Skip to content
NewsroomMarkets

August 3, 2026, 7:38 PM · Data Story · 10 min read

Treasury's $87 Billion Revision Ranks Third in Dollars, and Fourth or Fifth After Scaling

Treasury raised its July-September borrowing estimate by $68 billion and said the increase was $87 billion after excluding a $19 billion opening-cash offset. Across 12 comparable updates since 2015, $87 billion ranks third as a signed dollar revision, fifth as a share of prior financing need, and fourth against a like-for-like cash-adjusted funding gap.

By Cumulant Research

Hover or tap an underlined term to see its definition.

The neoclassical U.S. Treasury Building in Washington, viewed from the side with the Washington Monument in the background.
The U.S. Treasury Building in Washington, where the department released its revised July-September borrowing estimate. Photo: MeanieHyaena, CC BY 4.0, via Wikimedia Commons

The quick version

  • Treasury raised projected July-September borrowing from $671 billion to $739 billion.
  • A $19 billion higher opening cash balance offset part of the increase, so the published revision was $68 billion while the cash-adjusted revision was $87 billion.
  • The $87 billion increase ranks third among 12 signed July-September revisions from 2015 through 2026.
  • It equals 15.6 percent of the prior $557 billion financing need, which ranks fifth.
  • It equals 14.0 percent of the prior $621 billion cash-adjusted funding gap, which ranks fourth. The denominator changes the answer.

Figure

One revision, three rankings

The position of 2026 among 12 signed July-September forecast revisions from 2015 through 2026

Measure2026 calculation2026 valueRank
Signed dollar revision$87 billion$87 billion3 of 12
Share of prior financing need$87B / $557B15.6%5 of 12
Share of prior cash-adjusted gap$87B / $621B14.0%4 of 12

Ranks order signed revisions from the largest upward change to the largest downward change. The percentage rows use different denominators, neither of which is an official Treasury ranking.

Source: https://home.treasury.gov/policy-issues/financing-the-government/quarterly-refunding/quarterly-refunding-archives/quarterly-refunding-financing-estimates-by-calendar-year · $ billions and percent · 2015-2026

Why it matters

Treasury's financing estimate helps bond investors gauge the amount of government debt the market may need to absorb during the quarter. The analysis shows that the latest revision is historically large in dollars, but less exceptional after scaling, and that different defensible baselines produce different rankings. The article does not claim that the revision caused a market move or changed economic output.

By Cumulant Research

TreasuryTreasuryThe U.S. Department of the Treasury, which manages federal finances and borrows funds needed to operate the federal government. published two correct figures on August 3. Its July-September estimate for privately held net marketable borrowingprivately held net marketable borrowingThe net amount Treasury expects to raise through tradable securities after excluding Federal Reserve auction rollovers but including financing caused by Federal Reserve redemptions. rose from $671 billion in May to $739 billion, a $68 billion increase. Treasury also said the increase was $87 billion after excluding the quarter's higher-than-assumed opening cash balanceopening cash balanceThe cash Treasury has at the beginning of a quarter.. Nothing was concealed, and the two figures answer different accounting questions. Sources: https://home.treasury.gov/news/press-releases/sb0584 and https://home.treasury.gov/system/files/136/Sources-Uses-Public-Table-August-2026.pdf

The narrow answer is third in signed dollars, fifth against prior financing need, and fourth against a like-for-like cash-adjusted funding gap.

We reconstructed one comparable July-September update for every year from 2015 through 2026. The $87 billion increase ranks third among those 12 signed dollar revisions. Its scaled rank is not unique because a percentage needs a denominatordenominatorThe baseline amount used underneath a fraction, such as the earlier financing forecast against which a revision is compared., and two reasonable denominators produce different answers. Treasury's calendar-year archive links the underlying releases: https://home.treasury.gov/policy-issues/financing-the-government/quarterly-refunding/quarterly-refunding-archives/quarterly-refunding-financing-estimates-by-calendar-year

Figure

One revision, three rankings

The position of 2026 among 12 signed July-September forecast revisions from 2015 through 2026

Measure2026 calculation2026 valueRank
Signed dollar revision$87 billion$87 billion3 of 12
Share of prior financing need$87B / $557B15.6%5 of 12
Share of prior cash-adjusted gap$87B / $621B14.0%4 of 12

Ranks order signed revisions from the largest upward change to the largest downward change. The percentage rows use different denominators, neither of which is an official Treasury ranking.

Source: https://home.treasury.gov/policy-issues/financing-the-government/quarterly-refunding/quarterly-refunding-archives/quarterly-refunding-financing-estimates-by-calendar-year · $ billions and percent · 2015-2026

Two revisions, not a contradiction

The published borrowing estimate increased by $68 billion. That is the direct difference between $739 billion and $671 billion. The larger $87 billion figure removes a separate cash advantage: Treasury entered the quarter with $919 billion rather than the $900 billion previously assumed. The planned September cash balance remained $950 billion. Source: https://home.treasury.gov/system/files/136/Sources-Uses-Public-Table-August-2026.pdf

Checking-account analogy

Starting a month with more cash than expected lets a household borrow less even when the month's underlying funding gap has widened. Treasury's two figures reflect that same distinction.

Figure

How the $87 billion cash-adjusted change became a $68 billion published revision

May and August estimates for July-September 2026

StepMay 2026Aug. 2026Change
Financing need55763376
All other sources-64-75-11
Cash-adjusted funding gap62170887
Opening cash90091919
Cash effect on borrowing5031-19
Published borrowing estimate67173968

Cash-adjusted funding gap equals financing need minus all other financing sources. Published borrowing equals that gap plus the cash effect on borrowing. Parentheses in Treasury's table are represented here as negative values.

Source: https://home.treasury.gov/system/files/136/Sources-Uses-Public-Table-August-2026.pdf · $ billions · July-September 2026 forecast

Reconstructing the $87 billion

Treasury's projected financing needfinancing needTreasury's estimate of the cash federal operations require before accounting for marketable borrowing, other financing sources and changes in the government's cash balance. rose from $557 billion to $633 billion, an increase of $76 billion. Its estimate for all other financing sourcesall other financing sourcesTreasury accounting items other than marketable borrowing that can either provide cash or increase the amount that must be borrowed. moved from negative $64 billion to negative $75 billion. Because the second number became $11 billion more negative, those other sources left $11 billion more for marketable borrowing to cover. Source: https://home.treasury.gov/system/files/136/Sources-Uses-Public-Table-August-2026.pdf

The calculation is therefore $76 billion minus negative $11 billion, which equals $87 billion. In equivalent form, the cash-adjusted funding gapcash-adjusted funding gapAn analytical measure calculated as financing need minus all other financing sources, before adding the effect of a planned cash-balance change. rose from $621 billion to $708 billion. Treasury's extra $19 billion of opening cash then reduced the change in the published borrowing estimate from $87 billion to $68 billion. Source: https://home.treasury.gov/system/files/136/Sources-Uses-Public-Table-August-2026.pdf

Building a comparable history

For each year, we selected Treasury's initial July-September forecast published around late April or early May and its update published around late July or early August. We then applied the same formula: change in financing need minus change in all other financing sources. The result removes the effect of changed cash-balance assumptions. Source archive: https://home.treasury.gov/policy-issues/financing-the-government/quarterly-refunding/quarterly-refunding-archives/quarterly-refunding-financing-estimates-by-calendar-year

Treasury reports these tables in rounded billions. For 2016 and 2019, subtracting the displayed endpoints does not exactly reproduce Treasury's displayed revision row. We therefore used Treasury's revision row when it was available, rather than manufacturing precision that the published table does not contain. Sources: https://home.treasury.gov/system/files/136/archive-documents/sources-and-uses-aug-2016.pdf and https://home.treasury.gov/system/files/136/SourcesUsesJuly2019Public07-29-19.pdf

Figure

Cash-adjusted revisions, 2015-2020

Two percentage baselines show why the scaled ranking is not unique

YearCash-adjusted revisionRevision / prior needRevision / prior gap
201575.3%4.7%
2016115.2%5.4%
20173320.4%17.9%
20182914.4%10.2%
201931.2%0.9%
20201192179.8%176.1%

Revision divided by prior need uses the financing-need column alone. Revision divided by prior gap uses prior financing need minus prior all-other-sources. Values are rounded to one decimal place.

Source: https://home.treasury.gov/system/files/136/archive-documents/Sources-and-Uses---2015_May.pdf | https://home.treasury.gov/system/files/136/archive-documents/Sources-and-Uses-November-2015.pdf | https://home.treasury.gov/system/files/136/archive-documents/sources-and-uses-aug-2016.pdf | https://home.treasury.gov/system/files/136/archive-documents/Sources-and-Uses-Public-July-2017-Final.pdf | https://home.treasury.gov/system/files/136/SUApr2018Final.pdf | https://home.treasury.gov/system/files/136/Sources%20and%20Uses%20Oct%202018%20Public.pdf | https://home.treasury.gov/system/files/136/SourcesUsesJuly2019Public07-29-19.pdf | https://home.treasury.gov/system/files/136/SU_Jul2020.pdf · $ billions and percent · Successive July-September forecasts

Figure

Cash-adjusted revisions, 2021-2026

The 2026 observation is compared with the preceding 11 years

YearCash-adjusted revisionRevision / prior needRevision / prior gap
2021-96-11.2%-11.0%
202224478.5%73.5%
20237717.6%11.3%
2024-79-16.3%-10.6%
20256012.5%10.8%
20268715.6%14.0%

Negative revisions indicate that the cash-adjusted forecast moved down. Values are rounded to one decimal place.

Source: https://home.treasury.gov/system/files/136/Sources-and-Uses-July-2021.pdf | https://home.treasury.gov/system/files/136/SourcesandUsesJuly2022.pdf | https://home.treasury.gov/system/files/136/SourcesUsesJul2023.pdf | https://home.treasury.gov/system/files/136/Sources_Uses.July_2024.pdf | https://home.treasury.gov/system/files/136/Sources-and-Uses-Table-July-2025.pdf | https://home.treasury.gov/system/files/136/Sources-Uses-Public-Table-August-2026.pdf · $ billions and percent · Successive July-September forecasts

Why fifth is true but incomplete

Dividing $87 billion by the prior $557 billion financing need gives 15.6 percent. Four observations are higher on that measure: 2020 at 179.8 percent, 2022 at 78.5 percent, 2017 at 20.4 percent and 2023 at 17.6 percent. That places 2026 fifth. The annual values come from the Treasury tables linked in the two historical charts.

The numerator, however, combines changes in financing need and all other financing sources, while the $557 billion denominator includes financing need alone. A more internally matched baseline subtracts prior all-other-sources from prior financing need. For 2026, that gives $621 billion: $557 billion minus negative $64 billion. Dividing $87 billion by $621 billion gives 14.0 percent and a fourth-place rank. Source: https://home.treasury.gov/system/files/136/Sources-Uses-Public-Table-August-2026.pdf

Why the denominator matters

A percentage is not an intrinsic property of the revision. It describes the revision relative to a selected baseline. Changing that baseline can change the historical ordering even when the $87 billion numerator never moves.

Neither percentage is an official Treasury measure. The prior-financing-need ratio answers how large the revision was relative to the government's projected cash requirement. The prior-gap ratio answers how large it was relative to the cash-adjusted amount that market borrowing needed to cover. Reporting both is more accurate than calling either one the definitive scaled rank.

The two larger revisions were unusual

The 2020 cash-adjusted revisioncash-adjusted revisionThe change in the borrowing forecast after removing the effect of a changed opening or closing cash balance. was $1.192 trillion. Treasury said that forecast assumed $1 trillion of additional borrowing need in anticipation of legislation responding to the COVID-19 outbreak. Source: https://home.treasury.gov/news/press-releases/sm1077

The 2022 cash-adjusted revision was $244 billion. Treasury attributed the published borrowing change primarily to revised projections of fiscal activity and an estimated $120 billion of Federal ReserveFederal ReserveThe central bank of the United States. System Open Market AccountSystem Open Market AccountThe Federal Reserve portfolio that holds Treasury securities and other assets used in monetary-policy operations. redemptions. Source: https://home.treasury.gov/news/press-releases/jy0902

Those episodes are not routine benchmarks, but removing them after seeing the results would make the sample depend on the desired conclusion. We retain every selected year in the primary ranking and describe the exceptional context instead.

What the release does not show

Treasury said the 2026 increase was primarily due to lower projected net cash flows, partially offset by the higher opening cash balance. The August release and reconciliation table do not break that lower net cash flownet cash flowCash coming into the government minus cash going out during a period. into separate changes in receipts, outlays or payment timing. Source: https://home.treasury.gov/news/press-releases/sb0584

The data set used here contains cash forecasts and financing-accounting entries. It contains no Treasury yield, security-price, economic-output or employment series. This article therefore makes no claim about the bond market's reaction or the revision's economic effect.

The ranking is descriptive rather than causal. It shows where one forecast change sits in a small historical sample. It does not show why federal cash flows changed, whether the forecast will be accurate, or how investors will respond.

The answer

After removing the opening-cash effect, Treasury's July-September 2026 borrowing revision is $87 billion. It is the third-largest signed cash-adjusted dollar revision among the 12 observations from 2015 through 2026. It ranks fifth when divided by prior financing need and fourth when divided by the prior cash-adjusted funding gap. The dollar result is stable within this sample; the scaled rank depends on the denominator.

What to watch

  • Treasury's additional quarterly-refunding details scheduled for August 5, 2026. https://home.treasury.gov/news/press-releases/sb0584
  • Whether subsequent cash-flow data explain the lower projected net cash flows behind the revision.
  • Whether actual July-September borrowing and the quarter-end cash balance match Treasury's forecasts.
  • How future revisions rank when the same cash-adjusted methodology is applied consistently.

How we did this

  • We verified the August 3, 2026 event against Treasury's press release and its one-page sources-and-uses reconciliation table.
  • We selected one successive July-September forecast comparison for every year from 2015 through 2026, producing 12 observations.
  • We converted values shown in parentheses in Treasury's tables to negative numbers.
  • For each year, cash-adjusted revision equals the change in financing need minus the change in all other financing sources.
  • The first scale measure divides the revision by the prior financing-need forecast.
  • The second scale measure divides the revision by the prior cash-adjusted funding gap, calculated as prior financing need minus prior all-other-sources.
  • We ranked signed revisions from the largest upward change to the largest downward change. We did not rank absolute forecast errors.
  • Treasury tables are rounded to billions. Where Treasury displayed an explicit revision row, including in 2016 and 2019, we used that row rather than differences that can be distorted by rounding.
  • We retained the 2020 pandemic and 2022 Federal Reserve-redemption observations in the primary sample and identified their exceptional context separately.
  • We did not use market-price or macroeconomic data and drew no conclusion about market reaction, economic effect or causation.

What this cannot establish

  • The sample contains only 12 observations and should not be treated as a stable statistical distribution.
  • The comparison covers July-September updates only. Results may differ for other quarters.
  • The rankings preserve the direction of each revision. Ranking absolute forecast errors would answer a different question.
  • The scaled rank changes with the denominator: fifth against prior financing need and fourth against the prior cash-adjusted funding gap.
  • Dollar rankings use nominal amounts and do not adjust earlier observations for inflation.
  • The 2020 pandemic assumption and 2022 Federal Reserve redemptions make those years unusual, but they remain in the primary sample.
  • Treasury's tables are rounded to the nearest billion, so displayed endpoints can differ slightly from displayed revision rows.
  • Treasury attributed the 2026 change broadly to lower projected net cash flows but did not publish a receipts-versus-outlays decomposition in the cited release and table.
  • The analysis contains no bond-yield, security-price or macroeconomic outcome data.

This is AI-assisted analysis under stated assumptions; it is not investment advice or a price target. Figures are as of the publication date and trace to the cited sources; markets and disclosures change.

Sources

  1. 01Treasury Announces Marketable Borrowing Estimates, U.S. Department of the TreasuryPrimary
  2. 02Sources and Uses Reconciliation Table, August 2026, U.S. Department of the TreasuryData
  3. 03Treasury Announces Marketable Borrowing Estimates, May 2026, U.S. Department of the TreasuryPrimary
  4. 04Sources and Uses Reconciliation Table, May 2026, U.S. Department of the TreasuryData
  5. 05Quarterly Refunding Financing Estimates by Calendar Year, U.S. Department of the TreasuryData
  6. 06Sources and Uses Reconciliation Table, May 2015, U.S. Department of the TreasuryData
  7. 07Sources and Uses Reconciliation Table, November 2015, U.S. Department of the TreasuryData
  8. 08Sources and Uses Reconciliation Table, August 2016, U.S. Department of the TreasuryData
  9. 09Sources and Uses Reconciliation Table, July 2017, U.S. Department of the TreasuryData
  10. 10Sources and Uses Reconciliation Table, April 2018, U.S. Department of the TreasuryData
  11. 11Sources and Uses Reconciliation Table, October 2018, U.S. Department of the TreasuryData
  12. 12Sources and Uses Reconciliation Table, July 2019, U.S. Department of the TreasuryData
  13. 13Sources and Uses Reconciliation Table, July 2020, U.S. Department of the TreasuryData
  14. 14Treasury Announces Marketable Borrowing Estimates, August 2020, U.S. Department of the TreasuryPrimary
  15. 15Sources and Uses Reconciliation Table, July 2021, U.S. Department of the TreasuryData
  16. 16Sources and Uses Reconciliation Table, July 2022, U.S. Department of the TreasuryData
  17. 17Treasury Announces Marketable Borrowing Estimates, August 2022, U.S. Department of the TreasuryPrimary
  18. 18Sources and Uses Reconciliation Table, July 2023, U.S. Department of the TreasuryData
  19. 19Sources and Uses Reconciliation Table, July 2024, U.S. Department of the TreasuryData
  20. 20Sources and Uses Reconciliation Table, July 2025, U.S. Department of the TreasuryData
  21. 21Role of the Treasury, U.S. Department of the TreasuryPrimary
  22. 22Financing the Government, U.S. Department of the TreasuryPrimary
U.S. Treasuryfederal borrowingdebt managementfiscal policyTreasury marketdata journalismtreasury-borrowingquarterly-refundingUnited States

Related

Markets

ASC 842 Study Finds More Reported Non-Lease Leverage, Not Proof That Total Debt Fell

A new Federal Reserve working paper estimates that companies strongly affected by lease recognition reported a 3.0 percentage-point increase in non-lease debt relative to assets, while indicators of borrowing between reporting dates declined. The pattern is consistent with debt becoming more visible, but the study does not observe the proposed short-lived loans or establish that companies reduced their total economic debt. [Federal Reserve paper](https://www.federalreserve.gov/econres/ifdp/files/ifdp1441.pdf)

Glass-and-metal Pinnacle office tower rising above downtown Nashville.
Markets

The Record JGB Auction That Was Really the Sell-Off Wearing a Disguise

On 7 July 2026, Japan sold a 30-year bond with its first-ever 4% coupon to the strongest demand since 2019, in the middle of a bond-market rout. Read who actually bought and what they did next, and the record demand turns out to be the same fact as the record yield: a bond cheap enough to flip, not proof anyone trusts the fiscal path.

The Record JGB Auction That Was Really the Sell-Off Wearing a Disguise
Markets

Quantinuum's IPO Implies Triple-Digit Revenue Growth For Five Years. Its Signed-Order Backlog Is $77 Million.

At $60 a share, the largest stock-market debut by a company that does nothing but quantum computing is valued at roughly $15.7 billion, about 507 times last year's sales. Work the math backwards and, under a generous set of assumptions, that price only makes sense if sales grow about 119 percent a year through 2030 (in plain terms, more than doubling every single year). The orders Quantinuum has actually signed, as disclosed in its own filing with regulators, would cover only a small slice of that climb. The same test flags Quantinuum as the company with the thinnest cushion of signed orders relative to its price among similar firms, but that is a softer criticism than it first looks, because the entire quantum-computing sector is priced far beyond the orders it has booked.

Close-up of a NIST trapped-ion quantum computing apparatus showing a gold ion-trap chip surrounded by copper enclosure and a gold wire mesh used to manipulate beryllium ions.
Markets

The Shock Absorber That Made the Last Treasury Cash Rebuild Painless Is Empty. The Fed Built a New One, and It Works Differently Than People Think.

The Treasury is refilling its checking account at the Fed this quarter, and in 2023 a $2 trillion-plus pile of parked cash quietly absorbed almost the entire drain so bank reserves barely moved. That pile is now close to zero, but the Fed stopped shrinking its balance sheet in December 2025 and began buying Treasury bills to keep reserves topped up, so the real question is not whether reserves get hit but whether an active, flow-based tool can lean against a quarter-long refill the way a passive stock of idle cash once did. On the numbers, this looks like a manageable test the Fed most likely passes, not a cliff.

The neoclassical white marble facade of the Federal Reserve's Eccles Building in Washington, D.C., seen from the street under a clear sky.