We use two historical data sources – the Health and Retirement Study and the Medicare Current Beneficiary Study – to consider the patterns in older Americans’ severe disability and their use of long-term services and supports (LTSS) by age and socioeconomic status. We then use a dynamic microsimulation model to project how the effects of various interventions to support those with severe disabilities and their caregivers would be distributed across the income distribution. The interventions that we examine fall into three broad classes: tax credits for caregiving expenses, respite care for people in the community with family caregivers, and new social insurance programs. Within each broad class of policies, we examine how sensitive outcomes are to changes in policy details (such as, in the case of tax credits, deductible levels, refundability, and income phase-outs).
This paper found that:
- Older adults with less education and less wealth are more likely to report disabilities and service use than their more educated and wealthier counterparts.
- This pattern persists when we look at people at a point in time but also, more robustly, when we look at their disabilities prospectively. In a sample of older adults who do not report disabilities at baseline, we find that those with fewer economic resources earlier in life are generally more likely to develop disabilities and use paid LTSS over the next two decades, but the differences narrow when we restrict the sample to people who do not develop disabilities until their late 70s.
The policy implications of this paper are:
- The uneven distribution of disability risks across the population poses challenges for developing effective LTSS policies. Those most likely to need LTSS often lack enough resources to contribute to LTSS programs, and programs that try to contain costs by using underwriting or imposing work requirements often disqualify those who most need coverage.
- Certain classes of policies, such as respite care benefits, tend to direct much of their benefits to those in lower income quintiles, according to our projections. Caregiver tax credits and social insurance programs generally distribute benefits more proportionally, although impacts vary depending on how the policies are specified.
- Policy design details can significantly affect distributional outcomes. Provisions’ effects can be sensitive to the stacking order in which they are implemented.
- It can be useful to examine trends and proposals not only cross-sectionally but also over longer time periods. For example, the distributional effects of social insurance programs depend on the relatively high early-life mortality of those with less education and lower earnings and wealth.