Can family caregiving substitute for nursing home care?

Abstract

Informal care should be a substitute for nursing homes but empirical evidence often suggests the opposite. This may be because informal care receipt is positively correlated with unobserved negative health characteristics. We exploit variation in children's characteristics as instruments for informal care to provide Two-Stage Least Squares (TSLS) estimates of nursing home use among a sample of 6855 individuals from the 1993–2000 waves of the AHEAD survey. While OLS results suggest informal care is associated with greater future nursing home risk, TSLS estimates show that receipt of informal care statistically and substantially reduces the risk of nursing home entry. This finding has implications for Medicaid and private long-term care insurance markets.

Introduction

Nursing homes are a costly and generally undesirable source of long-term care for the disabled elderly. As a result, they are often considered an option of last resort by individuals, families, and the state. Despite various attempts by researchers to understand the determinants of nursing home use, many aspects of the transition are not well understood. For example, while health status is known to be a very strong determinant of nursing home entry, many seriously disabled individuals reside in non-institutionalized settings, while many less seriously disabled persons are institutionalized. This paper presents new evidence on the relationship between informal home care receipt and nursing home entry risk by following respondents in the Study of Asset and Health Dynamics among the Oldest Old (AHEAD) from 1993 to 2000.

Informal care provided in the home by family or friends should be a substitute for nursing homes but the available evidence in support of this proposition is quite weak; indeed there is some evidence that the receipt of informal care and nursing home entry are positively related. One obvious explanation for this empirical finding is that receipt of home care is positively correlated with unobserved negative health characteristics that increase the probability of nursing home entry. The use of lagged measures of home care may reduce but not eliminate this endogeneity problem, as health and disability are likely highly correlated over time.

To address this problem, we estimate instrumental variables models of the effect of informal care-giving on nursing home risk. Given a large literature documenting the higher propensity of daughters to provide informal care relative to sons (McGarry, 1999, Wolf et al., 1997), we use a variety of family structure characteristics to instrument for receipt of unpaid help with ADLs in 1993. We argue that conditional on marital status and the number of children, the gender composition and other characteristics of those offspring are, for all practical purposes, exogenous, and are thus valid instruments for receipt of informal care. We find that OLS estimates show a positive association between nursing home entry and the receipt of informal care. However, when we account for the endogeneity of the receipt of informal care using instrumental variables, we find that nursing home entry and informal care are negatively related – that is, they are substitutes.

Empirical evidence that informal care can reduce nursing home use has important policy implications. Evaluations of demonstration projects like the Channeling Demonstration in the past few decades concluded that investments in home and community based long-term care could not reduce total long-term care expenses by reducing nursing home care. Nevertheless, state governments are continually looking for ways to reduce Medicaid expenditures. Given that most disabled elderly people would prefer to receive care in the community rather than be institutionalized, our finding suggests that an effective and possibly efficient way to reduce nursing home expenses may be to subsidize informal caregivers. This is particularly important as the number of disabled elderly continues to increase, and Medicaid places a larger strain on state budgets.

The remainder of the paper is organized as follows. In the next section we summarize some of the previous literature on this question. Section 3 provides a description of the AHEAD data we use in our analysis. Section 4 presents OLS estimates of the associational relationship between informal care and nursing home use. In Section 5 we use instrumental variables strategies to account for the endogeneity of informal care. Two-Stage Least Squares (TSLS) and bivariate probit results are presented in this section. Section 6 concludes.