Abstract
Data collected on self-employed women and men in one county allow examination of work effort, housework effort, housework hours, and preference for flexible work on earnings. Regressions indicate housework effort of self-employed women contributes to their lower earnings. Housework hours do not supporting the view women select self-employment to find flexible work. Housework hours do reduce the earnings of self-employed men, which could reflect their stronger commitment to housework combined with less flexible work. A Oaxaca decomposition suggests less tenure and greater housework effort are important contributors to lower earnings of self-employed women. Ranges that measure earnings may contribute to the insignificance of work effort, normalized work effort, and preference for flexible work hours. (J16, J23)
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
These studies suggest the female/male median earnings ratio is 30% when both full-time and part-time self-employed are included.
Hundley (2000, 2001) found housework hours did not influence the earnings of self-employed men. Since self-employed men specialize in market work we expect housework hours do not impose time constraints that reduce their earnings. In the analysis below, then, we anticipate the housework hours coefficient will not be significant in the earnings estimates of self-employed men.
St. Croix County is located in western Wisconsin. The county forms a border with Minnesota and is included in the Minneapolis-St. Paul Metropolitan Statistical Area. St. Croix County has the highest population growth rate in the state of Wisconsin and ranks tenth in per capita income (Wisconsin Department of Workforce Development 2005). Also, in this study a person is self-employed if they are a sole proprietor, partner, owner of a unincorporated business, owner of an incorporated business, or an independent contractor.
Reference USA provides information that includes the firm’s contact person, title of contact person, and a phone number. In the list generated for St. Croix County most cases indicated “owner” under title of contact person. This designation identified the firm as self-employed. Some cases, however, were ambiguous because title of the contact person indicated “manager” or was blank. In these instances, firms were contacted by phone to determine if they should be categorized as self-employed. Self-employment status was also determined using the web page of the Wisconsin Department of Financial Institutions (www.wdfi.org). This provides information on owners of incorporated self-employed in the state. An additional problem involved identification of real estate agents and financial representatives (categorized as independent contractors). The Reference USA directory indicates the company and only one agent as the contact person. The web page of each company allowed identification of the remaining agents.
To maximize the response rate six tickets to the Minnesota Vikings-Green Bay Packers football game (which took place December 24, 2004 in Minneapolis) were purchased. The cover letter informed respondents that returning a completed survey would make them eligible to win a pair of the tickets in a random drawing. The 44% response rate, which is high for surveys of small businesses, suggests this was an effective tool.
Effort was defined as “the energy you put forth to perform a task competently in terms of mental requirements, physical requirements, and stress.” Beside each activity (work, household chores, and watching television) was a scale beginning at one and ending at eleven. The range 1 to 5 was identified as “less effort” and 7 to 11 as “more effort.”
“Provides flexible work hours” was the factor indicated in the questionnaire. Other factors included “Can be my own boss,” Had a good idea for a product,” and “Can advance my career.” To promote consistency with the effort measure (and facilitate ease of survey response) the question also employed an eleven-point scale. In this case, however, the range 1 to 5 was identified as “less important” and 7 to 11 “more important.”
The questions measuring starting capital and housework hours also used ranges. Starting capital used the same categories as net income with 37.4% of self-employed men and 22.8% of self-employed women identifying $77,000+. This suggests self-employed men also have a higher degree of understatement in the value of starting capital. The housework hours question presented seven categories. The first six increased by four-hour increments and the seventh indicated 30+ hours. A higher proportion of self-employed women (12.8%) identified 30+ hours than men (5.8%). Housework hours of self-employed women, however, do not appear to be understated. Table 2 (see below) indicates self-employed women in this sample spend 16.7 h per week at housework. This is lower than the 20.48 h per week in Hundley (2001). At the same time, self-employed women in this study work more hours per week (44.65) compared to 29.13 reported in Hundley (2001). The lower value of housework hours is consistent with the greater number of work hours.
When examining H1 using housework effort and normalized housework effort the alternative form suggests a negative relationship between these variables and net income. Becker (1985) assumes the energy available to an individual is limited. As a result, the more effort an individual expends on housework implies less effort for market work.
The correlation coefficients also do not suggest a tradeoff. For self-employed women the correlation between work effort and housework effort is .17. The correlation between normalized work effort and normalized housework effort is .69 indicating a stronger positive relationship. The value of these coefficients for self-employed men is .097 and .66. The correlation between work hours and housework hours does indicate a tradeoff. In this case the correlation coefficients for self-employed women and men are, respectively, -.16 and -.145.
Hundley (2001) indicated a higher proportion of self-employed men (7%) located in finance/real estate compared to only 4% for self-employed women.
Self-employed women operating on a small scale may not list their business in local directories, which is a main source of information for Reference USA. Devine (1994) used the 1990 CPS March Supplement while Hundley (2000, 2001) employed the NLS-72 and the 1989-90 wave of the PSID. These probability samples, drawn from national populations, may be more likely to include small scale self-employed.
Results indicate a correlation between housework hours and flexible work hours of .17, suggesting multicollinearity is not contributing to the insignificance of these coefficients in the women’s estimates. Regressions were also estimated dropping one of the variables and including the other. The coefficients remained insignificant.
Hundley (2001) reported annual average housework hours for self-employed men of 338. Dividing 52 into 338 we obtain 6.5 h per week.
The 2000 Census classifies 57% (36,201 out of 63,155) of residents in St. Croix County as rural. Information provided by Office of Social and Economic Trend Analysis (SETA) at Iowa State University. Retrieved July 19, 2007, from http://seta.iastate.edu.
Examination of education indicated over 50% of high-income ($77,000+) self-employed women and men identified schooling at a bachelor’s degree or above. Regarding tenure, 54% of high-income self-employed men owned their firms for more than the mean value of 15 years. Work hours indicated 70% of high income self-employed women worked more than 40 h per week. In each case, the coefficients could indicate positive effects on earnings depending on the degree to which actual income exceeds the $77,000 cap.
Reference USA provides information on the start year for a firm. This allows estimation of the distribution of tenure in the population. The population and sample distributions of tenure for self-employed women indicate a much closer correspondence. In the population, the distribution of years of firm ownership for women is 41.9% (5 years or less), 40.5% (6 to 15 years), and 17.6% (over 15 years). The sample indicates 38.5% (5 years of less), 39.7% (6 to 15 years), and 21.8% (over 15 years). We note the population estimates for self-employed women and men exclude real estate agents and financial representatives not listed in the Reference USA directory.
The Census Business Information Tracking Series (BITS) 2003–04 was examined to determine if a higher failure rate among self-employed men could be a reason for the low survey response among those with less tenure. The BITS provides information on “deaths” and “births” of firms by two-digit NAICS occupation at the county level. To estimate gender differences in deaths (and births) occupations were collapsed into the six categories used in this study. The Reference USA list generated for this study (which also provides NAICS occupation codes) was used to estimate the proportion of women to men in each occupation. Deaths and births in each occupation, by gender, were estimated based on the Reference USA women to men proportions.
In 2003–04, total deaths in St. Croix County were 135 compared to births of 206. Estimation of gender differences suggest men had a higher death rate (65% compared to 35%) and birth rate (66% compared to 34%) than women. If these estimates reflect trends among self-employed then the higher death rate may be contributing to the low response of self-employed men with less tenure. The higher birth rate, however, suggests self-employed men with less tenure were not inclined to respond to the survey. A possible reason could be that, as new entrants, these self-employed men felt additional competitive pressure to gain a foothold in their occupation. As a result, they may have been less willing to reveal information considered proprietary. In addition, they may have viewed the time and energy needed to complete the survey questionnaire as too costly. The U.S. Census BITS 2003–04 retrieved December 18, 2008, from http://www.census.gov/csd/susb/susbdyn.htm.
Oaxaca’s (1973) method assumes the difference in average earnings between men and women can be decomposed into:
\( \ln \overline{Wm} - \ln \overline{Wf} = (\overline{Xm} - \overline{Xf} )\beta f + (\beta m - \beta f)\overline{X} m \)
or
\( \ln \overline{Wm} - \ln \overline{Wf} = (\overline{Xm} - \overline{Xf} )\beta m + (\beta m - \beta f)\overline{X} f \)
The first term in each equation represents the portion of the difference in average wages between men and women due to their different attributes. The second term indicates the portion of the difference in average wages due to differences in rates of return to attributes men and women receive in the market. The analysis above focuses on the first term in these equations.
The total decomposition is included because of wide variation in results when applying coefficients from the female and male equations. This decomposition employs \( \ln \overline{Wm} - \ln \overline{Wf} = (\overline{Xm} - \overline{Xf} )\beta T \) where β T indicates coefficients estimated after pooling the samples of self-employed women and men. β T includes, in addition to the controls in model (3), a gender dummy variable. Consistent with the previous estimations, observations are limited to those in model (4) which has the largest number of missing values. Table B1 (in Appendix B) reports these regression results. The calculations determining results in Table 4 are based on:
Column (1)—Total Coefficients—\( (\overline{Xm} - \overline{Xf} )\beta T/(\ln \overline{W} m - \ln \overline{W} f) \)
Column (2)—Female Coefficients—\( (\overline{Xm} - \overline{X} f)\beta f/(\ln \overline{W} m - \ln \overline{W} f) \)
Column (3)—Male Coefficients—\( (\overline{Xm} - \overline{X} f)\beta m/(\ln \overline{W} m - \ln \overline{W} f) \)
Calculations in column (1) do not include the percent of the earnings gap attributed to the difference in the proportion of women to men in the sample. This is consistent with the focus on attribute differences when applying the female and male coefficients. In addition, this calculation would understate the contribution of gender differences to the earnings gap since the sample overstates the proportion of women to men who are self-employed in St. Croix County.
Additional regressions examined if the larger work hours coefficient in earnings model for self-employed men (compared to self-employed women) is sensitive to model specification. In these estimates flexible work hours, housework effort, and housework hours were successively dropped with no change in the work hours coefficient.
Additional regressions examined the sensitivity of housework effort (in the women’s estimates) and housework hours (in the men’s estimates) to model specification. After successively dropping flexible work hours and housework hours the housework effort coefficient increased to -.063 (from -.11) suggesting some sensitivity. The housework effort coefficient remained significant at the 5% level. The contribution of housework effort to the earnings gap based on the -.063 coefficient is still relatively large at 15.2%. The housework hours coefficient in the men’s estimate did not change after successively dropping flexible work hours, housework effort, and work hours.
References
Aronson RL (1991) Self-employment: A Labor Market Perspective. ILR, Ithaca, N.Y
Barzel Y (1987) The Entrepreneur’s Reward for Self-Policing. Economic Inquiry 25(1):103–16
Becker GS (1985) Human Capital, Effort, and the Sexual Division of Labor. Journal of Labor Economics 3(1):S33–S58
Bianchi SM, Robinson JP, Milkie MA (2006) Changing Rhythms of American Family Life. Russell Sage Foundation, New York
Blau FD, Kahn LM (2000) Gender Differences in Pay. Journal of Economic Perspectives 14(4):75–99
Bielby DD, Bielby WT (1988) She Works Hard for the Money: Household Responsibilities and the Allocation of Work Effort. American Journal of Sociology 93(5):1031–59
Carr D (1996) Two Paths to Self-Employment. Work and Occupations 23(1):26–53
Dillman DA (1978) Mail and Telephone Surveys: The Total Design Method. Wiley, New York
Devine, Theresa J. 1994. “Characteristics of Self Employed Women in the United States,” Monthly Labor Review, March: 20-34.
Duncan GJ, Holmlund B (1983) Was Adam Smith Right After All? Another Test of the Theory of Compensating Wage Differentials. Journal of Labor Economics 1(4):366–79
Hersch J, Stratton LS (2000) Household Specialization and the Male Wage Premium. Industrial and Labor Relations Review 54(1):78–94
Hundley G (2000) Male/Female Earnings Differences in Self-Employment: The Effects of Marriage, Children, and the Household Division of Labor. Industrial and Labor Relations Review 54(1):95–114
Hundley G (2001) Why Women Earn Less than Men in Self-Employment. Journal of Labor Research 22(4):817–29
Jacobs, Jerry. 1998. “Measuring Time at Work: Are Self-Reports Accurate?,” Monthly Labor Review, December: 42-53.
Korenman S, Neumark D (1991) Does Marriage Make Men More Productive? The Journal of Human Resources 26(2):282–307
Korenman S, Neumark D (1992) Marriage, Motherhood, and Wages. The Journal of Human Resources 27(2):233–55
Krueger AB, Summers L (1988) Efficiency Wages and the Inter-Industry Wage Structure. Econometrica 56(2):259–93
Kuhn P, Lozano F (2006) Why High Earners Work Longer Hours? NBER Working Paper, Number 11895:1–47
Lombard KV (2001) Female Self-Employment and Demand for Flexible, Nonstandard Work Schedules. Economic Inquiry 39(2):214–37
MacPherson DA (1988) Self-Employment and Married Women. Economics Letters 28(3):281–84
Oaxaca R (1973) Male-Female Wage Differentials in Urban Labor Markets. International Economic Review 14(4):693–709
Stratton LS (2001) Why Does More Housework Lower Women’s Wages? Testing Hypothesis Involving Job Effort and Hours Flexibility. Social Science Quarterly 82(1):67–76
Schumpeter JA (1934) The Theory of Economic Development: An Inquiry into Profits, Capital, and Credit. Harvard University, Cambridge
Taniguchi H (2002) Determinants of Women’s Entry into Self-Employment. Social Science Quarterly 83(3):875–93
U.S. Bureau of the Census. 2004. Business Information Tracking Series, http://www.census.gov/csd/susb/susbdyn.htm). Retrieved December 18, 2008.
Wisconsin Department of Workforce Development. 2005. “St. Croix County Profile.” Retrieved September 1, 2006, from http://dwd.wisconsin.gov/oea/cp_pdf/2005/s109cpw.pdf
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix A
A total of 30 self-employed women and 70 self-employed men in the sample identified net income of $77,000. Table A1 indicates the proportions of these high income self-employed who identified values in the upper range (8–11) of the work effort and normalized work effort measures. Also indicated are the proportions of these self-employed identifying values in the lower range (1–4) of housework effort, normalized housework effort, and preference for flexible work hours.
Table A1 indicates a high percentage of these self-employed women and men identified values in the upper range of work effort and normalized work effort. If the actual net income of these respondents greatly exceeds the $77,000 cap then the coefficients in the regression estimates will not fully capture the positive effect higher levels of work effort have on earnings. Results in Table A1 also indicate a substantial proportion of high income self-employed with values in the lower range of normalized housework effort. Again, if actual net income exceeds the $77,000 cap the normalized housework effort coefficient will not fully capture the inverse relationship between less effort expended on housework and higher earnings. This is relevant to self-employed women since Table 3 indicates the normalized housework coefficient in model (4) is negative and close to an acceptable level of significance.
Table A1 indicates smaller proportions of high income self-employed that identified low values of preference for flexible work hours. In addition, a small proportion of high income self-employed men indicated low values of housework effort. Even though the percentages are low, we cannot rule out the actual income these self-employed significantly exceed the $77,000 cap. In this case, the average inverse relationship between low values of these measures and net income could be significant.
Appendix B
Rights and permissions
About this article
Cite this article
Walker, J.R. Earnings, Effort, and Work Flexibility of Self-Employed Women and Men: The Case of St. Croix County, Wisconsin. J Labor Res 30, 269–288 (2009). https://doi.org/10.1007/s12122-009-9067-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12122-009-9067-4