Short Rates Desk
Thu, Aug 20, 2026

Notes for Thu, Aug 20, 2026

Morning brief

· generated 07:07 ET

Trim the 2s10s steepener as the buyback eats half our buffer

Treasury's doubled long-end buyback flattened 2s10s to 46bp; the July minutes read hawkish but data-contingent. We cut the steepener, keep the front leg.

Curve

2s10s flattened to 46bp Wednesday[5], in from 52bp Tuesday[6]. Treasury drove it. The desk doubled the maximum size of its long-end buyback operations, from $2 billion to at least $4 billion per operation across 10y to 30y, effective September 9 through November 4[7]. 30Y yields fell 6bp on the announcement and 8bp on the day, and 2s30s flattened 7bp[8]. A weak 20Y auction in the same session, pricing 5.204% with indirect takedown at 62.9% and the softest since February[9], would ordinarily steepen the curve; the buyback bid overwhelmed it. The $40 trillion debt-crossing headline is settlement mechanics[10], and we ignore it.

On our record, the first break test, the front pushing through 4.25% and pulling up with the long end, did not trip: 2Y sat at 4.19% Tuesday[11] and the minutes left it there. The second is now half-live. The buyback is the durable long-end cap we flagged, and it pulled 2s10s to 46bp, 6bp from our 40bp stop.

We hold the term-premium steepener but cut size. The buyback is a standing bid on 10s to 30s through November 4, a persistent cap on the back leg that a soft 20Y cannot offset operation for operation, and the front leg is where the edge now sits. The steepener's original premise, term premium re-expanding, is the weak link: ACM 10Y term premium was 84bp last week[12], and Stalwart argues it stays structurally depressed even as Treasury shifts toward bills[13], so the recent steepening owes more to supply and positioning than to a term-premium regime shift. We treat the buyback cap as binding rather than fading it. This still assumes Treasury holds its coupon pace; the buyback changes demand while leaving the coupon schedule intact, so the refunding math holds, and a coupon upsize at the next refunding is what forces a full re-baseline. We're wrong if the buyback takes 2s10s through 40bp before Thursday, or if a strong claims print pushes 2Y through 4.25%; out of the front leg on either.

FOMC

The July minutes read hawkish. Participants judged inflation risks skewed to the upside[1], several favored a hike over the hold[2], and several tied AI investment to broader price pressure[3]. Timiraos reads the hike option as live but explicitly contingent on incoming labor and inflation data[4], and we agree: the minutes do not foreclose a hike, they hand the next move to Thursday's jobless claims. Don't fade the front-end rally on the minutes alone. The front held at 4.19% into the print, which is why the near-term risk sits with the data, not the record.

Funding

Funding is unchanged and out of the story. SOFR fixed 3.65% Tuesday[14], level with IORB at 3.65%[15] and EFFR at 3.63%[16], so no pressure through the front. RRP take-up is negligible at $0.3 billion[17] and reserves at $2.94 trillion[18], which leaves no scarcity signal. The one dislocation is idiosyncratic: the February 18 2027 bill (912797TV9) drew a 403.3bp SOMA lending fee Wednesday[19], up from 253.6bp Tuesday[20] against the 5bp program minimum. That is single-issue demand and says nothing about GC.

Sources read

4 sources read

  • Commentary items: 4

Citations

  1. [1]Participants judged inflation risks skewed to the upside (FED: PARTICIPANTS JUDGED INFLATION RISKS WERE SKEWED TO UPSIDE) — Commentary · zerohedge.com
  2. [2]several favored a hike over the hold (Most participants at Fed's July 28-29 meeting supported keeping interest rates unchanged, but several favored an increase) — Commentary · zerohedge.com
  3. [3]several tied AI investment to broader price pressure (FED: SEVERAL SAW AI INVESTMENT HAVING BROADER EFFECT ON PRICES) — Commentary · zerohedge.com
  4. [4]Timiraos reads the hike option as live but explicitly contingent on incoming labor and inflation data (Fed's wait-for-clarity language doesn't foreclose hikes, but does condition them on interim data.) — Commentary · twitter.com
  5. [5]2s10s flattened to 46bp Wednesday (0.46) — FRED T10Y2Y · Aug 19, 2026
  6. [6]in from 52bp Tuesday (0.52) — FRED T10Y2Y · Aug 18, 2026
  7. [7]from $2 billion to at least $4 billion per operation across 10y to 30y, effective September 9 through November 4 (doubling of buyback operation sizes for longer-dated securities (10y to 30y), effective September 9, 2026 through November 4, 2026.) — Commentary · zerohedge.com
  8. [8]30Y yields fell 6bp on the announcement and 8bp on the day, and 2s30s flattened 7bp (30Y yields fell 6bp immediately on announcement and 8bp total for the day.; 2s30s curve flattened 7bp on the day) — Commentary · zerohedge.com
  9. [9]pricing 5.204% with indirect takedown at 62.9% and the softest since February (20Y auction priced at 5.204% yield; Indirects awarded 62.9% of the auction, down sharply from 69.1%, the lowest since February) — Commentary · zerohedge.com
  10. [10]The $40 trillion debt-crossing headline is settlement mechanics (Treasury debt crossed $40T overnight (+$60.8B in one day)) — Commentary · twitter.com
  11. [11]2Y sat at 4.19% Tuesday (4.19) — FRED DGS2 · Aug 18, 2026
  12. [12]ACM 10Y term premium was 84bp last week (0.8393) — FRED THREEFYTP10 · Aug 14, 2026
  13. [13]Stalwart argues it stays structurally depressed even as Treasury shifts toward bills (ACM model suggests TP remains structurally depressed even if Treasury shifted duration to bills) — Commentary · twitter.com
  14. [14]SOFR fixed 3.65% Tuesday (3.65) — NY Fed SOFR · Aug 18, 2026
  15. [15]IORB at 3.65% (3.65) — FRED IORB · Aug 20, 2026
  16. [16]EFFR at 3.63% (3.63) — NY Fed EFFR · Aug 18, 2026
  17. [17]RRP take-up is negligible at $0.3 billion (0.317) — FRED RRPONTSYD · Aug 19, 2026
  18. [18]reserves at $2.94 trillion (2944059) — FRED WRESBAL · Aug 12, 2026
  19. [19]403.3bp SOMA lending fee Wednesday (403.3 bp) — Observation · observation:seclend_observations:912797TV9:2026-08-19
  20. [20]up from 253.6bp Tuesday (253.6 bp) — Observation · observation:seclend_observations:912797TV9:2026-08-18

Generated by Short Rates Desk. Informational only. Not investment advice.

Close brief

· generated 16:23 ET

Buyback rally gone by the close; we keep the trimmed 2s10s steepener

Both break tests held. Claims at 206k anchored the front while oil and term premium pushed the long end back toward pre-buyback yields.

Session

The Treasury buyback rally faded through the session and the long end finished near where it traded before the announcement. The 10Y sat at 4.71% Tuesday[1] and the 30Y at 5.28%[2], with the 30Y near cycle highs[3], and afternoon reporting put 10Y back to 4.69% as Brent pushed above $94[4]. The front did not follow: SOFR fixed at 3.62%[5] and EFFR at 3.63%[6], both flat, and the 2Y held 4.19%[7]. Initial claims came in at 206k with continuing claims at 1,799k[8], a beat too marginal to hand doves cover. We read the day as a term-premium move in the long end and keep the trimmed 2s10s steepener on.

Standing view

Both break tests on our record held. The first, the front pushing through 4.25% and pulling the long end up with it, did not trip: the 2Y sat at 4.19% Tuesday[9], and a 6k claims beat is not the strong print that would reprice it. The second, the buyback taking 2s10s through 40bp before Thursday, also held: 2s10s printed 46bp Wednesday[10], down 6bp on the day[11] but well clear of 40. We stay in the trimmed steepener. The assumption behind it, that Treasury holds its coupon pace, is intact: the $23B 30Y sale Thursday is matched by a $23B buyback Friday, net duration neutral[12], and the buyback doubling to $4B a week from September[13] shifts the demand mix without touching the coupon schedule.

Long end

The long-end backup is term premium, not a policy repricing. If it were the latter, the front would move with the back, and it did not, so we attribute the selloff to term premium, fiscal supply, and the oil bid. Bessent's push that yields sit above fundamentals has not held, with 30s back above their pre-intervention levels[14], and the Street is reading the buyback as a floor under duration risk, with Nomura calling it cosmetic against structural long-end supply[15]. JPM's point that visible official hands in Treasury pricing widen term premium[16], and Goldman's that Operation Twist-style duration management cannot compress structurally driven term premium[17], both fit the reversal. We are wrong on the term-premium read if the front breaks higher with the long end; that shock has not shown up.

Funding

Funding stayed easy. SOFR at 3.62% fixed 3bp under IORB at 3.65%[18], and GCF Treasury repo held 3.66%[19], so overnight rates give no sign of reserve scarcity. The RRP is effectively empty at $0.3B[20], cash fully deployed out of the facility, and reserves fell $49.3B on the week[21] to $2.94T[22]; that is the buffer to watch as bill supply builds, though with SOFR under IORB it is not binding. TGA rose $30B on the week[23] to $959B ahead of the coming bill settlement[24].

Into tomorrow

We keep the trimmed 2s10s steepener and lean to a modest bear-steepening bias into tomorrow. The front is anchored by a Fed on hold at 3.50-3.75%[25] and a claims print that gave doves nothing, while term premium and the oil bid keep the long end under pressure. We add back the front leg only on a pullback. We are out of the front leg if 2s10s flattens through 40bp or the 2Y breaks 4.25%, both our thresholds off the standing view. The bear-steepener is off if Trump pivots and the Hormuz premium drains within 48 hours[26], or if the front reprices higher with the back. We re-baseline the whole structure if the next refunding upsizes coupons; the current buyback leaves that schedule intact.

Sources read

2 sources read

  • Commentary items: 2

Sources read

2 sources read

  • Commentary items: 2

Citations

  1. [1]The 10Y sat at 4.71% Tuesday (4.71) — FRED DGS10 · Aug 18, 2026
  2. [2]the 30Y at 5.28% (5.28) — FRED DGS30 · Aug 18, 2026
  3. [3]with the 30Y near cycle highs (Stalwart flags 30-year yields at cycle highs; argues fiscal consolidation, not Fed cuts, is the real driver to bring them down.) — Commentary · twitter.com
  4. [4]afternoon reporting put 10Y back to 4.69% as Brent pushed above $94 (Bessent's buyback intervention already erased, 10Y back to 4.69%, above pre-announcement levels, as Trump's Iran escalation pushed Brent above $94.) — Commentary · zerohedge.com
  5. [5]SOFR fixed at 3.62% (3.62000) — NY Fed SOFR · Aug 19, 2026
  6. [6]EFFR at 3.63% (3.63000) — NY Fed EFFR · Aug 19, 2026
  7. [7]the 2Y held 4.19% (4.19) — FRED DGS2 · Aug 18, 2026
  8. [8]Initial claims came in at 206k with continuing claims at 1,799k (Initial claims printed 4k below forecast at 206k, marginally tighter labor market, but continuing claims creep 11k above expectations to 1799k.) — Commentary · twitter.com
  9. [9]the 2Y sat at 4.19% Tuesday (4.19) — FRED DGS2 · Aug 18, 2026
  10. [10]2s10s printed 46bp Wednesday (0.46) — FRED T10Y2Y · Aug 19, 2026
  11. [11]down 6bp on the day (-0.06 d/d) — FRED T10Y2Y · Aug 19, 2026
  12. [12]the $23B 30Y sale Thursday is matched by a $23B buyback Friday, net duration neutral (Treasury announced $23B 30Y sale Thu, immediate $23B buyback Fri, net duration neutral but signals front-load of supply absorption.) — Commentary · twitter.com
  13. [13]the buyback doubling to $4B a week from September (Treasury doubling buyback program to $4B/week (9/9-11/4) is largely neutral for curve positioning, the cash drain is offset by concurrent issuance.) — Commentary · twitter.com
  14. [14]Bessent's push that yields sit above fundamentals has not held, with 30s back above their pre-intervention levels (Bessent's pushback on yields, claiming they don't reflect fundamentals, has failed to stick; 30s are back above pre-intervention levels.) — Commentary · twitter.com
  15. [15]the Street is reading the buyback as a floor under duration risk, with Nomura calling it cosmetic against structural long-end supply (Treasury buybacks priced as cosmetic amid structural long-end supply dynamics, Nomura positioning this as a hard floor for duration risk, not a valuation cushion.) — Commentary · twitter.com
  16. [16]JPM's point that visible official hands in Treasury pricing widen term premium (JPM calling out Treasury bond-market intervention as credibility-erosive... term premium reprices wider and curve steepeners get cheaper.) — Commentary · twitter.com
  17. [17]Goldman's that Operation Twist-style duration management cannot compress structurally driven term premium (Goldman pushes back on the notion that Operation Twist-style duration sales can sustainably compress long-end yields; argues term premia is structural.) — Commentary · twitter.com
  18. [18]SOFR at 3.62% fixed 3bp under IORB at 3.65% (3.65000) — FRED IORB · Aug 20, 2026
  19. [19]GCF Treasury repo held 3.66% (UST 3.65900) — DTCC GCF repo · Aug 19, 2026
  20. [20]The RRP is effectively empty at $0.3B (0.317) — FRED RRPONTSYD · Aug 19, 2026
  21. [21]reserves fell $49.3B on the week (-49,290M w/w) — FRED WRESBAL · Aug 12, 2026
  22. [22]to $2.94T (2,944,059M) — FRED WRESBAL · Aug 12, 2026
  23. [23]TGA rose $30B on the week (+30,080M w/w) — Treasury General Account · Aug 12, 2026
  24. [24]to $959B ahead of the coming bill settlement (959,405M) — Treasury General Account · Aug 12, 2026
  25. [25]a Fed on hold at 3.50-3.75% (3.50-3.75%) — FRED DFEDTARU · Aug 20, 2026
  26. [26]the Hormuz premium drains within 48 hours (Watch whether Hormuz closure scenario sticks or Trump pivot kills the bid within 48h.) — Commentary · zerohedge.com

Generated by Short Rates Desk. Informational only. Not investment advice.