Pathways to retirement and the self-employed

Self-employment is popular among older workers. For older workers who have spent their careers as wage earners, self-employment can play the role of a “bridge job” as they transition to retirement, providing flexible working hours and conditions. A number of papers have examined the factors – such as health insurance and wealth – that influence transitions to self-employment at older ages. One question that has received less attention is what happens to the incomes of older workers when they transition into and out of self-employment. This is an important question given policy makers’ interest in strengthening the financial security of retirees. The role that self-employment plays in extending working lives depends critically on the extent to which self-employment earnings replace earnings from wage employment. We investigate this question using a rich administrative dataset consisting of the tax returns of a large random sample of individuals from the 1945 and 1965 birth cohorts over 1999-2013. Our data allow us to observe receipt of wage income (reported on form W-2) as well as self-employment income (reported on schedules C and SE) in each year. This allows us to observe transitions into self-employment and retirement for the 1945 cohort, which turned 54 in 1999 and 68 in 2013. The younger cohort (aged 34 to 48 over the sample period) is used as a control group. We begin by examining how the probability of a self-employment transition varies with age, observable demographics, and local labor market conditions. We then examine how income changes upon transition to self-employment. Preliminary results indicate that a transition from wage employment to self-employment is associated with a substantially larger drop in income for the older group relative to the younger one, suggesting that older individuals may use self-employment as a “bridge job” that provides, flexible part-time work as they transition to retirement. We examine how this drop in income varies with observable demographics. We supplement our analysis using data from the Health and Retirement Study (HRS), a panel dataset that is representative of the older population. The HRS provides a much smaller sample compared to the administrative dataset, but it includes richer information about demographics. It also includes self-reports of hours worked that can be used to examine how hours worked – in addition to income – change upon transition to self-employment.