Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

This paper examines the effect of private long-term care insurance (LTCI) on utilization of long-term care services among disabled older adults, using pooled multiple-year (1998-2004) data from the Health and Retirement Study. long-term care (LTC) refers to health and personal care provided to assist with everyday activities such as eating, bathing, dressing, etc. Financing of LTC services in the U.S. is primarily through Medicaid and out-of-pocket payments. Unlike other medical services, LTC is not adequately insured in the private market. As the demand for LTC services is expected to increase rapidly with increased life expectancy and aging of the Baby Boomer cohort, policy makers have shown continued interest in promoting LTCI as a potential alternative to public funding. In order to assess the effectiveness of LTCI and determine future directions of public policy regarding LTC financing, it is important to have a thorough understanding of how LTCI changes the patterns of utilization of LTC services. Economic theories suggest that LTCI may affect not only the total amount of care used, but also the composition of care. This paper focuses on three major types of care: nursing home care, paid home health care, and informal care. The effect of insurance depends crucially on the relationships among them. Since most LTCI policies cover both nursing home care and home health care, the effect of insurance is a mix of 'own-price' effect and 'cross-price' effect, which is determined by the substitutability of nursing home care and home health care. The effect of LTCI on informal care utilization, on the other hand, depends on whether informal care substitutes or complements formal (paid) care. Much work in the literature has established the endogeneity of health insurance in models for medical care use. LTCI may be subject to the same argument. To address this issue, I formally test for endogeneity of LTCI purchase in each model, using the price (premium) of LTCI policy and whether an individual has life insurance as instruments. Endogeneity is detected in the model for paid home health care, which calls for IV estimation. Another econometric issue concerns the distribution of the dependent variable in the models that are estimated. Because a large proportion of people do not use any LTC services at all, there is a significant occurrence of the value 0. In addition, the distribution of non-zero values is highly skewed towards the right tail due to a small number of heavy users. To account for this, each model is estimated using a two-part formulation, where the first part is a probit model for the probability of using any care, and the second part is a log-linear model for the amount of care. The study sample is restricted to older adults ages 65 and older who have limitations in at least two activities of daily living. A rich set of control variables are used in the regression analyses, including demographic characteristics, financial resources, health status and functionality, and health behaviors and preventive care. Preliminary results indicate that LTCI significantly changes the patterns of care utilization. It reduces the use of nursing home care, increases the use of paid home health care, but has no significant effect on the use of informal care. Based on these findings, LTCI allows older adults to avoid or at least delay nursing home entries, and not surprisingly, enables them to remain in the community and consume more home health care, which is a less costly alternative to institutional care. In the mean time, the level of informal care is maintained, rather than reduced. Failure to observe a significant decrease in informal care utilization suggests that informal care is different in nature and cannot be largely replaced by formal care even though the latter is now paid by insurance. By directing resources to home health care without discouraging family care-giving, LTCI can help reduce unnecessary nursing home stays, and more efficiently finance LTC services.