National Business Regulations and City Entrepreneurship in Europe: A Multilevel Nested Analysis

Abstract

This article provides and tests a theoretical framework with a multilevel (country–city) nested model to analyze the relationship between national business regulations (NBRs) and city level entrepreneurship. While public interest theory predicts a positive relationship between NBR and city level entrepreneurship, public choice theory predicts the opposite, a negative relationship. Based on multilevel analysis for a matched country–city panel of 228 cities across 20 European countries for the years 2004 to 2009, the empirical evidence from panel data estimation explains how changes in NBRs influence changes in city level entrepreneurial activity over time.

Keywords

entrepreneurship city government policy regulation Europe multilevel JEL: L26 M13 R10 O17

Research on the drivers of systematic variations in entrepreneurship can be grouped into two streams. The first is entrepreneurship and regional economic development (Audretsch, Belitski, & Desai, 2015; Bosma & Sternberg, 2014; Boschma & Fritsch, 2009), and the second is cross-country research on institutions and entrepreneurship (Acs, Desai, & Klapper, 2008a; Ardagna & Lusardi, 2010; Autio, Kenney, Mustar, Siegel, & Wright, 2014; Bruton, Ahlstrom, & Li, 2010; Spigel, 2017; Stenholm, Acs, & Wuebker, 2013; Van Stel, Storey, & Thurik, 2007).

Despite theoretical underpinnings on the importance of institutions for entrepreneurship (Baumol, 1990; North, 1990), little theoretical work asks how country level institutions impact entrepreneurship in cites (Audretsch & Belitski, 2017; Hundt & Sternberg, 2016). This represents a knowledge gap on the impact of the country context on entrepreneurship in subnational units, such as cities (Charron, Dijkstra, & Lapuente, 2014). City level studies on entrepreneurship and economic development often ignore the country context in which cities are embedded (Charron et al., 2014; Spigel, 2016, 2017), while country level studies on regulations and entrepreneurship suffer from potential loss of nuance due to aggregation. Although the entrepreneur is embedded in the city context (Autio et al., 2014; Fritsch & Storey, 2014), he or she also operates in the country (Autio et al., 2014; Braunerhjelm, Desai, & Eklund, 2015; Whitley, 1999). Thus, country level conditions like regulations are not optional: Entrepreneurs in all cities must comply.

To rectify this, we offer and test a theoretical framework with a multilevel (country–city) nested model linking national business regulations (NBRs) and city level entrepreneurship. We argue that public interest theory implies a positive relationship between NBR and city level entrepreneurship, whereas public choice theory implies the opposite. Rooted in these theories, we hypothesize that the relationship between NBR and city level entrepreneurship is an inverted U shape. We also argue that there is an optimum level of NBR for city entrepreneurship.

For the purposes of our study, we include four types of NBRs related to entry, property registration, taxes, and contract enforcement. For each, we use one measure for the procedures needed to meet regulatory requirements and one measure for the costs of meeting requirements. City level entrepreneurship reflects new ventures, between 1 and 3 years old, as a share of total businesses in a city. We test our inverted U-shaped hypothesis using multilevel analysis on a matched country–city panel of 228 cities across 20 European countries, over the period 2004 to 2009. We find partial support for our hypothesis. We also find that country level characteristics explain more of the systematic variation in city level entrepreneurship than city characteristics.

We use panel data techniques, which enable us to infer causality on how changes in NBR affect changes in city entrepreneurship over time. We advance cross-sectional research on regulation and entrepreneurship (Ardagna & Lusardi, 2009, 2010; Stenholm et al., 2013), and we extend past research, which usually assumes a linear negative impact of regulation on entrepreneurship (Braunerhjelm et al., 2015). Our study answers calls for multilevel research (Ding, Au, & Chiang, 2015; Hundt & Sternberg, 2016; Westlund, Larsson, & Olsson, 2014), insight on cross-national institutional variation (Welter, 2011), dynamics of entrepreneurs in broader context (Autio et al., 2014), and drivers of urban entrepreneurship (Glaeser, Rosenthal, & Strange, 2010b). Our results can inform location choices of entrepreneurs considering city or country factors and can help policy optimize the impact of NBR on entrepreneurship in cities.

Theoretical Framework

Entrepreneurship in Cities

Research on entrepreneurship in cities has benefited from insights from several fields, including regional economics and economic geography (Jacobs, 1970; Marshall, 1890), and has been tested in city level empirical studies. While studies have analyzed why entrepreneurial activity varies across cities (Armington & Acs, 2002; Belitski & Desai, 2015; Spigel, 2016; Reynolds, Storey, & Westhead, 1994), they have not considered the particular country context and country-specific characteristics. These are rooted in agglomeration economies, which present opportunities for entrepreneurs, tied to factors like clustering, financial services depth, and human capital (Armington & Acs, 2002).

Past research has found that local demand, market size (Boschma & Fritsch, 2009), and regional purchasing power (Bergmann & Sternberg, 2007; Reynolds et al., 1994) can attract entrepreneurs. It is reasonable to expect this differs across cities located within a country. Bigger cities offer large market opportunities despite possibly burdensome national regulations (i.e., profits larger than regulatory costs). Larger agglomeration economies offer spatial concentrations of human capital (Audretsch et al., 2015), which is conducive to potential entrepreneurs and new firms seeking employees (Boschma & Fritsch, 2009). Industry mix in a city matters (Belitski & Desai, 2015; Jacobs, 1970; Marshall, 1890) because some industries are more salient for entrepreneurs (see Davidsson, Lindmark, & Olofsson, 1994). For example, higher concentration of financial, professional, and business industries can boost entrepreneurship (Keeble & Walker, 1994).

Country Level Regulation and City Level Entrepreneurship

Research on regulation and entrepreneurship at the country level suggests that a functional regulatory environment incentivizes entrepreneurs (Acs et al., 2008a; Ardagna & Lusardi, 2010; Estrin, Korosteleva, & Mickiewicz, 2013; Klapper, Laeven, & Rajan, 2006; Stenholm et al., 2013). The nuances of the regulatory environment, however, are not well understood, such as which particular regulations matter for specific entrepreneurship outcomes (Audretsch et al., 2015).

The concept of a national business system (Whitley, 1999) contends that firms do not exist in a vacuum. Rather, they are economic units affected by the business environment. This is intuitive: Entrepreneurs operate in markets that are governed by laws and regulations. These are closely tied to local context, including institutional and geographic constraints. Whitley (1999) argues that some influences are geographically restricted to the home region of the firm, while another set of influences is linked to broader institutions. The national business systems concept (Whitley, 1999) highlights the need to consider links between the country and city levels. Consistent regulatory design governing all entities within a country reflects the national business system, which shapes entrepreneurship (see Whitley, 1999). At the same time, entrepreneurship occurs within a local context, with constraints like availability of workers and infrastructure. In this way, entrepreneurship is also an urban event (Bosma & Sternberg, 2014).

We therefore ask: How do NBRs influence city level entrepreneurship? Our nested multilevel model is shown in Figure 1 in the Appendix). Entrepreneurship occurs in a city (Feldman, 2001), embedded inside a country, and is thus subject to NBRs (Whitley, 1999). Our goal is to identify the size and direction of the relationship between heterogeneous NBRs and city level entrepreneurship, rather than analyze differences in national and city level entrepreneurship. We point out that country level studies capture net effects, but city level entrepreneurship needs deeper study of trends within and between countries, which can vary greatly (Charron et al., 2014). Prior comparative research on entrepreneurship in cities suffers from a key limitation: It has been largely done within, not between, countries. Most studies have been conducted within one region or country, for example, Germany (Fritsch & Mueller, 2008), the United States (Armington & Acs, 2002), and Scotland (Spigel, 2016). Few studies examine cities in several countries, despite the need for contextual insight (Belitski & Desai, 2015; Hundt & Sternberg, 2016). We rectify this by injecting the country level into the question of variations in city level entrepreneurship. We focus on NBR, which is consistent for all cities in a country.

National institutions can set rewards and expected payoffs for entrepreneurship (Baumol, 1990), and NBR represents a subset of formal institutions, which are explicit and codified, and govern businesses (Williamson, 2000). Entrepreneurs adapt to their institutional context (Van Stel et al., 2007) by responding to institutions governing the entry and early life of a firm (Acs et al., 2008a). Transactional trust (Fogel, Hawk, Morck, & Yeung, 2006) reflects the extent to which entrepreneurs trust regulations that affect firm transactions, such as how they raise capital, enter new markets, enforce contracts, and pay taxes (Djankov, La Porta, Lopez-de-Silanes, & Shleifer, 2002). Regulation is heterogeneous (Braunerhjelm et al., 2015; Estrin et al., 2013; Williamson, 2000) and can vary by type (e.g., entry regulation, property registration regulations), highlighting the need to consider that the relationship between regulations and entrepreneurship could manifest in different ways.

Two contrasting economic theories of regulation explain why we can expect a complex nonlinear relationship between NBR and entrepreneurship. Public interest theory (Pigou, 1938) posits that regulation is generally devised to protect the public at large (Hantke-Domas, 2003). Regulation is enacted by benevolent politicians in order to correct market failures, ensure the quality of products, and improve public welfare (Pigou, 1938). In this view, regulation helps the public by weeding out “bad” entrepreneurs, who would have wasted precious resources and failed or taken advantage of consumers, from “good” entrepreneurs, who create welfare gains. Regulations also protect entrepreneurs by ensuring things like access to working courts, clear knowledge about tax policies, and equitable access to rules and information. A public interest perspective suggests we should expect a positive relationship between NBR and city level entrepreneurship.

In contrast, public choice theory argues that regulation is used by powerful agents to improve their own economic gains (see Djankov et al., 2002; Peltzman, 1976; Tullock, 1967). Agents can be incumbent firms that pursue regulation to protect profits and engage in lobbying and regulatory capture to raise entry barriers and limit competition. Agents can also be politicians who enact regulation to create rents and govern the process of collecting them. More complex regulations can benefit rent-seeking politicians (De Soto, 1989) if they yield direct revenues and offer more opportunities to extract bribes (Belitski, Chowdhury, & Desai, 2016). This “politics without romance” view reflects a realistic picture of regulation (Eskridge, 1988: 276, quoting Buchanan), especially in some developing countries (Chowdhury, Desai, Audretsch, & Belitski, 2015). Access to political and judicial power can influence decisions; public officials with this power can design and enforce rules with discretion. This is similar to a private toll road (see Djankov et al., 2002), which all entrepreneurs must use, but where officials are toll collectors and enforcers. For example, if an entrepreneur must file six forms to pay taxes, this offers six opportunities for a government agent to extract payment. A public choice perspective suggests that we should expect a negative relationship between NBR and city level entrepreneurship.

Both theories offer insight into the origins of NBR, suggesting two dynamics: Effective, conducive NBR encourages entrepreneurship (public interest), but excessive, cumbersome NBR discourages entrepreneurship (public choice). These opposing effects could occur across different types of NBRs. For example, a public-serving regulator in a country’s land registration agency could enact conducive regulations, whereas a self-serving regulator in the same country’s national tax authority could enact cumbersome regulations. Also, job change and staffing replacements, such as during changes in government administration or even the natural career trajectories of individual regulators, can bring a significantly different approach and style of regulating within the same agency at different times. We expect that the relationship between NBR and city level entrepreneurship is likely to take the form of an inverted U shape. We thus hypothesize that:

Hypothesis: The relationship between national business regulations (NBRs) and city level entrepreneurship is an inverted U shape.

Data and Variables

Our matched sample is built from Eurostat’s European Urban Audit, the World Bank Doing Business database, and Transparency International’s Corruption Perceptions Index (CPI). The Urban Audit is a standardized database of European cities, which covers the period 1994 to 2009. It offers a rich range of economic, political, social, and cultural indicators for European cities, with a core city defined to contain at least 50,000 residents. All European capitals and major regional cities are included, so it is appropriate to study trends in cities over time and space. The Urban Audit is the source for our dependent variable and for many of our control variables.

The Doing Business database measures a wide range of NBRs beginning in 2003 and is standardized to be comparable across countries. Given that the Urban Audit is collected every 3 years (2000–2004, 2004–2006, 2007–2009; reporting years 2005, 2007, 2010), we take the equivalent years of Doing Business data for our explanatory variables capturing NBRs (2004, 2006, 2008). This links entrepreneurship data from the Urban Audit with regulatory data from Doing Business for the appropriate periods. Our final sample is longitudinal data comprising 228 cities in 20 European countries and 496 observations, over the years 2004 to 2009. Table 1 contains correlations and descriptive statistics for our variables.

Table 1.

Correlation Matrix.

	Mean	SD	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18
1. City entrepreneurship	14.64	9.95	1.00
2. Entry procedures	7.52	2.23	0.09*	1.00
3. Property registration procedures	5.77	1.80	−0.14*	−0.10*	1.00
4. Paying taxes procedures	24.68	25.08	−0.09	−0.04	0.38*	1.00
5. Enforcing contracts procedures	31.99	3.95	−0.22*	0.50*	−0.17*	0.16*	1.00
6. Entry cost	9.32	8.12	−0.28*	0.46*	−0.21*	0.14*	0.64*	1.00
7. Property registration cost	4.69	3.07	−0.08	−0.34*	0.00	−0.46*	−0.23*	0.03	1.00
8. Paying taxes cost	50.43	8.78	−0.02	−0.17*	0.27*	−0.01	−0.07	−0.09	0.41*	1.00
9. Enforcing contracts cost	17.95	5.07	−0.09*	−0.40*	−0.18*	−0.06	−0.45*	−0.43*	−0.02	−0.04	1.00
10. CPI	6.36	1.94	0.23*	−0.36*	−0.25*	−0.67*	−0.56*	−0.50*	0.40*	−0.01	0.25*	1.00
11. GDP	9.84	0.57	0.31*	−0.17*	−0.14*	−0.58*	−0.49*	−0.43*	0.31*	0.00	0.09*	0.69*	1.00
12. Density	7.27	0.85	0.06	0.14*	0.14*	0.17*	−0.01	0.04	−0.10*	−0.17*	−0.04	−0.08	0.16*	1.00
13. Population at work	12.08	0.83	0.07	−0.04	0.10*	−0.05	−0.04	−0.10*	0.01	0.08	−0.01	0.09*	0.40*	0.55*	1.00
14. Agriculture	1.08	1.62	0.00	0.09*	−0.12*	0.02	0.28*	0.14*	0.19*	0.24*	0.00	−0.09*	−0.20*	−0.27*	−0.20*	1.00
15. Manufacturing	18.91	9.41	−0.13*	0.12*	0.02	0.57*	0.42*	0.39*	−0.27*	−0.04	−0.14*	−0.68*	−0.69*	−0.06	−0.30*	0.13*	1.00
16. Retail	18.30	3.61	0.03	0.10*	−0.14*	−0.17*	0.13*	−0.01	0.08	−0.13*	0.15*	0.01	−0.01	0.08*	0.08	0.17*	−0.01	1.00
17. Finance	15.69	7.06	0.26*	−0.15*	−0.07	−0.43*	−0.49*	−0.28*	0.19*	−0.16*	0.11*	0.63*	0.69*	0.27*	0.44*	−0.33*	−0.65*	−0.04	1.00
18. Public	33.43	7.35	0.00	−0.25*	0.04	−0.34*	−0.36*	−0.13*	0.39*	0.28*	−0.06	0.41*	0.24*	−0.28*	−0.15*	−0.09*	−0.46*	−0.25*	0.12*	1.00
19. ICT	1.01	1.68	0.11*	−0.02	−0.21*	−0.17*	−0.03	−0.05	0.08	0.09*	−0.03	0.03	0.05	−0.06	−0.04	0.03	0.20*	0.10*	0.00	−0.05

Note. The matrix is calculated using the full sample of 496 observations. (228 cities, 20 countries). Level of statistical significance is * 0.05%. Correlations for the 12 city typologies are suppressed to save space. CPI = Corruption Perceptions Index; GDP = gross domestic product.

Source: UAP = Urban Audit Project (Eurostat Statistics on European cities: 2004–2009); DB = World Bank's Doing Business Data: 2004–2009; TI = Transparency International (2012).

Dependent Variable

City level entrepreneurship is our dependent variable, measured as new firm formation in a city, and calculated using the ratio of new business ventures between 1 and 3 years old to the total number of existing businesses. The data come from the Urban Audit and include all periods in our study. A graphical depiction of our dependent variable can be found in Figure 2 in the Appendix.

Explanatory Variables

Our explanatory variables capture NBR and are all taken from the Doing Business data. Specifically, we focus on four aspects of NBR, related to entry, property registration, taxes, and contract enforcement. We use two dimensions of each regulation: procedures, measured as the number of procedures needed to complete the regulatory requirements, and cost, measured as the financial costs associated with completing the requirements. For each of the NBR variables, we also take their squared terms. We enter the variables into one model and investigate them at the single variable level (not as an index). Although the four aspects of regulation do not cover all possible NBRs that can influence individual entrepreneurial behavior (e.g., tariffs and nontariff barriers, safety and environmental standards; Ardagna & Lusardi, 2010), they include some of the most important regulatory constraints across countries.

Entry regulation reflects the entry burden for an entrepreneur to incorporate a business. It is measured using the number of procedures required to start a local limited liability company and the cost of the process as percentage of income per capita. Prior research found that more entry regulation deters entrepreneurship (Ciccone & Papaioannou, 2007; Djankov et al., 2002, Djankov, Ganser, McLiesh, Ramalho, & Shleifer, 2009; Klapper et al., 2006). Entry regulation imposes costs early on, often before cash flows, so entrepreneurs with strong prospects may be willing to take on the costs. For this reason, some entrepreneurs may be willing to accept higher entry regulation, while others may not.

Registering property matters because secure property rights can create channels for collateralization (see De Soto, 1989; Williamson, 2000), which could improve access to finance for entrepreneurs. This is measured using the number of procedures required to register property and as the cost of complying with the regulations as a percentage of the property value. Transparent and clear property rights could reduce risks to lenders or investors (Estrin et al., 2013), and security of property can raise confidence about investing in assets like buildings (Desai, Acs, & Weitzel, 2013).

Tax regulation has been found to negatively affect entrepreneurship (Acs et al., 2008b). Entrepreneurs may see paying taxes as cutting into business resources (Baliamoune-Lutz & Garello, 2014). Procedures to pay taxes is measured as the number of required tax payments, and costs reflects the total tax rate as a percentage of profits. In a study of corporate income taxes in 85 countries, Djankov et al. (2009) found a negative impact on new business. However, it is also possible that “a benevolent social planner who wants to spend significant resources on screening new entrants may choose to finance such activity with broad taxes rather than with the direct fees” (Djankov et al., 2002: 8). Taxes provide public revenues to finance resources that can help entrepreneurs like schools and infrastructure. Tax morale and willingness of entrepreneurs to comply with regulations could be shaped by what they “get back” (Belitski et al., 2016).

Contract enforcement regulation can shape confidence about forming relationships (e.g., with suppliers) and the security of claims to receive returns (Johnson, McMillan, & Woodruff, 1999). It is measured as the number of procedures needed to enforce contracts and as the costs of enforcement as a percentage of the claim value. New firms and incumbents face the same costs to enforce contracts, but incumbents can exploit information asymmetries due to longer standing in the market and more opportunities to cheat (Estrin et al., 2013). If enforcement costs are low, entrepreneurs may conclude they will be able to enforce contracts. However, if costs are too low or it is too easy to go to court, entrepreneurs may fear vulnerability to arbitrary claims.

Control Variables

We use several city level controls from the Urban Audit Project to reflect local conditions related to industry structure, agglomeration economies, workforce, and corruption (Bosma & Sternberg, 2014; Fritsch & Mueller, 2008; Glaeser, Kerr, & Ponzetto, 2010a). Population at work (16–65 years old), taken in logarithm, is used to capture the level of employment (Estrin et al., 2013). We use population density in logarithm, to account for market size (Boschma & Fritsch, 2009) because regional purchasing power (Bergmann & Sternberg, 2007) can attract entrepreneurs. Higher population density could mean more market opportunities and access to higher concentration of skilled people (Audretsch et al., 2015; Reynolds et al., 1994).

Employment structure across city industry is measured using the European standard industry classification (NACE) to control for human capital returns to entrepreneurship across different industries (Belitski & Desai, 2015). We select six aggregated sectors, measured by proportion of employment per sector, including agriculture; manufacturing; retail; finance; public (administration, health care, education); and information and communication technology (ICT). The six sectors are compared against a combined reference category (construction, creative industry, transport, other services). Grouping of the reference category is due to small shares of each sector in total employment and is done to avoid perfect multicollinearity.

Economic development, measured as gross domestic product (GDP) per capita in purchasing power parity (PPP; Nomenclature of Territorial Units for Statistics, grade 3 = cities, counties & districts [NUTS3] level), is as an economic performance indicator for cities (Ardagna & Lusardi, 2010). Since economic performance proxies can have a nonlinear relationship with entrepreneurship (Estrin & Mickiewicz, 2012), we also include the squared term for economic development. To account for the possibility that corruption raises costs and delays entry (Estrin et al., 2013; Klapper et al., 2006), we use Transparency International’s CPI. This is measured at the country level with a minimum value of 0 and maximum of 10. CPI is highly correlated with European Union (EU) integration and Eurozone status, so we omit these indicators.

To address potential omitted variable bias related to unobserved city characteristics, we use city-type fixed effects, proxied using 13 city typology dummies identified by the European Commission (2007). Three typologies represent international hubs with pan-European or global influence: knowledge hubs, established capitals, and reinvented capitals (e.g., Milan, Stockholm). Another six typologies represent cities serving as specialized poles, which can play an important role in at least some aspect of the urban economy: national service hubs, transformation poles, gateways, modern industrial centers, research centers, and visitor centers (e.g., Marseille, Seville). The remaining four typologies represent regional poles: deindustrialized cities, regional market centers, regional public service centers, and satellite towns (e.g., Poznan, Dresden, Manchester). The gateways typology serves as reference category. The city typologies have low correlations with the other variables (<0.25), except for knowledge hubs with economic development (0.66) and established capitals with population density (0.55). We also used time fixed effects to control for factors that vary within a city over time and can be applied to all cities (Certo, Withers, & Semadeni, 2016).

Data Analysis

We apply a nested multilevel model where cities are located (clustered) in countries. Our dependent variable is at the city level and not nested within a city. The multilevel regression contains fixed and random effects. Fixed effects are directly estimated, in addition to indirectly estimated by covariances of random intercepts and slopes (see McCulloch, Searle, & Neuhaus, 2008; Rabe-Hesketh & Skrondal, 2012).

The panel data include information on within-city and between-city variance. This lets us consider whether conditions across cities (between-city) and changes in conditions within cities (within-city) affect city entrepreneurship (see Certo et al., 2016). Multilevel regression yields a mixed model where we estimate variation within and between cities. This model specification has its origin in Papke and Wooldridge (2008). Formally, a two-level model was estimated with the dependent variable $y_{ijt}$ and the independent variable $x_{ijt}$ such that:

y_{ijt} = β_{0} + β_{1} x_{ijt} + ɛ_{ijt}

(1)

where i is the city level-1, j is the country level-2, and t serves to index the wave survey. Dependent variable

y_{ijt}

is city level new business formation. Explanatory variables, described earlier, are presented by

x_{ijt}

. Finally,

ɛ_{ijk}

is an error term that, in the hierarchical model, consists of three components:

ɛ_{ijt} = γ_{i \cdot \cdot} + μ_{ij \cdot} + v_{ijt}

(2)

where

γ_{i \cdot \cdot}

represents the omitted variables that vary across cities but not over countries and time,

μ_{ij \cdot}

denotes the omitted variables that vary over countries but are constant across cities and over time, and

v_{ijk}

is the error term.

Our two-level model includes 228 cities over 2004 to 2009. The random intercept captures unobserved heterogeneity at the country level and is assumed to be independent and normally distributed with zero mean and constant variance. The error terms are assumed to be independent and can therefore be directly estimated (Goldstein, 2005). There is no statistical justification for imposing a particular covariance structure between random effects at the start, so we estimate a model with unstructured random-effects covariance matrix (Rabe-Hesketh & Skrondal, 2012).

Post-estimated predictive margins were calculated from predictions of a previously fit model at fixed values of some covariates, and averaging or otherwise integrating over the remaining covariates. We do this for the four aspects of NBR, using both the procedures and costs measures. Post-estimated predictive margins are a tool to explain a relationship when the sign changes. They estimate the margins of responses for specified values of covariates using 95% confidence intervals to measure boundaries of the effect of NBR on city level entrepreneurship.

We point out that Lind and Mehlum (2010) argued that conventional means for testing nonlinearities are inadequate, and they proposed additional testing beyond significance and signs of coefficients. In line with their approach, after calculating predictive margins, we also test for inverted U-shaped relationships and inflection points.

Results

Hypothesis Testing

Table 2 contains results from the mixed-effects estimation testing the hypothesized inverted U-shaped relationship between NBR and city level entrepreneurship. The baseline model (specification 1, Table 2) has a statistically significant random intercept, which indicates multilevel modeling is most appropriate for our purposes (McCulloch et al., 2008). Also, the likelihood ratio (LR) test comparing the model with one level ordinary linear regression is significant, advocating multilevel modeling. Interclass correlation coefficients (ICC) demonstrate that a large part of the variance in city level entrepreneurship is due to between-country differences. ICC is between 89% for procedures related to NBR (specification 2) and 78% for costs of NBR (specification 3). We also perform poolability test of regression coefficients, checking that 228 cities can be pooled together. F-statistic varies from 12.36 to 21.32 (fixed effects), implying rejection of the null hypothesis of constant effects, justifying the panel data approach (Wooldridge, 2002).

Table 2.

Multilevel Multivariate Regression Main Effects Model (DV—City Level Entrepreneurship).^†

			Full sample		Reduced sample
	Model type	Baseline	Procedures	Cost	Procedures	Cost
Sample	Specification	(1)	(2)	(3)	(4)	(5)
Level 2
Entry		6.37** (1.44)	−1.11** (0.29)	0.96^† (0.44)	−1.16** (0.26)
Entry squared		−0.48** (0.10)	0.02** (0.01)	−0.20* (0.09)	0.02** (0.00)
Property registration cost		−21.43** (3.23)	−4.53** (0.99)	−28.39** (2.81)	−5.94** (0.89)
Property registration cost squared		1.57** (0.29)	0.28** (0.10)	1.97** (0.26)	0.43** (0.09)
Paying taxes		−0.17 (0.16)	−2.87** (1.03)	−0.12 (0.14)	−2.99** (0.92)
Paying taxes squared		0.89 (0.00)	0.03** (0.01)	−0.57 (0.00)	0.03** (0.01)
Enforcing contracts		−10.46 (8.49)	0.80 (2.07)	−10.40 (9.25)	0.68 (1.76)
Enforcing contracts squared		0.16 (0.13)	−0.02 (0.05)	0.17 (0.14)	−0.01 (0.04)
CPI		0.95 (0.69)	0.53 (0.66)	−2.67** (0.65)	−1.10^† (0.60)
Level 1
GDP		−39.36** (13.91)	−34.10* (15.18)	−9.73 (13.52)	−21.15 (15.98)
GDP squared		1.83** (0.70)	1.60* (0.77)	0.41 (0.70)	1.01 (0.83)
Density		−0.04 (0.39)	−0.01 (0.41)	0.14 (0.33)	0.10 (0.39)
Population at work		0.70^† (0.41)	0.77^† (0.46)	0.07 (0.39)	0.34 (0.46)
Agriculture		0.06 (0.17)	0.12 (0.19)	0.18 (0.15)	0.17 (0.17)
Manufacturing		0.0375 (0.05)	0.0880 (0.06)	0.0423 (0.04)	0.0838 (0.05)
Retail		0.021 (0.07)	0.02 (0.08)	0.15* (0.07)	0.07 (0.08)
Finance		0.14^† (0.08)	0.18* (0.08)	0.08 (0.07)	0.12 (0.08)
Public		−0.04 (0.05)	0.02 (0.06)	−0.06 (0.05)	0.03 (0.06)
ICT		−0.293* (0.14)	−0.304* (0.15)	−0.342* (0.16)	−0.395* (0.19)
Constant	12.55** (1.35)	418.6** (148.11)	248.3** (77.58)	321.5 (154.10)	195.8* (77.71)
Variance components
Within country (estimate)	32.65	148.33	78.26	232.22	54.74
Within country (SE)	10.79	64.21	31.5	108.32	27.95
Intercept variance (estimate)	29.19	18.14	20.84	11.93	16.78
Residual variance (SE)	1.88	1.19	1.36	0.86	1.21
ICC	0.53	0.89	0.78	0.93	0.76
LR test vs. linear regression		444.13	359.86	118.42	42.02
Wald χ² test	144.49	276.45	180.66	411.71	174.96
Log likelihood	−1569.5	−1470.4	−1497.2	−1175.4	−1230.1

Note. Full sample = 496 observations, 228 cities, and 20 countries. Reduced sample = 422 observations, 189 cities, and 19 countries. City type fixed effects and year dummies control for unobserved heterogeneity across cities. City fixed effects not included in the base model (with random intercept to test validity). Output is suppressed to save space. CPI = Corruption Perceptions Index; DV = dependent variable; GDP = gross domestic product; ICC = interclass correlation coefficients; ICT = information and communication technology; SE = standard error.

Significance ^†0.1%, * 0.05%, ** 0.01% do not include zero. 90%, 95%, 99% confidence intervals do not include zero. Given the nonlinear model, significance may vary within an interval at level 2.

Source: Eurostat Statistics on European Cities: 2004–2009; World Bank's Doing Business Data: 2004–2009; Transparency International (2012).

We found that urban industry mix, employment in services, transport, financial intermediation, and creative sectors encourage city level entrepreneurship. Level of economic development has nonlinear association with city level entrepreneurship (Thurik, Carree, van Stel, & Audretsch, 2008). Finally, the employed population positively affects city level entrepreneurship.

The results in Table 2 partially support an inverted U-shaped relationship between four NBRs and city level entrepreneurship. Our results for NBR related to entry provide partial support for our hypothesis and identify a nonlinear influence of NBR related to entry on city level entrepreneurship, with contrasting effects for entry procedures and entry costs. An increase in entry procedures has an inverse U-shaped relationship with city level entrepreneurship (β = 6.37, p < .01; β = −0.48, p < .01). Up to seven entry procedures encourages entrepreneurship, after which it falls, indicating that seven procedures is an optimal policy target. In contrast, we found a U-shaped relationship between the cost of entry regulation and entrepreneurship (β = −1.11, p < .01 and β = 0.02, p < .01). Entry costs deter entrepreneurship up to a threshold of 20% of income per capita.

Our results for NBR related to property registration do not support our hypothesis. We found a U-shaped influence for property registration procedures (β = −21.43, p < .01 and β = 1.57, p < .01) and costs (β = −4.53, p < .01; β = 0.28, p < .01) on city level entrepreneurship, with the effect being weaker for costs. NBR related to property registration discourages city level entrepreneurship, which falls to 0 at six procedures and at 9% of costs, after which it grows with more procedures and higher costs. This is in line with Ardagna and Lusardi (2010), who found a varied effect of regulation on necessity- and opportunity-driven entrepreneurship in a country. For example, in countries that are more heavily regulated but also have lower economic development status, the threshold for regulation could be higher (see Thurik et al., 2008).

We do not find support for our hypothesis when it comes to NBR related to tax regulation. The relationship between tax payments and city level entrepreneurship is negative (β = −0.17, p > .10 and β = 0.89, p > .10). This is consistent with some previous research (Acs et al., 2008b; Baliamoune-Lutz & Garello, 2014) that has found more tax payments reduces entrepreneurship. City level entrepreneurship is likely to be 0 when the number of tax payments exceeds 35. The cost of taxes has a U-shaped relationship (β = −2.87, p < .01; β = 0.03, p < .01) with city level entrepreneurship: It falls when the tax rate reaches 35% to 40%, and then it rebounds again. The result is driven by Eastern European cities with high entrepreneurship rates but also low financial development and high taxes.

Our findings provide support for NBR related to enforcing contracts. Although beta coefficients in Table 2 are not statistically significant, predicted margins demonstrate an inverted U-shaped relationship holds for a very specific interval. The inverted U-shaped relationship for contract procedures and city level entrepreneurship was found at between 36 and 42 procedures, while the relationship between contract costs and city level entrepreneurship was found between 12% and 28% of the claim value. This is consistent with research on the need for a conducive level of costs to enforce contracts (Johnson et al., 1999) to protect entrepreneurs. Graphical representation of the results is illustrated in Figure 3 in the Appendix.

Robustness Checks

We conduct four types of robustness checks. High levels of entrepreneurship in German cities compared to other European cities (Figure 2 in the Appendix), coupled with generally more complex NBRs, suggests an additional robustness check of our findings. We reduced our sample to 189 cities in 21 countries excluding German cities (Table 2, specifications 4 and 5). The country level gains explanatory power in explaining city entrepreneurship: We find an ICC of 92% for procedures and 76% for costs related to NBR. The reduced sample yields similar results for the direction and significance of explanatory variables, with different magnitudes and inflection points for inverted U-shaped and U-shaped relationships. Notably, the inflection point for entry procedures falls to four procedures when German cities are excluded (seven procedures in the full sample). The result for tax procedures is similar: 20 payments when German cities are excluded (30 payments in the full sample).

Results for predictive margins provide partial support for an inverted U-shaped relationship in two types of NBRs (related to entry, contracts) with city level entrepreneurship. In contrast, a U-shaped relationship is found for cost of paying taxes and cost of property registration. This brief exercise without German cities suggests that future research consider the role of unique country context for entrepreneurship in Germany (Audretsch & Lehmann, 2016).

Beyond the analyses shown in Table 2 and Figure 3, we statistically verify the inverted U-shaped relationships (Lind & Mehlum, 2010). We did this for eight potential relationships related to our main hypothesis. We corroborate the findings in Figure 3. The size of the effect of the U-shaped relationship remains consistent across all NBRs related to entry, property registration, and contract enforcement. Beta-coefficients and U-test jointly show that across NBRs related to entry cost and contract enforcement, the size of the inverted U-shaped relationship remains stable. For NBRs related to entry procedures and property rights, a U-shaped relationship was found. More specifically, the test calculates an optimum point (number of procedures or cost) where the NBR contributes most to city level entrepreneurship.

In addition, we explore whether four types of NBRs have a nonlinear association with city level entrepreneurship. We started by estimating each NBR measure in levels, then using squared terms. This allowed us to identify nonlinearities and establish a form of relationship. We also calculate plotted predicted margins for procedures and costs of each NBR, using Delta method. We calculate the F-test for nonlinearities and plot predictive margins of each regulation type using locally weighted regressions in Figure 3. Results confirm previous findings.

Finally, we standardize all country level variables. We created standardized values and specified the desired mean of 0 and standard deviation of 1, producing a new variable with mean zero and standard deviation unity. We do this so NBR is comparable across countries and cities (Wooldridge, 2002). Confidence intervals and predictive margins built on standardized values of NBR procedures and costs all support the findings in Table 2 and Figure 3.

Discussion

Our study highlights the importance of using multilevel modeling to understand the impact of NBR on city level entrepreneurship. The nested approach applied in our analysis provides clear insights on the multilevel embeddedness of entrepreneurship. Given that both city and country characteristics are important for city level entrepreneurship, we suggest that city policymakers can actively enhance or mitigate the influence of NBR for their own cities. We tested the hypothesis that NBR has an inverted U-shaped relationship with city level entrepreneurship. We found partial support for entry regulation and weak support for contract regulation (where the hypothesis is valid within a limited interval). In contrast, we found a U-shaped relationship for paying taxes (cost only), for property registration, and entry cost. We suggest that heterogeneity in the effect of NBR on city level entrepreneurship is an important question for future research, and this could be investigated with the lens of opportunity and necessity entrepreneurship in a city and country (Van Stel et al., 2007).

Although we find partial support for our hypothesis, we establish evidence that the relationship between NBR and city level entrepreneurship is nonlinear and this varies across the type of NBR as well as the way it is undertaken (procedures or cost). Nonlinearities tell much of the story about the influence of NBR in city level entrepreneurship than the usually linear negative effect that emerges from the country level research (Ardagna & Lusardi, 2010; Estrin et al., 2013; Klapper et al., 2006; Stenholm et al., 2013). Deeper investigation of nonlinearities should be a priority for future research.

Our results show that NBRs related to property registration and taxes are more consequential for city level entrepreneurship than NBRs related to entry and contract enforcement. In other words, the marginal increase/decrease in city level entrepreneurship corresponding to one unit increase in regulation is larger for property registration and taxes than for entry and contract enforcement (see Figure 3). We also find that city level entrepreneurs are more sensitive to changes in procedures to comply with regulation than to those related to financial costs: A unit change in procedures related to entry and property registration has a higher marginal effect than a unit change in costs (Figure 3). This implies that policymakers seeking fast results could prioritize reforms in procedural or administrative requirements. Our results highlight the importance of considering not only the type of NBR (e.g., entry versus taxes) but also the way regulations are implemented (e.g., by making changes to procedures or financial costs).

We contribute to entrepreneurship research in three ways. First, we use an institutional lens to link NBR and city level entrepreneurship (Baumol, 1990; Djankov et al., 2009; North, 1990; Whitley, 1999), advancing sparse research on country factors and subnational outcomes (Bosma & Sternberg, 2014). In doing so, we determine that key assumptions about size and role of regulation on entrepreneurship should be reconsidered (Autio et al., 2014; Spigel, 2017).

Second, we advance knowledge on institutions and entrepreneurship by delineating how public interest and public choice theories of regulation can explain the effect of NBR on city level entrepreneurship. We urge regulators, particularly in regions like Europe where moving between countries is easier, to reconsider the conventional “less is better” approach in favor of “smarter” regulation providing a favorable policy mix for entrepreneurship.

Third, we make a methodological contribution to entrepreneurship research by using a multilevel model comprising both country level and city level characteristics (Certo et al., 2016). Our findings demonstrate the value of exploiting multilevel analysis in entrepreneurship research (Ding et al., 2015; Hundt & Sternberg, 2016; Westlund et al., 2014).

Our work highlights the need for scholars to carry out a research agenda on within- and between-country effects of NBR and their relationship with subnational entrepreneurship outcomes. Some theoretical and methodological limitations of our study can be addressed in further research. First, our dependent variable does not capture if the motivation for entrepreneurship is opportunity or necessity driven, which could shift the threshold entry across various NBRs (see Ardagna & Lusardi, 2010). The Urban Audit data do not allow us to measure the overall direct size of the effect in employment terms, since the data on new business formation does not include jobs. Future research could attempt to connect employment with new business formation, to assess how NBRs influence specific types of city level entrepreneurs.

A fruitful path for future research is to further unpack the heterogeneity of NBRs beyond the four types covered in our study. These could include, for example, NBRs related to tariffs, exports, labor markets, finance, and bankruptcy. Related to this is the need to unpack other types of cross-national institutional variations (see Welter, 2011), like corruption. This could be particularly interesting given the differences between cities in Germany and those in other countries.

Also, the sectoral distribution of entrepreneurial activity across cities is not available in our sample. Some cities attract entrepreneurs in some sectors, for example, information technology (IT) in Berlin, and some cities may be marked by entry in other sectors or by founders, for example, immigrants (e.g., Birmingham, Paris, London, Marseille).¹¹ NBRs can have varied industrial and geographic effects on different types of entrepreneurs, which could vary based on where they tend to cluster. Future data efforts could enable fine-grained analysis of these questions.

The endogeneity of regulation may be a less relevant challenge for our study, but we are only able to consider our estimates with one period (3 years) lagged explanatory and control variables. Our results are robust but the significance of coefficients drops and the size of the confidence intervals increases once we apply lagged values. Future research should attempt more frequent data collection to lengthen the time period to address potential endogeneity issues.

Footnotes

Appendix

Acknowledgments

The authors thank Zoltan Acs, Johan Eklund, Sergio Fernandez, Leora Klapper, Ashlyn Nelson, Bob Rijkers, and participants at the World Bank’s 2014 Doing Business conference. Sean Webeck provided valuable research assistance.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Note

Author Biographies

David B. Audretsch is a Distinguished Professor and Ameritech Chair of Economic Development at the School of Public and Environmental Affairs, Indiana University.

Maksim Belitski is an Associate Professor at Henley Business School, University of Reading.

Sameeksha Desai is an Associate Professor at the School of Public and Environmental Affairs, Indiana University.

References

Acs

Desai

Klapper

(2008a) What does entrepreneurship data really show? Small Business Economics 31(3): 265–281.

Acs

Glaeser

Litan

Fleming

Goetz

Kerr

Strange

(2008b) Entrepreneurship and urban success: Toward a policy consensus, Kansas City, MO: Kauffman Foundation.

Ardagna

Lusardi

(2009) Where does regulation hurt? Evidence from new businesses across countries (No. w14747), Cambridge, MA: National Bureau of Economic Research.

Ardagna

Lusardi

(2010) Heterogeneity in the effect of regulation on entrepreneurship and entry size. Journal of the European Economic Association 8(2–3): 594–605.

Armington

Acs

(2002) The determinants of regional variation in new firm formation. Regional Studies 36(1): 33–45.

Audretsch

Belitski

(2017) Entrepreneurial ecosystems in cities: Establishing the framework conditions. The Journal of Technology Transfer 42(5): 1030–1051.

Audretsch

Belitski

Desai

(2015) Entrepreneurship and economic development in cities. The Annals of Regional Science 55(1): 33–60.

Audretsch

Lehmann

(2016) The seven secrets of Germany, Oxford: Oxford University Press.

Autio

Kenney

Mustar

Siegel

D. S.

Wright

(2014) Entrepreneurial innovation: The importance of context. Research Policy 43(7): 1097–1108.

10.

Baliamoune-Lutz

Garello

(2014) Tax structure and entrepreneurship. Small Business Economics 42: 165–190.

11.

Baumol

(1990) Entrepreneurship: Productive, unproductive and destructive. Journal of Political Economy 98(5): 893–921.

12.

Belitski

Chowdhury

Desai

(2016) Taxes, corruption, and entry. Small Business Economics 47(1): 201–216.

13.

Belitski

Desai

(2015) What drives ICT clustering in European cities? Journal of Technology Transfer 41(3): 430–450.

14.

Bergmann

Sternberg

(2007) The changing face of entrepreneurship in Germany. Small Business Economics 28(2–3): 205–221.

15.

Boschma

Fritsch

(2009) Creative class and regional growth: Empirical evidence from seven European countries. Economic Geography 85(4): 391–423.

16.

Bosma

Sternberg

(2014) Entrepreneurship as an urban event? Empirical evidence from European cities. Regional Studies 48: 1016–1033.

17.

Braunerhjelm

Desai

Eklund

J. E.

(2015) Regulation, firm dynamics and entrepreneurship. European Journal of Law and Economics 40(1): 1–11.

18.

Bruton

Ahlstrom

(2010) Institutional theory and entrepreneurship: Where are we now and where do we need to move in the future? Entrepreneurship Theory and Practice 34(3): 421–440.

19.

Certo

S. T.

Withers

M. C.

Semadeni

(2016) A tale of two effects: Using longitudinal data to compare within-and between-firm effects. Strategic Management Journal 38(7): 1536–1556.

20.

Charron

Dijkstra

Lapuente

(2014) Regional governance matters: Quality of government within European Union member states. Regional Studies 48(1): 68–90.

21.

Chowdhury, F., Desai, S., Audretsch, D., & Belitski, M. (2015). Erratum to: Does corruption matter for international entrepreneurship? International Entrepreneurship and Management Journal, 4(11): 981–981.

22.

Ciccone, A., & Papaioannou, E. (2007). Red tape and delayed entry. Journal of the European Economic Association, 5(2–3): 444–458.

23.

Davidsson

Lindmark

Olofsson

(1994) New firm formation and regional development in Sweden. Regional Studies 28: 395–410.

24.

Desai

Acs

Weitzel

(2013) A model of destructive entrepreneurship: Insight for conflict and postconflict recovery. Journal of Conflict Resolution 57(1): 20–40.

25.

De Soto

(1989) The other path, New York, NY: Harper and Row.

26.

Ding

Chiang

(2015) Social trust and angel investors’ decisions: A multilevel analysis across nations. Journal of Business Venturing 30(2): 307–321.

27.

Djankov

Ganser

McLiesh

Ramalho

Shleifer

(2009) The effect of corporate taxes on investment and entrepreneurship. AEJ: Macroeconomics 2: 31–64.

28.

Djankov

La Porta

Lopez-de-Silanes

Shleifer

(2002) The regulation of entry. Quarterly Journal of Economics 117(1): 1–37.

29.

Eskridge

Jr (1988) Politics without romance: Implications of public choice theory for statutory interpretation. Virginia Law Review 74(2): 275–338.

30.

Estrin

Korosteleva

Mickiewicz

(2013) Which institutions encourage entrepreneurial growth aspirations? Journal of Business Venturing 28(4): 564–580.

31.

Estrin

Mickiewicz

(2012) Shadow economy and entrepreneurial entry. Review of Development Economics 16(4): 559–578.

32.

European Commission. (2007). Adding the value to the European Urban Audit. The State of European cities report (May 2007). Brussels: Inforegio.

33.

Feldman

(2001) The entrepreneurial event revisited. Industrial and Corporate Change 10: 861–891.

34.

Fogel

Hawk

Morck

Yeung

(2006) Institutional obstacles to entrepreneurship. In: Casson

Yeung

Basu

Wedeson

(eds) The Oxford handbook of entrepreneurship, Oxford: Oxford University Press, pp. 540–579.

35.

Fritsch, M., & Mueller, P. (2008). The effect of new business formation on regional development over time: the case of Germany. Small Business Economics, 30(1), 15–29.

36.

Fritsch

Storey

(2014) Entrepreneurship in a regional context. Regional Studies 48(6): 939–954.

37.

Glaeser

Kerr

Ponzetto

(2010a) Clusters of entrepreneurship. Journal of Urban Economics 67(1): 150–168.

38.

Glaeser

Rosenthal

Strange

(2010b) Urban economics and entrepreneurship. Journal of Urban Economics 67(1): 1–14.

39.

Goldstein

(2005) Multilevel models, Chichester: John Wiley & Sons, Ltd.

40.

Hantke-Domas

(2003) The public interest theory of regulation: Non-existence or misinterpretation? European Journal of Law and Economics 15: 165–194.

41.

Hundt

Sternberg

(2016) Explaining new firm creation in Europe from a spatial and time perspective. Papers in Regional Science 95(2): 223–257.

42.

Jacobs

(1970) The economy of cities, New York, NY: Vintage Books Edition, Random House.

43.

Johnson

McMillan

Woodruff

(1999) Contract enforcement in transition (EBRD Working Paper No. 45), London: European Bank of Reconstruction and Development.

44.

Keeble

Walker

(1994) New firms, small firms and dead firms: Spatial patterns and determinants in the United Kingdom. Regional Studies 28(4): 411–427.

45.

Klapper

Laeven

Rajan

(2006) Entry regulation as barrier to entrepreneurship. Journal of Financial Economics 82: 591–629.

46.

Lind

J. T.

Mehlum

(2010) With or without U? The appropriate test for a U-shaped relationship. Oxford Bulletin of Economics and Statistics 72(1): 109–118.

47.

Marshall

(1890) Principles of economics, London, UK: Macmillan.

48.

McCulloch

C. E.

Searle

S. R.

Neuhaus

J. M.

(2008) Generalized, linear, and mixed models, 2nd ed. Hoboken, NJ: Wiley.

49.

North

(1990) Institutions, institutional change and economic performance, Cambridge: Cambridge University Press.

50.

Papke

L. E.

Wooldridge

J. M.

(2008) Panel data methods for fractional response variables with an application to test pass rates. Journal of Econometrics 145(1): 121–133.

51.

Peltzman

(1976) Toward a more general theory of regulation. Journal of Law and Economics 19(2): 211–240.

52.

Pigou

A. C.

(1938) The economics of welfare, 4th ed. London, UK: Macmillan.

53.

Rabe-Hesketh

Skrondal

(2012) Multilevel and longitudinal modeling using stata, 3rd ed. College Station, TX: Stata Press.

54.

Reynolds

Storey

Westhead

(1994) Cross-national comparison of the variation in new firm formation rates. Regional Studies 28: 443–456.

55.

Spigel

(2016) Developing and governing entrepreneurial ecosystems: The structure of entrepreneurial support programs in Edinburgh, Scotland. International Journal of Innovation and Regional Development 7(2): 141–160.

56.

Spigel

(2017) The relational organization of entrepreneurial ecosystems. Entrepreneurship Theory and Practice 41(1): 49–72.

57.

Stenholm

Acs

Wuebker

(2013) Exploring country-level institutional arrangements on the rate and type of entrepreneurial activity. Journal of Business Venturing 28(1): 176–193.

58.

Thurik

Carree

van Stel

Audretsch

(2008) Does self-employment reduce unemployment? Journal of Business Venturing 23: 673–686.

59.

Transparency International. (2012). Corruption perceptions index 2012. Berlin: Transparency International Secretariat.

60.

Tullock

(1967) The welfare cost of tariffs, monopoly, and theft. Western Economic Journal V: 224–232.

61.

Van Stel

Storey

Thurik

(2007) The effect of business regulations on nascent and young business entrepreneurship. Small Business Economics 28(2–3): 171–186.

62.

Welter

(2011) Conceptualizing entrepreneurship: Conceptual challenges and ways forward. Entrepreneurship Theory and Practice 35: 165–184.

63.

Westlund

Larsson

J. P.

Olsson

A. R.

(2014) Start-ups and local entrepreneurial social capital in the municipalities of Sweden. Regional Studies 48(6): 974–994.

64.

Whitley

(1999) Divergent capitalisms, Oxford: Oxford University Press.

65.

Williamson

(2000) New institutional economics. Journal of Economic Literature 38: 595–613.

66.

Wooldridge

(2002) Econometric analysis of cross section and panel data, Cambridge, MA: MIT Press.