CENTER FOR INNOVATION, RESEARCH AND COMPETENCE IN THE LEARNING ECONOMY Local Clusters of Entrepreneurs - neighborhood peers effects in entrepreneurship Martin Andersson* and Johan P Larsson** *CIRCLE, Lund University and Blekinge Institute of Technology (BTH) **CEnSE, Jönköping International Business School (JIBS) CIRCLE, Lund University, Sweden Our question • Does living in a neighborhood where many residents are established entrepreneurs induce entrepreneurial behavior? • Local social interactions: individuals’ behavior depend on the behavior of others in their environment (Glaeser and Scheinkman 2003) • Peer/network effects in entrepreneurship (Minniti 2005, Nanda and Sorensen 2010, Bosma et al 2012) Our finding: • Yes! – Local clusters of entrepreneurs at the neighborhood level – The fraction of neigborhood residents that are established entrepreneurs has an economic significant and robust effect on the probability that other residents transcend from employment to entrepreneurship CIRCLE, Lund University, Sweden Motivation and background • ’Local peer effects in entrepreneurship an ”old” question: – Social dimension of the decision to become an entrepreneur (Shapero and Sokol 1982, Aldrich 2005, Licht and Siegel 2006, Nanda and Sorensen 2010), Giannetti and Simonov 2009). • … but it is an important one – Geography of growth = f(geography of entrepreneurship) – Persistent local clusters of entrepreneurship (cf. Fritsch and Whywhich 2013, Anderson and Koster 2011) CIRCLE, Lund University, Sweden .02 0 .01 Start_up_rate .03 .04 PERSISTENCE OF REGIONAL START-UP RATES .01 .02 L20.Start_up_rate .03 .04 Figure 2. The relationship between start-up rates in 2007 (Start_up_rate) and in 1987 (L20.Start_up_rate) across Swedish municipalities (new establishments per inhabitant 16-64 years of age). CIRCLE, Lund University, Sweden Motivation and background • Peer effects put forth as an explanation of the evolution and persistence of entrepreneurship clusters: Minniti (2005): – “relatively simple assumptions about peer effects and learning behavior suffices to produce distinct local clusters of entrepreneurial activity”. • Anna Lee Saxenian: – Maintained that one important explanation for the divergent performance of Silicon Valley (California) and Route 128 (Boston) is rooted in differences in regional entrepreneurship culture “In Boston, if I said I was starting a company, people would look at me and say: ‘Are you sure you want to take the risk? You are so well established. Why would you give up a good job as vice president at a big company?’ In California, I became a folk hero when I decided to start a company. It wasn't just my colleagues. My insurance man, my water deliverer – everyone was excited. It’s a different culture out here.” – Social interactions and peer effects one way in which ”culture” persists and transfers between individuals in a locality. CIRCLE, Lund University, Sweden Motivation and background • Policy relevant: social multiplier (Glaeser et al 2003): – an exogenous shock induces not only a direct effect on individual behavior, but also an indirect effect mediated by people adopting the behavior of their peers. – Potential for long-term policy effect CIRCLE, Lund University, Sweden Motivation and background • Empirical evidence of local peer effects in entrepreneurship is still limited: – Survey-based evidence uninformative as regards the magnitude of the peer effects in quantitative terms (relative to other explanations), and few studies link local peer effects to geographic outcomes. • Regional analyses of persistence of entrepreneurship – Peer effects often cited as an explanation, but: » Political AND social dimension of ’culture’ (separation difficult) » Regions NOT homogeneous to several important fundamentals – Identification issue: • Manski’s (2000) reflection problem: separating the effects of the behavior of peers on individual behavior from the effects of spatial sorting CIRCLE, Lund University, Sweden Our contribution – Focus within-city clusters of entrepreneurs neigborhoods (1 square kilometer) • (1) Comes much closer to the conceptual notion of a neighborhood as an arena for social interactions. • corresponds to established findings of the distance-decay of inter-personal contacts. 42% of frequent contacts occur between individuals who lives less than 1 mile apart (Wellman 1996). • (2) Identification: neighborhoods homogeneous with regard to any determinant operating at the city (or municipality) level spatial differentiation in outcomes in the absence of differences in fundamentals is a key feature of any model of social interactions (Glaeser and Scheinkman 2003, Minniti 2005) CIRCLE, Lund University, Sweden #1: EMPIRICAL REGULARITY clusters of entrepreneurs across neighborhoods within regions CIRCLE, Lund University, Sweden d within-city neighborhood clusters of entrepreneurs Figure 2. Distribution of entrepreneurs within the Stockholm metropolitan area (left), and the Jönköping urban region CIRCLE, Lund(right). University, Sweden within-city neighborhood clusters of entrepreneurs 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 0 200 400 Stockholm average 600 800 1000 Fraction entrepreneurs 0 20 40 60 Jonkoping average 80 100 120 140 Fraction entrepreneurs Figure 3. The fraction of entrepreneurs across neighborhoods in Stockholm (left) and Jönköping (right) CIRCLE, Lund University, Sweden • Patterns consistent with local social interactions • variance in entrepreneurship across neighborhoods within one and the same city cannot be explained by city-wide fundamentals, since those are shared by all neighborhoods in the city. – “Standard” supply- and demand-side determinants likely to operate at the city (or region) level » ex. local policy regime, market-size, labor supply CIRCLE, Lund University, Sweden #2: MICROECONOMETRIC ANALYSIS - Does living in a neigborhood where a large fraction of the residents are established entrepreneurs influence the probability of transcending from employment to entrepreneurship? CIRCLE, Lund University, Sweden – Individuals that become entrepreneurs – Extensive controls (individual, employer, geography) – Inclusion of municipality-specific effects • Parameters identified from variations across neighborhoods within cities – Isolate sub-groups to test of robustness of the results with regard to the underlying identifying assumption (Lindbeck et al 2007) • Age groups, immigrants, local market-dependent sectors CIRCLE, Lund University, Sweden IDENTIFICATION STRATEGY Pr Ei ,t 1 x i,t 1 xi,t 1 Γ (1) xi,t 1 Γ Ii,t 1β Zi,t 1 γ Ωi,t 1θ Ri,t 1σ i,t Individual Employer Neighborhood Region • Leave full-time employment for full-time entrepreneurship. • All employees in 2007 (N= about 2.7 million) • Full population matched employer-employee dataset for Sweden CIRCLE, Lund University, Sweden Table 4. Determinants of leaving employment for entrepreneurship. Variable Fraction entrepreneurs in the neighborhood Neighborhood density (ln) Human capital (neighborhood) Fraction entrepreneurs in the municipality Size of municipality (ln) Stockholm (dummy) Years of schooling Tenure Wage (ln) Establishment exit Establishment employment size (ln) Age (ln) Age squared (ln) Male (dummy) Immigrant (dummy) 0.0323*** (0.00137) -0.00994*** (0.00309) 0.451*** (0.0359) 0.00717*** (0.00273) 0.00511 (0.00326) 0.0753*** (0.0103) 0.0137*** (0.00170) -0.00931*** (0.000669) -0.232*** (0.00373) 0.114*** (0.0139) -0.120*** (0.00321) 4.679*** (0.350) -0.598*** (0.0470) 0.335*** (0.00748) 0.00496 (0.00813) Observations 2,735,407 Pseudo R-squared .146 Note: The table report coefficient of the model in (1) using a Probit estimator. The underlying data is a matched CIRCLE,employer-employee Lund University, dataset Sweden for Sweden for the years 2007, covering all employees in the age interval 25- Issues • Sorting? – Individuals move to certain neighborhoods once the decision to start a firm is taken. • Migration of entrepreneurs and employees • Immigrants • Additional controls • Start-ups with very local market (neighborhood)? • Split by sectors (cafés, hairdressers etc.) • Driven by agglommerated areas? • Sample split (cities // countryside) • Entrepreneurship/self-employment? • incorporated business / self-proprietorship • Artifact of age composition in neighborhood? • Estimations by age groups CIRCLE, Lund University, Sweden Mobility of entrepreneurs and non-entrepreneurs across neighborhoods 30% 25% 20% 15% 10% 5% 0% 1 2 3 Entrepreneurs CIRCLE, Lund University, Sweden 4 Non-entrepreneurs 5 17+ Test #1 Immigrants, neighborhood tenure, agglomeration and ”local-demand” sectors Table 7. Estimated effects of neighborhood’s fraction of entrepreneurs on the decision to enter entrepreneurship, by different sub-groups. (1) (2) (3) (4) (5) (6) Excluding local demand driven sectors and retail Excluding local demand driven sectors 50% least dense neighborhoods Neighborhood tenure ≤ 2 years Neighborhood tenure ≤ 5 years Immigrants arrived after 2002 Immigrants only Selection (7) Fraction entrepreneurs .0423*** .0381*** .0376*** .0360*** .0260*** .0382*** .0380*** in the neighborhood (0.00329) (.00985) (.00287) (.00197) (.00161) (.00123) (.00128) Average marginal effect .0007 .0007 .0007 .0007 .0005 .0006 .0006 N 437,844 40,458 575,588 1,144,774 1,333,577 2,718,206 2,716,150 Note: The model is identical to (1), the coefficients of which are presented in Table 4. Robust standard errors are presented in parentheses. *** p < .01. Local demand driven sectors are defined as NACE 93, including restaurants, and NACE 82, including hair dressers and beauty salons. CIRCLE, Lund University, Sweden Test #2 Additional neighborhood controls (4) All of (1)-(3) Region human capital Neighborhood mean wage (ln) Added control(s) (3) Neighborhood fraction entrepreneurs 1991 Table 6. Sensitivity analysis of the main specification in Table 4. (1) (2) Fraction entrepreneurs .0302*** .0343*** .0328*** .0296*** in the neighborhood (.00138) (.00125) (.00131) (.00142) Average marginal effect .0005 .0006 .0006 .0005 Note: Aside from the added control variables the model is identical to (1), the coefficients of which are presented in Table 4. Robust standard errors are presented in parentheses. *** p < .01. CIRCLE, Lund University, Sweden Test #3 Split by start-up type Table 5. Estimated effects of neighborhood’s fraction of entrepreneurs on the decision to enter entrepreneurship by start-up type. (2) Startup of incorporated business Startup of sole proprietorship Start-up type (1) Fraction entrepreneurs .0352*** .0312 *** in the neighborhood (.00138) (.00202) Average marginal effect .0004 .0002 Note: Aside from the added control variables the model is identical to (1), the coefficients of which are presented in Table 4. Robust standard errors are presented in parentheses. *** p < .01. Out of all startups, 61 percent are sole proprietorships, and 39 percent are incorporated businesses. CIRCLE, Lund University, Sweden Test #4 Estimations by age intervals Table 8. Estimated effects of neighborhood’s fraction of entrepreneurs on the decision to enter entrepreneurship, by age interval. (1) (2) (3) Age 56-64 Age 36-55 Age 25-35 Age interval Fraction entrepreneurs in the .0266*** .0303 *** .0312*** neighborhood (0.00299) (.00183) (.00311) Average marginal effect .0004 .0005 .0005 N 745,201 1,446,622 525,481 Note: The model is identical to (1), the coefficients of which are presented in Table 4. Robust standard errors are presented in parentheses. *** p < .01. CIRCLE, Lund University, Sweden CONCLUSIONS • Local social interactions (or peer effects) in entrepreneurship may explain persistent local clusters of entrepreneurship – emphasized in theoretical work on the emergence and evolution of clusters (eg. Minniti 2005) as well as in cluster case studies, such as in Saxenian’s (1994) work on the strengths of the Silicon Valley region. • Imply potentially large policy effects (social multiplier): – direct effects amplified by peer effects CIRCLE, Lund University, Sweden CONCLUSIONS • We employed geo-coded matched employer-employee data and showed: – (i) clusters of entrepreneurs at the neighborhood level within cities => consistent with local social interactions. – (ii) micro-econometric evidence of a significant feedback effect in which existing entrepreneurs in a neighborhood breeds new local entrepreneurs • Overall => consistent with local social interactions effects. • Social interactions appear as relevant in explaining the emergence and persistence of local clusters of entrepreneurs. • Provides an example of how characteristics of a local environment induce entrepreneurial behavior at the individual-level, that then feeds back on the environment (social multiplier) CIRCLE, Lund University, Sweden
© Copyright 2026 Paperzz