Young Kyung Do1, Chetna Malhotra. 1. Program in Health Services and Systems Research, Duke-NUS Graduate Medical School Singapore, Singapore. young.do@duke-nus.edu.sg
Abstract
OBJECTIVES: To estimate the causal effect of coresidence with an adult child on depressive symptoms among older widowed women in South Korea. METHOD: Data from the first (2006) and second (2008) waves of the Korea Longitudinal Study of Ageing were used. The analysis was restricted to widowed women aged ≥65 years with at least one living child (N = 2,446). We used an instrumental variables (IVs) estimation exploiting two child characteristics as IVs (the number of sons and whether the eldest child is a daughter). Specification tests for IVs showed that these instruments predict the probability of an older widowed woman's coresidence with an adult child but do not directly affect depressive symptoms. RESULTS: Our IV two-stage least-squares estimator suggested that coresidence with an adult child has a protective effect on depressive symptoms among older widowed women in South Korea. Coresidence was found to be endogenous in our statistical model of depressive symptoms. DISCUSSION: The IV estimation method can be a useful approach to addressing the potential endogeneity between intergenerational coresidence and elderly health. Rapidly decreasing rates of intergenerational coresidence may raise public health concerns among older widowed women in South Korea.
OBJECTIVES: To estimate the causal effect of coresidence with an adult child on depressive symptoms among older widowed women in South Korea. METHOD: Data from the first (2006) and second (2008) waves of the Korea Longitudinal Study of Ageing were used. The analysis was restricted to widowed women aged ≥65 years with at least one living child (N = 2,446). We used an instrumental variables (IVs) estimation exploiting two child characteristics as IVs (the number of sons and whether the eldest child is a daughter). Specification tests for IVs showed that these instruments predict the probability of an older widowed woman's coresidence with an adult child but do not directly affect depressive symptoms. RESULTS: Our IV two-stage least-squares estimator suggested that coresidence with an adult child has a protective effect on depressive symptoms among older widowed women in South Korea. Coresidence was found to be endogenous in our statistical model of depressive symptoms. DISCUSSION: The IV estimation method can be a useful approach to addressing the potential endogeneity between intergenerational coresidence and elderly health. Rapidly decreasing rates of intergenerational coresidence may raise public health concerns among older widowed women in South Korea.