Abstract
This article presents the spatial patterns of general and high-tech start-up rates and explores regional factors associated with entrepreneurship in U.S. micropolitan areas. Regression results show that general entrepreneurship in these small cities is predicted by population growth, the middle-age population group, the presence of small businesses, and natural amenities. Additionally, high-tech start-up activities are positively associated with human capital, creative knowledge (instead of technological knowledge), high-tech clustering, and proximity to a large metropolitan area. These findings are compared with the patterns in larger metropolitan areas. This research sheds light on local entrepreneurship policy in the small-city context.
Over the past few decades, the knowledge economy has created a “great divergence” in economic development between large, innovative cities and the majority of cities and towns in the United States (Moretti, 2012). Due to the lack of a critical mass for agglomeration economies, small cities outside of metropolitan areas are particularly vulnerable to brain drain and population loss, creating serious challenges for economic development policy. Short-sighted smokestack-chasing strategies are still common in these places (Hill, 1998; Partridge et al., 2009). Nevertheless, from the Midwest to the Mountain West, small cities such as Jasper in Indiana and Bozeman in Montana have built vibrant local economies in recent years by adopting entrepreneurship-based economic development strategies (DeVol & Wisecarver, 2018; Motoyama et al., 2017).
Indeed, fostering entrepreneurship is one of the common economic development practices adopted by all kinds of local jurisdictions, not just in innovative hubs such as Silicon Valley, Boston, and New York. New business formation plays an increasingly important role in job creation, innovation, and economic growth (Acs et al., 2008; Haltiwanger et al., 2013; Qian, 2018). Economic performance differences across regions are in part attributable to varying business start-up rates (Audretsch et al., 2008), which are further affected by various regional environmental factors (Florida et al., 2017; Qian et al., 2013). Understanding these factors helps inform policy makers on entrepreneurship policy that creates friendly regional environments for entrepreneurs and business start-ups.
While scholars have extensively studied regional variations in entrepreneurial activities, most existing studies focused on metropolitan areas (e.g., Adler et al., 2019; Lee et al., 2004; Qian et al., 2013), and limited efforts have been made to study varying new firm formation across small and medium-sized urban communities. Undoubtedly, the most visible entrepreneurial cities are all large metropolitan areas (Moretti, 2012). But it is also important to study small-city entrepreneurship, at least for two reasons. First, despite in a disadvantageous position during the great divergence (Moretti, 2012), small and medium-sized cities and towns outside of metropolitan areas still account for a significant share of population and economic contribution to the U.S. urban system (Partridge et al., 2008). Entrepreneurship potential in these communities therefore may not be overlooked. Second, nonmetropolitan entrepreneurial activities exhibit different patterns from metropolitan entrepreneurship and research findings based on the latter may not necessarily apply to the former. For instance, Shrestha et al. (2007) found that the contribution of new business formation to job growth is stronger in U.S. metropolitan areas than in nonmetropolitan areas. Similarly, Audretsch et al. (2015) showed that the economic effects of entrepreneurship are different between small and medium-sized cities and large cities in Europe. It is also expected that regional factors predicting small-city entrepreneurship, which have been understudied, may not be the same with those predicting entrepreneurship in metropolitan areas.
The focus of this research, therefore, is on regional variations in entrepreneurial activities in U.S. micropolitan areas, one representative type of economically functional small cities. Micropolitan statistical areas, according to the U.S. Office of Management and Budget’s 2013 definition, consist of at least one urbanized area with a population size between 10,000 and 50,000 and the surrounding areas where residents largely travel to the urban core to work. We first present the spatial patterns of start-up rates in all industries and in high-tech industries across 532 U.S. micropolitan areas. We then use multivariate regression analysis to explore regional factors associated with entrepreneurship in these small cities. Independent variables cover population density and growth, female employment, foreign born, age structure, average firm size, unemployment and poverty, utility and design patents, human capital, natural amenities, social capital, industrial structure, proximity to large metropolitan areas, and the local presence of federal Small Business Innovation Research (SBIR) grants. Though the focus of this research is on micropolitan areas, we also include results based on metropolitan areas as a comparison. This study addresses the pressing need for research on entrepreneurial development in a small-city context. Our results suggest that local policy makers should recognize their own community characteristics and advantages to better inform entrepreneurship policy rather than simply replicating strategies taken in large cities.
This article is organized as five sections. The literature on the geographical patterns of entrepreneurship and on small-city economic development is reviewed. This paves the way for identifying explanatory variables used in regression models. We then explain the variables, data, and methods. Following that, we present the empirical results and then summarize the research and discuss policy implications.
Literature Review
This study focuses on regional variations in entrepreneurship in the small-city context. In this section, we first review the theoretical and empirical literature on regional factors associated with entrepreneurship. Even though findings in existing studies are mostly based on larger cities or regions, this review still sheds light on factors that could be relevant to the small-city context. Then, we narrow the discussion to the small-city context.
Factors Predicting Regional Variations in Entrepreneurship
Following a common approach in regional studies (e.g., Acs & Armington, 2006; Lee et al., 2004; Qian et al., 2013), we define entrepreneurship as new firm formation. Entrepreneurial activities vary by regions and are affected by various demographic, social, economic, knowledge, and amenities factors (Qian & Liu, 2018).
For demographic factors, population size and population growth may both have a significant association with regional entrepreneurial activities. Large cities facilitate knowledge spillovers across different sectors and encourage new firm formation (Acs et al., 2009; Duranton & Puga, 2001; Jacobs, 1969). Population growth in cities creates new market opportunities for entrepreneurs and also benefits them with increased labor force availability in volume and skills (Reynolds et al., 1994). We suspect that small cities, with the lack of a critical mass, may not enjoy such agglomeration effects. Social diversity is also considered an important factor that attracts and retains talent in cities (Florida, 2012) and facilitates innovation and entrepreneurship due to new ideas arising from diverse perspectives or new combinations of diverse ideas (Audretsch et al., 2010; Florida, 2012; Jacobs, 1969; Niebuhr, 2010; Qian, 2013). However, social diversity could create barriers for entrepreneurial activities due to high communication costs among different cultural groups (Lazear, 1999; Qian, 2013). There is also a gender dimension in entrepreneurship. Conventional wisdom shows that women face more obstacles in starting and growing their own businesses (Merrett & Gruidl, 2000; Minniti, 2009; Poczter & Shapsis, 2018). Nevertheless, female business ownership rates have been catching up quickly (Merrett & Gruidl, 2000; Minniti, 2009).
In terms of economic factors, unemployment—an indicator for economic stagnation—can either positively or negatively affect new firm formation. On one hand, a higher unemployment rate indicates limited employment opportunities in the labor market, encouraging necessity-based entrepreneurship when alternative employment options are unavailable (Acs & Armington, 2006). On the other hand, a higher unemployment rate relates to economic downturn and decreased demand for goods and services, which create barriers to entrepreneurial success in the market (Qian & Liu, 2018). The poverty level is another indicator for local economic performance. Rupasingha and Goetz (2011) found that self-employment rates are associated with poverty reduction in nonmetropolitan counties but not in metropolitan areas. Average establishment size is often found to be negatively associated with regional start-up activities in that clustering of smaller establishments may signify a friendly environment for business entry (Acs & Armington, 2006; Chinitz, 1961; Glaeser et al., 2010; Parajuli & Haynes, 2017; Qian, 2017a, 2017b). Moreover, the presence of industrial clusters represents competitive advantage for regions (Porter, 1998) and contributes to the success of entrepreneurs (Delgado et al., 2010). Small cities and towns, however, often lack a diverse industrial base, and their economy often depends on only one or a few industries (Mulligan & Vias, 2006).
Recent studies have also found evidence of knowledge spillovers in explaining variations in entrepreneurship across geographical regions. Audretsch (1995) and Acs et al. (2009) introduced the knowledge spillover theory of entrepreneurship. The theory considers new knowledge as one source of market opportunities for entrepreneurs and can best explain academic or high-tech entrepreneurship, which involves technological innovation and has high growth potential (Capello & Lenzi, 2014). Entrepreneurs can facilitate knowledge spillovers through exploiting new market opportunities and commercializing new knowledge in new firms. Human capital is related to the knowledge spillover theory and has been examined in empirical studies to explain high-tech entrepreneurship (Lee et al., 2004; Qian et al., 2013). The presence of skilled labor plays a critical role in creating new knowledge and appropriating its market value through entrepreneurial absorptive capacity (Qian et al., 2013). It should be noted that small cities typically do not have strong capacity to generate new knowledge and attract skilled workers.
Amenities or quality of life plays an important role in attracting talent, including entrepreneurs (Florida, 2012; Qian & Liu, 2018). Albouy (2012) found that cities with better quality of life (particularly natural amenities), regardless of population size, are more attractive to educated households. Glaeser et al. (2001) also reported that people are willing to pay higher rents to live in cities with higher amenities after considering wage levels. Quality of life is likely to play a critical role in small urban communities because of their comparative advantage in natural amenities.
From the policy perspective, SBIR has been identified as a potential entrepreneurship policy by various scholars (Audretsch et al., 2002b; Qian & Haynes, 2014; Shane, 2009). SBIR 1 is a federal program introduced to assist small and medium enterprises (SMEs) in their research and development (R&D) efforts and improve private commercialization of technologies. Previous studies have found that SBIR is an effective policy in promoting SME performance in terms of attracting venture capital, sales and employment growth, and/or technological commercialization (Audretsch et al., 2002a; Lerner, 1999; Link & Scott, 2009). A comprehensive study by Qian and Haynes (2014) focused on the role of SBIR in entrepreneurial activities in the technology sector across U.S. metropolitan counties. They reported a positive association between the number of SBIR grants and high-tech entrepreneurship. SBIR has not been studied in the small-city context but this direction is worth exploring.
Entrepreneurship and Economic Development in Small Cities
The geographical patterns of entrepreneurship among small cities are expected to be different from large cities. Behrens et al. (2014) argued that large cities are more attractive to talented individuals who are more likely to become productive entrepreneurs. In large cities, productive entrepreneurs can gain more due to agglomeration economies, whereas less productive entrepreneurs find it difficult to survive. Following this logic, it is expected to witness different types of entrepreneurship (by productivity) between small cities and large metropolitan areas. In another study, Renski (2008) identified varying intraregional entrepreneurship patterns across urban, suburban, and rural areas in the United States. For instance, he reported that technology start-ups in manufacturing and advanced-services industries are most often seen in suburban areas. This contradicts Jacobs (1961), who considered high density and diversity (which better describe central cities) more inducive to new work. Renski (2008) also found that small central cities or urban areas overall have the highest start-up rates in traditional manufacturing and the lowest start-up rates in technology-based advanced services. Though from the intraregional but not our interregional approach, the work by Renski (2008) also implied differentiated geographical patterns between small and large cities.
Despite being unattractive to entrepreneurs or new entries in productive industries (Behrens et al., 2014; Renski, 2008) due to the lack of agglomeration economies (Behrens et al., 2014, Partridge et al., 2008), small cities have comparative advantage in attracting people who value nature amenities; for instance, retirees and tourists (Kilkenny & Partridge, 2009) and the creative class (McGranahan & Wojan, 2007). The former group tends to increase the demand for entrepreneurs and the latter group tends to increase the supply for entrepreneurs. Therefore, natural amenities appear to be able to predict entrepreneurial activities in small cities. The importance of natural amenities to nonmetropolitan population and economic growth has been widely tested (e.g., Kilkenny & Partridge, 2009; McGranahan et al., 2011; D. Rickman & S. Rickman, 2011), but evidence on the relationship between natural amenities and entrepreneurship in small cities has been missed. This gap will be addressed in the following empirical analysis.
It is worth noting another factor that is unique to nonmetropolitan areas—proximity to metropolitan areas, which could exert both positive and negative impacts on entrepreneurship in small cities. In the broad economic development context, scholars described these effects as spread effects and backwash effects (e.g., Ganning et al., 2013; Partridge et al., 2007). Spread effects indicate that growth in metropolitan regions spills over positively to nearby regions, bringing resources and economic opportunities (Henry et al., 1997). Additionally, geographic proximity facilitates knowledge spillovers (Döring & Schnellenbach, 2006). It is reasonable to assume such spillover effects exist from metropolitan areas to neighboring areas, creating entrepreneurial opportunities for the latter (Acs et al., 2009). However, geographical proximity could also lead to backwash effects. Large metropolitan regions can draw resources such as human capital and businesses from nearby peripheral communities (Anderson, 2000; Barkley et al., 1996; Kilkenny & Partridge, 2009) or compete directly with the latter for the location of new firms. Therefore, entrepreneurs from small cities may suffer from the lack of resources and then be attracted to the metropolitan areas nearby.
Methodology
The empirical part of this research explores regional factors associated with entrepreneurship in small cities. Independent variables are identified through existing literature on regional studies of entrepreneurship (most of which focus on larger cities) and on small-city economic development (most of which focus on nonentrepreneurship economic indicators), as reviewed in the previous section. Although small cities are the primary interest of this study, we provide regression results for larger metropolitan areas using the same explanatory variables (except for proximity to a large metropolitan area), which allowed us to assess the uniqueness of small-city entrepreneurial development.
Micropolitan Areas as Economically Functional Small Cities
This study examines regional variations in entrepreneurial activities through the lens of micropolitan (and, for comparison, metropolitan) statistical areas. We use the 2013 delineation of micropolitan and metropolitan areas by the U.S. Office of Management and Budget, which states that a micropolitan or metropolitan area contains the territory that has a high degree of social and economic integration with an urban core. These geographical units are suitable for regional economic analysis as economically functional regions (Qian et al., 2013). A micropolitan statistical area has “at least one urban cluster of at least 10,000 but less than 50,000 population,” while a metropolitan statistical area consists of “at least one urbanized area of a population of 50,000 or more” (Office of Management and Budget, 2013, p. 2). According to the 2010 Census, about 10% of the U.S. population lives in micropolitan areas, compared with 83.7% in metropolitan areas. Population growth between 2000 and 2010 occurred in both metropolitan (accounting for 92.4% of all growth) and micropolitan regions (accounting for 6.3% of all growth; Wilson et al., 2012). In this study, we consider all 532 micropolitan areas and 377 metropolitan areas in the lower 48 states and the District of Columbia.
Variables and Data
The dependent variable in this study is entrepreneurship in micropolitan areas measured by the 2010 to 2015 annual average number of new single-unit establishments in all industries, and separately in high-tech industries, standardized by micropolitan employment (per 10,000 employees). We include only new single-unit establishments but not multiunit ones because the latter are mostly new branches of existing firms. We consider the annual average of the 5-year period to control for yearly fluctuation. It also allows us to have time lags with independent variables, most of which use 2010 data. We are separately interested in high-tech entrepreneurship because of the greater contribution of the high-tech sector to regional economies (Moretti, 2012). The definition of high-tech industries is obtained from Hecker (2005), who identified 46 four-digit North American Industry Classification System (NAICS) industries as high tech based on the share of science and technology workers. New establishment data are available from the U.S. Census Bureau’s Business Information Tracking Series. Employment data, available from County Business Patterns (CBP), are aggregated from the county level to the micropolitan level.
The geographical distributions of general and high-tech entrepreneurship are shown in Figures 1 and 2. These four maps describe the new firm formation rates in micropolitan and metropolitan areas separately. As shown in Figure 1, new firm formation rates of all industries are unevenly distributed across U.S. micropolitan areas, and the pattern is similar to the distribution across metropolitan areas. The Midwest cities, in general, have lower new-firm formation rates compared with the West and Southeast. The top five micropolitan areas with the highest new-firm formation rates across all industrial sectors are Key West, FL; Jackson, WY–ID; Williston, ND; Vineyard Haven, MA; and Heber, UT. The distribution pattern of micropolitan high-tech firm formation rates in Figure 2 is similar to the distribution of general entrepreneurship in micropolitan areas shown in Figure 1.
New-firm formation rates (all industries) in U.S. micropolitan and metropolitan statistical areas.
High-tech new firm formation rates in U.S. micropolitan and metropolitan statistical areas.
A detailed description of all variables for empirical analysis is shown in Table 1. Table 2 presents the descriptive statistics of these variables by micropolitan and metropolitan areas. Based on the mean differences in the table, micropolitan areas are significantly higher in general entrepreneurship but significantly lower in high-tech entrepreneurship than metropolitan areas. Table 3 further examines the heterogeneity between micropolitan and metropolitan entrepreneurship by sector. Besides primary industries such as agriculture and mining, micropolitan cities also show higher percentages of new firms in Construction, Retail Trade, Transportation and Warehousing, and Accommodation and Food Services. Not surprisingly, metropolitan regions have a much higher percentage of new firms in Professional, Scientific and Technical Services.
Description of Variables.
Note. NAICS = North American Industry Classification System.
Descriptive Statistics of Variables.
Percentages of New Firms by Two-Digit NAICS Code in Micropolitan and Metropolitan Areas.
Note. NAICS = North American Industry Classification System.
Independent variables are identified based on the literature reviewed in the previous section. Details of these variables are discussed below.
Demographic Indicators
Demographic indicators include population density, population change, female employment, foreign-born population, and population age structure. Population density is used as a measure of urbanization economies. It is the ratio of the 2010 population to the land area of a micropolitan area. Population and land area data are available from the 2010 Census. The second demographic variable is population growth measured by the percentage change of micropolitan population between the 2000 Census and the 2010 Census. Due to the gendered differences in entrepreneurship, we also include female employment as an explanatory variable, measured by the share of females among the civilian employed population of at least 16 years old. Its data source is the 2009 to 2013 5-year estimates of the American Community Survey (ACS). Foreign-born population is used to represent cultural diversity. Some studies found that foreign-born immigrants contribute disproportionately more to the skilled workforce and are more likely to become entrepreneurs in the high-tech sector (Fairlie & U.S. Small Business Administration, 2008; Saxenian, 1999). This variable is measured by the percentage of the foreign-born population in the micropolitan area, also based on the ACS data. Finally, we include the percentages of population between 25 and 44 and between 45 and 64 years old as two additional demographic variables. The former group is the young generation who may already have completed higher education and are active in the workforce. This group is more likely to migrate from small or rural regions to metropolitan areas for job opportunities and urban amenities (McGranahan & Wojan, 2007). The older generation between ages 45 and 64 years who are still in the workforce but closer to the retirement age may value small-city quality of life more than the younger generation. Small and rural communities have welcomed an increasing number of retirees in recent years (McGranahan & Wojan, 2007). This population age group, who is likely to have accumulated wealth and stronger purchasing power, may positively contribute to entrepreneurial activities in micropolitan regions. Population age structure data are retrieved from the 2010 Census.
Economic Factors
The second group of explanatory variables includes economic factors such as establishment size, poverty, unemployment, and industrial agglomeration. Establishment size is used as a proxy for market competition and is calculated by dividing the 2010 total micropolitan employment by the total number of establishments. The data source is CBP. Unemployment and poverty are used to represent the macroeconomic conditions in the micropolitan economy. Unemployment is measured by the unemployment rate and poverty is measured by the share of the micropolitan population below the poverty level. Data for both variables are retrieved from the 2009 to 2013 ACS.
Industrial structure/specialization is another economic indicator that can affect local entrepreneurship. We use the 2010 location quotients (LQ) of all two-digit NAICS industries as proxies for industrial structure/specialization. They are calculated based on the 2010 establishment data from CBP, using the nation as the benchmark region. The LQ method allows for comparing the regional share of an industry with its national share. Differentiated micropolitan industrial structures may lead to varying start-up rates. A high LQ often signifies the presence of an industrial cluster in a local economy, which benefits local entrepreneurs associated with that industry (Porter, 1998). However, specializing in an industry with a lower start-up rate may negatively affect micropolitan entrepreneurial activities. Understandably, it is more difficult to start a manufacturing firm than a retail firm due to the higher levels of capital investment and mechanical skills needed for the former. For the high-tech models, we used the LQ of the high-tech sector defined by Hecker (2005).
Knowledge Factors
We include two knowledge factors to measure potential knowledge-based entrepreneurial opportunities at the micropolitan level (i.e., patents and human capital), following Acs et al. (2002) and Qian et al. (2013). The patents variable is measured by the total number of patents in 2010 standardized by the 2010 micropolitan population. We distinguish two types of patents: design patents and utility patents. A utility patent “protects the way an article is used and works” and a design patent “protects the way an article looks” (U.S. Patent and Trademark Office, 2018, p.1500-2). Utility patents are commonly adopted to explain regional variations in high-tech entrepreneurship (Qian, 2017b), and design patents capture regional creative activities and better address the cultural and creative economy (Qian & Liu, 2018). Patents data are retrieved from the 2010 U.S. Patent and Trademark Office Inventor Database. Human capital is measured by the share of adults (age 25+ years) with a bachelor’s degree or higher. Data are available from the 2009 to 2013 ACS.
Amenities/Quality of Life
We include social capital and natural amenities as two indicators for quality of life. For social capital, we used the measure adopted in Rupasingha et al. (2006). Their study defined a county-level social capital index incorporating four key elements of social capital: associational density, nonprofit organizations, voter turnout in elections, and participation in the decennial census. The authors used principal component analysis on these four elements to generate the social capital index. Details can be found in Rupasingha et al. (2006). We convert the county-level measure to the micropolitan level by taking the mean of z-scores of the four factors. Higher z-scores reflect higher levels of social capital in micropolitan areas.
The second quality-of-life variable is the natural amenity scale defined by the U.S. Department of Agriculture, which is also our data source. The amenity scale is calculated by aggregating six measures: warm winter, winter sun, temperate summer, low summer humidity, topographic variation, and water area (McGranahan, 1999). These measures reflect the environmental quality suitable for human habitats. The latest 1999 data are available at the county level. Natural amenities, such as climate, typology, and water area, are unlikely to change significantly at the regional level in a decade, so it should be reliable to use the 1999 data. We derive the micropolitan-level data by taking the mean of the county-level amenity scale.
Metropolitan Proximity
Adjacency to large metropolitan areas is considered in this study. We measure whether a micropolitan area is adjacent to at least one metropolitan area that has a population size greater than 1 million (1: adjacent; 0: not adjacent). We obtain the micropolitan values for this variable using ArcGIS.
SBIR
Following the regional study of entrepreneurship by Qian and Haynes (2014), we include the SBIR variable. It is measured by the number of SMEs that received at least one Phase I SBIR award located in a micropolitan area. The address data of SBIR recipients are available from the Small Business Administration. We rely on the 2010 ZIP Code Tabulation Area to County Relationship File Layout from the Census Geography Program to match cities and zip codes with counties and states. We then calculate the total number of firms within each micropolitan area that received at least one SBIR during 2000 to 2010. Since the primary purpose of SBIR grants is to support small businesses R&D and technological innovation, our analysis of SBIR is geared toward high-tech new-firm formation. Nevertheless, we also include this variable in the general model for a comparison purpose, and there could be multiplier or spillover effects from high-tech start-ups to other sectors.
U.S. Regions
Figures 1 and 2 show metaregional variations in new firm formation rates. Therefore, we include U.S. region dummy variables. The definition and data of U.S. regions are acquired from the U.S. Census, which assigns the states into four regions based on geographical adjacency and socioeconomic homogeneity. 2
Method
We estimate ordinary least squares regressions using the variables discussed above, separately for micropolitan and metropolitan areas. In all regression models, we use variance inflation factor (VIF) to test for multicollinearity. The results indicate high collinearity between human capital and industrial structure. More specifically, the professional service LQ variable and human capital have VIF values between 7 and 8 in the micropolitan model (but all other variables have VIF values below 5). Therefore, we first estimate the regression without the LQ variables, and then include them in the following model to examine their true effects on entrepreneurship after controlling for human capital. Significance levels are reported based on robust standard errors to account for heteroscedasticity.
Regression Results
The regression results for general and high-tech entrepreneurship models are shown in Tables 4 and 5. Models A and B represent the general entrepreneurship models in micropolitan areas. Models C and D represent the general entrepreneurship models in metropolitan areas (for the purpose of comparison). Models B and D incorporate the LQ variables for the reason stated before. Models E to H in Table 5 are the regression results for high-tech entrepreneurship. The explanatory variables are the same as those in the general entrepreneurship models except for the high-tech LQ used here instead. In this section, we first discuss the results on the significant regional factors associated with new firm formation across all sectors in micropolitan areas. Then, we compare results between micropolitan and metropolitan areas. Finally, results based on high-tech entrepreneurship are discussed.
Regression Results (All Industries).
Note. Robust standard errors are displayed in parentheses.
Significance levels: *p < .10. **p < .05. ***p < .01.
Regression Results (High-Tech Industries).
Note. Robust standard errors are displayed in parentheses.
Significance levels: *p < .10. **p < .05. ***p < .01.
Regional Factors Associated With Entrepreneurship in Micropolitan Areas
Based on the results shown in Model A, demographic factors that are significantly associated with micropolitan entrepreneurial activities include population growth (significant at the 0.01 level) and the population age group between 45 and 64 (significant at the 0.05 level). Population growth is related to new-firm formation on both the supply side and the demand side, consistent with existing literature (Reynolds et al., 1994). The middle age generation, generally with more stable and accumulated wealth than the younger members of the workforce, appears to be more likely to start their own businesses and/or have higher demand for services that create new market opportunities for entrepreneurs (e.g., in medical services, food and accommodation services, and convenience stores; Glasmeier & Howland, 1993).
For economic factors, neither unemployment nor poverty shows a significant relationship with micropolitan entrepreneurship in Model A. These results indicate that micropolitan entrepreneurs are less sensitive to the macro environment of the regional economy. New-firm formation rates in micropolitan areas are higher in industries such as Retail, Transportation and Food Services (see Table 3), and these start-ups are often in smaller scales and mainly serve the local market. Establishment size is negatively associated with new firm formation in small cities at the 0.01 significance level, consistent with existing literature based on metropolitan regions that suggests the clustering of smaller businesses as a symbol of a more friendly environment for business entry (e.g., Acs & Armington, 2006).
For knowledge variables, human capital shows a positive and significant relationship with micropolitan start-up activities at the 0.05 level in Model A, while the utility and design patent variables present insignificant relationships. The utility patent variable, which is used as a proxy for technological knowledge, even has a negative coefficient. It indicates that micropolitan entrepreneurship does not seem to be built on technological knowledge.
Among quality-of-life factors, the natural amenity scale is positively associated with micropolitan entrepreneurship at the 0.05 significant level in Model A, supporting existing literature on the critical role of natural amenities in nonmetropolitan growth (e.g., Partridge, 2010; D. Rickman & S. Rickman, 2011). Social capital shows an insignificant relationship.
Finally, adjacency to a large metropolitan region 3 positively predicts general entrepreneurship development in micropolitan areas at the 0.10 significance level. It indicates that compared with those isolated micropolitan regions, micropolitan areas adjacent to large metropolitan areas are likely to have higher general new-firm formation rates during the study period, holding other variables constant. This relationship supports possible spread or geographical spillover effects from large metropolitan regions that benefit micropolitan entrepreneurs.
Next we turn to Model B, which addresses the relationship between industrial structure and entrepreneurship in micropolitan areas. After including the LQ variables, there are some significance changes compared with the results in Model A. The positive coefficient of poverty now turns significant at the 0.05 level, indicating its compounding relationship with industrial structure. The positive impact of human capital is no longer significant after controlling for the professional service LQ variable.
The signs and significance levels of industrial structure variables are mixed. Among all two-digit NAICS code industries in Model B, all but the Construction sector have either an insignificant or a significant and negative relationship with entrepreneurship. The latter group of industries include Retail Trade, Information, Management Companies, and Health Care. The reasons behind these negative associations might vary by industries. The Management and Information sectors are export based. Kilkenny and Partridge (2009) found the share of export industries negatively associated with rural growth because they may crowd out other local economic activities. For retail and health care industries, which primarily serve the local market, a higher LQ means that the local market demand may already be met and therefore does not encourage new entries.
Comparing Micropolitan and Metropolitan Areas
Although the focus of this study is on entrepreneurship in micropolitan areas, we included results from metropolitan areas using the same explanatory variables (except that proximity to a large metropolitan area is no longer applicable and not included). This helps understand the uniqueness of the micropolitan entrepreneurship patterns.
Among demographic factors, population growth is positively associated with entrepreneurship in both metropolitan and micropolitan areas. The differences are found in population age groups and population density. Our results suggest that the population age group between 25 and 44 years has a significant, negative association with metropolitan new-firm formation rates in Models C and D, which is different from the conclusion drawn in other regional studies (Reynolds et al., 1994) that younger workforce members may contribute positively to new-firm formation. The positive coefficient of this age group is not significant in micropolitan entrepreneurship models. Due to the availability of various opportunities, especially those in well-paid incumbent businesses, the younger members of the metropolitan workforce appear to be less likely to start their own businesses. Population density is positively associated with metropolitan entrepreneurship activities but not with micropolitan entrepreneurship. Larger cities with a high density of population often signify urbanization economies that benefit entrepreneurship. By contrast, small regions, even with a high-density center like micropolitan areas, often lack the critical mass needed to have major agglomeration effects.
For economic factors, the results from establishment size and unemployment are similar between micropolitan and metropolitan entrepreneurship models. A difference is found in the poverty factor. Poverty is negatively associated with metropolitan start-up activities at the 0.01 significance level in Model C, but shows a positive relationship with general entrepreneurship in micropolitan areas in Model B. It indicates that entrepreneurial activities in metropolitan areas are more sensitive to the macroeconomic situation than in micropolitan areas.
Among knowledge factors, human capital is positively associated with new-firm formation rates at the 0.01 significance level in metropolitan areas in Model C, while it is significant at the 0.05 level in micropolitan areas as indicated in Model A, reflecting the relevance of a skilled workforce in both large and small cities. Neither utility patents nor design patents show a significant relationship with metropolitan entrepreneurship in Model C. The coefficient of utility patents is positive and significant in Model D, but only at the 0.10 significance level. As noted in Qian et al. (2013), most start-up activities in the general economy have little to do with commercializing new knowledge.
Natural amenities also show a positive association with metropolitan new-firm formation, significant only in Model C but not Model D where industrial structure is considered. Furthermore, the coefficients are not as high as those in the micropolitan models, suggesting that natural amenities play a greater role in attracting entrepreneurs in small cities than large cities. This is consistent with what has been discussed in the literature. Social capital, in contrast, shows a negative and significant relationship with metropolitan entrepreneurship in Model C. It is not a significant predictor for micropolitan entrepreneurship.
The results of industrial structure measures are largely different between metropolitan and micropolitan areas. Based on the results in Model D, LQs in Real Estate, Professional Services, and Arts, Entertainment and Recreation show positive and significant relationships with metropolitan entrepreneurial activities, while others present insignificant results at the 0.05 level. These industries are key predictors of metropolitan entrepreneurship, which can be the result of agglomeration economies and/or high demand for products and services in these industries for our study period. Different from the metropolitan scenario, the coefficients of industrial structure variables in the micropolitan entrepreneurship models are mostly negative.
Results From High-Tech Entrepreneurship
Based on results in Table 5, population growth, the population age group of 45 to 64 years, design patents, human capital, high-tech agglomeration, and adjacency to large metropolitan areas are positively associated with high-tech new-firm formation in micropolitan areas. Establishment size is negatively associated with high-tech entrepreneurship. These patterns are mostly similar to those for general entrepreneurship in micropolitan areas as shown in Table 4.
Different from general micropolitan entrepreneurship models, female employment, which is not significantly associated with general entrepreneurship, presents a significant, negative relationship with high-tech entrepreneurship in micropolitan areas. It implies the gendered difference that women are less likely to become entrepreneurs in the high-tech field. In addition, high-tech entrepreneurship in micropolitan areas is not significantly affected by the natural amenity scale, as reported in both Models E and F. It shows an interesting pattern that, while natural amenities play an important role in attracting entrepreneurs in small cities, they are a less appealing factor for high-tech entrepreneurs.
When compared with the metropolitan high-tech models in Models G and H, different results can be found in demographic factors including population density and age. Population density shows a highly significant, positive coefficient in metropolitan high-tech entrepreneurship, but it is not a significant factor in micropolitan areas. Therefore, the benefits of urbanization economies do not seem to hold for small cities. In terms of population age, the middle-age generation positively predicts micropolitan high-tech entrepreneurship but not for metropolitan areas, similar to the general entrepreneurship patterns.
Noteworthy differences between metropolitan and micropolitan areas in high-tech entrepreneurship are found in knowledge factors. The design patent variable is positively related to micropolitan high-tech entrepreneurship at the 0.05 significance level, while the utility patent shows an insignificant relationship. In contrast, high-tech new-firm formation in metropolitan areas is significantly associated with utility patents but not with design patents. The utility patent variable is often used as a proxy for technological innovation (Malecki, 2010) and found to be positively associated with metropolitan high-tech entrepreneurship in other studies (e.g., Qian, 2017b). In general, high-tech entrepreneurship at the micropolitan level does not necessarily benefit from technological knowledge but tends to be associated with human capital and creative/cultural knowledge reflected in design patents.
For SBIR, our results indicate a positive relationship with metropolitan high-tech entrepreneurship, but it is only significant in Model G and becomes insignificant after controlling for high-tech specialization in Model H. This may indicate that SBIR grants are absorbed by existing high-tech small firms rather than by high-tech entrepreneurs in metropolitan areas. Furthermore, the number of SBIR recipients shows an insignificant relationship with high-tech entrepreneurship in micropolitan areas. Consistent with the insignificant result based on utility patents, we provide more evidence that the spillover effects in technological innovation apply to larger metropolitan areas but not smaller micropolitan areas.
Summary and Policy Discussion
Entrepreneurship plays an important role in urban and regional economic development. Regional variations in business start-up rates across the U.S. metropolitan regions are well studied in the literature. Nonetheless, limited efforts have been made to understand the context of smaller urban areas. Economic development strategies in metropolitan areas that target new business development may not be equally applicable to small and peripheral regions because of the lack of agglomeration economies and the weak entrepreneurial ecosystem for the latter. Understanding factors associated with entrepreneurship development in smaller cities could help local policy makers and practitioners make better decisions to promote entrepreneurship in their communities. In this background, our study examines the spatial distributions of start-up rates across U.S. micropolitan regions. More importantly, we apply multivariate analysis to understand regional factors associated with entrepreneurship in these small cities.
Our results show that population growth and human capital are positively associated with both general and high-tech entrepreneurship in micropolitan areas, while business establishment size presents a negative association. These patterns are also found in metropolitan entrepreneurship. Furthermore, we find natural amenities to be a more important factor for general entrepreneurship in micropolitan areas than metropolitan areas, a pattern supported by existing literature on economic development in nonmetropolitan areas (Kilkenny & Partridge, 2009; Mulligan, 2013; Partridge et al., 2007). However, the natural amenity scale is not significantly associated with micropolitan high-tech entrepreneurship. Additionally, the clustering of high-tech industries positively predicts high-tech new-firm formation in micropolitan areas, while female employment is a negative predictor. Geographically, proximity to a large metropolitan area shows a positive relationship with micropolitan entrepreneurship, which indicates possible spread or knowledge spillover effects from nearby largest metropolitan areas to micropolitan regions.
Some differences are observed between micropolitan and metropolitan areas in factors such as population density, age structure, and patenting. First, population density is not significantly associated with general or high-tech entrepreneurship in micropolitan areas, but it positively predicts metropolitan start-up activities, reflecting the critical mass needed for agglomeration economies to take effect. Second, the middle age group (45 to 64 years) is positively associated with micropolitan start-up activities but not in the metropolitan case. Increasing the in-migration of this age group to small cities is likely to bring wealth and economic resources that support local new business formation. Third, the intensity of design patents is significantly associated with micropolitan high-tech entrepreneurship; by contrast, the intensity of utility patents is significantly associated with metropolitan high-tech entrepreneurship. This reflects the differentiated knowledge bases of the high-tech economy between micropolitan and metropolitan areas.
From the policy perspective, local policy makers in small cities may prioritize development strategies that impose less or no harm to their comparative advantages, including natural amenities and low density. Paradoxically, entrepreneurial activities and general economic development may exert pressure toward higher density and fewer natural amenities in small cities. Careful planning and growth management are needed to maintain the advantages of small cities. Our analysis also implies that economic development and entrepreneurship policy in small cities might be more successful when engaging with the middle age generation. Facing an aging population, many small cities are making efforts to retain young families. These efforts are not necessarily effective in fostering entrepreneurship. Instead, attracting affluent middle-age families may bring market opportunities and needed resources for entrepreneurs in small cities.
More specifically about supporting high-tech start-ups in small cities, it is clear from our study that small cities cannot adopt the same strategies with larger metropolitan areas. For instance, the agglomeration of high-tech industries is a significant predictor of high-tech entrepreneurship in micropolitan areas. However, building a strong high-tech cluster takes tremendous efforts and is unrealistic for most small communities to achieve. Moreover, our study demonstrates that the technological innovation agenda in larger metropolitan areas is not applicable in small cities. Neither utility patents nor SBIR is a significant and positive factor in micropolitan entrepreneurship. These results point to the lack of sufficient technological infrastructure and weaker entrepreneurship ecosystems in small localities (Roundy, 2017). Local policy makers should recognize this reality. Stephens et al. (2013, p. 808) have correctly suggested that, “economic development policy in lagging regions may be more successful if it focuses on cultivating existing industries or supporting homegrown entrepreneurs, and not worrying about attracting the latest ‘hot’ industry or worrying about an innovation agenda.”
Growing high-tech entrepreneurship in small cities will benefit from supporting the creative economy, as indicated by the significance of design patents. In line with this economic development strategy, McGranahan et al. (2011, p. 550) found that nonmetropolitan counties with “both higher proportions of creative class and richer entrepreneurial contexts . . . tended to have greater gains in establishments and jobs.” Promoting the creative economy has gained some popularity in small and rural communities in recent years (Markusen, 2007). Waitt et al. (2009), using a case study in Wollongong, Australia, suggested that creativity-led urban regeneration and the creative economy are embedded in small cities regardless of their size. Adding to this argument, our study indicates that creativity in small cities is also friendly to high-tech entrepreneurship.
We recognize some limitations of this research. Most of all, our cross-sectional regression analysis, even with lagged independent variables, validates associations only but not causal relationships. The changing definitions of metropolitan and micropolitan statistical areas make it difficult to integrate the temporal dimension in our regression models. One of the future research directions is to tackle the causality issue. Moreover, the exploratory nature of this research has led us to include a variety of variables without in-depth theoretical or hypothesis development. The mechanisms through which those significant factors contribute to entrepreneurship in small cities also deserve more scholarly attention. We hope this study spurs more discussion on economic development strategies that support small-city development.
Footnotes
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.
