In private long-term care insurance markets, moral hazard is central to pricing and long-run robustness of the market, yet there is remarkably little evidence on the extent to which moral hazard is present in long-term care insurance.
The authors found evidence of significant moral hazard in home care use and a potentially meaningful but noisy effect on nursing home use. Policymakers designing incentives to promote private long-term care insurance should consider the consequences of moral hazard.
In the US the financing of long-term care will become increasingly challenging as the population ages and the prevalence of chronic conditions grows. Since the passage of prescription drug coverage for seniors, long-term care is arguably the single largest uninsured health expenditure risk facing the country. In 2016 the median monthly cost was over $7,000 for nursing home care and $4,000 for home-based care, and only a small minority of individuals can finance an extended nursing home stay out of their own pockets.
Medicaid is the largest payer of long-term care by default, but with long-term care consuming more than a third of state Medicaid budgets, states find this burden increasingly challenging. Other developed countries in Europe and Asia face similar challenges, with the population in many countries ageing at a faster pace than in the US.
Long-term care insurance is not a perfect solution
Long-term care insurance (LTCI) is often posited as one solution to the financing challenge, because an increase in private responsibility could offset some of the public burden. However, only 13% of elderly individuals have LTCI in the US.
In the presence of this fairly small private market, the US government passed a national LTCI programme called Community Living Assistance Services and Supports Act (CLASS) as part of the Affordable Care Act in 2010. CLASS was intended to be a self-sustaining, premium-financed, government-administered national insurance programme for long-term care services but was repealed after policymakers found the mandated terms of the programme made it impossible to avoid adverse selection and meet the sustainability requirement. The demise of CLASS renewed the interest in private LTCI among policymakers.
A rigorous estimate of moral hazard in LTCI has been markedly absent from these policy debates and the research literature. Yet, an estimate of the extent that insurance affects subsequent utilisation is crucial to pricing and sustainability of the private LTCI market. Our paper aims to fill this gap. We examine the changes in utilisation once people acquire LTCI, providing a rigorous estimate of ex post moral hazard in LTCI markets.
The current situation in the US
The ageing of the US population points to an increasing need for health care services in the coming decades, especially among elderly individuals with long-term care needs.
Medicare pays for short-term rehabilitation in a nursing home but not for chronic care; Medicaid, a payer of last resort with strict income and asset requirements, traditionally paid only for nursing home care but increasingly covers home and community-based alternatives. Private LTCI is meant to fill this gap in systematic public funding, but the market for it remains limited.
A number of LTCI papers have investigated the reasons for the limited market size and adverse selection. Among the elderly and near-elderly, the younger, healthier, and more educated people are more likely to have LTCI. While the highest income groups may not purchase LTCI because they can self-insure, those in the lowest income and asset groups are also unlikely to purchase it because it is expensive (typically several thousand dollars per year) and because they face a lower implicit ‘price’ of Medicaid in terms of spending down assets to qualify.
While the research discussed above moves towards a better understanding of consumers’ purchase behaviour, insurance ownership is only one factor in determining the viability of the market. Equally important is how insured individuals behave once they become insured. The importance of ex-post moral hazard—defined as additional (and potentially socially inefficient) utilisation of long-term care services due to the presence of insurance—parallels the importance of adverse selection, because both affect the amount of pay-outs relative to premiums.
The premiums for LTCI are set in part by estimating the expected long-term care spending of the insured population. If the pay-outs are greater than expected, either because adverse selection has attracted a disproportionate number of sicker individuals among the enrolees, or because moral hazard has generated a greater amount of spending per enrolee (due to increased utilisation) than would be expected without insurance, then the financial viability of the insurance programme would be called into question. Indeed, long-term care insurers have noted higher-than-expected claims as a fundamental reason for pulling out of the market in recent years.
We find evidence of significant moral hazard in home care use financed by private LTCI. The magnitude of the effect in our baseline specification - a 9.9 percentage point increase in the probability of home care use - is meaningful, given the overall mean usage rate of 12.4%. The direction of the effect is consistent with theoretical predictions when home health care is viewed as a normal good.
This is in marked contrast to Li and Jensen, who found no overall significant effect on home care use using similar instruments, but without matching. This different finding supports our contention that lack of balance after applying the IV renders these prior estimates potentially biased. As a desirable option for elderly individuals needing long-term care, a sizable moral hazard effect in the use of home care is to be expected and is confirmed by our findings.
Thus, it may be the case that the response of nursing home care to LTCI may represent the underlying combination of an income effect of an inferior good combined with a price effect that is in the conventional direction. At the same time, the magnitude of the effect is potentially meaningful—an increase of 3.3 percentage points when the mean use rate is approximately 1%. These large but insignificant results on nursing home use make interpretation complicated, as they may also indicate that our analysis simply lacks sufficient power, perhaps due to the low rate of nursing home use.
Viewed together, we find stronger evidence for moral hazard in home care use than in nursing home use. This may be because nursing home use is less income-elastic than home care use if the LTCI payout is viewed as an income transfer. It may also be the case that home care use is more price-elastic than nursing home use if the LTCI payout is viewed as a lowering of the price of the service. Our analysis cannot distinguish between these two interpretations. It may also be that the income effect of LTCI reinforces a conventional demand-increasing price effect for home care use, but these effects conflict for nursing home use.
Our results should be interpreted in the light of several limitations. First, our analysis may be limited in its power to detect significant changes in long-term nursing home use and may be subject to measurement error, given self-reported responses in the HRS. We have conducted robustness checks to assess the potential effects of measurement error, which appears to have minimal effect on our results.
Second, our estimates should be viewed as local average treatment effects and may not be generalizable. A