Do Expectations of Nursing Home Use Translate Into Long-Term Care Insurance Purchase and Savings?

Background and Purpose: The health care system and state budgets are straining with the rising costs of long-term care (LTC) and increasing reliance on Medicaid rather than private savings or long-term care insurance (LTCI) to pay LTC costs. Models of life cycle behavior suggest that expectations about future events, including health events, may affect individuals’ savings, insurance, and retirement planning, and policymakers have speculated that only those who know they will have a LTC need will purchase LTCI (adverse selection). The Health and Retirement Study (HRS) has collected subjective expectations of nursing home use at two points in time: 1992 and 2006. Previous work explored whether individuals accurately predict lifetime nursing home use and found that women and men incorporate their subjective experience, personal characteristics, and health conditions, and that their expectations conform reasonably well to previous empirical literature about predictors of lifetime nursing home use. However, there has not been any exploration of whether these self-reported expectations about future nursing home use translate into purchases of long term care insurance (LTCI) or contributions to savings.  This paper explores whether savings and LTCI purchase patterns correlate to expectations of long-term nursing home use.

Methods: Using data from nine waves of the HRS (1992 – 2008), we ran multiple regressions with each wave of data, controlling for age, education, gender, and self-rated health, with expectation of nursing home use (NH probability) and possession of LTCI in the current wave as the predictor variables of interest with assets as the dependent variable. We also ran similar logistic models with LTCI purchase as the dependent variable, and assets as a predictor. We use the RAND imputations of wealth and income measures in all analyses, and assets include common retirement savings products and a primary residence.

Results: We find that, in general, individuals who report higher expectations of NH use do not invest more in savings accounts, nor are they are more likely to purchase LTCI.  Instead, LTCI purchase is associated with higher assets. In Wave 5 (2000), for instance, a higher expectation of NH use was associated with lower levels of asset accumulation (b=-0.22, t=-2.09, p=.04), but possession of LTCI was strongly associated with higher assets (b=0.84, t=0.17, p<.0001). In all waves, age, education, and self-rated health are also significantly and positively associated with assets, though, interestingly, the association with gender is not always significant.

Conclusions and Implications: The results suggest that factors typically found to be predictors of investments in assets and LTCI (e.g., education, income) are significant predictors in the HRS sample.  The findings confirm that expectations, recorded long before respondents experienced a need for nursing home care, do not translate into planning for LTC utilization, especially under uncertainty. Though insurers and policymakers assume that adverse selection may contribute to making LTCI products unsustainable, our findings suggest that affordability of these products (relative to individuals’ assets), rather than subjective predications of future use, may play a more important role. We conclude that policy interventions should facilitate LTCI purchase by addressing affordability.