The impact of family characteristics on timing of retirement. A longitudinal analysis of the HRS in the United States between 1992 and 2006

Kim Boudiny, University of Antwerp
Karel Neels, University of Antwerp

This paper examines the influence of family obligations and relationships on the timing of retirement. Special attention is paid to the presence of and care for (grand)children and parents. The study makes use of the Health and Retirement Study (HRS), a multidisciplinary panel research representing all persons over 50 in the United States. Data are drawn from eight biennial waves of the HRS between 1992 and 2006. For the purpose of the current study, HRS is preferred above the European SHARE and ELSA. SHARE and ELSA build on HRS, but have fewer waves of data available. ELSA was first executed in 2002 and conducted biannually since then. SHARE took off in 2004. Like HRS, ECHP is characterized by a longer observed time period. ECHP-data are, however, for most countries only available till 2001. In addition, questions concerning family structure are less elaborated as in HRS. A subsample of 6,603 respondents born between 1931 and 1941 is analysed (n = 6603), the so-called HRS-cohort which has been included in the survey since the first data collection period in 1992. The analysis uses discrete-time event-history models for late entrants. This analysis strategy offers added value to a large amount of labour market studies performed on the HRS that make use of logistic regression analysis as the latter analyses cannot adequately address research questions concerning the timing of transition. Models are fitted with age as the exposure. Analyses are stratified by gender. Due to the limited public provision of care for either elderly, ill adults or children and the high costs related to private provisions in the United States, we expect that the presence of care demands contributes to an early withdrawal from the labour market.