance, 60: 577-614.
Hannan, M., & J. Freeman. (1977). The population ecology of organizations. The American Journal
of Sociology, 82 (5):929-964.
Hannan, M., & J. Freeman. (1989). Organizational ecology. Cambridge, MA: Harvard Press.
Ioannou, I. (2010). The Impact of Different Types of Spinoffs on Firm Survival: Evidence from the
US Automobile Industry 1890-1986. SSRN working paper.
Klepper, S. (2001). Employee startups in high-tech industries. Industrial and Corporate Change,10:
639-674.
Klepper, S. (2002). The capabilities of new firms and the evolution of the US automobile industry.
Industrial and Corporate Change, 11: 645-666.
Klepper, S. (2007). Silicon Valley: A chip off the old Detroit bloc. In Entrepreneurship, growth, and
public policy, D. Audretsch & R. Strom, eds., Cambridge Press.
Klepper, S. (2009). Spinoffs: A Review and Synthesis. European Management Review, 6:159-171.
10
Frontiers of Entrepreneurship Research, Vol. 32 [2012], Iss. 12, Art. 2
Posted at Digital Knowledge at Babson
http://digitalknowledge.babson.edu/fer/vol32/iss12/2

FRONTIERS OF ENTREPRENEURSHIP RESEARCH 2012
Klepper, S. & S. Sleeper. (2005). Entry by spinoffs. Management Science, 51: 1291-1306.
Klepper, S., & P. Thompson. (2010). Disagreements and intra-industry spinoffs. International
Journal of Industrial Organization, 28: 526-538.
Kogut, B., & U. Zander. (1992). Knowledge of the firm, combinative capabilities and the replication
of technology. Organization Science, 3: 383-397.
Levitt, B. & J. March (1988). Organizational learning. Annual Review of Sociology, 14: 319-340
Nelson, R., & S. Winter. (1982). An evolutionary theory of economic change. Harvard Press.
Nelson, R. (1991). Why do firms differ, and how does it matter? Strategic Management Journal,
12: 61-74.
Paarsch, H. (1982). Censored Regression Models. Journal of Econometrics, 24, 197 – 213.
Phillips, D. (2002). A genealogical approach to organizational life chances: The parent-progeny
transfer. Administrative Sciences Quarterly, 47: 474-506.
Stuart, T., & O. Sorenson. (2003). Liquidity events and the geographic distribution of
entrepreneurial activity, Administrative Sciences Quarterly, 48: 175-201.
Teece, D. & G. Pisano. (1994). The dynamic capabilities of firms: An introduction. Industrial and
Corporate Change, 3 (3): 537–56.
Tuma, N., M. T. Hannan & L. Groeneveld. (1979). Dynamic analysis of event histories. American
Journal of Sociology, 32: 842-854.
Tsiatis, A. (2006). Semiparametric theory and missing data. Springer Series on Statistics.
Winter, S. (1987). Knowledge and competence as strategic assets. In D. Teece (Ed.), The competitive
challenge: Strategy for industrial innovation and renewal: 159-184.
FIGURE 1: Stylized Schematic of the Data Truncation Effect.
In compiling data related to new market entry by nascent-stage firms an unobserved, non-linear
relationship results in the truncation of early-stage failures, dramatically changing the empirical
and theoretical implications of the data. Spinoff data analysis is highly prone to these effects.
11
Hunt and Lerner: REASSESSING THE ENTREPRENEURIAL SPINOFF PERFORMANCE ADVANTAGE
Frontiers of Entrepreneurship Research 2012

STRATEGY
FIGURE 2: Hypothesized Model of Spinoff Performance.
In the context of a complete population, spinoffs are predicted to underperform de novo entrants.
Spinoff founders possessing only technical knowledge will exacerbate this effect, while founders pos-
sessing general business knowledge will reduce the negative effect of being a spinoff. If correct, these
predictions shift the attention away from parental lineage to individual spinoff founder attributes.
FIGURE 3: Hypothesized Model of Spinoff Performance Variance.
Hypothesis 2 (i.e. the triangular region denoted as H2) predicts that the average performance vari-
ance for the cohort of spinoffs spawned by the same parent will exceed the performance variance
for the entire population of spinoffs. Line H0, the null hypothesis, involves no difference in vari-
ance. A result of H2 > H0, functionally indicates that low and high-performing parents produce
both low and high-performing spinoffs.
12
Frontiers of Entrepreneurship Research, Vol. 32 [2012], Iss. 12, Art. 2
Posted at Digital Knowledge at Babson
http://digitalknowledge.babson.edu/fer/vol32/iss12/2

FRONTIERS OF ENTREPRENEURSHIP RESEARCH 2012
TABLE 1: Bivariate Correlations.
TABLE 2: Descriptive Statistics.
N Minimum Maximum Mean s.d.
Year of Firm Founding 612 1986 2010 1997 7.38
Year of Firm Failures 508 1987 2011 2001 7.77
Currently Operating (Yes = 1) 612 0 1 0.17 0.37
Firm Lifespan (Years) 612 0 25 3.73 4.45
Entry Mode (Spinoff = 1) 612 0 1 0.73 0.21
Completed Projects (Lifetime) 612 0 2817 89 287
Completed Projects (Avg. Annual) 612 0 166 9 21
Spinoff Frequency - by Parent 100 1 20 4.48 4.41
Founder Experience (Non-Tech = 1) 448 0 1 0.26 0.44
New Firm Entry Cohort 612 14 41 26.49 7.41
Population at Entry 612 29 134 91 24.41
Entries as Percent of Population 612 13% 100% 33% 19%
Entry Cohort Average Lifespan 612 1 15 4 3
TABLE 3: Survival and Performance Comparisons by Entry Mode
Entry Mode # Firms
% of
Firms Average Lifespan
Average Projects Completed
Per Year
De novo 110 18% 6.6*** 37.3***
Spinoffs 448 73% 3.1*** 18.1***
De alio 54 9% 3.1 12.0
All Firms 612 100% 3.7 23.8
*** Focal mean differences (spinoff vs. de novo) were highly significant, p < .001.
13
Hunt and Lerner: REASSESSING THE ENTREPRENEURIAL SPINOFF PERFORMANCE ADVANTAGE
Frontiers of Entrepreneurship Research 2012

STRATEGY
TABLE 4: Effect of Entry Mode on Lifespan (OLS Regression Estimation)
For ease of use, the units in this table are expressed in Years. When data is truncated (Models 2a-c),
spinoffs survive 0.362 years longer than de novo firms. But with complete, non-truncated data,
the situation is reversed. Being a spinoff entrant actually reduces a firm’s lifespan by 1.242 years.
Predictor
Models
Non-Truncated (n = 612) Truncated Data (n = 380)‡
1a 1b 1c 2a 2b 2c
(Constant) 4.635*** 4.737*** 5.856*** 4.113** 4.464*** 4.935**
(1.066) (.906) (1.113) (.824) (.807) (.741)
Macro Controls -0.034 -0.031 -0.022 -0.102* -0.093 -0.091
(.013) (.010) (.007) (.117) (.110) (.110)
Pop at Entry -0.021** -0.022* -0.023* 0.142* 0.137* 0.131*
(.011) (100) (.011) (.457) (.452) (.448)
Entry Cohort Size 0.002 0.002 0.018 (-.089)* (-.033)* (-.024)*
(.028) (.028) (.028) (.031) (.017) (.024)
Cohort Lifespan 0.011 -0.009 0.001 -0.001 -0.001 0.000
(.025) (.030) (.025) (.002) (.001) (.001)
Founder Experience 0.077* 0.049* 0.136 0.122
(.051) (.027) (.034) (.029)
Year of Entry -0.079 -0.074 0.084 0.079
(.060) (.057) (.055) (.052)
Avg. Annual Projects -0.044** -0.030* -0.075*** -0.069***
(.021) (.013) (.018) (.015)
Total Projects 0.014*** .012*** 0.016*** 0.014***
(.001) (.001) (.002) (.001)
Entry Mode (Spinoff =1) -1.242*** 0.362*
(.361) (.394)
Adj. R2 0.331 0.585 0.780 0.414 0.534 0.551
F-value 34.7*** 111.2*** 118.8*** 50.8*** 80.9*** 57.7***
‡ - Truncation of firms failing to complete survive one year and/or complete at least five projects
Standard errors in parentheses. * p < .05, ** p < .01, *** p < 0.001.
TABLE 5: Spinoff Founder Comparison – Technical vs. Non-Technical Knowledge
Founder Ty pe Average Lifespan Average Projects Per
Firm-Year
Founder with Only Non-Technical Experience 5.3*** 32.09***
Founders with Only Technical Experience 2.3*** 6.75***
Founders with Both Tech & Non-Tech Experience 5.1 30.82
All Spinoffs 3.1 18.19
*** Mean differences (Technical vs Non-Technical) were highly significant, p < .001.
14
Frontiers of Entrepreneurship Research, Vol. 32 [2012], Iss. 12, Art. 2
Posted at Digital Knowledge at Babson
http://digitalknowledge.babson.edu/fer/vol32/iss12/2

FRONTIERS OF ENTREPRENEURSHIP RESEARCH 2012
TABLE 6: Heterogeneity of Performance – Cohort Variance vs. Population Variance
For parent-firms producing ten or more spinoffs, the performance variance for each parent-firm’s
spawn-cohort is compared to the performance variance for the entire population of spinoffs, which
was 17.5 projects per firm-year. In all but one case (i.e. MDR), the cohort performance variance
exceeded the population variance. This provides strong support for Hypothesis 2, which predicted
that both low and high-performing parent-firms spawned both low and high-performing spinoffs,
a finding that confounds the attempt to link parent-firm quality to spinoff quality.
Parent Name
# of
Spinoffs
in
Cohorts
Parent
Performance
(# Projects per
Firm-Year)
Average
Spawn
Performance
(# Projects per
Firm-Year)
Spawn
Performance
Range
(# Projects per
Firm-Year)
Cohort
Variance
(Std. Dev.)
Cohort
Variance
Minus
Population
Variance‡
American 20 112.7 13.1 0 – 89.4 24.23 6.72
RRI 17 90.0 17.6 0 – 116.9 34.72 17.21
LVI 17 166.3 12.5 0 – 97.9 24.51 7.00
Dominion 16 97.2 14.8 0 – 53.5 25.57 8.06
Great Plains 15 5.4 10.7 0 – 77.2 21.43 3.92
ACM Removal 14 88.3 11.7 0 – 58.7 19.68 2.17
Mac-Bestos 10 57.1 10.5 0 – 60.6 18.85 1.34
MDR 10 53.5 11.5 0 – 47.4 16.59 (1.08)
Schauer 10 51.0 12.3 0 – 28.9 21.51 4.00
Asbestos Tech 10 16.5 10.4 0 – 86.5 26.76 9.25
Onyx 10 33.3 23.5 0 – 133.1 20.38 2.87
Misers 10 52.6 14.3 0 – 87.9 27.53 10.02
A.R.C. 10 14.4 13.6 0 – 52.8 17.79 0.28
13 Largest Cohorts
(avg.) 169 64.5 12.6 0 – 133.1 24.26*** 6.75***
All Spinoff
Cohorts (avg.) 448 30.3 11.8 0 – 133.1 22.78*** 5.27***
‡ The standard deviation in projects completed per firm-year for all 448 spinoffs is 17.5
*** Mean differences (Average Cohorts Variance versus Population Variance) were highly significant, p < .001.
15
Hunt and Lerner: REASSESSING THE ENTREPRENEURIAL SPINOFF PERFORMANCE ADVANTAGE
Frontiers of Entrepreneurship Research 2012