Too Cold to Venture There? January Temperature and Immigrant Self-Employment Across the United States

Abstract

Immigrants in the United States have higher self-employment rates than native-born Americans. However, immigrant self-employment rates vary considerably across areas of the country. The authors examine the percentage of immigrant workers in local areas who are self-employed (i.e., the self-employment rate for the foreign born). Areas with colder winter temperatures have especially low self-employment rates among their immigrant populations compared to warmer areas. The relationship between winter temperature and immigrant self-employment persists after controlling for numerous individual and local area characteristics. The relationship holds for numerous subsamples of immigrants but is strongest for immigrants arriving to the United States as adults. Child immigrants and native-born Americans exhibit a weaker relationship, possibly because of previous exposure and attachment to particular locations chosen by their parents that constrain the migration responses of potential entrepreneurs. Entrepreneurial immigrants arriving to the country as adults appear especially footloose and particularly responsive to January temperatures in their location decisions.

Keywords

self-employment entrepreneurship immigrants amenities temperature J61 L26 and R23

Immigrant entrepreneurs are critical to regional and national economies (Azoulay et al., 2022). Immigrants in the United States have higher self-employment rates than native-born Americans, and immigrants have made outsized contributions as founders of numerous highly successful firms (Lofstrom & Wang, 2019). However, immigrant self-employment rates vary considerably across areas of the country in ways not previously well known. This paper documents that immigrant self-employment rates are higher in areas of the United States with warmer January temperatures and lower in colder areas. Our main measure is the percentage of foreign-born workers in an area who are self-employed. Areas in the northern parts of the country have colder winters and lower self-employment rates among immigrants. This pattern holds for numerous immigrant subsamples including new arrivals, longer-term immigrants, noncitizens, naturalized citizens, men, women, Hispanics, Asians, other races, college graduates, and noncollege graduates. Notably, the relationship is stronger for immigrants arriving as adults than for immigrants arriving as children. We also briefly examine self-employment patterns for persons native to the United States and find a weaker relationship with January temperatures than that for immigrants.¹

The relationship between immigrant self-employment rates and warmer January temperatures may be partly attributable to other characteristics of the local area or differential preferences and resources for immigrants in different areas. We use multivariate regression analysis to examine the roles of observable local areas and individual-level factors. Local area variables include amenities, the industrial structure, and the percentage of the local population that is foreign born. Individual controls include demographics, country of birth, education level, industry of employment, and other characteristics. Observable characteristics explain some of the relationship between immigrant self-employment and January temperatures, but a large portion of the relationship is unexplained by observable factors and therefore due to factors such as preferences that are not strongly related to observable individual characteristics. For example, immigrant entrepreneurs may be especially forward looking and sensitive to preferences for local amenities when deciding where to start a business because of the potential lock-in effect of the business.

The strong positive relationship between immigrant self-employment rates and January temperature is a new and important finding that, to our knowledge, has not been documented in previous literature. Natural amenities are well documented to affect location choices for the general population (Graves, 1980; Partridge, 2010; Rappaport, 2009), but their effect on immigrant entrepreneurs has been largely overlooked. Given the importance of immigrant entrepreneurs for regional economies, our study provides important new insights. Winter temperatures appear to be an especially important natural amenity for immigrants making entrepreneurial investments.

Empirical Motivation

We first document differences in immigrant self-employment rates across the United States. We combine individual years of the American Community Survey (ACS) microdata to compute the immigrant self-employment rate over the 2012–2019 period.² We limit the main sample to foreign-born persons ages 18–61 who are either self-employed or working for an employer.³ Our main measure for the foreign-born self-employment rate for area a is as follows:

F o r e i g n S e l f E m p l o m e n t R a t e_{a} = \frac{# F o r e i g n S e l f E m p l o y e d_{a}}{# F o r e i g n S e l f E m p l o y e d_{a} + # F o r e i g n P a i d E m p l o y e d_{a}}

(1)

F o r e i g n S e l f E m p l o m e n t R a t e_{a}

is a ratio with the numerator equal to the number of foreign-born self-employed workers in the area and the denominator is the sum of the number of foreign-born self-employed workers and the number of foreign-born employer-paid workers. Figure 1 maps the percentage of foreign-born workers who are self-employed for each state. It exhibits a strong pattern with lower rates in northern states, especially in the middle of the country, and higher immigrant self-employment rates in states in the South and Southwest regions of the United States.

Next, we define local areas using identifiable geography in the ACS microdata as a combination of metropolitan areas and state-specific nonmetropolitan residual areas.⁴ Figure 2 maps the foreign-born self-employment rates for metropolitan areas. Figure 3 includes all local areas. The greater geographic specificity provides additional variation relative to Figure 1, but the same pattern still holds. Immigrant self-employment rates tend to be lower in the northern parts of the country and higher in the southern parts.

The observed pattern in immigrant self-employment rates is strongly correlated with winter temperature differences. Figure 4 maps the mean of January temperatures (in Fahrenheit) by local area. Winter temperatures are colder in the North and warmer in the South. Figure 5 presents a scatterplot of the relationship between local area foreign-born self-employment rates and mean January temperatures. The scatterplot is weighted through the sum of foreign-born worker survey weights for each local area. We also include a line indicating the linear fit of the data.

Figure 5 documents that there is a strong positive relationship between the January temperature and self-employment rates of foreign-born workers. Linear regression indicates that an additional 10 degrees of mean January temperature is associated with a 1.46 percentage point increase in the self-employment rate of foreign-born workers (e.g., from 10% to 11.46%). The bivariate regression has an R-squared of 0.49. Notably, this relationship may not represent an unbiased causal effect. Colder and warmer areas differ in many ways besides temperature that could affect immigrant self-employment decisions. However, there are theoretical reasons to expect a positive effect of January temperature on immigrant self-employment that we discuss in the next section. Furthermore, the magnitude is quite large. The mean January temperature has a weighted standard deviation of 12.9 and ranges from 3.8 to 66.7 in our data. Scaling the Figure 5 coefficient estimate by the dispersion in January temperatures indicates a large influence on immigrant self-employment rates. For example, going from 10 degrees to 60 degrees raises the predicted immigrant self-employment rate from 6.6% to 13.9%.

Figures 6 and 7 provide scatterplots of self-employment rates for two groups of native-born Americans against mean January temperature. Figure 6 illustrates the relationship for persons aged 18–61 born in the United States and residing in their birth state at the time of the ACS. There is no relationship in Figure 6; the linear fit coefficient estimate is small, negative, and not statistically significant. Figure 7 illustrates the relationship for persons aged 18–61 born in the United States and residing outside their birth state.⁵ Figure 7 exhibits a significant positive relationship with a January temperature coefficient of 0.0006 and an R-squared of 0.102 (i.e., the slope is about 40% as large as that for immigrants in Figure 5). Thus, native-born Americans living in their birth state do not exhibit any relationship between their self-employment rates and January temperatures, but Americans who moved away from their home state exhibit a similar, albeit not as strong, pattern as immigrants.

Conceptual Framework

The immigrant self-employment rate in an area depends on immigrant location decisions, self-employment decisions, paid employment decisions, and their interactions. We first discuss location decisions independent of self-employment and paid employment decisions. We then discuss self-employment and paid employment decisions independent of location decisions. We then synthesize the combined decisions and discuss the observed relationship between January temperatures and immigrant self-employment rates across local areas.

Location Decisions

A long and notable literature has examined location decisions for workers and firms (Berry & Glaeser, 2005; Ellis, 2012; Graves, 1980; Greenwood et al., 1991; Mueser & Graves, 1995; Partridge & Rickman, 2003; Rijnks et al., 2018; Roback, 1982; Sjaastad, 1962). This literature argues that workers seek to maximize their own well-being subject to their endowments and constraints, and firms seek to maximize profits subject to their resources and constraints. Local areas differ in prices and location-specific amenities that affect worker utility and firm profitability. Firms prefer to pay lower input prices for labor, workspace, and physical materials, but they are willing to pay higher input prices to be in areas with productive amenities or better access to consumers (Chen & Rosenthal, 2008; Gabriel & Rosenthal, 2004). Competition in input and output markets will drive firm profits toward zero for firms making marginal location decisions. Some firm production may require location-specific investments that create barriers to entry and exit and allow their profits to deviate from zero in the short run and medium run.

All else equal, workers prefer higher wages, lower costs of living, and better location-specific amenities. Spatial equilibrium forces will cause adjustments to wages and costs of living so that marginal migrants are indifferent across areas. Better local amenities will lead to compensating differentials in labor and housing markets and result in lower “real wages,” (i.e., wages adjusted for cost of living). While marginal migrants are indifferent across areas, heterogeneous endowments and preferences will cause individuals to sort into the local area that gives them the highest possible well-being. Individuals are often strongly influenced by prior residential locations through location-specific human capital and moving costs (Deryugina et al., 2018; Kennan & Walker, 2011; Koşar et al., 2022; Krupka, 2009; Krupka & Donaldson, 2013; Ransom, 2022; Yu & Artz, 2019). Location decisions for children are made by their parents or guardians, and young people often develop attachments to familiar people, places, and activities that increase their preferences for living in or near their home area (Winters, 2020).⁶

In making location decisions, rational individuals also internalize that their current decisions will have effects later in life. Individuals can make multiple moves, but future moves involve additional moving costs. Migration is a human capital investment with costs and benefits in both the present and future (Sjaastad, 1962). Individuals form expectations about future wages, living costs, amenities, and moving costs and choose their current location to maximize their expected lifetime utility. Potential migrants may be somewhat risk averse and try to reduce the likelihood of getting stuck in a less preferred location due to future moving costs.

Foreign-born workers have already left their home area and often have less attachment to any particular area of the United States and lower costs of moving within the country. Thus, foreign-born workers are likely to be especially responsive to differences in local wages, living costs, and amenities (Cadena & Kovak, 2016).⁷ Immigrants may also have different preferences and skill endowments than native-born workers. Many immigrants may come from countries with relatively warmer climates and may have a preference to locate in warmer areas. Furthermore, adverse health effects from vitamin D deficiency may be especially acute for immigrants from warmer and sunnier countries who locate in areas of the United States with less winter warmth and sunshine (Andersen et al., 2021). Many immigrants also choose to locate in ethnic enclaves, which may enhance access to social, cultural, consumption, and employment opportunities that align with their preferences and endowments (Borjas, 2002). Finally, foreign-born college graduates are especially likely to be educated in science, technology, engineering, and mathematics (STEM) fields and would, therefore, tend to sort into tech hubs where those skills are highly rewarded. Their less-educated counterparts may be especially likely to work in less skilled jobs in agriculture, construction, and manufacturing, and may sort into areas offering those opportunities.

As regional economies evolve over time, so do prices, employment levels, and population distributions. In the last half of the 20th century and early part of the 21st century, the North and Midwest regions of the United States experienced major population redistribution from their states toward the South and West regions (Graves, 1980; McGranahan, 1999; Partridge, 2010). While preferences vary, there is a general tendency for individuals to prefer moderate temperatures and to dislike cold winters and warm summers (Mueser & Graves, 1995; Rappaport, 2009). Widespread availability and adoption of air conditioning likely made areas with warm summers relatively more attractive than before and encouraged net population flows to warmer areas (Graves, 2013; Rappaport, 2009). Less housing regulation and more affordable housing in the South have also influenced regional population flows (Glaeser & Tobio, 2008). Rising incomes have fueled increased demand for housing and locational amenities (Rappaport, 2009). Notably, recent immigrants have contributed considerably to net population flows to the South and West and are especially concentrated in these regions.

Self-Employment and Paid Employment Decisions

Workers can choose to work as either a self-employed business owner or a paid employee.⁸ Individuals will choose the employment path that offers them the highest expected lifetime utility given their endowments and constraints. In making this decision, individuals weigh the expected relative benefits including pecuniary and nonpecuniary and short-term and long-term. Individuals have heterogeneous skills, preferences, and resources that lead to some people becoming self-employed while others work in paid employment.

Self-employment is often riskier and exhibits more income variance than paid employment (Åstebro et al., 2014). Many businesses fail, and most self-employed workers end up earning less than they could in paid employment (Acs et al., 2016; Hamilton, 2000). However, some businesses are very successful and yield their proprietors substantial income.⁹ Many potential and actual small business owners are especially drawn to self-employment for the nonpecuniary benefits. They value being their own bosses, controlling their own work conditions, and setting their own schedules (Pugsley & Hurst, 2011). They are often willing to trade income for work satisfaction and accept lower income in self-employment than in paid employment for the more enjoyable work environment.

The relative benefits of self-employment and employer-paid jobs can also differ over the time horizon. Specifically, some of the benefits of self-employment can occur further into the future relative to paid employment. While many businesses do not succeed, those that do may be eventually sold at a capital gain or passed on to children many years later. Entrepreneurship exhibits intergenerational persistence with entrepreneurs often passing to their children not only their businesses but also the knowledge, skills, social networks, habits, and motivation to succeed in business ownership (Blumberg & Pfann, 2016; Hoffmann et al., 2015; Li & Goetz, 2019; Lindquist et al., 2015). A relatively long time horizon may be necessary to recoup early investments for many small businesses.

Immigrants to the United States have been documented to be especially entrepreneurial compared to native-born Americans, and this may reflect a number of factors (Azoulay et al., 2022; Borjas, 1986; Elo et al., 2018; Fairchild, 2010; Fairlie & Lofstrom, 2015; Liu, 2012; Yuengert, 1995).¹⁰ First, immigrants may typically have worse opportunities in paid employment due to factors like discrimination, credentialism, language skills, social connections, and less familiarity with U.S. employer expectations during job interviews and on the job. If immigrants are undercompensated in paid employment, self-employment may be an especially attractive option because they can make their own way and determine their own success. Second, differing self-employment rates may reflect differences in unobservables and selection into immigration. Immigrants may be less risk averse, more ambitious, more experienced with self-employment (in their home country), and more confident in their ability to succeed in self-employment compared to native-born Americans, though there is likely substantial heterogeneity within groups (Obschonka & Stuetzer, 2017). Immigrants may also be more forward-looking. For example, migration to the United States and immigrant investment decisions may be made, in part, to provide a better life for their children (Abramitzky et al., 2021). Third, immigrants locating in enclaves may exploit opportunities to provide goods and services targeted to their ethnic group through entrepreneurship (Li et al., 2018). Finally, some immigrants may not be legally authorized to work in the United States so choose self-employment to avoid detection, apprehension, and deportation (Amuedo-Dorantes et al., 2022). The motives for self-employment undoubtedly vary across individuals among both immigrants and native-born Americans.

Location and Self-Employment Decisions

Self-employment rates across areas depend on numerous factors that affect location, self-employment, and employer-paid decisions independently and jointly (Armington & Acs, 2002; Bosma & Schutjens, 2011; Feldman, 2001; Fritsch et al., 2019; Malecki, 1993; Sternberg, 2022; Yang & Zhang, 2023). Individuals will choose the combination of location and employment type that offers them the highest expected utility. Individuals with strong preferences for particular areas or amenities may make location decisions first and then choose among paid employment and self-employment options (Sorenson, 2018). Many individuals live much of their life in the same area and largely take their location as given (Winters, 2017, 2022). Alternatively, footloose entrepreneurs may be adamant about owning their own business and seek out the best location to maximize their utility as a business owner. Amenities and access to complementary inputs are likely critical factors in the location decisions of footloose entrepreneurs (Goetz & Rupasingha, 2009; McGranahan et al., 2011; Rupasingha & Marré, 2020). Other workers may be intent on paid employment and seek out the area paying wages that yield the highest utility for their skill set. Some individuals may make simultaneous decisions; for example, their top two options may be self-employment in City A and paid employment in City B. Furthermore, both location and self-employment decisions are dynamic. Individuals can change locations multiple times and make multiple transitions between employer-paid jobs and self-employment. Successful self-employment in some industries may benefit from familiarity with the local area and local business practices; individuals may move to these areas and initially work in employer-paid positions but later transition to self-employment after gaining valuable knowledge and skills.

Self-Employment Rates and January Temperatures

In this section we discuss why immigrant self-employment rates across areas may increase with mean January temperature. Recall that our preferred measure is the percentage of foreign-born workers aged 18–61 in an area who are self-employed. The percentage is restricted to immigrant workers for both the numerator and denominator in Equation (1) above. Figure 5 documents that areas with colder winter temperatures have lower immigrant self-employment rates, indicating that they are relatively unattractive locations for immigrant self-employment compared to paid employment. Various factors may explain this phenomenon.

One possible explanation is that areas with colder winters may offer relatively better earning opportunities for employer-paid immigrants than self-employed immigrants. This could occur for multiple reasons including local industry composition and network effects. For example, a local economy may have a strong manufacturing base that pays good wages but has high financial barriers to entry for potential entrepreneurs. More generally, the local industry composition may lead to high wages for employees but result in low profits for potential entrepreneurs, especially among immigrants. Network effects also may be important in areas where immigrant workers are especially concentrated in a subset of industries and even in a few large firms. This can increase opportunities for newer immigrants to find paid employment in the local area through referrals and information about job openings. Immigrant entrepreneurial investment and success in an area likely depend on how embedded the potential entrepreneur is in the local community (Sequeira et al., 2009), which can depend on how long they have resided there and the depth and breadth of their social networks. A thin local network of immigrant entrepreneurs may make it especially hard for potential immigrant entrepreneurs to make critical connections and succeed in the local business environment. Similarly, immigrant entrepreneurs may depend on a thick network of other immigrants to serve as customers, input suppliers, and employees (Drori et al., 2009).

A second possible explanation is that self-employed immigrants may have stronger preferences for warm winters than employer-paid immigrants. This could result if self-employed and paid-employed immigrants come from countries with very different climates. This could also result if self-employed immigrants have vastly greater wealth and human capital and warmth is a normal good (i.e., greater wealth and income could induce them to accept lower real income to live in warmer areas). Self-employed and employer-paid immigrants could also differ on individual characteristics in ways that generate differing preferences for winter temperatures.

Additionally, self-employed immigrants may be especially sensitive to local amenities in their initial location decisions to avoid getting locked into areas with few amenities. Starting a business is often a location-specific investment that ties an individual to the local area. Physical capital is often not portable and may have to be sold at a loss if the entrepreneur wishes to exit the local market (Chen et al., 2021; Yu et al., 2011). An individual who starts a business will need to build up a reputation and relationships with local customers, suppliers, and financing institutions. Paid employees can more easily pick up and move to a new location and find a new job. Some immigrants may be willing to work temporarily in a less preferred location as a paid employee if it is their best immediate option. But starting a business in an ex ante less-preferred location may lock the entrepreneur into the area for the long run, with consequences for future generations as well. Untethered foreign-born workers may be especially averse to tethering themselves to low amenity areas and less likely to start a business in places with cold winters.

Immigrants also differ sharply from native-born Americans in that they had some control over their initial location decisions in the United States. Birth places and childhood residences for native-born Americans are chosen by their parents and these initial location experiences can have persistent effects into adulthood. Immigrants can often choose where they want to initially reside in the United States. Of course, not all immigrants have a blank slate when it comes to choosing a location. Many immigrants come to the country as children or young adults and have their location decisions decided by family. Even immigrants arriving as adults may be influenced by the location decisions of family, friends, and other contacts. However, adult immigrants are still unique in that their initial location in the United States was not decided for them, which may allow immigrant entrepreneurs to be especially responsive to January temperature.

The conceptual framework helps motivate our subsequent empirical analysis. Individual and local area characteristics may at least partially explain the relationship between January temperatures and immigrant self-employment rates. However, preferences cannot be fully observed, and relevant aspects of preferences may not be strongly related to observable characteristics like education and country of origin. The next section discusses our empirical framework that regresses immigrant self-employment on several characteristics of local areas and individuals.

Regression Framework and Data

We use multivariate linear regression to examine the factors related to immigrant self-employment differences across areas. Specifically, we are interested in how controlling for numerous local area and individual characteristics that affect the relationship between immigrant self-employment and January temperatures. We estimate variants of the following linear probability model:

S e l f E m p l o y e d_{i a c t} = θ J a n T e m p_{a} + γ Z_{a} + β X_{i a c t} + π_{c} + δ_{t} + ε_{i a c t}

(2)

The dependent variable,

S e l f E m p l o y e d_{i a c t}

, is an indicator for individual i living in local area a, born in country c, and observed in the ACS year t that equals 1 if self-employed and 0 if employer paid; our analysis excludes persons not working in either. The main sample is also limited to individuals aged 18–61 who were born in a foreign country and reside in the continental United States during ACS years 2012–2019.¹¹ We define local areas as a combination of metropolitan areas and state-specific nonmetropolitan residential areas. Our analysis includes 401 local areas. We use survey weights and cluster standard errors by local area.

The main explanatory variable of interest, $J a n T e m p_{a}$ , is the mean January temperature in the local area. This variable is constructed using the United States Department of Agriculture (USDA) Economic Research Service's (ERS) natural amenity scale county-level data. We use 2010 county population data to construct a weighted average for each local area. $Z_{a}$ is a vector of local area characteristic controls; $X_{i a c t}$ is a vector of individual characteristic controls; $π_{c}$ includes origin country fixed effects; $δ_{t}$ includes survey year fixed effects; and $ε_{i a c t}$ is a mean zero error term. The local area characteristics include additional data on natural amenities from the USDA ERS and described in McGranahan (1999). Labor market characteristics and industry composition are computed from the pooled 2000 census 5% sample and 2005–2011 ACS obtained from IPUMS (Ruggles et al., 2021), the 1980–2010 population growth rate, the natural log of the 2010 Core Based Statistical Area (CBSA) population, log median housing values computed from the pooled 2000 census and 2005–2011 ACS, and three variables measuring urban hierarchy proximity.

We estimate variants of Equation (2) with and without individual controls. The regressions without individual controls exclude year and country of origin fixed effects and are, therefore, equivalent to the following local-area-level linear regression equation,

F o r e i g n S e l f E m p l o m e n t R a t e_{a} = θ J a n T e m p_{a} + γ Z_{a} + ε_{a}

(3)

where

F o r e i g n S e l f E m p l o m e n t R a t e_{a}

is computed as in Equation (1) and the regression is weighted using the sum of individual survey weights in each local area as shown in Figure 5.¹²

Table 1 presents sample means for all local area variables in the main specification. Additional natural amenities from the USDA ERS include mean January sunlight hours, mean July temperatures, mean July relative humidity, mean topography scores, and percentage of water area. We pool the 2000 census and 2005–2011 ACS to compute regression-adjusted relative incomes between the self-employed and employer- paid, labor force participation (LFP), and unemployment. The latter two variables are local area fixed effects from linear probability model regressions of immigrant LFP and unemployment that control for detailed indicators for survey year, age, sex, race, Hispanic ethnicity, and highest education level. The relative income measure involves regressions of immigrant log annual income on local area fixed effects, survey year, age, sex, race, Hispanic ethnicity, and education level, estimated separately for immigrants working in self-employment and paid employment. We then compute the relative income variable as the local area fixed effects for the self-employed minus the local area fixed effects for the paid employed. We also use the pooled Census/ACS to compute the percentage of foreign-born workers in each local area employed in agriculture, mining, construction, manufacturing, transportation, communications, and utilities, wholesale trade, retail trade, and services; the omitted industry category is public administration and national defense. We use the Census/ACS data to compute median housing values (adjusted over time for inflation through the Consumer Price Index) among owner-occupants and then convert to logs. We use decennial census data to compute 1980–2010 population growth. The 2010 CBSA population is obtained from IPUMS. CBSAs include both metropolitan and micropolitan areas; each of our local areas has at least one CBSA. We include three variables measuring proximity to the urban hierarchy: distance to the nearest metropolitan area with population greater than 250,000 (250 K), incremental distance to the nearest metropolitan area with population greater than 500,000, and incremental distance to the nearest metropolitan area with population greater than 1.5 million (1500 K). These proximity variables are included following findings in Partridge et al. (2008, 2009) that proximity to the urban hierarchy influences labor markets through commuting, consumption, and trade flows. Finally, we also include a control variable for the percentage of local area population that is foreign born in the pooled 2000 census and 2005–2011 ACS. This latter variable is highly aggregated and does not capture the influence of specific co-ethnic populations. We include additional variables discussed below as individual characteristics that depend on an individual's country of birth and primary language spoken at home.

Table 1.

Selected Variable Means for Immigrant Self-Employed and Employer-Paid.

	Self-employed	Employer-paid
Local area characteristics
Mean January temperature (Fahrenheit)	43.82	41.38
Mean January sunlight hours	173.97	168.35
Mean July temperature (Fahrenheit)	76.04	75.76
Mean July relative humidity	60.39	59.35
Mean ERS topography score	9.52	9.44
Percent water area	12.53	12.15
Regression-adjusted SE/PE relative income	0.055	0.055
Regression-adjusted labor force participation	−0.001	0.001
Regression-adjusted unemployment	0.016	0.015
Share of local employment in agriculture	0.038	0.040
Share of local employment in mining	0.003	0.002
Share of local employment in construction	0.100	0.097
Share of local employment in manufacturing	0.127	0.132
Share of local employment in trans., comm., & util.	0.064	0.062
Share of local employment in wholesale trade	0.034	0.032
Share of local employment in retail trade	0.179	0.177
Share of local employment in services	0.428	0.428
Percentage population change, 1980–2010	0.624	0.599
Ln CBSA population, 2010	15.05	14.93
Ln median housing values	12.23	12.22
Distance to nearest metro w/ pop. > 250K	4.07	5.02
Incremental dist. to nearest metro w/ pop. > 500K	4.71	5.15
Incremental dist. to nearest metro w/ pop. > 1500K	21.30	21.77
Share of local population foreign born	0.282	0.263
Selected individual characteristics
Ln distance from local area to home country	7.890	7.921
Share of local population from own home country	0.053	0.050
Share of local population same foreign language	0.123	0.106
Age	44.49	40.57
Years in the USA	21.32	18.85
Naturalized citizen	0.494	0.471
Female	0.378	0.438
Married	0.679	0.604
Number of own children	1.291	1.105
High school diploma	0.249	0.225
Some college but less than a bachelor's degree	0.203	0.210
Bachelor's degree	0.166	0.185
Master's degree	0.065	0.099
Professional degree	0.033	0.021
Doctorate degree	0.014	0.025
Black and not Hispanic	0.055	0.093
Asian and not Hispanic	0.223	0.261
Other race and not Hispanic	0.020	0.018
Hispanic	0.491	0.482
Speaks only English at home	0.133	0.150
Speaks another language and English very well	0.349	0.400
Speaks another language and English well	0.259	0.215
Speaks another language and English but not well	0.196	0.170
Homeowner	0.486	0.388
Homeowner × home value (in $1000s)	235.0	153.3
Family member income (in $1000s) excluding own income	43.75	43.47

Note. The sample includes 1,527,208 individuals aged 18–61 who were born in a foreign country and reside in the United States during ACS years 2012–2019. Local areas are a combination of metropolitan areas and state-specific nonmetropolitan residual areas.

These additional local area characteristics are intended primarily as control variables, though the signs for some are difficult to predict. However, we do have some expectations for many of these variables. For example, July temperature and humidity are expected to be disamenities and reduce immigrant self-employment rates, while the other ERS variables are amenities and expected to increase immigrant self-employment rates. The relative income variable increases with self-employment earnings and decreases with employer-paid earnings and should make self-employment more attractive. However, relative incomes are endogenously determined through supply and demand for entrepreneurs and employees, so a positive relationship might not hold. Higher regression-adjusted immigrant LFP rates may reflect a stronger local labor market and reduce the likelihood that immigrants turn to self-employment out of necessity. However, a stronger labor market may also attract both self-employed and employer-paid migrants to the area, and the relative effect is unclear a priori. Local unemployment is expected to have opposite effects to the LFP but the expected effect is again somewhat unclear. Industrial structure is likely to be a partial factor. For example, an area with relatively high manufacturing employment may be more attractive to paid employees than the self-employed. Alternatively, areas with relatively high retail employment may be especially attractive to self-employed immigrants who can open their own businesses with relatively low barriers to entry.

To the extent that self-employed immigrants are more forward looking than the employer-paid, one might expect positive effects of prior population growth and CBSA population and negative effects from increasing distance to the urban hierarchy. Housing value effects are largely ambiguous. Individual housing wealth could help finance business ventures, but the self-employed may seek out areas where they can afford to buy a home. Finally, the share of the local population of immigrants is expected to increase immigrant self-employment rates because immigrant entrepreneurs can create businesses to provide goods and services targeted to the tastes of immigrants that are not well served by mainstream businesses.

Table 2 presents bivariate correlation coefficients for the immigrant self-employment rate and mean January temperature between each other and with the other local area characteristic variables. The correlations use local area weights computed as the sum of individual weights for foreign-born workers in the area. The correlation between the immigrant self-employment rate and mean January temperature is 0.703, which is the strongest correlation in the table. January temperature is also strongly but imperfectly correlated with January sunlight with a correlation coefficient of 0.649. Cold places vary in the amount of sunlight received, which may impact the desirability and immigrant self-employment rates. Our regression analysis includes both January temperature and January sunlight. Table 2 also indicates that the immigrant self-employment rate and the share of the local population that is foreign born has a coefficient of 0.519, possibly suggesting notable influences of co-ethnic complementarities on self-employment decisions. The other correlations in the table vary in magnitude including some that are negative, and others close to zero. Overall, the correlations suggest a unique and important relationship between the immigrant self-employment rate and mean January temperature, but the multivariate regression results will provide more rigorous analysis.

Table 2.

Local Area Characteristic Correlations with Immigrant Self-Employment Rate and January Temperature.

Correlation with:	Immigrant self-employment rate	Mean January temperature
Mean January temperature (Fahrenheit)	0.703
Mean January sunlight hours	0.498	0.649
Mean July temperature (Fahrenheit)	0.169	0.258
Mean July relative humidity	0.263	0.217
Mean ERS topography score	0.040	0.169
Percent water area	0.130	−0.128
Regression-adjusted SE/PE relative income	0.010	0.128
Regression-adjusted labor force participation	−0.237	−0.174
Regression-adjusted unemployment	0.040	0.128
Share of local employment in agriculture	−0.114	0.141
Share of local employment in mining	0.093	0.097
Share of local employment in construction	0.260	0.348
Share of local employment in manufacturing	−0.293	−0.271
Share of local employment in trans., comm., & util.	0.315	0.192
Share of local employment in wholesale trade	0.427	0.375
Share of local employment in retail trade	0.358	0.176
Share of local employment in services	−0.013	−0.210
Percentage population change, 1980–2010	0.170	0.434
Ln CBSA population, 2010	0.279	0.054
Ln median housing values	0.090	0.074
Distance to nearest metro w/ pop. > 250K	−0.169	−0.155
Incremental dist. to nearest metro w/ pop. > 500K	−0.081	−0.002
Incremental dist. to nearest metro w/ pop. > 1500K	−0.029	−0.079
Share of local population foreign born	0.519	0.497

Note. The analysis includes 401 local areas. Local areas are a combination of metropolitan areas and state-specific nonmetropolitan residual areas.

Individual controls in Equation (2) include a combination of continuous and indicator variables. We include continuous variables for the natural log of distance from the local area to the immigrant's home country, the share of the local area population that is from the immigrant's home country, and the share of the local population that speaks the same language at home as the immigrant; this last variable is coded as zero for immigrants who only speak English.¹³ Immigrant entrepreneurs may prefer to start businesses in areas closer to their home country, but negative effects of distance may be minimal since immigrants are typically already travelling very long distances. The other two variables are intended to capture different dimensions of enclave effects. Enclave effects may be especially strong for immigrants from the same country, but a common language and similar culture may extend enclave effects. We also include continuous variables for real family income excluding the individual's own income and own housing value for homeowners; the latter variable equals zero for renters.¹⁴ The full model also includes indicator variables for homeownership, sex, survey year, origin country, age, race, education level, college major, citizenship status, years in the United States, English ability, marital status, number of children, age of youngest child in the household, and industry, and detailed interaction terms for interactions of a female indicator variable with indicators for survey year, origin country, age, race, education level, college major, citizenship status, years in the United States, English ability, marital status, number of children, age of youngest child in the household, and industry.¹⁵

Table 1 also presents sample means for selected individual variables. Categorical variables for origin country, college major, industry, and age of youngest child are excluded from Table 1 due to limited space and the large number of indicators. Similarly, Table 1 reports the mean for age, years in the United States, and number of children, but the regression analysis includes detailed indicator variables for these. We also include more detailed indicators for marital status and education level than the variables reported in Table 1. Not surprisingly, there are some notable differences in individual characteristics between self-employed and employer-paid immigrants. For example, the self-employed are older and have lived more years in the United States. They are also more likely to be male, married, and homeowners.

Regression Results

Main Results

Table 3 presents our main regression results. Column 1 reports an individual regression that only includes mean January temperature as an explanatory variable. This yields a coefficient of 0.00146, the same as the linear fit for Figure 5. Column 2 includes local area characteristics controls but not individual controls. Adding the local area controls reduces the coefficient on January temperature to 0.00089 in column 2, but the coefficient is still statistically significant and has a large and important magnitude. At face value, this suggests that local area characteristics explain 39% of the raw relationship between immigrant self-employment rates and January temperature. Adding individual controls further reduces the January temperature coefficient to 0.00082 in column 3, but it is significant and still meaningfully large; 56% of the raw relationship between immigrant self-employment rates and January temperature remains unexplained. Recall that the weighted standard deviation for January temperature is 12.9. Multiplying by the 0.00082 coefficient, a one standard deviation increase in January temperature corresponds to an increase in immigrant self-employment of 0.0106. Approximately 11.2% of immigrant workers are self-employed, so an increase of 1.06 percentage points due to a one standard deviation increase in January temperature corresponds to roughly 10% of the sample mean.

Table 3.

Individual-Level Regressions of Immigrant Self-Employment Probability.

	(1)	(2)	(3)
Mean January temperature	0.00146***	0.00089***	0.00082***
Mean January temperature	(0.00021)	(0.00021)	(0.00013)
Mean January sunlight hours		0.00004	0.00005
Mean January sunlight hours		(0.00004)	(0.00003)
Mean July temperature		−0.00057	−0.00080**
Mean July temperature		(0.00057)	(0.00032)
Mean July relative humidity		0.00018	0.00003
Mean July relative humidity		(0.00013)	(0.00008)
Mean ERS topography score		0.00067**	0.00027
Mean ERS topography score		(0.00033)	(0.00019)
Percent water area		0.00044***	0.00016*
Percent water area		(0.00017)	(0.00009)
Regression-adjusted self-emp/emp-paid relative income		−0.0162	−0.01121
Regression-adjusted self-emp/emp-paid relative income		(0.01172)	(0.00710)
Regression-adjusted labor force participation		−0.08813*	−0.10216***
Regression-adjusted labor force participation		(0.05090)	(0.02898)
Regression-adjusted unemployment		0.09368	−0.08106
Regression-adjusted unemployment		(0.10007)	(0.06639)
Share of local employment in agriculture		−0.07793*	−0.01331
Share of local employment in agriculture		(0.04665)	(0.03191)
Share of local employment in mining		0.01767	0.08074
Share of local employment in mining		(0.17254)	(0.11215)
Share of local employment in construction		0.11881*	−0.03817
Share of local employment in construction		(0.06357)	(0.03575)
Share of local employment in manufacturing		−0.07989	−0.00613
Share of local employment in manufacturing		(0.05164)	(0.02923)
Share of local employment in trans., comm., & util.		−0.19918	−0.27087***
Share of local employment in trans., comm., & util.		(0.13572)	(0.07626)
Share of local employment in wholesale		0.45637***	0.43105***
Share of local employment in wholesale		(0.17325)	(0.12035)
Share of local employment in retail		0.24273***	0.12948**
Share of local employment in retail		(0.08335)	(0.05062)
Share of local employment in services		0.02443	0.01607
Share of local employment in services		(0.04667)	(0.02820)
Percentage population change, 1980–2010		−0.00365	0.00095
Percentage population change, 1980–2010		(0.00420)	(0.00262)
Ln CBSA population, 2010		0.00181	−0.00122
Ln CBSA population, 2010		(0.00216)	(0.00118)
Ln median housing values		−0.02188***	−0.02198***
Ln median housing values		(0.00803)	(0.00499)
Distance to nearest metro w/ pop. > 250K		0.00005	0.00006
Distance to nearest metro w/ pop. > 250K		(0.00006)	(0.00004)
Incremental dist. to metro w/ pop. > 500K		0.00001	1.14E–06
Incremental dist. to metro w/ pop. > 500K		(0.00005)	(0.00003)
Incremental dist. to metro w/ pop. > 1500K		0.00003	0.00002
Incremental dist. to metro w/ pop. > 1500K		(0.00002)	(0.00001)
Share of local population foreign born		0.05892*	0.01899
Share of local population foreign born		(0.03530)	(0.01962)
Ln distance from local area to home country			0.00151
Ln distance from local area to home country			(0.00513)
Share of local pop. from own home country			−0.01574
Share of local pop. from own home country			(0.03273)
Share of local population same foreign language			0.06689***
Share of local population same foreign language			(0.01108)
Additional controls	No	No	Yes

Note. Standard errors are clustered by local area. Additional controls include continuous variables for family member income (excluding own income) and housing values for homeowners (zero for renters), and indicator variables for homeownership, sex, survey year, origin country, age, race, education level, college major, citizenship status, years in the United States, English ability, marital status, number of children, age of youngest child in the household, and industry, and detailed interaction terms for interactions of a female indicator variable with indicators for survey year, origin country, age, race, education level, college major, citizenship status, years in the United States, English ability, marital status, number of children, age of youngest child in the household, and industry. *Significant at 10% level; **Significant at 5% level; ***Significant at 1% level.

Results for the local area characteristic controls are suggestive but sometimes inconsistent across the two specifications. The other natural amenity variables all have expected signs except for July humidity, but only the percentage of water area is statistically significant in both columns 2 and 3. The regression-adjusted labor force participation variable has a consistently negative and significant coefficient that is consistent with stronger local labor markets being relatively more attractive to the paid employed. Employment shares in wholesale and retail both have consistently positive coefficients. The latter may reflect low barriers to entry in retail and the wholesale coefficient may reflect supply linkages between foreign-born retail and wholesale entrepreneurs. Log median housing values have a negative coefficient, consistent with immigrant entrepreneurs especially preferring locations where they can purchase an affordable home. The share of the local population that is foreign born has a positive coefficient in column 2 that is significant at the 10% level, but the coefficient shrinks and loses significance in column 3.

We report results for the three continuous individual control variables related to distance and enclave effects in column 3. We do not report the other individual characteristic results because they are very numerous. Our focus is on spatial variables. The log of distance from the individual's home country to their local area and the share of the local population from their home country are both not statistically significant. However, the share of the local population speaking the same foreign language as the individual has a significant positive effect. These results suggest that positive enclave effects on self-employment are primarily language based and not based strictly on national origin or on the share of the population that is foreign born.

Alternative Samples and Heterogeneity

We next examine heterogeneous impacts of January temperature on self-employment for several alternative samples through regression coefficients shown in Figure 8. Each bar of Figure 8 is from a separate regression and indicates the January temperature coefficient estimate on a self-employment indicator dependent variable. The specification is otherwise the same as column 3 of Table 3 except it is limited to the specific subsample indicated. Some controls are excluded to prevent perfect collinearity for the particular subsample. The coefficient estimates are significant at the 1% level for all subsamples in Figure 8. Standard errors are reported in supplementary online Appendix Table A1.

Figures 5 through 7 indicate that the bivariate relationship between self-employment and January temperature is much stronger for immigrants than for native-born Americans. The bivariate relationship differs among native-born Americans by whether the individual resides in their birth state. To facilitate comparison to the main regression results for immigrants in Table 3, we conduct similar multivariate regression analyses for native-born Americans and again separate them into those who do and do not reside in their birth state. Figure 8 indicates that both groups of native-born Americans have self-employment decisions significantly related to January temperature after including the other regression controls. However, the coefficient magnitudes are notably different. The coefficient estimates are 0.00027 for birth-state residents and 0.00043 for birth-state nonresidents. These are both much smaller than the baseline estimate of 0.00082 for the full immigrant sample. Thus, controlling for other factors in the regressions, both immigrants and natives have self-employment rates positively related to warmer winter temperatures, but the effects are much larger for immigrants than natives and larger for native migrants than native nonmigrants.

We next explore several alternative samples of immigrants with results for the January temperature coefficient in Figure 8. We first divide immigrants into groups based on the age at which they arrived in the United States. We consider immigrants arriving between birth and age 14 as child immigrants whose initial location decisions were overwhelmingly decided by their parents or other family members. Persons arriving between ages 15 to 24 have more say in their migration decision, and persons arriving after age 25 are adults who may be able to make location decisions independent of their parents; though many new immigrants join relatives who immigrated earlier and provide support to the newcomers (e.g., finding work and housing). Notably, we find that the January temperature coefficient on the self-employment probability varies in a systematic way. The coefficients are 0.00042 for immigrants arriving between ages 0–14, 0.00080 for those arriving between ages 15–24, and 0.00104 for immigrants arriving at the age of 25 and older. Thus, the relationship between January temperature and immigrant self-employment strongly differs between immigrants arriving as children and adults, with the effect strongest for adult immigrants. One plausible explanation is that adult immigrants interested in self-employment disproportionately choose areas with warmer winters. However, child immigrants have initial location decisions chosen by their parents, with little concern for the child's future entrepreneurial intentions. Child immigrants form networks and attachments that make their future location decisions sticky and may reduce the relationship between self-employment decisions and January temperature. Child immigrants are somewhat like native-born Americans in that they did not choose their initial locations and have January temperature coefficients more similar to native-born Americans than to immigrants arriving as adults.

We next examine potential differences among immigrants based on the amount of time living in the United States. Notably, immigrants living in the country for less than 1 year have a coefficient that is close in magnitude to the coefficient for the full immigrant sample. The coefficient estimates appear somewhat smaller for those who have been in the United States for 6–10 and 11–15 years compared to the full sample, but standard errors are such that the confidence intervals largely overlap. The coefficients for those in the United States for 16–20 and 21-plus years are very similar to the new arrivals and the full sample coefficient. The most notable result here is that the self-employment probability of newly arriving immigrants is very responsive to January temperature. Newly arriving immigrant entrepreneurs are disproportionately drawn to areas with warmer winters.

Figure 8 also reports that the relationship between January temperature and the probability of self-employment is large and significant for both noncitizens and naturalized citizens, though the effect appears larger for noncitizens. Additionally, we split the sample into three mutually exclusive race and ethnic groups comprising Hispanic, non-Hispanic Asian, and persons neither Hispanic nor Asian. The coefficient estimates are significant for all groups but are largest for persons who are neither Hispanic nor Asian, though these are all very heterogeneous groups.

We next consider heterogeneity by country of origin. We first separately examine immigrants from Canada and Mexico. We also consider immigrants from China and India, the two most populous countries in the world and important sources of high-skilled immigrants to the United States. Finally, we collect winter temperature for all origin countries and divide immigrants based on whether the winter temperature in their origin country is above or below the median. All these origin country subsamples yield positive and statistically significant coefficient estimates for the effect of January temperature on immigrant self-employment probability. However, there is some variation. The coefficient for Canadians is 0.00140, while that for Mexican immigrants is only 0.00052. The coefficient for Chinese is 0.00095, while that for Indian immigrants is only 0.00032. Finally, the coefficient estimates are similar for immigrants with origin country winter temperatures above and below the median but perhaps slightly larger for those below the median. We do not have strong explanations for differences by country of origin. However, it is notable that the effect is not simply driven by self-employed immigrants from warmer winter countries being especially averse to cold winter temperatures in the United States. If anything, self-employed immigrants from colder countries may exhibit a stronger relative response. For example, Canada is a cold winter country and self-employed immigrants from Canada appear especially attracted to warmer areas of the United States. More generally, it is especially notable that the coefficient is positive and significant for the four countries of origin and the two winter temperature groups. Thus, the results are broad-based and not driven by any particular country or group.

Figure 8 also reports that men and women have similar coefficients with maybe a slightly larger coefficient for women. Noncollege graduates have larger coefficients than college graduates, but both are positive and statistically significant. Renters have slightly larger coefficient estimates than homeowners. We next include immigrants residing in Alaska and Hawaii but exclude controls for other natural amenities due to lack of data¹⁶; results are very similar to the main specification. We also expand the main sample to include nonworkers and code them as not self-employed; this makes the coefficient slightly smaller but does change the qualitative results. We next restrict the main sample to exclude individuals working in agriculture and construction industries, which are both seasonal and may draw different types of immigrants; results are similar to the main sample. Finally, we restrict the sample to ages 25–54, which is sometimes characterized as prime ages for attachment to the workforce; results are very similar to the full sample.

Industry-Specific Self-Employment Impacts

We next take a deeper look into the relationship between January temperature and self-employment by industry with results in Table 4. We first replicate Table 3 column 3 but exclude all industry-related explanatory variables; results are in column 1, Panel A of Table 4. The additional panels in Table 4 use industry-specific explanatory variables measuring the joint outcome of being self-employed and in a specific industry. For example, the dependent variable in Panel B equals 1 if the individual is self-employed and in the agriculture industry, and zero otherwise. The column 1 results in Table 4 indicate that January temperature has a statistically significant effect on industry-specific self-employment for six of the eight broad industries considered including agriculture, mining, construction, wholesale, retail, and services. There is no significant effect for manufacturing nor transportation, communication, and utilities. The broad significance in six of eight industries indicates that the January temperature impact is broad-based and not due to a single industry.

Table 4.

Jan. Temp. Impacts on Industry-Specific Immigrant Self-Employment (Excluding Industry Controls).

	(1)	(2)	(3)
	Industry-specific	Industry-specific	Industry-specific
	Self-emp. Jan. temp.	Self-employment	Self-employment
	Regression results	Sample means	Reg. coeff / Sample mean
A. Any industry	0.00122***	0.11194	0.01090
A. Any industry	(0.00017)	0.11194	0.01090
B. Agriculture	0.00015***	0.00644	0.02329
B. Agriculture	(0.00003)	0.00644	0.02329
C. Mining	0.000004*	0.00007	0.05685
C. Mining	(0.000002)	0.00007	0.05685
D. Construction	0.00016**	0.02028	0.00789
D. Construction	(0.00007)	0.02028	0.00789
E. Manufacturing	0.00001	0.00285	0.00351
E. Manufacturing	(0.00001)	0.00285	0.00351
F. Trans. comm., & utilities	0.00002	0.00999	0.00200
F. Trans. comm., & utilities	(0.00003)	0.00999	0.00200
G. Wholesale trade	0.00003**	0.00253	0.01185
G. Wholesale trade	(0.00001)	0.00253	0.01185
H. Retail trade	0.00021***	0.01550	0.01355
H. Retail trade	(0.00004)	0.01550	0.01355
I. Services	0.00063***	0.05427	0.01161
I. Services	(0.00009)	0.05427	0.01161

Note. Each panel includes the full foreign-born sample as in Table 3 but uses a different dependent variable and excludes all industry variables from the controls; other controls are the same as Table 3 column 3. The Panel A dependent variable is an indicator for self-employment in any industry, which corresponds to the main self-employment indicator in other tables. Panels B–I use industry-specific self-employment indicators (i.e., the dependent variable in Panel B equals 1 if the individual is self-employed AND in the agriculture industry, and zero otherwise. The column 1 January temperature coefficients in Panels B–I add up to the coefficient in Panel A subject to slight rounding error, and column 2 sample means in Panels B–I add up to the sample mean in Panel A). Column 3 illustrates relative impacts computed as the industry-specific regression coefficients divided by their sample means.

The coefficients in Table 4 column 1 are somewhat hard to compare because some industries are larger than others. Column 2 includes the sample means for the overall self-employment variable in Panel A and for the industry-specific self-employment indicators in Panels B through I. Services account for roughly half of self-employed immigrants. Mining self-employment is very small overall. To help compare industry-specific impacts, column 3 reports industry-specific coefficients divided by corresponding sample means. Relative impacts appear particularly large for agriculture and mining, though their overall impacts were small, because of very small sample means. The relative impacts for services, retail, and wholesale are similar to each other and to that for overall self-employment in Panel A. Thus, there are some differences but also some similarities in relative impacts by industry.

Additional Analysis

The ACS has only partial information on migration history. For example, we do not know the initial destinations of immigrants to the United States, and we do not generally observe intermediate locations. We observe their country of birth, their current location in the United States, and their location 1 year prior to the survey. Additionally, we know an individual's self-employment status during the survey period but nothing prior. While limited, this does allow us to examine self-employment differences between immigrants who have lived in the same local area for at least 1 year and those who moved to their current local area in the previous 12 months. We refer to the latter group as recent migrants. Table A2 in the supplementary online appendix reports three regression columns using the same immigrant sample as the main analysis. The first column regresses the self-employment dummy on an indicator for being a recent migrant and a constant term with no controls. The recent immigrant indicator is negative and significant indicating that recent migrants are considerably less likely to be self-employed. However, it is important to note that the relationship likely flows in both directions. Starting a business in the past is likely to reduce subsequent migration and increase the likelihood of future self-employment. Adding detailed additional controls in column 2 and adding local controls in column 3 reduces the magnitude of the negative coefficient for the recent migrant dummy, but it is still negative and significant. These results are admittedly difficult to interpret, so we relegate them to the supplementary online appendix, but the fact that recent migrants have lower self-employment rates is consistent with self-employment creating frictions that lower migration. Migration frictions from self-employment may deter individuals from pursuing self-employment in a local area unless they are confident they want to be there long term.

As another exercise, we examine the effects of controlling for state economic freedom measures from the Fraser Institute.¹⁷ Results for the January temperature variable are reported in supplementary online Appendix Table A3. Consistent with our specification for the local area characteristic controls, we only control for cross-sectional variation in economic freedom prior to our sample period by using the data for 2010. This is intended to capture long-term differences in economic policy but not recent changes. Because these economic freedom measures are available by state but not for our local areas, we can only account for state-level differences. Thus, our analysis is not intended to rigorously examine the effects of economic freedom on immigrant self-employment. We are simply examining the robustness of our main result to cross-sectional controls for economic freedom. The results in supplementary online Appendix Table A3 indicate that the main result is very robust to these controls for economic freedom. Thus, the main result is not simply driven by southern states being warmer and more business friendly.

We also consider the relationships with occupation and industry. We use the full (immigrant and native) sample to calculate the share of workers who are self-employed by occupation and industry, then match these percentages to individuals based on their own occupation and industry and include this as another explanatory variable in the immigrant self-employment regressions. As expected, this variable is a strong predictor of immigrant self-employment. The coefficient in Table A4 in the supplementary online appendix is 0.945 and is statistically significant at the 1% level. Notably however, adding this control only modestly impacts the coefficient on January temperature; it goes from 0.00082 in Table 3 column 3 to 0.00077 in supplementary online Appendix Table A4. Given that this variable is likely endogenous and given the challenges with measuring occupation for self-employed persons, we relegate Table A4 to the supplementary online appendix.

Conclusion

This paper documents a new and important relationship between immigrant self-employment rates and winter temperature. Immigrant entrepreneurs are disproportionately drawn to areas with warmer winters. All immigrant subsamples examined have higher self-employment rates in areas with warmer January temperatures, but we find some evidence of heterogeneous responses with immigrants arriving to the United States as adults exhibiting a much stronger response than immigrants arriving as children. We also find that persons born in the United States exhibit some relationship between self-employment and January temperature, but the relationship is generally weaker for native-born Americans than immigrants. Among self-employed native-born Americans the effects are stronger for birth-state out-migrants than for persons living in their birth state.

Our conceptual framework suggests that the observed relationship between January temperature and immigrant self-employment rates may be correlated with better paying employment opportunities relative to self-employment opportunities. Controlling for a large set of local area characteristics, including industrial structure, partially explains the January temperature coefficient on immigrant self-employment, but a large portion remains unexplained. Thus, there is some support for local economic conditions as an explanation but not the only explanation. Our framework also suggests that self-employed immigrants may have stronger preferences for warm winters than employer-paid immigrants. This motivates us to control for numerous individual characteristics including country of origin, education, housing, wealth, and enclave effects. These factors often affect an individual's likelihood of being self-employed. For example, we find evidence of increased immigrant self-employment from enclave effects measured by the share of the local population speaking the same language as the individual. Linguistic compatibility may enhance networks and increase entrepreneurship through access to suppliers, customers, and mentors. However, adding observable individual characteristics as regression controls only moderately alters the relationship between January temperature and immigrant self-employment rates. Thus, much of the observed relationship between January temperature and immigrant self-employment remains unexplained and is therefore attributable to unobservable factors such as preferences not strongly tied to the observable individual characteristics we included as explanatory variables. For example, immigrant entrepreneurs may be especially forward-looking and averse to getting locked into areas with cold winters and few amenities.

Our conceptual framework also suggests that prior location experiences likely create local attachments that reduce the responsiveness of some potential entrepreneurs to January temperature. This may explain why immigrants arriving as adults have a stronger relationship between self-employment and January temperature than native-born Americans and child immigrants. Native-born Americans and child immigrants may be somewhat attached to local areas where they lived previously. Entrepreneurial immigrants arriving to the United States as adults have the most control over their initial location decisions and seem particularly responsive to January temperature.

Our findings have important implications for policy makers. Amenities are an important factor in attracting entrepreneurs from outside the local area, with especially strong effects among adult immigrant entrepreneurs. While local areas cannot easily change their winter temperatures, they can work to make cold seasons more attractive or at least more tolerable to outsiders. They can invest more heavily in nonclimate amenities that potential residents value. Policy makers can seek to create an inclusive, welcoming community and business environment to attract potential entrepreneurs and improve quality of life. Policy makers may also be able to help migrant entrepreneurs achieve better visibility and enhanced relationships with customers, suppliers, and other local individuals and entities who can help their businesses succeed.

Potential entrepreneurs in the process of choosing a location often have incomplete information about the bundle of amenities and quality of life that various areas offer. Local amenities are often such that an individual cannot assess their quality or desirability before they experience them. Enhanced marketing to potential entrepreneurs is potentially useful for communities with colder winters, but information does not easily substitute for experience. State and regional organizations interested in enhancing economic development in areas with colder winters may need additional efforts focused on having potential entrepreneurs actually visit or temporarily reside in their areas such as during conferences, training programs, networking events, and recreational activities.

To the extent that policy leaders can quantify the benefits of immigrant entrepreneurs to their local communities, there is some potential rationale for incentives. However, an abundance of caution is likely warranted. The current study does not provide formal estimates of moving costs, but previous research suggests that these are likely to be substantial (Kennan & Walker, 2011; Ransom, 2022). A relatively modest locational incentive likely would not solicit a significant migration response and the benefits would flow disproportionately to inframarginal migrants (i.e., immigrants who would start a business there anyway). Nationally administered place-based policies such as immigrant visas for starting businesses in disadvantaged local areas may have more potential, but they also involve important opportunity costs and may have unintended consequences. Furthermore, national-level policies are outside the control of local leaders. There is still much that is unknown about whether and how such policies can be effectively designed and implemented. More research and policy experimentation seem warranted.

Policy leaders in areas with colder winters may find nurturing “homegrown” entrepreneurs to be more effective than trying to attract outside entrepreneurs. Homegrown entrepreneurs need not be the only people who have lived their entire lives in an area. Persons who moved to the area for work or higher education or family reasons, including immigrants, may have developed a taste for the local amenities. Nurturing homegrown entrepreneurs through accelerators, training, and support programs is still an emerging area of interest to research and policy communities. Future research is clearly warranted on the best strategies to nurture homegrown entrepreneurs, but it is also important to recognize that the best strategies may differ across areas. Even homegrown entrepreneurs may eventually leave an area they view as less preferable if they can easily do so. Thus, policy makers nurturing homegrown entrepreneurs in colder areas should recognize the future mobility of potential entrepreneurs and structure investments and incentives to maximize the return on their investments.

Supplemental Material

sj-docx-1-edq-10.1177_08912424241271142 - Supplemental material for Too Cold to Venture There? January Temperature and Immigrant Self-Employment Across the United States

Supplemental material, sj-docx-1-edq-10.1177_08912424241271142 for Too Cold to Venture There? January Temperature and Immigrant Self-Employment Across the United States by Jun Yeong Lee and John V. Winters in Economic Development Quarterly

Footnotes

Declaration of Conflicting Interests

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

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

John V. Winters

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

John V. Winters is professor of economics at Iowa State University in Ames, Iowa. He is also a research affiliate with the Center for Agricultural and Rural Development (CARD), the Program for the Study of Midwest Markets and Entrepreneurship (PSMME), Global Labor Organization (GLO). and Institute of Labor Economics (IZA).

Jun Yeong Lee is an assistant professor at Pusan National University, with a Ph.D. in economics from Iowa State University. His research interests include industrial organization, labor economics, experimental economics, and behavioral economics. He is currently researching collusive behavior, principal-agent problem, and COVID-19.

References

Abramitzky

Boustan

Jácome

Pérez

(2021). Intergenerational mobility of immigrants in the United States over two centuries. American Economic Review, 111(2), 580–608. https://doi.org/10.1257/aer.20191586

Acs

Åstebro

Audretsch

Robinson

D. T.

(2016). Public policy to promote entrepreneurship: A call to arms. Small Business Economics, 47(1), 35–51. https://doi.org/10.1007/s11187-016-9712-2

Amuedo-Dorantes

Lofstrom

Wang

(2022). Immigration policy and the rise of self-employment among Mexican immigrants. ILR Review, 75(5), 1189–1214. https://doi.org/10.1177/00197939211057765

Andersen

T. B.

Dalgaard

C. J.

Skovsgaard

C. V.

Selaya

(2021). Historical migration and contemporary health. Oxford Economic Papers, 73(3), 955–981. https://doi.org/10.1093/oep/gpaa047

Armington

Acs

Z. J.

(2002). The determinants of regional variation in new firm formation. Regional Studies, 36(1), 33–45. https://doi.org/10.1080/00343400120099843

Åstebro

Chen

Thompson

(2011). Stars and misfits: Self-employment and labor market frictions. Management Science, 57(11), 1999–2017. https://doi.org/10.1287/mnsc.1110.1400

Åstebro

Herz

Nanda

Weber

R. A.

(2014). Seeking the roots of entrepreneurship: Insights from behavioral economics. Journal of Economic Perspectives, 28(3), 49–70. https://doi.org/10.1257/jep.28.3.49

Azoulay

Jones

B. F.

Kim

J. D.

Miranda

(2022). Immigration and entrepreneurship in the United States. American Economic Review: Insights, 4(1), 71–88. https://doi.org/10.1257/aeri.20200588

Berry

C. R.

Glaeser

E. L.

(2005). The divergence of human capital levels across cities. Papers in Regional Science, 84(3), 407–444. https://doi.org/10.1111/j.1435-5957.2005.00047.x

10.

Blumberg

B. F.

Pfann

G. A.

(2016). Roads leading to self-employment: Comparing transgenerational entrepreneurs and self-made startups. Entrepreneurship Theory and Practice, 40(2), 335–357. https://doi.org/10.1111/etap.12227

11.

Borjas

G. J.

(1986). The self-employment experience of immigrants. The Journal of Human ResourcesJournal of Human Resources, 21(4), 485–506. https://doi.org/10.2307/145764

12.

Borjas

G. J.

(2002). Homeownership in the immigrant population. Journal of Urban Economics, 52(3), 448–476. https://doi.org/10.1016/S0094-1190(02)00529-6

13.

Bosma

Schutjens

(2011). Understanding regional variation in entrepreneurial activity and entrepreneurial attitude in Europe. The Annals of Regional ScienceAnnals of Regional Science, 47(3), 711–742. https://doi.org/10.1007/s00168-010-0375-7

14.

Bosquet

Overman

H. G.

(2019). Why does birthplace matter so much? Journal of Urban Economics, 110(1), 26–34. https://doi.org/10.1016/j.jue.2019.01.003

15.

Cadena

B. C.

Kovak

B. K.

(2016). Immigrants equilibrate local labor markets: Evidence from the great recession. American Economic Journal: Applied Economics, 8(1), 257–290. https://doi.org/10.1257/app.20140095

16.

Cai

Winters

J. V.

(2017). Self-employment differentials among foreign-born STEM and non-STEM workers. Journal of Business Venturing, 32(4), 371–384. https://doi.org/10.1016/j.jbusvent.2017.03.003

17.

Chen

Orazem

P. F.

(2021). Churning in urban and rural markets: Evidence from firm entry and exit, 1999–2015 [Iowa State University Working Paper]. https://www.econ.iastate.edu/files/events/files/churning_final_8_2020final–1.docx.

18.

Chen

Rosenthal

S. S.

(2008). Local amenities and life–cycle migration: Do people move for jobs or fun? Journal of Urban Economics, 64(3), 519–537. https://doi.org/10.1016/j.jue.2008.05.005

19.

Deryugina

Kawano

Levitt

(2018). The economic impact of hurricane Katrina on its victims: Evidence from individual tax returns. American Economic Journal: Applied Economics, 10(2), 202–233. https://doi.org/10.1257/app.20160307

20.

Drori

Honig

Wright

(2009). Transnational entrepreneurship: An emergent field of study. Entrepreneurship Theory and Practice, 33(5), 1001–1022. https://doi.org/10.1111/j.1540-6520.2009.00332.x

21.

Ellis

(2012). Reinventing US internal migration studies in the age of international migration. Population, Space and PlaceSpace and Place, 18(2), 196–208. https://doi.org/10.1002/psp.666

22.

Ellis

Wright

Townley

(2014). The great recession and the allure of new immigrant destinations in the United States. International Migration Review, 48(1), 3–33. https://doi.org/10.1111/imre.12058

23.

Elo

Sandberg

Servais

Basco

Cruz

A. D.

Riddle

Täube

(2018). Advancing the views on migrant and diaspora entrepreneurs in international entrepreneurship. Journal of International Entrepreneurship, 16(2), 119–133. https://doi.org/10.1007/s10843-018-0231-x

24.

Fairchild

G. B.

(2010). Intergenerational ethnic enclave influences on the likelihood of being self-employed. Journal of Business Venturing, 25(3), 290–304. https://doi.org/10.1016/j.jbusvent.2008.10.003

25.

Fairlie

R. W.

Lofstrom

(2015). Immigration and entrepreneurship. In Handbook of the economics of international migration (Vol. 1, pp. 877–911). Elsevier.

26.

Feldman

M. P.

(2001). The entrepreneurial event revisited: Firm formation in a regional context. Industrial and Corporate Change, 10(4), 861–891. https://doi.org/10.1093/icc/10.4.861

27.

Fritsch

Obschonka

Wyrwich

(2019). Historical roots of entrepreneurship-facilitating culture and innovation activity: An analysis for German regions. Regional Studies, 53(9), 1296–1307. https://doi.org/10.1080/00343404.2019.1580357

28.

Gabriel

S. A.

Rosenthal

S. S.

(2004). Quality of the business environment versus quality of life: Do firms and households like the same cities? Review of Economics and Statistics, 86(1), 438–444. https://doi.org/10.1162/003465304774201879

29.

Glaeser

E. L.

Tobio

(2008). The rise of the sunbelt. Southern Economic Journal, 74(3), 609–643664. https://doi.org/10.1002/j.2325-8012.2008.tb00856.x

30.

Goetz

S. J.

Rupasingha

(2009). Determinants of growth in non-farm proprietor densities in the US, 1990–2000. Small Business Economics, 32(4), 425–438. https://doi.org/10.1007/s11187-007-9079-5

31.

Graves

P. E.

(1980). Migration and climate. Journal of Regional Science, 20(2), 227–237. https://doi.org/10.1111/j.1467-9787.1980.tb00641.x

32.

Graves

P. E.

(2013). Spatial equilibrium in labor markets. In: Fischer

M. M.

Nijkamp

(Eds) Handbook of regional science. Springer, pp. 539–555. https://doi.org/10.1007/978-3-662-60723-7_10

33.

Greenwood

M. J.

Hunt

G. L.

Rickman

D. S.

Treyz

G. I.

(1991). Migration, regional equilibrium, and the estimation of compensating differentials. American Economic Review, 81(5), 1382–1390. https://www.jstor.org/stable/2006927

34.

Hamilton

B. H.

(2000). Does entrepreneurship pay? An Empirical Analysis of the Returns to Self-Employment. Journal of Political Economy, 108(3), 604–631. https://doi.org/10.1086/262131

35.

Han

Winters

J. V.

(2020). Industry fluctuations and college major choices: Evidence from an energy boom and bust. Economics of Education Review, 77(1), Article 101996. https://doi.org/10.1016/j.econedurev.2020.101996

36.

Hoffmann

Junge

Malchow-Møller

(2015). Running in the family: Parental role models in entrepreneurship. Small Business Economics, 44(1), 79–104. https://doi.org/10.1007/s11187-014-9586-0

37.

Kennan

Walker

J. R.

(2011). The effect of expected income on individual migration decisions. Econometrica, 79(1), 211–251. https://doi.org/10.3982/ECTA4657

38.

Koşar

Ransom

Van der Klaauw

(2022). Understanding migration aversion using elicited counterfactual choice probabilities. Journal of Econometrics, 231(1), 123–147. https://doi.org/10.1016/j.jeconom.2020.07.056

39.

Krupka

D. J.

(2009). Location-specific human capital, location choice and amenity demand. Journal of Regional Science, 49(5), 833–854. https://doi.org/10.1111/j.1467-9787.2009.00614.x

40.

Krupka

D. J.

Donaldson

K. N.

(2013). Wages, rents, and heterogeneous moving costs. Economic Inquiry, 51(1), 844–864. https://doi.org/10.1111/j.1465-7295.2012.00475.x

41.

Goetz

S. J.

(2019). The intergenerational persistence of self-employment across China’s planned economy era. Journal of Labor Economics, 37(4), 1301–1330. https://doi.org/10.1086/703164

42.

Isidor

Dau

L. A.

Kabst

(2018). The more the merrier? Immigrant share and entrepreneurial activities the more the merrier? Entrepreneurship Theory and Practice, 42(5), 698–733. https://doi.org/10.1177/1042258718795344

43.

Lindquist

M. J.

Sol

Van Praag

(2015). Why do entrepreneurial parents have entrepreneurial children? Journal of Labor Economics, 33(2), 269–296. https://doi.org/10.1086/678493

44.

Liu

C. Y.

(2012). Intrametropolitan opportunity structure and the self-employment of Asian and Latino immigrants. Economic Development Quarterly, 26(2), 178–192. https://doi.org/10.1177/0891242412443126

45.

Lofstrom

M. W.

Wang

(2019). Immigrants and entrepreneurship. IZA World of Labor, 85(1), 1–11. https://doi.org/10.15185/izawol.85.v2. https://wol.iza.org/articles/immigrants-and-entrepreneurship/long

46.

Malecki

E. J.

(1993). Entrepreneurship in regional and local development. International Regional Science Review, 16(1–2), 119–153. https://doi.org/10.1177/016001769401600107

47.

Marshall

(1890). Principles of Economics. Macmillan.

48.

McGranahan

D. A.

(1999). Natural amenities drive rural population change [Agricultural Economic Report No. (AER–781)]. United States Department of Agriculture, Economic Research Service.

49.

McGranahan

D. A.

Wojan

T. R.

Lambert

D. M.

(2011). The rural growth trifecta: Outdoor amenities, creative class and entrepreneurial context. Journal of Economic Geography, 11(3), 529–557. https://doi.org/10.1093/jeg/lbq007

50.

Mueser

P. R.

Graves

P. E.

(1995). Examining the role of economic opportunity and amenities in explaining population redistribution. Journal of Urban Economics, 37(2), 176–200. https://doi.org/10.1006/juec.1995.1010

51.

Obschonka

Stuetzer

(2017). Integrating psychological approaches to entrepreneurship: The entrepreneurial personality system (EPS). Small Business Economics, 49(1), 203–231. https://doi.org/10.1007/s11187-016-9821-y

52.

Partridge

M. D.

(2010). The dueling models: NEG vs amenity migration in explaining US engines of growth. Papers in Regional Science, 89(3), 513–537536. https://doi.org/10.1111/j.1435-5957.2010.00315.x

53.

Partridge

M. D.

Rickman

D. S.

(2003). The waxing and waning of regional economies: The chicken–egg question of jobs versus people. Journal of Urban Economics, 53(1), 76–97. https://doi.org/10.1016/S0094-1190(02)00501-6

54.

Partridge

M. D.

Rickman

D. S.

Ali

Olfert

M. R.

(2008). Lost in space: Population growth in the American hinterlands and small cities. Journal of Economic Geography, 8(6), 727–757. https://doi.org/10.1093/jeg/lbn038

55.

Partridge

M. D.

Rickman

D. S.

Ali

Olfert

M. R.

(2009). Agglomeration spillovers and wage and housing cost gradients across the urban hierarchy. Journal of International Economics, 78(1), 126–140. https://doi.org/10.1016/j.jinteco.2009.02.004

56.

Pugsley

B. W.

Hurst

(2011). What do small businesses do? Brookings Papers on Economic Activity, 43(2), 73–142. https://www.jstor.org/stable/41473598

57.

Ransom

(2022). Labor market frictions and moving costs of the employed and unemployed. Journal of Human Resources, 57(S), S137–S166. https://doi.org/10.3368/jhr.monopsony.0219-10013R2

58.

Rappaport

(2009). The increasing importance of quality of life. Journal of Economic Geography, 9(6), 779–804. https://doi.org/10.1093/jeg/lbp009

59.

Rijnks

R. H.

Koster

McCann

(2018). Spatial heterogeneity in amenity and labor market migration. International Regional Science Review, 41(2), 183–209. https://doi.org/10.1177/0160017616672516

60.

Roback

(1982). Wages, rents, and the quality of life. Journal of Political Economy, 90(6), 1257–1278. https://doi.org/10.1086/261120

61.

Ruggles

Flood

Foster

Goeken

R. E.

Pacas

Schouweiler

Sobek

(2021). IPUMS USA: Version 11.0 [Data set]. IPUMS. https://doi.org/10.18128/D010.V11.0

62.

Rupasingha

Marré

A. W.

(2020). Moving to the hinterlands: Agglomeration, search costs and urban to rural business migration. Journal of Economic Geography, 20(1), 123–153. https://doi.org/10.1093/jeg/lby057

63.

Sequeira

J. M.

Carr

J. C.

Rasheed

A. A.

(2009). Transnational entrepreneurship: Determinants of firm type and owner attributions of success. Entrepreneurship Theory and Practice, 33(5), 1023–1044. https://doi.org/10.1111/j.1540-6520.2009.00333.x

64.

Sjaastad

L. A.

(1962). The costs and returns of human migration. Journal of Political Economy, 70(5, Part 2), 80–93. https://doi.org/10.1086/258726

65.

Sorenson

(2018). Social networks and the geography of entrepreneurship. Small Business Economics, 51(3), 527–537. https://doi.org/10.1007/s11187-018-0076-7

66.

Sternberg

(2022). Entrepreneurship and geography—some thoughts about a complex relationship. Annals of Regional Science, 69(3), 559–584. https://doi.org/10.1007/s00168-021-01091-w

67.

Winters

J. V.

(2011). Why are smart cities growing? Who moves and who stays. Journal of Regional Science, 51(2), 253–270. https://doi.org/10.1111/j.1467-9787.2010.00693.x

68.

Winters

J. V.

(2017). Do earnings by college major affect graduate migration? The Annals of Regional ScienceAnnals of Regional Science, 59(3), 629–649. https://doi.org/10.1007/s00168-016-0748-7

69.

Winters

J. V.

(2020). In-state college enrollment and later life location decisions. Journal of Human Resources, 55(4), 1400–1426. https://doi.org/10.3368/jhr.55.4.0916-8255R2

70.

Winters

J. V.

(2022). No place like home: Place-based attachments and regional science. Review of Regional Studies, 52(2), 159–175. https://doi.org/10.52324/001c.37967

71.

Yang

Zhang

(2023). Persistence of culture: How the entrepreneurial culture of origin contributes to migrant entrepreneurship. Small Business Economics, 61(3), 1179–1204. https://doi.org/10.1007/s11187-022-00699-2

72.

Artz

G. M.

(2019). Does rural entrepreneurship pay? Small Business Economics, 53(3), 647–668. https://doi.org/10.1007/s11187-018-0073-x

73.

Orazem

P. F.

Jolly

R. W.

(2011). Why do rural firms live longer? American Journal of Agricultural Economics, 93(3), 673–692. https://doi.org/10.1093/ajae/aaq173

74.

Yuengert

A. M.

(1995). Testing hypotheses of immigrant self-employment. The Journal of Human ResourcesJournal of Human Resources, 30(1), 194–204. https://doi.org/10.2307/146196

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB