SSA computed in 2008 that a hearing system without a backlog would produce a decision in 270 days — the average was 333 days in FY2022, and the expectation that each judge dispose of 500 to 700 cases a year was set in 2007 and unreviewed as of 2021

What is happening?

A claim for federal disability benefits passes through as many as five stages: a nonmedical eligibility determination at a Social Security Administration field office, a medical determination at a state Disability Determination Services agency, reconsideration, a hearing before an administrative law judge, and review by the Appeals Council. In 2008 an SSA workgroup calculated that a hearing system without a backlog would produce a decision in 270 days. That number became the target the agency carried into its 2016 CARES Plan, published at a point when SSA recorded that more than 1 million people were waiting for a hearing decision for the first time.

Average processing time for a hearing has not been at or under 270 days in any year the opened sources report. It was 543 days in FY2016 and rose to 605 days in FY2017, then fell to 506 in FY2019, 386 in FY2020, 326 in FY2021 and 333 days in FY2022. The audit does not report FY2018. The FY2020 fall coincides with the suspension of in-person and video hearings during the pandemic rather than with added capacity.

Two measures moved in opposite directions at the end of that series, and the difference is definitional rather than cosmetic. Pending hearings fell from about 1,100,000 in FY2016 to 346,567 in FY2022. The hearings backlog, which SSA defines as the pending cases that push average processing time past 270 days, grew 59 percent between FY2021 and FY2022. A shorter queue and a larger backlog are compatible, because the second is measured against a target and the first is not.

SSA could not say what produced the decline. An audit published by the Office of the Inspector General in September 2023 examined the 45 initiatives in the CARES Plan and found that 42 of them carried no documentation showing a direct causal link to the fall in the backlog or in processing time, and that the agency could not separate the effect of its own initiatives from a smaller number of cases arriving at the hearing stage out of reconsideration.

The capacity side of the equation has an older date on it. The expectation that an administrative law judge dispose of 500 to 700 cases a year was set in 2007 and had never been formally reviewed as of 2021, when GAO examined it. SSA had about 1,350 judges that year. Of 54 judges GAO surveyed, 47 said the expectation was too high, and SSA data showed the average case record had grown to roughly five times its 2007 size.

One measure had already doubled before any of the 2025 decisions were taken. The time a field office needs to make a nonmedical eligibility determination went from 120 days in FY2019 to 231 days in FY2024, a fiscal year that closed before the staffing reduction was announced.

In February 2025 a plan was announced to bring SSA staffing to about 50,000, a reduction of roughly 7,000 positions. A national business magazine reported in June 2026 that 7,100 people had left, more than 13 percent of the workforce, and that 6 of the 10 regional offices had closed. Telework ended in January 2025 and the share of staff working remotely fell from 71 percent in December 2024 to 13 percent in April and May 2025. The share of initial claims given an appointment within 30 days fell from 78.1 percent around July 2025 to 64.6 percent as of 2026-07-06, with some regional averages below 45 percent.

Whose problem is this?

The people who wait the longest are the ones least able to wait, because an application for disability benefits is filed by someone who has already stopped being able to work. Nobody standing in the queue can shorten it, and the decisions that set its length are taken in a budget cycle and a personnel policy rather than in the case.

Where does this problem end?

The line drawn here is between how long a decision takes and what the decision says. Both matter to the person waiting, and only the first is the subject of this document.

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

Now

Needs a new measurementthe target state: the 270-day figure is the only stated target any opened source carries, it applies to the hearing stage alone, and it was computed in 2008 rather than adopted as a current commitment. No opened source gives a target processing time for an initial claim or a reconsideration, a date by which service levels are to be restored, or a staffing level the agency states it needs. The indicators that would let a reader check progress against any such target were withdrawn from the public website in June 2025, and the one figure published since, a statement in November 2025 that initial claim processing time had fallen from 240 days in January 2025 to 209 days, cannot be set against the withdrawn series because no opened source says whether the two rest on the same definition.

How big is it?

What is confirmed is one stage at one date. 346,567 hearings were pending in FY2022, down from about 1,100,000 in FY2016. That figure covers the hearing stage alone, it is four fiscal years old, and it closed before the staffing reduction was announced, so it cannot be read as the number of people waiting now.

The capacity that number runs against is known for one year. About 1,350 administrative law judges were in post in 2021, each expected to dispose of 500 to 700 cases. The expectation was written in 2007, and by 2021 the average case record had grown to about five times its 2007 size, so the same annual number describes a larger amount of work than it did when it was set.

Two other quantities describe the size of the program rather than the size of the queue, and neither substitutes for it. About 16 million people were receiving SSI or SSDI benefits in 2026, which counts people already awarded benefits rather than people waiting for a decision. New disability applications fell about 7 percent in the first half of 2025 against the same period a year earlier, which is a rate of change with no level attached to it.

Needs a new measurementthe number of people waiting. No opened source counts, at a single date, the people whose disability claim is pending anywhere across the five stages. SSA reports the hearing stage and the initial claim stage in different places, and the population file for this document therefore records the estimate as not derivable rather than assembling one out of parts that do not share a date.

Under what conditions does it arise?

1. The volume of work is set outside the agency and the capacity is set inside a budget. Applications arrive as a function of the economy, the aging of the workforce and the incidence of disabling conditions. Staff and judges arrive as a function of appropriations and personnel policy. Nothing binds the second to the first, so a gap opens whenever either moves without reference to the other.

2. The productivity expectation is a fixed number attached to a growing unit of work. A judge is expected to dispose of 500 to 700 cases a year, a figure set in 2007. By 2021 the average case record had grown about fivefold, and 47 of the 54 judges GAO surveyed said the expectation was too high. A quota that does not move as the case grows becomes a quota on how closely each case can be read.

3. The queue can be measured in ways that improve while the shortfall widens. Pending hearings fell to 346,567 in FY2022 while the backlog measured against the 270-day target grew 59 percent in the same year. A reader looking at either number alone reaches the opposite conclusion from a reader looking at the other.

4. Removing capacity takes one decision and shows up across every stage. The reduction was announced in February 2025 and 7,100 people had left by the time it was reported in June 2026. Between July 2025 and July 2026, the share of initial claims given an appointment within 30 days fell from 78.1 percent to 64.6 percent, and in some regions the average fell below 45 percent.

None of these four conditions is about whether a particular claim deserves benefits, and all four are about whether anyone is available to read it.

What has been tried?

The first four attempts were aimed at the wait and the last two were not. A staffing reduction and the withdrawal of the measures that would show its effect are not attempts to shorten a queue, and they sit in this table because they are the most recent things done to the system that produces it.

What was found?

Why is it still unsolved?

Supply shortage — the number of people available to decide a claim is set by a budget and a personnel policy, and neither is set by the number of claims arriving.

The clearest evidence that this is a capacity problem rather than a management problem is that the agency itself could not show otherwise. SSA ran 45 initiatives across three phases of the CARES Plan and an audit found that 42 of them had no documented causal link to the improvement that occurred. The agency also could not rule out that the improvement came from fewer cases arriving at the hearing stage. When a decade of process work cannot be distinguished from a change in inflow, what is left is the number of people available to decide.

The second reason is that the standard the work is measured against was set once and left. A judge is expected to dispose of 500 to 700 cases a year, a figure written in 2007, when the average case record was about a fifth of what it had become by 2021. GAO found no formal review of that expectation in the fourteen years to 2021 and recommended that SSA build a process to review it periodically. The absence of a review is not a disagreement about the right number. It is the absence of anyone whose job is to ask.

The third reason is that the inputs move quickly and the measures move slowly. A staffing reduction of 7,100 was announced in February 2025 and carried out inside a year. The processing-time series that would show its effect ends at FY2022 in the audit, and the indicators SSA published in the meantime were withdrawn from its website in June 2025. A system whose inputs can be changed in a month and whose outputs are published years later cannot be corrected by the people reading the outputs.

What observation would mean it is solved?

Candidates — (a) average processing time for a hearing falls to 270 days or below and stays there across several consecutive years (b) SSA publishes a target processing time for every stage of the determination process and reports against it (c) the staffing and judge complement is set by a stated relationship to the volume and complexity of claims rather than by an annual appropriation alone.

(a) alone is weak for a reason this document has already recorded. Average processing time fell from 605 days in FY2017 to 326 in FY2021 without anyone being able to show what caused it, and part of the FY2020 fall came from hearings being suspended rather than decided. A number that improves when the system stops working is not a measure of the system working.

(b) is a measurement condition rather than an outcome, and it is a necessary one. The indicators that would let anyone check (a) were withdrawn in June 2025, and the single figure published since cannot be compared with the withdrawn series because no opened source says whether the definitions match. An observation that cannot be made is not an observation.

(c) is the only candidate that addresses the failure named here, and it is the least visible from outside. Nothing opened here describes such a relationship, and the one capacity standard that does exist, the expectation of 500 to 700 cases, is itself the example of a number that stopped tracking the work it describes. The three have to be read together, and (a) has to be read against what was happening to hearings in the year it improved.

What is it connected to?

Fills with researchthe trust fund solvency projections that govern the same agency, the accuracy of disability determinations as distinct from their speed, state-level variation in the capacity of Disability Determination Services agencies, federal telework policy across other agencies, and the standing of the federal disability programs on the GAO High-Risk List. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• How many people are waiting for a determination now. SSA reports the hearing stage and the initial claim stage separately, and no opened source combines them at a single date. Figures for pending initial claims surfaced during the research stage but were not opened, so they are not carried here.
• What the hearings backlog did after FY2022. The audit series ends there. Whether the increase of 59 percent from FY2021 to FY2022 continued, reversed or accelerated after the staffing reduction is stated by nothing opened here.
• How many administrative law judges are in post now. The last directly confirmed count is about 1,350 in 2021. Reports of a decline of about 13 percent and of a twenty-year low rest on a source the research stage could not open.
• Whether the indicators published before and after June 2025 are the same indicators. SSA withdrew phone wait times and disability processing times from its website in June 2025. A statement in November 2025 that initial claim processing time had fallen from 240 days in January 2025 to 209 days is cited in a GAO report, and no opened source says whether it rests on the definition the withdrawn series used.
• Why the public indicators were withdrawn in June 2025. No opened source gives a reason for the withdrawal, and no opened source connects it to the staffing reduction announced four months earlier. The two appear together in the attempts table because they are the most recent decisions recorded about this system, and their order in time is the only relation anything opened here establishes between them.
• Why new disability applications fell about 7 percent in the first half of 2025. The analysis reported shows the decline and identifies no cause. Fewer applications can mean less need or a harder path to the counter, and nothing opened here separates the two.
• What the staffing reduction did to each stage. The figure is agency-wide. No opened source divides it between field offices, state determination agencies, hearing offices and the Appeals Council, so the stage at which the loss lands is unknown.
• What SSA says it needs. No opened source carries a staffing level, a judge complement or an appropriation figure that the agency states would meet the 270-day target.

See the evidence

No primary SSA dataset was read directly. The two GAO reports and the OIG audit are federal oversight documents and they carry every figure here that predates 2025, but each was reached through its published file rather than through the underlying SSA series, and no figure was taken from an SSA statistical table. The 2025 and 2026 figures come from two press reports. Where those overlap with the GAO account they agree in direction and differ in value, and both values are kept rather than merged: GAO carries the announced reduction of about 7,000 toward a target of about 50,000, and the business magazine report carries 7,100 as the number who left. Where the sources are silent the silence is left visible rather than filled — the processing-time series stops at FY2022, the judge count stops at 2021, and no source states what the agency needs. Four sources in this table returned HTTP 403 to automated fetching during the research stage, so they were not opened, their URL column is blank and no claim rests on them; the most consequential of the four is the congressional justification, which is where a stated staffing requirement would be found. This is a Path A output, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range Not derivable

The reason and what is missing are listed under “What is missing” below