Federal Reserve rate hikes pressure the U.S. short end; Canada's entire yield curve retreated.

Interest Rate Weekly · 2026-09-14 to 2026-09-20

One-line summary: The main theme last week was clear: the Fed’s first rate hike in three years pushed U.S. front-end rates higher, but the 10Y and 30Y did not spike in the same week, causing the U.S. curve to bulge around the 2Y/3Y area; Canada’s long end fell back more noticeably, and policy divergence among the UK, Japan, and Eurozone further reinforced cross-country rate dispersion.

I. Review of Last Week

On the U.S. front, there was essentially one core event: the Fed was confirmed on 9/16 to 9/17 to have initiated a new rate-hiking cycle, and multiple reports mentioned that "there may be more hikes to come." The accompanying macro backdrop was consistent: Reuters noted strong U.S. retail sales and persistent economic resilience with building inflationary pressure; Warsh publicly pointed to inflation as the issue and interpreted rising bond yields as a sign of an overly strong U.S. economy. Looking at policy and data together, the market’s first reaction was a stronger repricing at the front end while the long end was relatively restrained.

Curve data and the news align perfectly. The U.S. 2-year yield closed at 4.67%, up 11bp over 5 days, up 48bp over 20 days, up 51bp over 60 days, z-value 2.62; the 5-year yield was 4.78%, up 3bp over 5 days, up 43bp over 20 days, up 51bp over 60 days, z-value 2.59. The front and belly have been at significantly high levels over the past one to two months. By contrast, the 10-year yield was 4.94%, down 1bp over 5 days, and the 30-year yield was 5.29%, down 8bp over 5 days. Connecting these points shows the U.S. curve did not shift up uniformly; instead the short end and belly continued to push higher while the ultra-long end steadied within the week. US 2s10s reached +27bp, and the positive slope persisted. The 9/20 report that "2-year and 3-year approaching 5%, 10-year holding near 5%, curve bulging" is essentially a verbal summary of this week’s market action.

That the U.S. long end did not follow the front end into a runaway move can also be traced in the news. On one hand, descriptions that "the Fed finally hiked and bond markets breathed a sigh of relief" indicate that prior market fears about policy lagging inflation were at least temporarily eased once the hike occurred. On the other hand, the 9/18 report that "U.S. Treasuries are becoming less attractive to foreign central banks and governments" is a reminder that the long end is not without risks: if official foreign demand continues to weaken, the ultra-long end will face supply-demand pressure going forward. This helps explain why the 30-year yield, though lower this week, still rose 35bp over 60 days and has a z-value of 1.98, so it is not at a low level.

On the Canada front, there were no comparably weighty domestic policy actions during the week; news focused more on real estate and mortgages. On 9/16 BMO noted that the phase of mortgage-rate relief in Canada had ended, pointing to a still-tight household financing environment. The curve’s behavior diverged sharply from the U.S. during the week: the Canada 2-year yield was 3.27%, down 4bp over 5 days; the 5-year was 3.54%, down 9bp over 5 days; the 10-year 3.83%, down 11bp over 5 days; the 30-year 4.17%, down 11bp over 5 days. In other words, the whole Canadian curve fell within the week, with larger declines at the long end, keeping CA 2s10s at +56bp and making the curve steeper than the U.S. But looking over a longer horizon, Canada still shows general increases over 20 and 60 days: the 2-year is +29bp over 20 days and +46bp over 60 days; the 10-year is +14bp over 20 days and +40bp over 60 days, which suggests the intraweek decline is more of a correction after prior rises rather than a trend reversal.

On the global front, policy divergence became more pronounced than in the prior phase. The Bank of Japan hiked on 9/20, indicating that Japan is not the only major central bank staying on the tighter side. The UK was reported on 9/17 to 9/18 as "holding rates at 3.75%," and market feedback included "borrowing costs fell after pausing gilt sales," meaning the UK chose to stand pat and even delivered marginal easing via balance-sheet operations. The Eurozone had no direct rate decision, but the 9/18 ECB consumer expectations survey was released, showing Europe continues to track inflation expectations and household behavior for policy space. Putting a Fed hike, a BoJ hike, and the UK standing pat together shows clear differences in the G3 policy rhythm; cross-market spreads and exchange-rate constraints will continue to affect long-end pricing in each country.

On credit markets, only descriptive comments are possible. U.S. HY OAS was 2.70%, tightening 5bp over 20 days; IG OAS was 0.78%, tightening 4bp over 20 days. Given that credit spreads did not widen during the Fed’s hike week, credit markets have not yet interpreted policy tightening as an immediate hit to growth and defaults; risk appetite has not visibly deteriorated. But these two credit series have only 3.1 years of history, providing a limited statistical basis, so stronger conclusions are unwarranted.

II. Views on Future Rates

First, clarify the methodological boundaries. Directional judgments for the next three weeks can only be made for U.S. 2Y, 5Y, and 10Y where momentum signals are marked as "proven" in the data; Canada’s maturities and the U.S. 30Y are close to random walks, so directions are unpredictable and no judgment is offered. Carry represents expected return "if yields do not move" and is a tailwind, not a directional forecast; if yields instead rise, price losses will still overwhelm carry.

Start with the short end in the U.S. The momentum evidence is strongest for the U.S. 2Y and 5Y. US_2Y is labeled [proven] with dir=up, hitting 56.2%, significantly above the 50.5% baseline; US_5Y is also [proven] with dir=up, hitting 53.9%, above the 50.7% baseline. Combined with spot moves—2Y up over 20 and 60 days by more than 48bp, z-value 2.62; 5Y up about 51bp over 20 and 60 days, z-value 2.59—and given the Fed just started the first hike cycle in three years and communicated "there may be more hikes," these points support the view that the U.S. front end and belly will continue to face upward pressure over the next roughly three weeks. In other words, the directional bias remains "higher rates, lower bond prices," especially for the 2Y and 5Y.

The 10-year U.S. Treasury can also offer a directional view, but with less conviction than the front end. US_10Y is likewise [proven] with dir=up, hitting 52.7%, only slightly above the 50.5% baseline, so the edge is modest. In spot terms, the 10Y is up 29bp over 20 days and 44bp over 60 days, z-value 2.48, so the trend remains; however, the recent 5-day modest decline of 1bp shows the long end has not been pushed immediately by policy like the front end has. Based on the evidence level, one can retain the view that the 10Y has a slightly higher probability of rising over the next three weeks, but with lower conviction than the 2Y and 5Y. For trading, this is more like "follow the trend but lower confidence," not treating the 10Y as the cleanest directional bet.

The 30-year U.S. Treasury offers no directional call. US_30Y is statistically close to a random walk, and the data explicitly demand no directional prediction. What can be said is static yield characteristics: US_30Y annualized carry+roll is 5.815, highest in the sample, but corr is only 0.077 and the long-only win rate is 53%, so its edge is marginal. The implication is clear: if the 30Y yield remains sideways over the next three weeks, holding ultra-long bonds will provide relatively high coupon and roll-down returns; but if foreign demand weakens, supply pressure increases, or markets continue to re-price long-term inflation and term premium higher, price drawdowns could easily swallow that carry. The ultra-long end is currently only a "high-carry" asset, not a directionally certain long.

Canada’s entire curve cannot provide directional judgments. CA_2Y, CA_5Y, CA_10Y, and CA_30Y are all labeled as close to random walks, so directions over the next three weeks are unpredictable and no judgment is offered. The intraweek Canadian yield decline did contrast with the U.S., but that should be seen as a one-week behavior and cannot be extrapolated into a tradable directional signal. Looking only at static carry, Canada’s long-end carry is not negligible: CA_30Y annualized carry 4.926, CA_10Y 4.41, and both are marked as "proven" win-rate ranges, indicating holding them when yields do not move provides some buffer. The problem remains that if the North American rate center continues to lift, Canadian bonds will also face price pressure; carry can only reduce the felt volatility, not offset a directional miss.

Cross-country divergence is worth continued monitoring. The U.S. front-end up signal is the clearest, Canada has no directional edge, the UK is standing pat, and Japan is hiking. This mix suggests that over the next three weeks global rates are more likely to show "different countries and different maturities moving independently" rather than a highly synchronized easing or tightening cycle. For medium-term positioning, monitoring relative value across curve segments and countries will be more important than a single-duration exposure, because pure duration bets can easily pick up country and maturity noise.

On curve shape, the U.S. is closer to a continuation of "bear steepening" or a "belly bulge." 2s10s is already +27bp; the short end and 5-year segment remain more vulnerable after the Fed hike, while the 30-year fell within the week, indicating the most direct policy impact is concentrated at the policy-sensitive end. If US 2Y, 5Y, and 10Y follow their proven momentum and rise, with the front end stronger than the long end, the U.S. curve will most likely keep a positive slope and could even see 2s10s steepen further. Canada’s 2s10s at +56bp means its curve is already steeper, but because directions at each maturity are unpredictable, we can only describe its shape as "relatively steep" without judging whether it will steepen further or flatten.

Putting carry into portfolio language, the U.S. 2Y still offers the most balanced cost-benefit. US_2Y annualized carry 5.21, corr 0.326, long-win rate 68%, making it the most reliable and high-carry option among U.S. predictable tenors; US_10Y carry 5.26, win rate 56%, US_5Y carry 4.963, win rate 59%. But the boundary must be repeated: these win rates and carry figures are better interpreted as explaining "if the trend does not reverse, holding results in lower losses or even tailwind" rather than direct buy recommendations. Because the directional signals for US 2Y and 5Y are for rising rates, static carry is not friendly to long positions and implies some negative carry for longs while short positions would earn carry.

III. Risk Warnings

First, the Fed has just started a new hiking cycle, and market repricing of terminal rates and pace can easily be over-interpreted at the front end. Although U.S. 2Y and 5Y show upward directional signals, their win rates are only in the 53.9%–56.2% range, not an overwhelming one-sided edge; short-term pullbacks and event-driven rebounds are normal.

Second, there is substantial disagreement in the U.S. long end. One side sees 10Y momentum still biased up, while the other sees the 30Y as directionally unpredictable and it already declined within the week. The 9/18 report about declining foreign central bank and government interest in Treasuries, if more widely priced in, could raise long-term term premia; but if the market interprets the Fed’s hike as credible anti-inflation action, the long end could remain more stable than the front end. Long-end volatility has more sources, so a single logic is insufficient.

Third, Canada and the UK showed relatively milder moves versus the U.S. during the week, but statistically they lack directional edges. Mistaking a one-week divergence for a trend continuation risks misreading random noise. Especially in Canada, the whole curve is still up over 20 and 60 days while only down over the last 5 days; this "medium-term rise, short-term pullback" structure is prone to creating direction illusions.

Fourth, credit spreads are currently tightening, indicating risk assets have not treated higher policy rates as an imminent stressor. If growth expectations weaken later, credit spreads and risk-free rates may decouple, and correlations inside the bond market will change. Current credit data have a short history and are suitable only for monitoring sentiment, not for robust forecasting.

This report is for research purposes and not investment advice.