The SSI countable resource limit has stood at USD 2,000 since 1989 — about two months of the benefit it pays to 7.4 million people

What is happening?

Supplemental Security Income pays a monthly benefit to aged, blind and disabled people who have almost no income and almost no property. To remain eligible a recipient may hold no more than USD 2,000 in countable resources, or USD 3,000 for an individual with a spouse present. Both figures were fixed by a statutory schedule whose final step took effect on 1989-01-01, and as of 2026-08-08 neither has moved since. About 7.4 million people received SSI in January 2026.

The section that carries the ceiling, 42 U.S.C. 1382(a)(3), writes it as a schedule of five annual steps rising from USD 1,500 to USD 2,000, ending on 1989-01-01. That section contains no cost-of-living provision and no successor adjustment mechanism. The number did not fail to be raised. The schedule ran out of instructions and nothing replaced it.

The benefit itself is indexed. A cost-of-living increase of 2.8 percent set the 2026 maximum federal payment at USD 994 a month for an individual and USD 1,491 for an individual with an eligible spouse, published in the Federal Register on 2025-11-03. That same annual notice re-prices 14 program parameters, one of which is an SSI eligibility parameter that is indexed, the student earned income exclusion at USD 2,410 a month. The word resource appears in the notice zero times. Measured against the 2026 benefit rate the ceiling is worth about 2.0 months of the payment it guards, a ratio computed here from the two published figures rather than stated by any source.

Whose problem is this?

The body that would have to raise the number is the only body that can, and it has not taken the question up. The agency that administers the rule has spent the interval doing the one thing left available to it, which is to check bank balances more closely.

Where does this problem end?

The boundary matters here because the program already exists and works, and only one number inside it has stopped moving. What is missing is not a program and not a proposal but an amendment to a single dollar figure.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state: no source opened here states what the limit should be, nor what condition would count as this being settled. A bill introduced in three consecutive Congresses proposes USD 10,000 for an individual and USD 20,000 for a couple with indexing thereafter, and NASI states that a limit indexed from 1972 would be approximately USD 10,000 for an individual today. That sentence names no price index and no end year, and no opened source shows the arithmetic behind it. No agency or committee document names a figure at all.

How big is it?

About 7.4 million people, and plausibly one to two million more. The Congressional Research Service published the recipient count for January 2026 as roughly 7.4 million — 1.0 million children under 18, 3.8 million adults aged 18 to 64 and 2.5 million people aged 65 or older. Every one of them is subject to the ceiling in every month of eligibility, so this is a floor and not an estimate of harm.

Above that floor sit the deemors. SSA counts an ineligible spouse or parent whose income and resources it considers when deciding whether the recipient qualifies, and the 1.0 million child recipients each sit in a household with one or two such adults. No opened source publishes a deemor count. Taking at most one to two per household containing a child recipient adds one to two million as an upper bound, which is deliberately generous, because a child whose parents both receive SSI has no deemor at all and siblings share the same parents. The range therefore runs from 7.4 million to about 9.4 million, and the upper figure is a bound rather than a measurement.

The count cannot reach the part that matters most. The rule binds continuously rather than as an event. Its cost is paid every month by people who stay below the line and never appear in any count of people above it, and a headcount of recipients currently over the ceiling would be close to zero while telling the reader nothing.

Under what conditions does it arise?

1. The ceiling is a statutory dollar figure, so the agency cannot move it. Every other lever SSA holds over SSI is regulatory, and the agency has used those levers. In a single month of 2024 it finalized three rules loosening what SSI counts, on food in kind support, on rental subsidy, and on the definition of a public assistance household. The resource ceiling is the part it cannot touch.

2. The freeze tightens by itself. The benefit rate is indexed, the student earned income exclusion is indexed, and the substantial gainful activity thresholds are indexed. The resource ceiling is not. Every annual increase therefore widens the distance between what the program pays and what it permits a recipient to hold, without anyone deciding anything.

3. The only response left to the agency is to verify harder. SSA identifies financial accounts holding countable resources above the limit as a leading cause of SSI overpayments, and those overpayments averaged about USD 2 billion a year across FY2020 through FY2024, more than the roughly USD 1.3 billion a year attributed to earnings discrepancies over the same span. The tolerance for bank matching fell from USD 400 to USD 0 for aged initial claim allowances in August 2025 and for all initial, appealed and re-opened claim allowances effective May 2026.

4. Holding a reserve is itself the disqualifying act. Qualitative research funded by SSA and a secondary analysis covering more than 75 working-age disabled recipients report the same behavioral pattern — vigilance about account balances, spending an inheritance down to the limit within thirty days, and declining life insurance because policy value counts. The rational response to the rule is to hold nothing, and holding nothing is what the program exists to prevent.

5. Marriage compresses the allowance rather than doubling it. Two single recipients may hold USD 2,000 each; married, the household ceiling is USD 3,000. NASI describes this as a 25 percent reduction in the asset limit, measured against twice the individual figure. Measured instead against the 2026 benefit rates the couple ceiling and the individual ceiling stand in about the same ratio to their own benefit, roughly two months each.

What has been tried?

Two directions have run at once and they point opposite ways. The agency loosened what it counts wherever the statute allowed, and tightened how it verifies whatever remains countable. The one direction that would move the number itself requires a vote, and across three Congresses that vote has not been scheduled.

What was found?

Why is it still unsolved?

Incentive inversion — the program exists to keep aged and disabled people out of destitution, and the most effective private defense against destitution is a cash reserve, which this rule makes a disqualifying holding.

The freeze is not neutral, because it tightens on its own. The benefit rate moves with the cost of living, the student earned income exclusion moves with it, and the substantial gainful activity thresholds move with it. The resource ceiling does not, because the section that sets it contains no clause that could carry an adjustment. Every annual increase therefore widens the distance between what the program pays and what it lets a recipient keep. Nobody has to act for the rule to become stricter. Someone has to act for it to stay where it is.

The lever sits in a body that has never pulled it. Only Congress can change a dollar figure written into statute, and the record of what happens when it is asked is very short. S.2767 arrived on 2023-09-12 with 11 cosponsors, was read twice and referred to the Committee on Finance the same day, and stopped there. There is no hearing, no markup, no roll call and no committee report. The bill was not defeated, it was never taken up, and a question that is never taken up leaves behind no adverse finding for anyone to answer.

The third part is what keeps the pattern stable. The only lever the agency holds points at the recipient rather than at the number. SSA cannot raise the line, so the sole thing it can do about resources is find them faster, and it has. Financial accounts above the limit are a leading cause of SSI overpayments, those overpayments run about USD 2 billion a year and exceed the earnings-related total, 78 percent of the FY2024 amount arose after the initial claim was processed, and the agency response was to lower the bank-matching trigger to zero. That zero-tolerance verification arrives in the same period in which SSA cut more than 7,100 jobs, about 13 percent of its workforce, closed six of ten regional offices, and removed customer-service metrics from its website. The threshold being enforced to the dollar was last set in 1989.

Put together, every actor is behaving sensibly and the aggregate is the opposite of the stated purpose. Recipients spend down because holding the money would end the benefit. The agency audits balances because overpayments are what it is measured on. Congress does nothing because nothing forces the question onto a calendar. A competing reading fits the same facts, that this is an institutional gap, since the adjustment schedule expired with no successor and there is literally no annual document into which the number could be written. That reading is kept here as the secondary one, because a pure gap would produce erosion alone, and what makes this live is that the erosion falls precisely on the behavior that would reduce dependence on the program.

What observation would mean it is solved?

Candidates — (a) Congress amends the statute to raise the ceiling and attaches an adjustment mechanism to it (b) SSI overpayments tied to financial accounts fall as a share of program spending (c) recipients report holding a cash reserve without expecting to lose eligibility for it.

(a) alone is weaker than it looks. A one-time increase without an adjustment mechanism restarts exactly the clock that produced this, and the level chosen decides how long the reprieve lasts. The bill that exists proposes USD 10,000 with indexing, and no opened source explains why that figure rather than another, so the reader cannot tell whether it is a restoration or a round number. An increase that arrives without indexing is 1989 again in slow motion.

(b) alone is worse. Overpayment dollars move with verification intensity as much as with recipient behavior, so a fall could mean fewer people holding more than the limit or simply fewer of them being detected, and the trigger just moved from USD 400 to USD 0. SSA was also asked why the figure rose between FY2023 and FY2024 and answered that its stewardship review does not capture the data needed to analyze national economic factors, so it has no explanation. A measure whose owner cannot account for its movement cannot carry this weight alone.

(c) alone counts a feeling, and nobody publishes it. The behavioral evidence rests on 17 in-depth interviews in one state plus a secondary analysis of qualitative work covering more than 75 recipients, which is enough to establish that the pattern exists and not enough to track it over time. A survey also cannot separate a change in the rule from a change in what people expect the rule to be. The three have to be read together, and (a) has to be read against whether an adjustment mechanism came with it.

What is it connected to?

Fills with researchasset tests in other means-tested programs, take-up and reach of the ABLE program, administrative burden research across benefit programs generally, and resource tests in other countries. Relation type and evidence grade were not confirmed in this round, and no comparison across programs is asserted here.

What these sources do not say

• Nobody publishes how many people the rule actually cuts off. Not one opened source gives a count of SSI applications denied for excess resources, or of recipients suspended or terminated for that reason, in any year. CRS states the limit and states that it is not indexed and says nothing about how many people it binds. The Inspector General counts dollars and never counts people.
• Nobody publishes the distribution of countable resources among recipients. Without it the public record cannot say whether the ceiling binds a few people severely or nearly everyone slightly, and those two shapes call for different remedies.
• The figure of approximately USD 10,000 if indexed travels without a method. NASI states 1972 as the base year and names no price index and no end year, and shows no arithmetic. Enactment in 1972, first payments in 1974 and the last increase in 1989 are all defensible base years and give materially different answers. The number appears in advocacy, in polling wording and in the dollar amounts of the bill, and its derivation is nowhere in the opened record.
• There is no target anywhere. The bill proposes USD 10,000 and USD 20,000 and the polling asks about at least those amounts, but no opened source explains why those figures rather than an indexed value, and no agency or committee document states what the limit should be or what would count as the matter being settled.
• No opened source gives a cost estimate. Across four years of coverage from 2022 to 2026, no opened source carries a budget score or any figure for what raising the limit would cost. Both news articles on the 2022 bill contain none, and the CRS In Focus opened here has no legislative-proposals section at all. Cost is the usual reason a threshold does not move, and that number is absent from the reachable public record.
• SSA states that it cannot explain its own trend. Asked why financial-account overpayment dollars rose from FY2023 to FY2024, the agency answered that the SSI stewardship review does not capture data to conduct analysis of national economic factors, and therefore it has no explanation. It measures the dollars and does not measure the cause.
• The Inspector General publishes its rates only as pictures. The SSI improper-payment and unknown-payment rates and the year-by-year financial-account deficiency dollars appear in the report as chart images. The narrative gives five-year averages and the compliance threshold of less than 10 percent, and never prints the rates on which it found the agency non-compliant.
• The annual re-pricing document never mentions the frozen parameter. The Federal Register notice that adjusts fourteen parameters for 2026, including the one SSI eligibility parameter that is indexed, uses the word resource zero times. The natural place to record that an SSI eligibility parameter is not indexed does not acknowledge that the parameter exists.
• Nobody prices the asset test. CRS reports USD 4.6 billion in FY2024 administrative and other costs as a single lump. The Inspector General describes bank-account matching at every application and redetermination, with the tolerance lowered to zero in 2026, without one figure for what that machinery costs to run or what it recovers net of that cost.

See the evidence

Two primary documents were read directly and they carry the load. The statutory text at Cornell LII establishes the mechanism, which is that the ceiling is a schedule that terminated rather than a figure that lags, and the full raw text of the Federal Register notice of 2025-11-03 establishes both the 2026 benefit rates and the absence of the word resource from the document that re-prices everything else. The Inspector General report was fetched as a PDF and read directly and is the source of every overpayment and verification figure here. The recipient counts and the statement that the limits are not indexed come from two Congressional Research Service products read through EveryCRSReport, because the Congressional Research Service site itself refused every request in this round. Everything else is second hand — advocacy, trade press and one tertiary encyclopedia entry used only to surface conflicts. Where sources disagree the disagreement is left visible rather than resolved: the last increase is dated 1989 by the statute, by CRS and by NASI and 1984 by one news article; the pre-1985 couple limit is USD 2,250 in the statute and USD 2,500 for 1974 in the tertiary source; and the poverty figures of 40 percent and about 3.3 million rest on different bases and different dates with neither traced to a primary table. Three things are carried as claims rather than as findings, and are marked as such above — the 2025-04-01 reintroduction, the figure of approximately USD 10,000 if indexed, and the polling internals. Seven evidence rows carry no URL because the host refused this session or the file exceeded the fetch limit; those rows are kept rather than deleted, and none of them is treated anywhere in this document as evidence that the public record is silent. This is a Path A output based on research rather than on a linked observation, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 7,400,000–9,400,000 As of January 2026

Sensitivity The width of the range is entirely the deemor term, and it is the distance between a published count and an upper bound built here, not a confidence interval. The low bound is close to exact for recipients as of January 2026; the high bound should be read as at least 7.4 million and plausibly one to two million more. What this number fails to count is the larger part of the effect. Nobody publishes how many applicants are denied for excess resources, or how many recipients are suspended or terminated for that reason, in any year. People who spent savings down in order to qualify appear in the 7.4 million only after the rule has already cost them the reserve, so the count records them as beneficiaries rather than as affected. People who never applied because they hold more than the limit are invisible by construction, since the effect of the rule on them is absence from every dataset. Former recipients terminated for resources are not counted anywhere, and neither are relatives who cannot make a gift or leave a bequest. The limit in the opposite direction is that this is not a count of people currently above the ceiling and must not be read as one. The binding effect is continuous rather than an event, and the interview evidence shows the cost is borne every month by people who stay below the line by not saving, by declining life insurance, and by spending an inheritance down within thirty days. A headcount of people currently over the line would be near zero and would be the wrong number.

Regional breakdown The resource limit is a single federal figure set in statute and applies identically in every state, so there is no state-level variation in the rule itself. No opened source publishes SSI recipient counts by state for January 2026, and no opened source publishes any state-level count of denials, suspensions or terminations for excess resources. State supplementary payment programs vary, but nothing opened in this round ties them to countable resources, so no regional figure could be confirmed and none is estimated.