Lapse in the Long-Term Care Insurance Market: Evidence from the Health and Retirement Study

Long-term care (LTC) represents a significant and growing source of healthcare expenditures for elderly in the U.S. with approximately $320 billion in 2011 or about 14% of all total  healthcare spending. The financing of LTC therefore is becoming a significant challenge as the US population ages and the need for LTC is expanding.  In an effort to expand the market for private LTC, an equally important issue is policy lapse, defined as the termination of a policy prior to claiming benefits. Termination of a policy can be a loss to a consumer because of lost premium and the future financial risk of LTC needs. Despite of this loss, about 20-30% policies are terminated each year. Previous studies explaining lapse in the LTC insurance ignores important factors contributing to a LTIC policy lapse such as distribution of LTC utilization risk based on episode of care.  The current study investigates factors explaining decision to terminate a LTC policy.

Using a standard expected utility model of demand for health insurance, a policyholder of long-term care insurance (LTCI) decides to lapse if and only if the expected utility from terminating the LTCI policy is greater than the expected costs of retaining the policy.  The study uses data from the Health and Retirement Study (HRS), nationally representative longitudinal survey of individual aged 50 or over. Data from wave 3-12 (1996-2012) were used to observe the dynamics of LTCI policy purchase over a long-period of time. Due to the study focus on LTCI policy lapse , only observations from individuals with at least two consecutive surveys during any time between 1996-2012 who reported having private LTCI in the first year of the two-year transition period and remained in the sample in the year 2 to answer LTCI questions.  The total of 4617 unique respondents and 12015 person-years observations were used in the analysis.  The dependent variable was the lapse of LTCI, based on point-in-time responses defined as whether a respondent has purchased LTCI.  To account for unobserved individual heterogeneity, Mundlak-Chamberlin’s (1984) correlated random effect (CRE) model was estimated.

Overall 2-year lapse rate was estimated as 38%. The care transition model estimated that an average 65 year old female will spend just over 5 months in a nursing home, about 4 months in an assisted living facility and just over 8 months with home health care. An average 65 old individual pays $1490 as annual LTCI premium with maximum daily benefits of $100. Preliminary results indicate that individuals with higher expected use of nursing home care are less likely to lapse te policy (10% with p<0.001).  self-reported use of expected nursing home utilization and life expectancy over 85 years both are negatively associated with the probability of policy lapse.  In all model mean of time varying variable , total wealth remain statistically significant indicating a greater degree of individual heterogeneity and appropriateness of CRE model. Among demographic variables, being married, female, White, non Hispanic are less likely to drop LTCI policy.