This paper examines the sources of the decline in self-employment among near-retirees over 1994–2012. Using Current Population Survey data, tabulations imply that the decline was driven by an increase in the exit rate to wage and salary employment, a decline in the rate of self-employment among new entrants into this age cohort, and an increase in the share of these new entrants. Multinomial logits suggest that health insurance coverage and after-tax prices of health insurance were significantly associated with these three rates. However, counterfactual simulations suggest that only the changes in after-tax prices of health insurance were found to appreciably influence the trends in these rates, though in the opposite direction of the actual declining trend for the rate of self-employment of new entrants.
This is a preview of subscription content, log in to check access.
Buy single article
Instant access to the full article PDF.
Price includes VAT for USA
Subscribe to journal
Immediate online access to all issues from 2019. Subscription will auto renew annually.
This is the net price. Taxes to be calculated in checkout.
In the literature, several different age cuts have been used when examining behavior of individuals nearing retirement. Generally, the upper bound is age 64, since age 65 is the age of eligibility for Medicare and was for many years the full retirement age for receiving Social Security. A variety of ages have been used for the lower bound, from 51 in studies like Zissimopoulos and Karoly (2007) to 55 for tabulations in Hipple (2010). Since the data used in this study is most similar to those used in Hipple (2010), this study uses the higher age cut of 55.
Since these observations are in the sample twice, they comprise 208,636 observations.
Tabulations for individuals in the second year of the matched sample are similar.
Note that this is the fraction of the total population that is self-employed, while Fig. 1 presents tabulations of the fraction of non-agricultural workers that are self-employed, and so the latter figure is higher than the former.
Weights were adjusted using a process described in Nichols (2007) using code graciously provided by Stuart Craig at Yale University. In this procedure, a logit is estimated to generate probabilities of being in the matched sample, and weights are then adjusted to match the age, race, sex, marital status, and employment mode distribution of the full sample.
These trends are presented in Figure A1 of an Appendix that is available upon request from the author.
These rates were calculated using the reweighted matched CPS sample. Details of the reweighting procedure are presented in the Appendix.
In tabulations by industry, in order to ensure sufficient sample sizes in each group, respondents were separated according to whether they reported their industry being:
manufacturing (including mining, construction, manufacturing, transportation, and utilities),
wholesale and retail trade, or
service (including information, financial activities, professional and business, educational and health services, leisure and hospitality, and other services).
For transition rates, observations are classified according to the industry of the individual in the first year of the 2 years pair.
In tabulations by region of the country, in order to ensure sufficient sample sizes in each group, respondents were separated into four regions:
Northeast (including Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, and Vermont),
Midwest (Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota and Wisconsin),
South (Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Virginia, Texas and West Virginia), and
West (Arizona, Alaska, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington and Wyoming).
In an appendix that is available upon request from the author, these trends are presented in Figure A2. In this figure, continuation rates do not exhibit a clear declining trend, and instead appear to be relatively constant. Continuation rates also did not differ noticeably across incorporation status, industry, or region of the country.
Similar trends were also found within incorporation statuses, industries, and regions.
These trends are presented in Figure A3 in an appendix that is available upon request from the author. Similar trends were found within incorporation statuses and industries. Looking by region, exits to not working increased in the Northeast, but these increases were offset by declines in other regions.
Similar results are seen in Figure A3 in an Appendix available from the author, in which the exit rate to wage and salary work among the self-employed in 1995 was below 14 %, while toward the end of the sample, this rate hovered around 18 %.
These results are presented in Figure A4 in an Appendix available from the author.
These results are presented in Figure A5 in an Appendix available from the author.
These results are presented in Figure A6 in an Appendix available from the author.
These results are also presented in Figure A7 in an appendix available from the author. This figure demonstrates that the self-employment rate among 55-year-olds declined during the sample period, from about 17 % at the beginning of the sample to around 13 % toward the end of the sample.
These results are presented in Figure A8 in an appendix available from the author. Tabulations by region of the country showed that the self-employment rate among 55-year-olds declined across all regions of the country.
To examine exits from self-employment, the sample will be cut to include only self-employed workers. To examine entries into self-employment, the sample will be cut to include wage and salary workers or individuals who are not working.
Unfortunately, the CPS does not contain detailed wealth information, and so using such a proxy is necessary. One should note that investment income is a relatively poor proxy for wealth, as it includes wealth for which a return was not realized in the given year, as well as housing wealth.
From the U.S. Bureau of Labor Statistics, Local Area Unemployment Statistics. (www.bls.gov/lau).
From the U.S. Bureau of Labor Statistics, Quarterly Census of Employment and Wages. (www.bls.gov/cew).
The after-tax price of a dollar of health insurance premiums paid if self-employed is calculated as (1 − τ p − τ f − τ s + (τ s·τ f)*item)/(1 + τ p), where τ p is the payroll tax rate, τ f and τ s are federal and state marginal tax rates, and item is an indicator for the individual itemizing deductions. The after-tax price of a dollar of employer-sponsored health insurance premiums is calculated as (1 − τ p − τ f − τ s + (τ s·τ f)*item)/(1 + τ p). Since the after-tax price for wage and salary workers is proportional to the share of income that an individual would keep after taxes (i.e. one minus their combined marginal federal, state, and payroll tax rates), it is not possible to enter tax rates directly into the estimation equation, as the two variables are almost perfectly collinear. Instead, the after-tax price for wage and salary workers will capture the combined effects of changes in the cost of health insurance while a wage and salary worker and changes in tax rates per se.
These include whether a state’s individual insurance market had community rating regulations (which limit the extent to which insurance companies may charge different premiums based on health status) or guaranteed issue regulations (which prevent insurance companies from excluding anyone because of preexisting conditions), or both. These data come from Heim and Lurie (2014a).
These variables include the state-level average Disability Insurance benefit among disabled workers, and the state-level Disability Insurance disabled worker receipt rate among adults age 18–64. See OASDI Beneficiaries by State and County (Various Years), Social Security Administration, Office of Policy and Office of Research, Evaluation and Statistics. Tables 1, 2, 3. http://www.ssa.gov/policy/docs/statcomps/oasdi_sc/index.html.
Because these variables are partly a function of whether an individual is self-employed, a wage and salary worker, or not working, the approach is to use the value of these variables in the first year of the two-year pair (that is, the year before a transition, if any, is made).
Similar findings are reported in Heim and Lurie (2010).
See footnote 24.
Unfortunately, it is not possible in the CPS to determine whether employer-sponsored health insurance is from a current or former employer or union.
This is in contrast to Bruce et al. (2000), who find that health insurance variables have insignificant effects on exits from self-employment. One possibility for the difference could be the earlier time period (1992–1996) examined in their study.
These results are consistent with Heim and Lurie (2010).
See footnote 24.
Similar results are reported in Zissimopoulos and Karoly (2007).
This could take the form of retiree health insurance through a former employer.
Because the health insurance and pension variables come from the previous year, in order to be in the estimation sample, an individual must be in the matched sample and be 55 in the second year of the matched pair (so that health insurance and pension information from the prior year is available). However, the results in this specification are similar to those when the full sample of 55-year-olds and health insurance variables from the contemporaneous year are used.
Similar simulations were performed in Heim and Lurie (2013) to simulate counterfactual changes in Earned Income Tax Credit caseloads.
Counterfactual simulations for the continuation rate, entry rates, and exit rate to not working are presented in an appendix available from the author in Figures A9–A11. In Figure A9, the simulations imply that changes in the after-tax prices of health insurance and health insurance coverage would have served to increase the continuation rate, but this impact was dampened by changes in the generosity and coverage of Disability Insurance (though the estimated coefficients on these variables were not statistically significant). In Figure A10, which presents counterfactual simulations for entry rates, the simulations imply that changes in Disability Insurance generosity and coverage, wages and unemployment, health insurance coverage, and pension coverage are simulated to have decreased entry from wage and salary employment (though only the health insurance coverage variables were statistically significant), but these were offset by changes in the after-tax price of health insurance. Similarly, changes in Disability Insurance and wages and unemployment variables, though estimated to be statistically insignificant, would have led to declining rates of entry from not working, but changes in the after-tax prices of health insurance offset this trend. Finally, in Figure A11, changes in the after-tax prices of health insurance are simulated to have decreased exits to not working, as are changes in health insurance coverage, while this simulated impact was offset by changes in Disability Insurance generosity and coverage, wages and unemployment (though these variables carried statistically insignificant coefficients).
For example, by allowing self-employed health insurance premiums to be deductible when figuring Self Employment Contributions Act (SECA) taxes.
Bruce, D., Holtz-Eakin, D., & Quinn, J. (2000). Self-employment and labor market transitions at older ages. Center for Retirement Research at Boston College Working Paper 2000-13.
Dunn, T. A., & Holtz-Eakin, D. J. (2000). Financial capital, human capital, and the transition to self-employment: Evidence from intergenerational links. Journal of Labor Economics, 18(2), 282–305. doi:10.1086/209959.
Evans, D. S., & Jovanovic, B. (1989). An estimated model of entrepreneurial choice under liquidity constraints. The Journal of Political Economy, 97(4), 808–827.
Evans, D. S., & Leighton, L. S. (1989). Some empirical aspects of entrepreneurship. American Economic Review, 79(3), 519–535.
Fairlie, R. W. (2004). Self-employed business ownership rates in the United States: 1979–2003. Report for Office of Advocacy, Small Business Administration.
Fairlie, R. W., Kapur, K., & Gates, S. (2011). Is employer-based health insurance a barrier to entrepreneurship? Journal of Health Economics, 30(1), 146–162. doi:10.1016/j.jhealeco.2010.09.003.
Fuchs, V. R. (1982). Self-employment and labor force participation of older males. The Journal of Human Resources, 17(3), 339–357. doi:10.2307/145584.
Giandrea, M.D., Cahill, K.E., & Quinn, J.F. (2008). “Self-Employment Transitions among Older American Workers with Career Jobs.” BLS Working Paper 418.
Heim, B. T., & Lurie, I. Z. (2010). The effect of self-employed health insurance subsidies on self-employment. Journal of Public Economics, 94(11–12), 995–1007. doi:10.1016/j.jpubeco.2010.08.007.
Heim, B.T., & Lurie, I.Z. (2013). The dynamics of earned income tax credit receipt among single mothers. Working Paper.
Heim, B. T., & Lurie, I. Z. (2014a). Did reform of the non-group health insurance market affect the decision to be self-employed? Evidence from state reforms in the 1990s. Health Economics, 23(7), 841–860. doi:10.1002/hec.2960.
Heim, B. T., & Lurie, I. Z. (2014b). Does health reform affect self-employment? Evidence from massachusetts. Small Business Economics, 43(4), 917–930. doi:10.1007/s11187-014-9572-6.
Hipple, S. F. (2010). Self-employment in the United States. Monthly Labor Review, 133(9), 17–32.
Holtz-Eakin, D., Joulfaian, D., & Rosen, H. S. (1994a). Sticking it out: Entrepreneurial survival and liquidity constraints. The Journal of Political Economy, 102(1), 53–75.
Holtz-Eakin, D., Joulfaian, D., & Rosen, H. S. (1994b). Entrepreneurial decisions and liquidity constraints. The Rand Journal of Economics, 25(2), 334–347.
Holtz-Eakin, D., Penrod, J. R., & Rosen, H. S. (1996). Health insurance and the supply of entrepreneurs. Journal of Public Economics, 62(1–2), 209–235. doi:10.1016/0047-2727(96)01579-4.
Kautonen, T. (2008). Understanding the older entrepreneurs: Comparing third age and prime age entrepreneurs in finland. International Journal of Business Science and Applied Management, 3(3), 3–13.
Kautonen, T., Luoto, S., & Torniko, E. T. (2010). Influence of work history on entrepreneurial intentions in ‘prime age’ and ‘third age’: A preliminary study. International Small Business Journal, 28(6), 583–601. doi:10.1177/0266242610368592.
Klerman, J. A., & Haider, S. J. (2004). A stock-flow analysis of the welfare caseload. Journal of Human Resources, 39(4), 865–886. doi:10.2307/3559030.
Klerman, J. A., & Haider, S. J. (2005). Dynamic properties of the welfare caseload. Labour Economics, 12(5), 629–648. doi:10.1016/j.labeco.2004.01.006.
Madrian, B.C., & Lefgren L.J. (1998). The effect of health insurance on transitions to self-employment. Unpublished mimeo. University of Chicago Population Research Center.
Meyer, B. (1990). Why are there so few black entrepreneurs? NBER Working Paper 3537.
Nichols, A. (2007). Causal inference with observational data. Stata Journal, 7(4), 507–541.
Shane, S. A. (2008). Illusions of entrepreneurship: The costly myths that entrepreneurs, investors, and policy makers live by. New Haven: Yale University Press.
Stodola, S. (2012). “The American Entrepreneur: A Dying Breed?” The Fiscal Times, May 3, 2012. http://www.thefiscaltimes.com/Articles/2012/05/03/The-American-Entrepreneur-A-Dying-Breed
Svaleryd, H. (2015). Self-employment and the local business cycle. Small Business Economics, 44(1), 55–70. doi:10.1007/s11187-014-9592-2.
Tervo, H. (2008). Self-employment transitions and alternation in Finnish rural and urban labour markets. Papers in Regional Science, 87(1), 55–76. doi:10.1111/j.1435-5957.2007.00150.x.
Tervo, H. (2011). Self-employment transitions at older ages in urban and rural labor markets. Working Paper.
Velamuri, M. (2012). Taxes, health insurance and women’s self-employment. Contemporary Economic Policy, 30(2), 162–177. doi:10.1111/j.1465-7287.2011.00256.x.
Wang, Q., & Liu, C. Y. (2015). Transnational activities of immigrant-owned firms and their performances in the USA. Small Business Economics, 44(2), 345–359. doi:10.1007/s11187-014-9595-z.
Wellington, A. J. (2001). Health insurance coverage and entrepreneurship. Contemporary Economic Policy, 19(4), 465–478. doi:10.1093/cep/19.4.465.
Zissimopoulos, J., & Karoly, L. (2007). Transitions to self-employment at older ages: The role of wealth, health, health insurance, and other factors. Labour Economics, 14(2), 269–295. doi:10.1016/j.labeco.2005.08.002.
Zissimopoulos, J., Maestas, N., & Karoly, L. (2007). The effect of retirement incentives on retirement behavior: Evidence from the self-employed in the United States and England. Michigan Retirement Research Working Paper 2007-155.
I wish to thank Ed Gerrish and LeeKai Lin for excellent research assistance, and Jules Lichtenstein, three anonymous reviewers, Adam Copeland, and participants at the WEAI 89th Annual Conference for helpful advice and comments.
An earlier draft of this study entitled “Understanding Self-Employment Dynamics Among Individuals Nearing Retirement” was developed under contract SBAHQ-13-M-0055 for the Small Business Administration, Office of Advocacy. The final conclusions of the report do not necessarily reflect the views of the Office of Advocacy.
About this article
Cite this article
Heim, B.T. Understanding the decline in self-employment among individuals nearing retirement. Small Bus Econ 45, 561–580 (2015). https://doi.org/10.1007/s11187-015-9660-2
- Health insurance
- Markov chain