What happened during the September 2019 repo crisis?

The short answer

Until mid-September 2019, the U.S. money market had been calm for years. Then on September 16, two predictable events drained reserves from the banking system: a Treasury debt settlement (around $54 billion) and corporate tax payments (a similar amount). These flows had occurred many times without incident.

This time, the system overshot violently. Overnight repo rates spiked from around 2 % to 10 % intraday on September 17, dragging the federal funds rate above the top of the Fed’s target range. The Fed responded with emergency repo operations of $75 billion that morning — but only $53 billion was bid for, suggesting that the constraint was not just the absolute amount of cash available.

The episode forced the Fed to recognise that its post-crisis reserve regime had become fragile in ways that had been invisible until then.

→ Background: Why do repo markets matter for financial stability?

What the data shows

The chronology compiled by the Fed of New York and the Office of Financial Research:

• September 13, 2019 (Friday): SOFR at 2.30%, EFFR at 2.14%, both well within the Fed’s target range.
• September 16: SOFR at 2.43%, EFFR at 2.25% — at the top of the range. Reserves at $1.4 trillion (multi-year low).
• September 17 morning: SOFR jumped to 5.25% for the day, with intraday transactions at 10%; EFFR moved above the target range.
• Fed response: $75 billion in overnight repo offered at 9:30 a.m.; primary dealers bid for $53 billion. The Fed continued daily operations for the rest of the week.

The exception worth noting: between September 16 and September 17, aggregate reserves actually rose by $35 billion. The crisis happened despite reserves growing, which rules out a simple aggregate-shortage explanation.

→ Dataset: U.S. bank reserves

Why it happens — the macro mechanism

The September 2019 spike was not a single-cause event. It was the intersection of three slow-building structural shifts that became binding all at once.

Channel 1 — the reserve regime transition. Between October 2017 and July 2019, the Fed shrunk its balance sheet from roughly $4.5 trillion to $3.8 trillion, draining reserves from a peak of $2.8 trillion to $1.4 trillion. This was the Fed’s first quantitative tightening, and the level at which reserves became binding was not known in advance.

Channel 2 — the distribution problem. Here is the angle that distinguishes serious analysis of the episode. Aggregate reserves at $1.4 trillion were not absolutely scarce — they were comparable to levels seen earlier in the year without incident. The problem was that reserves had become concentrated in a small number of large banks (the four primary dealer banks), which under post-Basel III liquidity rules were reluctant to lend them out. Dealers facing tight intraday liquidity constraints chose to forgo the opportunity to borrow at 2.1 % from the Fed and lend at 5 %+ in repo, suggesting that balance-sheet costs, not the price of cash, were the binding constraint.

Channel 3 — the LCR feedback. The Liquidity Coverage Ratio requires large banks to hold high-quality liquid assets to cover 30-day stress outflows, and these holdings must be operationally available within hours. Internal liquidity stress tests at major banks effectively penalised intraday lending of reserves, even when the headline LCR ratio looked comfortable.

Synthesis by regime: in the scarce-reserves regime that prevailed before 2008, the Fed actively managed reserves through small open-market operations and rates were stable. In the ample-reserves regime adopted after the financial crisis, the Fed assumed that simply having abundant aggregate reserves would prevent rate spikes. September 2019 revealed a third regime: enough aggregate reserves on paper, but not enough at the right institutions at the right time of day to prevent dysfunction.

The September 2019 spike was not a shortage of reserves but a failure of distribution — a lesson that aggregate measures can hide the location of stress.

→ Framework: Liquidity, financial conditions and monetary plumbing

What it means for different economic actors

Banks and dealers learned that internal liquidity policies designed to satisfy regulators could conflict with their traditional role as intermediaries in money markets. Many large institutions reassessed how they would respond to similar episodes.

The Fed reversed course quickly. It restarted Treasury bill purchases (the so-called organic balance-sheet expansion), launched temporary repo operations, and ultimately created the Standing Repo Facility in July 2021 to provide a permanent backstop.

Money-market participants took away that the post-crisis regime was less robust than advertised. Many built operational buffers to manage potential repeats — buffers that proved valuable in March 2020 and again in March 2023.

A common error is to read the September 2019 spike as a one-off technical event. It was not — it was the first visible signal that the post-crisis monetary plumbing had developed structural fragilities that would resurface in different forms in subsequent episodes.

Practical observation

Go deeper

Related questions

• Why do repo markets matter for financial stability?
• What is the Fed’s Standing Repo Facility?
• How does funding liquidity differ from market liquidity?
• Why does the Treasury General Account drain bank reserves?

Frequently asked questions

Was the September 2019 spike caused by a reserve shortage?

The conventional narrative says yes, but the evidence is more nuanced. Aggregate reserves stood at $1.4 trillion on the day of the spike — not exceptionally low compared to other periods that did not produce dislocations. The spike happened because reserves were concentrated in a few large banks that were unwilling or unable to redistribute them through repo, partly because of LCR-related internal constraints. Calling it a reserve shortage is shorthand for a more complex distribution failure.

Why did primary dealers not bid for the full $75 billion of Fed liquidity?

Primary dealers bid for $53 billion, leaving $22 billion on the table even though they could have borrowed from the Fed at 2.1 % and lent at much higher rates. This pattern suggests that balance-sheet capacity, not the price of cash, was the binding constraint. Even with cheap funding available, dealers could not expand their balance sheets enough to redistribute the liquidity throughout the system.

What permanent changes resulted from the episode?

The Fed restarted balance-sheet expansion in late 2019, then dramatically accelerated during COVID. It created the Standing Repo Facility in July 2021 to provide a permanent ceiling on overnight repo rates. It also began publishing reserve demand elasticity research to better understand the level at which reserves become binding. The lessons of September 2019 directly shaped the COVID-era response.

Last updated — 12 May 2026