The effects of the COVID-19 pandemic on U.S. entrepreneurship

Abstract

We study the effects of the COVID-19 pandemic on U.S. entrepreneurial activities, as measured by the overall number of new business applications, high-propensity business applications, business applications from corporations, and business applications with paid wages. However, the number of business applications increased significantly after the lockdown. Also, the portion of high-propensity business applications as a share of total business applications declined considerably during and after the lockdown. Our findings could partially explain the tight labor market in the U.S. during the pandemic.

Introduction

COVID-19 severely affected people’s lives, health, employment, and the economy. The pandemic disrupted economic activities, increased the unemployment rate, and reduced economic growth. Policymakers and individuals faced a trade-off between maintaining health and maintaining an income, while balancing employment and possibly limited mobility. Many policies, including lockdowns, were implemented to curb the spread of the virus. Nevertheless, these policies were contentious and received many critiques about their impacts on the U.S. labor market and entrepreneurship.

In this paper, we study entrepreneurship in the U.S. during the COVID-19 pandemic. Using state-level data, we measure entrepreneurship based on the number of new business applications, including total business applications, applications with a high-propensity of turning into a business with a payroll, applications from a corporation or a personal service corporation, and applications with a planned date for paying wages. We find that the local COVID-19 lockdowns significantly reduced all types of new business applications, specifically in the Northeast region. However, lifting the lockdown policies significantly raised all types of new business applications, particularly in 2021.

While the number of new business applications increased significantly in total, we also find that the proportion of high-propensity business applications as a share of total business applications declined significantly during and after economic lockdowns. These high-propensity business applications contain applications from corporations and applications with planned wages.1 The decline in the share of these applications suggests that many of these new businesses, referred to as low-propensity business applications, might not turn into businesses with a payroll in the long run. The increase in business applications suggests that people become increasingly entrepreneurial during the pandemic. On the one hand, people may try to run their businesses due to positive beliefs of entrepreneurial opportunity (Seah 2021). On the other hand, necessity-based entrepreneurship, whereby people become self-employed to avoid unemployment, could also increase the number of new business applications (O’Leary 2022). Our findings might partially explain tight labor markets in the U.S. after the lockdowns, where there have been shortages of available workers for plentiful vacant jobs (Andolfatto & Birinci 2022). All other things being equal, increasing new businesses will lead to higher demand for workers. Simultaneously, increasing self-employment can lower the labor supply in the market. Therefore, a sudden increase in entrepreneurship after the lockdowns could contribute to tight labor markets. Nevertheless, our findings suggest that many new entrepreneurial activities are unlikely to become businesses with payrolls.

Our paper is related to the emerging literature focusing on entrepreneurship in the COVID-19 era. Belitski et al. (2022) conduct a literature review on the effects of exogenous shocks, specifically COVID-19, on entrepreneurship and small businesses. Zhang et al. (2022) find that working from home improved small businesses’ performance during the pandemic. Maurer (2022) show that during the pandemic, entrepreneurs accessed support from online communities, such as sharing digital marketing tools, waterproofing ideas in early venture stages, reflecting on new emerging issues, and providing tailor-made plans for entrepreneurs.

Our study is also related to research analyzing the impacts of the COVID-19 pandemic on economic outcomes. Employing economic modeling, Capello and Caragliu (2021) predict declining European regional disparity trends in a post-COVID era. Using U.S. county-level data, Yilmazkuday (2021) finds about 11% reduction in the average welfare during the pandemic in 2020. Brooks et al. (2021) find that rural labor markets outperformed urban markets during the early period of the pandemic. Hean and Chairassamee (2022a) find differential responses by race in labor nonparticipation during the pandemic.

Finally, other studies examine the effects of COVID-19-related policies. Rodríguez-Pose and Burlina (2021) find that strong institutions significantly limit the number of COVID-19 deaths. Hean and Chairassamee (2022b)  show that an economic lockdown could significantly affect consumption, especially in developing regions. Martin et al. (2020) project that in the absence of social protection programs and a three-month shelter-in-place period, the poverty rate would increase by around 9%, from 17.1%, and lower-income people would suffer the most. Glaeser et al. (2021) find that relaxing stay-at-home orders could lead to a sharp increase in mobility and the infection rate.

The remainder of the paper is organized as follows. Section 2 discusses our data and highlights several empirical facts. Sections 3 and 4 show a regression analysis and results, respectively. Section 5 provides additional analysis, while Sect. 6 concludes.

Data

The data is monthly, from 2017 to 2022, including the 50 states and the District of Columbia. There are 3,276 observations in total. We use monthly business formations provided by the U.S. Census Bureau, including the total number of business applications, high-propensity business applications, business applications from corporations, and business applications with planned wages.

The Census Bureau defines business applications (BA) as those core business application series that correspond to a subset of all Employer Identification Number (EIN) applications. High-propensity business applications (HBA) are a subset of BA that contains all applications with a high-propensity of turning into a business with a payroll, based on various factors. High-propensity business applications comprise applications from corporations and applications with planned wages. Business applications from corporations (CBA) are applications that come from a corporation or a personal service corporation, while applications with planned wages (WBA) are those that indicate a planned date for paying wages.

The quarterly state gross domestic product (GDP) is sourced from the Bureau of Economic Analysis (BEA), and monthly unemployment rates are from the Bureau of Labor Statistics (BLS). We also obtain the number of total COVID-19 cases from the Centers for Disease Control and Prevention. Descriptive statistics for the data are shown in Table 1.

Table 1.

Descriptive statistics

Variable	Mean	Std. Dev	Min	Max
Total business applications	6,682.41	8,485.03	335.00	62,946.00
High-propensity business applications	2,346.43	3,306.93	107.00	21,842.00
Business applications from corporations	938.72	1,889.85	18.00	12,627.00
Business applications with planned wages	875.19	1,072.83	40.00	6,813.00
State GDP (millions of current dollars)	416,923.8	526,542.70	30,758.40	3,568,888.00
Unemployment rates	4.75	2.45	1.30	27.50
Total COVID-19 case rates	129.29	238.25	0.00	1,335.94

All business application variables, unemployment rates, and total COVID-19 case rates are monthly data. The state GDP is quarterly data

Most states implemented a local lockdown for a month. During the initial lockdown period from March to April 2020, the total number of business applications in all regions changed slightly (Fig. 1a). After the lockdown, however, the number of business applications increased significantly. Figure 1b shows that the Northeast region experienced the highest reduction in total business applications by almost 21% during the initial lockdown period, but also had the highest increase in total business applications by 33% immediately after the lockdown relaxation in May 2020. In the Northeast, New York State drove the changes in the number of applications (details of the applications by state are given in Appendix B).

Fig. 1.

Total business applications (BA) by region a number of BA b month-to-month changes in BA

The positive changes in the total number of business applications increased continuously until July 2020. The total number of business applications in the South and Midwest regions increased by more than 50%, compared to those in June 2020. An increase in the number of business applications in the South after the lockdown was driven by Florida and Texas, while Illinois saw the highest change in the number of applications in the Midwest region.

Similar to the number of total business applications, the number of high-propensity business applications, business applications from corporations, and business applications with planned wages slightly decreased during the lockdown (Fig. 2a), while the Northeast region experienced the greatest reduction in all application types (Figs. 2b–4b).

Fig. 2.

High-propensity business applications (HBA) by region a number of HBA b month-to-month changes in HBA

Fig. 4.

Business applications with planned wages (WBA) by region a number of WBA b month-to-month changes in WBA

In May 2020, the beginning of the lockdown relaxation in most states, the number of high-propensity business applications in every region increased. The increases reached their highest point three to four months after the lockdown ended. The increase in high-propensity business applications is a result of increases in both business applications from corporations (Fig. 3a) and business applications with planned wages (Fig. 4a).

Fig. 3.

Business applications from corporations (CBA) by region a number of CBA b month-to-month changes in CBA

Methodology

To study the effects of the COVID-19 pandemic on entrepreneurial activities, we estimate the following regression specification:

$${\mathrm{logAPP}}_{\mathrm{smt}}=\alpha +{\sum }_{\tau =1}^3{\beta }_{\tau }\times I\left(\tau \right)+\delta X_{\mathit{st}}+{\eta }_m+{\nu }_s+{ϵ}_{\mathit{smt}},$$ 1

where ${\mathrm{logAPP}}_{\mathrm{smt}}$ is the natural log of the number of business applications in state $s$, in month $m$, in year $t$. ${\mathrm{APP}}_{\mathrm{smt}}$ is either the number of total new business applications, high-propensity business applications, business applications from corporations, and applications with planned wages. $\tau$ is a set of dummy variables for one of the two phases of the pandemic: the strict lockdown period and the reopening of each state. The pre-pandemic period is, therefore, a reference period. It is important to note that not every state implemented a lockdown. For states that did not implement lockdowns, we set the lockdown dummy to zero throughout the entire period, while the after-lockdown dummy for these states is equal to 1. Details for the lockdown period in each state are in Appendix A. The regression controls $X_{\mathit{st}}$, which includes the natural log of the quarterly state GDP, monthly unemployment rates, and total COVID case rates.2 We also include the month-fixed effects $\left({\eta }_m\right),$ to capture seasonality, and the state-fixed effects $\left({\nu }_s\right),$ to capture local socioeconomic conditions.3

We estimate specification (1) using the Ordinary Least Squares (OLS) regression. We also conduct a sensitivity analysis using negative binomial regressions. The results, shown in Table 5 of Appendix C, are similar to the baseline (OLS) results in Table 2.

Table 5.

Negative binomial regressions

	Overall (1)	Northeast (2)	Midwest (3)	South (4)	West (5)
Panel A: BA
Lockdown	− 0.071**	− 0.262***	− 0.195*	0.073	− 0.155**
	(0.034)	(0.081)	(0.118)	(0.055)	(0.061)
After lockdown, in 2020	0.306***	0.219***	0.257***	0.476***	0.151***
	(0.014)	(0.036)	(0.037)	(0.021)	(0.020)
After lockdown, in 2021	0.387***	0.325***	0.600***	0.547***	0.230***
	(0.018)	(0.033)	(0.051)	(0.029)	(0.027)
After lockdown, in 2022	0.368***	0.316***	0.881***	0.523***	0.221***
	(0.031)	(0.053)	(0.080)	(0.048)	(0.052)
Pseudo R2	0.23	0.26	0.22	0.23	0.26
Panel B: HBA
Lockdown	− 0.154***	− 0.274***	− 0.290***	− 0.069	− 0.176***
	(0.031)	(0.080)	(0.101)	(0.047)	(0.059)
After lockdown, in 2020	0.245***	0.219***	0.204***	0.349***	0.135***
	(0.012)	(0.035)	(0.032)	(0.019)	(0.020)
After lockdown, in 2021	0.338***	0.327***	0.444***	0.453***	0.207***
	(0.016)	(0.038)	(0.046)	(0.026)	(0.024)
After lockdown, in 2022	0.290***	0.306***	0.574***	0.412***	0.135***
	(0.027)	(0.058)	(0.074)	(0.046)	(0.046)
Pseudo R2	0.28	0.32	0.27	0.27	0.30
Panel C: CBA
Lockdown	− 0.285***	− 0.404***	− 0.297***	− 0.289***	− 0.204***
	(0.032)	(0.086)	(0.080)	(0.047)	(0.059)
After lockdown, in 2020	0.068***	0.117***	0.073**	0.072***	0.035
	(0.013)	(0.042)	(0.029)	(0.019)	(0.030)
After lockdown, in 2021	0.117***	0.150***	0.171***	0.115***	0.063*
	(0.018)	(0.037)	(0.045)	(0.027)	(0.036)
After lockdown, in 2022	0.061*	0.101*	0.162**	0.055	− 0.042
	(0.034)	(0.054)	(0.083)	(0.051)	(0.077)
Pseudo R2	0.32	0.37	0.32	0.32	0.31
Panel D: WBA
Lockdown	− 0.214***	− 0.381***	− 0.345***	− 0.135***	− 0.217***
	(0.033)	(0.080)	(0.105)	(0.045)	(0.066)
After lockdown, in 2020	0.197***	0.153***	0.156***	0.302***	0.094***
	(0.013)	(0.036)	(0.036)	(0.019)	(0.022)
After lockdown, in 2021	0.278***	0.244***	0.415***	0.420***	0.124***
	(0.016)	(0.036)	(0.050)	(0.026)	(0.025)
After lockdown, in 2022	0.200***	0.183***	0.534***	0.374***	0.002
	(0.028)	(0.053)	(0.083)	(0.043)	(0.046)
Pseudo R2	0.30	0.33	0.27	0.29	0.33
Controls	YES	YES	YES	YES	YES
Month-fixed effects	YES	YES	YES	YES	YES
State-fixed effects	YES	YES	YES	YES	YES
Obs	3,276	567	756	1,134	819

For states without lockdown measures, the lockdown dummy is equal to zero, and, for the same period, the after-lockdown dummy is equal to 1. All models control for the log of the quarterly state GDP, monthly unemployment rates, and monthly COVID case rates. State- and month-fixed effects are included. Robust standard errors are in parentheses. *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively

Table 2.

The effects of the COVID-19 pandemic on the number of business applications

	Overall (1)	Northeast (2)	Midwest (3)	South (4)	West (5)
Panel A: BA
Lockdown	− 0.068** (0.33)	− 0.248*** (0.086)	− 0.122 (0.118)	− 0.076 (0.055)	− 0.164*** (0.058)
After lockdown, in 2020	0.292*** (0.013)	0.220*** (0.037)	0.256*** (0.033)	0.461*** (0.021)	0.148*** (0.020)
After lockdown, in 2021	0.382*** (0.017)	0.327*** (0.034)	0.595*** (0.049)	0.532*** (0.029)	0.238*** (0.026)
After lockdown, in 2022	0.366*** (0.030)	0.314*** (0.054)	0.861*** (0.080)	0.500*** (0.048)	0.241*** (0.050)
R2	0.9872	0.9929	0.9835	0.9881	0.9902
Panel B: HBA
Lockdown	− 0.153*** (0.030)	− 0.245** (0.082)	− 0.224** (0.100)	− 0.063 (0.048)	− 0.189*** (0.053)
After lockdown, in 2020	0.229*** (0.012)	0.216*** (0.034)	0.193*** (0.028)	0.333*** (0.020)	0.131*** (0.020)
After lockdown, in 2021	0.334*** (0.016)	0.338*** (0.039)	0.432*** (0.046)	0.434*** (0.027)	0.214*** (0.023)
After lockdown, in 2022	0.291*** (0.027)	0.325*** (0.060)	0.546*** (0.076)	0.379*** (0.046)	0.154*** (0.044)
R2	0.9913	0.9953	0.9898	0.9899	0.9921
Panel C: CBA
Lockdown	− 0.285*** (0.032)	− 0.354*** (0.091)	− 0.238** (0.080)	− 0.288*** (0.050)	− 0.228*** (0.062)
After lockdown, in 2020	0.048*** (0.014)	0.097** (0.044)	0.052* (0.029)	0.052** (0.021)	0.017 (0.029)
After lockdown, in 2021	0.119*** (0.019)	0.149*** (0.044)	0.175*** (0.049)	0.088*** (0.029)	0.067* (0.041)
After lockdown, in 2022	0.072** (0.036)	0.089 (0.071)	0.164* (0.095)	0.004 (0.055)	− 0.017 (0.076)
R2	0.9894	0.9947	0.9875	0.9897	0.9847
Panel D: WBA
Lockdown	− 0.216*** (0.031)	− 0.338*** (0.084)	− 0.259** (0.104)	− 0.129*** (0.046)	− 0.236*** (0.060)
After lockdown, in 2020	0.173*** (0.013)	0.149*** (0.037)	0.130*** (0.032)	0.284*** (0.020)	0.090*** (0.022)
After lockdown, in 2021	0.266*** (0.016)	0.240*** (0.038)	0.386*** (0.051)	0.398*** (0.027)	0.141*** (0.024)
After lockdown, in 2022	0.194*** (0.057)	0.180*** (0.061)	0.476*** (0.088)	0.340*** (0.046)	0.049*** (0.044)
R2	0.9882	0.9923	0.9820	0.9884	0.9912
Controls	YES	YES	YES	YES	YES
Month-fixed effects	YES	YES	YES	YES	YES
State-fixed effects	YES	YES	YES	YES	YES
Obs	3,276	567	756	1,134	819

For states without lockdown measures, the lockdown dummy is equal to zero, and, for the same period, the after-lockdown dummy is equal to 1. All models control for the log of the quarterly state GDP, monthly unemployment rates, and monthly COVID case rates. State- and month-fixed effects are included. Robust standard errors are in parentheses. *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively

Empirical results

The results in Panel A of Table 2 show the effects of the COVID-19 pandemic on the total number of new business applications. On average, compared to the pre-pandemic period, the lockdown reduced business applications by 6.57% (i.e., ${(e}^{-0.068}-1)\times 100$). Importantly, the Northeast and West regions were significantly affected by the lockdown, with almost 22% and 15% reductions in business applications, respectively.

The number of total applications increased after the lockdown period. Relative to the pre-pandemic period, overall new applications increased by almost 34% in 2020, and hit 47% in 2021. Later, these business applications slowed down and reached about 44% in 2022, still above the pre-pandemic level. Significantly, the total number of business applications in the Midwest jumped to 81% and more than 100% in 2021 and 2022, respectively, compared to the pre-pandemic.

In Panel B of Table 2, the effects of the pandemic on high-propensity applications are similar to those on the total business applications. The lockdown decreased the number of high-propensity applications by 14%. Almost all regions were affected by the lockdown; yet, the Midwest region experienced the highest decrease, with 22% of the number of the applications, compared to the pre-pandemic period. The number of high-propensity business applications increased immediately after the lockdown. The Midwest is the only region having a continuous increase in these applications until 2022; other regions reached their highest point in 2021.

Panels C and D in Table 2 present the estimated effects of the pandemic on business applications from corporations and with planned wages, which are the two components of HBA. A decrease in high-propensity business applications during the lockdown is a result of a reduction of business applications from corporations (Panel C) and applications with planned wages (Panel D).

The lockdown led to declines of about 25% and 20% in the overall applications from corporations and applications with planned wages, respectively. The Northeast region was affected the most. After the lockdown, the number of business applications from corporations was higher in 2021; however, in 2022, the number of this type of application in almost every region was not significantly different from that in the pre-pandemic period. In contrast, applications with planned wages in every region markedly increased during the reopening period.

Additional analysis

Figure 5 plots high-propensity business applications as a percentage of total business applications. Even though the total number of all business application types increased after the lockdown, the percentage of high-propensity business applications during the pandemic was lower than before the pandemic. During the lockdown, the percentage of high-propensity business applications dropped in all regions, especially in the Northeast, South, and Midwest. The evidence implies that during and after the lockdown, the percentage of low-propensity business applications increased.

Fig. 5.

Percentage of high-propensity business applications by region

To consider the effects of the pandemic on the proportion of high-propensity business applications, we estimate the following regression specification:

$${\mathrm{HBA}}_{\mathrm{smt}}=\alpha +{\sum }_{\tau =1}^3{\beta }_{\tau }\times I\left(\tau \right)+\delta X_{\mathit{st}}+{\eta }_m+{\nu }_s+{ϵ}_{\mathrm{smt}},$$ 2

where ${\mathrm{HBA}}_{\mathrm{smt}}$ is the percentage of high-propensity business applications in state $s$, during month $m$, in year $t$; the denominator is the total number of business applications in the same state and time period. Other variables are the same as in specification (1).

The results in Table 3 show that the COVID-19 pandemic reduced the proportion of high-propensity business applications. Particularly, the Midwest and South regions experienced a significant decrease in the percentage of high-propensity business applications during the lockdown period. Moreover, in almost all regions, the share of the applications decreased even after the lockdown. This decline in the proportion of high-propensity business applications suggests that many of these new businesses are unlikely to become businesses with payrolls.

Table 3.

The effects of the COVID-19 pandemic on the percentage age of high-propensity business applications

	Overall (1)	Northeast (2)	Midwest (3)	South (4)	West (5)
Lockdown	− 2.579*** (0.420)	0.188 (1.288)	− 3.045** (1.259)	− 3.931*** (0.055)	− 1.066 (0.787)
After lockdown, in 2020	− 1.987*** (0.217)	− 0.057 (0.715)	− 2.091*** (0.485)	− 3.810*** (0.264)	− 0.708* (0.379)
After lockdown, in 2021	− 1.584*** (0.363)	0.198 (1.024)	− 5.337*** (0.934)	− 2.803*** (0.402)	− 0.945* (0.495)
After lockdown, in 2022	− 2.546*** (0.700)	− 0.347 (2.015)	− 10.213*** (1.855)	− 3.349*** (0.755)	− 3.039*** (0.911)
Controls	YES	YES	YES	YES	YES
Month-fixed effects	YES	YES	YES	YES	YES
State-fixed effects	YES	YES	YES	YES	YES
Obs	3,276	567	756	1,134	819
R2	0.7493	0.9929	0.9835	0.9881	0.9902

For states without lockdown measures, the lockdown dummy is equal to zero, and, for the same period, the after-lockdown dummy is equal to 1. All models control for the log of the quarterly state GDP, monthly unemployment rates, and monthly COVID case rates. State- and month-fixed effects are included. Robust standard errors are in parentheses. *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively

Discussion and conclusion

Our study examines entrepreneurship in the U.S. during the COVID-19 pandemic. We find significant decreases during the lockdown in the number of new business applications, including the number of total business applications, high-propensity business applications, business applications from corporations, and applications with planned wages. This decline was particularly pronounced in the Northeast region. However, the number of new business applications increased significantly after the lockdown.

Even though the overall number of business applications increased after the lockdown, the percentage of high-propensity applications declined both during and after the lockdown. The decreased percentage of high-propensity business applications indicates that many of these new businesses might not turn into businesses with payrolls.

The U.S. labor market is associated with new and young businesses (Benedetti-Fasil Sedláček & Sterk 2022). Our findings could partially explain tight labor markets in the U.S. after the lockdowns, which have been characterized by job vacancies far exceeding available workers (Andolfatto & Birinci 2022). Ceteris paribus, new businesses will increase the demand for workers. Also, increasing self-employment can lower the labor supply in the market. Therefore, a sudden increase in entrepreneurship after the lockdowns could contribute to tight labor markets. The increase in the total number of business applications and the fall in the share of high-propensity business applications could suggest that during the pandemic, people became increasingly entrepreneurial. However, these entrepreneurial activities might not have become businesses with payrolls.

Appendix A: Local Lockdown Period in 2020 and 2021

See Table4.

Table 4.

Local lockdown period in 2020

The data were collected by the authors. Black-colored boxes denote lockdown in 2020

Appendix B: The Number of Total Business Applications by Region and State

See Figs. 6, 7, 8, 9.

Fig. 6.

The number of total business applications in Northeast

Fig. 7.

The number of total business applications in Midwest

Fig. 8.

The number of total business applications in South

Fig. 9.

The number of total business applications in West

Appendix C

See Table 5.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declarations

Conflict of interest

The authors have no potential conflict of interest.