J Evol Econ (2004) 14: 605 616
DOI: 10.1007/s00191-004-0228-6
© Springer-Verlag 2004
Entrepreneurship and regional growth:
an evolutionary interpretation
David B. Audretsch and Max Keilbach
Max Planck Institute for Research into Economic Systems, Kahlaische Straße 10, 07745 Jena, Germany
(e-mail: {audretsch,keilbach}@mpiew-jena.mpg.de)
Abstract. While the neoclassical growth theory considered economic growth as a
process of mere accumulation of production capital, the endogenous growth the-
ory shifted the lens to the importance of knowledge in the production process and
its potential to create spillovers. We argue in this paper that there is a gap be-
tween knowledge and exploitable knowledge or economic knowledge. Economic
knowledge emerges from a selection process across the generally available body
of knowledge, actively driven by economic agents. This paper suggests that en-
trepreneurship is an important mechanism in driving that selection process hence
in creating diversity of knowledge, which in turn serves as a mechanism facilitating
the spillover of knowledge. We provide empirical evidence that regions with higher
levels of entrepreneurship indeed exhibit stronger growth in labor productivity.
Keywords: Regional growth  Labor productivity  Entrepreneurship capital 
Selection process
JEL Classification: M13, O32, O47
1 Introduction
Why do some regions grow more than others? One answer was provided by neoclas-
sical economics. Robert Solow (1956) argued that economic growth is determined
by the stocks of capital and labor. More recently Paul Romer (1986) and others

We are grateful to the ZEW in Mannheim for research support and to the German Science Foun-
dation (DFG) for financial support under research grant number STA 169/10-2. We are indebted to two
anonymous referees and the editor of the special issue for their helpful comments.
Correspondence to: D.B. Audretsch
606 D.B. Audretsch and M. Keilbach
extended the neoclassical model of growth by suggesting that not only is knowl-
edge is an important factor generating growth, but because it spills over for use by
third-party firms it is actually the most potent factor.
The purpose of this paper is to argue that this assessment of the role of knowl-
edge overlooks some of the most fundamental mechanisms driving the process of
economic evolution. In particular, it is important to make a distinction between
knowledge that is generally available and economic knowledge, i.e. knowledge
that is economically exploitable and thus enters the production process. Economic
knowledge emerges from a selection process across a diversity of ideas or across
the generally available body of knowledge. Thus, the spillover process that Romer
and the endogenous growth theory assumes to be automatic is not at all automatic,
rather it is a process that is actively driven by economic agents. This paper suggests
that entrepreneurship is an important mechanism in driving that selection process
hence in creating diversity of knowledge, which in turn serves as a mechanism
facilitating the spillover of knowledge.
In the Section 2 of the paper regional growth is viewed through the framework of
an evolutionary model. Entrepreneurship emerges as playing an important role by
creating diversity through the spillover of knowledge. In Section 3 the link between
entrepreneurship and economic growth is tested across German regions. Finally,
a summary and conclusions are provided in Section 4. In particular, the empirical
evidence based on German regions suggests that the neoclassical approach does not
adequately explain growth. Rather, inclusion of entrepreneurship provides a more
adequate explanation of why some regions grow more than others.
2 Entrepreneurship and the selection of new ideas 
an evolutionary perspective
Evolutionary economics focuses on two central principles  diversity and selection
(Nelson and Winter, 1982). The process of evolution takes place by a process of
selection among diverse entities, which propels an economy into a new direction.
An economy with no diversity and no selection will not evolve. It will remain
permanently locked in a long-run steady state equilibrium.
While Nelson and Winter (1982) made significant progress in identifying the
role that diversity and selection play in shaping economic evolution, they were
more vague about the sources of diversity. What are the sources of diversity, and
why does it pay economic agents to invest in diversity?
One answer, provided by Nelson and Winter (1982), is that diversity emanates
from investments in research and development (R&D). Firms have an incentive to
invest in the creation of new economic knowledge. Thus, investments in R&D and
human capital are an important source of diversity by generating new economic
knowledge.
However, Arrow (1962) suggested that knowledge was different than other
types of economic goods. In particular, knowledge is characterized by greater un-
certainty, increased asymmetries and high costs of transaction. This means that
knowledge is not the equivalent of economic knowledge, but rather a gap exists
between knowledge and economically commercialized knowledge. This implies
Entrepreneurship and regional growth: an evolutionary interpretation 607
that firm investment in R&D and human capital does not automatically result in the
generation of diversity, but will be constrained by these underlying characteristics
creating the filter between knowledge and economic knowledge.
An implication ofArrow s characterization of the gap distinguishing knowledge
from economic knowledge is that the valuation of new ideas will be distributed dif-
ferently across individuals, or economic agents. Those economic agents placing
a high value on knowledge that is not valued as highly by the hierarchical deci-
sion making organizations in incumbent firms will face an incentive to become
entrepreneurs and start a new firm in order to appropriate the value of that knowl-
edge. When economic agents recognize economic opportunities emerging from
knowledge generated but not commercially exploited in incumbent firms, and act
on that economic opportunity, they become entrepreneurs.
Thus, when a new firm is launched, its prospects are shrouded in uncertainty.
If the new firm is built around a new idea, i.e., potential innovation, it is uncertain
whether there is sufficient demand for the new idea or if some competitor will have
the same or even a superior idea. Even if the new firm is formed to be an exact replica
of a successful incumbent enterprise, it is uncertain whether sufficient demand for
a new clone, or even for the existing incumbent, will prevail in the future. Tastes
can change, and new ideas emerging from other firms will certainty influence those
tastes.
Finally, an additional layer of uncertainty pervades a new enterprise. It is not
known how competent the new firm really is, in terms of management, organiza-
tion, and workforce. At least incumbent enterprises know something about their
underlying competencies from past experience. By contrast, a new enterprise is
burdened with uncertainty as to whether it can produce and market the intended
product as well as sell it. Thus, the degree of uncertainty will typically exceed that
confronting incumbent enterprises.
Casper and Whitely (2002) point out that the economic opportunity provided
by new knowledge is conditioned by the institutional context. Innovative firms may
face different risks in different regimes and develop the links that these risks have
with institutions prevalent in particular sectors and countries. In particular, labor
laws and access to finance will both create and constrain, but in any event, create
the conditional context for the recognition of possible opportunities and the cost of
creating a new enterprise to appropriate the economic value of those opportunities.
This initial condition of not just uncertainty, but greater degree of uncertainty
vis-Ä…-vis incumbent enterprises in the industry is captured in the theory of firm
selection and industry evolution proposed by Boyan Jovanovic (1982). Jovanovic
presents a model in which the new firms, which he terms entrepreneurs, face costs
that are not only random but also differ across firms.A central feature of the model is
that a new firm does not know what its cost function is, that is its relative efficiency,
but rather discovers this through the process of learning from its actual post-entry
performance. In particular, Jovanovic (1982) assumes that entrepreneurs are unsure
about their ability to manage a new-firm startup and therefore their prospects for
success. Although entrepreneurs may launch a new firm based on a vague sense of
expected post-entry performance, they only discover their true ability  in terms of
managerial competence and of having based the firm on an idea that is viable on the
608 D.B. Audretsch and M. Keilbach
market  once their business is established. Those entrepreneurs who discover that
their ability exceeds their expectations expand the scale of their business, whereas
those discovering that their post-entry performance is less than commensurate with
their expectations will contact the scale of output and possibly exit from the industry.
Thus, Jovanovic s model is a theory of noisy selection, where efficient firms grow
and survive and inefficient firms decline and fail.
The role of learning in the selection process has been the subject of considerable
debate. On the one hand is what has been referred to as the Larackian assumption
that learning refers to adaptations made by the new enterprise. In this sense, those
new firms that are the most flexible and adaptable will be the most successful in
adjusting to whatever the demands of the market are. As Nelson and Winter (1982,
p. 11) point out, "Many kinds of organizations commit resources to learning; or-
ganizations seek to copy the forms of their most successful competitors." On the
other hand is the interpretation that the role of learning is restricted to discovering
if the firm is producing a good or offering a service that is compatible with mar-
ket viability. Under this interpretation the new enterprise is not necessarily able to
adapt or adjust to market conditions, but receives information based on its market
performance with respect to its fitness in terms of meeting demand most efficiently
vis-Ä…-vis rivals. The theory of organizational ecology proposed by Michael T. Han-
nan and John Freeman (1989) most pointedly adheres to the notion that, "We assume
that individual organizations are characterized by relative inertia in structure." That
is, firms learn not in the sense that they adjust their actions as reflected by their fun-
damental identity and purpose, but in the sense of their perception. When viewed
from this evolutionary perspective, the startup of a new firm injects diversity into
the market. The process of entrepreneurship, or starting a new firm, is therefore a
mechanism generating diversity and the spillover of knowledge. As a result of the
startup, knowledge is transformed into new approaches that otherwise would have
remained unexplored.
Thus, entrepreneurship is an important source of diversity by transforming
knowledge into economic knowledge that otherwise would have remained uncom-
mercialized. This would suggest that those regions with a greater amount of en-
trepreneurial activity also have a greater degree of diversity, which should result in
higher rates of growth.
Nelson and Winter (1982) and Winter (1984) have suggested that more diversity
will be generated under what Winter (1984) terms as the entrepreneurial regime
compared to what he terms as the routinized regime. Under the routinized tech-
nological regime innovative activity will tend to be more incremental in nature
and thus be characterized by less diversity. By contrast, under the entrepreneurial
regime innovative activity tends to be characterized by a greater degree of diversity.
the agent will tend to appropriate the value of his new ideas within the boundaries
of incumbent firms. Thus, the propensity for new firms to be started should be
associated with a greater degree of diversity and therefore greater growth.
Entrepreneurship and regional growth: an evolutionary interpretation 609
3 Linking entrepreneurship to regional growth
Several previous studies have attempted to link entrepreneurship to regional growth.
The unit of observation for these studies is at the spatial level, either a city, region,
or state. The most common and most exclusive measure of performance is growth,
typically measured in terms of employment growth. These studies have tried to
link various measures of entrepreneurial activity, most typically startup rates, to
economic growth. Other measures sometimes used include the relative share of
SMEs, and self-employment rates.
Holtz-Eakin and Kao (2003) also examine the impact of entrepreneurship on
growth. Their spatial unit of observation is for states. Their measure of growth is
productivity change over time. A vector autoregression analysis shows that varia-
tions in the birth rate and the death rate for firms are related to positive changes in
productivity. They conclude that entrepreneurship has a positive impact on produc-
tivity growth, at least for the case of the United States.
Audretsch and Fritsch (1996) analyzed a database identifying new business star-
tups and exits from the social insurance statistics in Germany to examine whether
a greater degree of turbulence leads to greater economic growth, as suggested by
Schumpeter in his 1911 treatise. These social insurance statistics are collected for
individuals. Each record in the database identifies the establishment at which an
individual is employed. The startup of a new firm is recorded when a new establish-
ment identification appears in the database, which generally indicates the birth of a
new enterprise. While there is some evidence for the United States linking a greater
degree of turbulence at the regional level to higher rates of growth for regions
(Reynolds, 1999), Audretsch and Fritsch (1996) find that the opposite was true for
Germany during the 1980s. In both the manufacturing and the service sectors, a
high rate of turbulence in a region tends to lead to a lower and not a higher rate
of growth. They attribute this negative relationship to the fact that the underlying
components  the startup and death rates  are both negatively related to subsequent
economic growth. Those areas with higher startup rates tend to experience lower
growth rates in subsequent years. Most strikingly, the same is also true for the death
rates. The German regions experiencing higher death rates also tend to experience
lower growth rates in subsequent years. Similar evidence for Germany is found by
Fritsch (1997).
Audretsch and Fritsch (1996) conjectured that one possible explanation for the
disparity in results between the United States and Germany may lie in the role that
innovative activity, and therefore the ability of new firms to ultimately displace the
incumbent enterprises, plays in new-firm startups. It may be that innovative activity
did not play the same role for the German Mittelstand as it does for SMEs in the
United States. To the degree that this was true, it may be hold that regional growth
emanates from SMEs only when they serve as agents of change through innovative
activity.
The empirical evidence suggested that the German model for growth provided
a sharp contrast to that for the United States. While Reynolds (1999) had found
that the degree of entrepreneurship was positively related to growth in the United
States, a series of studies by Audretsch and Fritsch (1996) and Fritsch (1997) could
610 D.B. Audretsch and M. Keilbach
not identify such a relationship for Germany. However, the results by Audretsch
and Fritsch were based on data from the 1980s.
Divergent findings from the 1980s about the relationship between the degree
of entrepreneurial activity and economic growth in the United States and Germany
posed something of a puzzle. On the one hand, these different results suggested that
the relationship between entrepreneurship and growth was fraught with ambiguities.
No confirmation could be found for a general pattern across developed countries.
On the other hand, it provided evidence for the existence of distinct and different
national systems. The empirical evidence clearly suggested that there was more
than one way to achieve growth, at least across different countries. Convergence
in growth rates seemed to be attainable by maintaining differences in underlying
institutions and structures.
However, in a more recent study,Audretsch and Fritsch (2002) find that different
results emerge for the 1990s. Those regions with a higher startup rate exhibit higher
growth rates. This would suggest that, in fact, Germany is changing over time, where
the engine of growth is shifting towards entrepreneurship as a source of growth. The
results of their 2002 paper suggest an interpretation that differs from their earlier
findings. Based on the compelling empirical evidence that the source of growth in
Germany has shifted away from the established incumbent firms during the 1980s
to entrepreneurial firms in the 1990s, it would appear that a process of convergence
is taking place between Germany and the United States, where entrepreneurship
provides the engine of growth in both countries. Despite remaining institutional
differences, the relationship between entrepreneurship and growth is apparently
converging in both countries.
The positive relationship between entrepreneurship and growth at the regional
level is not limited to Germany in the 1990. For example, Foelster (2000) examines
not just the employment impact within new and small firms but on the overall link
between increases in self-employment and total employment in Sweden between
1976-1995. By using a Layard-Nickell framework, he provides a link between
micro behavior and macroeconomic performance, and shows that increases in self-
employment shares have had a positive impact on regional employment rates in
Sweden.
Hart and Hanvey (1995) link measures of new and small firms to employment
generation in the late 1980s for three regions in the United Kingdom. While they
find that employment creation came largely from SMEs, they also identify that
most of the job losses also came from SMEs.
Callejon and Segarra (1999) use a data set of Spanish manufacturing indus-
tries between 1980-1992 to link new-firm birth rates and death rates, which taken
together constitute a measure of turbulence, to total factor productivity growth in
industries and regions. They adopt a model based on a vintage capital framework in
which new entrants embody the edge technologies available and exiting businesses
represent marginal obsolete plants. Using a Hall type of production function, which
controls for imperfect competition and the extent of scale economies, they find that
both new-firm startup rates and exit rates contribute positively to the growth of total
factor productivity in regions as well as industries.
Entrepreneurship and regional growth: an evolutionary interpretation 611
Estimation approach
Viewed through the lens of neoclassical economics, regional growth is shaped by
the traditional factors of labor and capital. The contribution of the new endogenous
growth theory was to suggest an additional source of growth, knowledge. How-
ever, the previous section argued that the presence of knowledge is not sufficient
to generate growth. Rather, what matters is not knowledge, per se, but rather the
propensity for that knowledge to create diversity, which, in turn, results in growth.
Entrepreneurship is one mechanism, although certainly not the only one, transform-
ing knowledge into diversity. To measure the impact of entrepreneurship capital on
the dynamics of labor productivity we estimate the following simple growth equa-
tion1

log (yi,t1/yi,t0) =Ä… - 1 - e-² log(yi,t0) +Xg + ui,t1 (1)
where i denotes regions, yi is GDP in region i divided by the regions number of
employees, hence labor productivity, t0, t1 are time instances (in our case 1992
and 2000) and X is a set of variables that might account for regional differences in
the growth rate of labor productivity2.
The dependent variable, regional growth of labor productivity, might depend
on the structure of the regional economy. That is, regional growth might be more
pronounced in regions where a larger proportion of fast growing industries are
located. Now, a priori, a fast growing industry is also one where a large number
of startups can be observed. If this holds, then both, startups and growth depend
on a third variable, the industry structure of the region. To avoid the resulting
endogeneity bias in the regression process, we control for this industry structure
using two steps. First, we include the level of the regional R&D activity in X. This
will correct for the fact that knowledge based industries are usually those who show
higher growth rates. Second, we introduce a second equation of the form
Ei = f(yi,t0, Hi) (2)
that explains entrepreneurship capital in region i as a function of the region s labor
productivity and the region s human capital level. Both equations are estimated
simultaneously using three stage least squares regressions. By specifying explicitly
both equations as recursive model, we eliminate an endogeneity bias that would
occur due to the fact that a startup activity might depend on the growth dynamics
of a region (see e.g. Intriligator et al., 1996).
Measurement issues
Measurement of entrepreneurship is no less complicated than is measuring the tradi-
tional factors of production. Just as measuring capital, labor and knowledge invokes
1
See e.g. Barro and Sala-I-Martin (1995, p. 384) and the subsequent literature on convergence.
2
See Barro and Sala-I-Martin (1992, 1995) and Mankiw, Romer and Weil (1992) for a further
discussion of this approach.
612 D.B. Audretsch and M. Keilbach
numerous assumptions and simplifications, creating a metric for entrepreneurship
presents a challenge. Many of the elements determining entrepreneurship defy
quantification. In any case, entrepreneurship is multifaceted and heterogeneous.
However, entrepreneurship manifests itself in a singular way  the startup of new
enterprises. Thus, we propose using new-firm startup rates as an indicator of en-
trepreneurship. Ceteris paribus, higher startup rates reflect higher levels of en-
trepreneurship. Our data will consist in a cross-section of 327 West-German regions
or Kreise for the year 1992 if not indicated otherwise. Sources and construction of
the data is as follows.
Output is measured as the Kreise s Gross ValueAdded of all industries corrected for
purchases of goods and services, VAT and shipping costs. Statistics are published
every two years for Kreise by the Working Group of the Statistical Offices of the
German Länder, under  Volkswirtschaftiche Gesamtrechnungen der Länder  .
Physical capital. The stock of capital used in the manufacturing sector of the
Kreise has been estimated using a perpetual inventory method which computes
the stock of capital as a weighted sum of past investments. In the estimates we
used a ?-distribution with p=9 and a mean age of q=14. Type of survival function
as well as these parameters have been provided by the German Federal Statistical
Office inWiesbaden. This way, we attempted to obtain maximum coherence with the
estimates of the capital stock of the German producing sector as a whole as published
by the Federal Statistical Office. Data on investment at the level of German Kreise
is published annually by the Federal Statistical Office in the series  E I 6 . These
figures however are limited to firms of the producing sector, excluding the mining
industry, with more than 20 employees. The vector of the producing sector as a
whole has been estimated by multiplying these values such that the value of the
capital stock of Western Germany  as published in the Statistical Yearbook  was
attained. Note that this procedure implies that estimates for Kreise with a high
proportion of mining might be biased. Note also that for protection purposes, some
Kreise did not publish data on investment (like e.g. the city of Wolfsburg, whose
producing sector is dominated by Volkswagen). Therefore five Kreise are treated
as missing.
Labor. Data on labor is published by the Federal Labor Office, Nürnberg that
reports number of employees liable to social insurance by Kreise. A region s labor
productivity is simply the region s output divided by labor input.
R&D Activity is expressed as number of employees engaged in R&D in the public
(1992) and in the private sector (1991). With this approach we follow the exam-
ples of Grilliches (1979), Jaffe (1989), and Audretsch and Feldman (1996). With
these data, it was impossible to make a distinction between R&D-employees in the
producing and non-producing sectors. Regression results therefore will implicitly
include spillovers from R&D of the non-producing sector to the producing sectors.
We presume however that this effect is rather low.
Human Capital is expressed as the percentage of employees with a high level
of qualification (Masters degree). Data on qualification levels of employees are
published for 1989 by the Federal Research Institute for Regional Geography and
Entrepreneurship and regional growth: an evolutionary interpretation 613
Regional Planning (Bundesforschungsanstalt für Landeskunde und Raumordnung)
in their Volume 47. Since that data is not available for 1992 we use data for 1989 as
a proxy. These data refers to the Kreis as a whole, i.e. the figures do not refer to the
producing sector alone. Thus (as in the case of R&D - personnel) estimates implic-
itly include spillovers of human capital from the non-producing to the producing
sectors.
Entrepreneurship capital is measured as the number of startups in the respective
region relative to its population, which reflects the propensity of inhabitants of
a region to start a new firm. While such a measure clearly does not capture the
universe of knowledge worker mobility, this measure surely reflects the mobility of
knowledge workers. After all, mobility is typically inherent in the startup of a new
firm, since the founder moves from a situation in an incumbent organization to start
the new enterprise. The data on startups is taken from the ZEW foundation panels
that are based on data provided biannually by Creditreform, the largest German
credit-rating agency. This data contains virtually all entries  hence startups  in the
German Trade Register, especially for firms with large credit requirements as e.g.
high-technology firms.3 By now, there are 1.6 million entries for Western-Germany.
Since number of startups is subject to a greater level of stochastic disturbance over
short time periods, it is prudent to compute the measure of entrepreneurship capital
based on startup rates over a longer time period. We therefore used the number of
startups between 1989-1992. Lagged values of start-up rates are used in order to
avoid problems of simultaneity between output and entrepreneurship. This lagged
relationship reflects causality between entrepreneurship capital in one period and
economic output in subsequent periods.
To test the argument that entrepreneurs are the driving force in the selection
of economically viable ideas, we use different measures of entrepreneurship. Our
measure of general entrepreneurship corresponds to startup activities in all indus-
tries. This implies that more than half of the startups are in the retail and catering
industries hence shops and restaurants. Assuming that startups in innovative in-
dustries correspond rather to the process of the selection of ideas discusses above,
we considered two modified measures of entrepreneurship. The first one restricts
entrepreneurship capital to include only startup activity in high-technology man-
ufacturing industries (whose R&D-intensity is above 2.5%). The second measure
restricts entrepreneurship capital to include only startup activity in the ICT in-
dustries, i.e. firms in the hard- and software industries. The both latter measures
overlap, i.e. some of the ICT industries are also classified under high-technology
manufacturing.
Table 1 reports on the regression results from estimating Equations (1) and (2)
simultaneously. First, from the upper part of Table 1, it is noticeable that the co-
efficient of log(yi,t0)is negative and within a region that has often been reported
within this kind of equation.4 This finding implies that regions with a higher level
3
Firms with low credit requirements, with a low number of employees or with unlimited legal forms
are registered only with a time lag. These are typically retail stores or catering firms. See Harhoff and
Steil (1997) for more detail on the ZEW foundation panels.
4
See again Barrow and Sala-i-Martin, 1995.
614 D.B. Audretsch and M. Keilbach
Table 1. Results of three stage least squares regressions of Equations (1) and (2)
Dependent variable: Growth rate of labor productivity
Constant 0.1379""" 0.1443""" 0.1423"""
(5.98) (6.17) (6.16)
Log(Y/L) 1992 -0.0384""" -0.0365""" -0.0356"""
(-4.81) (-5.14) (-5.18)
R&D Activity 0.0021""" 0.0020""" 0.0020"""
(3.49) (3.37) (3.55)
General entrepreneurship 2.7309"
(1.89)
High tech entrepreneurship 16.4787""
(2.04)
ICT entrepreneurship 19.6472""
(2.02)
Ç2 48.48""" 47.65""" 47.74"""
(324)
(p-value) (0.0000) (0.0000) (0.0000)
Dependent variable:
General entr. High tech entr. ICT entr.
Constant -0.0342""" -0.0063""" -0.0051"""
(-6.54) (-8.71) (-9.08)
Log (Y/L) 1992 0.0021"" 0.0002"" 0.0002
(2.27) (1.97) (1.58)
Human capital 0.0029""" 0.0005""" 0.0004"""
(5.91) (7.29) (7.79)
Ç2 73.37""" 97.83""" 103.34"""
(325)
(p-value) (0.0000) (0.0000) (0.0000)
of labor productivity show a lowd subsequent growth rate of this variable. The esti-
mated impact of regional R&D input is positive and significant for all estimations,
implying that R&D activity exersts a positive impact on the region s growth rate of
labor productivity.
All our measures of entrepreneurship capital exert a positive influence on the
dependent variable. The estimated level is at first sight strikingly high. This is of
course a measurement issue, since these variables are expressed as intensities, and
therefore range between zero and one. Thus, the high coefficients merely reflect
the low values of the entrepreneurship measures. It is noteworhty however that the
impact of our general measure of entrepreneurship is estimated as being smaller
and less significant as compared to our more high-tech oriented measures of en-
trepreneurship capital. Hence, given that we have corrected for R&D input, we find
that innovative startups exert a stronger impact on regional productivity growth.
The bottom part of Table 1 shows the results for equation (2). This estimation is
included to correct for an endogeneity bias in regressing entrepreneurship capital,
measured as startups, against productivity growth. We find that startup activity is
higher in regions with a higher level of labor productivity and in regions with a
higher level of human capital. Hence, the findings of our stylized reduced form
model are in accordance with the arguments on the relationship between human
Entrepreneurship and regional growth: an evolutionary interpretation 615
capital and the commercialization of economic ideas through the start up of new
firms given in Section 2.
Overall, the results provide compelling evidence that entrepreneurial activity
in the high-tech industries fosters economic growth. We take this as evidence in
support of our argument that entrepreneurial activity fosters the selection and trans-
formation of generally available knowledge into economic knowledge. This would
suggest that, while the Romer growth model assumed that knowledge capital is both
necessary and sufficient for knowledge spillovers, in fact entrepreneurship plays
an important role in creating diversity. Knowledge may be important for economic
growth, but the capacity for that knowledge to be transformed into diversity is also
important. Entrepreneurship is one such mechanism generating such diversity.
Conclusions
The propensity for growth to vary across geographic space is not a novel obser-
vation. Neoclassical economics attributes spatial variation in economic growth to
variation in capital and labor. More recently, endogenous growth theory has added
knowledge as a determinant of growth.
This paper has invoked the lens provided by evolutionary economics to argue
that knowledge is not sufficient to generate the diversity that is the driving force of
economic growth. Rather, additional mechanisms are required to transform knowl-
edge into diversity. One such mechanism is entrepreneurship. The startup of a new
firm contributes to diversity by attempting to commercialize knowledge that oth-
erwise would not be commercialized. Thus, ceteris paribus, a greater amount of
entrepreneurship is expected to be associated with more diversity and therefore
higher growth.
The results of this paper support the view that entrepreneurship enhances
growth. Those German regions with a greater degree of entrepreneurial activity
exhibit higher rates of growth of labor productivity. Whether these results hold for
other countries or for other time periods can only be ascertained through subsequent
research. However, the results from this study suggest that a more evolutionary lens
may be appropriate in understanding why growth rates vary across regions, at least
for the case of Germany.
References
Arrow KJ (1962) Economic welfare and the allocation of resources for invention. In: Nelson RR (ed)
The rate and direction of inventive activity, pp 609 626. Princeton University Press, Princeton, NJ
Audretsch D, Fritsch M (2002) Growth regimes over time and space. Regional Studies 36: 113 124
Audretsch DB, Keilbach M (2003) Entrepreneurship capital and economic performance. CEPR Discus-
sion Paper no. 3678, London, www.cepr.org
Callejon M, Segarra A (1999) Business dynamics and efficiency in industries and regions: the case of
Spain. Small Business Economics 13(4): 253 271
Cohen W, Levinthal D (1989) Innovation and learning: the two faces of R&D. Economic Journal 99(3):
569 596
Dosi G (1988) Sources, procedures, and microeconomic effects of innovation. Journal of Economic
Literature 26(3): 1120 1171
616 D.B. Audretsch and M. Keilbach
Griliches Z (1979) Issues in assessing the contribution of R&D to productivity growth. Bell Journal of
Economics 10: 92 116
Hannan MT, Freeman J (1989) Organizational ecology. Harvard University Press, Cambridge, MA
Intriligator M, Bodkin R, Hsiao C (1996) Econometric models, techniques and applications, 2nd edn.
Prentice Hall, Englewood Cliffs, NJ
Jovanovic B (1982) Selection and evolution of industry. Econometrica 50(2): 649 670
Nelson, RR, Winter SG (1982) An evolutionary theory of economic change. Harvard University Press,
Cambridge
Romer, PM (1986) Increasing returns and long-run growth. Journal of Political Economy 94(5): 1002
1037
Solow R (1956) A contribution to the theory of economic growth. Quarterly Journal of Economics 70:
65 94
Winter SG (1984) Schumpeterian competition in alternative technological regimes. Journal of Economic
Behavior and Organization 5: 287 320