The current empirical entrepreneurship literature mainly shows a positive correlation between entrepreneurship (measured as the number of startups) and economic growth. However, the mechanisms by which entrepreneurship exerts its positive influence are not obvious. This chapter studies the connections between startups and local development at the municipal level in Sweden 2000–08.We use a unique database including not only total startups, but also data on startups divided into six branches to study the impact of entrepreneurship on population and employment growth. Analyses are performed on all municipalities as well as by municipality type. In contrast to previous research, our results indicate that for several branch groups startup effects on growth may be more pronounced in low density areas than in urban agglomerations.
The strategy of fostering growth by promoting entrepreneurship is ever more popular. Nationally and locally, we can see policy support for entrepreneurship, and startups in particular, including major investments of public funds. At the local government level, expenditures for business promotion activities were on average about €30 per inhabitant in 2009, and varied between €0 and €490 (http://www.kolada.se). In Sweden, Regional Growth Programs (RTP) emphasize entrepreneurship as a prioritized strategy and indicator, and local governments in Sweden are using a broad range of measures to increase the share of startups, including building supporting infrastructure and cooperating with other public and private sector actors (Rader Olsson and Westlund, 2011).
The term “entrepreneurship” is increasingly used broadly to denote a wide range of activities related to social welfare and political goals as well as new firm creation (see Westlund, 2010). However, in this chapter we focus only on entrepreneurship in the form of startups.
Notwithstanding the interest among governments in supporting entrepreneurship through growth, there are only a limited number of studies in the current literature that are focused on the effect of startups on local growth in population and employment. Empirical studies of the determinants of variations in startup rates generally use regional data (Gries and Naudé, 2008). Early studies of regional variations in startups were Johnson (1983) and Wood et al. (1993), and Storey and Johnson (1987) was an early contribution on the causes of startups.
As to the effects of entrepreneurship, much of the literature deals with firms and establishments and their performance. As noted by, among others, Wennekers and Thurik (1999), Carree and Thurik (2003), and Fritsch and Mueller (2004), the effects of startups at firm level can be distinguished as direct effects and indirect supply-side effects (intermediate linkages). Direct effects are the startups' new employment and new production, and startups' direct contributions to in-migration and increased regional productivity. Direct effects also include decline or closure of incumbents that cannot face the new competition. The indirect supply-side effects are divided by Fritsch and Mueller (2004) into four types:
1 secured efficiency: startups force incumbents to behave more efficiently;
2 acceleration of structural change: incumbents are substituted by new firms;
3 amplified innovation: new firms may introduce innovations; and
4 greater variety: new firms may lead to a greater variety of products and problem solutions.
Together these indirect supply-side effects enhance the regional competitiveness and growth.
Over the last 10–15 years a growing number of studies have analyzed regional or local effects of entrepreneurship. Within this field of research, several studies have indicated a clear positive influence of startups on regional employment in the USA (for example, Reynolds, 1999; Acs and Armington, 2004). European studies have shown more ambiguous results. Early studies of West Germany and the Netherlands (Audretsch and Fritsch, 1996; Fritsch, 1996, 1997; EIM, 1994) showed no correspondence between entrepreneurship and regional economic growth during the 1980s. However, later German studies (for example, Audretsch and Fritsch, 2002; Fritsch and Mueller, 2004) showed the opposite results for the 1990s. These and other contributions (for example, Acs and Mueller, 2008; Fritsch and Mueller, 2008; Andersson and Noseleit, 2011) conclude that the net effect of startups may be negative in the short term but thereafter turn positive in processes that might have a significant impact on growth for up to 10 years.
Swedish research of the impact of entrepreneurship on regional employment and other regional economic variables has showed significant positive effects (Davidsson et al., 1994; Fölster, 2000; Braunerhjelm and Borgman, 2004; Borgman and Braunerhjelm, 2007).2 Andersson and Noseleit (2011) have also confirmed that the wave pattern of first negative and then positive effects of startups on employment seem to hold for Sweden as well. Another result of theirs is that when they divide the startups in three sector aggregates – manufacturing, low-end services, and high-end services – the effects of startups on employment vary.
This chapter focuses on the problem of why the effects of startups might vary between different regions. Fritsch and Schroeter (2011) have highlighted this problem when they analyzed regional variations in the employment effects of startups. In a study of Germany they include a number of regional characteristics as control variables. The most important variable seems to be population density, i.e. The positive effects of startups on employment growth are more pronounced in high-density areas than in rural regions. Another conclusion is that the positive effects of startups diminish with increasing startup rates.
In contrast to most of the existing literature on entrepreneurship at regional level, we in this chapter focus on the local government (municipality) level. The reason is that in Sweden the municipalities are the most important policy actors concerning promoting local entrepreneurship. By focusing on the municipalities we focus on the level where entrepreneurship most clearly can be influenced by policy measures.
We study the impact of startups at municipal level on two variables, employment, and population. We are, of course, aware of that the causation chain might also go in the opposite direction, i.e. that growth of employment and population should have a positive impact on entrepreneurship. However, we use startups per capita as our entrepreneurship variable, i.e. we standardize for population. Moreover, to avoid obvious endogeneity problems, the dependent variables, employment growth and population growth, are measured for a later time period than the startup measure.
As shown above, there are recent studies of the regional impact of startups divided by sectors and of regional variations of the effects. To our knowledge, there are no studies that combine these approaches and examine the effects of startups in various sectors on local economic development and in different types of regions.
Our aim is to test the following hypotheses:
1 Startups, over a time period of about 6–8 years, have, in general, a positive impact on local employment and population.
2 Startups' impacts on employment growth should be stronger than impacts on population growth, since the effects on employment are of a more direct nature.
3 The effects of startups on employment and population growth vary by branch groups. We expect that startups in service sectors have a stronger impact on employment growth for two reasons: a) low costs of entry for startups in services, and b) most new employment in Sweden occurs in service sectors.
4 The effects of startups on employment and population growth vary by type of municipality. In line with Fritsch and Schroeter (2011) we expect the effects of startups to be strongest in urban areas.
We use data on startups provided by the Swedish Agency for Growth Policy Analysis (Tillväxtanalys), the official provider of statistics on startups of new firms and bankruptcies in Sweden. To avoid the effects of coincidental occurrences in a certain year, the startups data covers the period 2002–08. Only genuinely new firms are included in the statistics. The number of startups is divided per capita and, besides the total sum, they are divided into six branch groups:
1 manufacturing;
2 construction;
3 trade, hotels, and restaurants;
4 transportation and communications;
5 financial and business services (excluding real estate service);
6 education, health and medical service, and other public and personal services.
The other data are provided by Statistics Sweden (http://www.scb.se).
As has been shown by, among others, Fritsch and Mueller (2008) and Andersson and Noseleit (2011), the effects of startups on employment may be negative in the short run but positive in a longer perspective. For this reason we use startup data over a period of six years. In order to avoid the endogeneity problems that can occur when data for both the independent and the dependent variables covers the same period, the data for the dependent variables, changes in employment and population respectively, are for a later period than the data for the independent variables.
The analyses are performed with all municipalities and with the municipalities distinguished as urban or rural. These are based on classifications developed by the Swedish Board of Agriculture and reflect population, population density, and labor market areas.3
This analysis focuses on the local effects of startups. The impact of startups and control variables on employment ratio and population growth, respectively, is analyzed for startups in total and for the six branch groups. Also, we analyze the impact of startups for urban and rural municipalities separately.
Two models were constructed estimating the effects of startups on employment growth and population in Swedish municipalities (2009–10). In both cases, startups per capita were used as the independent variable, along with several control variables for year 2002. Control variables are the share of university educated
population, employment in 2002, intra- and inter-municipal accessibility,4 and dummies distinguishing municipalities with an industrial specialization in raw materials or tourism. These two branches have expanded during the 2000s and, in contrast to most other branches, mining and tourism are mainly based on place-bound resources like ore and specific nature (see Table 3.1).5
Our main variable of interest measures startups per 10,000 inhabitants over the period 2002–08. The branch measures measure the same startups but disaggregated by six broad levels of industry specialization. Hence, the mean values of the decomposed startups variables add up to the grand mean of the overall startup variable. The lower number of observations for the disaggregated measures is because not all municipalities had startups in every level of industry during the reporting period. Share of university educated is the share of workers in each municipality with a college education of at least three years, or the equivalent of a bachelor's degree. The share of labor force employed refers to the share of the total active labor force currently employed.
Intra-municipal accessibility is a distance-decay weighted accessibility to purchasing power (wage sums), within each municipality, while inter-municipal accessibility refers to a similar measure for municipalities in the same integrated labor region, as well as to all other municipalities in the country. The measures represent a continuous view of geography, where all activities affect each other in space, but where the effects attenuate with distance. The measures have been shown to act as catch-all variables, e.g. for industry diversity and spillover effects (Andersson and Klaesson, 2009; Johansson et al., 2003).
Entrepreneurship, as predicted, has an effect on employment growth in municipalities. This variable is highly significant in our model when considering all Swedish
Variables | All municipalities | Urban municipalities | Rural municipalities |
Startups 2002–08 | 3.656*** | 6.505*** | 0.316** |
(4.453) | (3.145) | (2.321) | |
Share of university educated | −731.1 | −7.219 | 45.53 |
(−0.290) | (−1.228) | (0.0924) | |
Share of labor force employed | −4,791*** | −3.565 | 331.4 |
(−2.648) | (−0.651) | (1.102) | |
Intra-municipal accessibility | 609.1*** | 1,046*** | 94.49*** |
(7.176) | (4.720) | (5.991) | |
Inter-municipal accessibility | −8.350 | −31.57 | −13.04 |
(−0.131) | (−0.150) | (−1.495) | |
Mining dummy | 217.7 | 370.0*** | |
(0.256) | (3.727) | ||
Tourism dummy | −300.6 | 571.4 | −34.33 |
(−0.990) | (0.396) | (−0.872) | |
Constant | −10,738*** | −20,728*** | −1,839*** |
(−5.483) | (−2.680) | (−6.136) | |
Observations | 290 | 93 | 197 |
R-squared | 0.344 | 0.413 | 0.339 |
Note:t-statistics in parentheses: *** p<0.01, ** p<0.05, * p < 0.1
Entrepreneurship is measured as startups per inhabitant and year, 2002–08. Control variables are measured as the percentage change 2002–08.
municipalities. It is also a highly significant variable explaining employment growth in both urban and rural communities. This is consistent with the findings of Fritsch and Schroeter (2011) and our hypothesis (see Tables 3.2 and 3.3).
The basic model also supports the hypothesis that entrepreneurship has a significant and positive effect on population change. Highly significant results were also evident in rural areas. However, entrepreneurship does not contribute to explaining population growth in urban areas.
The models estimating the effect of startups on population and employment growth in Swedish municipalities was also used to test the hypothesis that startup effects vary by branch groups. The presentation of tables of regression analyses is focused on the effects on employment. The coefficients for the impact of startups on population are presented in the text, but tables are in most cases not shown.
Manufacturing accounts for about 20 percent of non-farm jobs in Sweden. About three percent of manufacturing jobs are in sparse rural areas, and 43 percent are
All municipalities | Urban municipalities | Rural municipalities | |
Startups 2002–08 | 9.94e-06** | −3.17e-06 | 1.78e-05*** |
(2.376) | (−0.523) | (2.996) | |
Share of university educated | 0.0307** | 0.00170 | 0.0609*** |
(2.395) | (0.0985) | (2.831) | |
Share of labor force employed | 0.0529*** | 0.0561*** | 0.0480*** |
(5.740) | (3.498) | (3.655) | |
Intra-municipal accessibility | 0.00278*** | 0.00360*** | 0.00250*** |
(6.429) | (5.540) | (3.633) | |
Inter-municipal accessibility | 0.00112*** | 0.00258*** | 0.000671* |
(3.433) | (4.173) | (1.764) | |
Mining dummy | −6.24e-05 | −0.000600 | |
(−0.0144) | (−0.138) | ||
Tourism dummy | 0.00177 | 0.000368 | 0.000467 |
(1.144) | (0.0871) | (0.272) | |
Constant | −0.120*** | −0.162*** | −0.107*** |
(−11.98) | (−7.144) | (−8.198) | |
Observations | 290 | 93 | 197 |
R-squared | 0.618 | 0.501 | 0.420 |
t-statistics in parentheses: *** p<0.01, ** p<0.05, * p<0.1.
Entrepreneurship is measured as startups per inhabitant and year, 2002–08. Population and control variables are measured as the percentage change in population 2002–08.
in areas classified as rural. The rest are rather evenly distributed among urban and metropolitan areas (about 27 percent each).
Specifying the model for this branch indicates a strongly positive effect of new manufacturing startups on employment growth in all municipalities, and further specifying by municipality type reveals this variable as highly significant in urban communities but not in rural communities.
Manufacturing startups did not have a significant effect on population growth in Swedish municipalities (table not shown here). This might indicate that manufacturing startups to a large extent are employing workers already residing in the local labor market (see Table 3.4).
The construction industry accounts for about seven percent of all non-farm jobs in Sweden. Jobs in this sector are expected to increase between 2008 and 2030 (Statistics Sweden, 2008). As in most countries, the share of jobs in this sector is highest in metropolitan areas (36 percent of all construction jobs in Sweden) and lowest in sparse rural areas (3 percent).
The model does not indicate that construction startups affect employment growth in Swedish municipalities, excepting a weak significance in rural municipalities (see Table 3.5).
All municipalities | Urban municipalities | Rural municipalities | |
Startups 2002–08 | 19.15*** | 68.98*** | 1.112 |
(2.691) | (2.907) | (1.153) | |
Share of university educated | 5,367** | −2,562 | 465.5 |
(2.436) | (−0.471) | (0.964) | |
Share of labor force employed | −3.700* | −1.677 | 335.8 |
(−1.867) | (−0.307) | (1.014) | |
Intra-municipal accessibility | 636.3*** | 1,321*** | 97.02*** |
(7.111) | (5.637) | (5.942) | |
Inter-municipal accessibility | 31.77 | 228.1 | −12.01 |
(0.412) | (1.129) | (−1.039) | |
Mining dummy | −52.86 | 343.7*** | |
(−0.0596) | (3.300) | ||
Tourism dummy | 191.4 | 1,925 | 6.120 |
(0.664) | (1.281) | (0.175) | |
Constant | −12.816*** | −33.634*** | −1.894*** |
(−6.082) | (−4.416) | (−6.032) | |
Observations | 287 | 92 | 195 |
R-squared | 0.317 | 0.405 | 0.318 |
t-statistics in parentheses: * p < 0.01, ** p < 0.05, *** p < 0.1.
All municipalities | Urban municipalities | Rural municipalities | |
Startups 2002–08 | 2.907 | 8.632 | 1.034* |
(0.798) | (0.874) | (1.932) | |
Share of university educated | 5,357** | 1.553 | 305.8 |
(2.401) | (0.279) | (0.637) | |
Share of labor force employed | −3,485* | −1.172 | 378.5 |
(−1.707) | (−0.195) | (1.167) | |
Intra-municipal accessibility | 608.3*** | 1,157*** | 98.43*** |
(6.669) | (4.882) | (6.077) | |
Inter-municipal accessibility | 8.221 | 167.0 | −18.13 |
(0.104) | (0.784) | (−1.534) | |
Mining dummy | −34.03 | 350.0*** | |
(−0.0379) | (3.379) | ||
Tourism dummy | 175.4 | 701.8 | −13.25 |
(0.576) | (0.453) | (−0.360) | |
Constant | −11.370*** | −28.087*** | −1.810*** |
(−5.514) | (−3.616) | (−5.934) | |
Observations | 287 | 92 | 195 |
R-squared | 0.301 | 0.352 | 0.326 |
Standard errors in parentheses: * p < 0.01, ** p < 0.05, *** p < 0.1.
However, as shown in Table 3.6, the model indicates that entrepreneurship in the construction industry has a positive effect on population growth in Swedish municipalities. Estimating the model by municipality type reveals that this effect is concentrated in rural areas. New firm formation in construction should normally be regarded as a reflection of population growth, i.e. the reverse causal relationship than the one we examine here. However, as the startup variable here precedes population growth in time, an impact of startups in construction on population growth seems to occur too, at least in rural areas.
The trade, hotels, and restaurant branches account for about 18 percent of all employed in Sweden. Here again, jobs are rare in sparse rural areas (2 percent of all Swedish jobs in these sectors) but fairly evenly divided among rural and urban areas (24 and 29 percent, respectively, with the expected highest share (44 percent) in the metropolitan areas. Trade, hotels, and restaurants are branches with a traditionally high rate of startups compared to mature firms.
The model indicates that startups in these dynamic branches have only a weakly significant effect on employment growth in Swedish municipalities (see Table 3.7). These effects are, as indicated by the model, strongest in urban areas. The model also shows a significant effect of startups on population (coefficient 3.26e–05; p < 0.05; R-squared 0.618).
All municipalities | Urban municipalities | Rural municipalities | |
Startups 2002–08 | 3.68e-05** | −3.26e-05 | 6.89e-05*** |
(2.056) | (−1.202) | (2.966) | |
Share of university educated | 0.0433*** | −0.00582 | 0.0733*** |
(3.957) | (−0.382) | (3.513) | |
Share of labor force employed | 0.0459*** | 0.0576*** | 0.0474*** |
(4.575) | (3.489) | (3.366) | |
Intra-municipal accessibility | 0.00293*** | 0.00354*** | 0.00270*** |
(6.552) | (5.444) | (3.845) | |
Inter-municipal accessibility | 0.00140*** | 0.00262*** | 0.000517 |
(3.597) | (4.487) | (1.007) | |
Mining dummy | 0.000934 | −0.000928 | |
(0.212) | (−0.206) | ||
Tourism dummy | 0.00222 | 0.00157 | 0.00136 |
(1.485) | (0.369) | (0.853) | |
Constant | −0.125*** | −0.161*** | −0.107*** |
(−12.34) | (−7.552) | (−8.049) | |
Observations | 287 | 92 | 195 |
R-squared | 0.618 | 0.512 | 0.421 |
Standard errors in parentheses: *** p < 0.01, ** p < 0.05, p < 0.1.
All municipalities | Urban municipalities | Rural municipalities | |
Startups 2002–08 | 5.686* | 18.62* | 0.595 |
(1.814) | (1.925) | (1.355) | |
Share of university educated | 4.908** | 2.019 | 343.1 |
(2.194) | (0.370) | (0.712) | |
Share of labor force employed | −2.769 | 1.138 | 459.0 |
(−1.338) | (0.205) | (1.396) | |
Intra-municipal accessibility | 590.5*** | 1,083*** | 93.81*** |
(6.665) | (4.675) | (5.811) | |
Inter-municipal accessibility | −3.573 | 75.01 | −13.92 |
(−0.0454) | (0.347) | (−1.203) | |
Mining dummy | 27.29 | 356.0*** | |
(0.0305) | (3.407) | ||
Tourism dummy | 62.92 | 1.064 | −9.707 |
(0.205) | (0.707) | (−0.256) | |
Constant | −11.434*** | −27,118*** | −1,855*** |
(−5.571) | (−3.543) | (−6.026) | |
Observations | 287 | 92 | 195 |
R-squared | 0.308 | 0.374 | 0.319 |
*** p < 0.01, ** p < 0.05, * p < 0.1.
The transportation and communications branches together comprise about seven percent of Swedish jobs with an average of five employees per facility but including both large and small employers. Almost half of all employees in this branch (46 percent) are employed metropolitan areas, 29 percent in urban areas, 22 percent in rural areas, and only three percent in sparse rural areas.
These branches appear to be a significant explanatory variable for employment growth in Swedish municipalities taken as a whole. The model also shows a highly significant effect on population growth when considering all municipalities, though the coefficient is small (0.000134; p < 0.001; R-squared 0.618) (see Table 3.8).
This branch deserves further study to determine whether or not startups are hiring existing workers or leading to the restructuring of skill profiles in local municipalities.
The financial and business services sectors (excluding real estate service) account for about 15 percent of jobs in Sweden. As might be expected, this branch is highly concentrated in metropolitan areas (60 percent of jobs in these sectors) and is almost non-existent in sparse rural areas (one percent of jobs). Twelve percent
All municipalities |
Urban municipalities |
Rural municipalities |
|
Startups 2002–08 | 27.07*** | 53.09 | 0.819 |
(2.618) | (1.651) | (0.498) | |
Share of university educated | 3.699 | 2.093 | 316.8 |
(1.610) | (0.380) | (0.638) | |
Share of labor force employed | −3,377* | −1.017 | 399.3 |
(−1.701) | (−0.179) | (1.214) | |
Intra-municipal accessibility | 628.7*** | 1,040*** | 97.55*** |
(7.004) | (4.376) | (5.849) | |
Inter-municipal accessibility | 17.53 | 25.45 | −11.49 |
(0.227) | (0.110) | (−0.959) | |
Mining dummy | −243.4 | 339.3*** | |
(−0.273) | (3.242) | ||
Tourism dummy | 158.8 | 1.134 | 10.12 |
(0.546) | (0.748) | (0.287) | |
Constant | −12.104*** | −22.844*** | −1.911*** |
(−5.826) | (−2.714) | (−5.766) | |
Observations | 286 | 92 | 194 |
0.317 | 0.367 | 0.313 |
of jobs in this sector are located in rural areas and 26 percent in non-metropolitan urban areas.
The model indicates that startups in these sectors positively affect employment growth in urban areas, as might be expected given the relatively low cost of entry and large customer base in urban areas. Less obvious is the significant effect of this variable on employment growth in rural areas. These sectors comprise a number of services that can, thanks to advances in IT, effectively serve customers from remote locations–particularly in Sweden where IT infrastructure is well developed. In other words, our model may be capturing the effect of startup business and financial consulting firms registered in rural communities but serving predominantly “urban” clients. In other words, urban professionals may be moving to rural communities and taking their jobs with them. The model explaining population growth indicates significant, but much smaller effects of startups in these branches when considering all municipalities and in rural communities, but not in urban areas (p < 0.001) (see Table 3.9).
The service sector (excluding business and financial services) represents over a third of all Swedish jobs (34 percent). Many of these jobs are those that require personal contact with clients: teachers, health care professionals, and many
Variables | All municipalities |
Metro and urban municipalities |
Rural and sparse rural municipalities |
Startups 2002–08 | 9.171*** | 10.52** | 0.891** |
(4.951) | (2.534) | (2.150) | |
Share of university educated | −4.869 | −10.709 | −138.4 |
(−1.631) | (−1.510) | (−0.257) | |
Share of labor force employed | −5.101*** | − 3.5 | 255.8 |
(−2.610) | (−0.614) | (0.774) | |
Intra-municipal accessibility | 616.7*** | 1,061*** | 94.28*** |
(7.207) | (4.646) | (5.885) | |
Inter-municipal accessibility | −35.64 | 28.08 | −13.48 |
(−0.472) | (0.131) | (−1.176) | |
Mining dummy | 37.65 | 362.0*** | |
(0.0437) | (3.492) | ||
Tourism dummy | −234.5 | 509.4 | −30.47 |
(−0.793) | (0.342) | (−0.771) | |
Constant | −9.248*** | −20.629** | −1.734*** |
(−4.566) | (−2.536) | (−5.649) | |
Observations | 287 | 92 | 195 |
R-squared | 0.356 | 0.392 | 0.329 |
of the other types of jobs associated with both public and private services in Sweden.
Our model indicates that startups in these service sectors affect employment growth in all municipalities (see Table 3.10). The population model indicates that these startups may also have a slight positive effect on population growth in Swedish municipalities (coefficient 4.52e–05, p < 0.05; R-squared 0.623) and in rural areas (coefficient 6.89e–05, p < 0.001; R-squared 0.420).
When considering all Swedish municipalities, startups in all non-farm sectors except construction and trade, hotels and restaurants had a net positive effect on employment growth. Startups also had a positive effect on population growth in four of six sectors. Where the models indicate significant effects, the magnitude of employment effects is higher than population growth effects for all sectors. It is hardly surprising that employment effects are larger in magnitude than population effects. Employment change is a direct effect of startups, and employment is also affected through the indirect supply-side effects that startups bring. Population change is in this perspective an effect of employment change, but not necessarily a proportional effect, as an increase in local employment can happen by people already living in the municipality, for example the unemployed and students. The fact that population and employment in certain cases seem to grow at a similar
Variables | All municipalities |
Urban municipalities |
Rural municipalities |
Startups 2002–08 | 14.47*** | 26.50*** | 1.423*** |
(4.460) | (3.422) | (2.631) | |
Share university educated | 318.5 | −7.103 | 103.2 |
(0.130) | (−1.240) | (0.212) | |
Share of labor force employed | −4,153** | −3.874 | 376.2 |
(−2.138) | (−0.708) | (1.169) | |
Intra-municipal accessibility | 563.8*** | 1,012*** | 88.66*** |
(6.531) | (4.535) | (5.517) | |
Inter-municipal accessibility | −10.01 | 83.28 | −13.63 |
(−0.132) | (0.414) | (−1.197) | |
Mining dummy | 111.5 | 366.4*** | |
(0.128) | (3.556) | ||
Tourism dummy | −148.0 | 1.241 | −30.95 |
(−0.501) | (0.861) | (−0.819) | |
Constant | −10.044*** | −22.183*** | −1.733*** |
(−4.980) | (−2.946) | (−5.696) | |
Observations | 287 | 92 | 195 |
R-squared | 0.346 | 0.425 | 0.337 |
Branch | Positive effect on employment growth in all municipalities (p < 0.01)? | Municipal type where effect is most evident? | Positive and effect on population growth in all municipalities (p < 0.01)? | Municipal type where effect is most evident? |
Manufacturing | Yes | Urban*** | No | |
Construction | No | Rural** | Yes | Rural*** |
Trade, hotels and restaurants | No | Urban* | Yes | – |
Transportation and communications | Yes | – | Yes | – |
Financial/business services | Yes | Urban** | No | Rural*** |
Education, health/medical service, other public and personal service | Yes | Urban*** | Yes | – |
Rural*** |
rate may reflect an in-migration effect, as many entrepreneurs “bring their jobs with them” to new areas but then remain small or grow slowly.
However, when this data was analyzed by municipal type, some important differences were revealed. In urban areas, startups in all branches except construction and transportation/communications were highly significant (p < 0.05) in explaining employment growth (hotels/restaurants were significant in urban areas at the p < 0.1 level). In urban areas, none of the branch groups had significant population effects. In rural areas, the construction and education/health services branches had the only significant employment effect. In rural areas, startups in construction and financial/business services were significant in explaining population growth (see Table 3.11).
This study has focused on the effect of startups on key development goals such as employment and population growth. The study was based on four hypotheses. The first one, that startups over a time period of about 6–8 years had a positive impact on local employment and population, was clearly confirmed. The second hypothesis, that startups' impact on employment growth should be stronger than impacts on population growth was also confirmed. We found considerable evidence that startups affected employment growth positively at the municipal level, and some, smaller evidence that population also was affected.
The third hypothesis was that the effects of startups on employment and population were stronger for startups in the service sectors. This was not confirmed. The coefficient for startups in manufacturing effect on employment growth was the second highest among the six branch groups, when all municipalities were included and when the urban municipalities were analyzed. Even if most new employment occurs in the service sectors, startups in manufacturing have a strong impact on employment growth.
The final hypothesis suggested that the effects of startups would vary by municipality type and that the effects, in line with earlier research, should be stronger in urban areas. Neither point in this hypothesis was confirmed. Analyzing the data by municipal type and by branch revealed some important differences in the ways in which startups affect municipal development. Perhaps most interesting was the indication that the marginal effects of entrepreneurship in several branch groups appear to affect employment growth positively in rural areas. In contrast to previous research, our results indicate that for several branch groups startup effects on growth may be more pronounced in low density areas than in urban agglomerations. This is good news for development specialists frustrated by the seemingly intractable challenges of development outside of metropolitan and urban areas. Entrepreneurship may have an important role to play in restructuring areas dominated by sectors with declining employment trends.
1 The work with this chapter has partly been financed by the research council, Formas, grant No 251 – 2007–2038.
2 Some of these studies used the share of self-employed without employees as a measure of entrepreneurship, assuming that this would be an approximate value of the share of new firms. However, the correlation between firms per capita 2000 and startups 2002–08 in the Swedish municipalities was not higher than 0.31.
3 The analyses are performed with all municipalities and with the municipalities divided into two types. We use the division elaborated by economists at the Swedish Board of Agriculture, according to which the municipalities are classified into four different groups: municipality type (MT) 1, 2, 3, and 4. (MT 1) metropolitan areas (N = 46), (MT 2) urban areas (N = 47), (MT 3) rural areas/countryside (N = 164), and (MT 4) sparse populated rural areas (N = 33). The four types of areas are defined as follows: Metropolitan areas (MT 1): Includes municipalities where 100 percent of the population lives within cities or within a 30 km distance from the cities. Using this definition, there are three metropolitan areas in Sweden: the Stockholm, Gothenburg, and Malmöregions. Urban areas (MT 2): Municipalities with a population of at least 30,000 inhabitants and where the largest city has a population of 25,000 people or more. Smaller municipalities that are neighbors to these urban municipalities will be included in a local urban area if more than 50 percent of the labor force in the smaller municipality commutes to a neighbor municipality. In this way, a functional-region perspective is adopted. In practice, this group contains regional centers outside the metropolitan areas and their “suburb municipalities.” Rural areas/countryside (MT 3): Municipalities that are not included in the metropolitan areas and urban areas are classified as rural areas/countryside, given they have a population density of at least five people per square kilometer. Sparse populated rural areas (MT 4): Municipalities that are not included in the three categories above and have fewer than five people per square kilometer.
On account of the relatively small number of municipalities in MT 1, 2, and 4, we merge MT 1 and 2 into one metropolitan/urban group and MT 3 and 4 to a rural group.
4 The accessibility measure used is the product of three market potential measures, each discounted by time travelling distances. The three components are local, intra-regional, and inter-regional accessibility:
where each municipality is situated in one of Sweden's 81 functional regions (R), and where time-distances tii, tir, and tir are measuring average commuting times within each municipality, within regions, and outside of regions, respectively. The distance-decay parameter λ is based on commuting flows and is estimated in Johansson et al. (2003). The measure represents a continuous view of geography, and apart from capturing market potential originating outside of each municipality, it also alleviates the problems involved with using observational units of different sizes.
5 The mining dummy was given to five municipalities (Askersund, Hedemora, Skellefteå, Gällivare, and Kiruna). Twenty-three municipalities had an employment over 10 percent of total employment in tourism activities (hotels and restaurants) and were given the tourism dummy.
Acs, Z. J. and Armington, C. (2004), “Employment Growth and Entrepreneurial Activity in Cities,” Regional Studies, 38: 911–27.
Acs, Z. J. and Mueller, P. (2008), “Employment Effects of Business Dynamics: Mize, Gazelles and Elephants,” Small Business Economics, 30: 85–100.
Andersson, M. and Klaesson, J. (2009), “Regional Interaction and Economic Diversity: Exploring the Role of Geographically Overlapping Markets for a Municipality's Diversity in Retail and Durables,” in C. Karlsson, B. Johansson, and R. R. Stough (Eds), Innovation, Agglomeration and Regional Competition, Cheltenham, UK: Edward Elgar.
Andersson, M. and Noseleit, F. (2011), “Start-ups and Employment Dynamics Within and Across Sectors,” Small Business Economics, 36 (4): 461–83.
Audretsch, D. B. and Fritsch, M. (1996), “Creative Destruction: Turbulence And Economic Growth,” in Helmstädter, E. and Perlman, M. (Eds), Behavioral Norms, Technological Progress, and Economic Dynamics: Studies in Schumpetarian Economics, Ann Arbor: University of Michigan Press: 137–50.
Audretsch, D. B. and Fritsch, M. (2002), “Growth Regimes Over Time and Space,” Regional Studies, 36: 113–24.
Borgman, B. and Braunerhjelm, P. (2007), “Entrepreneurship and Local Growth: A Comparison of the US and Sweden,” CESIS WP 103, Stockholm: Royal Institute of Technology.
Braunerhjelm, P. and Borgman, B. (2004), “Geographical Concentration, Entrepreneurship, and Regional Growth: Evidence from Regional Data in Sweden 1975–1999,” Regional Studies, 38: 929–47.
Carree, M. A. and Thurik, A. R. (2003), “The Impact of Entrepreneurship on Economic Growth,” in Acs, Z. and Audretsch, D. (Eds), International Handbook of Entrepreneurship Research, London: Kluwer Academic Publishers: 437–71.
Davidsson, P., Lindmark, L., and Olofsson, C. (1994,) “New Firm Formation and Regional Development in Sweden,” Regional Studies, 28: 395–410.
EIM (1994), Kleinschalig Ondernehmen 1994, Vol II: Regionalekonomische Dynameik en Werkgelegenheidscreatie, Zoetemeer: Small Business Research and Consultancy.
Fölster, S. (2000), “Do Entrepreneurs Create Jobs?,” Small Business Economics, 14: 137–48.
Fritsch, M. (1996), “Turbulence and Growth in West Germany: A Comparison of Evidence by Regions and Industries,” reviews in Industrial Organisation, 11: 231–51.
Fritsch, M. (1997), “New Firms and Regional Employment Change,” Small Business Economics, 9: 437–48.
Fritsch, M. and Mueller, P. (2004), “Effects of New Business Formation on Regional Development over Time,” Regional Studies, 38 (8): 961–75.
Fritsch, M. and Mueller, P. (2008), “The Effects of New Business Formation on Regional Development Over Time: The Case of Germany,” Small Business Economics, 30: 15–29.
Fritsch, M. and Schroeter, A. (2011), “Why Does the Effect of New Business Formation Differ Across Regions?,” Small Business Economics, 36 (4): 383–400.
Gries, T. and Naudé, W. (2008), Entrepreneurship and Regional Economic Growth: Towards a General Theory of Start-Ups, Helsinki: United Nations University UNUWIDER, Research Paper No. 2008/70.
Johansson, B., Klaesson, J., and Olsson, M. (2003), “Commuters' Non-Linear Response to Time Distances,” Journal of Geographical Systems, 5 (3): 315–29.
Johnson, P. S. (1983), “New Manufacturing Firms in the UK Regions,” Scottish Journal of Political Economy, 30 (1): 75–9.
Rader Olsson, A. and Westlund, H. (2011), “Measuring Political Entrepreneurship: An Empirical Study Of Swedish Municipalities,” paper presented at the fiftieth, golden anniversary, meeting of the Western Regional Science Association, Monterey, California, February 27–March 2, 2011.
Reynolds, P.D. (1999), “Creative Destruction: Source of Symptom Of Economic Growth?,” in Acs, Z. J., Carlsson, B., and Karlsson, C. (Eds), “Entrepreneurship, Small and Medium-Sized Enterprises and TheMacroeconomy,” Cambridge: Cambridge University Press: 97–136.
Storey, D. J. and Johnson, S. (1987), “Regional Variations in Entrepreneurship in the UK,” Scottish Journal of Political Economy, 34 (2): 161–73.
Wennekers, S. and Thurik, R. (1999), “Linking Entrepreneurship and Economic Growth,” Small Business Economics, 13: 27–55.
Westlund, H. (2010), “Multi-Dimensional Entrepreneurship: Theoretical Considerations and Swedish Empirics,” Paper Presented at the International Workshop on Modeling Innovation, Entrepreneurship and Regional Development, 17–18 May, 2010, Tinbergen Institute and VU University, Amsterdam, and at the European Regional Science Association's 50th Congress in Jönköping, Sweden, 19–23 August, 2010 Available at: http://www.kolada.se (accessed March 23, 2011).
Wood P. A., Bryson J., and Keeble, D. (1993), “Regional Patterns of Small Firm Development in the Business Services: Evidence from the United Kingdom,” Environment and Planning A, 25: 677–700.
13.59.232.9