1 Introduction

Despite several decades of progress, women, on average, earn less than men and are significantly underrepresented in leadership positions (Bertrand & et al., 2011; Bertrand, 2018). This is true whether they are self-employed or work for a firm. The earnings gap between genders is not only due to differential occupational sorting, but is also present within occupations (Goldin, 2014).

A particular challenge that women face in the labor market is the relatively higher demand for their time outside of the labor market (Bertrand, 2018). The most significant components of this non-market work are childbearing, childcare, and household production activities. Although technological change has decreased the amount of time spent on household tasks, the amount of time devoted to parenting activities has increased, especially among the well educated (Cortés & Pan, 2019). Traditional societal gender roles, though weakening, are still present, and society expects women to be the main provider of non-market work (Fortin, 2005; Patrick et al., 2016).

The tension between workplace and household responsibilities may result in women seeking flexible schedules at work or exploring entrepreneurship, which gives them the option to work from home. Women are more likely to work from home than men, both as employees and business owners (Edwards & Field-Hendrey, 2002; Hughes et al., 2012; Carter et al., 2015). Research finds that home-based female owners experience less work-to-family conflict than their counterparts (Loscocco & Smith-Hunter, 2004). However, these businesses are, on average, smaller and often run on a part-time basis (Thompson et al., 2009). Overall, the existing literature provides little analysis of the factors contributing to the performance of home-based businesses or of the advantages of operating such businesses for female business owners.

Motivated by these insights, we seek to examine the performance of female-owned businesses operated both in and outside of the home, relative to their male counterparts. By doing so, we also investigate whether the increased autonomy and flexibility of owning a home-based business improves female business owners’ relative performance. We use the 2007 Survey of Business Owners and Self-Employed Persons (SBO) data to create a firm-level dataset of 663,385 single-owner firms that contains information about business operations and the owners’ demographic characteristics. We calculate the effect of female-owned and home-based firms on firm performance. In isolation, the female owner effect and the home-based firm effect on performance are negative and significant. However, being being both female-owned and home-based has an incrementally positive and significant effect on firm performance, indicating the presence of the positive synergy effect. This result consistently holds in other model specifications.

Our work is related to the literature on gender gaps in economic success. Specifically, we are motivated by the literature that examines the challenges that women face in the labor market because of the relatively higher demand on their time outside of the labor market. Women value flexibility and shorter hours more than men (Mas & Pallais, 2017; Wiswall & Zafar, 2018). Loscocco & Bird (2012) posit that women face a trade-off between work and personal lives which hinders their career prospects as men and women respond to different societal expectations. However, this increased demand for more flexibility in the workplace results in labor market penalties in the corporate setting (Bertrand et al., 2010; Goldin & Katz, 2011; Goldin, 2014; Bertrand, 2018). The need for flexibility often results in women pursuing self-employment (Still & Walker, 2006). To further alleviate the work life balance conflict, these businesses are often home-based (Walker et al., 2008). Our paper extends the findings of Kiefer et al. (2022) who study gender differences in small business outcomes. In their paper, Kiefer et al. (2022) confirm the existence of the gender gap, but point out that it may be due to the omission of difficult-to-measure factors such as preferences for growth, risk aversion, networks, lending discrimination, and consumer discrimination against female-led firms. Our paper helps identify one of these factors: preference for flexibility.

Second, our work is related to the literature on women in entrepreneurship. Female entrepreneurs face greater capital constraints than men, and relaxing these constraints leads to substantial profit gains (De Mel et al., 2009). Bernhardt et al. (2019) study micro-entrepreneurs in low-income countries and demonstrate that the gender gap in performance is not due to a gap in aptitude, but rather the result of the household-level distribution of capital. Yang and del Carmen Triana (2019) find that women-led businesses are more likely to fail due to gender beliefs that discount women’s leadership. However, there is little literature on the performance of home-based businesses based the owner’s gender. The closest study to this paper is Loscocco and Smith-Hunter (2004) which studied a relatively small sample of female home-based business owners located in upstate New York and observed that such businesses had lower sales volumes than female non-home-based business owners. None of the existing literature sought to evaluate female business owners’ performance relative to their male counterparts operating businesses of similar size, within the same sectors, with similar start-up capital, and other factors. Our paper, using a large cross-sectional sample of US firms, finds that, ceteris paribus, the gap in performance between men and women entrepreneurs narrows considerably among home-based businesses.

Third, our work is related to corporate finance literature examining the impact of gender on corporate performance. Khan & Vieito (2013) show that firms with female CEOs are associated with an increase in performance compared to firms managed by male CEOs when the CEO is female. Niessen-Ruenzi & Ruenzi (2019) examine US mutual funds run by female managers and find no differences in performance between female and male managers. Moreover, they find that female managers, despite being able to raise less funds from investors, achieve more stable performance than male managers. Finally, Amore et al. (2014) find that Italian family-controlled firms led by female CEOs are more profitable if there are female directors on their boards. Our paper identifies a key factor that allows female firm owners to narrow gaps in performance their male counterparts in the market. That condition is being able to work from home. Identifying this condition will help inform policy decisions.

The rest of this paper is organized as follows. Section 2 presents our theory development, while Section 3 outlines the empirical framework and methodology. Section 4 presents our results, and Section 5 summarizes our conclusions.

2 Theory development

Our study focuses on examining whether the location of a business, either at home, or at an outside office, impacts its performance across genders. To address this question, we must first examine what motivates entrepreneurs to become entrepreneurs. Then we need to understand why some entrepreneurs choose to operate their business from home rather than from a separate location. Finally, we also need to establish whether the factors that influence this decision are different for men and women.

Analyzing the decisions of individuals about whether to become entrepreneurs is central to occupational choice models. These models focus on the decisions of individuals with heterogeneous abilities and tastes to become employees or entrepreneurs. Individuals may derive utility from financial and non-financial factors (Taylor, 1996). Some models concentrate solely on the utility of pecuniary rewards (Evans & Jovanovic, 1989). We draw from models that focus not only on financial returns but incorporate non-financial considerations. Individuals may prefer greater flexibility in organizing their time or other non-financial benefits associated with being their own boss (Pugsley & Hurst, 2011; Yurdagul, 2017). Since women are more likely to have constraints on their time due to household responsibilities, the flexibility motive should be stronger for women (Bento et al., 2021). This insight informs our hypothesis development below.

2.1 Gender and entrepreneurship

Empirical research shows that men and women start their own businesses for different reasons (Redmond et al., 2017). Men are more likely to cite making money as the motivating factor when starting a business, whereas women are more likely to be motivated by the flexibility, which will allow for work and family balance (Edwards & Field-Hendrey, 2002; Walker et al., 2008; Craig et al., 2012; Loscocco & Bird, 2012). Hughes (2006) studies Canadian entrepreneurs and finds that the gender differences in motivations are significant. In her study, 31.9% of female entrepreneurs cite work-family balance as their primary motivator as opposed to only 7.9% of men.Footnote 1

The entrepreneurs that cite work-family balance as their primary motivation are the most likely to earn lower income and yet are the least likely to report financial difficulties or express a willingness to become an employee of another firm. These individuals may have greater access to alternative sources of household income or are otherwise more willing to forfeit income in exchange for non-pecuniary quality of life benefits. Women are more likely to be pushed rather than pulled into self-employment, or small business ownership, meaning that some women may go into entrepreneurship by necessity and not because it is the most attractive career opportunity (Pines et al., 2010). During the COVID-19 crisis, women entrepreneurs were more likely to be at risk of reduced hours and earnings than men due to increased childcare responsibilities during lockdowns and school closures (Reuschke et al., 2021). Also, despite of being hit harder in some domains, women-led businesses were less likely to receive public support (Torres et al., 2021).

Several studies to date have examined female-led business outcomes. Women business owners earn less than men and less than employees of firms regardless of gender (Hundley, 2000; Bertrand, 2018). Lim & Suh (2019) show that women are more likely to establish solo or family-only firms than a non-family businesses in comparison to their male counterparts. They also show that businesses run by women tend to display lower initial performance compared to businesses run by men. One reason for this is that women do not start their business on an equal footing with their male counterparts. Women-led businesses have less start-up capital and less experience (Carter et al., 2015). In addition, they also have less prior work experience in family businesses (Fairlie & Robb, 2009). Female-owned firms are, on average, more likely to close and have lower levels of sales, profits, and fewer employees (Robb, 2002; Robb & Wolken, 2002; Kiefer et al., 2022). Overall, self-employed women, on average, earn less than self-employed men. Some of this discrepancy has been attributed to self-employed women operating more frequently in lower-paying sectors such as service and retail, which are more competitive and also exhibit lower business survival rates (Yang & del Carmen Triana, 2019). For all these reasons, one would expect that female-led businesses, on average, will achieve lower performance than male-led enterprises.

H1: Female-led businesses achieve lower firm performance compared to male-led businesses.

2.2 Home-based entrepreneurs

Relatively few studies to date have researched home-based entrepreneurs. Kim & Parker (2021) find that self-employed home-based entrepreneurs suffer an earnings penalty relative to self-employed non-home-based entrepreneurs and employees. Home-based businesses are less likely to hire employees and are more likely to operate part-time (Mason et al., 2011; Kim & Parker, 2021). Home-based businesses tend to be smaller in turns of employment (Mason et al., 2011). For these reasons, one would expect that home-based entrepreneurs are likely to achieve lower performance than non-home-based entrepreneurs:

H2: Home-based entrepreneurs achieve lower performance than non-home-based entrepreneurs.

2.3 Gender differences in home-based entrepreneurs’ performance

Several studies analyze gender differences in home-based enterprises and find that women are drawn to small business ownership as it provides them with the opportunity to better juggle their home and work responsibilities. Kim & Parker (2021) show that home-based businesses are positively correlated to homes with dependent children, and that this correlation is stronger for female than male entrepreneurs. Loscocco & Smith-Hunter (2004) study women entrepreneurs in upstate New York and find that women who own home-based enterprises have less work-family conflict than those who run firms away from home. Peters et al. (2020) examine female freelancers and reveal that worktime control is related to career satisfaction. Justo et al. (2015) show that female-owned businesses are not more likely to financially fail than their male-owned counterparts. Instead, the study found female entrepreneurs are more likely to exit voluntarily, and generally due to personal reasons. This finding by Justo et al. (2015) challenges the female underperformance hypothesis. Powell and Eddleston (2013) study entrepreneurship outcomes across genders. They link female entrepreneurial success to positive linkages of family-to-business enrichment and support, finding that women benefit from such linkages whereas men do not. This is likely because of the female gender roles that encourage women to pursue work-family synergies. Madanoglu et al. (2020) show that home-based family firms with spousal ownership are less likely to exit that other firms. They posit that spousal owners in home-based firms achieve attain efficiencies through economizing and risk aversion driven by a desire to preserve family assets. Rodríguez-Modroño (2021) argue that the home-based self-employed male and female profiles are completely different in terms of job quality, earnings and prospects, and time spent on paid work and domestic work.

In light of this literature, we posit that female entrepreneurs working from home are different from male entrepreneurs in that they are more likely to be motivated by flexibility that allows them to more efficiently fulfill the expectations of their gender role and the needs of their business. This efficiency gain should enable women to narrow the overall gender-based performance gap with respect to the management of home-based business.

H3: The negative relationship between being home-based and performance is weaker for female than male entrepreneurs.

3 Data and methods

3.1 Data

We use the 2007 SBO Public Use Microdata Sample (PUMS) to create a firm-level dataset containing information about business operations and the owner’s demographic characteristics. The US Census Bureau conducts a firm-level economic survey once every five years to collect the data, however, the 2007 dataset is the latest iteration that was fully released for public use at this time. The SBO PUMS provides the only comprehensive, regularly collected source of information on selected economic and demographic characteristics for businesses and business owners. The sample firms in the SBO PUMS dataset were randomly selected by the Census Bureau from a list of all non-agricultural firms operating during the survey year with receipts of $1,000 or more. The responses to the survey are mandatory and authorized by Title 13 of the United States Code. Information about the sample firms’ business operations, such as receipts, payrolls, and employment, come from the Internal Revenue Service (IRS) tax filing records. The forms include the 1040 Schedule C, Firm 1065, Corporation Tax Form 1120, Form 941, or Form 944.Footnote 2,Footnote 3 Other information about the sample firms’ business and demographic characteristics are collected via mailed surveys. The response rate to the questionnaire was approximately 62%.

Table 1 presents descriptive statistics for the 2007 SBO data. The initial dataset included 2,165,683 firm records. We include only single-owner firms in our sample, which reduces the number of firms to 663,385. The first two rows of Table 1 reveal that about 36.3% of the firms are female-owned and about 57.6% of the firms are home-based businesses.Footnote 4 The revenue, labor cost, and employment variables are obtained from IRS tax records. The revenue is the total receipts in 2007 US dollars, and the labor cost is the total payroll. The Start-up capital variable is coded as a categorical variable which takes a value of 1 for start-up capital less than $ 5,000; 2 for start-up capital between $5,000 and $10,000; etc. We convert the start-up capital variable with eight categories into a dollar value by assigning the mid-range value for each category. The highest category of $1,000,000 or more gets assigned the value of $1,000,000. Female is a dummy variable equal to 1 if the owner is female. Education is a categorical variable that takes a value from 1 to 7 depending on the level of education that the owner completed. Age is a categorical variable that takes a value from 1 to 6 depending on owner’s age. Years of operation is the age of the business in years. Nonwhite is a dummy variable equal to 1 if the owner did not choose “white” as their race classification and 0 otherwise.

Table 1 Descriptive statistics: SBO
Full size table

The bottom five variables are indicator variables for firm ownership. Founder is a dummy variable equal to 1 if the owner is a founder of the business; Purchase is a dummy variable equal to one if the owner purchased the business; Inherit is a dummy variable equal to 1 if the owner inherited the business; Manage is a dummy variable equal to 1 if the owner’s primary function is managing the business; and Financial Control is a dummy variable equal to 1 if the owner’s primary function is the financial control over the firm.

The descriptive statistics reported in Table 1 indicate that firms in our sample are relatively small.Footnote 5 The average number of employees is 1.67, and the average revenue and labor cost are $286,000 and $52,400, respectively. The average educational attainment of the owners is 4.5, which is between a college dropout and an associate degree. The average age group is 3.9, which translates to between 45 and 54 years old. The average years of operation is 4.43, which is a measure of firm duration. About 11.5% of the owners are non-white. The fraction of owners who are founders of their businesses is 88.3%. The fraction of owners who reported that their primary function is to manage the business is 46.9%, whereas 37% of the owners indicated that financial control is their primary function.

3.2 Empirical framework

In order to examine the relationship between owners’ gender, workplace location, and firm performance, we employ a linear regression model. An indicator variable for home-based firms is used to measure the workplace location, along with a female-owner indicator variable for a gender effect. We are particularly interested in the relationship between gender and the firm being home-based. Therefore, an interaction term between gender and being home-based is included in the model. The results will indicate to what extent being female is associated with firm performance, and, in particular, how this association is contingent on being home-based.

We estimate the following model:

$$ \begin{array}{@{}rcl@{}} y_{i} = \beta_{0} +\delta_{1}\cdot T_{1i}+\delta_{2}\cdot T_{2i}+\delta_{3}\cdot (T_{1i} \cdot T_{2i}) +\mathbf{X}_{i}\boldsymbol{\beta} +\epsilon_{i}, \end{array} $$
(1)

where yi is the return on assets (ROA) for firm i, T1i is an indicator variable that takes a value of 1 if the business owner is female, T2i is the indicator variable that takes a value of 1 if the business is home-based, and Xi is a vector of control variables. Therefore, δ3 is our parameter of interest. The dependent variable yi is firm i’s ROA calculated as (Revenue - Labor cost)/Start-up capital. The control variables in Xi are as follows: Education, Age, Years of operation, and indicator variables for Nonwhite, Founder, Purchase, Inherit, Manage, Financial control, and the natural logarithm of Start-up capital. We also include state and industry fixed effects.

In Table 2, we present descriptive statistics for the four firm size variables (Revenue, Labor cost, Employment, and Start-up capital) and ROA, a measure of firm performance We present these statistics for home-based (top panel) and non-home-based (bottom panel) subgroups. In each subgroup, we further divide the sample into male and female-owned firms. The analysis of the top panel reveals that male-owned, home-based firms have, on average, almost double the start-up capital and more than twice the revenue and labor cost compared to the female-owned, home-based businesses. The bottom panel demonstrates that the average differences between male and female-owned businesses are greater for non-home-based firms than for home-based firms.Footnote 6 This result is consistent with the literature that finds that women business owners have less access to capital (Hundley, 2000; Loscocco & Bird, 2012) and that home-based businesses are typically smaller (Loscocco & Smith-Hunter, 2004; Kim & Parker, 2021).

Table 2 Descriptive statistics by gender and workplace location
Full size table

There are significant differences in the average ROA between male and female-owned businesses and between and home or non-home-based firms. The mean difference in ROA between female and male owners is -8.05 (= 9.95 − 18) for home-based firms, and -29.5 (= 20.5 − 50.0) for non-home-based firms. These results suggest that the gender difference in managerial performance is endogenous due to the firm size differential, as the four variables, revenue, labor cost, employment and start-up capital, proxy for firm size. In other words, we stipulate that male-owned firms, on average, perform better because their firms are bigger in terms of production inputs and outputs, and have more factors of production than female-owned firms. These findings are consistent with Edwards and Field-Hendrey (2002) who show that home-based workers are more likely to have characteristics that are associated with higher fixed costs of working on-site or a need for joint household and market productivity. Therefore, in a way, these home-based workers are more closely related to women that are out of the labor force than they are to on-site workers. Home-based workers are more likely to be living in rural areas, married, have children under age 18, or be disabled (Edwards & Field-Hendrey, 2002).

Next, we aggregate the data by state and industry and plot the average ROAs against the percentages of female-owned firms grouped by workplace location. The industries are defined using 2-digit NAICS codes. Figure 1 presents the scatter plots for state in panel (a) and for industry in panel (b). Panel 1(a) reveals some patterns: (i) home-based and non-home-based firms form two separate clusters, (ii) a downward linear trend can be observed over the two clusters. Home-based firms are clustered in the bottom right corner, and non-home-based firms are clustered in the upper left corner of the scatter plot. Panel 1(b), which shows the scatter plot by industry, does not exhibit similar clustering. However, a downward linear trend can be observed in the data. The patterns identified in Fig. 1 indicate that a higher proportion of female-owned firms in the state is negatively correlated with firm performance. Moreover, this negative relationship becomes more pronounced when separating the data into home-based and non-home-based firms. However, this negative relationship may be the result of differences in firm size, as we know that firms that are either led by women or are home-based are, on average, smaller than male-owned and non-home-based firms, as described in Table 2. Similarly, the average ROAs of female-owned firms and home-based firms are smaller than their counterparts.

Fig. 1
figure 1

Scatter plots: ROA vs female firm ratio

Full size image

Our findings in the scatter plot for state in panel 1(a) are in line with the analysis in Amore et al. (2014), who study female CEOs in Italy and their performance by region. They report that firms with a lone female CEO achieve significantly lower performance if they are located in southern Italy. This region is characterized by a more traditional view of the gender roles, where women are expected to assume the role of homemaker and men are expected to provide for the family. This implies that women may value their home production activities differently across different states and their reservation wages and hours may differ depending on their location. Therefore, women in some states may be more willing to become self-employed or open their own businesses at home, even though they can earn more by either being employed or running their businesses outside of the home.

4 Results and discussion

Table 3 presents the key results of our empirical analysis. We use the return on assets (ROA) as the dependent variable. We employ ordinary least squares (OLS) regression with state and industry fixed effects specified by Eq. 1. To evaluate the impact of home-based female ownership on a firm’s value, we compare differences in firm values between non-home-based male-led and non-home-based female-led firms to the differences in home-based male-led and home-based female-led firms. Our main coefficient of interest is therefore an indicator interaction variable HBF (Home-based× Female-owned), which takes the value of 1 for firms that are both home-based and female-owned and 0 otherwise. The HBF coefficients which are estimates of the target parameter δ3 from Eq. 1 are reported in the first row of Table 3.

Table 3 Main model estimates
Full size table

We report the coefficient estimates for firm owner’s gender (Female) and workplace location (Home-based) in the second and third rows of Table 3, respectively. Each column in Table 3 reports model estimates using different combinations of the state and industry fixed effects. We also include a set of control variables: Employment, Start-up Capital, Education, Age, Years of operation, Nonwhite, Founder, Purchased, Inherit, Manage, and Financial Control. The standard errors are clustered by state in all regression specifications. We use the natural logarithm of ROA as a dependent variable in all specifications to to account for skewness in the data.Footnote 7 Thus, we employ the (Halvorsen & Palmquist, 1980) correction to calculate the average effect of ROA as follows: 100 × [exp(δj) − 1], j = 1,2,3.

The coefficients on Female are all significantly negative at the 1% significance level. Similarly, the coefficients on Home-based, reported in the third row, are significantly negative. The size of these coefficients do not substantially vary across columns (1) to (4). Therefore, we can conclude that the state and industry variations are exogenous. We will use the results from column (4) in our discussion.

The case of T1i = 0 and T2i = 0 is the base group (male-owned and non-home-based). The coefficients on Female and Home-based are − .4652 and − .9461, respectively. After applying Halvorsen and Palmquist (1980) correction, the estimated ROA differences are − 37.20% and − 61.17%, respectively. The first result indicates that female-owned, non-home-based firms, on average, have over 37% lower ROA than male-owned, non-home-based firms. This lends support to Hypothesis 1, which states that female-led businesses achieve lower performance compared to male-led businesses. The second result implies that home-based, male-owned firms have their average ROA over 61% lower than their non-home-based male-owned counterparts, which supports Hypothesis 2, which posits that home-based entrepreneurs achieve lower performance than non-home-based entrepreneurs.

Our most interesting result is the coefficient on HBF. It is significantly positive at a 1% significance level in each of the four model estimates. The HBF coefficient in the full model estimate, reported in column (4), is 0.3331. The estimated ROA difference is therefore 39.53% (= 100 × [exp(0.3331) − 1]). This implies that if the firm is home-based, the gap in performance between men and women shrinks. To calculate the impact of ROA from being female-led and home-based we need to add the three coefficients δ1 + δ2 + δ3 = -37.20% - 61.17% + 39.53% = − 58.85%. We compare this value to the impact of being home-based on ROA for male-owned firms of − 61.17%, and conclude that the gap in ROA for female entrepreneurs vs. male entrepreneurs shrinks, or even disappears, for home-based businesses and that there is a different relationship between gender and firm performance for home-based vs. non-home-based firms.

To visualize the main result, Fig. 2 presents a margin plot of predicted ROA (non-log scale) versus location type (home-based and non-home-based). The gap between male and female for home-based firms is very small and insignificant, whereas the gap between male and female-owned firms for the non-home-based group is larger and significant. This confirms that the combination of being female-owned and home-based produces a positive synergy and that some firms can do better if they operate their businesses at home. This result provides support for Hypothesis 3 and indicates that home-based businesses led by women perform closer to their non-home based peers than those led by men. This suggests that women entrepreneurs attain advantages in the home-based setting relative to their male counterparts.

Fig. 2
figure 2

Predictive margins of home-based and female-run firms

Full size image

Turning to control variables, we find a positive association between performance and financial control, inheritance of a company, years of operation, and the educational attainment of the owner. Being non-white, a founder of the business, or having purchased the business is negatively related to firm performance. Interestingly, the age of the owner is also negatively related to small business performance. Start-up capital is also significantly negative in all model specifications, and these results are consistent with the definition of ROA where it decreases with increasing asset (capital) size for given return (profit).

4.1 Propensity score matching

Propensity score matching (PSM) ensures that home-based firms with female owners are as similar as possible to the control firms with respect to a number of observable characteristics. The model estimates the probability that a given firm with specific characteristics is a female-owned and home-based firm (Dehejia & Wahba, 2002). This method allows us to verify whether the main result we obtained in Table 3 is biased due to a sample selection problem. As shown in Table 2 and Fig. 1, the assignment of owners’ gender and workplace location is neither balanced nor random. Female-owned firms and home-based firms are smaller than their counterparts in terms of start-up capital size, revenue, labor cost, and employment. By using PSM, we obtain a subset of the control group firms with balanced propensities, or likelihoods, of being female-owned, home-based firms. Then, the mean comparison of ROA between the female-owned, home-based firms and control group firms can identify and estimate the effect consistently. We obtain the predicted propensities from a probit regression with the female-owned and home-based firm indicator as a dependent variable. All the control variables used for estimating the model Eq. 1 are used as control variables in the probit regression estimation.

Figure 3 presents the steps of the PSM process. The histograms in Fig. 3 are the predicted likelihoods from the probit regression. Panel (a) presents the “treatment group” distribution, which we aim to replicate using the control group observations. The control group observations distribution for the full sample is presented in panel (b). Panel (b) has a different shape and interval from the one in panel (a). PSM first eliminates the control group observations that are out of the treatment group interval. The trimmed control group is presented in panel (c). The interval in (c) becomes identical to panel (a), but the distribution is still different than (a). Using this cut-off control group sample, the PSM then chooses observations from the control group to make an identical shape of the distribution in panel (a). Panel (d) presents the matched control group’s distribution. The shape of the new propensity -score matched distribution looks quite close to the target distribution in panel (a) (Table 3).

Fig. 3
figure 3

Propensity scores: distribution

Full size image

The PSM estimate is consistent with the full sample estimates and significantly positive. Table 4 reports the mean comparison of the natural logarithm of ROA in female-owned, home-based firms and controls of the matched and unmatched sample group. The mean difference in the unmatched group is − 0.0729 and statistically significant at the 1% significance level. In contrast, the mean difference of the matched sample is 0.063 and is statistically significant at the 5% significance level. The PSM estimation suggests that estimation of the effect of female-owned and home-based firms might be biased due to sample selection.

Table 4 Propensity score matching estimation
Full size table

Next, we replicate the analysis from Table 3 and estimate Eq. 1 using the propensity score-matched sample. Table 5 reports the results. In our new sample, we matched 9,275 treatment observations to 7,711 control observations. The Treated observations are female-owned and home-based. The Controls may be male-owned and home-based, male-owned and non-home-based or female-owned and non-home-based. The results in Tables 3 and 5 are identical in terms of coefficient signs and significance. HBF coefficients in all four models in columns (1) to (4) are significantly positive, all of the four coefficients of Female are significantly negative, and all of the four coefficients on Home-based are significantly negative. These results indicate that the OLS estimation of the model Eq. 1 is not affected by endogeneity due to sample selection.

Table 5 Main model estimates for the matched sample
Full size table

4.2 Subsample analysis

The literature indicates that the industry in which a firm competes is an important dimension when examining female-owned businesses and their success (Loscocco & Bird, 2012; Yang & del Carmen Triana, 2019). Thus, we examine whether the results are driven by the fact that a firm belongs to a specific industry. Industries are defined using North American Industry Classification System (NAICS) 2-digit industry classifications. Table 6 reports the results of the regressions specified in Eq. 1 calculated by industry. As expected, we find a notable heterogeneity in our results across different industries. The results imply that the estimates differ substantially by industry in terms of coefficient signs and significance. The HBF estimates are significantly positive in the following: retail trade, arts and entertainment, other services, information, real estate, and professional, scientific and technical services. The HBF estimates of the other six industries: wholesale trade, educational services, health care, finance and insurance, management of companies and enterprises, and accommodation and food services are insignificant at the 10% significance level. In contrast, the coefficient estimates of Female and Home-based are consistently negative at the 1% significance level for all industries but the management of companies and enterprises.

Table 6 Main model estimates by industry
Full size table

This heterogeneity in our results could be driven by the presence of industry-specific production technology. Some industries are more compatible with home-based work; some require an on-site location. To explore this further, we look at the female owner firm ratio by industry, both for home-based and non-home-based businesses. Figure 4 reports the results. The proportion of home-based firms within female-owned firms is relatively large in, for example, both retail and health-care industries, but their HBF estimates are quite different. Also, educational services seem to be dominated by women, as the proportion of female owners exceeds 50% for both home-based and non-home-based firms. However, the HBF estimate for this industry is not statistically significant. This shows that different proportions of female-owned and home-based firms do not explain the heterogeneity in our results.

Fig. 4
figure 4

Female-owned firm ratio by industry

Full size image

To further ensure the robustness of our results, we estimate Eq. 1 by age and education level. The model estimates by owner’s educational attainment are reported in Table 7. Overall, the coefficient estimates are consistent with the main result in Table 3 in terms of sign and significance. All of the HBF coefficients are significantly positive, whereas the coefficients for Female and Home-based are significantly negative. The HBF coefficient for graduate degrees (master or higher) is smaller than for other categories, even though the graduate degree is the highest educational attainment category. This might be partially explained by the fact that the coefficient for Female is greater (− 0.3691) for this education category than for any other education level.

Table 7 Main model estimates by educational attainment
Full size table

The ROA difference between home-based and non-home-based firms is greater when the owner is male, for all education levels. Figure 5 presents the weighted average ROA for home-based firms and non-home-based firms by owner’s gender and educational attainment. For every category, the average ROA for male-owned firms is greater than for their female-owned counterparts. In addition, the average ROA difference between home-based firms and non-home-based firms is greater for male owners. Higher education does not seem to either narrow the ROA gap between genders or between home-based vs. non-home-based.

Fig. 5
figure 5

Bar plots: ROA by gender and workplace

Full size image

The model estimates by firm owner’s age, reported in Table 8, are also consistent with the main result in Table 3 in terms of the coefficient’s sign and significance. All HBF coefficients are significantly positive, except for the youngest group, under 25. Also, the effect increases in firm owners’ age from age 26 until age 64. The group over 65 exhibits a sizable effect, but not as high as the firm owners between 45 and 64 years old. The HBF coefficient for the youngest and smallest group, under 25, is insignificant. The coefficients for the indicator for female-owned and home-based firms are significantly negative for all age groups.

Table 8 Main model estimates by age group
Full size table

4.3 Study limitations

There are certain limitations to the study. The dataset is from 2007. It is possible that the labor market circumstances of women and the division of their time between market and non-market work have changed due to technological advances in recent years or changes in gender role expectations.Footnote 8 The SBO data is collected every five years; however, the 2007 dataset is the only set of US Census data that has been fully released for public consumption (Kiefer et al., 2022). Other years, including the more recent ones, have heavy use and access restrictions. Despite this shortcoming, due to the richness of the dataset, our paper offers unique and novel insights.

Another limitation is that although the dataset is very comprehensive, it does not provide certain individual factors such as marriage, number of children, age of the children, or the accessibility of childcare. These variables would allow for a more complete understanding of the differences in the performance of firms due to family obligations. A recent UK study finds that women entrepreneurs who have childcare responsibilities are more likely to be home-based than their male-counterparts (Kim & Parker, 2021). Therefore, it is possible that homeworking provides a more efficient way for women to juggle business and caregiving duties.

Lastly, we employed a cross-sectional design. This means that no insight could be gained into how the relationships may change over time. Using longitudinal data would allow us to examine how female entrepreneurs’ relative performance changes overtime due to technological changes in the marketplace and examine the effects on being business that transfers to/from home-based or female owned. These tasks are left for future research.

5 Conclusion

Our study provides a comprehensive analysis of small businesses in the USA which confirms previous findings that men benefit from an advantageous position in society which translates into a gender-based performance gap among small-business entrepreneurs. While this result was consistent across much of the sample, we discovered a striking reverse of this effect with respect to female-owned businesses operating from home. This surprising result can best be explained in the context of literature establishing that a significant source of women’s disadvantages originates from a lack of sufficient workplace flexibility needed to fulfill their societal expectations, as well as the inequitable distribution of these expectations between men and women.

Our study addresses a mostly overlooked aspect of entrepreneurship. The literature to date shows that women are disadvantaged when establishing their own businesses in a multitude of ways, such as access to funding, social capital, and cultural capital. We add to these discussions by showing that working-from-home is affecting women-led businesses differently, and positively, relative to their male-led counterparts.

Our research identifies a business structure in the economy (home-based businesses) which enables women to narrow the gap in performance with their male counterparts. The research suggests that granting women the autonomy and flexibility serves to mitigate disadvantages resulting from the greater demands society places on their time with respect to non-market household work. Our results imply that significant efficiency gains may be obtained by granting women greater flexibility over their schedules.

Policymakers, lenders, and investors should look to alleviate obstacles home-based business owners, particularly female business owners, face in gaining access to capital. Local governments and home-owner associations should consider removing or lessening restrictions on operating home-based businesses and/or increasing tax benefits for such businesses. Finally, employers may benefit by ensuring that female employees, particularly those leading semi-autonomous business units or projects, are encouraged, and not penalized, to opt to work from home.