Independent workers constitute an increasingly visible feature of the American economy, but employment that is temporary and precarious in nature is not new. From the freelance writer to the “temp” hired for the holiday season to the warehouse driver whose employment has been contracted out, nontraditional work has existed in our economy for decades.
Independent workers, who aren’t part of a traditional employer-employee relationship, generally don’t have access to retirement plans or health insurance through their employers. New data from the U.S. Bureau of Labor Statistics (BLS) show that 18.4 percent of “contingent workers” (BLS describes them as workers who do not expect their jobs to last) participated in employer-sponsored pension or retirement plans in 2017, compared to 43.4 percent of workers in more traditional and permanent employment arrangements. Other researchers have found similar results.1 Comparing different researchers’ measurements of independent workers is complicated by the various definitions they use and by shortcomings in survey and administrative data.
The Pew Charitable Trusts analyzed the University of Michigan’s Health and Retirement Study (HRS)2 to better understand the financial condition of self-employed workers on the cusp of retirement. The assessment examined various types of self-employment, in single- and multi-person firms, to see how different arrangements affect retirement readiness measures such as retirement plan assets, prospective Social Security wealth, and participation in workplace retirement plans—either a defined contribution3 or a defined benefit plan.4 The research sought to understand whether having a spouse or partner who participates in a workplace retirement plan might affect a self-employed worker’s retirement readiness. Pew looked at whether the covered spouse saved more than average in order to improve retirement readiness for the couple.
About 8.6 percent of workers ages 50-64 in the HRS database were self-employed in single-person firms (the “solo” self-employed), while another 8.3 percent worked in firms of two or more people. Self-employed workers in this age group were more likely to be male, white, and well educated, and this did not vary with firm size. Self-employed women were much more likely to be in couples than self-employed men or traditional workers. Overall, many self-employed workers near retirement, no matter the firm size, appear to be less prepared for retirement than workers in traditional employment relationships. The self-employed ages 50-64 have lower rates of participation in workplace retirement plans. They also claim Social Security earlier, which reduces monthly benefits for life.
Pew selected workers who reported being self-employed in both 2012 and 2014, in order to exclude retirees and those who were self-employed just temporarily. Pew divided the self-employed into those who worked in single-person firms versus in multi-person firms in 2014. Multi-person firm arrangements, including legal, medical, and other practices, may be sufficiently large to suggest long-term employment stability because many larger firms enjoy relatively stable clientele and income. However, this methodology cannot definitively distinguish between financially insecure self-employed workers versus those who may have more financial stability in the form of long-term clientele or practices. There is no practical way to use HRS variables to pull out these different workers. Breakouts by profession and industry suffer from very small sample sizes, even when collapsed into broader categories such as the “knowledge” professions.5
This study finds that those who are self-employed in single-person firms have a more precarious financial situation than other workers, which could feed retirement insecurity.
- The solo self-employed had the lowest median total household incomes and the highest household poverty rates of all workers.
- About 13 percent of self-employed workers in single-person firms reported participating in retirement plans at their current jobs, compared to almost three-quarters of traditional workers.
- Among workers who do participate in a workplace retirement plan, the solo self-employed had lower mean (but not median) savings balances in defined contribution plans than traditional workers.
- The solo self-employed were more likely than traditional workers to be married, and their spouses were slightly more likely than the spouses of other workers to be employed full time. Their spouses were also more likely to participate in pension plans in their own jobs but had lower mean (although not median) balances than the spouses of traditional workers.
Self-employed workers in multi-person firms do the best of all—surpassing traditional employees—on measures such as income. Although they are less likely than traditional employees to participate in workplace retirement savings plans, when they do have these plans, their average balances are higher than those of all other workers.
- Those at multi-person firms work more hours a week than either traditional employees or the solo self-employed, although they worked fewer weeks per year than traditional employees. They earned the highest hourly and weekly wage rates of all workers.
- About 29.9 percent of self-employed workers in multi-person firms reported participating in retirement plans at their current jobs.
- These workers were more likely to be married than other workers in self-employment and traditional jobs. They were also more likely to have a working spouse than traditional (but not solo self-employed) workers. The working spouse was more likely to be self-employed than the working spouses of other respondents. As a likely consequence of being selfemployed themselves, these working spouses were less likely to have their own pension coverage (a defined benefit or defined contribution plan) than the spouses of solo self-employed and traditional workers.
- For spouses that had defined contribution plans, however, savings balances surpassed those of the spouses of the solo self-employed and traditional workers.
Gender, race, and ethnicity
Men are more likely to be self-employed than women; they are also more likely to work in multi-person firms than in solo self-employment. When women are self-employed, they are more likely to be in solo self-employment.6
White and Hispanic workers are more likely to be self-employed than non-white, non-Hispanic workers, a category that includes non-Hispanic blacks, Asians, and Pacific Islanders (data limitations prevented further disaggregation).7
Examining the data within each employment category (not shown), men constitute more than half (56.7 percent) of the self-employed workers in single-person firms and two-thirds (67.9 percent) of the self-employed workers in multi-person firms.
Mean age was about 59 years and 8 months for solo self-employed workers, and about 59 years and 2 months for the self-employed in multi-person firms. Traditional workers in the dataset were about 59 years old.
The self-employed tend to be more highly educated than those who were never self-employed, although the differences are not highly significant.8 Education categories were combined to build up cell sizes and statistical significance. Within educational groups, about 21.3 percent of those with law degrees, medical degrees, or doctorates were self-employed in either single-person firms (6.0 percent) or multi-person firms (15.3 percent), compared to about 17.0 percent of those with a high school education or less. Those with MBAs or other master’s degrees were the least likely (12.6 percent, including 7.8 percent in single-person firms and 4.8 percent in multi-person firms) to be self-employed.
Looking at the data within employment categories (not shown), about 40.1 percent of self-employed workers in single-person or multiperson firms had some college or more, compared to 37.0 percent of workers who were not self-employed in 2012 or 2014.
Those who were self-employed in 2012 and 2014 were slightly more likely to be coupled (in marriages or partnerships) than uncoupled (separated or divorced, widowed, or had never married) in 2014, although the analysis cannot reject the null hypothesis that there was no difference.9 Self-employed women ages 50-64, however, were significantly more likely to be in couples.10
These results raise a question as to whether, for some self-employed workers, a spouse or partner’s employment might provide a cushion against the vagaries of self-employment. A spouse’s job might shield against volatile or low income, or it might bring benefits such as health insurance or the availability of a workplace retirement plan to the household. Having a spouse may be particularly beneficial to women in self-employment, because they tend to earn less than women in traditional work and compared to men in general.
Labor force status
The self-employed are more likely to work part time or to describe themselves as partially retired than workers who were not self-employed in 2012 and 2014. About half (49.7 percent) of self-employed workers in single-person firms described themselves as part-time workers or as partially retired, as did 27.5 percent of workers who were self-employed in multi-person firms. By contrast, just 18.9 percent of workers who had not been self-employed in either 2012 or 2014 described themselves as working part time or as being partially retired in their main jobs.11
These results may reflect confusion over the meaning of “retirement” in the survey and in general. But these results may also provide evidence that some view their current self-employment as different from career employment, for example, as a bridge between career work and full retirement.12
Hours and weeks worked
Workers who are self-employed in single-person firms spend fewer hours per week, and fewer weeks per year, at their jobs than other workers. This is consistent with the finding reported in Figure 3, that self-employed workers in singleperson firms are more likely to work part time, or to be partially retired, than other workers. Those who are self-employed in multi-person firms log the highest average weekly hours of all.
The self-employed who worked in multi-person firms in their primary jobs realized the highest hourly and weekly wage and salary rates of all workers, surpassing those in traditional employment in 2014.13 Median hourly wage rates (gross) are lowest for self-employed workers in single-person firms ($16.67 per hour).14
Spousal work status
The spouse or partner of a self-employed respondent is somewhat more likely to be working either full-time or part-time, and less likely to be out of the workforce (retired, unemployed, or disabled) than the spouse or partner of one in a traditional work relationship.15 This may suggest that couples in which one member is self-employed are more likely to have a cushion as one spouse works a full-time job. This cushion may provide a steady or higher income, smoothing out the uncertainty that can stem from self-employment, or access to a workplace retirement plan for the household.16
The spouses of self-employed workers are more likely to be self-employed than the spouses of traditional workers. As noted earlier, self-employed respondents are slightly more likely than traditional workers to have spouses or partners (Figure 2), and their spouses are in the workforce at somewhat higher rates than the spouses of traditional workers (Figure 4). Among respondents in traditional work relationships, only 16.8 percent of employed spouses or partners were self-employed in either single- or multi-person firms (the percentages given here sum to 100 percent by each type of respondent work status). But among respondents who were self-employed in singleperson firms, about 26.4 percent of their spouses or partners were also self-employed. Notably, over half (51.6 percent) of respondents who were self-employed in multi-person firms had spouses or partners who were also self-employed in either single- or multi-person firms.17 This may reflect the prevalence of family businesses, or there could be another unidentified explanation.
Household income and poverty
Household-level analysis offers another perspective on the financial stability of self-employed workers. When median total household earnings are calculated, including spousal income, households where the respondent was self-employed in a single-person firm had the lowest median total incomes and the highest household poverty rates.18 Traditional workers had the lowest poverty rates. Households where the respondent was self-employed in a multi-person firm fared well, with the highest median total household income.19 Their poverty rates fell to between those of traditional workers and self-employed workers in singleperson firms. This may be due, in part, to the fact that respondents who worked in multi-person firms were more likely to be married and to have working spouses.
Social Security benefits
Among those ages 50-64 who have not yet retired, workers who are self-employed in single-person firms are more than twice as likely to be receiving any type of Social Security benefits (old age or disability) as those who were not self-employed in 2012 or 2014.20 They were also much more likely to be receiving Social Security than those who were self-employed in multi-person firms.
Claiming Social Security retirement benefits before normal retirement age reduces these benefits by about 5 percent to 6 percent a year.21
There are a few reasons why self-employed workers might claim Social Security before retiring from the workforce. The self-employed— particularly those working in single-person firms—often have lower earnings. The nature of many self-employment jobs suggests their incomes may also be more volatile and claiming Social Security may provide a steady cushion of income. Additionally, lower incomes are less likely to be subject to the Social Security retirement earnings limit.22
Social Security wealth
The Health and Retirement Study predicts the Social Security wealth of individual workers and their households. This is effectively a present value calculation of expected future Social Security payments and is one measure of future benefit levels.
Self-employed workers in singleperson firms have the lowest personal and household median Social Security wealth. This may be because they lacked Social Security coverage through an employer; because they worked sporadically in Social Security-covered employment; because many self-employed workers had lower incomes than traditional workers; or other traits related to being self-employed. In addition, one study found that more than 3 million self-employed people did not disclose some or all of their earnings to the IRS in 2014, which would translate to lower future Social Security benefits.23 Traditional workers and their households have the largest median predicted Social Security wealth.
Participation in workplace retirement plans
The self-employed are often less likely to be included in a pension— either a defined contribution plan or a defined benefit plan—at their current jobs. The self-employed who work in single-person firms have the lowest workplace pension participation rates of all.24
Many self-employed workers consequently lack the behavioral “nudges” that can be attached to employer-sponsored retirement plans to encourage participation and garner significant levels of savings. Employer-based nudges include auto-enrollment, default contribution rates and investments, and sometimes even auto-escalation of contribution rates. By contrast, self-employed workers need to identify, establish, and contribute to a Savings Incentive Match Plan for Employees or a Simplified Employee Pension plan,25 or individual retirement accounts, on their own. While they may represent a smaller group of the self-employed, those who do so may be motivated to make regular contributions.
Spousal participation in workplace retirement plans
When self-employed workers in single-person firms do not participate in workplace retirement plans (defined benefit or defined contribution plans), their spouses are slightly more likely than the spouses of traditional workers to participate in a workplace retirement plan.26 These results suggest that spousal benefits packages may provide some cushion for the self-employed in single-person firms.
When self-employed workers in multi-person firms do not participate in a workplace retirement plan, however, their spouses are somewhat less likely than the spouses of traditional workers to participate.27 A partial explanation may be that more than half of self-employed workers’ spouses are self-employed themselves, with similarly low workplace retirement plan access and participation (Figure 5).
It is possible that some respondents weren’t aware of their spouses’ participation in a workplace retirement plan.
Respondent’s retirement plan assets
Among workers with defined contribution plans at their current jobs, self-employed workers in single-person firms have lower mean balances than traditional workers. The self-employed in multi-person firms have the highest mean balances.28
Median balances tell a slightly different story. Workers who were self-employed in single-person firms had roughly equal median balances as traditional workers. Both had much lower median balances than the self-employed in multi-person firms.29 Some explanation may be that self-employed workers in multiperson firms work longer hours (Table 2) and earn higher wages (Table 3) than other workers.
Self-employed workers in singleperson firms averaged about one year more at their current jobs than traditional workers, but their balances were lower.30
Spouse’s retirement plan assets
Spouses who participated in a workplace defined contribution plan had lower mean balances when the respondent they were married to was self-employed in a single-person firm. The spouses of self-employed workers in multi-person firms had the highest mean balances.31
Median figures again tell a different story. This time, among spouses who contributed to a workplace defined contribution plan, the spouses of workers in single self-employment had higher balances than spouses of workers in traditional employment.32
When spouses participated in workplace retirement plans, the spouses of workers in singleperson firms had higher retirement savings balances as a ratio to total household income (80.6 percent) than the spouses of traditional workers (61.1 percent of household income). The spouses of workers in multi-person firms were less likely to participate in workplace retirement savings plan than other spouses (Figure 9), but when they did participate, their balance ratios were 133.3 percent of household income.
Some self-employed workers ages 50-64 appear to be less prepared for retirement than workers who were not self-employed. The self-employed ages 50-64 have lower rates of participation in workplace retirement plans. The self-employed also claim Social Security earlier, which reduces monthly benefits for life.
Among those who are self-employed in single-person firms, the combination of fewer work hours, fewer weeks worked per year, lower hourly and weekly wage rates, and lower rates of participation in workplace retirement plans suggests a more precarious financial situation that could feed retirement insecurity. Among workers who do have workplace defined contribution plans, those who were self-employed in single-person firms have lower average (mean) retirement savings balances than traditional workers or those in multi-person firms, although their median balances were similar to those of traditional workers.
Self-employed workers in multi-person firms do best of all—surpassing traditional workers—on measures such as income. Although self-employed workers in multi-person firms are less likely than traditional workers to participate in workplace defined contribution plans, when they do participate, they have higher balances than other workers. Although Pew did not have enough observations to look at employment by industry or occupation, it is possible that lawyers, doctors, engineers, architects, or other “knowledge professions” predominate in this group.
Self-employed workers are somewhat more likely than traditional workers to be part of a couple, which brings some benefits that may help shield against low incomes or income volatility. For example, the spouses and partners of self-employed workers are more likely to work full time than the spouses and partners of traditional workers. Spousal retirement savings are a mixed story. When workers who are self-employed in single-person firms do not participate in a workplace retirement savings plan, their spouses may participate in workplace pensions at a slightly higher rate than other spouses. The spouses or partners of self-employed workers in single-person firms also have higher median balances than the spouses of traditional workers (although not when balances are calculated as means), and their savings balances are higher as a ratio to household income than the balances of the spouses of traditional workers. The spouses of workers who are self-employed in multi-person firms have much lower rates of participation in workplace retirement savings plans than other spouses. But when these spouses do participate in a retirement savings plan, their balances are higher than other spouses’ balances.
To capture those who were self-employed for more than just a brief spell, Pew limited the study to respondents ages 50-64 in 2014 who were self-employed as their primary job in both 2012 and 2014. The analysis pool excludes retirees.
Then, we divided the self-employed into those who worked in single-person firms (the “solo” self-employed) versus in multi-person firms in 2014. Multi-person firm arrangements, including legal, medical and other practices, may be sufficiently large to suggest long-term stability, a hypothesis we test by examining income, assets, and participation in workplace retirement plans. The third category of workers consists of those in traditional employer-employee relationships—the great majority of workers—as well as those who were self-employed in either 2012 or 2014 but not in both years.
About 16.9 percent of the workforce described themselves as being self-employed in their primary jobs.33 This group was fairly evenly split between those working alone and those working in multi-person firms.
Although a simple breakout of single- versus multi-person self-employment was necessary to preserve sample sizes for most of the analyses in this report, there is likely to be considerable variability within multi-person firms. Therefore, Pew also examined different firm size breakouts within the group of firms that had two or more employees. Pew examined firm sizes of one, two, and three or more employees. Pew also examined firm sizes of one employee, two to five employees, and six or more employees. These finer breaks reduce the number of observations in each category, however: About 2.5 percent of workers were self-employed in firms with two employees, and about 4.6 percent of workers were self-employed in firms of between two and five employees. When these observations are crossed with other variables (e.g., education, marital status, pension participation), cell counts frequently drop below statistically reliable thresholds. Where warranted by adequate sample sizes and statistical significance, however, these additional results are reported in endnotes.
- Pew Charitable Trusts, “How Well Are Independent Workers Prepared for Retirement?” (2019), https://www.pewtrusts.org/en/researchand-analysis/issue-briefs/2019/06/how-well-are-independent-workers-prepared-for-retirement.
- The University of Michigan’s Health and Retirement Study (HRS) is a longitudinal study that follows older Americans as they retire and age. It is conducted every two years since 1992, with 2016 the most recent study. This report uses the 2012 and 2014 waves. Population is found by applying HRS respondent weights to the observations remaining in each employment category after selecting for age and employment status. The HRS is a rich mine of information on older Americans’ finances, work histories, demographic characteristics, and health. The survey has no data, however, on participation in workplace retirement plans through second jobs, which are often an important part of self-employed workers’ overall financial picture. There are also insufficient data to analyze the occupation, industry, or household wealth of workers by our self-employment categories.
- A defined contribution plan is a type of retirement plan where a worker contributes a percentage or a share of her paycheck to an account that is held in her name. Employers generally contribute matching percentages as an incentive to participate. The accounts are intended to be used in retirement, and restrictions apply to pre-retirement withdrawals.
- Defined benefit plans provide a fixed benefit for employees at retirement. The benefit is often determined by a formula that incorporates salary and number of years of service.
- A simple breakout of single- versus multi-person self-employment was necessary to preserve sample sizes for most of the analyses that follow. There is likely to be considerable variability within multi-person firms, however. Therefore, Pew also examined different firm sizes within the group of firms that had two or more employees.
- A chi square test of significance indicated significant differences in participation rates across generations at p < 0.001. More men work in self-employment in two-person firms (3.7 percent) than women (2.0 percent) (p<0.0001).
- A chi square test of significance, at p=0.0955, could not rule out the possibility that variations across race and ethnicity were not different from zero.
- A chi square test of significance, at p=0.0904, could not rule out the possibility that variations by educational attainment were not different from zero.
- A chi square test of significance, at p < 0.1158, could not rule out the possibility that variations by marital status were not different from zero.
- A chi square test of significance, at p < 0.0233, indicated significant differences in women’s marital status across employment categories.
- 11 A chi square test of significance indicated significant differences in rates of part-time status across employment categories at p<0.0001.
- J.F. Quinn and K.E. Cahill, “The New World of Retirement Income Security in America” (working paper, Boston College, 2015), https://www.bc.edu/bc-web/schools/mcas/departments/economics/people/faculty-directory/joseph-quinn.html.
- Dollar figures for weekly wages do not equal the product of average hours per week and average weekly wages because of the interaction between different respondents’ different hours worked and hourly wages.
- For the self-employed in two-person firms, the median hourly wage rate was $19.23 and the median weekly wage rate was $769.23.
- A chi square test of significance indicated significant differences in spousal employment status across the respondents’ employment categories at p=0.0051.
- Among self-employed respondents who worked in two-person firms, about 56.4 percent of their spouses worked full time, 18.5 percent worked part time, and 12.0 percent reported being partially retired.
- A chi square test of significance indicated significant differences in spousal self-employment rates across the respondents’ employment categories, at p<0.0001.
- A chi square test of significance indicated significant differences in poverty rates across employment categories at p=0.0009.
- This result was apparently characteristic of respondents who were self-employed in firms of three or more workers; respondents who were self-employed in two-person firms had median household incomes of $80,036, the lowest of any firm size group. Poverty rates among households with a self-employed respondent in a two-person firm were about 8.0 percent (p=0.0015), similar to those with a selfemployed respondent in a one-person firm.
- A chi square test of significance indicated significant differences in Social Security claiming rates across employment categories at: p<0.0001. The sample for this figure, as in the rest of this report, includes those under age 65 who are still working for pay but who may or may not have already claimed Social Security. This group includes full-time workers, part-time workers, and workers who describe themselves as being partially retired.
- The full retirement age is the age at which a person may first become entitled to full or unreduced retirement benefits. For workers born before 1937, the full retirement age is 65. The retirement age increases gradually for workers born in later years and is 67 for workers born in 1960 and later. Workers can take benefits as early as age 62, but benefits taken before the full retirement age are reduced by 5 percent to 6 percent a year to reflect the greater number of years over which benefits will be paid.
- For workers under the full retirement age, the earnings limit reduces benefits by $1 for every $2 earned up to $17,640 in 2019. See Social Security Administration, “Exempt Amounts Under the Earnings Test,” https://www.ssa.gov/OACT/COLA/rtea.html. Workers with earnings over the Social Security earnings limit have part of their benefit withheld. They receive the withheld money back when they fully retire, however, in the form of higher benefit payments.
- Caroline Bruckner and Thomas L. Hungerford, “Failure to Contribute: An Estimate of the Consequences of Non- and Underpayment of Self-Employment Taxes by Independent Contractors and On-Demand Workers on Social Security” (working paper, Center for Retirement Research at Boston College, Chestnut Hill, Massachusetts, 2019), https://crr.bc.edu/working-papers/failure-to-contribute-an-estimateof-the-consequences-of-non-and-underpayment-of-self-employment-taxes-by-independent-contractors-and-on-demand-workers-onsocial-security/.
- A chi square test of significance indicated significant differences in rates of participation in workplace pension plans across employment categories at p<0.0001. About 17.8 percent of the self-employed in two-person firms participated in a workplace retirement plan (p<0.0001).
- A Savings Incentive Match Plan for Employees (SIMPLE plan) may be set up by small employers (less than 100 employees) and selfemployed workers. SIMPLE plans usually take the form of a tax-deferred IRA, with simpler and less costly administration rules than employer-provided 401(k)s. A Simplified Employee Pension (SEP) IRA may be adopted by business owners to provide retirement savings coverage for themselves and their employees. SIMPLE IRAs allow employees to make contributions, unlike SEP IRAs where the employer is required to make the contributions.
- A chi square test of significance indicated significant differences in spousal rates of participation in workplace pension plans across the respondents’ employment categories at p=0.0869.
- When self-employed workers in two-person firms do not participate in workplace pension plan, about 34.6 percent of their spouses do participate in a workplace plan. A chi square test cannot rule out that the results are not statistically different from zero (p=0.0935), however.
- The number of observations was 1,101.
- Self-employed workers in two-person firms had median balances of $170,000, higher than those of traditional workers and the selfemployed in single-person firms.
- Self-employed workers in single-person firms reported being at their current jobs for an average (mean) of 14.2 years, compared to 17.8 years for the self-employed in multi-person firms and 13.3 years for traditional workers.
- The number of observations was 674.
- The spouses of self-employed workers in two-person firms had retirement plan balances of $200,000, similar to other self-employed workers in multi-person firms.
- The HRS’ estimated share of self-employed workers (16.9 percent) is higher than the U.S. Bureau of Labor Statistics’ (BLS’) estimates for either independent or alternative workers in the same 50-64 age group. One reason may be BLS’ restrictions on defining independent and self-employed workers.
America’s Overdose Crisis
Sign up for our five-email course explaining the overdose crisis in America, the state of treatment access, and ways to improve care