A framework, evidence review blog, and research and analysis plan for assessing how long-term care subsidies could affect US federal and state healthcare and long-term care spending and financing capacity.

Status: Conceptual framework (research in progress, model building in 2027) Last updated: October 3, 2026.

Model question

What LTC subsidy policy would yield optimal 10-year federal and state fiscal health over 10 years?

Empirical questions

Do increases in LTC subsidies:

• reduce long-term healthcare spending?
• increase government tax revenue?
• reduce long-run total public LTC costs? Hypothesis: yes to all three questions above. The pursuit of answers to those questions will be impartial and factual. In the absence of strong evidence, the model will allow users to fill those evidence gaps with hypothetical inputs.

Core Scenarios

This project will compare changes in subsidy eligibility against a baseline. These scenarios may change depending on availability of evidence.

• Baseline: maintain current LTC subsidy eligibility and benefit rules. Fiscal projections are driven by changing needs and costs.
• Scenario A: expand current eligibility to people with a lower level of LTC need but otherwise meet other current eligibility requirements
• Scenario B: expand current eligibility to people with more financial resources but otherwise meet current eligibility requirements.

In the model the subsidy options will be shown with varying degrees of expansion and with the resulting increase in LTC subsidization (in dollar terms). The project will highlight the optimal level of LTC subsidization for long-term fiscal sustainability, under the model. These will be the most favorable fiscal outcome for the user’s selected perspective (federal plus state, or federal or state alone), after projected scenario-attributable changes in healthcare and LTC spending and in relevant tax revenue.

How the framework works

Baseline build-up:

Projected demographics
        ↓
People with LTC needs, by need category
        ↓
Hours of assistance needed, by type
        ↓
Resource value of current subsidy-eligible needs ($)
        ↓
Subsidy-to-need ratio under current eligibility rules
        ↓
Subsidized service utilization
        ↓
LTC spending

Scenario build-up:

Eligibility rules' change in eligible people
+
Resource value of newly subsidy-eligible needs ($)
        ↓
Newly eligible people's needs met by subsidy-to-need ratio
        ↓
Subsidized service utilization
        ↓
LTC spending

Changes in subsidized services will also be assessed through two linked pathways:

Changes in LTC subsidies
        ├──→ Changes in healthcare utilization and spending
        └──→ Labor-market and economic effects
                         ↓
                 Relevant tax revenue

These relationships are hypothetical at this point, so the project may drop some proposed components of the model.

Baseline and scenarios

The baseline reflect current benefit and eligibility rules amid changing projected care utilization and prices.

The first version will vary one eligibility dimension at a time:

Scenario	Policy change
Financial eligibility expansion	Lower income or asset requirements for people who meet current care-need requirements.
Care-need eligibility expansion	Lower care-need requirements for people who meet current financial eligibility requirements.

Each scenario will include a sliding scale of eligibility changes; exactly how the scale will work is TBD.

Newly eligible people’s needs will be subsidized at the current aggregate subsidy-to-need ratio (SNR) for home care users, which this project will calculate as:

SNR = [Total (I)ADL LTC subsidy amount] / [Total need ($) for subsidized eligible (I)ADL LTC users]

A major effort will be converting LTC need to a dollar amount, reflecting the private-pay value of services estimated to be needed at current (I)ADL dependence levels. Each person’s hours of assistance needed by service type will be proxied by wealthy people’s hours used given a certain ADL level. Then, the value of those hours is calculated at average private-pay prices. This analysis, which likely will use HRS data (or MCBS PUF data), still needs to be planned.

Financial accounting

All model results will be measured relative to baseline. If a modeled scenario has a $100 billion smaller fiscal gap than the baseline fiscal gap (spending vs revenue), the output of this project would describe the scenario as achieving a modeled $100 billion 10-year net savings.

The annual change in the financing gap will be defined as:

$$\Delta Gap_t = \Delta \text{LTC spending}_t - \Delta \text{Healthcare spending}_t -\Delta \text{Relevant tax revenue}_t$$

A negative value indicates an improvement relative to baseline.

This reflects the total scope of this project’s model:

• Changes in LTC subsidy levels, in dollars, under scenarios’ policy changes
• Resulting impacts on healthcare spending (evidence based)
• Tax revenue changes (typically payroll and general-revenue), such as those caused by labor market shifts (for both the LTC workforce and would-be caregivers), income effects, and estate taxes

Federal and state results will be reported separately and combined. To do so, the model will split Medicaid spending by federal and state.

Outputs

The model will produce annual estimates over ten years, alongside cumulative net present value and net future value figures.

The main 10-year fiscal trajectory visualization will show baseline spending and signed scenario adjustments for spending and revenue. (For now, the model results will be point estimates.) Supporting webpages will explain the mechanisms, assumptions, and evidence.

Evidence and review status

Regular blog posts will explore new evidence. An initial batch of evidence reviews generated by AI will be reviewed by the author over time. Evidence gathering will focus on the model framework’s key mechanisms and assumptions, directionally and in terms of magnitude. Evidence review may change the model’s scenarios, effects, and structure, where the evidence overturns the author’s initial hypothesis.