Medicaid and crowd-out of long-term care insurance

Although long-term care costs represent a substantial financial risk for retired households, few purchase insurance. Previous research shows that it would not be optimal for most single individuals to purchase coverage, due to crowd-out by the means-tested partial
insurance provided by Medicaid. Married couples pool risk (and so should value insurance less)but also face a greater cost in case the care of an infirm spouse impoverishes a healthy spouse (and so should value insurance more); recognizing this, Medicaid offers greater income and asset protection for married couples than for singles, increasing the implicit tax on private insurance.
We construct a model in which retired households decide whether to insure themselves via a private policy or implicitly via saving or anticipated future Medicaid use, given their expectation of needing care. We also make use of new estimates of the likelihood of needing care, which differ in important dimensions from earlier estimates. Using numerical optimization techniques, we calculate that only married couples in the two top wealth deciles will be willing to purchase an actuarially fair insurance policy, implying even more crowd-out than for singles.
Nevertheless, the absence of comprehensive insurance results in substantial welfare losses, amounting to an average of 10 percent of age-65 financial assets, because Medicaid provides only partial insurance while crowding out private purchases fully. We calculate that eliminating Medicaid spousal protection rules might increase coverage by XX percentage points. In contrast, plausible premium subsidies would have little effect on long-term care insurance coverage, given the current protection offered by Medicaid.