REGIONAL DIVERSITY AND HIGH-GROWTH ENTREPRENEURSHIP FREDRIK ANDERSSON STATISTICS SWEDEN NEDIM EFENDIC STOCKHOLM SCHOOL OF ECONOMICS KARL WENNBERG STOCKHOLM SCHOOL OF ECONOMICS & RATIO BACKGROUND • While unevenly distributed, we know that High-Growth Firms (HGFs) exist in all industries and all regions (Delmar, Davidsson & Gartner, 2003) • Yet, research has merely began to grapple with regional factors that may correlate with the emergence of HGFs (Teruel & de Wit, 2011) • Immigrants tend to start more companies than natives (Dana, 2007) – but whether these firms grow or not contingent on several other factors (Hart & Acs, 2011) THEORY AND PURPOSE • Theories of regional science suggests that regional diversity, both in economic and non-economic terms, is conducive for economic growth (Quigley, 1998) • While migration has been shown to facilitate firm formation, we do not know what type of firms (low-growth, high-growth) (Levie, 2007; Pennings, 1982) (1) How do diversity in income, ethnicity, and education shape the number of HGFs in a region? Purpose (2) Do the same factors affect the likelihood that individual firms within a region will become an HGF? DATA • Matched employee-employer data from Statistics Sweden • Sample: All Swedish incorporated firms 2004-2009: – Fewer than 100 employees in 2004 – No public involvement – At least one employee in addition to founder N=43,199 • Analyses: (1) Region-level analyses on # of HGFs (2) Firm-Level analyses on probability of becoming an HGF (same predictors as in analysis 1) MEASURING GROWTH • Gibrat-type regression (size-independent growth) where growth is a relative measure: (Sorenson, 2003; Delmar & Wennberg, 2010) S t+1/St= Sty exp (βΧt + ε) Where: Xit= predictor variable X at time t S = Size in turnover at time t y = firm’s relative size to the industry (MES) ε = error term HIGH-GROWTH FIRMS IN A REGION AND INDUSTRY Region level: Panel models on % HGFs in each municipality Firm level: Growth among all firms… Firm level: Quantile regression on each ”snippet” of growth… Firm level: Logit models on 10% most rapidly growing firms (Stam & Wennberg, 2009) 6,000 5,000 4,000 3,000 2,000 1,000 0 -… -79% -58% -37% -16% 5% 26% 47% 68% 89% 110% 130% 150% 169% 190% 209% 229% 250% 272% 293% 314% 335% 355% 384% 410% 450% 488% 531% 589% 650% 716% 826% 108… 152… 236… 345… 1. 2. 3. 4. FIRMS AND CEO BACKGROUND: DESCRIPTIVES 2005 2006 2007 2008 2009 Native Swedish 90,1% 90,4% 91,1% 90,9% 90,1% Immigrant 7,1% 7,4% 6,9% 6,6% 7,8% Second generation immigrant 2,7% 2,2% 1,9% 2,3% 2,0% REGIONS AND HGFS: DESCRIPTIVES “Star Gazelles” - 1995-2002 Industry Motor vehicle manufacture Investigation and security Secretarial and translation Private Health Care Software consultancy Software consultancy Data processing Municipality Employees Stockholm 702 Stockholm 962 Stockholm 1414 Malmö 1847 Göteborg 217 Uddevalla 142 Karlskoga 143 Debt collecting / credit rating Stockholm Engineering consultancy Västerås Software consultancy Motala Source: Fredric Delmar and Karl Wennberg (2010) 180 152 172 Turnover (mil.€) 176 57 52 44 19 15 11 11 9 9 REGIONAL-LEVEL PREDICTORS Variable Definition Gini coefficient Diversity in income in a focial municipality, where 1=toally unequal and 0=totally equal Median Income Median Income^2 Median income in a focial municipality, and it’s squared term (Davidsson et al. 1994) % first generation immigrants % of municipality residents born abroad whose parents were also born abroad % second generation immigrants % of municipality residents born in Sweden whose parents were also born abroad ln(inhabitants) Inhabitants in municipality (natural log) (Braunerhjelm & Borgman 2004) % employees in service sector Share of individuals employed in the service industry in relation to the municipality’s population (Fritsch and Falck, 2007; Braunerhjelm and Borgmann, 2004; Van Stel and Storey, 2004) % employed Share of individuals in municipality with paid employment % post-secondary education Share of individuals in municipality with 3-year or longer College Degree Page 9 REGIONAL-LEVEL ANALYSIS OLS (pooled) Variables: ln(inhabitants) % HGFs in Swedish Municipalities, 2005-2008 % employees in service sector % employed 0.35** (2.145) -0.0031 (-0.202) -0.032*** (-3.023) % post-secondary education Gini coefficient % HGFs= #HGFs / all firms Median Income Median Income^2 % immigrants % 2nd generation immigrants Constant Observations R2 11.8*** (2.745) 1,160 0.119 -0.21 (-0.998) -0.016 (-0.968) -0.022*** (-2.832) 0.060 (1.479) 0.10* (1.792) 0.22* (1.909) -0.001* (-1.886) -0.071 (-1.382) 0.56*** (2.614) -10.5 (-0.796) 1,160 0.155 Random Effects -0.20 (-0.937) -0.013 (-0.747) -0.023*** (-2.832) 0.064 (1.567) 0.094* (1.709) 0.21* (1.879) -0.001* (-1.876) -0.079 (-1.524) 0.59*** (2.667) -9.84 (-0.757) 1,160 0.167 Fixed Effects -6.35 (-0.479) 0.29** (2.534) -0.045 (-0.372) -0.54 (-1.386) 0.074 (0.821) 0.14 (0.840) -0.001 (-0.948) -0.76 (-1.822) 2.99* (1.951) 38.0 (0.304) 1,160 0.175 FIRM-LEVEL VARIABLES Variable Definition First generation immigrant Born abroad, parents born abroad Second generation immigrant Born in Sweden, parents born abroad Post-High School Education Post-High School Education, 3 years or more Firm’s relative size Number of employees / average number of employees in industry Lagged DV (endogeneity control) Size at t-1 Industry controls SIC-2 equivalent Firm-level Analysis DESCRIPTIVE RESULTS Mean growth Mean turnover 14,000 25% Native CEO Thousand SEK 12,000 20% 10,000 15% 8,000 6,000 Second generation immigrant CEO 10% 4,000 5% 2,000 0 2004 • First generation immigrant CEO 2005 2006 2007 2008 2009 Firms with Native CEO 1020% larger 0% 2005 2006 2007 2008 2009 Firms with Native CEO grows slower (”catching up effect”) FIRM-LEVEL ANALYSIS: LOGIT ON BECOMING HGF Variables: CEO: Post-High School Education CEO: Women Firm-level Variables Firm’s relative size First generation immigrant Second generation immigrant Firm Age Gini coefficient Region-level Variables Median Income Median Income ² % post-secondary education OECD Worskhop Other controls: Industry Controls: Observations 2 Logit (1) 0.18*** (5.284) -0.36*** (-11.46) 0.099*** (14.361) 0.055 Logit (2) 0.18*** (5.262) -0.36*** (-11.46) 0.099*** (14.33) 0.055 (1.043) (1.046) 0.26*** (5.145) -0.005*** (-4.990) 0.014*** (2.601) 0.002* (1.724) -0.001* (-1.924) -0.0020 (-0.477) no yes 181,380 0.26*** (5.150) -0.005*** (-5.132) 0.017*** (3.054) 0.002 (1.601) -0.001** (-2.273) 0.003 (0.679) yes yes 181,380 RESULTS • Regional factors exhibit a strong influence on the emergence of HGFs • Diversity in incomes (gini) as wells as ethnicity (second generation immigrants) positively associated with %HGFs in a region • The same region-level factors also affect the growth of individual firms CONCLUSIONS AND FURTHER WORK • Results indicate the regional aspects of HGFs is an underexplored are requiring more empirical and theoretical work • Regional characteristics (e.g. diversity) important both for firm dynamics in the region and for the growth chances of the individual firm • Analysis of growth patterns of individual firms also need to consider geographic factors (e.g. multi-level analysis) • Current definition of HGFs limited to firms at the top of the cross-sectional growth rate distributions additional analyses needed to distinguish between ”persistent HGFs” (3 year+) and ”temporary growth firms” QUESTIONS, COMMENTS, CRITICISM? Thank you! WHAT INDUSTRIES DO NON-NATIVE CEOS START BUSINESSES IN? • • First generation immigrants often start firms in ”personal services” (=restaurants) Industry Second gen. immgrant Swedish 1,7% 4,2% 12,1% 15,2% 1 – Forestry and Agriculture 2 – Manufacturing Second generation 3 – Energy, water, and waste immigrants often start in 4 - Construction oftare in ”Financial Services and Consulting” 5 – Trade and communication 6 – Financial Services and as well as in “Education Consulting and Research” 7 – Education and Research 8 – Health Care 9 – Personal Services Totalt: First gen. immgrant 1,2% 12,4% 0% 12,8% 33,7% 0,3% 17,1% 36,0% 0,1% 8,1% 28,8% 20,5% 1,7% 4,8% 12,6% 16,8% 1,3% 3,5% 5,6% 16,3% 1,5% 7,9% 23,6% 4 029 208 685 15 826