Repo Fails

What Are Repo Fails?

A repo fail occurs when the seller in a repurchase agreement cannot deliver the agreed-upon securities to the buyer by the contractually specified settlement date. In a standard repo transaction, the seller borrows cash by pledging securities as collateral, agreeing to repurchase them at a slightly higher price on a future date. When the seller lacks possession of those securities — typically on-the-run U.S. Treasuries, agency debt, or agency MBS — the trade fails to settle, and the buyer does not receive the collateral they contracted for. The fail rate is measured as the aggregate dollar value of unsettled repo and outright securities transactions on a given day, reported weekly by the Federal Reserve Bank of New York across primary dealer activity.

The structural reason fails can persist is rooted in incentive economics. Prior to 2009, the fails charge was effectively zero: a seller who failed to deliver still owed the agreed-upon repo interest to the buyer, but at a near-zero short-term rate that charge was trivial. This created a perverse incentive — if you couldn't source the security, it was cheaper to fail than to pay up in the special repo market to borrow the specific collateral. Once a large enough number of participants adopt this posture, fails become self-reinforcing, as the same scarce security cycles through multiple chains of failed settlement.

Why It Matters for Traders

Repo fails are among the most sensitive real-time gauges of collateral scarcity and systemic stress in fixed income markets. When fails spike in a specific Treasury issue, it typically reflects one or more of three causes: aggressive short-selling of an on-the-run security, dealer balance sheet constraints preventing securities from flowing freely through the intermediation chain, or outright infrastructure bottlenecks in Fedwire Securities settlement. Each of these causes carries different implications for traders.

Elevated fails in on-the-run Treasuries directly compress the repo specialness of that issue — the additional yield a holder can earn by lending the scarce security. As fails climb, the spread between general collateral (GC) repo rates and the special rate for the affected issue can widen significantly, sometimes to several hundred basis points for the most acutely squeezed securities. This feeds directly into the Treasury basis trade: futures traders short the cheapest-to-deliver Treasury and hold the physical, so when that CTD goes on special or experiences chronic fails, the carry on the position deteriorates sharply. Wider bid-ask spreads and reduced market depth in the affected issues typically follow.

At the macro level, sustained fails signal that the Treasury market's plumbing is under strain in ways that can impair monetary policy transmission. The Federal Reserve's open market operations rely on smooth securities settlement; widespread fails complicate the Fed's ability to drain or add reserves efficiently, a concern that became operationally real during the post-GFC era of large-scale asset purchases.

How to Read and Interpret It

The New York Fed publishes weekly primary dealer fails data broken out by Treasuries, agencies, and MBS, typically with a one-week lag. Practical thresholds traders use as reference points:

• Under $50 billion/week: Normal functioning; routine settlement frictions only.
• $50–$150 billion/week: Elevated scarcity in specific issues; monitor on-the-run specialness spreads actively.
• $150–$500 billion/week: Systemic stress signal; commonly coincides with quarter-end dealer balance sheet compression, large auction settlements, or acute collateral shortages.
• Above $500 billion/week: Crisis-level dysfunction; immediate cross-market risk implications.
• Sustained elevation (>3 consecutive weeks): Suggests structural collateral dysfunction rather than a transient mechanical event.

Fails data gains analytical power when cross-referenced with related signals. The spread between SOFR and EFFR captures general collateral stress in overnight funding. CTD specialness in Treasury futures — the annualized premium to borrow the cheapest-to-deliver note in the repo market — often leads the aggregate fail data by several days. Net primary dealer Treasury positioning from the Fed's H.4.1 and related reports helps triangulate whether stress is idiosyncratic to one issue or market-wide.

Historical Context

The most catastrophic repo fail episode in U.S. history erupted in the weeks following the Lehman Brothers collapse in September 2008, when weekly Treasury fails peaked above $2.7 trillion — a figure that represented a staggering share of the entire outstanding Treasury market. Settlement chains froze as counterparties refused to extend credit to one another and custodians locked down positions; the ripple effects spread into money market funds, commercial paper markets, and ultimately the broader financial system.

But fails were problematic well before 2008. In August 2003, weekly fails exceeded $1 trillion as speculative short positions in on-the-run 10-year Treasuries outpaced the float available for borrowing, illustrating how even a benign macro environment can produce severe fails through concentrated positioning. A smaller but instructive episode emerged in late 2015 and early 2016, when post-crisis regulations constraining dealer balance sheets combined with heavy hedge fund positioning in specific off-the-run notes to produce persistent fails that lasted several weeks and notably widened Treasury-OIS spreads in those maturities.

In direct response to the 2008 crisis, the Treasury Market Practices Group (TMPG) implemented a fails charge of 3% annualized in May 2009, applied when the relevant federal funds target rate falls below 3%. This charge created a meaningful economic penalty for allowing fails to persist, and the data confirms it worked: chronic multi-week fail episodes became substantially rarer in the post-2009 era, though acute spikes around quarter-end and large settlement dates remain common.

Limitations and Caveats

Repo fails data has meaningful blind spots that traders must account for. The weekly New York Fed report covers primary dealer activity only, leaving the broader bilateral repo market — which includes significant inter-custodian and non-primary-dealer activity — largely invisible. The true market-wide fail rate could be materially higher than what the reported data suggests, particularly in agency MBS where bilateral settlement is common.

Mechanical calendar effects regularly distort the signal. Treasury auction settlement dates cluster fails artificially as new securities enter the system, and quarter-end window dressing by bank dealers compresses balance sheets in ways that elevate fails without implying any deterioration in underlying collateral availability. Analysts who fail to strip out these effects routinely misread the data.

During periods of Federal Reserve quantitative easing, the Fed's concentration of ownership in specific on-the-run issues can paradoxically suppress reported fails even when underlying conditions are tight — the Fed's Securities Lending Facility provides an emergency circuit-breaker that mutes the market signal. Conversely, during quantitative tightening, the removal of that buffer can amplify fail sensitivity to the same degree of underlying stress.

What to Watch

• New York Fed weekly primary dealer fails (released with ~one-week lag): watch for trend breaks and sustained elevation above $150 billion in Treasuries.
• SOFR–EFFR spread: a widening spread in tandem with rising fails confirms systemic funding stress rather than issue-specific scarcity.
• CTD specialness in Treasury futures: often leads aggregate fails data by two to five trading days and is available in near real-time through broker screens.
• Treasury issuance calendar: heavy auction settlement weeks — especially 10-year and 30-year — mechanically elevate fail risk; adjust thresholds accordingly.
• Fed Securities Lending Facility usage: a meaningful uptick signals market participants actively attempting to cure fails by borrowing from the Fed's portfolio, confirming scarcity is real rather than administrative.
• Dealer balance sheet proxies around quarter-end: falls in GC repo volumes and widening FRA-OIS spreads in the final days of a quarter are reliable leading indicators of an impending fail spike.