US Treasury Weekly Assessment — Week Ending 18 September 2026

Current Treasury regime

The regime is a front-end-led selloff with positive but flattening curve slopes. The 2-year yield rose 13bp to 4.76% during the week, compared with increases of 8bp at 5 years and 5bp at 10 years; the 20-year was unchanged and the 30-year declined 1bp. At the close, 2s10s, 5s30s and 10s30s were positive at 25bp, 48bp and 33bp, respectively.

This was a bear flattening through 10 years rather than a broad duration selloff. The relative firmness of the ultra-long end should not be treated as evidence of strong primary-market duration demand, as the supplied 20-year and near-10-year TIPS auction results were weak.

Executive summary

The available labour signal was resilient: initial claims fell to 196,000, the four-week average declined to 203,250, and insured unemployment fell to 1.73 million. This is consistent with resistance to near-term easing, but it is one weekly labour release and does not establish the broader growth trend.

Inflation compensation split by horizon. Five- and 10-year breakevens fell 9bp and 3bp, respectively, while 5y5y forward compensation rose 3bp. Energy-infrastructure disruption presents an upside inflation risk, but the supplied record contains neither an observed oil-price move nor a measured Treasury-flow response.

Primary-market conditions were mixed but weak in duration. The 20-year reopening had below-average indirect participation and elevated dealer absorption, while the near-10-year TIPS reopening had 4th-percentile bid-to-cover and very weak indirect participation. Next week's 2-, 5- and 7-year offerings will test whether that softness broadens.

Review of the previous thesis

The prior relative-maturity framework was broadly confirmed: the curve remained positively sloped, the selloff slowed, and the 2-year and 5-year sectors underperformed the long end. The unchanged 20-year yield and 1bp decline in the 30-year yield extended relative ultra-long resilience.

The prior auction mechanism requires revision. This week's 20-year reopening showed weak indirect participation and above-average dealer awards, while the near-10-year TIPS reopening had exceptionally weak coverage and indirect participation. Thus, long-end resilience continued despite—not because of—supportive supplied duration-auction evidence.

The contribution of expected policy rates and term premium remains unresolved. The available fitted one-year forward rate and Kim-Wright term-premium estimates are dated 11 September and do not cover most of the subsequent weekly move.

What changed this week

The weekly adjustment became more concentrated in policy-sensitive maturities. Same-series changes imply roughly 8bp of 2s10s flattening, 9bp of 5s30s flattening and 6bp of 10s30s flattening. Five-year real yields rose 8bp through 17 September, whereas 20- and 30-year real yields declined 2bp and 3bp.

Near-term inflation compensation softened while forward compensation firmed. Five-year and 10-year breakevens were observed on 18 September, one day after the supplied real-yield observations. These differently dated observations are not an exact nominal-real-inflation decomposition.

Auction demand and dealer absorption

Duration-auction demand weakened. The 20-year reopening's bid-to-cover ranked at the 42nd percentile of its preceding 24 exact-series auctions; indirect awards were 1.75 standard deviations below average and dealer awards were 1.23 standard deviations above average. The near-10-year TIPS reopening was weaker still, with coverage at the 4th percentile and indirect participation 2.32 standard deviations below average.

Bill auctions were heterogeneous. The 4-week reopening had 79th-percentile coverage, above-average indirect participation and below-average dealer awards; the 8-week reopening also had above-median coverage. The 13- and 26-week reopenings had weaker coverage rankings, although allocation measures were close to their exact-series averages.

Indirect awards are only an imperfect proxy for foreign demand, while direct awards do not identify domestic ownership. Auction allocations also do not establish subsequent dealer inventory, balance-sheet capacity or secondary-market liquidity.

Yield curve and market structure

The front end and belly underperformed, while the ultra-long end was stable to firmer. This supports a maturity-dependent rate move rather than uniform long-duration pressure. The 2s10s, 5s30s and 10s30s slopes remained positive despite their weekly narrowing.

Futures positioning was polarized. Leveraged funds remained heavily net short in 5- and 10-year futures, while asset managers held an extreme net-long 10-year position. Dealer-intermediary positions became more short at 2 and 10 years and were near the low end of their 52-week distributions. These data can signal amplification risk in either direction but do not reveal cash holdings, trade intent or financing sensitivity.

Current volatility and secondary-market liquidity conditions are unknown. The only supplied volatility observation, VIXTLT at 85.9 on 30 July, is stale and is not a whole-curve Treasury volatility measure.

Macro, Federal Reserve, issuance and fiscal backdrop

Claims data point to labour resilience rather than emerging deterioration. That is consistent with front-end pressure, but no meeting-specific policy pricing, current Federal Reserve communication or broader activity data are supplied. The fitted one-year forward rate was unchanged at 4.6662% as of 11 September, before most of the week's yield move.

Treasury projected privately held net marketable borrowing of $739 billion for July–September and $628 billion for October–December. It expects nominal coupon and FRN auction sizes to remain stable for several quarters, with shorter-bill reductions during September and increases across the bill curve anticipated in October.

Announced offerings for 21–24 September total at least $457 billion across identified bills and notes, including $69 billion of 2-year notes, $70 billion of 5-year notes and $44 billion of 7-year notes. This defines a material near-term absorption test, though the maturity-weighted duration of the complete forward calendar is not supplied.

Geopolitical and cross-market considerations

Supplied geopolitical records describe conflict escalation and disruption to Saudi oil infrastructure. These developments create plausible energy-inflation and safe-haven transmission channels, but they do not establish an observed oil-price effect, Treasury safe-haven flow or causal explanation for the week's yield moves.

Market evidence was not consistent with a broad Treasury haven rally: the 2-, 5- and 10-year yields rose, while the 20-year was unchanged and the 30-year declined marginally. The result is better characterized as maturity-specific long-end resilience under uncertainty than as measured broad-based safe-haven demand.

Current base case

The base case is for a positively sloped curve with a continuing, though potentially interrupted, flattening bias. Policy-sensitive maturities remain relatively vulnerable given resilient claims, the 2-year-led selloff and the rise in the 5-year real yield. Stable-to-lower ultra-long nominal and real yields support relative long-end outperformance rather than an outright bullish-duration conclusion.

Confidence is moderated by weak duration auctions, substantial forthcoming note supply and the absence of current meeting-specific policy pricing. Large opposing futures positions may amplify either a rally or a selloff without identifying the eventual direction.

Bullish-duration scenario

A duration rally would require materially weaker subsequent labour evidence, a current policy-path measure indicating greater expected easing, and broader real-yield declines. Improved coupon-auction coverage and end-investor participation with lower dealer absorption would provide primary-market confirmation.

Heavily net-short leveraged-fund positioning in the 5- and 10-year sectors could amplify a rally. This scenario is presently low confidence because the supplied claims release was resilient, the front end sold off, and recent duration auctions were weak.

Bearish-duration scenario

A broader selloff would require continued labour resilience, current evidence of reduced expected easing, rising real yields beyond the intermediate sector, and repeated weakness in 2-, 5- and 7-year auction demand with higher dealer awards. Fiscal financing and energy disruption would add to this case only if accompanied by observed market or inflation measures.

The current evidence supports only parts of this scenario: intermediate yields and the 5-year real yield rose, duration auctions weakened and financing needs are sizable. Counterweights include lower 20- and 30-year real yields, lower 5- and 10-year breakevens and stable-to-lower ultra-long nominal yields.

Thesis risks and invalidation

The base case would be challenged by a sustained front-end rally alongside materially weaker labour evidence and a lower current policy-path measure. It would also be challenged by a long-end-led selloff that widens 5s30s and 10s30s, particularly if 20- and 30-year real yields rise.

Strong 2-, 5- and 7-year auctions across coverage, end-investor participation and dealer containment would reduce the supply-and-absorption risk. Conversely, further weak coupon auctions with materially higher dealer awards and rising long yields would challenge the view that current auction softness is localized.

A broad decline in nominal and real yields accompanied by lower inflation compensation would invalidate the continuing bearish-duration impulse. Updated policy-path or term-premium measures could materially alter the proposed interpretation because current supplied estimates predate most of the week.

Next-week catalysts

On 21 September, $92 billion of 13-week and $79 billion of 26-week bills will retest weak coverage outcomes in those series. On 22 September, $75 billion of 6-week bills and $69 billion of 2-year notes will test short-end demand after the 2-year-led selloff.

On 23 September, $70 billion of 5-year notes and $28 billion of 1-year-10-month notes are scheduled; the announced 7-year offering is $44 billion on 24 September. The 24 September claims release will test whether the latest improvement in initial and insured claims persists. The 4- and 8-week bill reopenings also merit attention, although their offering amounts were not supplied by the cutoff.

Evidence and source notes

Data cutoff: 18 September 2026, 21:15:46 UTC. Official Treasury daily par yields provide the weekly nominal-rate and curve observations. Official auction records provide exact-series comparisons for bid-to-cover, indirect and dealer allocations. Treasury quarterly-refunding material supplies borrowing, issuance-policy and buyback guidance.

Real yields were observed through 17 September, while nominal yields and inflation-compensation measures were observed through 18 September. Accordingly, nominal, real and inflation-compensation changes should not be presented as an exact same-date decomposition. Breakevens include inflation, risk and liquidity premia.

Indirect auction awards are not a clean foreign-investor measure; direct awards do not establish domestic ownership. Futures positioning is not cash-market ownership or trader intent. Geopolitical evidence identifies risks and possible channels, not observed Treasury flows or causation.

Methodology and disclosure

This report synthesizes only the supplied evidence available by the stated cutoff. Weekly curve changes use same-series Treasury observations; auction assessments use the supplied exact-series comparison statistics. Interpretations are conditional where direct measures are unavailable.

AI assistance was used to structure and draft this assessment, subject to manual review. The report may be revised if source data are corrected, revised or supplemented. It is informational only, does not constitute personal financial advice, and does not recommend any security, strategy or transaction.