Cross-National Comparisons of the Variation in New Firm Formation Rates

Abstract

REYNOLDS P. D., STOREY D. J. and WESTHEAD P. (1994) Cross-national comparisons of the variation in new firm formation rates, Reg. Studies 28, 443–456. Births of new firms have been seen to be a major source of new employment creation in developed countries. In the European Community (EC), government policy for much of the 1980s focused upon encouraging the startup of new enterprises. This research examines birth rates of new firms by region in France, Germany (West), Italy, Ireland and the United Kingdom. Comparison is also made with similar studies which have been conducted in parallel for Sweden and the United States. The key findings of the study are twofold. First, the average new firm birth rates are roughly similar across countries and there are similar regional variations within all countries; the most fertile regions have annual new firm birth rates that are two to four times higher than the least fertile regions. Second, the underlying processes affecting new firm births at the regional level appear uniform across countries. Consequently, we suggest that the main empirical findings are likely to apply to populations of new firms in other advanced market economies, including those in the EC.

Full text

Regional
Studies,
Vol.
28.4,
pp.
345346.
Cross-national Comparisons
of
the
Variation in
New
Firm Formation Rates:
An Editorial Overview
GUEST
EDITORS:
PAUL
REYNOLDS*,
DAVID
J.
STOREYt
and
PAUL
WESTHEADt
*College
of
Business Administration, Marquette University, Milwaukee,
WI
53233,
USA
tCentre
for
Small and Medium Sized Enterprises, University
of
Warwick, Coventry CV4
7AL,
UK
No
feature of
a
region is more critical than
its
economic
well-being. Maintaining the prosperity of the citizens may
require continuous economic change and adaptation. Cent-
ral
to
such shifts is the birth of new firms-entrepreneur-
shp. Substantial evidence suggests that the implementation
of new firms has
a
central role in regional economic
advances
(STOREY,
1984).
But what processes lead
to
the birth of new firms? What
is
the relative effect of different ‘birth processes’ on the
implementation of new firms? Further, are these processes
universal? That is, can they be observed to have an impact
in
all
advanced market economies? Finally, does an
enhanced understanding of these mechanisms have
implications for government policy designed to enhance
regional-and national-economic growth? These ques-
tions form the basis for the cross-national project reviewed
in this special issue. The project involved comparisons of
standardized analyses completed for seven economically
advanced countries: France; Germany (West); Ireland; Italy;
Sweden; United Kingdom; and the United States.
The introduction will discuss the perspectives and strat-
egy for this approach, including
a
review of the major
assumptions and associated advantages. There were, of
course, disadvantages, and these
are
given some attention.
A
review of the organization and history of the project
is
provided, followed by
a
summary
of
the conclusions.
FIRM BIRTH PROCESSES
AND SPATIAL VARIATION
IN
NEW
FIRM BIRTHS
The research strategy for this collaborative work
assumes, first, that some type
of
causal mechanism
underlies the birth
of
a
new firm. That is, if
a
new
business entity is identified in
a
relatively compre-
hensive national database, something happened to
put it there. Second, if there is variation among the
regions within a given country in the annual birth
rates
of
new firms,
this may reflect the greater
importance of some firm birth processes, the pres-
ence
of
more processes creating firm births,
or
both
in the high firm birth regions -when compared with
low firm birth regions. Third, if it is possible to
identify characteristics
of
these regions that vary in
periods prior to those in which firm births are meas-
ured and there is
a
systematic relationship to vari-
ation among regions in the birth
of
new firms, then
these characteristics may reflect the presence
of
one
or more ‘firm birth processes’.
For example,
if
regions with higher per capita
income in one time period have higher firm births in
later time periods, then hgher per capita income
is
assumed to reflect some process-such as higher con-
sumer demand-that
is
leading to more firm births.
As
multiple firm birth processes are anticipated, mul-
tiple regional characteristics are included in the
efforts to account for variation in firm birth rates.
The major analytical technique to explore the pro-
cesses affecting firm births, then, is the development
of
linear models that will account for regional vari-
ation in new firm birth rates. Standardized data on
regional characteristics are used in linear regression
models to account for variation in firm birth rates
across regions within the same country. Those
regional characteristics found to be more important
in accounting
for
variation in firm birth rates are con-
sidered to reflect firm birth processes that are
of
some
importance.
So
while the focus
of
the linear models is on
explaining regional variation in firm birth rates, the
major objective
of
the research programme is to
identify the most important firm birth processes. It
is,
of
course, not possible
to
provide a detailed speci-
fication for a firm birth process based
on
aggregate
regional data. However, if consistent patterns are
found among regional characteristics and measures
of
new firm births, there is some guidance for more
precise studies of the birth processes, perhaps
through systematic studies
of
the entrepreneurs and
new firms involved.
Such a study can be carried out withm a single
country when several conditions are met. First, reli-
343

344
Paul Reynolds, David
J.
Storey
and Paul Westhead
able data are available for the important variables:
new firm births as well as various other contextual
features, e.
g.
unemployment, household wealth, etc.
Second, comparable data
on
independent variables
are available for individual regions within the coun-
try. Third, data are available for measures of firm
births for periods following the periods for which
the independent variables are measured. Such
a
time
lag increases confidence in
a
causal relationship.
It
turns out such information is available for
a
number of advanced economies, albeit only recently.
Very few countries have useful data on firm births
much before
1980.
Of course, if there were no
regional variations in new firm births, this strategy
could not be used to study firm birth processes.
As
there are such regional variations within most coun-
tries, this does not turn out
to
be a major problem.
CROSS-NATIONAL COMPARISONS
Progress in science usually starts with a single study
indicating empirical regularities in
a
specific time and
place. This is generally followed by additional studies
that may, if all goes well, confirm the same patterns
in other settings and
at
other times. Confidence in
the generality of
a
pattern grows with increases in the
diversity of settings and times in which the pattern
is observed. The capacity to expand the analysis of
regional factors affecting firm birth rates across time
is hampered by the lack of data on firm births prior
to the late
1970s.
However, as
a
number of countries
have recently developed reliable counts of new firm
births, it is possible to explore the relationship of
regional factors affecting new firm births across dif-
ferent countries.
This cross-national approach has both advantages
and disadvantages. The major advantage, of course,
is information about the general nature of the empir-
ical regularities related to firm births and, by infer-
ence, the general presence of firm birth processes.
If the same patterns are found among
a
number of
countries, then confidence in the universal nature of
these processes is substantially enhanced. The disad-
vantage lies in the diversity among the countries with
regard to the major features of the analysis:
(1)
definition and measurement of
a
new business entity;
(2)
rationale and procedures used to define
a
sub-
national geographic region;
(3)
definition and meas-
urement of major regional features, the independent
variables;
(4)
variation in the time lag between meas-
urement of regional features and firm births; and
(5)
the absence of some variables, dependent or inde-
pendent, in the data available for some countries.
Comparability among measures and procedures,
both conceptually and in operational terms, is always
desirable. Such comparability is critical if the phe-
nomena under study are subtle and affected by
a
diversity of complex processes. When such volatile
phenomena are given careful study, sensitivity to
diversity in context and procedures is reflected in
instability in the results or dramatically different
results
-
perhaps even reversals -in different
settings.
On
the other hand, if the phenomena are robust
and the processes potent, then the same patterns may
be observed even when diversity in context and
measurement procedures are present. Indeed, it was
clear-as mentioned above- that virtually every
aspect of the analysis would vary for the cross-
national comparison. To provide partial compensa-
tion for this variation and enhance the potential for
cross-national comparisons, two procedures were
employed to increase standardization. First, the same
conceptual framework was used for
all
analyses-
there was harmonization of the abstract models. This
meant that each country team made an attempt to
incorporate indicators of the same set of regional fac-
tors, even if the measures were different. Second, the
same analysis procedure was employed in all of the
studies-ordinary least squares regression analysis
with forced entry of
all
independent variables. This
facilitated the most basic of comparisons: determin-
ing which explanatory variables -and related firm
birth processes
-
made
a
statistically significant con-
tribution to explaining variation in firm births in the
different linear models (this ‘maximum comparabil-
ity approach’ is presented in the last paper in this
volume; the papers on each country reflect each
teams ‘optimal effort’).
Proceeding in spite of the lack of comparability
could lead
to
two
outcomes. If it turned out that
there was no consistency among the countries
involved and the relationships between regional fac-
tors and firm births were different for each country,
two conclusions would be justified. First, that the
startup processes were affected by
a
wide range of
subtle factors that were unique for each country.
Second, that precise standardization among the
measures would be required for useful cross-national
comprisons.
If, on the other hand, there turned out to be sub-
stantial cross-national similarity despite the diversity
in measures, then it would suggest that the startup
processes were robust and the same processes might
be expected in countries not included in the analysis.
This would be expected even if different procedures
were used to measure the same regional features or
firm births.
The latter was the case. There was considerable
success in the linear models developed
to
account
for new firm births and substantial similarity in the
models across the different countries. The major
problems that developed were related
to
the absence
of measures for some regional characteristics. There
appeared to be few problems associated with divers-
ity of measures of the same variables. Hence, it seems

Cross-national Comparisons
of
the Variation in New
Firm
Formation Rates
345
appropriate to conclude that the processes leading to
firm births are quite robust and are reflected through
a
diversity of measures or indicators.
The major results of this research programme sug-
gest
-at
least to the programme co-ordinators- that
the processes leading to firm births are broadly sim-
ilar across countries with advanced market econom-
ies. The reader is, of course, able
to
develop their
own conclusions after
a
review of the papers that
follow
.
BACKGROUND
OF
THE
PROJECT
This research programme reflects several stages.
Some of the initial work was completed on the
United States (REYNOLDS and
MAKI,
1990, 1991).
This led to
a
pilot project, supported by the Euro-
pean Commission (EC), completed for European
countries (KEEBLE
el
al.,
1990).
This pilot study sug-
gested that data were available in
a
number of EC
countries and Sweden that would facilitate the ana-
lysis.
A
second round of EC funding followed and
funds were provided by Swedish government
sources to maintain their role. Several meetings of
the country teams were held
to
consider how to pro-
ceed with
a
cross-national analysis. All country teams
eventually agreed to produce a standardized analysis,
which was presented
at
an Organization for Eco-
nomic Co-operation and Development (OECD)
workshop (December
1992)
with
a
subsequent report
(REYNOLDS and STOREY,
1993).
The comparisons
presented in the concluding paper are based
on
the
standardized analysis developed for the OECD
workshop.
But each team had completed, simultaneously,
additional analysis that varied from the ‘common
standard’. The articles on each country in this
volume of
Regional Studies
reflects each country
team’s judgement about the most appropriate ana-
lysis.
As
they are most familiar with the data for their
country as well as distinctive regional and national
features, this provides the reader with two versions
of each country analysis. The standardized analysis
is in the concluding paper in this volume and the
special analysis for each country is provided in the
individual papers.
The analysis for France (Guesnier) is organized
around changes in the perceptions of the major policy
making groups, as they discovered that an unques-
tioned support of Fordism (large national firms) may
have been responsible for
a
comparative lag in French
job creation in the late
1970s.
Regional differences in
firm births have accompanied
a
shift in the national
emphasis toward
a
support for entrepreneurship in
the
1980s
as
small and medium sized enterprises
began to generate the majority of new jobs, particu-
larly in Ile-de-France (around Paris) and the southern
regions.
The (West) German analysis by AUDRETSCH and
FRITSCH,
1994,
on
factors affecting firm births, finds
results similar to that of other studies. More small
establishments, greater unemployment,
a
more
skilled workforce, and greater population density
seems to enhance firm birth rates. However, in
a
separate analysis, AUDRETSCH and FRITSCH,
1993,
provide
a
more extended analysis of the relationship
between turbulence and regional economic growth.
They find that the most stable regions seem to have
higher subsequent economic growth, while more
firm turbulence (high firm births
as
well as high firm
deaths) appear to be followed by regional economic
decline. Such patterns have not been found in the
United States (REYNOLDS and
MAKI,
1992)
or in
Sweden, as reported in this issue. Clearly, some addi-
tional work is required to determine the basis for this
major difference between countries.
The analysis of Ireland (Hart and Gudgin) is lim-
ited to manufacturing,
as
that is the only sector for
which accurate counts of firm births are available.
The present discussion confirms an earlier analysis
(O’FARRELL and CROUCHLEY,
1984)
that indicates
that the proportion of small firms and urban dwellers
leads to more firm births and indicates that access
to higher education, demand growth, proportion of
professional and managerial workers, and govern-
ment assistance also has
a
significant effect.
Italy (Garofoli) appears to have had a very high
firm birth rate, and local specialization seems to pro-
mote new firm births, particularly in manufacturing.
The level of local demand does not seem to be
a
significant factor and increases in unemployment
appear to reduce firm births. The situation in Italy
may have some distinctive features.
Sweden (Davidsson, Lindmark and Olofsson) has
the advantage of
a
relatively complete and detailed
data set, which allowed identification of independent
firm births (referred to
as
‘simples’). They have also
had the benefit of measures of economic change
at
the regional level, which permitted attention to the
impact of firm births on subsequent economic
growth. Higher levels of firm dynamics (firm births
as well as firm deaths) appeared to have
a
positive
impact on subsequent economic growth. This differ-
ence from the German results clearly justifies more
comparative analysis. Structural characteristics of the
labour market areas used in the Swedish analysis
explain much of the variation in simple firm births.
The analysis also indicated that the relative import-
ance of the birth processes, as reflected
in
the inde-
pendent variables, varied for firm births in different
economic sectors.
The average annual births for the United Kingdom
(Keeble and Walker) for
1980-90
appear to be affec-
ted by population growth, housing wealth, profes-
sional expertise, presence
of
small firms, urbaniza-
tion and demand growth. There is some evidence of

346
Paul
Reynolds, David]. Storey and
Paul
Westhead
a
local enterprise culture and an impact from govern-
ment programmes and actions. Greater presence of
unemployment does not appear to increase firm
births. The study finds interesting differences
between manufacturing and producer services, while
growth in small business numbers strongly reflect
population growth and housing wealth. Linear
models predicting firm births and deaths were quite
similar.
Analysis of the United States (Reynolds) was
organized into
382
labour market areas and it was
possible to focus on the annual birth rates of auto-
nomous (independent) firms. Indicators of demand
growth had a major impact in all predictive models.
Statistically significant evidence of an impact of urb-
anization, unemployment, personal/household
wealth and small firm presence/economic specializa-
tion was present, but tended to vary for different
economic sectors and type of labour market area.
There appears to be a small, independent, positive
effect of manufacturing birth rates on subsequent
regional economic growth. This suggested that new
firm births were
a
necessary but not sufficient condi-
tion for creating regional economic growth.
The final paper in this volume, then, provides
a
standardized summary that complements the ana-
lyses of each of these seven countries. This begins
with a review of the firm birth processes selected
for attention, the general nature of the results, and
a
discussion of the more general implications for public
policy.
As
no country team was able to develop data
on the character of the new firms-whether they
were likely to export
or
provide ‘good’ jobs or have
a high growth trajectory-many policy issues could
not be addressed by these findings. However, this
research programme suggests that regional character-
istics are
a
major factor affecting variation in firm
births. The ability of governments to affect regional
characteristics is an open question.
There is still much unknown about entrepreneur-
ship and regional growth, but the research
pro-
gramme described in this volume does help guide the
way. Future research may focus on the critical steps
and transitions of the entrepreneurial process-firm
conception, firm gestation, firm birth and early firm
development (REYNOLDS,
1994).
The impact of
regional factors probably varies for each transition;
policy implications may also vary for each transition.
REFERENCES
AUDRETSCH
D.
B. and FRITSCH M. (1993)
Births, Deaths and Turbulence and the Link with Economic Growth in the Federal
AUDRETSCH
1).
B. and FRITSCH M. (1994) The geography of firm births in Germany,
Reg. Studies
28,
359-365.
KEEBLE
D.,
POTTER
J.
and STOREY
D.
J.
(1990)
Cross
National Comparisons
of
the
Role
of
SMEs
in Regional Economic
Growth in the European Community.
Report to the European Commission, Small Business Research Centre, University
of Cambridge, and Small Medium Enterprise Centre, University
of
Warwick.
O’FARRELL P.
N.
and
CROUCHLEY
R. (1984) An industrial and spatial analysis of new firm formation in Ireland,
Reg,
Studies
18,
221-36.
REYNOLDS P. and STOREY
D.
(1993)
Local
and Regional Characteristics A8ecting Small Business Formation:
A
Cross-national
Comparison.
Report submitted to the Organization for Economic Cooperation and Development (OECD), Paris.
REYNOLDS
P.
and MAKI
W.
(1990)
Business Volatility and Economic Growth.
Final project report submitted
to
the Office of
Advocacy, Contract SBA 3067-OA-88,
US
Small Business Administration.
REY
NOI
r)s
P. and MAKI
W.
(1991)
Regional Characteristics Affecting Business Growth: Assessing Stratexiesfor Promoting Regional
Economic Well-being.
Project report submitted
to
the Rural Poverty and Resource Programme. Grant 900-013, the Ford
Foundation.
REYNOLDS P.
D.
(1994, forthcoming) The role of entrepreneurship in economic systems: developed market and
post-socialist economies, in BREZINSKI
H.
and
FRITSCH
M. (Eds)
Economic
Impact
ofNew
Firms
in Post-Socialist Countries:
Bottom-Up Transformation
in
Eastern
Europe.
Edward Elgar, Aldershot, Hants.
Republic
of
Germany.
Final Report to
DG
XXIII,
European Community.
STOREY
D.
J.
(1984) Small firms in regional economic development,
Reg. Studies
18,
197-275.