FAA cut the controller availability factor to 1.87 for FY2026, setting a staffing target of 12,563 without publishing how trainees count

What is happening?

The Federal Aviation Administration publishes a controller workforce plan every year, and the edition dated 2026-06-01 sets the FY2026 total staffing target at 12,563 controllers. That number is the product of an input the agency changed in the same document. The plan states that FAA adopted a revised controller availability factor of 1.87, reduced from the prior factor of 2.14 used in the FY2025 plan, and that the FY2026 target of 12,563 follows from it.

The plan also changes how the workforce on the other side of that target is counted. It adopts a metric called CPC-Equivalent Workforce, which weights controllers still in training by equivalence ratios — and it does not publish the ratios. Its own appendix, listing all 313 air traffic control facilities with actual on board figures as of 2025-09-20, prints a target of 12,563, a certified professional controller total of 10,693, and a facility total of 14,223 counting every trainee. The document does not state which of those two on-board totals the target is meant to be read against.

The line most often used to describe an understaffed facility turns out not to be a staffing target at all. A DOT Office of Inspector General audit issued 2023-06-21 records that FAA officials stated that when the certified-controller staffing level at a facility falls below an 85-percent threshold, the agency prioritizes the placement of controllers at that facility. It is a rule for dispatching scarce people, agreed between the agency and the union that represents controllers. The same audit records that the underlying staffing levels were set as interim values in October 2014 and that after more than eight years no comprehensive review of them had been conducted for any facility, because the offices involved and the union did not agree on the availability factors used in the model.

Whose problem is this?

The body that sets the target is the same body that sets the rule for counting the workforce against it, and the audit office that measures the gap has to use whichever rule is in force at the time it measures.

Where does this problem end?

The boundary matters because two labour pools with different employers and different training clocks are routinely described as one shortage, and their gaps are reported in the same sentence as though they could be summed.

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

Now

Should be

The target state is published and dated. The FY2026 plan sets hiring targets of 2,200, 2,300 and 2,400 new controllers in FY2026, FY2027 and FY2028, and the audit office records that the agency announced it anticipates hiring at least 8,900 new controllers through 2028 to achieve full staffing levels and offset anticipated attrition. Full staffing itself is defined per facility: the plan prints a 2026 staffing target for each of the 313 facilities, summing to 12,563.

What is not published is the rule that connects the two sides. The plan adopts the CPC-Equivalent Workforce metric to weight trainees against the target and does not print the equivalence ratios, so the official reading of the plan cannot be reproduced from the plan.

How big is it?

What the record supports is a set of magnitudes, each carrying its own document and date. The FY2026 plan sets a target of 12,563 controllers and its appendix prints, as of 2025-09-20, a certified professional controller total of 10,693 and a facility total of 14,223 including trainees, with 608 more at the Academy for a grand total of 14,832. Separately, the audit office states the agency employs about 14,500 controllers in more than 300 facilities, of which almost 11,000 are certified, on employment data current as of 2026-03-07. An earlier audit gives about 13,300 controllers with about 10,600 certified in 2023. These are three counts from three documents with three dates, one of which includes the Academy and two of which do not, and they are not a series.

The second pool is counted separately and on a different clock. The contract tower program comprises 266 towers in 46 states staffed by more than 1,500 contractor controllers, handling more than 18 million tower operations in CY2024, about 30 percent of the total. That program was understaffed by 276 controllers, roughly 18 percent, as of April 2025. Those controllers are employed by contractors rather than by the agency, and the audit puts their replacement cycle at three to six weeks to hire and onboard plus an additional 30 to 60 days to certify.

The record does carry counts of facilities below a line, and each names the line it was measured against and the date it was measured on: 20 of 26 critical facilities below the 85 percent placement line and 22 of 26 below the workforce plan minimum with trainees excluded, both on March 2022 data, and 19 of the busiest facilities serving the Core 30 airports more than 15 percent below target, reported 2026-01-06. All three predate the revision of the availability factor in the plan dated 2026-06-01, and no document opened here restates any of them on the current baseline or prints a single system-wide shortfall figure.

Needs a new measurementthe affected population is not derivable as a count of people. No document opened here links controller staffing to a number of passengers, flights or delay minutes attributable to it, so the population file for this document records the estimate as not derivable rather than assembling one from parts that do not share a definition.

Under what conditions does it arise?

1. The measure was interim and then it moved. The audit issued 2023-06-21 records that interim certified-controller staffing levels were developed in October 2014 to be used until a comprehensive review was completed, and that after more than eight years no such review had been conducted for any facility. The measure that decides whether a facility is short was set as a temporary value and then left in place for more than eight years. One exception is recorded: a single route center was revised from 241 to 275 in 2022. In the plan dated 2026-06-01 the availability factor was revised from 2.14 to 1.87 and the FY2026 target of 12,563 follows from it.

2. Two baselines give different answers for the same facilities on the same date. Measured against the 85 percent placement line, 20 of 26 critical facilities were below it. Measured against the workforce plan staffing minimum with trainees excluded, 22 of 26 were below it. Both figures are as of March 2022 and both appear in the same audit, which records that considerable debate remained about how to account for the contributions of trainees and about the appropriate level of staffing.

3. Headcount and capacity are not the same quantity. The same audit records that as of August 2022 a controller who is not yet certified spends, on average, less than half of the time on position — 41 percent. A workforce total that counts trainees at full weight therefore describes something other than the capacity available to move traffic, and that is exactly the gap the unpublished equivalence ratios were adopted to bridge.

4. The constraint is throughput, not the calendar. The audit issued 2026-06-24 names two limits directly: the number of instructors at the Academy creates a practical limit on trainee throughput, and there are limits to how many trainees field facilities can absorb without delaying or degrading training at those facilities. Both are purchasable inputs rather than fixed durations.

5. The training system misses its own targets, and the targets were then adjusted. Across 2022 through 2024 only 2 of the 313 facilities met or exceeded their target training hours. Against targets the agency lowered for reported impediments, 65 percent of facilities met the adjusted target. Of more than 40 impediments identified since 2019, none had been removed as of the audit.

6. Two labour pools run on different clocks. The contract tower cycle is measured in weeks plus days and the federal cycle in years, and adding the two gaps together produces a number that describes neither.

7. The pipeline is exposed to appropriations. The FY2026 plan records that a lapse in funding at the beginning of FY2026 resulted in job declinations and trainee losses, and lists it alongside sequestration in FY2013, the FY2019 shutdown and the pandemic as disruptions to the same pipeline.

What has been tried?

Three directions ran at once — hire more, pay more to keep people, and change the model that says how many people are needed. The third arrived in the same document as the target it produced.

What was found?

Why is it still unsolved?

Measurement absent — the quantity that would say how short the system is has no settled definition, and the two documents that would fix it moved in the same year.

Start with the arithmetic that nobody performs. The plan prints a target of 12,563, a certified professional controller total of 10,693, and a facility total of 14,223 as of 2025-09-20, and it does not state which of the two on-board totals the target is to be read against. The metric it adopts to settle that question, CPC-Equivalent Workforce, weights trainees by ratios the plan does not print. So the document that sets the target does not supply the rule for reading it, and the reader is left holding three numbers and no operation.

The line that most descriptions of this problem rely on was never a measure of adequacy in the first place. The audit issued 2023-06-21 records that the 85 percent threshold exists so that the agency knows where to send controllers next. A dispatch rule tells you which facility is most starved relative to its peers. It does not tell you whether the system as a whole has enough people, and it cannot, because it is defined against a baseline that was itself provisional.

That baseline is the third part. It was set as an interim value in October 2014, it went more than eight years without the comprehensive review it was explicitly waiting for, and the reason recorded in the audit is that the offices involved and the union did not agree on the availability factors in the model. Then in the plan dated 2026-06-01 the availability factor was revised from 2.14 to 1.87, and the FY2026 target of 12,563 follows from the revision. A target that moves when one factor is revised does not settle the question it was built to answer.

The same shape appears one layer down, in training. Only 2 of the 313 facilities met or exceeded their target training hours across 2022 through 2024. The agency then credited facilities by allowing them to lower their hourly training targets, and against the lowered targets 65 percent of facilities met them. Two different numbers describe the same year of training depending on which target is in force, and the record does not carry a single figure that survives the change.

It also appears in the second labour pool, arriving from the opposite direction. The audit issued 2021-08-18 found that the program-wide minimum of four controllers per contract tower lacks any supporting analysis or study, and the 2026 audit that reports a shortfall of 276 controllers against that minimum also records that the agency faces challenges validating the staffing data contractors submit. The contract tower gap is measured against a standard the audit office itself called unsupported, using data the agency says it has difficulty validating.

Two mechanisms that are often blamed for this turn out to point elsewhere. Retirement is the smaller loss and it is shrinking: the plan projects 215 retirements in FY2026 against 743 losses at the Academy, and states that the long-anticipated retirement wave has passed. And the lead time is not a fixed constant either — the average weeks to certification fell from a December 2022 peak, and the audit office names the binding limits as Academy instructor supply and the number of trainees that field facilities can absorb. Both of those are quantities somebody buys. Which means the question of whether enough has been bought reduces, again, to a measurement: the training hours that facilities are missing, counted against a target that has already been adjusted once.

What observation would mean it is solved?

Candidates — (a) the agency publishes the equivalence ratios behind the CPC-Equivalent Workforce metric and a facility-level count of how many facilities sit below target under that rule (b) the number of facilities meeting or exceeding their target training hours rises from 2 of 313 without the target being adjusted downward (c) the average weeks to certification falls while the certified share of the workforce rises (d) the contract tower shortfall closes without federal hiring slowing.

(a) alone counts a publication, not a controller. Printing the ratios would make the gap readable; it would not change it by one person. Its value is that every later observation becomes checkable, which none of them are today.

(b) has a known escape route. The record already contains one instance of targets being adjusted downward when they were missed, and 65 percent compliance against the adjusted target sitting beside 2 of 313 against the original. Any future rise has to be read against which target was in force, or it measures the target rather than the training.

(c) is the cleanest of the four and still needs a caveat. The average weeks figure pools developmentals with already-certified controllers transferring between facilities, so it can fall because more transfers happened rather than because new hires certified faster. Read it with the certified share, and read both against the same date.

(d) can move for the wrong reason. The 2026 audit records that federal and defence hiring from the contract pool is the primary cause of attrition there, with one contractor reporting 35 percent of its controllers departing in CY2024 for such positions. The two numbers are coupled by construction: federal hiring success mechanically worsens the contract tower figure, so a closing contract gap may mean federal hiring stopped.

What is it connected to?

Fills with researchair traffic system modernisation and the state of control equipment · flight delay and cancellation statistics and how much of either is attributed to staffing · the continuity of federal appropriations, which the FY2026 plan names as a pipeline disruption · the repository problem `us-critical-legacy-system-modernization-stall`. Relation type and evidence grade were not established.

What these sources do not say

• Which on-board total the target is measured against. The FY2026 plan prints a target of 12,563, a certified controller total of 10,693 and a facility total of 14,223 on the same page, and states no rule for choosing between them. The metric it adopts to settle the question, CPC-Equivalent Workforce, depends on equivalence ratios the plan does not print, so the official reading of the plan cannot be reproduced from the plan.
• A single system-wide shortfall figure. None of the documents opened here states one. The FY2026 plan gives a target and separately reports about 11,000 certified controllers deployed with about 4,000 more in the training pipeline as of April 2026, and it performs no subtraction. The Fiscal Year 2026 Top Management Challenges report was searched in full text and carries no such figure either.
• A published washout rate. The FY2026 plan projects Academy attrition in absolute numbers — 743, 770 and 775 for FY2026 through FY2028 — and developmental losses of 208, 202 and 225, and it does not state the intake cohort those projections are computed against. Dividing them by the planned hire count for the same year mixes fiscal-year cohorts and would not produce the rate.
• Why the availability factors were disputed. The audit issued 2023-06-21 records that the offices involved and the union did not agree on them and that the comprehensive review therefore did not happen. It does not describe what the disagreement was about, and the FY2026 plan, which resolves the number, does not say whether the disagreement was resolved with it.
• What the total staffing target was before the revision. The workforce plan landing page carries only the current edition and offers no archive of prior editions, so the FY2025 total under the prior availability factor of 2.14 does not appear in anything opened here. The direction of the change is stated in the plan; the size of it is not.
• What the new equivalence ratios do to earlier counts. No document opened here restates any prior facility count under the FY2026 factor of 1.87, so the 20 of 26, the 22 of 26 and the 19 Core 30 facilities cannot be compared to anything measured after the revision.
• The count of facilities below their own target. The FY2026 plan prints per-facility targets and per-facility on-board figures and prints no count of how many facilities sit below. Neither audit prints one on the current baseline.
• How the two labour pools relate in the ledger. The 2026 contract tower audit records that federal and defence hiring is the primary cause of contract tower attrition, and no document opened here states what share of the 276 is a transfer within the national airspace system rather than a loss from it.
• What the independent review says in full. The findings on the distribution of staffing across facilities and on which staffing model to rebuild from are taken from the publisher project and publication summary pages. The report body sits behind a separate catalog flow and was not opened, and the sponsor restatement of the same report in the FY2026 plan is differently weighted from the publisher summary.

See the evidence

Three primary documents carry most of this file and all three were opened directly: the FAA controller workforce plan dated 2026-06-01, the DOT Office of Inspector General audit of on-the-job training issued 2026-06-24, and the audit of critical facility staffing issued 2023-06-21. Two further audits, on contract towers in 2021 and 2026, and the departmental challenges report of 2026-01-06 were opened as well. Where documents overlap they agree: the facility count of 313 appears both as an itemised footnote in the 2026 training audit and as 313 appendix rows in the workforce plan, and the age rules appear in both the recruiting page and the workforce plan. The certification lead time is stated in two places that do not share a starting point — 1.5 to 3 years in the training audit, measured from hire, and 1 to 3 years on the agency recruiting page, measured from after the Academy — so the two figures sit close together without being the same measure. Where they differ the difference is left visible rather than resolved: three controller headcounts of about 13,300, about 14,500 and 14,832 come from three documents with three dates and different treatment of the Academy, and the publisher summary of the independent review and the workforce plan restatement of the same review are differently weighted. Two findings rest on publisher summary pages rather than the report body and are graded medium for that reason. This is a Path A output (research-based definition), 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