Does Self-Employment Influence the Personal and Financial Health of Older Americans? Evidence from Six Waves of the Health and Retirement Study Using Propensity Score Analysis

Background and Purpose: Self-employment is a prominent form of work in later life. Among Americans in the labor force, more than one in five aged 50 and older and nearly one in three in their late 70s work for themselves. While studies have documented the correlates of self-employment in later life, there remains a need to understand how this form of work compares to more traditional wage-and-salary work while controlling for selection into self-employment. In response, this study asks, “How does self-employment influence the personal health and financial well-being of older Americans, in comparison to wage-and-salary work?” I hypothesize that self-employment leads to reduced earnings, yet given the underdeveloped nature of this scholarship, no hypotheses are provided for the remaining outcome variables.

Methods: This study includes data on Americans aged 50 and older from six waves of the nationally-representative and biennial Health and Retirement Study, from 2004 to 2014. The final sample includes nearly 6,500 individuals and more than 25,000 observations. Missing data were imputed using multiple imputation by chained equations, with 20 final datasets created and a single set of outcome models developed using Rubin’s combination rules. Two styles of inverse probability of treatment weighting, a form of propensity score analysis, were employed to estimate “treatment” effects of self-employment. Cluster-robust standard errors were used to account for clustering within households over time, and sampling weights were included to increase the sample’s representativeness. Among the outcome variables, personal health was operationalized as self-rated health, a four-point scale; and depressive symptoms, ranging from 0-8. Financial well-being was operationalized as individual earnings from work and total household wealth. The “treatment” variable was self-employment vs. wage-and-salary work, and covariates included measures of sociodemographic characteristics; human, social, and financial capital; and workplace characteristics. Outcome analyses utilized the appropriate form of regression. Ten models were developed for each outcome variable and compared for sensitivity analysis, a recommended step with propensity score analysis.

Results: Among older Americans, there was unanimous agreement among the models (all negative coefficients with significant relationships) that self-employment led to reduced individual earnings; that is, older self-employed Americans would have earned more, on average, if they had worked for another company, all else being equal. There was moderate agreement among the models (general agreement with the direction of relationships, not all significant) that self-employment led to increases in household wealth and self-rated health. There was little evidence of a relationship between self-employment and the number of depressive symptoms reported.

Conclusions and Implications: Self-employment—often used interchangeably with entrepreneurship—has been heralded as a win-win for individuals and society at large, increasing older adults’ financial security and workplace flexibility while driving economic growth. To a great extent, this may be true. However, this study paints a complex picture about the causal effects of later-life self-employment and challenges key arguments on its benefits, with the notable finding that self-employment decreases earnings. Program developers, policymakers, and the media must provide a more balanced view on the costs and benefits of self-employment in later life.