Interest Rate Weekly · 2026-08-17 to 2026-08-23

One-line summary: Last week the main theme in global interest rate markets was not a “disinflation trade,” but rather continued pressure on the long end of U.S. yields under the combined background of fiscal supply, policy communication and doubts about safe-haven status; the Canadian curve moved more coherently upward across the curve; U.K. and euro-area data provided an external reference of “inflation still sticky, growth marginally weakening.”

I. Review of Last Week

On the U.S. front, market focus was almost entirely on long-term Treasuries. On 8 August 19, the Federal Reserve left rates unchanged; the same day, multiple media outlets reported that the U.S. Treasury was stepping up debt buybacks to try to ease recent upward pressure on yields. From 8 August 20 to 21, reports about the Treasury’s “surprise buybacks,” “expanded buybacks,” and the “limited effect of plans to support the bond market” continued to circulate, but were soon overwhelmed by stronger supply and pricing pressure: the core refrain in the news was that absorption of new Treasuries was difficult, investors demanded higher yields, and the Treasury’s operations failed to stabilize the long end. Summary reports on 8 August 23 describing a “wild week in the bond market” and that the “turbulent week taught Treasury officials a lesson” essentially confirmed the dominant narrative of the week: the U.S. bond market was not derailed from basic pricing constraints by administrative technical measures.

The curve data matches the news narrative. U.S. 2-year, 5-year, 10-year, and 30-year yields rose 4bp, 7bp, 6bp, and 2bp respectively over the past 5 days, indicating this episode was not purely a long-end breakdown but also included short- and mid-end repricing of inflation and the policy rate path. In absolute terms, pressure remained concentrated at the long end: U.S. 10-year yield 4.69%, 1-year z-value 1.98; 30-year 5.23%, 1-year z-value 2.16, already at the high end of the past year’s distribution. Looking at the term structure, U.S. 2s10s is +50bp, maintaining a clearly positive slope, which indicates the market does not simply interpret current high rates as “much more Fed tightening at the short end,” but rather reflects term premium and long-end supply compensation to a greater extent.

Correspondingly, U.S. macro and policy news did not provide clear tailwinds for long bonds. On one hand, on 8 August 19 “many” Fed officials still viewed inflation as a problem; on the other hand, discussions about a possible Iran ceasefire affecting inflation and employment were more event-level uncertainty than realized evidence of disinflation. Reporting on 8 August 21 about a weakening of Treasuries’ safe-haven status further undermined the traditional linear logic that “in times of risk, long bonds are naturally sought.” The result is that Treasury buybacks can suppress volatility in the short term, but are unlikely to reverse long-end pricing.

The Canadian story is more straightforward. Without the same policy drama as in the U.S., Canadian yields moved up more uniformly and in a more neatly synchronized manner. Canadian 2-year, 5-year, 10-year, and 30-year yields rose 10bp, 12bp, 13bp, and 11bp respectively over the past 5 days; 13bp, 13bp, 15bp, and 15bp over the past 20 days; and 18bp, 23bp, 28bp, and 31bp over the past 60 days. Whether viewed weekly, monthly, or quarterly, Canada shows an “upward shift across the entire curve with a stronger long end.” Canada 10-year currently 3.75%, 30-year 4.15%, 1-year z-values are 2.25 and 2.49 respectively, positions that are clearly high; 2s10s is +73bp, a steeper curve than the U.S.

There was not much direct news for Canada, but one key item stood out: Bank of Canada research suggested that cutting rates might actually worsen housing affordability. For rate markets this is not a direct guide for the next meeting, but it at least indicates that Canadian policy discussion is not unidirectionally seeking lower rates and must weigh feedbacks among asset prices, housing and financial conditions. This discussion climate is not inconsistent with the recent steady steepening and long-end rise of the Canadian curve: if the market thinks rate-cutting space is constrained, or that cuts may not ease real constraints, then medium- and long-term yields will struggle to decline.

Global cues mainly came from the U.K., the euro area and Australia. In the U.K., energy bills pushed up inflation on 8 August 19; news on 8 August 20 showed inflation rising to 2.9% and on 8 August 22 a combination of “fewer payroll recipients, rising inflation” appeared — a typical configuration of weakening growth with inflation not fully receding, which is not friendly for global long-term yields. In the euro area, the ECB’s consumer expectations survey published on 8 August 21 at least indicates inflation expectations remain a policy focus rather than having exited the radar. In Australia, reports of a weakening labor market reinforced growth slowdown signals, but market discussion remained about “what that implies for the rate path” and did not coalesce into a one-sided easing consensus.

On credit, the report can only be descriptive. U.S. high-yield spreads narrowed 2bp on the 20th, while investment-grade spreads widened 3bp on the 20th, both modest moves. Combined with the rise in long-term Treasury yields, this looks more like volatility driven by the risk-free rate than a credit panic-driven risk event.

II. Views on Future Rates

First clarify methodological boundaries. Directional judgments for the next three weeks can only be built on the portion of signals in the data that are labeled “proven”; terms and countries labeled “≈ random walk” are unpredictable and not to be forecast — for those, only coupon, roll, and static yield can be discussed. The meaning of carry must also be viewed separately: it is the expected return if yields do not move, a tailwind, not a directional prediction; if yields rise further, price losses can entirely offset carry.

Start with the U.S. short and mid end. U.S. 2-year, 5-year and 10-year signals are all labeled “proven,” but their direction labels are mixed. This does not mean a strong bullish or bearish tilt; it means momentum information at these maturities has some statistical value but currently provides no clear one-sided direction. For the next three weeks, one cannot say “definitely down” or “definitely up”; rather: U.S. short- and mid-end rates are likely to remain range-bound at high levels and highly sensitive to data and policy language, so it is inappropriate to adopt an overly strong one-sided narrative.

If forced to tier within the short end, the U.S. 2-year is more suitable for harvesting static yield than the 5-year or 10-year. The reason is not greater directional optimism but firmer carry+roll: U.S. 2-year annualized carry is 4.59, with a reliability of 0.326 and a long score win rate of 68%, making it the most robust among the proven tenors. U.S. 5-year carry is 4.723, reliability 0.178, win rate 59%, also usable. U.S. 10-year carry is 5.29, a higher absolute level but reliability only 0.112, clearly weaker than the 2-year and 5-year. For a three-week holding period, this means that if yields trade sideways, the 2-year and 5-year are more likely to realize coupon and roll; but if yields continue to rise, especially if policy-induced term premium increases, the longer the duration the larger the drawdown.

The U.S. long end is more definite: the 30-year direction is unpredictable and we do not make a call. The data labels US_30Y as ≈ random walk; although current dir=up, the hit rate 52.7% versus a baseline 51.3% does not provide enough edge to draw a directional conclusion. Only the static layer can be stated: U.S. 30-year has the highest carry, annualized 6.04, but reliability is only 0.077 and the long-side win rate 53%, marginally effective. In other words, long bonds’ coupon looks attractive only if yields don’t continue to rise; last week’s real-world news precisely demonstrated that fiscal supply, policy credibility and safe-haven status are exerting pressure on the long end, so high carry cannot be treated as a directional buy signal for long bonds.

For Canada, be more restrained on direction. Canadian 2-year, 5-year, 10-year and 30-year are all labeled ≈ random walk, so their directions for the next three weeks are unpredictable and no directional calls are made. The entire curve is currently at high one-year percentiles, especially 10-year and 30-year z-values at 2.25 and 2.49 respectively — this only indicates high position, not automatically that “they should come down.” Under these constraints, Canadian trade value mainly comes from static yield rather than direction: Canadian 30-year carry is 5.11 and 10-year is 4.55, both listed as proven carry assets, indicating that if yields don’t move, holding returns have some statistical support. But again, if the Canadian long end continues to rise as it has over the past 60 days, carry will quickly be consumed by price losses.

Across-country comparison: both the U.S. and Canada are experiencing a bear-steepening, but the difference is that U.S. news-driven moves are concentrated in long-end supply and policy communication while Canada looks more like a general upward re-centering of rates. In curve shape, Canadian 2s10s at +73bp is steeper than the U.S. +50bp. For the next three weeks, the notable issue is not which country will necessarily catch up or reverse, but that the sources of long-end risk compensation differ: U.S. leans toward fiscal and term premium, Canada toward whole-curve repricing. Lacking directional statistical advantage, cross-country relative-value calls should be muted; one can only say U.S. long-end narrative noise is larger and has more volatility sources, while Canada is closer to a simple high-level steep curve.

Turning to the curve itself: both the U.S. and Canada have returned to a normal slope, with Canada steeper. For medium-term holding, a normal slope implies roll-down conditions are generally adequate, especially the 2-year to 5-year segment, which can provide decent carry and benefit from rolling down the curve; this is why U.S. 2-year and 5-year static holding attributes beat the long end. In contrast, the long end while nominally offering higher carry faces larger duration exposure and a more unstable term premium. Last week the U.S. market already demonstrated this: an administrative operation could deliver a one-day gain that was fully erased by long-end rises within two days.

Synthesizing a three-week framework yields three points. First, keep the U.S. short- and mid-end understanding as “high-level oscillation, carry-dominant,” and avoid strong one-sided bets on direction. Second, U.S. 30-year and all Canadian tenors are unpredictable in direction and we make no calls; acknowledge their high carry but also acknowledge price volatility can overwhelm carry. Third, prioritize curve shape and holding returns in cross-country analysis; do not mistake “high level” for “turning point.”

III. Risk Warnings

The biggest risk remains the U.S. long end. Last week showed that technical arrangements like Treasury buybacks can only produce transient disturbances in long-end yields and cannot substitute for the fundamental pricing drivers of supply, inflation and term premium. If policy statements about “stabilizing the bond market” continue in the coming weeks while supply pressures and inflation concerns are not simultaneously eased, long-end rates may repeatedly spike higher and weigh on all duration assets.

The second risk is a bumpy path of disinflation. U.K. data already reflected an inflation rebound driven by energy, and euro-area consumer expectations remain a focal point. In the U.S., Fed officials’ language on inflation has not loosened. If oil prices, energy bills, or geopolitical factors again lift inflation concerns, the market’s current shaky pricing for the short and mid end could reorient toward higher rates.

The third risk is “carry illusion.” Currently the annualized carry on U.S. 30-year, Canada 30-year, U.S. 10-year, and Canada 10-year is not small, but this is the expected return only if yields do not move. For a three-week holding horizon, a rise of a dozen or so basis points can wipe out a substantial portion of coupon and roll-down, especially at the long end. High carry means a cushion, not directional superiority.

The fourth risk is weakening safe-haven correlation. Reports have already suggested a “reduced safe-haven role for Treasuries.” If risk-asset volatility rises but long bonds do not benefit in step, the traditional stock-bond hedge could break down and rate-asset drawdowns could be more direct than historical experience suggests.

This report is for research purposes and is not investment advice.