Sixty Billion Dollars No One Claims: The Caregiver Benefits Gap

A woman in Georgia spends 35 hours a week caring for her mother with dementia. She manages medications, handles insurance calls, coordinates doctor visits, and drives her mother to adult day care three mornings a week. She earns $38,000 a year at a job she can't afford to lose. She qualifies for at least four federal and state programs that could ease her financial burden, provide respite care, and cover some of her out-of-pocket costs.

She doesn't know any of them exist.

This is not an edge case. This is the system working exactly as designed. Or rather, as it was never designed at all.

63 Million People, Zero Coherent Support System

The 2025 AARP/NAC Caregiving in the U.S. report confirmed what caregivers already knew: the crisis is accelerating. 63 million Americans now provide unpaid care, a 45% increase since 2015. Nearly one in four adults is a caregiver. They spend an average of 27 hours per week on care tasks. One in four provides 40 or more hours: a full-time job they will never be paid for.

The work has grown more complex. 55% of caregivers now handle medical or nursing tasks: wound care, injections, medication management. Only 11% received any training. Nearly half report major financial impact: debt, depleted savings, food insecurity. 64% report high emotional stress. Six in ten are employed, and half of those say caregiving has disrupted their work.

These numbers describe a population the size of France performing unpaid labor that the economy depends on and the government barely acknowledges.

The Benefit Programs Exist. The Problem Is Finding Them.

Here is what makes this crisis so frustrating: the money is already allocated. The programs already exist. And most eligible caregivers never access them.

Over $60 billion in benefits goes unclaimed every year across safety net programs in the United States. Participation rates range from 40-60% depending on the program. For some caregiver-specific supports, uptake is far lower. Only 12% of eligible caregivers access FMLA benefits. Only 7% use VA caregiver programs.

The reasons are structural, not personal. Benefits are scattered across three levels of government, dozens of agencies, and hundreds of local organizations, each with its own eligibility rules, application forms, and jargon:

Federal programs include SNAP (food assistance), Medicaid (healthcare), the National Family Caregiver Support Program (respite and counseling through 600+ Area Agencies on Aging), VA PCAFC (monthly stipends up to $3,200 for caregivers of disabled veterans), the GUIDE Model (new dementia care coordination), LIHEAP (energy assistance), and FMLA (unpaid job-protected leave).

State programs multiply the complexity. 14 states plus DC now offer paid family leave. Eight states provide caregiver tax credits. Eleven states run Self-Directed Care or Structured Family Caregiving programs that pay family members as care providers. Every state runs Medicaid differently. There are 50+ distinct Home and Community-Based Services waiver programs, each with its own income thresholds, enrollment caps, and waitlists.

Local programs fragment things further. The 600+ Area Agencies on Aging each serve different populations with different resources. County-level programs, nonprofit services, and faith-based supports add another layer.

No single tool, website, or hotline maps all of this. The caregiver in Georgia would need to search federal benefit screeners, her state's Medicaid portal, her local AAA, the VA website, and her county's social services office. That assumes she knows each exists, has the time to work through them, and can stomach applying separately to each.

Why People Don't Claim What They're Owed

The research on benefit take-up barriers is clear and consistent. It is not laziness. It is not ignorance. It is a system designed to be difficult.

Administrative burden is the primary barrier. Complex forms, documentation requirements, in-person verification, and multi-step processes disproportionately screen out the people most in need. A caregiver already spending 27 hours a week on care tasks does not have a spare afternoon to gather tax documents, proof of residency, medical records, and employer verification letters for one application. Let alone four.

Information gaps compound the problem. Over half of caregivers report lacking information in at least one area of caregiving. Many don't know caregiver-specific programs exist. SNAP and Medicaid have some public awareness; the National Family Caregiver Support Program, state-paid caregiver programs, and respite vouchers do not.

Stigma remains a real barrier. Research published in Health Affairs Scholar (2024) links welfare stigma to discrimination and stress that actively hinders access to benefits. Many caregivers don't identify as "caregivers." They're daughters, sons, spouses doing what family does. They resist engaging with systems they associate with charity.

Benefits cliffs create perverse incentives. An AEI study found that 22% of assistance recipients took negative actions (turning down raises, reducing work hours) to avoid losing benefits. 62% felt stuck in low-income jobs. For caregivers already balancing work and care, the risk of losing Medicaid coverage or SNAP benefits by earning a few hundred dollars more can make rational people turn down income they desperately need.

Cognitive load might be the most powerful barrier and the least discussed. Caregivers are exhausted. They make dozens of care decisions a day. Present bias (prioritizing immediate needs over long-term benefits) is not a character flaw. It's the predictable result of chronic stress and sleep deprivation. A caregiver who managed a 3 a.m. medication crisis is not spending the next morning researching eligibility criteria.

The Tools That Exist Today, and What They Miss

Several organizations have built benefit screening tools, and some are excellent at what they do:

• PolicyEngine models tax and benefit programs with sophisticated microsimulation. It received an OpenAI/PSL Foundation grant in 2025 for democratizing policy analysis. But it covers no caregiver-specific programs.
• MyFriendBen (from Propel) screens for SNAP, Medicaid, and other safety net programs with a user-friendly mobile interface. Caregiver programs are absent.
• mRelief focuses on SNAP and housing, primarily in Illinois. No caregiver coverage.
• Benefits.gov and USA.gov provide federal benefit information but offer limited personalized caregiver-benefits pre-screening and state-level detail.

These tools share a common gap: none of them model caregiver-specific programs. NFCSP, VA PCAFC, state-paid family leave, Structured Family Caregiving, caregiver tax credits, respite voucher programs. The benefits most relevant to 63 million Americans providing unpaid care are missing from every existing screener.

This gap is not accidental. Caregiver programs are genuinely harder to model. SNAP checks income against the federal poverty level. Straightforward. But the National Family Caregiver Support Program requires that the care recipient be 60 or older (or have Alzheimer's at any age), that the caregiver is an adult family member, and that priority be given to incomes under 200% FPL, with no hard cutoff and local AAA discretion baked into the rules. VA PCAFC requires a 70%+ disability rating, a specific relationship or cohabitation with the veteran, and demonstration that personal care services are needed for six or more continuous months. State Structured Family Caregiving programs each define "family caregiver" differently, set their own compensation rates, and impose different training requirements.

The eligibility rules for caregiver programs resist the simple threshold-based modeling that works for SNAP and Medicaid. They require structured data that captures relationship types, care intensity, geographic variation, and the interactions between programs. No one has built this.

What Needs to Exist

The gap calls for something specific: a structured, machine-readable database of every caregiver benefit program at federal, state, and local levels, paired with a benefits pre-screening engine that can ask simple questions and surface programs worth checking against official rules.

This is not a trivial data problem. The architecture needs to handle real-world complexity:

• Federal programs with 50 state variants. Medicaid HCBS waivers operate under federal authority but differ in every state. A single "Medicaid" entry is meaningless without the state-specific parameters.
• Many-to-many taxonomy mappings. A single program can serve multiple benefit types (financial assistance, respite care, training) and a single caregiver can qualify for programs across multiple categories.
• Shared eligibility parameters. Income, age, disability status, veteran status, and geographic location are checked repeatedly across programs. A well-designed system asks these questions once and applies them everywhere.
• Temporal data. SNAP thresholds change every October. Federal poverty levels update every January. State paid leave programs phase in over multiple years. The data must track when it was last verified and when it needs re-verification.

Established open taxonomies provide a foundation. The Open Eligibility Project offers a standardized human services taxonomy used by 211 systems nationwide. CMS publishes structured data on HCBS waiver programs. The National Strategy to Support Family Caregivers provides a federal framework. But no one has synthesized these into a unified, queryable system focused on caregiver benefits.

What GiveCare Is Building

We started building benefits-cli-tools: an open-source Python CLI for caregiver benefits discovery, pre-screening, and tracking.

The approach is research-first. Before writing a line of screening logic, we compiled over 1,000 lines of documented findings spanning federal and state programs, existing tools, open taxonomies, academic literature on knowledge graphs and eligibility modeling, and the Rules-as-Code movement that is gaining momentum across government technology. The Beeck Center at Georgetown, Code for America's Digital Benefits Network, and the Policy2Code Challenge at BenCon 2024 all point toward the same conclusion: eligibility rules should be structured, open, machine-readable, and interoperable. No one has applied this approach to caregiver-specific programs.

We're starting with five federal programs (SNAP, Medicaid, NFCSP, VA PCAFC, and LIHEAP), each modeled against a Pydantic schema that captures eligibility criteria, benefit details, application processes, and source provenance (including when data was last pulled from .gov sources and when a human verified accuracy). The schema uses flat-optional eligibility fields because each program tests a different subset of criteria: SNAP checks income, NFCSP checks care recipient age, VA PCAFC checks disability rating. Every field is documented, every threshold is real, every source URL points to a .gov or verified authority.

From five federal programs, we expand to state variants: the 14+ paid family leave programs, 50+ Medicaid HCBS waivers, 11 Structured Family Caregiving states, and 8 caregiver tax credit states. Then local: the 600+ Area Agencies on Aging and their program offerings.

The tools are designed from day one for integration with the GiveCare platform. A caregiver texting GiveCare at 11 p.m. shouldn't need to know that NFCSP exists, or that their state offers a caregiver stipend, or that they may be under the income threshold for SNAP. They should answer a few questions about their situation (who they care for, their relationship, their income, their state) and see which programs may be relevant, with clear next steps and reminders to confirm details on official program sites.

The Standard We're Holding Ourselves To

Benefits data is a domain where mistakes have real consequences. A caregiver who spends hours applying for a program she doesn't qualify for loses time she doesn't have. A caregiver who is steered away from a program that might fit loses money her family needs.

Every program entry in our database validates against a strict schema. Source URLs must point to .gov or verified authority sites. "Accessed" dates must be real: the actual date someone pulled the data. Eligibility thresholds must come from official program documentation, not estimates or approximations. We track a freshness interval for each program because federal poverty levels change in January and SNAP thresholds change in October. Stale data is flagged, not served.

The code is open source because the problem is not competitive. If another organization wants to build a caregiver benefit screener, they should be able to use our program data and schema. If a state agency wants to verify that we've modeled their programs correctly, they should be able to read our code and file corrections.

This Problem Is Solvable

The caregiving crisis has many dimensions that resist easy solutions. You cannot engineer your way out of insufficient Medicaid funding, or automate your way past a society that devalues care work, or build an app that replaces the structural support caregivers need from employers and government.

But the benefits gap, the space between "programs that exist" and "caregivers who access them," is a solvable information problem. The programs are funded. The eligibility rules are public. The data can be structured. The screening can be automated. The results can be delivered through channels caregivers already use.

Over $60 billion sits unclaimed every year because the system that distributes it is too fragmented for the people it's meant to serve. 63 million Americans provide care that holds families and communities together. The least we can do is make sure they know what help is available and how to get it.

No caregiver should miss benefits they're entitled to because the system is too fragmented to search. That is the gap. That is what we're building to close.

---

GiveCare is building open-source tools for caregiver benefits discovery. Follow our progress at github.com/amadad/benefits-cli-tools or reach us at info@givecareapp.com.