Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
net/publication/235999914
CITATIONS READS
48 643
1 author:
Michael Fritsch
Friedrich Schiller University Jena
337 PUBLICATIONS 9,058 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Michael Fritsch on 28 May 2014.
Contents
Acknowledgments 346
References 347
Foundations and Trends R
in
Entrepreneurship
Vol. 9, No. 3 (2013) 249–364
c 2013 M. Fritsch
DOI: 10.1561/0300000043
Michael Fritsch
Abstract
This monograph reviews the current state of knowledge about the
effects of new business formation on regional development. These effects
are diverse and include the creation and destruction of employment,
introduction of innovations, structural change, and increasing produc-
tivity, among others. Theory particularly emphasizes the role that some
new businesses play in the diffusion of knowledge and innovation as
drivers of economic growth. I provide an explanatory approach that
highlights the competitive challenge that start-ups pose to incumbent
firms and discuss important implications. The overview of empirical
research particularly deals with the development of start-up cohorts,
identification of different types of indirect effects and their magnitude,
differences based on characteristics of entry, and regional variation. A
general conclusion is that the diverse indirect effects of new business
formation on development are much more important than the growth
effects created by newcomers. The diverse indirect effects of entry on
development are currently less than fully understood. Finally, I draw
conclusions for policy and put forward a number of important questions
for further research.
1 See,
for example, Wennekers and Thurik (1999), OECD (2003), Commission of the
European Communities (2003, 2010), Reynolds et al. (2005), and the contributions in
Audretsch et al. (2007) and Leitao and Baptista (2009).
251
252 The Link Between New Business Formation and Growth
1 The English edition of this book was published in 1934, more than 20 years later, and was
based on a considerably revised and particularly shortened version of the original German
text.
253
254 Emergence of the Research Field
2 Wennekers and Thurik (1999) suspect that the enthusiastic reactions to the Birch study
may be partly explained by a shift in policy toward supply-side economics at that time
when the overall economic situation in many countries was characterized by stagflation
and high unemployment.
3 For example, Acs (1984), Acs and Audretsch (1990), Brock and Evans (1986), Piore and
Sabel (1984), Sengenberger et al. (1990), and Storey and Johnson (1987).
4 Carree and Thurik (2010) provide an overview on the possible reasons for this development.
5 A start-up can be either a new firm or a new establishment of a multiplant enterprise.
The term “new business” is used here as an overall category that encompasses the setup
of new headquarters as well as the creation of a new subsidiary establishment.
255
6 Birch (1979) used microdata from the Dun & Bradstreet credit rating agency for the
United States in the 1969–1976 period.
7 Audretsch et al. (2004), Evans (1987), Davis et al. (1996), Sutton (1997), Haltiwanger
256
3.1 The Nature of Entrepreneurship 257
2 For
example, Baldwin (1995), Caves (1998), Disney et al. (2003), Foster et al. (2001, 2006),
and OECD (2003).
260 Theory: The Role of Entrepreneurship in Economic Development
Number of firms
Shake out
Number of entries
Number of firms
Number of exits
Time
Net entry
Fig. 3.1 Stylized paths in the numbers of entries, exits, and firms over the industry life
cycle.
3 Empirical research, however, shows that many innovative industries tend to be clustered
in space and that the development of such industry clusters can play a main role in
the development of these regions. Empirical examples, such as the Silicon Valley or the
automobile industry in Detroit (Klepper, 2010) clearly, demonstrate that a main vehicle
for the emergence of geographic clusters is the formation of spin-offs from incumbent firms,
which tend to locate in close geographic proximity to the incubator (Klepper, 2009).
4 The common explanation for why entrepreneurs enter markets that are already crowded
is that they are overconfident with regard to their chance of success and ill-informed or
unaware of the risks (Arabsheibani et al., 2000; Koellinger et al., 2007). And, indeed,
overconfidence does seem to be common among firm founders; one could even argue that
it is a necessary ingredient of new ventures, given the high risk of failure that would
262 Theory: The Role of Entrepreneurship in Economic Development
otherwise deter entry (Arabsheibani et al., 2000; Koellinger et al., 2007). Excessive entry
can occur in markets with low barriers to entry (for example, certain service industries) or
if public subsidies are available that lead to reduced costs of venture creation. Individuals
particularly prone to founding such types of business may be those individuals who face
relatively low opportunity costs, for example, due to being unemployed.
5 An exception is the analysis by Carree and Thurik (1999) of the Dutch retailing sector.
3.2 The Industry Perspective: New Firms and the Evolution of Markets 263
6 Empirical research shows that many founders of innovative firms acquired the relevant
knowledge by working in a research organization or in an incumbent firm (Klepper, 2009).
A main motivation of many of these founders is to commercialize certain knowledge not
being exploited at their former workplace. A quite common pattern is that the management
and the potential entrepreneur disagree about the realization of an idea, be it because of
different perceptions of the economic potential of this idea, the risk involved in bringing
the innovation to the marketplace, or for other reasons (Klepper, 2009).
7 In the literature it is assumed that recognition of entrepreneurial opportunities is partic-
ularly likely by persons who have respective technological knowledge or knowledge about
customers’ wants and markets, such as R&D employees or sales managers. See Shane
(2003) and Eckhardt and Shane (2011) for a more detailed discussion.
266 Theory: The Role of Entrepreneurship in Economic Development
Direct
New effect
businesses
Growth
Incumbent
firms Indirect
effect
8 Diverseempirical studies confirm that low productivity firms are more likely to exit than
firms with relatively high productivity. See, for example, Baldwin (1995), Carreira and
Teixeira (2011), and Wagner (2009). These studies, however, also find exit of relatively
high productivity firms, suggesting that low productivity is not the only reason for exit.
On average, young firms do not have higher but productivity levels than incumbents (van
Praag and Versloot, 2007). Some studies (for example, Wagner, 2009) find relatively low
average productivity levels in new firms compared to older firms.
270 Theory: The Role of Entrepreneurship in Economic Development
Securing efficiency
Acceleration of
Market process (selection) structural change
Amplified innovation
Greater variety
New Exiting
capacities capacities Improved
(direct effect): (indirect effect): competitiveness
Development Decline or
of new closure of
Growth
businesses incumbents
9 The supply-side effects include all consequences for the competitiveness of the regional
economy. Accordingly, the demand of new businesses for resources such as labor may be
labeled a demand-side effect. However, empirical analyses suggest that such demand-side
effects are rather small so that they can be almost ignored (see Section 4.4.2 for details).
3.4 Entry, Competition, and Growth: An Eclectic Explanatory Approach 271
10 Such a process was observed in the transformation of the former socialist economies of
Central and Eastern Europe, where new firms — the bottom-up component — had a con-
siderably strong impact on structural change (cf. Brezinski and Fritsch, 1996; Pfirrmann
and Walter, 2002).
272 Theory: The Role of Entrepreneurship in Economic Development
11 Since it is not entirely clear what inputs or activities should be counted as R&D and
because not all innovations require any significant R&D, this method of defining an
innovative firm or start-up is somewhat imprecise.
12 Because there are innovative and not so innovative firms in all industries, a classifica-
tion based on industry affiliation leads to only a diffuse picture of innovative and non-
innovative entries. Given the limited availability of data on innovation, however, this is
often the only feasible way to identify such new businesses.
276 Theory: The Role of Entrepreneurship in Economic Development
Another factor that may shape the level of entry as well as its effect
on growth is the market structure and the intensity of competition.
Although economic theory is as yet ambiguous as to the relationship
between market structure and the intensity of competition, it seems
reasonable to assume that easy entry and a high number of competitors
lead to rather strong competition that results in relatively large effects
of new business formation. To what extent certain barriers to entry such
as minimum efficient size and capital requirements influence the effects
of new business formation is a priori unclear. On the one hand, entry
rates will be relatively low in industries with a high minimum efficient
size (Fritsch and Falck, 2007), which should lead to lower intensity of
competition and a less pronounced role of new businesses in industry
development. On the other hand, a high minimum efficient size can
induce larger size entries, which will tend to have a more pronounced
effect than smaller start-ups (Acs and Mueller, 2008). A large minimum
efficient size may also constitute higher sunk costs of exit, which can
motivate incumbents to seek a productive response to the challenge
of entry instead of simply quitting the market. Hence, start-ups in
the manufacturing sector, which is characterized by a relatively large
minimum efficient size, should have a stronger effect on growth than
new businesses in small-scale industries with low barriers to entry and
exit, such as many types of consumer-oriented services.
surprising that a number of studies find that the level of new business
formation tends to be higher in such clusters than in regions where firms
of the same industry are absent or relatively scarce (for an overview,
see Frenken et al., 2011). The available empirical evidence provides
only weak support for the hypothesis that new firms setup in clusters
of the same industry have higher survival rates and grow faster than
new businesses outside such clusters (Wennberg and Lindqvist, 2010;
Frenken et al., 2011). Although spatial proximity between firms of the
same industry may be conducive to the transfer of knowledge, clustered
firms are not found to be generally more innovative than firms located
outside of clusters (Martin and Sunley, 2003).
Relatively wide availability of resources on thick regional input
markets in cities can have two main advantages for new businesses
(Helsley and Strange, 2011). First, wide availability of inputs can lead
to faster project completion that may increase the success of new ven-
tures. Second, thick input markets may enable firms to successfully
manage rather complex projects such as innovations that would be
infeasible in regions with relatively thin markets. Generally, a richly
diverse supply of inputs can compensate for skill deficits in that it allows
entrepreneurs with a limited variety or “balance” of skills (Lazear,
2004, 2005) to hire local personnel with the appropriate qualifications,
thereby substituting market thickness for own skill imbalance. Due to
such agglomeration advantages, one may expect that larger cities have
a comparatively high share of entries that pose a significant challenge to
incumbent firms. However, large cities are also characterized by rather
intensive competition on input markets that — in contrast to com-
petition on output markets — is not necessarily limited to the same
industry. Since many input markets are much more local in character
than output markets, one may expect that the competition effects in
input markets will occur more often in the same region where a new
business is located than the competition effects in output markets.
In summary, there are many regional factors that can influence
the effect of new business formation on development. In particular,
there are a number of reasons why one may expect that the effects
of new business formation on regional development will be relatively
pronounced in large agglomerations characterized by high population
280 Theory: The Role of Entrepreneurship in Economic Development
density. The main factors that should lead to more pronounced effects
of start-ups in larger cities are the following:
All these points strongly suggest that the effects of new business for-
mation should be different across regions.
3.5 Summary and Conclusions 281
Regional environment
Supportive Qualification of
infrastructure the workforce
for start-ups
Industry-specific conditions
Quality of
Thickness Entry and other inputs
of local exit New
input barriers businesses
markets
Regional
Stage of growth
the industry
lifecycle Incumbent
firms
Fig. 3.4 Factors that may determine the effect of new business formation on regional devel-
opment.
3.5 Summary and Conclusions 283
285
286 Results of Empirical Research
for assessing the indirect effects. To account for such indirect effects, the
relationship between level of new business formation and some aggre-
gate performance measure, such as change in employment, gross domes-
tic product, or productivity, in the country, region, or industry, has to
be analyzed. To date, work on the effects of new business formation
on economic development has mainly focused on employment creation,
possibly due to the importance policymakers place on job generation
and the prevention of unemployment. Another reason may be the bet-
ter availability of employment data compared to other performance
indicators.
A meaningful comparison of regions of different size or economic
potential should in one or another way account for this economic poten-
tial. An easy way to fulfill this requirement is to calculate a start-up rate
that relates the number of start-ups to a measure of regional size. Most
commonly, the number of employees, the regional workforce (including
the unemployed), or the regional population of working age is chosen
as the denominator of the start-up rate, what Audretsch and Fritsch
(1994) call the “labor market” approach. This kind of start-up rate is
based on the notion that each member of the workforce is faced with
the decision to either work as an employee in someone else’s business, to
be unemployed, or to start his or her own firm. According to the labor
market approach, the entry rate may be viewed as the propensity of a
member of the regional workforce to start an own business.1 Many of
the analyses of the effect of new business formation on regional devel-
opment use sector-adjusted start-up rates that account for the fact that
start-up rates differ systematically across industries.2 Sector-adjusted
1 Because start-ups are usually located close to the residence of the founder (Dahl and
Sorenson, 2009; Figueiredo et al., 2002; Stam, 2007), the regional workforce can be
regarded as an appropriate measure of the number of potential entrepreneurs.
2 For example, start-up rates are higher in the service sector than in manufacturing indus-
tries. This means that the relative importance of start-ups and incumbents in a region is
confounded by the composition of industries in that region. If this fact is not appropriately
taken into consideration, the result will be an overestimation of the level of entrepreneur-
ship in regions that are home to a large number of industries for which start-ups play an
important role, and an underestimation of the role of new business formation in regions
that are home to a high share of industries characterized by relatively low start-up rates.
To correct for the confounding effect of the regional composition of industries on the
number of start-ups, a shift–share procedure is employed to obtain a sector-adjusted mea-
sure of start-up activity (for details, see Audretsch and Fritsch, 2002, Appendix). This
4.1 Methodological Issues 287
start-up rates often lead to somewhat clearer results and higher lev-
els of determination than do estimates using the non-adjusted start-up
rate, but the basic relationships are usually left unchanged.
The results derived from an analysis of the relationship between
start-up rates and the development of employment or turnover at the
industry level can be very difficult to interpret. The problem is that if
industries follow a life cycle, then the number of entries and the start-up
rate will be relatively high in the early stages of the life cycle when the
industry is growing, and relatively low in later stages when the industry
is in decline (see Section 3.2 for a detailed exposition). Can the result-
ing positive correlation between the start-up rate and development of
the industry in subsequent periods be regarded as an effect of entry on
growth? Probably not—and, indeed, entirely different results are found
if, for example, the relationship between the level of start-ups and sub-
sequent employment change is analyzed at the regional or industry
level (see Fritsch, 1996; Bos and Stam, 2011). One may try to solve
this problem by controlling for industry-specific time trends (see, for
example, Falck, 2007, 2009), but such a control cannot completely rule
out the danger of a misspecification. Another problem when analyzing
the effects of new business formation on the development of industries
is that the impetus for new business formation in one industry may
also have effects on the development of other industries (Andersson
and Noseleit, 2011). These results and considerations strongly suggest
that geographical units of observation are much better suited for ana-
lyzing the effects of new business formation on economic development
than are industries.
Empirical studies analyzing the impact of new business formation
on the development of regions or countries usually employ a start-up
measure that is based on gross entry as indicator of the level of new
3 Moreover, most of the available approaches that analyze the effects of entrepreneurship in
a production function framework are based on cross-section estimates and therefore may
neglect the effect of region-specific factors on growth that are not explicitly accounted for.
Hence, a fixed-effects panel approach that controls for such unobserved regional effects
may lead to quite different estimates of the contribution of entrepreneurship to economic
development.
4 An exemption is Neumark et al. (2006), who account for time lags of up to 10 years.
290 Results of Empirical Research
time lags for new business formation makes it possible to identify indi-
rect effects of entry on incumbent employment (see Section 4.4.2). For a
more reliable identification of the relevant causalities, it would be desir-
able to include regional fixed effects and apply instrumental variable
approaches.
5 Thestatistic is limited to those businesses that have at least one employee, that is, start-
ups consisting only of owners are not included.
4.2 The Direct Contribution of New Businesses to Employment 291
line with work using other kinds of data and for other countries.6 Ana-
lyzing the German Employment Statistics for West Germany, Fritsch
and Weyh (2006) and Schindele and Weyh (2011) show that overall
employment in entry cohorts first rises but then declines from the sec-
ond or third year on (Figure 4.1). After about eight years, it falls below
the initial level and after 20 years the overall number of employees in
a cohort is slightly less than 80 percent of that in the year the new
businesses were setup. After 29 years, the maximum length of the cur-
rently available time series, the overall number of employees is about 50
percent of the initial number of employees. Since most start-up activity
takes place in the service sector, employment development in services
is quite similar to the pattern observed in the private sector as a whole.
The results for manufacturing are somewhat different. The number of
6 See,
for example, Horrell and Litan (2010), Spletzer (2000), and Stangler and Kedrosky
(2010) for the United States, and the cross-country comparison in Bartelsman et al. (2009).
292 Results of Empirical Research
7 See,for example, Storey (1994, pp. 113–119) for a review of the cohort studies that were
available until the early 1990s.
4.2 The Direct Contribution of New Businesses to Employment 293
Fritsch and Schindele (2011) analyze the share that new businesses
contribute to regional employment in West Germany. Their measure
for the direct employment effect is the employment in regional start-up
cohorts two (short-term effect) and 10 years (long-term effect) after
entry divided by total regional employment in the year prior to the
start-up.8 On average, an entry cohort of a particular year adds 1.8
percent to overall employment after two years and 1.56 percent after
10 years. The difference between the long-term and the short-term
contribution reflects the declining employment in start-up cohorts
between the second year and the tenth year after entry. Both the
short-term and long-term contributions to employment are more
pronounced for start-ups in services than for those in manufacturing,
which is obviously a result of the higher level of new businesses
formation in the service sector.
Analyses of the direct contribution start-ups make to employment
clearly show that new businesses do create a considerable number
of jobs, but that the share of employment in new businesses in
overall employment is not overwhelming. Moreover, many new firms
exit the market soon after entry; only a small fraction of start-ups
exhibit considerable growth. Why these few gazelles succeed is still
unclear and deserves further investigation (Henrekson and Johansson,
2010). However, keeping in mind Schumpeter’s examples of innovative
entrepreneurs who initiate radical changes with important indirect
effects on other firms and other sectors of the economy (Schumpeter,
1911/1934, 1939), the development of start-ups tells only part of
the story. Hence, even if the number of jobs provided by gazelles or
the amount of value they add does make up a considerable part of
a region or sector, attempts to assess the impact of new business
formation on development should try to account for indirect effects
also. It is a severe disadvantage of cohort analyses that they do not
account for possible indirect effects of new business formation such as
the displacement of incumbent businesses or productivity enhancing
8 Employmentin the year prior to the start-up was used because this figure is not influenced
by employment in the newly created businesses.
294 Results of Empirical Research
9 Mueller (2007) uses a similar approach to assess the effects of the level of new business
formation on the GDP level in the regions of West Germany. She finds a weakly significant
positive coefficient for start-ups in technology- and knowledge-intensive industries, whereas
the coefficient for new business formation in the other industries is nonsignificant.
4.3 The Overall Effect of New Business Formation on Employment 295
10 Respondents were classified as “ambitious” entrepreneurs if they stated that they expect
their firm to grow to at least six employees within five years.
11 A venture was classified as having a “high-growth potential” if it fulfilled four criteria:
(1) the venture plans to employ at least 20 employees in five years; (2) the venture
indicates at least some market creation impact; (3) at least 15 percent of the venture’s
customers normally live abroad; and (4) the technologies employed by the venture were
not widely available more than a year ago (Wong et al., 2005, p. 345).
12 Necessity entrepreneurship is understood as a start-up that is founded due to a lack
of alternatives (for example, the founder cannot find any other kind of job). A new
business setup to pursue an opportunity is classified as “opportunity” entrepreneurship.
See Reynolds et al. (2005) for details.
296 Results of Empirical Research
entry into the market. Due to data restrictions, the analyses mentioned
above do not include any or only rather short time lags between the
founding of the start-ups and the respective effect on output. However,
if longer time lags play a role, these studies may have assessed the
effects on regional development only incompletely. In an analysis for
West German regions, Audretsch and Fritsch (2002) do, indeed, find
evidence for positive long-term effects of new business formation. In
this study, new business formation activity in the early 1980s could not
explain regional employment change in the rest of the decade, but did
provide an explanation of employment change in the 1990s.
Van Stel and Storey (2004) analyze the relevance of such time lags
more systematically and estimate a time lag structure of the effects
of new business formation on regional employment growth with data
for Great Britain. They confirm that there are considerable time lags
between new business formation and its effect on regional development,
which they find to be positive. A severe problem in analyzing the time
lag structure of the effect of new business formation on growth emerges
from a high correlation between yearly start-up rates. Because of this
high correlation, the original estimates may not reflect the “true” lag
structure. In dealing with this problem, van Stel and Storey (2004)
apply the Almon polynomial lag procedure. This procedure attempts
to approximate the lag structure by means of a polynomial function
(for a detailed description of this method, see Greene, 2008). In this
type of analysis, an assumption must be made about the order of the
polynomial to be used for estimating the lag structure. Fritsch and
Mueller (2004) apply the Almon polynomial lag procedure in an anal-
ysis of the effect of new business formation on regional development in
West Germany. They find that a statistically significant effect of new
business formation on employment is restricted to a period of about
10 years. Van Stel and Storey (2004) assume a second-order polyno-
mial for estimating the lag structure of new business formation rates;
Fritsch (2004), however, also apply higher-order polynomials. With a
third- and higher-order polynomial, they find a “wave” pattern of the
effects (see Figure 4.2).
Figure 4.2 depicts the original regression coefficients that have been
found without application of the Almon lag procedure as well as the
4.3 The Overall Effect of New Business Formation on Employment 297
Fig. 4.2 Effects of new business formation on employment change over time in West Ger-
many — regression coefficients for start-up rates and the results of the Almon lag procedure
assuming a third-order polynomial
(Source: Fritsch, 2008).
the United States (Acs and Mueller, 2008), as well as for a sample of
23 OECD countries (Carree and Thurik, 2008).14
Fritsch and Mueller (2004) suggest an interpretation of this wave
pattern that builds on the systematization of direct and indirect effects,
as discussed in Section 3.4.1. According to this interpretation, the posi-
tive employment impact for start-ups in the current year can be under-
stood as the additional jobs created by the newly founded businesses
at the time of inception. This direct employment effect is indicated
in Area I in Figure 4.2. It is well known from a number of analyses
that employment in entry cohorts tends to be stagnant or even decline
from the second or the third year onward (see Section 4.2). Therefore,
new firm formation in year t − 3 and more distant time periods should
not lead to any significant direct employment effect of the cohort as
a whole. As soon as a new business is set up, it is subject to market
selection and may gain market shares from incumbent suppliers. Thus,
the negative impact of the start-ups in years t − 1 to t − 5 (Area II in
Figure 4.2) is probably a result of market exit, that is, new businesses
that fail to be competitive and displacement of incumbents. The pos-
itive impact of new business formation on employment for years t − 6
to t − 10 (Area III in Figure 4.2) is probably due to a dominance of
indirect supply-side effects, that is, increased competitiveness of the
regional suppliers resulting from market selection. After about nine or
ten years, the impact of new business formation on regional employ-
ment fades away.
When they use a second-order polynomial for the Almon lag pro-
cedure, Fritsch and Mueller (2004) find the resulting lag structure to
be “u”-shaped. The interpretation of the “u”-shaped lag structure is
similar to that for the wave pattern that resulted from assuming a
higher-order polynomial. Fritsch and Mueller (2004) posit that the ini-
tial increase in employment is a result of the new businesses’ initial
14 Insome cases, the curve for the lag structure that is estimated by applying the Almond
polynomial lag procedure shows negative coefficients for the last one or two years of the
period under inspection. These negative coefficients in later periods should be regarded
as an artifact of the estimation procedure. It is difficult, if not impossible, to plausibly
argue for negative employment effects of new business formation after the supply-side
effects have phased out.
4.4 Indirect Effects of New Business Formation 299
15 Inthis analysis, new businesses are those that have been set up during the previous 10
years.
302 Results of Empirical Research
3.0
2.5
Employment change (in %) 2.0
1.5
1.0
0.5
0.0
-0.5 0 1 2 3 4 5 6 7 8 9 10
-1.0
-1.5
Years after start-up
Direct effect Indirect effect Overall effect
Fig. 4.3 Impact of start-ups on regional employment change — direct and indirect effects
(Source: Fritsch and Noseleit, 2012a).
16 The period of 10 years was chosen here because the analyses by Fritsch and Mueller (2004,
2008) for Germany find significant effects of new business formation for this length of time
(see Section 4.2).
4.4 Indirect Effects of New Business Formation 303
17 This result corresponds quite well with the estimates of Fritsch and Mueller (2008).
According to Fritsch and Mueller (2008), one additional start-up per 1,000 employees
leads to an overall employment increase of about 0.46 percent in the average region.
Given an average start-up rate of about nine new businesses per 1,000 employees, an
employment increase of 4.14 percent can be attributed to new business formation.
304 Results of Empirical Research
Table 4.1. The impact of new business formation on total employment change, employment
change in incumbent businesses, and employment change in new and young businesses.
regions. The study reveals that there are indeed pronounced indirect
effects that were the strongest for start-ups in manufacturing, followed
by start-ups in low-end services, with high-end service industries bring-
ing up the rear.
18 I am not aware of any study that relates regional new business formation to other indi-
cators of innovation output.
19 Mueller (2006) identifies a highly significant positive effect of the start-up rate on the
lags. They identify a wave pattern as has been found in many anal-
yses of the effect of entry on employment change (see Section 4.3).
There are some negative coefficients for first years after net entry but
these coefficients are not statistically significant. The overall effect of
net entry on productivity growth measured over a period of ten years
is clearly positive. Andersson et al. (2012), investigate the effect of
new business formation and turbulence (entry plus exit) on the pro-
ductivity of incumbent firms in Swedish regions accounting for time
lags of up to four years. In the various estimations they frequently find
significantly negative effects of entry and turbulence on incumbent’s
productivity for the current year and for the first two years after entry.
After these first years the effect becomes strongly positive in most cases.
Andersson et al. (2012) suggest that the negative effect of entry and
turbulence in the first years may be due to the effort of “adjusting rou-
tines and strategies etc. to respond to the newcomers on the regional
market.”
Bosma (2011), using GEM data for European regions, finds a
positive relationship between the level of early-stage entrepreneurial
activity and labor productivity. It is remarkable that this relation-
ship is weaker for early-stage entrepreneurial activity with low-growth
ambitions than for other types of entrepreneurship. The relationship is
particularly strong for early entrepreneurial activity with high-growth
ambitions and for those entrepreneurs with innovative ambition. Bosma
et al. (2011) analyze the effect of entry and turbulence (defined as the
number of entry plus the number of exits) on the change of total factor
productivity in the 40 NUTS III regions of the Netherlands. Turbu-
lence is defined as the number of entries plus the number of exits and
is intended to measure the level of business dynamics understood as
“creative destruction.” The number of entries was lagged by two years;
the number of exits was taken from the current year. They find a posi-
tive effect for both indicators in the service sector but not for start-ups
and turbulence in manufacturing. According to their estimates, the
marginal effect of entry and turbulence in the service sector on the
growth of total factor productivity decreases with a rising level of busi-
ness dynamics. They identify a certain rate of turbulence at which the
effect on productivity growth is at a maximum.
4.5 The Quality of Entry 307
of the founder’s entrepreneurial skill and the firm’s knowledge base and
innovativeness should be important for its effect. One fact that may
indicate certain aspects of the quality of a new business is its industry
affiliation. There are many classifications, including “high-tech man-
ufacturing,” “technologically advanced manufacturing,” “knowledge-
intensive services,” and “non-technical consulting,” that can be used
to distinguish the quality of start-ups. Other ways of distinguishing
between the quality of start-ups is to use information about their orga-
nizational status (headquarter vs. subsidiary location), the qualification
of the entrepreneur, and the personnel or characteristics based on self-
estimations that are available in the GEM data, such as “necessity,”
“opportunity,” or “ambitious” entrepreneurship.
Section 4.5.1 provides an overview of the results of analyses of start-
ups in different industries. Section 4.5.2 reports results for other types
of classifications.
20 The number of organizational foundings in Koster’s data is about half the number of
independent start-ups (Koster, 2011).
4.5 The Quality of Entry 313
22 Such a development is well illustrated by Neumark et al. (2008), who analyze the ini-
tial effect of new stores set up by a large supermarket chain (Wal-Mart) on local retail
employment in the United States. According to their estimates, the new competition has
on average led to a 2.7 percent reduction of average retail employment and a decline of
retail earnings of 1.5 percent in the short run.
4.6 The Effects of New Business Formation on Economic Development 315
0.40
Agglomerated
regions
Moderately
Impact of new business formation
0.00
Rural
regions
-0.20
-0.40
0 1 2 3 4 5 6 7 8 9 10
Lag (year)
Fig. 4.4 Impact of new business formation on regional employment change in agglomera-
tions, moderately congested regions, and rural regions (Fritsch and Mueller, 2008).
4.6 The Effects of New Business Formation on Economic Development 317
23 However, the results for rural regions should be viewed with caution because only two of
the 11 coefficients for start-up rates in the unrestricted model proved to be statistically
significant and the coefficients for the Almon lags remained insignificant.
318 Results of Empirical Research
4.0
3.0
Employment change (in %)
2.0
1.0
0.0
-1.0 0 1 2 3 4 5 6 7 8 9 10
-2.0
-3.0
-4.0
Years after start -up
Agglomerations Moderatly congested areas Rural regions
Fig. 4.5 Average effects of new business formation on incumbent employment in different
types of regions
(Source: Fritsch and Noseleit, 2012a).
26 This holds, for example, for large sections of the old industrialized Ruhr area in Germany.
326 Results of Empirical Research
Employment
change
Growth is result of
Growth is result of
stable enterprise
start-ups and highly
structure and
turbulent enterprise
development of
structure
incumbents
Routinized Entrepreneurial Start-up
rate
Downsizing Revolving door
Effects of low level of
High start-up rates but
new business format-
entries tend to be of
ion do not compensate
low quality and hardly
for losses of
induce any growth
incumbents
27 See, for example, Audretsch and Keilbach (2004), Beugelsdijk (2007), Beugelsdijk and
Noorderhaven (2004), Beugelsdijk and Maseland (2011), Chinitz (1961), Davidsson
(1995), Etzioni (1987), Freytag and Thurik (2007), Glaeser and Kerr (2009), Glaeser
et al. (2012), and Malecki (1994).
28 Beugelsdijk (2007) and Beugelsdijk and Noorderhaven (2004) have constructed an indica-
tor for regional entrepreneurial culture based on data from the European Values Survey
(EVS) and relate this indicator to measures for regional patenting activity. Comparing
the regional values of the indicator for an entrepreneurial culture with measures for the
level of regional patenting activity shows a pronounced positive relationship.
4.7 New Business Formation a Cause or a Symptom of Regional Development? 329
previous growth may generate a relatively large number of new entrepreneurial oppor-
tunities. Second, positive expectations about future growth can encourage individuals to
start an own business. Third, overall growth makes it easier for start-ups to survive their
first critical years and to establish themselves in the market.
330 Results of Empirical Research
31 See, for example, Audretsch and Keilbach (2007), Fritsch and Falck (2007), Reynolds
et al. (1994), and Sutaria and Hicks (2004).
4.7 New Business Formation a Cause or a Symptom of Regional Development? 331
1994 and 2000. The authors find that during the period of analysis,
the number of businesses in Ireland has almost tripled. However, the
number of start-ups in relation to the number of incumbent businesses
remained fairly constant in the longer run. The same holds for the num-
ber of new businesses in relation to the number of employees. Relat-
ing the time series of new business formation and employment showed
no statistically significant effect of employment growth on the level of
start-ups, but did show a weak effect of gross entry on employment.
These statistical tests, however, were restricted by the limited length
of the two time series. Anyadike-Danes et al. (2011) find that the stock
of businesses per regional population is fairly constant across regions,
supporting the idea of an equilibrium number of businesses per popu-
lation at a certain point in time (see Section 3.2). The authors suggest
that relatively high start-up rates in a region might be regarded as a
process of catching-up to this equilibrium rate.
Hartog et al. (2010) investigates the possible two-way relationship
between changes in the business ownership rate (= net entry) and
growth for 21 OECD countries for the period 1981–2006, employing a
simultaneous empirical approach. The authors identify a link between
the national welfare level and the business ownership rate, but find that
development during the previous periods has no statistically significant
effect. Analyzing the effect of changes in the business ownership rate on
GDP growth, Hartog et al. (2010) conclude that there are decreasing
marginal returns in terms of growth effects to entrepreneurship, which
confirms results of Fritsch and Schroeter (2011a) for German regions.
Hartog et al. (2010) explain this result with the notion of an equilib-
rium business ownership rate: an increasing level of entrepreneurship
will have relatively pronounced effects on growth if the initial business
ownership rate is below the equilibrium rate; the effects will be consid-
erably smaller if the initial rate is above the equilibrium rate. A main
limitation of Hartog et al.’s (2010) study is that it contains no informa-
tion on gross entry and thus nothing can be learned about the effects
of the number of entries on turbulence in the stock of businesses and
its effects on economic development.
In summary, work to date has not identified any, or only a rela-
tively weak, effect of growth in previous periods on the level of new
332 Results of Empirical Research
334
335
337
338 Avenues for Further Research
This catalog of open questions makes clear that much more needs to
be learned about how different types of new businesses impact economic
development in different regional environments.
7
Final Remarks
344
345
346
References
347
348 References