Science and industry evolution: evidence from the first 50 years of the German laser industry
Science and industry evolution: evidence from the first 50 years of the German laser industry
Guido Buenstorf 0 1 2 3 4
Dominik P. Heinisch 0 1 2 3 4
0 G. Buenstorf Institute of Innovation and Entrepreneurship, University of Gothenburg , Gothenburg , Sweden
1 G. Buenstorf
2 G. Buenstorf ( ) University of Kassel , Kassel , Germany
3 JEL classification L10
4 G. Buenstorf IWH Leibniz Institute of Economics Halle , Halle , Germany
Industry evolution is driven by innovation. Scientific research is an important source of innovation-relevant knowledge. To trace its impact on industry evolution, we follow the entry and exit of firms from various backgrounds over the first five decades of the German laser industry. We find that academic startups became increasingly competitive after substantial changes were introduced to the governance of university-industry relationships. Their enhanced performance helps explain why no shakeout in firm numbers has been observed to date. Doctorateholding inventors contributed to the performance of entrants, indicating that the impact of scientific knowledge on industry evolution goes beyond academic entrepreneurship.
Industry evolution; Scientific knowledge; Academic entrepreneurship; Doctoral training; Laser
Laser manufacturing is a prime example of a
. Following a prolonged
race involving universities, governmental research
facilities, and corporate R&D laboratories, the first
workable laser was built in 1960
Within a few years, commercial laser
manufacturers entered the US market, often directly drawing
on prior research efforts. Subsequently, the size of
the US laser firm population continued to increase
for several decades
(Klepper and Sleeper 2005)
entrants came from various backgrounds. Besides
diversifying pre-existing firms from high-tech
sectors such as aerospace and defense, they included
academic entrepreneurs from universities and
government labs as well as intra-industry spin-offs,
i.e., entrepreneurial firms started by individuals who
had previously worked for laser firms. Defying the
regularities of the industry life cycle
a shakeout in the number of active US producers
only set in after about 35 years, much later than in
most other industries. This shakeout had a pronounced
effect on the firm population. From 1996 to 2007, the
number of US laser producers fell by about 50% (from
172 to 87).
Bhaskarabhatla and Klepper (2014)
this drastic decrease to a major upheaval in laser
technology: the diode-pumped solid state (DPSS) laser
that replaced earlier laser types in a variety of
The development of the German laser industry
exhibits a number of parallels to the USA. Early
entry drew on direct or indirect links to laser research
, and an almost uninterrupted increase
in the number of active firms was observed for decades
. Academic entrepreneurship and—
to an even larger extent—intra-industry spin-offs were
also widespread in the German laser industry (ibid.).
This is notable because in the 1970s and 1980s,
Germany was characterized by a weaker entrepreneurial
culture, a less developed venture capital industry,
and stronger legal restrictions to employee
mobility than the USA and especially California (where
many US laser producers were located)—factors that
have been shown to affect the prevalence and
success of entrepreneurial activities
Marx et al. 2009; Samila and Sorenson 2011; Fritsch
and Wyrwich 2014)
. More recently, however, the
German laser industry has taken a different path from
that of the US industry. No significant decrease in
firm population size has been observed, but firm
numbers have stabilized in the late 1990s (cf. Fig. 1).
New entrants have continued to replenish the firm
population, compensating for incumbents that failed,
were acquired, or decided to concentrate on other
markets. This absence of a shakeout after 50 years
of evolution constitutes a remarkable deviation from
the pattern generally observed in industry evolution
(Klepper 1997; Buenstorf 2007)
The objective of this paper is to study the recent
evolution of the German laser industry in more
detail. Motivated by the science-based nature of the
industry, we particularly focus on the potential role
of scientific1 knowledge. In line with prior work
on industry evolution, we first trace recent entries
and exits and how they affected the composition
of the firm population in terms of pre-entry
experience. We identify new entrants based on academic
entrepreneurship and compare their longevity in the
laser industry with that of other types of entrants.
However, just tracing academic startups does not
fully account for the impact of scientific knowledge
on industry evolution. Entrepreneurs with other
preentry backgrounds may also have acquired scientific
knowledge, and university-educated employees
further add to the firms’ knowledge stocks. We therefore
identify doctorate holders among spin-off founders
and corporate inventors and analyze their impact
on firm longevity, thus contributing to the
understanding of how the backgrounds and early hires of
entrepreneurial entrants shape the evolution of
innovative industries. With doctoral training, we address a
potentially powerful channel of transferring academic
knowledge to the private sector that still is not
sufficiently well understood, and for which consistent
information can be assessed over long periods of time.
The remainder of the paper is structured as follows.
In Section 2, we discuss the theoretical and
institutional background of the study. Section 3 details the
1In what follows, “scientific” will be used as a generic term that
also covers engineering research.
data collection for the empirical analysis. Section 4
studies the performance of different types of entrants
in the evolution of the German laser industry,
focusing on academic entrepreneurship and inventors with
doctoral training. Section 5 concludes.
2 Theoretical background: industry evolution, entrepreneurship, and science
2.1 The evolution of firm populations in innovative
A sizeable literature explores empirical regularities in
the evolution of industries (cf., e.g.,
for surveys). The key stylized
fact established in this literature is that industries2
tend to experience a shakeout in the number of active
firms. Characteristically, the shakeout is observed at
an early age of the industry when its output is still
growing. It turns an atomistic market structure with
many small producers into an oligopolistic one
dominated by a few large firms. A prominent case in point
is the US automobile industry where the shakeout
began in 1907, the year before Ford introduced the
Model T that would spur mass motorization and the
phenomenal growth of the industry. Within 16 years,
the number of active firms was reduced by 60%, only
to fall further afterwards.
According to the leading models of industry
evolution, firm population dynamics are driven by
technological developments in terms of product
and Sua˙rez 1993)
or process innovation
and MacDonald 1994; Klepper 1996, 2002)
Klepper’s evolutionary model, the shakeout emerges from
differences in firms’ ability to spread the fixed cost
of process innovation over their output base
and Klepper 1996)
. Cost spreading induces an
escalation of R&D efforts
and gives rise to
success-breeds-success dynamics favoring large and
early entrants. Eventually, profitable entry is no longer
feasible and smaller incumbents (which often entered
late) are driven from the market.
2In the literature on industry evolution, industries are narrowly
defined, typically corresponding to individual product markets
such as automobiles, disk drives, or semiconductors. The laser
industry has been studied in this literature ever since
Not all industries experience a shakeout, and not all
at the same age. Variations in the evolutionary
dynamics of industries have been linked to the heterogeneity
of product designs. In particular, in industries whose
product is made in many variants that are poor
substitutes, individual submarkets may be too small to
induce an escalation of R&D efforts. Small producers
may then remain competitive, market structure may
remain atomistic, and profitable entry may be
feasible for decades. According to
, this description fits the early US
laser industry. Before the diode-pumped solid-state
(DPSS) laser, individual laser types could hardly be
substituted for each other because they differed in
wavelength, maximum power, and other user-relevant
characteristics. In essence, each application required
a specific type of laser. Pre-DPSS, lasers for
individual applications were often made by a small number
of producers, and most producers were only active in
a few application markets. New laser designs opened
up new applications and allowed for the entry of
new producers, which often faced little competition
from incumbents. Beginning in the late 1980s,
however, DPSS lasers were becoming available that were
more versatile in use and began to compete with other
Bhaskarabhatla and Klepper (2014)
suggest that the DPSS market was large enough to warrant
cost-reducing process innovations. An R&D
escalation set in, which challenged smaller DPSS producers
while further increasing DPSS competitiveness
vis-a`vis other laser types. This gave rise to a shakeout first
in the DPSS market, then in the US laser industry more
Based on the interest in industry evolution, a related
literature focuses on the composition of entry cohorts
in different industries and on how pre-entry
experience relates to post-entry performance (cf.
). Two types of entrants have been
in the focus of this literature: diversifiers from related
industries and intra-industry spin-offs. Both types of
entrants figured prominently in the laser industry.
Similar to other industries, they tended to outcompete
firms with alternative types of pre-entry experience
(Sleeper 1998; Buenstorf 2007)
The distinctive performance of diversifiers and
spin-offs suggests that capabilities can be reproduced
and transferred across markets and organizations.
Spin-off dynamics can be conceptualized as an
evolutionary process in which spin-offs inherit capabilities
and the underlying organizational routines
and Winter 1982)
from their parent firms
(Klepper 2001, 2016)
. This distinctively evolutionary
process is relevant beyond individual industries because
it shapes the composition of national and regional
economies and induces path-dependent development
trajectories (Frenken and Boschma 2012), which is
corroborated by findings on national product
(Hidalgo et al. 2007)
, regional clustering
(Buenstorf and Klepper 2010)
, and the branching of
(Neffke et al. 2011)
. Pinning down exactly what
is inherited has remained elusive, however. Likewise,
we do not fully understand how industry evolution is
shaped by the context of the industry.
2.2 Academic entrepreneurship and industry
Industry evolution does not happen in isolation but
is shaped by the industry’s environment. Scientific
research is an important part of this environment.
The literature on university-industry relationships
indicates that science may shape industry evolution
through a number of channels of impact and
various forms of scientist engagement (cf., e.g.,
et al. (2013)
). Traces of science are easily found in
the laser industry. Not only did the laser originate
from scientific research activities. Science continued
to affect the laser industry throughout its subsequent
Academic entrepreneurship generating new entrants
is perhaps the most direct channel through which
science affects industry evolution
(Heblich and Slavtchev
. It is quite significant in the laser industry.
Academic entrepreneurs from universities and
government labs accounted for 9% of all US laser entrants
from 1991 to 1994. They were the third largest group
of entrants, surpassed only by diversifiers (60%) and
intra-industry spin-offs (16%). In Germany, about
20% of all laser entrants before 2003 were based
on academic entrepreneurship (where the larger share
is mostly due to a smaller number of diversifying
entrants than in the USA). Even though early
academic laser startups were mostly outperformed by
diversifiers and spin-offs
(Sleeper 1998; Buenstorf
, individual success stories exist. In Germany,
they include Lambda Physik, which pioneered
excimer lasers and became one of the industry’s early
leaders. Lambda Physik went public in 2000 and was
eventually acquired by Coherent, one of the major US
Academic entrepreneurship is generally recognized
as a relevant field of innovation policy today, and
public support schemes have been established to foster
academic entrepreneurship. This is a relatively recent
development, however, which started well after the
beginning of the laser industry. The main instrument
to support academic entrepreneurship in Germany is
the EXIST program initiated by the federal
government in 1998
. EXIST combines the
funding of measures fostering a more entrepreneurial
culture at German universities with (since 2000)
personal support for (nascent) academic entrepreneurs.
It has been complemented by policy measures at the
level of individual federal La¨nder. Availability of
venture capital for academic entrepreneurs was increased
when the High-Tech Gru¨nderfonds (HTGF), a
privatepublic partnership that has grown into Germany’s
most important seed-stage investor, was established in
Initiatives to enhance knowledge and technology
transfer from universities have accompanied the
support of academic entrepreneurship. New infrastructure
for technology transfer offices was put into place, and
inventor ownership of academic patents was replaced
by university ownership, in 2002. While the available
empirical evidence does not suggest that these reforms
resulted in larger numbers of academic patents, they
seem to have increased the likelihood that academic
patenting leads to entrepreneurial activity
et al. 2016)
Through their various initiatives, German policy
makers hoped to increase the number and quality of
startups from universities and public research
organizations. If successful, the effects of these initiatives
should be particularly pronounced in science-based
industries such as laser manufacturing. One might
therefore expect to find increasing numbers, as well
as an enhanced performance, of academic startups
when they gained the support of public policy. We will
explore this conjecture below.
2.3 Doctoral training of corporate inventors
Accounting for academic start-ups is not sufficient to
fully capture the role of scientific knowledge in
industry evolution. Science does not only provide a seedbed
for new firms, but also impacts on established ones.
Various forms of academic engagement with the
(Perkmann et al. 2013)
have also been in
the focus of recent public policy initiatives, and the
German laser industry was a preferred target of these
initiatives. From its beginning, policy makers
considered laser technology to be of strategic importance
for the German manufacturing sector, and a
succession of large-scale programs addressed laser research
. Not least due to the active
lobbying of the laser industry itself, these programs often
focused on industrial applications and collaboration
with private-sector partners
As German universities use large shares of their
grant money to employ doctoral students in the funded
projects, doctoral training is an important outcome
of research funding. The majority of graduated
doctoral students subsequently work in the private sector,
often in corporate R&D laboratories. Their
mobility constitutes an important channel through which
scientific knowledge is disseminated. As shown by
, graduates from doctoral training not
only bring up-to-date knowledge about their
specific field to their private-sector employment.
Analytical problem-solving skills and familiarity with the
broader discipline seem even more important.
Embodied knowledge transfer based on graduate mobility
also provides firms with access to tacit research skills,
which are difficult to acquire from publications and
Recent work shows that the quality of early
employees is an important factor in the performance
of entrepreneurial entrants (e.g.,
Dahl and Klepper
). We follow numerous earlier studies and use
patent data to identify inventors working for laser
producers. Based on an intricate matching procedure (see
Section 3 for details), we are able to distinguish
corporate inventors who hold doctoral degrees from those
who do not. This allows us to study whether the (early)
3Tracing R&D collaboration in patents and scientific
publications is difficult, particularly over large periods of time. Based
on limited data,
Blankenberg and Buenstorf (2016)
modest numbers of co-authored papers and co-invented patents
in German laser research.
Buenstorf et al. (2015)
as well as
Fritsch and Medrano Echalar (2015) find that public research
and private-sector laser activities are positively related at the
hiring of R&D staff with doctoral research training
is systematically associated with the performance of
Information about doctoral training moreover
allows for a more fine-grained approach to trace the role
of scientific knowledge in the pre-entry experience of
entrants. By identifying entrepreneurs with doctoral
degrees, we begin to disentangle founder education
from pre-entry job experience in accounting for the
performance of entrants. Not all academic
entrepreneurs in our firm dataset hold doctoral degrees, but a
number of academic startups were organized by
technical staff or graduates of master programs. Nor is
doctoral training limited to academic entrepreneurs.
Founders of intra-industry spin-offs often completed
doctoral training before they started to work for laser
incumbents. We also observe diversifiers whose
founders hold doctoral degrees and still influence the laser
activities of the firm.
3.1 Entry, exit, and pre-entry experience in the
German laser firm population, 1964–2013
To trace the recent evolution of the German laser firm
population, we extended the dataset assembled by
to the year 2013. Given that the first
entrant was observed in 1964, this provides us with
data for the full population of commercial laser source
producers in Germany over the first 50 years of the
industry. The extension is based on the same data
sources and classification principles as the original dataset.
In particular, we employed the catalogs of the
biannual LASER trade fair in Munich, as well as annual
buyers’ guides, to identify new entrants after 2003
as well as firms that exited from the laser market.
The extended dataset covers 184 firms active from
1964 to 2013. It is restricted to domestic producers of
laser sources and does not include importers or
mere distributors. Laser system producers without
own production and sales of laser sources are also
Again replicating earlier efforts, we then identified
the pre-entry experience of the post-2003 entrants.
For diversifiers, we recorded their backgrounds in
other industries. De novo entrants were classified
into three broad categories: intra-industry spin-offs
started by individuals with work experience in the
laser industry (including serial laser entrepreneurs),
academic startups started by individuals coming from
universities and other public research organizations
(researchers, technical staff, as well as students), and
a small group of other startups including all
remaining entrepreneurial entrants. Firm and founder
backgrounds were established on the basis of all available
information, including laser trade magazines and other
publications, firm websites, incorporation files, and
professional online networks.
The German laser firm population has been
characterized by substantial heterogeneity throughout its
evolution. German laser producers are active in many
different application markets. They are best known
for industrial materials-processing lasers, but also
supply lasers for metering, signal processing, scientific
research, and other applications. Many produce not
only laser sources but also systems incorporating these
(Buenstorf et al. 2015)
. The available
evidence moreover suggests that firm sizes are highly
skewed. Some laser producers are large diversified
corporations, but the majority is small. The
ownership structure is likewise heterogeneous; it includes
many family-owned companies. As indicated by the
example of Trumpf Group, one of the world’s
leading makers of industrial laser sources and systems,
family-owned firms can grow into global players of
substantial size. (Trumpf currently has more than
10,000 employees, and about two-thirds of sales are
related to lasers; cf.
Trumpf Group (2016
firms are located all over Germany, with pronounced
clusters in and around Munich and Berlin. Notably,
the laser industry is one of the few high-tech
industries with a meaningful presence in Eastern Germany.
This reflects the long-term legacy of the optical
industry in this region, but also the substantial success that
socialist East Germany had in laser research before
Germany was united in 1990
3.2 Identifying corporate inventors and entrepreneurs
with completed doctoral training
A series of data collection, matching, and
disambiguation steps were required to extend the
firmlevel dataset to the level of individual inventors and
entrepreneurs. As a first step, we collected the names
of all inventors listed on patents by laser source
producers. To this purpose, all laser-related priority
patent applications4 filed by German applicants were
retrieved from PATSTAT (2014 Autumn Edition), a
total of 12,134 patents for the time period 1960 to
2012. Laser source producers among the applicants
were identified using an extended list of firm names
that includes several name variants of the 184 firms in
our firm-level dataset. Standardized applicant and firm
names (and where available previous or alternative
firm names) were matched using a fuzzy string
matching algorithm.5 Matches were manually checked for
false positives. A total of 110 firms included in the
list were identified as having filed at least one patent.
They account for 4189 patent applications or about
35% of all patents in the respective classes, reflecting
substantial laser-related patenting of large
manufacturing firms that are not commercial laser source
producers. To ensure completeness of firms’ patent
portfolios, the above procedure was repeated for the
firm founders. Entrepreneurs may have applied for
patents before they started their own firms, which
may then not be listed as patents of the respective
firms. Altogether, information is available on 244
founders. Applying the above matching procedure
resulted in 128 additional patent applications filed by
We next identified laser inventors and entrepreneurs
with completed doctoral training. To this purpose,
we matched inventor names with the author names
of all laser-related doctoral dissertations submitted to
German universities (1970–2015). Six thousand four
hundred seventeen distinct inventor names are listed
on the patent applications in the sample. Inventor
names were cleaned and academic titles listed in the
patents were retrieved. Since the “Dr.” is an official
part of the name in Germany, it is also frequently
found in patent documents. The set of inventor names
was disambiguated using information on common
co-inventors, applicants, addresses, and titles. This
procedure identified 2794 unique inventors whom we
4We selected all patents filed by German applicants in IPC
“H01S” and “B23K 26” and their subclasses. We also included
patents filed in IPC, “A61B 18/2,” “A61F 9/008,” “A61F 9/009,”
“A61N 5/068,” “A62D 3/17,” “B22F 3/105,” “B23H 7/38,”
“B29C 65/16,” “B41J 2/455,” “F21K 9/00,” “F21Y 115/30,”
“G01C 19/66,” “G01N 21,” “G02B 27/48,” “G11B 7/127,” and
“G11B 11/03” whenever the patent title contained “laser”.
5For all string matchings, a 2-gram Jaccard similarity was
employed as proposed by
Schoen et al. (2014)
patent data. We use a threshold of 0.8 because our sample allows
for manual data checking.
Rows one to four report number of firms, rows four to nine report averages (standard deviation in parentheses)
matched against the full list of dissertation authors
listed in the catalog of the Deutsche
Nationalbibliothek (DNB). Since 1969, the DNB has had a legal
mandate to collect all doctoral dissertations defended
in Germany. For our matching, we consider all
dissertations in physics, electrical engineering, and
mechanical engineering listed in the DNB catalog. Medical
dissertations were excluded. The final dataset contains
152,679 dissertations and their authors.
Several filters were applied to identify false-positive
matches. These filters include information on
laserrelated titles and unique author names. Dissertations
were classified as laser-related if the title includes
words strongly suggesting an association with laser
research.6 For all matched dissertation authors,
homonyms were searched in the dissertation sample. If no
homonym (dissertation author with exactly the same
combination of first and family names) was found, the
respective individual was classified as having a unique
name. In other words, they are the only person of
this specific name holding a doctoral degree from a
German university in one of the relevant disciplines.
Combined with a “Dr.” in the inventor name data, a
unique name is a strong predictor of a true positive
match. In addition, the lag between the year of
graduation and the year of the first patent filing was
calculated. All positive-matched inventor-author pairs were
manually processed to detect false name matchings.
6The following words were classified as indicating laser
science-related dissertations: “laser,” “light,” “spectroscopy,”
“spectral,” “pulse,” “optical,” “induced,” and their German
Additional information was used in cases where no
clear-cut decision was possible. After eliminating
falsepositive-matched author-inventor pairs, we obtained
665 inventors with completed doctoral training who
patented for German laser source producers. By
applying the same procedure to the 244 laser firm founders,
86 entrepreneurs were identified to hold a doctoral
Table 1 provides descriptive statistics of the main
variables used in the empirical analysis below. Table 2
contains pairwise correlations.
4 Empirical analysis
4.1 Academic startups in the evolution of the German
The updated dataset covering the full firm population
of the German laser industry over its first five decades
provides a rare opportunity to study the long-term
evolution of a contemporary science-based industry.
It also allows us to explore the impact of scientific
knowledge on this evolution, as well as how this
impact may have changed over time. As noted above,
the recent evolution of the firm population in the
German laser industry differed from what might have been
expected when using the US industry as a benchmark.
Given the modified governance of university-industry
relations in Germany, changes in the use of scientific
knowledge in entrepreneurship and innovation
plausibly contributed to the divergent development. To
begin to explore this conjecture, in this subsection,
we trace changes in the composition of entry and the
performance of entrants. In the next subsection, we
will investigate the role of inventors and entrepreneurs
with doctoral training.
Bhaskarabhatla and Klepper (2014)
beginning of the US shakeout to 1996. In our data for
Germany, the number of active laser firms increased
until 2001, and no clear-cut trend in the number of
active firms is discernible in later years. In the
following analysis, we will distinguish two phases of the
German laser industry. We define the second phase
as the time period when, based on the US
experience, a shakeout might have been expected to occur
in Germany. Specifically, we use 1998 as the
beginning of this second phase, i.e., 2 years after the US
shakeout had started and 3 years before the
continual growth of the German firm population ceased. The
year 1988 also was the starting year of the EXIST
program, which marks a watershed in the support for
academic entrepreneurship (
Egeln et al. (2010)
also Section 2).
As can be seen in Table 1, about half of all
academic startups entered in the post-1997 period.
Academic startups account for 24% of all entrants in this
period vs. 20% in the first period. On first glimpse,
this modest increase seems to suggest that the
changing governance of university-industry relations did not
have a strong effect on academic startups. Note,
however, that their share among all active firms in the
laser industry tended to increase throughout the
post1997 period (Fig. 1). This indicates that, counter to
prior findings for the earlier years of the German laser
industry, academic startups entering in the second
period performed relatively well.
To study the performance of entrants in a more
systematic way, we use longevity in the laser industry
as a proxy of firm performance and perform hazard
rate models specified in similar ways as in
. As has often been observed (e.g.,
and Montgomery (2013)
Weterings and Marsili
), longevity is an imperfect measure of firm
performance. Remaining in the market indicates only
that a minimal level of performance is attained, while
there tend to be pronounced differences in growth
and profitability among the firms that remain active
in a market. Even though these differences are not
captured in hazard rate models, hazard rate analyses
are nonetheless in widespread use in the literature on
industry evolution. This is owed to the lack of
alternative performance indicators that can be assessed for
the full population of firms, including those which are
too small to be required to disclose financial
information or employee numbers, over long periods of
observation. We face the same problems for the
German laser industry, which includes many small firms
that have ceased to exist long ago and for which no
information about size or profits can be retrieved.
Another shortcoming of longevity as a
performance indicator is that firms may exit for a
variety of reasons which are not equally
informative about performance (or the lack thereof).
Exit by merger of acquisition is particularly
worrisome. Being acquired may allow poor
performers to avoid impending bankruptcy. Alternatively,
it may reflect success, as up-and-coming firms
are taken over by larger competitors or industry
outsiders. The acquisition may then allow these
promising firms to finance their further growth. In the
literature on industry evolution, the conventional
approach to deal with the ambiguous nature of
acquisitions is to handle firms acquired by competitors, as
well as smaller partners in merger events, as censored
). Censoring makes
full use of the information about the firm’s survival up
to the acquisition event, while allowing the researcher
to remain agnostic about the cause of being acquired.
We adopt the same approach in our empirical
analysis. We also follow prior practice (ibid.;
) in specifying the hazard models as parametric
Gompertz models assuming proportionality of hazards
across types of entrants.
The first set of results from these models is
reported as Models 1–3 (Table 3). Model 1 only includes
two dummy variables denoting diversifiers and
intraindustry spin-offs, respectively. Negative but
insignificant coefficient estimates are obtained for both
variables. Accordingly, in contrast to
we do not find systematic differences in exit hazards
across the various types of entrants. Consistent with
, no systematic duration dependence
of hazard rates is indicated by our results. In other
words, entrants are not subject to a significant liability
In Model 2, we investigate how the exit hazard
developed over time, with a focus on academic
startups. The model includes a dummy variable indicating
academic startups, a second dummy variable denoting
post-1997 entrants, and an interaction between both.
Two aspects of the results are noteworthy. On the one
hand, we do not find that post-1997 entrants
generally had a lower exit hazard. The coefficient estimate
of the post-1997 dummy is very close to zero and far
from being statistically significant. On the other hand,
the exit hazard of academic startups differs drastically
between the two periods. While early academic
startups perform significantly worse than other entrants,
we estimate a much smaller coefficient for the late
academic startups (significant at the 10% level)
implying a reduction in the hazard of about 70% between
the two cohorts of academic startups. These results
suggest a positive development in the longevity of
entrants after 1997, which however was restricted to
Model 3 corroborates these findings. It differs
from Model 1 only in that a further dummy variable
denoting post-1997 academic startups is added. The
coefficient estimated for this variable implies a 64%
lower hazard compared to the reference group of early
academic startups and other entrants. This compares
to a 50% reduced hazard of diversifiers and a 44%
reduced hazard of spin-offs (irrespective of their entry
year) and indicates that, different from their earlier
peers, late academic startups were similarly
successful as diversifiers and spin-offs.7 The log-likelihood
of Model 3 is significantly above that of Model 1,
indicating that separating the two cohorts of academic
startups improves our ability to explain exit hazards.
We thus conclude that the inferior performance of
7Differences between the respective coefficient estimates are
not statistically significant at conventional levels.
academic startups reported by
limited to the first cohort. In contrast, post-1997 academic
startups were able to compete successfully with other
types of entrants.
4.2 Doctoral research training and the performance
The above results point to an increasing
competitiveness of academic startups after 1997, which might
provide some justification to the recent policy efforts
toward academic entrepreneurship in Germany.
However, as discussed above, not only academic startups
utilize scientific knowledge. In particular, we expect
that entrants of all types may benefit from having
founders and inventors trained in scientific research.
To test the relevance of founders’ scientific
knowledge, we distinguish spin-offs with at least one founder
who holds a doctoral degree from all other spin-offs
(Model 4 in Table 4). This provides no support for our
conjecture, as we do not find a significant difference
in the exit hazards of both groups. Model 5 is another
variant of Model 3 featuring two new variables.
“Opportunity entrepreneur” uses pre-entry patents held by
the firm’s founder as a proxy of opportunity
entrepreneurship, i.e., entry primarily driven by the motivation
to exploit a business opportunity rather than induced
by external factors such as job loss or deteriorating
Robust standard errors in
ap<0.1; bp<0.05; cp<0.01
Academic startup (post-1997)
(Reynolds et al. 2002; Buenstorf
.8 We also control for the firm’s overall patent
stock at entry (measured in the post-entry year to allow
for lags in patent application). We obtain a
significantly negative coefficient estimate for the founder
patent variable, suggesting that opportunity
entrepreneurship is systematically related to firm performance.
In contrast, the overall patent stock at time of entry has
no discernible effect on longevity.
The final two hazard rate models (Models 6 and
7 in Table 4) address the role of inventors with
completed doctoral training. To limit endogeneity
concerns, we again focus on early patents (entry year +
1). Controlling for the firm’s overall patent stock at
entry (Model 6) or the overall number of inventors
(Model 7), our estimates point to a systematic link
between early hires of doctorate-holding R&D staff
and firm performance (coefficient estimates are
significant at the 5 and 10% levels, respectively). The
hazard of academic startups is most strongly affected
by accounting for inventors with doctoral degrees. The
implied effect of being an academic startup is reduced
by about one-third (Model 7 vs. Model 3) and the
estimated coefficient is no longer significant. In
contrast, the coefficient estimated for spin-offs is reduced
by less than 10%, and that of diversifiers by about
15%. This suggests that among all entrants, (late)
academic startups benefitted most from hiring of R&D
staff who had received doctoral training at German
4.3 Further evidence on differences between types
and cohorts of entrants
Thus far, our empirical results indicate that academic
startups fared better in the post-1997 period of the
German laser industry than before, and that inventors
with doctoral training contributed to the performance
of entrants in this industry. While our research design
does not allow for causal inference, these findings are
consistent with a better utilization of scientific
knowledge after the governance of university-industry was
8Some diversifiers are small and of young age when they enter
the laser industry, and the available evidence suggests that their
founders are still active at the time of entry into lasers. In these
cases the prior patenting activity of founders is also included in
the “opportunity entrepreneur” measure.
changed.9 In this subsection, we provide further pieces
of evidence indicating that the German laser industry
became more “scientific” after 1997.
In Table 5, we compare pre- and post-1997
academic startups to other entrants whose founders did,
or did not, undergo doctoral training. The
comparison shows that “non-academic” entrants (i.e.,
diversifiers, intra-industry spin-offs) were more likely to
have founders with completed doctoral training after
1997. Academic startups entering in the post-1997
period were much more likely to patent, also prior
to entry, than both their earlier peers and the other
types of entrants. This stronger link between
patenting and academic entrepreneurship after 1997 is in
line with the findings of
Czarnitzki et al. (2016)
contrast, the share of patentees decreased among the
other types of entrants. Furthermore, while it is not
surprising that academic entrepreneurs had more
scientific publications than (doctorate-holding) founders
of other types of entrants, again there is a notable
increase from the first to the second period. Finally,
the time elapsed between completing doctoral
training and entry into the laser industry increased for both
groups of entrepreneurs with doctoral degrees, with
a constant difference of about 4 years between the
groups. These patterns provide further evidence that
the average quality of academic startups in the German
laser industry increased after 1997.
Further differences are shown in Table 6, which
lists the ten most common words found in the titles
of patent applications of the different groups
distinguished in Table 5.10 In spite of substantial
heterogeneity, some changes in the prevalence of title
words are noteworthy. First, “gas” is a common term
in patents from all types of entrants before 1997,
but more recently, it figures prominently only in the
patents of non-academic firms. In contrast,
“semiconductor” and “diode” have become more frequent over
time, but only in the patents of academic startups
and other entrants with doctorate-holding founders.
9The increased longevity of academic startups is also
consistent with earlier results for EXIST-funded startups in the optical
industries more generally
(Kulicke and Kripp 2013)
10About 60% of all priority patents are in English. For the
remaining ones, we identified the first family member with
English title in the DOCDB patent family. More than 95% of
all patent filings could be tracked in this way. Titles were
standardized and English stop words were excluded. Frequent terms
occurring in all subgroups were also excluded.
Rows one and two report number of firms, rows three to five report averages
These appear to have shifted to the new laser types
more rapidly. We moreover note that terms related
to materials-processing lasers, such as “producing,”
“machining,” “material,” or “ablation,” are not found
among the top ten title words of academic startups at
any time. Instead, their titles are more likely to refer to
specifics of laser design, such as “cavity,” “resonator,”
or “beam,” particularly in the post-1997 period.
Finally, in light of the crucial role that
Bhaskarabhatla and Klepper (2014)
attribute to the DPSS laser
in accounting for the recent evolution of the US
laser industry, we also explore its importance for
the German industry. Figure 2 shows the
development of DPSS-related patents in relationship to overall
patenting activities over time, as well as the
patenting activities of academic startups. We employ the
same measure for DPSS patents as
and Klepper (2014)
, i.e., the IPC classes
corresponding to USPTO 372 subclasses 40–50 (according to the
USPTO concordance scheme11). Numbers of
DPSSrelated patents are far from dominating overall
patenting activities in Germany. Apparently, there is still
significant innovation potential in other fields of laser
technology, which is consistent with the fact that the
DPSS laser did not choke entry into the German laser
This does not mean that the DPSS laser was
unrelated to entry. As a new and promising laser design,
it should have provided entrants with entrepreneurial
opportunities. DPSS patents can thus be interpreted as
indicators of opportunities discovered by various types
of firms in the laser industry. And if the performance
of post-1997 academic startups was related to
innovation, then they might stand out in their likelihood
of DPSS patenting. To probe into this conjecture, we
estimate a set of simple logit models pooling
patenting activities over a firm’s active years (conditional
on that it was active in the time period when DPSS
lasers were available) and having DPSS patenting as
the dependent variable. Model 8 (Table 7) studies
the likelihood of DPSS patenting for different types
of entrants, using pre-1997 academic startups as the
omitted reference group. We also control for the
duration that a given firm was active in the laser industry,
separating the years before and after emergence of the
DPSS laser. We find that post-1997 academic startups
11https://www.uspto.gov/web/patents/classification/ (last access
Table 7 Entry in DPSS technology
Logit regression model
(Dependent variable: patent filed in DPSS associated IPCs)
(8) (9) (10)
were most likely to have DPSS patents, whereas there
are no significant differences among the other types
of entrants. Earlier entrants were less likely to patent
DPSS technology. Model 9 separates spin-offs by
whether they have founders with doctoral training or
not, finding that, if anything, the latter were more
active in the DPSS field. Finally, in Model 10, we
again use founder-invented patents before the time
of entry as a proxy of opportunity entrepreneurship.
This variable indeed helps predict DPSS patenting.
It also changes the reference group of the analysis,
which now consists of those early academic startups
that were not based on opportunity (according to our
proxy), as well as the small group of entrants with
other or unknown backgrounds. Compared to this
group, the coefficient estimate for spin-offs turns
positive and marginally significant. We conclude from
this set of models that post-1997 academic startups
were disproportionately likely to exploit the
opportunities opened up by the DPSS laser. This provides
further evidence consistent with the conjecture that
these firms were more capable than their earlier peers.
5 Concluding remarks
The laser industry has received substantial attention
in the literature on industry evolution. One reason is
that, in contrast to the life cycle pattern observed in
many other industries, for decades, no shakeout was
observed in the US laser industry or in its German
counterpart. In addition, intra-industry spin-offs were
numerous and tended to perform well in both
countries. In this article, we focused on the recent evolution
of the German laser industry. Our main interest was
to trace how scientific knowledge, transferred to the
laser industry through academic entrepreneurship and
the labor mobility of inventors with doctoral training,
affected entry and survival in the population of laser
We found that the competitive performance of
academic startups increased over time. The available
evidence also suggests that they became more
innovative and disproportionately accounted for patents in
the novel DPSS lasers. These developments took place
after the governance of university-industry relations
changed in Germany, notably after support for
academic entrepreneurs was stepped up and new
infrastructure was provided for university technology
transfer. However, our empirical study is not designed as
a program evaluation able to identify causal effects,
and we also lack detailed data on the US laser
industry allowing for a systematic comparative analysis. We
thus consider our findings as no more than suggestive
regarding the link to policy changes, and we refer to
the USA only because it provides us with a benchmark
to assess the German development.
We moreover traced the role of doctoral training
as a source of firm capabilities in the laser
industry. According to our findings, inventors with doctoral
training are associated with higher firm performance,
and recent academic startups seem to have been most
successful in harnessing the scientific knowledge of
these inventors. In contrast, whether or not founders of
intra-industry spin-off had completed doctoral
training did not help account for the performance of their
Do these findings help us understand why there still
has not been a shakeout in firm numbers in the German
laser industry? Academic startups clearly helped
sustain the size of the firm population, and due to their
increased longevity, the share of academic startups
among all active firms has gradually gone up.
However, academic startups only account for a minority
of laser firms, and it would be far-fetched to attribute
the absence of a shakeout exclusively to them. That
the number of firms remained high in Germany also
seems to reflect that the DPSS laser has not (yet)
dominated the laser industry as much as it did in the
USA. Other submarkets have remained competitive to
date and continue to provide opportunities for entry.
This includes submarkets for industrial
materialsprocessing lasers, where German producers are among
the global market leaders, but also new submarkets for
ultra-high-frequency lasers for research applications.
We conclude by noting two important limitations
of our analysis. First, data availability has limited our
ability to trace the role of labor mobility in accounting
for firm performance. While we could identify founder
backgrounds and inventors with doctoral training, no
information could be retrieved for non-patenting
employees. We hope that future work will be able to do
so using linked employer-employee data, but to date,
this type of data is not available for industry
studies of the type performed above. Second, our analysis
relied on longevity in the laser industry as a measure
of firm performance. As discussed above, this only
provides a partial picture of firm performance, since
surviving firms may differ widely in profitability and
growth. Even though systematic information on
profitability and growth could not be collected for the
complete firm population, it is worth noting that none
of the recent entrants into the laser industry appears
to challenge the industry leaders. This holds for
academic startups but also for intra-industry spin-offs.
Specifically, there is a dearth of intra-industry
spinoffs organized by top-level employees of leading
firms, which according to prior research would be
most likely to grow into new market leaders. Though
there still are opportunities for entrepreneurial entry
into the laser industry, apparently these are not
exploited by those who would be best suited to do so.
Judging from the earlier experience of other industries
, the lack of high-level entrepreneurial
activity does not bode well for the long-term prospects
of the German laser industry.
Funding information The authors received funding from the
German Federal Ministry of Education and Research (BMBF)
under grant number 16FWN001.
Open Access This article is distributed under the terms of the
Creative Commons Attribution 4.0 International License (http://
creativecommons.org/licenses/by/4.0/), which permits
unrestricted use, distribution, and reproduction in any medium,
provided you give appropriate credit to the original author(s)
and the source, provide a link to the Creative Commons license,
and indicate if changes were made.
Albrecht , H. ( 1997 ). Laserforschung in Deutschland 1960-1970. University of Stuttgart, Unpublished Habilitation Thesis.
Beugelsdijk , S. ( 2007 ). Entrepreneurial culture, regional innovativeness and economic growth . Journal of Evolutionary Economics , 17 ( 2 ), 187 - 210 . https://doi.org/10.1007/978-3- 540 -87910-7. arXiv: 1011 . 1669v3 .
Bhaskarabhatla , A. , & Klepper , S. ( 2014 ). Latent submarket dynamics and industry evolution: lessons from the US laser industry . Industrial and Corporate Change , 23 ( 6 ), 1381 - 1415 . https://doi.org/10.1093/icc/dtt060.
Blankenberg , A.K. , & Buenstorf , G. ( 2016 ). Regional coevolution of firm population, innovation and public research? Evidence from the West German laser industry . Research Policy , 45 ( 4 ), 857 - 868 . https://doi.org/10.1016/j. respol. 2016 . 01 .008.
Bromberg , J.L. ( 1991 ). The laser in America, 1950 - 1970 . Cambridge: MIT Press.
Buenstorf , G. ( 2007 ). Evolution on the shoulders of giants: entrepreneurship and firm survival in the German laser industry . Review of Industrial Organization , 30 ( 3 ), 179 - 202 . https://doi.org/10.1007/s11151-007-9132-1.
Buenstorf , G. ( 2009 ). Opportunity spin-offs and necessity spinoffs . International Journal of Entrepreneurial Venturing , 1 ( 1 ), 22 - 40 .
Buenstorf , G. , & Klepper , S. ( 2010 ). Why does entry cluster geographically? Evidence from the US tire industry . Journal of Urban Economics , 68 ( 2 ), 103 - 114 . https://doi.org/ 10.1016/j.jue. 2010 . 03 .005.
Buenstorf , G. , Fritsch , M. , Medrano , L.F. ( 2015 ). Regional knowledge, organizational capabilities and the emergence of the West German laser systems industry, 1975 - 2005 . Regional Studies, 49 ( 1 ), 59 - 75 . https://doi.org/10.1080/ 00343404. 2012 . 711947 .
Cohen , W.M. , & Klepper , S. ( 1996 ). A reprise of size and R & D. Economic Journal , 106 , 925 - 951 .
Czarnitzki , D. , Doherr , T. , Hussinger , K. , Schliessler , P. , Toole , A.A. ( 2016 ). Knowledge creates markets: the influence of entrepreneurial support and patent rights on academic entrepreneurship . European Economic Review , 86 , 131 - 146 . https://doi.org/10.1016/j.euroecorev. 2016 . 04 .010.
Dahl , M.S. , & Klepper , S. ( 2015 ). Whom do new firms hire Industrial and Corporate Change, 24 ( 4 ), 819 - 836 . https://doi.org/10.1093/icc/dtv026.
Egeln , J. , Dinges , M. , Knie , A. , Simon , D. , Braun-Thu¨rmann, H., Fryges , H. , Mu¨ller, K. ( 2010 ). Evaluation des Existenzgru¨ndungsprogramms EXIST III . Baden-Baden: Nomos.
Fabian , C. ( 2011 ). Technologieentwicklung im Spannungsfeld von Industrie, Wissenschaft und Staat . Unpublished doctoral dissertation , Technical University Bergakademie Freiberg.
Frenken , K. , & Boschma , R.A. ( 2012 ). Economic development as a branching process , In: Buenstorf, G . (Ed.): Evolution, Organization and Economic Behavior (pp. 185 - 196 ). Cheltenham: Edward Elgar.
Fritsch , M. , & Medrano Echalar , L.F. ( 2015 ). New technology in the region-agglomeration and absorptive capacity effects on laser technology research in West Germany, 1960 - 2005 . Economics of Innovation and New Technology, 24 , 65 - 94 . https://doi.org/10.1080/10438599. 2014 . 897861 .
Fritsch , M. , & Wyrwich , M. ( 2014 ). The long persistence of regional levels of entrepreneurship: Germany, 1925 - 2005 . Regional Studies, 48 ( 6 ), 955 - 973 . https://doi.org/10. 1080/00343404. 2013 . 816414 .
Gort , M. , & Klepper , S. ( 1982 ). Time paths in the diffusion of product innovations . The Economic Journal , 92 ( 367 ), 630 - 653 .
Grupp , H. ( 2000 ). Learning in a science-driven market: the case of lasers . Industrial & Corporate Change , 9 ( 1 ), 143. https://doi.org/10.1093/icc/9.1.143.
Heblich , S. , & Slavtchev , V. ( 2014 ). Parent universities and the location of academic startups . Small Business Economics , 42 ( 1 ), 1 - 15 . https://doi.org/10.1007/s11187-013-9470-3. http://link.springer.com/article/10.1007/s11187-013-9470-3 http://link.springer.com/content/pdf/10.1007% 2Fs11187 - 013 - 9470-3.pdf.
Helfat , C.E. , & Liebermann , M.B. ( 2002 ). The birth of capabilities: market entry and the importance of prehistory . Industrial and Corporate Change , 11 ( 4 ), 725 - 760 . https://doi.org/10.1093/icc/11.4.725. arXiv: 1011 . 1669v3 .
Hidalgo , C.A. , Winger , B. , Baraba˙si, A.L. , Hausmann , R. ( 2007 ). The product space conditions the development of nations . Science , 317 ( 5837 ), 482 - 487 . https://doi.org/10. 1126/science.1144581. arXiv: 0708 . 2090 .
Jovanovic , B. , & MacDonald , G.M. ( 1994 ). The life cycle of a competitive industry . Journal of Political Economy , 102 ( 2 ), 322 - 347 . https://doi.org/10.1086/261934. -.
Klepper , S. ( 1996 ). Entry, exit and growth, and innovation over the product life cycle . American Economic Review1 , 86 ( 3 ), 562 - 583 .
Klepper , S. ( 1997 ). Industry life cycles . Industrial and Corporate Change , 6 ( 1 ), 145 - 182 . https://doi.org/10.1093/icc/6.1. 145.
Klepper , S. ( 2001 ). Employee startups in high-tech industries . Industrial and Corporate Change , 10 ( 3 ), 639 - 674 . https://doi.org/10.1093/icc/10.3.639.
Klepper , S. ( 2002 ). Firm survival and the evolution of oligopoly . The RAND Journal of Economics , 33 ( 1 ), 37 - 61 .
Klepper , S. ( 2016 ). Experimental capitalism: the nanoeconomics of american high-tech industries . Princeton: Princeton University Press.
Klepper , S. , & Sleeper , S. ( 2005 ). Entry by spinoffs . Management Science , 51 ( 8 ), 1291 - 1306 . https://doi.org/10.1287/ mnsc.1050.0411.
Kulicke , M. ( 2014 ). 15 Jahre EXIST Existenzgru¨ndungen aus der Wissenschaft Entwicklung des Fo˙rderprogramms von 1998 bis 2013 . Frauenhofer ISI , Karlsruhe.
Kulicke , M. , & Kripp , K. ( 2013 ). Ergebnisse und Wirkungen des Fo¨rderprogramms EXIST-Gru˙nderstipendium . Frauenhofer ISI , Karlsruhe.
Lieberman , M.B. , & Montgomery , D.B. ( 2013 ). Conundra and progress: research on entry order and performance . Long Range Planning , 46 ( 4-5 ), 312 - 324 . https://doi.org/10. 1016/j.lrp. 2013 . 06 .005.
Marx , M. , Strumsky , D. , Fleming , L. ( 2009 ). Mobility, skills, and the Michigan non-compete experiment . Management Science , 55 ( 6 ), 875 - 889 .
Neffke , F. , Henning , M. , Boschma , R. ( 2011 ). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions . Economic Geography , 87 ( 3 ), 237 - 265 . https://doi.org/10.1111/j.1944- 8287 . 2011 . 01121 .x.
Nelson , R.R. , & Winter , S.G. ( 1982 ). An evolutionary theory of economic change . Cambridge MA: Belknap Press. arXiv: 1011 . 1669v3 .
Peltoniemi , M. ( 2011 ). Reviewing industry life-cycle theory: avenues for future research . International Journal of Management Reviews , 13 ( 4 ), 349 - 375 . https://doi.org/10.1111/ j.1468- 2370 . 2010 . 00295 .x.
Perkmann , M. , Tartari , V. , Mckelvey , M. , Autio , E. , Brostro˙m, A. , Este , P.D. , Fini , R. , Geuna , A. , Grimaldi , R. , Hughes , A. , Krabel , S. , Kitson , M. , Llerena , P. , Lissoni , F. , Salter , A. , Sobrero , M. ( 2013 ). Academic engagement and commercialisation: a review of the literature on university-industry relations . Research Policy , 42 ( 2 ), 423 - 442 . https://doi.org/10.1016/j.respol. 2012 . 09 .007.
Reynolds , P.D. , Camp , S.M. , Bygrave , W.D. , Autio , E. , Hay , M. ( 2002 ). Global entrepreneurship monitor: 2001 executive report . Tech. rep., Business Council for the United Nations.
Samila , S. , & Sorenson , O. ( 2011 ). Venture capital, entrepreneurship, and economic growth . Review of Economics and Statistics , 93 ( 1 ), 338 - 349 .
Schoen , A. , Heinisch , D. , Buenstorf , G. ( 2014 ). Playing the 'Name Game' to identify academic patents in Germany . Scientometrics, 101 ( 1 ), 527 - 545 . https://doi.org/10. 1007/s11192-014-1400-x.
Sleeper , S.D. ( 1998 ). The role of firm capabilities in the evolution of the laser industry: the making of a high-tech market . Carnegie Mellon University, PhD Dissertation .
Stinchcombe , A.L. ( 1965 ). Social structure and organizations . In March, J.G. (Ed.) Handbook of organizations (pp. 142 - 193 ). Chicago: Rand Mc Nally.
Sutton , J. ( 1998 ). Technology and market structure: theory and history . Cambridge: MIT Press.
Trumpf Group . ( 2016 ). Annual report 2015/16. Ditzingen: Tech. rep.
Utterback , J.M. , & Sua˙rez, F.F. ( 1993 ). Innovation, competition, and industry structure . Research Policy , 22 ( 1 ), 1 - 21 . https://doi.org/10.1016/ 0048 - 7333 ( 93 ) 90030 -L.
Weterings , A. , & Marsili , O. ( 2015 ). Spatial concentration of industries and new firm exits: does this relationship differ between exits by closure and by M&A Regional Studies , 49 ( 1 ), 44 - 58 . https://doi.org/10.1080/00343404. 2012 . 726708 .
Zellner , C. ( 2003 ). The economic effects of basic research: evidence for embodied knowledge transfer via scientists' migration . Research Policy, 32 ( 10 ), 1881 - 1895 . https://doi.org/10.1016/S0048- 7333 ( 03 ) 00080 - 5 .