Study VIII · The Federal Dollars Series

Every Federal Job in the Region

Twelve pictures of what happened to the federal payroll in Washington, Maryland and Virginia between January 2025 and August 2026.

5 min read · 12 exhibits

The national study, Every Federal Job, follows every federal civilian job across the country, Exhibits 1 to 10.

The short version

The Washington region lost 63,900 federal jobs between January 2025 and August 2026 — 17 percent, against 10.9 percent nationally to September 2026 (BLS, all federal jobs, Postal Service included). The three states together lost 83,500.

Half of the national cut (166,000 of 328,000, January 2025 to September 2026) landed in a single month. October 2025, when the deferred-resignation cohort came off payroll, took 166,000 jobs (BLS, all federal jobs, Postal Service included) — 5.7 percent of the 2,914,000 federal jobs of September, in thirty days.

Maryland lost a larger share than the District: 18.1 percent against 14.6. Virginia lost 13.4 percent. The region’s share of all federal jobs fell from 12.5 to 11.6 percent (375,800 of 3,010,000 in January 2025; 311,900 of 2,683,000 in August 2026; the metro area not seasonally adjusted, the nation seasonally adjusted).

Federal employment is back to 1966. September 2026’s 2.68 million (BLS, all federal jobs, Postal Service included) is below every December since then (BLS December values, seasonally adjusted); the last month as low was May 1966.

Layoffs did almost none of the cutting. Of 386,826 separations from January 20, 2025 to January 2026 (Partnership for Public Service, from OPM data), 10,436 were reductions in force; 136,822 were buyouts; the other 239,568 were all other separations (outside the program and RIFs). Buyouts out-cut layoffs thirteen to one.

The federal civilian payroll is a quarter of the District’s economy (25.1 percent of GDP), a tenth of Maryland’s and one-twentieth of Virginia’s. Virginia’s federal money arrives mostly as contracts: $120.6 billion of federal contract obligations recorded to Virginia in FY2025 against $44 billion of payroll output.

The money lagged the jobs by a year. Payroll output barely moved in 2025 because departing workers were paid through September; the first quarter of 2026 shows DC federal earnings down $1.1 billion at an annual rate.

The DC CFO forecast 40,000 over the plan period; by June 2026 it counted 28,000 and called the trend on track. Oxford Economics forecast 33,700 for the metro in 2025; the MSA lost 50,400 from January to December 2025 (BLS, not seasonally adjusted), so the forecast was about two-thirds of it. The payroll has been flat since January 2026; the cutting phase, as the data see it, is over.

DOGE’s lease savings fell from a claimed $400 million to $53.5 million on DOGE’s own ledger, GAO found, against $5.75 billion of GSA’s annual lease rent; the terminations of 108 of the 264 leases were already under way before DOGE, GAO found. Metro trips rose 9 percent in FY2025 and rail ridership 9 percent in FY2026.

The count

Two instruments measure the federal workforce. The Bureau of Labor Statistics’ payroll survey counts jobs at establishments every month, Postal Service included. The Office of Personnel Management’s records count executive-branch civilians on board, Postal Service excluded, with a lag. They do not agree on levels and are never mixed in one chart here. Every monthly figure below is BLS.

Exhibit 11

The region fell faster than the country

Source. BLS Current Employment Statistics via FRED: CES9091000001 (US), SMS11000009091000001 (DC), SMS24000009091000001 (MD), SMS51000009091000001 (VA), SMU11479009091000001 (Washington MSA, not seasonally adjusted). January 2024 – September 2026 (states and MSA to August 2026).

Full size

Method. Federal government employment indexed to January 2025, monthly from January 2024. States and the US seasonally adjusted; the Washington MSA is not seasonally adjusted because BLS publishes no adjusted federal series for it. CES counts jobs at establishments and includes the Postal Service. The national series ends in September 2026 at 89.1; the Washington MSA in August 2026 at 83.0; Maryland at 81.9; the District at 85.4; Virginia at 86.6. The step in October 2025 is the deferred-resignation cohort leaving payroll. The levels (thousands) are the cited ex01 file; the index columns are the cited ex01b.

Exhibit 12

The region’s federal payroll, month by month

Source. BLS CES SMU11479009091000001 via FRED, not seasonally adjusted (BLS publishes no seasonally adjusted federal series for the MSA). January 2024 – August 2026.

Full size

Method. Federal jobs in the Washington–Arlington–Alexandria metropolitan area, thousands. 375.8 thousand in January 2025; 353.8 in September; 326.8 in October 2025, the month the deferred-resignation cohort came off payroll; 311.9 in August 2026 — a fall of 63.9 thousand, 17.0 percent. The series has been flat at 312–314 thousand since January 2026.

Exhibit 13

Maryland lost the largest share

Source. BLS Current Employment Statistics via FRED: CES9091000001 (US), SMS11000009091000001 (DC), SMS24000009091000001 (MD), SMS51000009091000001 (VA), SMU11479009091000001 (Washington MSA, not seasonally adjusted). Our arithmetic on the published series. January 2025 to August 2026 (US to September 2026).

Full size

Method. Percentage change in federal government employment from January 2025 to the latest month, with the levels in thousands. Every series counts all federal jobs, Postal Service included. State figures cover whole states — Baltimore’s Social Security Administration, Aberdeen, Norfolk’s Navy yards — so the three together (549.9 to 466.4 thousand, −83.5 thousand) are a larger universe than the MSA, and neither is the “DMV” of common speech.

The cut

The chronology: a government-wide buyout offer on January 28, 2025; terminations and reductions in force in February and March (nationally, 1,727 and 2,919 terminations, 1 and 24 RIFs, in OPM’s separations files); most buyout takers paid until September 30; a 43-day shutdown from October 1. The payroll data record the cut twice — a slow bleed through 2025 under a hiring freeze, then the October cliff.

Exhibit 14

Buyouts did thirteen times the cutting that layoffs did

Source. Partnership for Public Service, “The Federal Workforce One Year into the Trump Administration”, tabulating OPM Federal Workforce Data; the remainder is our arithmetic (386,826 − 136,822 − 10,436). January 20, 2025 – January 2026.

Full size

Method. Separations from the federal civilian workforce by reason, January 20, 2025 to January 2026, and new hires over the same period. The three separation rows partition the 386,826 separations: the 239,568 are all other separations — retirements, resignations, transfers, terminations, deaths and the rest — outside the program and RIFs; new hires are not separations. Reductions in force were 2.7 percent of separations. Four numbers circulate for “the buyout” and they are different things: about 75,000 acceptances in February 2025; 136,822 people who actually left under any buyout window; 386,826 separations for every reason; and a net change of roughly −264,000 after hiring. GAO-26-108583 gives 128,589 buyout and 6,331 RIF separations across 22 CFO Act agencies, January 2025 to January 2026.

Exhibit 15

Agency by agency: what agencies, auditors and OPM’s data report

Source. TIGTA 2025-IE-R027; VA, July 8, 2025; GAO-26-108155 via Government Executive; GAO-26-108916 (OPM); USAID from the Partnership’s OPM tabulation. 2025–26.

Full size

Method. Reductions stated by the agencies themselves or their auditors, and USAID’s from the Partnership’s tabulation of OPM data. The units differ — GSA’s is a headcount, the IRS figure is people separated or on a buyout as a share of the workforce (25,386, 25 percent, TIGTA), VA is a planned reduction reached by attrition after a canceled RIF — so the bars are each agency’s own claim, not one measure. Education’s and HHS’s stated reductions are not drawn: their releases and the press reports that give them could not be retrieved (What we could not verify).

Exhibit 16

The calendar of the cut

Source. OPM memoranda of February 4 and November 13, 2025; Federal Register, September 15, 2026; SpacePolicyOnline and CRA on the CR; BLS CES. January 2025 – December 2026.

Full size

Method. The sequence from the buyout offer to the FY2027 continuing resolution. Most buyout takers stayed on payroll until September 30, 2025; the shutdown began the next day; the ending legislation barred mass reductions in force until January 30, 2026; OPM’s rule weighting performance over tenure took effect September 2, 2026; and there is no shutdown as of October 2, 2026 — the CR, signed September 2, 2026 (P.L. 119-103), funds government to December 11. The vertical rule is October 2, 2026, the date of the study. Bars end at the close of the period shown; an open-ended rule is drawn to the end of 2026. The probationary terminations, the agency buyout windows and the 2026 lapses rest on agency releases or secondary sources and are listed under What we could not verify.

The money

What a federal job is worth to each economy is measured two ways: the payroll, which the Bureau of Economic Analysis records as federal civilian output, and the contracts, which USAspending records by recipient. They point at different states.

Exhibit 17

Source. BEA GDP by state, released September 30, 2026, via FRED (DCGOVFEDCIVNGSP ÷ DCNGSP, and the MD and VA equivalents); BEA, July 1, 2026. 2024 and 2025.

Full size

Method. Federal civilian government as a share of state GDP in 2024 and 2025, in millions of current dollars as the inputs show. The shares fell in all three because the rest of each economy grew; the dollar values barely moved (DC +1.0 percent, Maryland −1.2, Virginia −0.3) because government output is measured by compensation and most of the departing workers were paid through September 2025. The first quarter of 2026 is where the money shows: BEA reports DC federal civilian earnings down $1.1 billion at an annual rate, the leading contributor to a 1.5 percent fall in DC earnings.

Exhibit 18

Payroll in the District and Maryland, contracts in Virginia

Source. BEA, as Exhibit 17; USAspending state profiles (recipient/state/awards for DC, Maryland and Virginia, FY2025). 2025 (payroll) and FY2025 (contracts).

Full size

Method. Federal civilian payroll output in 2025 against federal contract obligations recorded to recipients in each state in FY2025. Virginia’s contract dollars are 2.7 times its payroll dollars ($120.6 billion ÷ $44.2 billion); the District’s and Maryland’s are below theirs. The two are not additive and contracts are counted where the recipient is registered, not where the work is done. FY2025 total prime awards fell in DC ($61.3 billion to $48.9 billion) and Maryland ($114.7 billion to $105.7 billion) and rose in Virginia ($189.1 billion to $214.2 billion).

Forecasts and the outturn

Exhibit 19

What was forecast, and what the payroll shows

Source. DC CFO, February 28, 2025 and June 30, 2026; Fuller Institute, GMU, as reported by GMU’s Schar School, January 2026; Oxford Economics as reported by CNN, March 16, 2025; BLS as Exhibit 11. 2025–26.

Full size

Method. Forecasts against outturns from the payroll data and official reporting. The DC CFO’s February 2025 letter projected 40,000 District federal jobs lost over the financial plan period; by June 2026 the CFO counted 28,000 — what the BLS District series shows — and called the trend in line with that forecast. Oxford Economics forecast 33,700 for the metro area in 2025; the MSA lost 50,400 from January to December 2025 (BLS, not seasonally adjusted, 375.8 thousand to 325.4 thousand), so the forecast was about two-thirds of the like-for-like outturn. The geographies differ: a “Washington” federal job loss can be 28,000, 64,000 or 83,500 depending on the boundary.

Exhibit 20

Eleven times the usual number of federal workers filed for unemployment in Maryland

Source. Maryland Department of Labor, high-frequency indicators, August 8, 2026. Weeks ending January 25, 2025 to August 8, 2026.

Full size

Method. Cumulative initial unemployment claims by federal employees (UCFE) from Maryland residents between January 19, 2025 and August 8, 2026, against what the state’s 2024 pace of 11 a week would have produced over the same 81 weeks (11 × 81 = 891); 10,132 ÷ 891 = 11.4. Maryland is the only state in the region publishing a federal-worker claims tracker.

Space and trains

Exhibit 21

Lease savings, claimed, and what GAO found

Source. GAO-26-108615 (August 2026); Federal News Network, December 2025. March 2025 – August 2026.

Full size

Method. GSA’s annual lease rent across all federal leases; the savings DOGE claimed in March 2025; the terminations GSA completed in 2025; and the total DOGE’s own Wall of Receipts showed for the 264 leases left after 415 were removed by October 5, 2025, as GAO found (August 2026). GAO did not verify that figure, and found that the terminations of 108 of the 264 leases, about $15.3 million of it, were already under way before DOGE was established.

Exhibit 22

The trains filled while the payroll emptied

Source. WMATA, FY2025 results; WMATA, FY2026 year-end results, September 10, 2026; BLS as Exhibit 12. FY2025 and FY2026; payroll January 2025 to August 2026.

Full size

Method. Metro trips, rail and bus, rose 9 percent in FY2025 (264 million); Metrorail ridership rose 9 percent in FY2026, while total trips rose from 264 million to 268 million (1.5 percent). The region’s federal payroll fell 17 percent.

What we could not verify

The first-round buyout acceptance figure of “approximately 75,000” is an OPM spokesperson quoted by NBC; no OPM release was found.
Shutdown furlough totals and the 2026 lapse dates (January 31 – February 3, 2026; for DHS, February 14 – April 30, 2026) rest on secondary sources; the CRS report that would settle them (R48832) blocks automated reading. They are not in Exhibit 16.
The probationary terminations and first RIF notices (February – March 2025) and the agency buyout windows and RIF announcements (March – July 2025) rest on agency releases that were not retrieved; they are not in Exhibit 16.
Brokerage tallies of DOGE lease postings in the region (29 leases, 1.75 million square feet) pre-date GAO’s corrections and are not used in a chart.
Education’s reduction (4,133 to about 2,183, March 11, 2025) and HHS’s (82,000 to 62,000 full-time employees, March 27, 2025, which may be FTE rather than headcount) are not in Exhibit 15: both releases refused automated requests on October 5, 2026, and the press reports that give the figures (WBNG and Michigan Public; ABC and Winston & Strawn) were seen only as search excerpts.
The USAspending state endpoint returns a “year: 2017” label and a 2017 population in its metadata regardless of the fiscal year requested, though the dollar amounts respond correctly; the contract figures should be re-pulled from the web interface before any republication.
Monthly OPM Federal Workforce Data files through July 2026 are parquet files of more than two million rows; the national page counts duty stations by county from them (Exhibit 10), for the records with a published location only.
BEA’s federal civilian wages as a share of personal income (SAINC7) was not retrieved; the GDP share in Exhibit 17 is the substitute and is labeled as such. Metro’s monthly ridership for 2024–26 was not scraped; fiscal-year totals are used.
Nationally, UCFE initial claims are reported at 1,580 in the week ending March 1, 2025, against 393 a year earlier, and at 404 by June 2026 (DOL); DOL’s releases refused automated requests on October 5, 2026, and only the 1,580 was corroborated, by a search excerpt.
The Fuller Institute’s report (January 2026) is cited through GMU’s Schar School announcement; the report’s own PDF link no longer resolves. Oxford Economics’ forecast is cited through a syndicated copy of CNN’s report, and CRS R47716 through a mirror.

Sources

Corrections

The District’s chief financial officer forecast 40,000 over its plan period and later called the trend on track; the claims that it “did not arrive” and that “every forecast was wrong” are withdrawn.
The 33,700 forecast for 2025 is compared with the region’s 50,400 for January–December 2025 (about two-thirds), not with the twenty-month outturn.
$53.5 million is DOGE’s own reported total after removals, not savings GAO verified.
Metro trips, rail and bus, rose 9% in FY2025 and Metrorail ridership 9% in FY2026; “under return-to-office orders” is withdrawn.
The last month as low as September 2026 was May 1966.
Maryland’s UCFE comparison runs 81 weeks (891 claims at the 2024 pace), not 82 (902).
The buyout remainder is all other separations, outside the program and reductions in force, and the 108 leases’ terminations were under way before DOGE.

Data for every exhibit: download (CSV)

Cite this study

Higgins Strategy. “Every Federal Job in the Region” Study VIII, The Federal Dollars Series. October 12, 2026. https://higginsstrategy.com/studies/every-federal-job-in-the-region/

@techreport{higginsstrategy2026everyfederaljobintheregion,
  author      = {{Higgins Strategy}},
  title       = {{Every Federal Job in the Region}},
  institution = {Higgins Strategy},
  type        = {Study},
  number      = {VIII},
  series      = {The Federal Dollars Series},
  year        = {2026},
  url         = {https://higginsstrategy.com/studies/every-federal-job-in-the-region/},
  month       = oct
}

Next: Study IX, Every Dollar of War

Every Federal Job in the Region

The short version

The count

The cut

The money

Forecasts and the outturn

Space and trains

What we could not verify

Sources

Read next