US Debt Ceiling: How TGA Drawdowns Inject Trillions in Liquidity
A weekly dataset of the US Treasury General Account since 2002, with episode tagging for every debt ceiling standoff. The seven episodes since 2015 drained $2.9 trillion of cash into the financial system, $2.0 trillion of it in the three since 2021, at a monthly pace comparable to the Fed’s QE1 to QE3 programs.
The seven debt ceiling episodes since 2015 have drawn down the Treasury General Account by $2.9 trillion in total; the three since 2021 account for $2.0 trillion, released at a monthly pace comparable to the Fed’s first three quantitative easing (QE) programs.
A weekly dataset of the Treasury’s cash balance at the Federal Reserve since 2002, with episode tagging for each debt ceiling standoff and quantified drawdown magnitudes. The nine standoffs since 2011 have released a total of $3.0 trillion into the financial system, and 95.1% of that total came from the seven episodes since 2015, the year the Treasury changed its cash balance policy.
The US Treasury General Account (TGA) is the federal government’s operational cash account held at the Federal Reserve. Since 2002, the TGA balance has ranged from $4 billion (October 2003) to $1,817 billion (July 2020), with a distribution that shifted profoundly after 2015, when the Treasury adopted a policy of holding higher precautionary cash buffers. Across the nine debt ceiling episodes since 2011, TGA drawdowns have released liquidity into the private financial system in amounts ranging from $70 billion in 2013 to $876 billion in 2021. This page provides the complete weekly series, episode-level drawdown statistics, and regime classifications for reproducible analysis.
The seven debt ceiling episodes since 2015 drew down the Treasury General Account by $2.9 trillion in total. The three since 2021 (2021, 2023, 2025) account for $2.0 trillion, at a monthly pace comparable to or above the Fed’s QE1 to QE3 programs, though far below the Fed’s purchases of March to June 2020. The two pre-2015 episodes produced drawdowns of just $77 billion and $70 billion, 19.5× smaller in aggregate than the seven since 2015, because the Treasury’s balance at the Fed then stayed mostly below $100 billion. Splitting the 2013 episode at its suspension raises the pre-2015 total to $250 billion and lowers the ratio to 11.5×. Note: this dataset measures gross TGA flows; it does not capture the reserve creation mechanism that distinguishes QE from cash account redistribution (see Methodology and Limitations).
Current TGA balance
4-week change
Historical percentile
Current regime
- The seven debt ceiling episodes since 2015 drew down the Treasury General Account by $2,865 billion in aggregate; the three since 2021 account for $1,998 billion, released at a monthly pace between $82 billion and $166 billion, against average paces of $75 billion to $108 billion for the Fed’s QE1 to QE3 programs.
- The 2021 episode drained the TGA at a pace of $166 billion per month, 54% faster than QE1’s average pace of $108 billion; the Fed’s securities purchases of March to June 2020 ran at $699 billion per month, 4.2 times the 2021 pace.
- The two pre-2015 episodes were 19.5× smaller in aggregate than the seven since 2015 ($147 billion combined vs. $2,865 billion), not because the political dynamics differed, but because the Treasury’s balance at the Fed stayed below $100 billion in 87% of weeks from 2009 to 2014, before its May 2015 cash balance policy.
- The TGA was above $500 billion in 40.4% of weeks after January 2015, versus 0.0% of weeks before that date, a structural break that follows the cash balance policy announced in the Treasury’s quarterly refunding statement of May 6, 2015.
- As of April 15, 2026, the TGA stands at $751 billion, the 88th historical percentile, with a 4-week drawdown of $102 billion; the classifier places this observation in the “Drawdown” regime, outside any debt ceiling episode.
- The dataset contains 1,218 weekly observations with a proprietary episode-tagging column and a TGA regime classifier (Accumulation / Stable / Drawdown / Post-ceiling rebuild) not available in public sources.
1,218 observations · Weekly · 2002-12-18 – 2026-04-15 · Study file not offered for download ·
Methodology ·
Cite this dataset
All-time peak (Jul. 2020)
All-time low (Oct. 2003)
Largest drawdown (2021)
Fastest drawdown pace (2021)
Post-2015 vs pre-2015 total
Weekly observations
Chart: Treasury General Account balance, 2002–2026
TGA balance with the nine debt ceiling episodes shaded
The pre-2015 episodes (2011, 2013) are barely visible on the chart; the four episodes of 2015 to 2019 are visible but modest; the three since 2021 dominate it.
The visual story is the 2015 structural break. Before that year, the TGA functioned as a checking account: small, volatile, mechanically tied to tax receipts and daily outlays. After 2015, it became a precautionary reserve, large enough that each of the seven debt ceiling episodes since then produced a drawdown of $120 billion or more, and each of the three since 2021 more than $500 billion.
Sources: Federal Reserve Economic Data (FRED, series WTREGEN, weekly averages for the week ending Wednesday). Chart: Eco3min Research.
How to read this chart
The vertical axis shows the Treasury General Account balance in billions of dollars, and the horizontal axis spans December 2002 to April 2026. Each shaded vertical band marks a period during which the federal debt ceiling was binding and the Treasury was drawing down its cash buffer to meet obligations. The dashed grey line in May 2015 marks the cash balance policy announced by the Treasury in its quarterly refunding statement of May 6, 2015: hold a level of cash generally sufficient to cover one week of outflows, subject to a minimum balance of roughly $150 billion.
The effect of that policy is visible in the post-2015 baseline level of the series. Where the TGA had previously oscillated between single-digit and low-hundreds of billions, it has since ranged from $30 billion (November 2015) to $1,817 billion (July 2020), with a median of $377 billion. This matters because it determines how much liquidity can be released during a debt ceiling episode, the precise variable this page is built around. For broader context on the liquidity environment, see our Net Liquidity Index and Fed Balance Sheet datasets.
The pre-2015 context: why earlier episodes barely register
The dominant narrative around debt ceiling standoffs frames them as primarily political events, high-stakes negotiations between Congress and the executive over budget authority, with market effects mediated by uncertainty, sovereign rating downgrade risk, and volatility. In that reading, the 2011 episode is the reference point: Standard & Poor’s stripped the United States of its AAA rating on August 5, 2011, and the S&P 500 fell 6.7% the following Monday.
The TGA data tell a parallel story that political analysis does not capture. Between the May 2011 onset of the standoff and its early-August resolution, the Treasury drew down its General Account from $115 billion (May 4) to $38 billion (June 8), a net release of $77 billion into the financial system. The 2013 standoff, which culminated in the October government shutdown, produced a $70 billion drawdown. In aggregate, the two pre-2015 episodes released $147 billion, 19.5 times less than the $2,865 billion released by the seven episodes since 2015.
The reason is not a shift in political mechanics; the legal architecture of the debt ceiling is substantially the same today as it was in 2011. The reason is that the Treasury changed how it manages operational cash in the years following 2011. Before the 2008 crisis, it targeted a balance of about $5 billion at the Fed and placed surplus cash in Treasury Tax and Loan accounts at commercial banks, as the Federal Reserve Bank of New York documents (weekly average of $5.3 billion from December 2002 to September 2008). From 2009 to 2014 the balance ran mostly between $26 billion and $108 billion (10th and 90th percentiles), and the Treasury relied on its ability to issue short-dated Treasury bills at any time to cover cash shortfalls. The 2011 standoff highlighted the risk of this setup: if the Treasury loses access to issuance (because it cannot issue new debt without breaching the ceiling), a thin cushion provides only a few days of runway.
Structural break in May 2015. On May 6, 2015, the Treasury announced in its quarterly refunding statement that it would hold a level of cash generally sufficient to cover one week of outflows, subject to a minimum balance of roughly $150 billion, to protect against a loss of market access. The Treasury Borrowing Advisory Committee had recommended a larger buffer, on the order of $500 billion, in 2014 and in February 2015. That policy is the most consequential regime change in TGA behavior since the start of the series. The TGA was above $500 billion in 0.0% of weeks before January 2015, versus 40.4% of weeks after. Any comparison between pre- and post-2015 debt ceiling episodes must account for this mechanical difference in the amount of cash available to draw down. A debt-ceiling episode is also a different object from a lapse in appropriations, which halts discretionary spending without touching debt service: the register of every US government shutdown since 1976 runs on a separate clock.
What this dataset does not measure. The TGA balance measures a gross cash position: the Treasury’s operational account at the Fed. It does not measure net fiscal impulse to the economy, reserve creation through asset purchases (as QE does), or a direct policy action. The reserve creation mechanism that distinguishes central bank asset purchases from Treasury cash redistribution is a material economic difference, discussed in the QE vs TGA drawdown section. Between the 2021 pre-episode peak (May 5) and trough (October 13), bank reserves at the Fed (FRED WRESBAL, H.4.1) rose by $321 billion, 37% of the $876 billion TGA drawdown over the same weeks, while the overnight reverse repo facility grew from $163 billion to $1,365 billion (Federal Reserve Bank of New York, via FRED RRPONTSYD). Readers using this dataset to infer market effects should consult the related work on Net Liquidity and the Overnight Reverse Repo, which together capture the broader monetary plumbing context.
The role of the debt ceiling as a liquidity event is a post-2015 phenomenon. The underlying mechanism (the Treasury drawing down cash to meet obligations while it cannot issue new debt) has always existed. What changed is the starting level.
How the drawdown mechanism works: from debt ceiling to market liquidity
The mechanism by which a debt ceiling episode produces a liquidity flow to the private sector is a chain of accounting entries, not a public policy decision. Understanding that chain is essential to interpreting the dataset correctly.
When the statutory debt ceiling is reached, the Treasury loses the ability to issue net new debt. Because federal spending continues on its normal schedule (Social Security payments, federal payroll, interest on existing debt, Medicaid transfers) the Treasury must fund these outflows with cash on hand. That “cash on hand” is the General Account at the Federal Reserve. As outlays are made, the TGA balance falls. The wider context: Full benefits through 2034, then 81%.
Each dollar of TGA drawdown is, in accounting terms, a transfer from the Treasury’s account at the Fed to a private-sector recipient: a retiree’s bank account, a federal contractor, a state Medicaid agency, a bondholder. Once the transfer settles, the recipient’s bank holds a reserve claim against the Fed that did not exist (on the private side) before. In aggregate, TGA drawdowns increase the stock of bank reserves by the amount drawn down, all else equal. This is the “liquidity” effect the dataset quantifies.
Two boundary conditions matter. First, the effect is mechanical only if “all else” actually remains constant. In practice, the Treasury typically offsets part of the drawdown by reducing Treasury bill issuance relative to baseline, which partially neutralizes the effect. Second, the flow is time-limited: once the ceiling is raised or suspended, the Treasury rebuilds the TGA by issuing bills, which withdraws reserves from the private sector. The post-ceiling rebuild is the mirror image of the drawdown. For more context on how these flows translate into observable market conditions, see our work on high yield spreads and the Financial Conditions Index.
A debt ceiling episode converts stored Treasury cash into circulating bank reserves at a pace determined by the spending schedule and the TGA starting level; the subsequent resolution reverses the flow as the Treasury rebuilds its cash buffer.
Point 3: a legitimate analytical qualification. Framing TGA drawdowns as “stealth liquidity” requires careful delimitation. Primary dealers, bank treasuries, and money market funds track TGA balances in real time via the Daily Treasury Statement and explicitly model drawdown trajectories. Sophisticated institutional participants do not experience these flows as hidden; they price them in advance, often weeks before drawdown intensification. What can fairly be called “stealth” is the absence of public discussion: these flows are not debated at the FOMC, presented in press conferences, or catalogued in Fed balance sheet commentary. But claiming these events are unknown to institutional participants would be overstating the case. They are known to the market plumbing; they are under-discussed in the financial press. The distinction matters for interpreting the dataset correctly.
The nine episodes in detail
The table below presents the pre-episode peak, intra-episode trough, and drawdown magnitude for each US debt ceiling episode since 2011 that produced a measurable TGA drawdown. Values are computed row-by-row from the CSV: the pre-episode peak is the maximum weekly TGA balance in the 90 days before the start date; the episode trough is the minimum value observed within the episode window.
| Episode | Start | Resolution | Pre-episode peak | Episode trough | Drawdown | Duration | Pace ($bn/month) |
|---|---|---|---|---|---|---|---|
| 2011 standoff | 2011-05-16 | 2011-08-02 | $115bn | $38bn | $77bn | 11 weeks | $67bn |
| 2013 standoff | 2013-01-01 | 2013-10-17 | $83bn | $14bn | $70bn | 42 weeks | $9bn |
| 2015 standoff | 2015-03-16 | 2015-11-02 | $201bn | $47bn | $154bn | 33 weeks | $17bn |
| 2017 standoff | 2017-03-16 | 2017-09-08 | $391bn | $38bn | $353bn | 25 weeks | $48bn |
| 2017–2018 standoff | 2017-12-09 | 2018-02-09 | $196bn | $76bn | $120bn | 9 weeks | $75bn |
| 2019 standoff | 2019-03-02 | 2019-08-02 | $403bn | $164bn | $240bn | 22 weeks | $40bn |
| 2021 standoff | 2021-08-01 | 2021-12-16 | $954bn | $78bn | $876bn | 20 weeks | $166bn |
| 2023 standoff | 2023-01-19 | 2023-06-03 | $635bn | $49bn | $586bn | 19 weeks | $82bn |
| 2025 standoff | 2025-01-02 | 2025-07-04 | $842bn | $306bn | $536bn | 26 weeks | $106bn |
| Pre-2015 subtotal (2 episodes) | $147bn | – | – | ||||
| 2015–2019 subtotal (4 episodes) | $867bn | – | – | ||||
| 2021–2025 subtotal (3 episodes) | $1,998bn | – | – | ||||
| Post-2015 total (7 episodes) | $2,865bn | – | – | ||||
Two points in the table deserve emphasis. The 2013 window is the longest (42 weeks) because it spans two binding periods: from January 1 to the No Budget, No Pay Act of February 4, 2013, which suspended the limit until May 18, and from the limit’s reinstatement on May 19 to the Continuing Appropriations Act of October 17, 2013. Treated as one window, 2013 shows a $70 billion drawdown. Split at the suspension, it shows $29 billion (January 2013) and $144 billion (May to October 2013, measured from the April tax-season peak of $157 billion), $173 billion in total: the pre-2015 subtotal would then rise to $250 billion, and the ratio of post-2015 to pre-2015 drawdowns would fall from 19.5× to 11.5×. The table keeps the single window, the page’s original convention.
The 2025 window runs from the reinstatement of the limit on January 2, 2025 to the law of July 4, 2025 (Pub. L. 119-21), which raised the limit by $5 trillion. Within that window the TGA reached its low of $306 billion in the week ending April 9, 2025, recovered on April tax receipts, then fell again; after the increase it touched $296 billion in the week ending July 16, 2025, before the rebuild. The 2025 monthly drawdown pace ($106bn/month) is exceeded only by the 2021 episode ($166bn/month) in the dataset’s history.
The 2021 drawdown pace, at $166 billion per month, is the fastest of the nine debt ceiling episodes and exceeds the average paces of QE1 ($108 billion per month: $1.725 trillion of announced purchases over 16 months), QE2 ($75 billion) and QE3 ($85 billion at its full pace). It remains far below the Fed’s purchases of March to June 2020: securities held outright rose by $2,091 billion between March 11 and June 10, 2020, or $699 billion per month (H.4.1, FRED WSHOSHO).
QE vs TGA drawdown: comparable flows, different mechanisms
Comparing the TGA drawdown pace to QE pace requires specifying exactly what is being compared. The numbers in the signature chart (below) represent monthly flow magnitudes, dollars of liquidity released into the private financial system per month, for both the Fed’s asset purchase programs and the Treasury’s cash drawdowns during debt ceiling episodes. On that narrow axis, the 2021 episode at $166 billion per month exceeds QE1, QE2 and QE3, but not the emergency purchases of March to June 2020, which ran at $699 billion per month.
The mechanism, however, differs materially. When the Federal Reserve purchases $100 billion of Treasury securities in a QE program, it creates $100 billion of new bank reserves as the offsetting liability on its own balance sheet. The total stock of private-sector financial assets changes: before the purchase, private holders had $100 billion of Treasury securities; after the purchase, they have $100 billion of bank reserves. Bank reserves are a claim on the Fed; Treasury securities are a claim on the US Treasury. The duration composition of the private sector’s portfolio changes, and the Fed’s balance sheet expands.
When the Treasury draws down the TGA by $100 billion, by contrast, no new reserves are created. The Treasury transfers existing reserves (which it held in its account at the Fed) to recipient banks. The Fed’s balance sheet does not expand. What changes is the composition on the liability side: reserves that were liabilities to the Treasury become liabilities to commercial banks. From the private sector’s standpoint, $100 billion of new bank reserves appear, but those reserves already existed “in the system” in the sense that they were already on the Fed’s balance sheet; they were simply in another Fed liability account.
Why does this distinction matter for interpreting the signature chart? Because QE aims to expand the aggregate stock of bank reserves, while a TGA drawdown redistributes existing reserves from the Treasury’s account to commercial bank accounts. In monthly flow terms to the private sector, the two are directly comparable. In stock terms, the liquidity available to the banking system, QE is a pure expansion while TGA drawdown is a transfer. Both effects have been documented as influencing short-term rates, money market fund flows, and collateral conditions, but through partially different transmission channels.
Debt ceiling drawdowns vs Fed asset purchases: monthly liquidity release pace
The three debt ceiling episodes since 2021 drained the TGA at a pace comparable to QE1 to QE3; the purchases of March to June 2020 ran 4.2 times faster than the fastest episode.
The 2021 debt ceiling episode drained the TGA faster than the average pace of QE1, QE2 and QE3. Reference release: the weekly H.4.1 release on the Federal Reserve balance sheet. The 2025 episode ran close to the QE1 pace. The Fed’s purchases of March to June 2020 dwarf every debt ceiling episode, and the six episodes before 2021 all ran at or below $75 billion per month.
Sources: FRED (WTREGEN) for TGA observations; FRED (WSHOSHO, securities held outright, H.4.1) for the 2020 purchases; Federal Reserve program announcements for QE1 to QE3 totals and dates. Chart: Eco3min Research.
Forward S&P 500 returns by TGA regime, 2002–2025
The table below presents forward S&P 500 returns conditional on the TGA regime at the observation date. The classifier uses a 4-week rolling change in the TGA balance, combined with a post-episode indicator (see Methodology). Each regime is associated with a different distribution of subsequent equity outcomes, although the spreads are narrower than the headline chart might suggest: a reminder that TGA flows are one input among many in market conditions, not the sole driver.
| Regime | n | Median 6-month return | Median 12-month return | 12m IQR (P25–P75) | % positive at 12m | Median 12m MDD |
|---|---|---|---|---|---|---|
| Accumulation (TGA rising) | 44 | +11.68% | +22.85% | +12.11% to +36.52% | 81.8% | −4.42% |
| Stable | 990 | +5.87% | +11.81% | +4.52% to +17.45% | 84.7% | −4.88% |
| Drawdown (TGA falling) | 94 | +7.64% | +13.85% | −2.99% to +19.08% | 72.3% | −6.06% |
| Post-ceiling rebuild | 40 | +4.37% | +12.86% | +2.62% to +22.86% | 77.5% | −7.88% |
When the TGA was in a “Drawdown” regime (n=94), the median 12-month S&P 500 return was +13.85% with a 72.3% positive-outcome rate. In a “Stable” regime (n=990), the median was +11.81% with 84.7% positive outcomes. The central tendencies are close; it is the tails of the distribution that differ: Drawdown observations show the lowest positive-outcome rate, the lowest 25th percentile (−2.99%) and a deeper median drawdown than the Stable and Accumulation regimes. The Accumulation regime (n=44) shows the highest median 12-month return (+22.85%); the Post-ceiling rebuild regime (n=40) shows the deepest median 12-month drawdown (−7.88%). Both rest on small samples drawn from a few episodes.
Methodological note: The regime statistics pool every weekly observation with a complete 12-month window (reference dates from December 2002 to April 2025, windows ending by April 2026, the study cutoff), so forward windows overlap. Overlapping windows inflate the statistical significance of any formal hypothesis test; the table is presented as descriptive statistics, not as inference. Forward returns are measured on monthly averages of daily S&P 500 closes (Robert J. Shiller, Online Data): each weekly observation starts from the average of its calendar month, the 6-month and 12-month returns compare it with the averages six and twelve months later, and the 12-month maximum drawdown is the largest peak-to-trough decline between monthly averages within the window. Weeks of the same month therefore share the same values, and monthly averaging makes drawdowns shallower than on daily closes. Regime transitions are handled by reclassification at each weekly observation. The Accumulation (n=44) and Post-ceiling rebuild (n=40) categories are the smallest samples and draw on a handful of episodes; their statistics carry that caveat. For comparable forward-return frameworks in other macro datasets, see our Real Rates vs CAPE study.
Past distributions do not predict future outcomes. The regime-conditional statistics describe historical patterns observed in the dataset, not expected returns.
- ▸ TGA at $751 billion (April 15, 2026): a reading below $500 billion would take the balance under a level it stayed below in 59.6% of weeks since January 2015. Last time the TGA was below $500bn: week ending August 6, 2025 ($421 billion).
- ▸ 4-week change at −$102 billion: a 4-week decline larger than $200 billion has occurred in 31 weeks since 2002, inside and outside debt ceiling episodes; the largest was −$469 billion (week ending March 24, 2021), and the largest within a debt ceiling window −$339 billion (week ending August 18, 2021). See our ON RRP dataset for the complementary flow on the Fed’s liability side.
- ▸ Next quarterly Treasury refunding announcement: early May 2026. The marketable borrowing estimates published with it state the cash balance the Treasury assumes for the end of the current and the following quarter.
TGA regime classification
The regime classification reduces the space of weekly observations to four interpretable categories, computed from the 4-week rolling change in the TGA balance and a post-episode indicator. Each category is a filter on the full dataset, not a forecast.
TGA rising significantly. Typically observed during tax collection periods (April, January) and post-ceiling rebuild phases. Historically associated with private-sector reserve absorption as the Treasury rebuilds its cash buffer.
TGA broadly flat over four weeks. The default regime: it describes 83% of the dataset since 2002. Treasury cash operations offset issuance and outflows without a directional trend.
TGA falling significantly. Observed both during debt ceiling episodes (the dataset’s high-magnitude cases) and during ordinary high-outflow weeks (Medicaid, bond coupons). The current observation is classified here.
TGA rebuilding after a resolved debt ceiling episode. Mechanical mirror of the prior drawdown: Treasury bill issuance returns to or above normal, withdrawing reserves from the private sector. The narrowest sample (n=40) and therefore the least precise forward statistics.
Historical inflection points
May–August 2011: The first modern standoff
The 2011 episode is the reference point for politics-centric analysis. The TGA entered the episode at $115 billion on May 4, 2011, bottomed at $38 billion on June 8, and ranged between $50 billion and $91 billion in July. The Budget Control Act, signed on August 2, resolved the statutory constraint. Drawdown magnitude: $77 billion. Pace: $67 billion per month, below the average paces of QE1 to QE3 shown in the comparison chart. The Standard & Poor’s downgrade of the US sovereign rating on August 5 came three days after resolution; the S&P 500’s August 2011 monthly average stood 10.6% below July’s. The TGA data do not indicate that the 2011 episode was a significant direct liquidity event; the market reaction was driven by credit-rating uncertainty and broad risk-off flows. A sell-off with no liquidity trigger, August 2011 is consistent with how little the statistical tie between Fed net liquidity and the stock market holds once it is measured in weekly changes rather than in levels.
January–October 2013: Two consecutive standoffs
The 2013 episode encompassed two linked binding periods. The No Budget, No Pay Act of February 4, 2013 suspended the limit until May 18; the limit was reinstated on May 19, and the second period ended with the Continuing Appropriations Act of October 17, 2013, which closed the October government shutdown and suspended the limit until February 7, 2014. Measured from its pre-episode peak ($83 billion on October 3, 2012), the TGA fell to $14 billion on May 29, 2013, ten days after the reinstatement. Combined-window drawdown: $70 billion over 42 weeks. Pace: $9 billion per month, the slowest in the dataset. As in 2011, the episode was constrained by the Treasury’s pre-2015 cash management posture.
August–December 2021: The post-COVID drawdown
The 2021 episode is the largest in the dataset. The TGA entered the episode at $954 billion on May 5, 2021, at the 95th percentile of the series to that date, after the COVID response had elevated Treasury cash balances throughout 2020. Secretary Yellen notified Congress on August 2 that extraordinary measures were beginning. The TGA bottomed at $78 billion on October 13, 2021. The episode was resolved by a ceiling increase enacted on December 16, 2021. Drawdown: $876 billion. Pace: $166 billion per month, 54% faster than QE1’s average pace. The S&P 500 rose 5.0% over the episode window (monthly averages, August to December 2021), consistent with the supportive liquidity environment created mechanically by the drawdown, though it is difficult to disentangle the TGA effect from simultaneous Fed balance sheet dynamics and the post-vaccine reopening.
January–June 2023: Yellen’s extraordinary measures
The 2023 episode hit the debt ceiling on January 19, 2023, triggering extraordinary measures that lasted until resolution via the Fiscal Responsibility Act of June 3, 2023. The TGA entered the episode at $635 billion on October 26, 2022 (pre-episode peak) and reached its window low of $49 billion in the week ending May 31, 2023, three days before resolution; the weekly average fell further, to $45 billion, in the week ending June 7. Drawdown: $586 billion. Pace: $82 billion per month. The episode coincided with the regional banking stress of March 2023 (Silicon Valley Bank, Signature Bank, First Republic), which complicates any attempt to attribute equity or credit market behavior to a single factor. The rebuild that followed was the second fastest in the dataset: from $45 billion to $531 billion in the six weeks to July 19, 2023, behind the $591 billion rebuilt in the six weeks after December 22, 2021. The balance slipped back to $414 billion in late August before passing $800 billion in late October, and the rebuild coincided with the rise in the 10-year Treasury yield, from 3.97% on July 31 to 4.34% on August 21, 2023. See also: Our breakdown of how QT and monetary plumbing shape market liquidity.
January–July 2025: The recent episode
The 2025 debt ceiling episode began with the reinstatement of the statutory ceiling on January 2, 2025 (after the 2023 suspension expired). The Treasury entered the episode at $842 billion on November 6, 2024. The TGA reached its window low of $306 billion in the week ending April 9, 2025. The law of July 4, 2025 (Pub. L. 119-21) raised the limit by $5 trillion. Drawdown: $536 billion. Pace: $106 billion per month. The 2025 window low was higher than the 2021 ($78bn) and 2023 ($49bn) lows, but the balance kept falling after the increase, to $296 billion in the week ending July 16, 2025, before a rebuild to $590 billion by August 27.
April 2026: Current observation
As of April 15, 2026, the TGA balance is $751 billion, at the 88th historical percentile, with a 4-week drawdown of $102 billion. This places the current reading in the “Drawdown” regime, although outside any active debt ceiling episode. The most likely explanation is routine Treasury cash management following the April 15 tax filing deadline: although April typically produces strong receipts, the drawdown reflects the timing gap between outflows and the rebuild of the cash position via post-April bill issuance. The next quarterly refunding announcement, in early May 2026, states the end-of-quarter cash balance the Treasury assumes.
Methodology
The dataset combines weekly Treasury General Account observations from FRED (series WTREGEN, weekly average of daily balances for the week ending Wednesday) with monthly averages of daily S&P 500 closes (Robert J. Shiller, Online Data) and weekly Federal Reserve total assets (FRED series WALCL). Derived columns include episode tagging, drawdown magnitudes, regime classification, and forward-return columns at standard 6-month and 12-month horizons.
Regime = f(4-week rolling TGA change, post-episode indicator)
Episode selection criteria
– T_bind = date the statutory ceiling becomes binding (Treasury notification to Congress)
– T_resolution = date of legislation raising or suspending the ceiling
Included episodes: 2011, 2013, 2015, 2017, 2017–2018, 2019, 2021, 2023, 2025 (all resolved before the April 15, 2026 cutoff)
Episode boundaries are set from public legislative dates and Treasury notifications to Congress, not from the TGA series itself. A purely algorithmic definition (e.g., “a drawdown greater than $X billion over Y weeks”) would conflate debt ceiling episodes with other large outflow events (tax refund season, coronavirus stimulus) and would be methodologically inferior for this specific use.
Sensitivity to boundary choice: Splitting the 2013 window at the suspension of February 4 to May 18, 2013 gives two drawdowns, $29 billion (January 2013) and $144 billion (May to October 2013, measured from the April peak of $157 billion), $173 billion in total against $70 billion for the combined window; the pre-2015 subtotal would rise from $147 billion to $250 billion. Shortening the 2011 episode to end at the Budget Control Act signing rather than the post-resolution rebuild does not change the reported drawdown.
Bibliographic anchoring: The Treasury’s quarterly refunding statement of May 6, 2015 sets out the cash balance policy, discussed in the TBAC minutes of May 5, 2015; Krishnamurthy and Vissing-Jorgensen (2015) document how the supply of Treasury securities affects the financial sector’s short-term debt and lending; Bianchi and Bigio (2022) model banks’ liquidity management and the transmission of monetary policy through reserves.
Dataset design
| Variable | Type | Unit | Source | Calculation |
|---|---|---|---|---|
| date | date | – | FRED | direct (week ending Wednesday) |
| tga_balance | float | $bn | FRED WTREGEN | FRED value / 1000 (millions to billions) |
| tga_weekly_change | float | $bn | derived | tga_balance[t] − tga_balance[t−1] |
| walcl | float | $bn | FRED WALCL | FRED value / 1000 |
| sp500 | float | index | Robert J. Shiller (Online Data) | monthly average of daily closes, month of the observation |
| debt_ceiling_episode | string | – | derived | episode label per legislative record |
| tga_drawdown_from_local_peak | float | $bn | derived | 90-day pre-episode peak − current balance |
| liquidity_injection_cumulative | float | $bn | derived | identical to drawdown during episodes |
| tga_regime | string | – | derived | f(4w change, post-episode indicator) |
| sp500_fwd_6m_pct | float | % | derived | ((SP500[m+6] / SP500[m]) − 1) × 100, m = month of t |
| sp500_fwd_12m_pct | float | % | derived | ((SP500[m+12] / SP500[m]) − 1) × 100 |
| sp500_fwd_12m_mdd | float | % | derived | min((SP500[s] / max(SP500[m:s])) − 1) over months [m, m+12] |
Python reproduction code
# Reproduce this dataset from primary sources import pandas as pd import numpy as np # 1. Fetch WTREGEN from FRED (weekly) tga = pd.read_csv('https://fred.stlouisfed.org/graph/fredgraph.csv?id=WTREGEN') tga.columns = ['date', 'tga_balance_m'] tga['date'] = pd.to_datetime(tga['date']) tga['tga_balance'] = tga['tga_balance_m'] / 1000 # millions -> billions # 2. Tag debt ceiling episodes episodes = [ ('2011_standoff', '2011-05-16', '2011-08-02'), ('2013_standoff', '2013-01-01', '2013-10-17'), ('2015_standoff', '2015-03-16', '2015-11-02'), ('2017_standoff', '2017-03-16', '2017-09-08'), ('2017_2018_standoff', '2017-12-09', '2018-02-09'), ('2019_standoff', '2019-03-02', '2019-08-02'), ('2021_standoff', '2021-08-01', '2021-12-16'), ('2023_standoff', '2023-01-19', '2023-06-03'), ('2025_standoff', '2025-01-02', '2025-07-04'), ] tga['episode'] = 'none' for name, start, end in episodes: mask = (tga['date'] >= start) & (tga['date'] <= end) tga.loc[mask, 'episode'] = name # 3. Compute drawdown from pre-episode peak for name, start, end in episodes: pre = tga[(tga['date'] >= pd.Timestamp(start) - pd.Timedelta(days=90)) & (tga['date'] < pd.Timestamp(start))] pre_peak = pre['tga_balance'].max() mask = tga['episode'] == name tga.loc[mask, 'drawdown'] = pre_peak - tga.loc[mask, 'tga_balance'] # 4. Classify regime tga['4w_change'] = tga['tga_balance'].diff().rolling(4, min_periods=1).mean() tga['regime'] = 'Stable' tga.loc[tga['4w_change'] > 25, 'regime'] = 'Accumulation' tga.loc[tga['4w_change'] < -25, 'regime'] = 'Drawdown' # 5. Post-ceiling rebuild: within 180 days after a resolution and rising faster than 25bn per week for name, start, end in episodes: post = (tga['date'] > end) & (tga['date'] <= pd.Timestamp(end) + pd.Timedelta(days=180)) tga.loc[post & (tga['4w_change'] > 25), 'regime'] = 'Post-ceiling rebuild'
Study file & reproducibility
The study file is not offered for download: it contains data whose owner restricts redistribution (S&P 500 index levels). The methodology lists every source.
Data sources & references
- Primary Federal Reserve Economic Data (FRED), Federal Reserve Bank of St. Louis. Series WTREGEN: “U.S. Treasury, General Account: Week Average”, weekly, not seasonally adjusted. Accessed April 2026.
- Primary Federal Reserve Economic Data (FRED). Series WSHOSHO (securities held outright) and WRESBAL (reserve balances with Federal Reserve Banks), H.4.1, weekly. Accessed October 2026.
- Primary Federal Reserve Economic Data (FRED). Series WALCL: “Assets: Total Assets: Total Assets (Less Eliminations from Consolidation): Wednesday Level”, weekly. Accessed April 2026.
- Primary Robert J. Shiller, Online Data (shillerdata.com): S&P 500, monthly averages of daily closes. Accessed September 2026.
- Reference US Department of the Treasury, Quarterly Refunding Statement, May 6, 2015 (cash balance policy), and Treasury Borrowing Advisory Committee minutes, May 5, 2015.
- Reference Public laws on the debt limit, 2011 to 2025 (Pub. L. 112-25, 113-3, 113-46, 114-74, 115-56, 115-123, 116-37, 117-73, 118-5, 119-21), GovInfo.
- Reference US Department of the Treasury, “Daily Treasury Statement”, historical archive. Daily TGA observations used for cross-validation of the FRED weekly series.
- Reference Federal Reserve H.4.1 release, “Factors Affecting Reserve Balances”, weekly. Used for cross-checking the TGA as a Fed liability.
- Research Krishnamurthy, A., and Vissing-Jorgensen, A. (2015). “The Impact of Treasury Supply on Financial Sector Lending and Stability.” Journal of Financial Economics, 118(3), 571–600.
- Research Bianchi, J. and Bigio, S. (2022). “Banks, Liquidity Management, and Monetary Policy.” Econometrica, 90(1), 391–454.
Methodological limitations
- The post-2015 sample contains seven episodes, three of them since 2021. Formal hypothesis tests comparing post-2015 to pre-2015 behavior are not appropriate at this sample size. The dataset is descriptive, not inferential.
- TGA drawdowns do not create new bank reserves the way Federal Reserve asset purchases do. Readers treating the QE comparison as mechanically equivalent should consult the QE vs TGA drawdown section for the distinction between reserve creation and reserve redistribution.
- Episode boundaries are defined from legislative dates, not from the TGA series itself. Alternative bounding rules change reported drawdowns, most for the 2013 episode: splitting it at its suspension raises its drawdown from $70 billion to $173 billion.
- Forward S&P 500 returns by regime use overlapping weekly observations. Reported statistics are descriptive of the historical distribution; any significance testing would require adjustment for serial correlation.
- The TGA is a gross cash balance, not a net fiscal position. Interpreting it as a fiscal-impulse proxy is incorrect; the TGA can rise or fall for reasons unrelated to underlying fiscal stance (issuance decisions, cash management targets, seasonal receipts).
- The study is cut at April 15, 2026: the 2025 episode, resolved on July 4, 2025, and its rebuild are covered; later observations are not part of this version.
Frequently asked questions
What is the current Treasury General Account balance?
As of April 15, 2026, the Treasury General Account balance is $751 billion, with a 4-week change of −$102 billion. This places the reading at the 88th historical percentile and classifies the current regime as “Drawdown” under the dataset’s 4-week change criterion. The balance is updated weekly by the Federal Reserve via the FRED series WTREGEN.
Why are pre-2015 debt ceiling drawdowns much smaller than recent ones?
The difference is rooted in Treasury cash management policy. Before 2015, the Treasury kept a small operational cash balance at the Fed (a $5 billion target before the 2008 crisis, mostly below $100 billion from 2009 to 2014), relying on its ability to issue short-dated Treasury bills to cover near-term outflows. On May 6, 2015, following the 2011 and 2013 episodes, the Treasury announced that it would hold a level of cash generally sufficient to cover one week of outflows, subject to a minimum balance of roughly $150 billion. In 2020–2021, TGA balances ranged between $500 billion and $1.8 trillion in 63.8% of weeks. Higher starting balances mechanically produce larger drawdowns when the ceiling becomes binding. The policy is set out in the Treasury’s quarterly refunding statement of May 6, 2015.
Is a TGA drawdown the same thing as quantitative easing?
No. Both produce comparable monthly liquidity flows to the private financial system, but through different mechanisms. QE (Fed asset purchases) creates new bank reserves by expanding the Fed’s balance sheet: the Fed issues a new liability (reserves) and acquires a new asset (a Treasury security). A TGA drawdown redistributes existing reserves: the Treasury transfers cash from its account at the Fed to recipient banks. The Fed’s total balance sheet does not expand, but the composition of its liabilities changes: reserves move from the Treasury’s account to commercial bank accounts. In monthly flow magnitude to the private sector, the two are directly comparable; in aggregate reserve stock terms, QE is a pure expansion while TGA drawdown is a composition change. Both are instruments of the same liquidity plumbing, one strand of the monetary regimes pillar on rates, liquidity and market cycles.
Do market participants already account for TGA flows?
Yes, in most professional contexts. Primary dealers, bank treasurers, and money market funds track TGA balances in real time via the US Treasury’s Daily Treasury Statement and produce explicit forward-balance projections. Sophisticated institutional participants anticipate TGA dynamics. What is less present is the public discussion: TGA flows are not debated at the FOMC, presented at press conferences, or catalogued in Fed balance sheet commentary. Framing TGA drawdowns as “stealth liquidity” is accurate in the sense that these flows are under-discussed in financial journalism, not in the sense that they are unknown to institutional participants.
Is the claim that 2021 was “faster than QE1” robust to different methods of computing pace?
The 2021 TGA drawdown pace reported here ($166 billion per month) is computed as the total drawdown ($876 billion) divided by the number of months between the pre-episode peak (May 5, 2021) and the episode trough (October 13, 2021), 5.3 months. QE1 is often cited at $108 billion per month based on $1.725 trillion of total purchases over 16 months. Using alternative windows on either side, the exact ratio shifts, but the order remains the same: computing 2021 over its formal standoff window (August–December, 4.5 months) yields $195 billion per month; computing QE1 over the initial MBS purchase window (January–June 2009) yields $131 billion per month (change in securities held outright, H.4.1). The 2021 pace exceeds QE1 under each of these window conventions; it does not exceed the Fed’s purchases of March to June 2020 ($699 billion per month).
What does this dataset not measure?
This dataset measures gross weekly TGA balances and episode-level drawdown magnitudes. It does not measure: (1) bank reserves: see FRED series WRESBAL or the weekly H.4.1 release; (2) net fiscal impulse: a different concept requiring a model of spending types and multipliers; (3) market impact: the relationship between TGA flows and asset prices is complex and mediated by simultaneous QE/QT, bill supply, and money market fund behavior; (4) effects through the Overnight Reverse Repo Facility: for that, see our ON RRP dataset and the Net Liquidity Index.
Source
Related Eco3min research
Last updated — 23 October 2026
Disclaimer – Financial and Legal Information: The analyses, commentary, and content published on eco3min.fr are provided for informational and educational purposes only. They do not constitute investment advice or a solicitation to buy or sell financial instruments. Past performance is not indicative of future results. All investment decisions involve risk and are the sole responsibility of the reader. Data, charts and calculated series are provided as is and may contain errors or lag behind their source: check any figure against the primary source cited before relying on it. Content describing laws, regulations or contracts reflects the law as of its publication date and does not constitute legal or tax advice; applying it to a specific situation requires a qualified legal or tax professional.
Regulated and tax values: the rates, indices, caps and brackets with legal effect cited on this page are set by official decision, change over time and apply according to individual circumstances. Only the official publication in force on the relevant date is authoritative (for example the Journal officiel, INSEE, Banque de France, the IRS or the Social Security Administration). To revise a rent, sign or challenge a contract, or file a return, check the value at the source or with a professional.