New Firms and Regional Employment Change

ABSTRACT. This article analyzes the relationship betweenregional start-up activities and employment change in subse-quent years for the West German planning regions. I find thatthe number of start-ups in a particular region is determinedto a high degree by the prevalent sectoral structure. Toeliminate this influence sectorially adjusted new-firm forma-tion rates are calculated. These rates reflect a clear core-to-periphery decline. The relationship between the level ofregional start-up activities and employment change comes outto be rather weak. In contrast to the common wisdom thecoefficients for the impact of the entry rate on regionalemployment indicates a negative contribution of regional start-up acitivities.

1. Introduction1

Setting up a new business establishment impliestaking up new activities and it seems obvious thatnew establishments lead to an increase of jobsupply in a region. Therefore, it is a widely sharedview that stimulating the founding of new firmswill have a considerable positive impact onregional employment. However, there have beenremarkably few attempts to empirically investigatethe contribution that new firms make to regionalemployment so that we still know very little aboutthe role of start-up activities in the job-generationprocess.

This paper analyzes the contribution of newestablishments to regional employment change inWest Germany. Section 2 summarizes the mainhypotheses and the empirical evidence to be foundin the literature. A short description of the database (Section 3) is followed by a discussion of

alternative ways to form an entry rate (Section 4).Section 5 provides a comparison of these alterna-tives. Based on a short overview on regional dif-ferences in new-firm formation in West Germanyduring the period 1986–89 (Section 6), the rela-tionship between new-firm formation and regionalemployment change in subsequent periods isanalyzed in Section 7.

2. Hypotheses and empirical evidence

The contribution new firms make to employmentin a certain region is quite unclear. In particular,we know only very little about the

relative impor-tance of new firm startup activities compared tothe performance of already established units andthe impact of firm exits. A number of argumentscast doubt on the common belief that new firmsplay a major role in the process of regionaldevelopment:

First: Many new firms have to leave the marketafter only a short time so that their contributionto regional job supply is rather temporary (Brüderland Schüssler, 1990; Brüderl, Preissendörfer andZiegler, 1992).2

Second: Most new firms start quite small and anumber of empirical studies show that only veryfew of the new firms generate substantial numbersof jobs (cf. Fritsch, 1990; König, 1994; Storey andJohnson, 1987).

Third: The impact of new establishments onregional employment depends critically on theconsequences for the incumbents. If successfulentry of newcomers forces incumbent firms toleave the market, or if entries are mainly inducedby established firms assigning specific activitiesor jobs to other businesses (cf. Storey and

New Firms and RegionalEmployment Change Michael Fritsch

Small Business Economics 9: 437–448, 1997. 1997 Kluwer Academic Publishers. Printed in the Netherlands.

Final version accepted on February 14, 1996

Technical University Bergakademie FreibergFaculty of Economics and Business AdministrationGustav-Zeuner-Strasse 8D-09596 FreibergGermany

Johnson, 1987), the effect of startups on regionaljob supply will be relatively limited. Moreover,to the extent that entries are associated withchanges in the division of labor between estab-lishments, this might result in rationalizationeffects and cause a reduction of employment.However, if such reorganization leads to anincrease in regional competitiveness one shouldexpect a positive impact on regional growth andemployment, at least in the long run.

Fourth: It may be of central importance for theimpact of a new establishments on regionalemployment:

– to what extent the newcomers receive theirinputs from inside or from outside that region,

– to what degree the new establishments supplyinputs to the regional economy that have beenbought from outside beforehand, and

– whether the successful entry influences theregional economy in some other way, e.g. byreinforcing agglomeration economies or disec-onomies.3

The first two of these objections aim at the direct-employment effects of new firms, i.e. new jobsthat are created in the new establishments. Theother two arguments concern influences that aremore indirect. All of them suggest that thecontribution of entries to regional employment ishighly dependent on the characteristics of the newfirms.4

Empirical analyses of the contribution of newfirms to employment are rather rare. A possiblereason for this lack in research may be the lackof appropriate data. Information on the number ofnew firms by region is available only in a fewcountries and often only from very recent sources(cf. Reynolds, Storey and Westhead, 1994). Asevere shortcoming of nearly all such analyses isthat they measure only the direct impact of newfirms on employment and completely neglect theindirect effects. Those studies that try to estimatethe direct-employment effects of new firms arriveat diverging results, depending on the region(s)and sector(s) studied and on the period of timeunder inspection. While, for example, Birch(1979) in his highly controversial analysis founda strong positive relationship between the numberof jobs created by new firms and overall regional

employment change in the U.S.A., other studies(cf. Boeri and Cramer, 1992; Dunne, Roberts andSamuelson, 1989; Gerlach and Wagner, 1993;1994; König, 1994; Hart and Hanvey, 1995)indicate that the performance of the survivingincumbent firms often has a higher impact onregional employment than the number of jobscreated by the new firms. In many cases thenumber of firms that exit the market in a certainperiod is about as large as the number of entriesso that the net change in the population of firmsis relatively small when compared to the grossflows.5 Correspondingly, the positive direct-employment effects generated by the new firmsare frequently more or less compensated by thenegative effects due to the exit of firms.

Profound analyses of the relationship betweennew-firm formation and regional employment thataccount for the indirect effects have been carriedout only recently and reflect diverging results.Reynolds and Maki (cf. Reynolds and Maki, 1990;Reynolds, 1994) for the U.S.A. arrive at theconclusion that there was a definite positiveimpact of the regional birth rate (number of newfirms divided by the number of incumbent firms)on employment change in subsequent years.However, the intensity of this relationship variedconsiderably between time periods. Ashcroft andLove (1994) analyzed the impact of new-firmformation on regional employment change inGreat Britain in the 1980s. They estimated thaton average each case of gross-entry induced a netemployment growth of about 4.1 jobs within thefirst four and a half years after the firm wasfounded. Davidsson, Lindmark and Olofsson(1994a, 1994b) related the birth rates for Swedishregions to indicators of economic well-being insubsequent years and found some positive coeffi-cients. Remarkably, there was no statistically sig-nificant relationship between the entry rate andthe change in unemployment.6 An analysis forthe Netherlands (Economisch Instituut voor hetmidden- en kleinbedrijf, 1994, pp. 41–47) foundno impact of the new-firm formation rate onregional employment change in subsequent years.Summarizing the empirical evidence it can be con-cluded that the importance of new firms forregional employment remains unclear.

438 Michael Fritsch

3. Data base

The analyses presented here are based on 1986–89data on new establishments in the West Germanplanning regions. These data were generated byusing social insurance statistics (for details Fritsch,1992). The data contain only those establishmentsthat have at least one employee working theminimum number of hours required for compul-sory insurance contributions, i.e. they comprise nobusinesses entirely based on self-employment,family labor or marginal workers. In order to avoidlarge distortions due to errors in the data onlythose new establishments that had existed for atleast one year and that had no more than 50employees during the first two years were con-sidered as start-ups.7 The analysis encompasses thestart-ups of the years 1986–89. Although the database is limited to the establishments, the new unitsare referred to as ‘firms’ to indicate that the focusis on business formation.

The regional basis for the analyses are the 75planning regions (‘Raumordnungsregionen’) ofWest Germany. Figure 1 shows the commonclassification of West German planning regionsaccording to their population density into agglom-erations, moderately congested regions and ruralareas.8

4. How to calculate entry rates

Regions often differ considerably in theireconomic potential (e.g. number of establishmentsand employees). Therefore, the absolute numberof new firms may be no meaningful indicator forinterregional comparisons of new-firm formationprocesses and an entry rate has to be determined.9

There are a number of alternative ways of how tocalculate such a rate with regard to the measure-ment of new firms (Section 4.1), the choice of thereference figures (Section 4.2) and the question,of how to account for the effects of divergentsectoral structures in the regions (Section 4.3).

4.1. Measuring entries

The birth of firms can be measured in two ways.One alternative is to simply count the number ofnew firms, the other one is to weigh the new-comers by their size (e.g. by the number of

employees). Weighing the new firms by their sizecreates at least two problems: First, it is verydifficult to identify new firms at all and in manycases, reliable information on their size is notavailable. Second, since at least some of the newfirms develop quite dynamically within the earlyperiod of their existence, a specific point in timein relation to the date of birth would have to bedetermined so that size may be measured in acomparable manner. To avoid such problems theanalyses presented here made use only of theinformation on the number of new firms regard-less of their size.10

4.2. ‘Labor-market’ and ‘ecological’ approach

There are also alternatives concerning the denom-inator of the new-firm formation rate. One possi-bility for calculating an entry rate is to standardizethe number of start-ups by the number of estab-lishments that already exist in the respectiveregion. In this case, the entry rate can be lookedat as an indicator for the ‘fertility’ of the popula-tion of business establishments in the region(‘ecological’ approach). An alternative way offorming an entry rate is to divide the number ofnew firms by the regional workforce; this mightbe labeled the ‘labor-market’ approach. It is basedon the notion that all members of the workforceare faced with the decision to work as dependentemployees in someone else’s business or to starttheir own firm. Because start-ups are usuallylocated near the founder’s residence (cf. Gudgin,1978; O’Farrell, 1986), the regional workforce canbe seen as an appropriate measure of the numberof potential entrepreneurs. The entry rateaccording to the labor market approach maybe interpreted as the propensity of a member ofthe regional workforce to start his or her ownbusiness.11

The choice of the denominator for the entryrate does not only affect the results of empiricalanalyses of interregional differences in new firmformation (see Audretsch and Fritsch, 1994a and1994b for details), but has also implications foranalyses confined to specific sectors. If appro-priate data are available, there is no problem incalculating the entry rate for certain industriesaccording to the ecological approach. This is doneby dividing the number of new firms in an industry

New Firms and Regional Employment Change 439

440 Michael Fritsch

Fig. 1. Geographical distribution of West German planning regions.

by the number of incumbent firms in the sameindustry. Such an analysis becomes far moreproblematic when the labor-market approach isapplied: One may divide the number of new firmsin a specific industry by the number of employeesin that particular industry, but it is not possible tounequivocally allocate the unemployed individualsto the different industries.12

4.3. Adjustment for effects of the sectoral structure

New-firm formation rates differ between industriesfor several reasons, e.g. diverging change ofdemand for the goods supplied or differences inbarriers to entry. As a consequence, one mayexpect the sectoral structure of a region to have aneffect on the region’s entry rate: If there is a highshare of establishments in industries which onaverage have a relatively high new-firm formationrate, the regional new-firm formation rate shouldalso be relatively high. Vice versa, a high shareof establishments in industries with a low entryrate should lead to a correspondingly low regionallevel of new-firm formation.

To account for such sectoral effects an adjustedentry rate is formed indicating the regional factorin founding activities according to a shift-shareapproach. In the first step of this procedure, theaverage entry rate in the industries is calculatedon a national level. Since only the ecologicalapproach allows for a meaningful differentiationby industry, the average entry rate is calculatedas “number of new firms in industry i/number ofestablishments in industry i”. In a second step, ahypothetical number of entries is determined foreach region by multiplying the number of estab-lishments in each industry by the average entryrate of that particular industry. This hypotheticalnumber of start-ups tells us how many new firmsin a region are to be expected according to itssectoral structure, provided the regional entry ratein the industries equals the national average. Thesector-adjusted number of entries is then calcu-lated by subtracting the hypothetical number ofnew firms from the actual number. To form anentry rate this figure can be related to differentdenominators, e.g. regional workforce or numberof incumbent establishments. This sectoriallyadjusted entry rate permits determining the degree

to which new-firm formation in a region is aboveor below the national average when the effects ofthe sectoral structure are taken into account.

The analyses presented here use four alterna-tive versions of a regional new-firm formationrate. They are formed by combining the actual andthe sector-adjusted number of start-ups with theregional workforce and with the number of firmswhich already exist. Therefore, we have:

Entry rate A (labor market approach):

number of new firmsworkforce

Entry rate B (ecological approach):

number of new firmsnumber of establishments

Entry rate A adjusted:

number of new firms – hypothetical number of new firmsworkforce

Entry rate B adjusted:

number of new firms – hypothetical number of new firmsnumber of establishments

In the calculations reported here a classificationinto 55 private industries was used for estimatingthe hypothetical number of entries for all sectors.The estimations of the hypothetical number ofentries in manufacturing are based on a divisioninto 35 industries and for calculating the hypo-thetical number of new firms in the service sectora classification into 15 subsectors was utilized.

5. Comparing alternative new-firm formationrates

Comparing the actual number of entries with thehypothetical number of new firms reveals a highdegree of agreement (Figure 2). In the calculationsfor all sectors the hypothetical number of newfirms deviates from the actual number by no morethan 10% in nearly 70% of all cases (regions). Inabout 85% of all cases the deviation is less than15% (see Table I). The maximum deviation of theactual number of entries from the hypothetical oneis –30% and + 35% with these cases being rare


exceptions as can be seen from Figure 2. For theaverage region the difference between the hypo-thetical and the actual number of entries amountsto about 10% of the hypothetical number, i.e. 90%of the regional start-ups can be explained by thesectoral structure of the respective region.Calculations limited to manufacturing and theservice sector arrive at rather similar results.13

These comparisons of the hypothetical and theactual number of entries show clearly that thenumber of new firms set up in a region is largelydetermined by the sectoral structure in that region.Therefore, it seems to be important to account forthe effects of the sectoral structure on the entryrate when analyzing interregional differences innew-firm formation. Without such an adjustmentof entry rates there is great danger that the resultsreflect the differences in the sectoral structurerather than the differences in the level of founding-activities or entrepreneurship.14

Table II shows the correlation coefficients forthe relationship between the entry rates in differentdefinitions for all sectors as well as for manufac-turing and services separately. The entry rates forall sectors are positively correlated with the lowestcoefficients between entry rate A (labor marketapproach) and the sector-adjusted rates. Thestrongest correlation exists between the two sec-torially adjusted entry rates. As can be seen inTable III there is a close statistical relationshipbetween the entry rates for the different years

442 Michael Fritsch

Fig. 2. Percentage deviation of the actual number of births from the hypothetical number in the West-German planning regions1986–89.

TABLE IReference numbers for the dispersion of differences between

actual and hypothetical number of new firms*

Deviation All sectors Manufacturing Servicessmaller than

± 1% 07.6 07.3 07.0± 5% 36.3 31.3 43.3± 10% 69.6 57.6 75.0± 15% 85.3 73.6 89.6

* Pooled birth rates for the years 1986–89 (N = 300). Thetable contains the percentage of cases (regions) for which thedifference between the actual and the hypothetical number ofnew firms is below the level given in the first column.

within the 1986–89 time period: Regions that arecharacterized by a relatively high birth rate in oneyear are very likely to exibit a relatively high birthrate in the subsequent year. All coefficients ofcorrelation for the relationship of regional entryrates in the different years show high positive

values and are statistically significant at the 1%-level. This indicates that the factors determiningregional differences in new-firm formation are notsubject to short-term variations but rather continueto be effective over time. It may be concluded that,to a high degree, new-firm formation rates relyon respective regional context conditions.

6. Survey of regional differences in new firm formation 1986–89

According to the definition of a new establishmentused in the context of this paper (Section 3) therewere about 121,000 to 135,000 new establish-ments founded per year in West Germany withinthe 1986–89 time period. This accounts for abouteight to ten percent of the stock of all establish-ments at the beginning of the respective year.There are, however, considerable differencesbetween economic sectors in this respect (Figure3): The number of start-ups in relation to the stockof already existing firms (entry rate B) is consid-erably higher in the service sector than in manu-facturing. There were, on average, 5.3 new firmsper 1,000 members of the regional workforce ineach year.

As can be seen from Figure 3, more than onehalf of all start-ups occured in the highly con-gested regions. This concentration of thefounding-activities in the agglomerations isprobably a result of the larger economic potentialthere. The rural areas accounted for only 14% ofthe new firms. The overwhelming majority of newfirms (68%) was in the service industries. Theshare of the service sector in the total number ofnew firms is relatively high (71.4%) in congestedareas and lowest (60.6%) in rural areas. 62% ofall new service-sector firms were founded inagglomerations; only 12% were in rural regions.

If the number of new firms is related to theregional workforce (entry rate A) we find anegative relationship between new firm formationand regional population density, i.e. the level ofstart-up activity is below-average in the agglom-erations and relatively high for rural areas (TableIV). Standardizing the number of start-ups by thenumber of incumbent establishments (entry rateB) leads to exactly the opposite pattern: There isa positive relationship between new-firm forma-tion and regional population density.15 The also


TABLE IICorrelations between the alternative new firm formation

ratesc

Entry rate B A B adjusted adjusted

a: All sectorsA (labor market approach) 0.341a 0.178a 0.176a

B (ecological approach) – 0.709a 0.730a

A adjusted – – 0.984a

b: ManufacturingA (labor market approach) 0.106 0.117b 0.113b



c: ServicesA (labor market approach) 0.523a 0.253a 0.327a



a Statistically significant at the 1%-level.b Statistically significant at the 5%-level.c Pooled data (N = 300).

Table IIICorrelations of new firm formation rates in different yearsa

Entry rate Year t + 1 Year t + 2 Year t + 3

a: All sectorsA (labor market) 0.847 0.871 0.877B (ecological) 0.582 0.610 0.679A adjusted 0.545 0.511 0.458B adjusted 0.564 0.555 0.495

b: ManufacturingA (labor market) 0.794 0.762 0.762B (ecological) 0.622 0.546 0.586A adjusted 0.569 0.498 0.514B adjusted 0.573 0.465 0.469

c: ServicesA (labor market) 0.716 0.867 0.857B (ecological) 0.201 0.298 0.494A adjusted 0.533 0.571 0.456B adjusted 0.539 0.581 0.490

* Pooled data. Year t + 1: 225 cases; year t + 2: 150 cases;year t + 3: 75 cases. All coefficients are statistically signifi-cant at the 1%-level.

positive correlation coefficients between thesector-adjusted entry rates and population densityindicate that the level of new-firm formationactivity is clearly above the average in the agglom-erations while we find negative deviations fromthe number of start-ups expected in the ruralareas. There is obviously a clear core-peripherydownward trend in founding activities if entryrates account for the effect of a region’s sectoralstructure.

7. The impact of new-firm formation activities on regional employment

Is there an identifiable link between new-firm for-mation activities and regional employment changein subsequent years?

Though regional employment may be deter-mined by a number of factors other than new firmformation, some kind of multivariate analysiswould be desirable to test the relative importanceof entry rates for changes in the number of jobssupplied. However, attempts to identify such vari-ables have not been successful, i.e. no meaningfulapproach to explain regional employment changein West German planning regions by factors suchas regional labor market conditions (e.g. qualifi-cation of the regional workforce, regional wagelevel, unemployment rate), industry structure, per-capita income, establishment-size structure ordegree of agglomeration could be found.16

Therefore, in order to account for unobservedregion-specific effects on employment changeregressions were estimated, in which the birth rateand a dummy variable for each planning region

444 Michael Fritsch

TABLE IVCorrelations between the alternative new firm formation

rates and regional population densityc

Entry rate All sectors Manufacturing Services

A (labor market) –0.234a –0.315a –0.016B (ecological) –0.365a –0.318a –0.189a

A adjusted –0.235a –0.153a –0.220a

B adjusted –0.280a –0.223a –0.240a

a Coefficient statistically significant at the 1%-level.b Coefficient statistically significant at the 5%-level. c Pooled data (N = 300).

Fig 3. Average yearly number of new establishments by type of region.

(all together 74 regional dummy variables) wereincluded as explanatory variables. Table V showsthe beta-coefficients for the influence of the entryrate in various definitions on employment changeand the adjusted R2 for the respective regressionmodel.17 The coefficients for the regional dummyvariables are omitted due to space limitations.

In the calculations for all sectors many of thecoefficients and of the regression equations proveto be statistically significant. While there seemsto be a clear positive relationship between entryrate and employment change in year 0, the periodwhen the start-up occurs, the impact of new firmformation on employment change turns out to benegative in the subsequent years t + 1 and t + 2.For year t + 3 the unadjusted entry rates indicatea positive relationship between entry and employ-ment change but we find negative signs for theadjusted rates with relatively high and statisticallysignificant values of the respective beta-coeffi-cients. The results for year t + 4 are also contra-dictory, however, the respective coefficients aresmall and not significant. The estimates for theregressions limited to start ups in the manufac-

turing sector reflect a rather diffuse pictureshowing positive as well as negative values for theimpact of entry on employment change in a certainyear. We find a statistically significant negativesign for the relationship between entry rate A andemployment change in year t + 1 and t + 2,however, in those cases where the beta-coeffi-cients show a positive value they are not statisti-cally significant. In the service sector three of thefour alternative entry indicators have a statisticallysignificant positive sign for the relationshipbetween new firm formation and regional employ-ment change in year 0. All the other coefficientsthat prove to be statistically significant show anegative sign here.

All in all, it can be concluded that there is asomewhat positive impact of new firm formationon regional employment in the year where the newbusinesses are set up, but that the relationshipbetween entry and regional employment change insubsequent periods is negative. This finding is insharp contrast to common wisdom that is based onthe fact that new firms obviously create new jobs.The most plausible explanation for this seemingly


TABLE VBeta-coefficients for the relationship between entry rates and employment changee

Entry rate Employment change (%)

Year 0 Year t + 1 Year t + 2 Year t + 3 Year t + 4

a: All sectorsA –0.323b (0.251c) –0.972a (0.191c) –0.943a (0.104d) –0.332 (0.029) –0.079 (0.514c)B –0.071 (0.236c) –0.781a (0.217c) –0.530a (0.056) –0.320a (0.001) –0.035 (0.514c)A adj. –0.267a (0.272c) –0.095 (0.050) –0.391a (0.017) –0.640a (0.202c) –0.042 (0.514c)B adj. –0.292a (0.278d) –0.119 (0.048) –0.422a (0.026) –0.649a (0.202d) –0.059 (0.576d)

b: ManufacturingA –0.008 (0.133c) –0.400a (0.145c) –0.386a (0.083) –0.175 (0.156) –0.121 (0.229)B –0.001 (0.133c) –0.058 (0.108d) –0.039 (0.118) –0.001 (0.163) –0.280 (0.192)A adj. –0.011 (0.133c) –0.036 (0.107d) –0.052 (0.118) –0.068 (0.161) –0.046 (0.230)B adj. –0.003 (0.133d) –0.034 (0.107e) –0.082 (0.116) –0.072 (0.160) –0.018 (0.231)

c: ServicesA –0.766a (0.132c) –0.396a (0.022) –0.640a (0.064) –0.492a (0.013) –0.108 (0.467c)B –0.054 (0.011) –0.296a (0.025) –0.314a (0.059) –0.107 (0.071) –0.041 (0.464c)A adj. –0.264a (0.018) –0.057 (0.059) –0.021 (0.038) –0.098 (0.077) –0.000 (0.463c)B adj. –0.241b (0.012) –0.056 (0.060) –0.030 (0.038) –0.079 (0.079) –0.057 (0.464d)

a Coefficient statistically significant at the 1%-level.b Coefficient statistically significant at the 5%-level.c Regression model statistically significant at the 1%-level.d Regression model statistically significant at the 5%-level.e Results of regressions with regional dummy-variables (adjusted R2 values for the whole equation in parantheses).

contradictory phenomenon may be simply that,during the period under examination, the positiveimpact of new firms on regional employmentchange has been overcompensated for by otherfactors. Presumably, many of the new servicefirms represent an element of a reorganizationprocess in which activities and jobs are allocatedto the establishments within and outside therespective region in a different way.18

8. Conclusions

The results of the analyses presented here suggesta number of conclusions for future research andfor economic policy.

It was demonstrated that the number of start-ups in a region is highly influenced by the sectoralstructure in that region and that much more thanhalf of the number of start-ups can be ‘explained’by the sectoral structure. Consequently, interre-gional comparisons of new-firm formation shouldaccount for this influence and operate with secto-rially adjusted entry rates. The results of the rela-tionship between new firm formation and regionalemployment change are well suited to question thestimulation of new firm formation as a strategy fora policy aimed at promoting regional employmentgrowth. This concerns particularly non-specificforms of such a policy which try to promote allkinds of new firms. Policies that are aimed atstimulating economic development by promotingnew firm formation should be selective and tryto concentrate on job-generating start-ups. Theresults indicate that, at least for West Germanyduring the period of analysis, new firms obviouslyhave been far less important for regional employ-ment than is commonly assumed. Therefore, theendogenous potential of the regions, i.e., theincumbent firms, should not be neglected.

Notes

1 Two anonymous referees provided a number of very helpfulsuggestions for improvements.2 Boeri and Cramer (1992) analyzed the job development incohorts of newly founded West German establishments overa ten year period of time and found, that the number of jobsin these cohorts at the end of this period was in many casessmaller than at the time these establishments have been set up.The maximum number of jobs in a cohort of new establish-ments was normally attained in the second year. The decrease

in the number of jobs in the later years is mainly due to exitsfrom the market. According to Boeri and Cramer only about40% of all newly founded establishments survived the first tenyears. The average number of jobs in the surviving units was6.1 employees.3 If a newcomer has only a low rate of value added, receiveslarge amounts of inputs from outside the region and iscrowding out a competitor that bought his input mainly fromregional suppliers, then the setting up of a new establishmentleads to decrease in the regional supply of jobs.4 Cf. the empirical studies by Storey, Keasey, Watson andWynarczyk, 1987; Storey et al., 1989; Brüderl, Bühler andZiegler, 1993; Barkham, 1994.5 For the relationship between firm births and deaths seeAudretsch and Fritsch, 1996; Fritsch, 1996; Keeble andWalker, 1994.6 Davidsson, Lindmark and Olofsson (1994a, 1994b) give noinformation on the relationship between the new firm forma-tion rate and regional employment change.7 Most of the new establishments recorded by the socialinsurance statistics are very small: In the 1986–89 time periodmore than 80% had fewer than ten employees and more than90% had fewer than 50. But there were also several estab-lishment-numbers new to the data set in a certain year withmore than 100 employees. It can be assumed that these estab-lishments were new branch plants of large firms, that theyreflect errors in the social insurance statistics or that in thesecases the identification number changed for non-economicreasons. Because the analysis did not focus on new branchplants but rather on new firms, which usually start off as rel-atively small establishments, new identification numbers rep-resenting establishments with 50 or more employees were notconsidered to indicate the founding of an enterprise. However,omitting this cut-off criterion or modifying the number ofemployees from which on a new identifcation number is notconsidered to indicate the setting up of a new enterprise maylead to a different number of startups in a region but does notchange the results of the analysis in any significant way.8 The West German planning regions are based on smalleradministrative subunits (‘Kreise’) that are aggregatedaccording to the functional relationship between them. Eachplanning region consists of at least one core city and therespective suburban ring. Although planning regions aresomewhat larger than the official labor market regions, manyof them may be seen as a plausible approximation of regionallabor markets. The subunits, on the other hand, are muchsmaller than regional labor markets.9 In order to avoid the calculation of entry rates Harhoff(1995) in his analysis of regional differences of new firmformation chose the absolute number of startups as the depen-dent variable and related it to (among other independentvariables) the absolute number of establishments as well asto the absolute number of employees. However, this is not aconvincing solution to the problem of accounting for differenteconomic potentials in an interregional comparison becausethe introduction of a variable for agglomeration effects (e.g.population density) in a regression that already contains thenumber of employees would obviously lead to considerablemulticollinearity. We know from empirical work (for a surveysee Reynolds, Storey and Westhead, 1994) that such agglom-

446 Michael Fritsch

eration effects have a significant influence on new firmformation processes, i.e. that a high degree of agglomerationis conducive to the formation of new businesses.10 One reason for not using the information on the size ofthe new establishments is that many of those cases thatrepresent errors in the social insurance statistics are largerunits. Therefore, if new establishments would be weighed bytheir size, the error in the data caused by questionable caseswould be much more severe compared to analyses that arebased on the number of new establishments.11 Note that the definition of an entry rate in the waydescribed here does not necessarily imply the assumption thatthere is a linear relationship between the number of new firmsand the number of incumbents or the size of the regional work-force. In fact, if a high degree of agglomeration is conduciveto new firm formation, then a nonlinear relationship betweenthe number of startups and the size of the regional workforceis to be expected.12 One solution to that problem would be to allocate theunemployed to the different industries according to theemployment share of the industries in the respective region.However, such a procedure would be quite problematic,though, since there are large differences with respect to theemployment performance of industries; consequently they donot all contribute to unemployment to the same degree.13 In addition to manufacturing and services the calculationfor all sectors comprise agriculture, mining, energy andconstruction.14 However, empirical tests with these sector-adjusted entryrates lead to the same variables that determine regional dif-ferences in new firm formation than are found on the basis ofnon-sectorially adjusted rates.15 The reason for this difference obviously lies in the factthat agglomerations exhibit a higher average establishment-size and therefore a larger number of potential founders isinvolved with each business. For a more detailed analysis seeAudretsch and Fritsch (1994a, 1994b).16 This corresponds to the results of a study by Bade (1991),who tried to explain and to predict employment change in theWest German planning regions by all kinds of socio-economicfactors and did not find any reasonable estimate.17 The data on employment were also taken from socialinsurance statistics.18 The analyses for certain sectors did not yield clear sector-specific patterns in the relationship between entries andeconomic growth. An exeption were calculations restricted tocertain ‘high-tech’ sectors that showed a relatively strongpositive link between entries and regional employment changeduring periods to follow. It might be concluded that start-upsin these sectors are more likely to survive and generate aconsiderable number of jobs than do new firms in othersectors. However, even in ‘high-tech’ sectors many new firmsfail or stay small so that – as is the case in all sectors – fastgrowing firms are a rare exeption. Therefore, the sector a newfirm belongs to is not an adequate criterion to identify job-generating start-ups with any certainty.

References

Ashcroft, Brian and James H. Love, 1994, ‘Modelling FirmBirths and Employment Change in the British Counties’,Paper presented to the 34th Congress of the RegionalScience Association, Groningen, The Netherlands, August22–26.

Audretsch, David B. and Michael Fritsch, 1994a, ‘On theMeasurement of Entry Rates’, Empirica 21, 105–113.

Audretsch, David B. and Michael Fritsch, 1994b, ‘TheGeography of Firm Births in Germany’, Regional Studies27, 359–365.

Audretsch, David B. and Michael Fritsch, 1996, ‘CreativeDestruction: Tubulence and Economic Growth’, in ErnstHelmstädter and Mark Perlman (eds.), Behavioral Norms,Technological Progress, and Economic Dynamics: Studiesin Schumpeterian Economics, Ann Arbor: University ofMichigan Press.

Bade, F.-J., 1991, ‘Regionale Beschäftigungsprognose 1995’,Mitteilungen aus der Arbeitsmarkt- und Berufsforschung24, 25–44.

Barkham, R. J., 1994, ‘Entrepreneurial Characteristics and theSize of the New Firm: A Model and an Econometric Test’,Small Business Economics 6, 117–125.

Birch, David, 1979, The Job Generation Process,Cambridge/Mass: mimeo.

Boeri, Tito M. and Ulrich Cramer, 1992, ‘EmploymentGrowth, Incumbents, and Entrants – Evidence fromGermany’, International Journal of IndustrialOrganization 10, 545–565.

Brüderl, J. P. and R. Schüssler, 1990, ‘OrganizationalMortality: The Liability of Newness and Adolescence’,Administrative Science Quarterly 35, 530–547.

Brüderl, J., P. Preisendörfer and R. Ziegler, 1992, ‘SurvivalChances of Newly Founded Business Organizations’,American Sociological Review 57, 227–242.

Brüderl, Josef, Christian Bühler and Rolf Ziegler, 1993,‘Beschäftigungswirkungen neugegründeter Betriebe’,Mitteilungen aus der Arbeitsmarkt- und Berufsforschung26, 521–528.

Davidsson, Peer, Leif Lindmark and C. Olofsson, 1994a, ‘NewFirm Formation and Regional Development in Sweden’,Regional Studies 27, 395–410.

Davidsson, Peer, Leif Lindmark and C. Olofsson, 1994b,‘Entrepreneurship and Economic Development: The Roleof Small Firm Fomation and Expansion for RegionalEconomic Well-Being’, Journal of Enterprising Culture 1,347–365.

Dunne, P., M. J. Roberts and L. Samuelson, 1989, ‘Plantturnover and gross employment flows in the US manufac-turing sector’, Journal of Labor Economics 7, 48–71.

Economisch Instituut voor het midden- en kleinbedrijf, 1994,Kleinschalig ondernemen 1994, deel II: Regional-economische dynamiek en werkgelegenheidscreatie,Zoetermeer.

Fritsch, Michael, 1990, Arbeitsplatzentwicklung inIndustriebetrieben, Berlin: de Gruyter.

Fritsch, Michael, 1992, ‘Regional Differences in New FirmFormation: Evidence from West Germany’, RegionalStudies 25, 233–241.


Fritsch, Michael, 1996, ‘Turbulence and Growth in West-Germany: A Comparison of Evidence for Regions and onthe Industry Level’, Review of Industrial Organization 10.

Gerlach, Knut and Joachim Wagner, 1993, ‘Gross andNet Employment Flows in Manufacturing Industries’,Zeitschrift für Wirtschafts- und Sozialwissenschaften 113,17–28.

Gerlach, Knut and Joachim Wagner, 1994, ‘Entries, Exits, andthe Dynamics of Employment: Lower Saxony,1978–1990’, in Friedrich Buttler, Joachim Schwarze andGerd Wagner (eds.), Labor Market Dynamics in PresentGermany, pp. 240–256, Frankfurt a.M., New York:Campus.

Gudgin, Graham, 1978, Industrial Location Processes andRegional Employment Growth, Westmead: Saxon House.

Harhoff, Dietmar, 1995, ‘Firm Formation and RegionalSpillovers – Evidence from Germany’, Zentrum füreuropäische Wirtschaftsforschung, Discussion Paper 95-11(Industrial Economics and International ManagementSeries), Mannheim.

Hart, Mark and Eric Hanvey, 1995, ‘Job Generation and Newand Small Firms: Some Evidence from the Late 1980s’,Small Business Economics 7, 97–109.

Keeble, David and Sheila Walker, 1994, ‘New Firms, Small

Firms and Dead Firms: Spatial Patterns and Determinantsin the United Kingdom’, Regional Studies 27, 411–427.

König, Andreas, 1994, Betriebliche Beschäftigungsdynamikund personeller Strukturwandel – Eine Longitudinal-analyse, Frankfurt a.M., New York: Campus.

O’Farrell, Patrick, 1986, Entrepreneurs and IndustrialChange, Dublin: Irish Management Institute.

Storey, David J. and Steve Johnson, 1987, Job Generation andLabour Market Change, London: MacMillan.

Storey, David J., K. Keasey, R. Watson and P. Wynarczyk,1987, The Performance of Small Firms – Profits, Jobs andFailures, London: Croom Helm.

Storey, David J. et al., 1989, Fast Growth Small Businesses.Case Studies of 40 Small Firms in NE England,Department of Employment Research Paper No. 67.

Reynolds, Paul D., 1994, ‘Autonomous Firm Dynamics andEconomic Growth in the United States, 1986–90’, RegionalStudies 27, 429–442.

Reynolds, Paul D. and Wilbur R. Maki, 1990, BusinessVolatility and Economic Growth (Final Project Report),Minneapolis: mimeo.

Reynolds, Paul D., David J. Storey and Paul Westhead, 1994,‘Cross National Comparisons of the Variation in New FirmFormation Rates’, Regional Studies 27, 443–456.

448 Michael Fritsch

Documents

New Firms and Regional Employment Change