US short- and medium-term yields are expected to rise, but the direction of the long end and of additional bond issuance is unclear.

Interest Rate Weekly · 2026-08-31 to 2026-09-06

One-line summary: Last week the global rates narrative was clear — US data continued to support pricing of "growth not weak, inflation pressures not dissipated," the Bank of Canada held rates steady and emphasized inflation concerns, global long bonds remained in high-level oscillation in the wake of sell-offs, and the curve continued to steepen overall.

I. Review of Last Week

On the US front, the core story is "employment recovering, wages lagging inflation, the Fed still constrained by inflation." On September 4 multiple outlets reported US August nonfarm payrolls increased by 162,000, a rebound in hiring momentum that refocused market attention on how the Fed will handle inflation next. Reuters on September 3 also noted the US labor market remains stable and that rising service-sector input prices point to elevated inflationary pressure; multiple outlets between September 3 and 4 similarly framed coverage around "moderate economic growth but persistent price pressure." At the same time, The Washington Post on September 5 and NBC on September 4 both pointed out that wage growth continues to lag inflation, implying real purchasing power has not kept pace with nominal employment improvements.

The rate market response to this information was fairly uniform: the short and belly continued to rise, and the long end also moved up from high levels but by a smaller amount. Curve data show US 2-year yield at 4.34%, up 14bp over the past 5 days, up 9bp over 20 days, up 21bp over 60 days; 5-year 4.52%, 5-day +14bp; 10-year 4.77%, 5-day +10bp; 30-year 5.25%, 5-day +6bp. In other words, the whole US Treasury curve is moving higher, but the front and belly are rising faster — 2s10s reached +43bp — which typically reflects the market pricing in "policy rates staying higher for longer."

Two background factors last week amplified the market’s perception of rising Treasuries. First, on September 2 New York Fed President Williams publicly linked higher long-term yields to a strong US economy and mentioned large-scale AI-related investment. Such remarks provided no new numbers but reinforced market views that "higher yields do not necessarily mean bad news and at least partly reflect growth resilience." Second, Reuters on September 3 noted that this round of bond selling may have been amplified by a relatively obscure economic interest-rate variable; on September 4 there were further reports pointing to large institutional reductions in global government bond holdings. For trading, this implies that the long end was not driven solely by macro data; positioning and supply-demand dynamics also heightened volatility.

The response in US mortgage financing also highlights a detail: on September 4 there were reports discussing why mortgage rates barely fell after the stronger-than-expected jobs report, and on September 3 an article noted the spread between the 10-year Treasury and 30-year mortgage rates remained stuck at around 2 percentage points. This indicates that even if the market occasionally trades on a "rates peaked" impulse, transmission to household financing costs is not necessarily smooth and financial conditions are not clearly loosening.

The Canada story is simpler than the US but not easier. On September 2 the Bank of Canada left rates unchanged, citing renewed inflation concerns. On the curve, Canadian rate increases were not smaller than the US: 2-year 3.10%, 5-day +14bp, 20-day +18bp, 60-day +25bp; 5-year 3.41%, 5-day +11bp; 10-year 3.79%, 5-day +9bp; 30-year 4.16%, 5-day +3bp. Canada 2s10s reached +69bp, a steeper slope than the US, indicating markets are likewise pricing "tight for longer," and the long end did not visibly pull down as a flight-to-safety response relative to the front.

The global picture was characterized by "Japan moving up, UK corporates lacking confidence in inflation control, and global bond markets still adjusting to higher yields." Reports on September 2 showed Japan's 10-year JGB yield rose to 3.02% and the 30-year to 4.18%, with markets describing Japan's bond market repricing since YCC exit as ongoing. On the UK side, Bloomberg on September 3 noted firms were not convinced by plans to reduce inflation. The BBC on September 2 likened this round of global bond market volatility to a "wildfire" that policy makers across countries could not ignore. These reports, together with the US and Canada curves, form a common backdrop: major economies have not entered an environment that would allow long yields to fall steadily.

In credit markets, risk appetite did not materially deteriorate. US high-yield spreads narrowed 6bp over 20 days, while investment-grade spreads widened 3bp over 20 days — overall this looks more like rate-driven than credit-driven pricing for the week. For rates trading, this combination typically means the rise in nominal yields is coming more from higher risk-free rates rather than sudden economic weakening that would create credit stress.

II. Views on Future Rates

First, clarify the methodological boundary. Directional judgments for the next 3 weeks can only be based on validated momentum signals. In the current dataset, only the US 2-, 5-, and 10-year tenors have direction signals labeled as "validated"; the US 30-year and all Canadian tenors are labeled as near-random walk. Correspondingly, we can give directional judgments for the US short and belly; the US ultra-long end and the full Canadian curve are not suitable for directional predictions and can only be discussed in terms of static yields, curve structure, and cross-country relative shapes.

Start with the US short end. Signals for the US 2-year and 5-year both point to further yield increases. The 2-year is currently 4.34%, 1-year z-score 1.90, up 14bp over 5 days and up 21bp over 60 days; the validated signal gives dir=up, hit rate 56.2%, above the base rate 50.4%. The 5-year is currently 4.52%, z-score 2.09, up 14bp over 5 days and up 26bp over 60 days; validated signal dir=up, hit rate 53.9%, above base rate 50.7%. Combined with last week's news, this directional view is consistent with fundamentals: nonfarm payrolls rebounded, service-sector price pressures are elevated, and Fed officials have explained higher long yields as a sign of economic strength, so the market will naturally continue to push down near-term rate-cut expectations or at least tolerate a longer period of tight policy.

The US 10-year also has a validated upward signal. It is currently 4.77%, 5-day +10bp, 60-day +24bp, z-score 2.11; directional hit rate 52.7%, above base rate 50.6%. This implies that over the next roughly 3-week positioning window, the probability of further upside in the US 10-year is tilted higher. However, be more cautious here than for the short end because the 10-year already embeds term premium, supply, and global allocation factors; the press reports about bond selling being amplified by technical factors serve as a reminder. Directionally it can still be biased toward higher, but certainty is lower than for the 2- and 5-year.

The US 30-year does not allow a directional judgment. Data explicitly label its directional forecast as close to a random walk, and the hit-rate edge is not statistically persuasive, so the direction is unpredictable and no directional call should be made. What can be discussed is carry: the US 30-year annualized carry + roll expected return is 5.97, the highest in the given ranking, but its associated reliability is only 0.077 and the "marginal" tag already indicates weak stability. More importantly, carry simply means the static return if yields do not move; it is a tailwind, not a directional signal. Once yields continue to rise, long-duration price losses can easily outweigh that carry, so high carry alone should not be treated as a buy signal. For the 30-year, it is more meaningful to view it as a high-volatility, high-duration exposure rather than a stable coupon-like instrument.

For Canada, directions for the next 3 weeks are unpredictable and no directional calls are advised. The 2-, 5-, 10-, and 30-year are all labeled as near-random walk, meaning current momentum evidence cannot say whether yields are more likely to go up or down. Two things can be confirmed: first, curve levels are elevated — the Canada 10- and 30-year 1-year z-scores are both at 2.26, indicating yields are at notably high positions relative to the past year; second, static carry is not bad, notably 30-year annualized carry 5.06 and 10-year 4.55. But again, carry only provides positive returns when yields are flat or move mildly; if Canadian rates continue to rise, price drawdowns will still overwhelm carry. The Bank of Canada standing pat and restating inflation concerns provides an environment conducive to higher yields, but the data do not authorize a directional bet.

On cross-country divergence, the US and Canada currently look more like "same direction but asymmetric evidence." Both countries’ curves are rising and positively sloped, indicating markets are not trading recession-style rate cuts; but the US short and belly have validated upward momentum while Canada does not. This will make relative-value judgments over the next 3 weeks tilt toward "US direction is clearer, Canada is more about static yield and relative spread changes." If this framework continues, US front-end upward pressure is more likely to form sustainable pricing than Canada’s front end, because last week US payroll and service-price information directly reinforced Fed constraints while Canada's most important event was the central bank holding steady, which added relatively limited incremental information.

On curve shape, both the US and Canada are maintaining normal positive slopes, and Canada is steeper. US 2s10s is +43bp, Canada 2s10s is +69bp. Over the three maturities, this positive slope implies "holding the belly and long end offers coupon and roll-down support," especially given US 10-year carry 5.27 with win rate 56%, US 5-year carry 4.82 with win rate 59%, US 2-year carry 4.8 with win rate 68% — all reasonable. But ranking cannot replace direction: with US 2-, 5-, and 10-year directional signals all pointing up, even if carry is high, further yield increases could produce price losses that erode or exceed static returns. Shorter duration has the advantage of higher carry reliability and smaller price volatility, so under a "rates biased upward" framework the front end typically endures less passive pressure than the long end.

Compressing these points into the next 3 weeks’ watchlist yields three items. First, the US short and belly remain the main battleground; markets will continue to adjust the Fed path around "whether employment remains resilient, why wages lag inflation, and whether service prices stay elevated." Second, the long end is more affected by supply, positioning, and global allocation disturbances, and macro alone cannot explain all moves, so avoid making deterministic directional calls on the US 30-year and Canadian long end. Third, the global backdrop of rising Japanese long yields and high-level oscillation in European and US bond markets persists, meaning there is no clear short-term catalyst for a broad decline in global term premia.

III. Risk Warnings

The primary risk over the next 3 weeks remains nonlinear news shocks to the long end. Two types of signals appeared last week: one is US political figures continuously pressuring the Fed and linking trade threats with demands for rate cuts; the other is reporting of large funds reducing government bond holdings. These items may not stably alter policy paths but are sufficient to amplify trading volatility, especially for the 30-year.

The second risk is that the validated US upside momentum itself only carries a statistical edge, not high certainty. While the hit rates for the 2-, 5-, and 10-year are above base rates, the edge is not dramatic, and any materially weaker data could quickly reverse rates. Directional judgments should be understood as probability tilts, not single deterministic paths.

The third risk is misreading carry as a safety cushion. US and Canadian belly and long-end carry may look attractive now, but that holds only if yields are flat or move mildly. If global bond selling continues, duration losses will erode carry faster than it accrues, especially on the ultra-long end.

The fourth risk is cross-market contagion. Rising Japanese long yields, shaky UK inflation control expectations, and global markets still digesting higher nominal rates can transmit to US and Canadian bonds through term premia and asset-allocation channels. Even if domestic data are not surprising, global long ends can be carried higher by external yields.

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