Health and Retirement Among Older Working Couples in Australia
Posted: 16 Jun 2007
Date Written: June 2007
Abstract
This paper investigates the causal relationship between ill-health, health shocks and retirement among older working Australian individuals using the first four waves (2001- 2005) of the Household, Income and Labour Dynamics in Australia (HILDA) Survey. We represent the transition to retirement as a discrete-time hazard model which enables us to estimate the effect of measures of health status and a number of socio-economic characteristics on the probability of retirement.
Traditionally, among retirement models, the focus has been on male behaviour modelled as independent decision-makers, ignoring the retirement decision of a spouse. Our paper considers retirement also as a joint decision made by couples by examining the effects of a spouse's health and labour market status on an individual's retirement decision.
In order to overcome the problems related to measurement error (reporting bias) and endogeneity of self-assessed measures of health, the model considers the construction of a latent "health-stock" variable which is used as an indicator of health and is purged of reporting bias. This is obtained by estimating a model of SAH as a function of more objective measures of health using pooled ordered probit models. The estimated coefficients show the expected negative signs, indicating that measures of health problems and limitations are associated with lower reporting of self-assessed health.
To identify health shocks we condition on the initial period health-stock when estimating the coefficient of lagged health. This allows us to interpret the estimated coefficient on lagged health status as representing a deviation from the underlying health-stock.
We use the stock sampling approach of Jenkins (1995) to define our sample of interest. The selected 1328 working individuals at risk of retirement (50 years old or above) in the first wave are followed through the first four waves of the HILDA survey until they retire or are censored. This method represents the transition to retirement as a discrete-time hazard model and changes the unit of analysis from the individual to the time at risk of an event (in this case, retirement). This econometric framework allows complex sequence likelihoods to be simplified to a standard estimation for a binary outcome.
Our estimates show that together with household income and the financial incentives provided by the social security system (state age pension), own health is an important determinant of labour supply among older working individuals. This is especially true for men. Negative shocks to health that lead to health limitations increase the hazard of retiring, although the effect is not marked for women. Among couples, for female individuals, a significant determinant of retirement is a partner's labour force status. For both male and female individuals, partner's health appears less significant in the couple's decision making process.
Keywords: health, discrete-time hazard model, retirement
JEL Classification: I12, C23
Suggested Citation: Suggested Citation