When Bond Yields Spike, Duration Decides: The September 2026 Selloff in Preferred Shares

September 30 made the point with unusual clarity: as long-term bond yields climbed toward 2007-era levels, no-reset perpetual preferreds drew down hard into the quarter-end — and then recovered into the close — while premium FixedResets barely budged. Same asset class, same day, wildly different sensitivity. The difference is duration, and understanding it explains most of what happened to preferred share prices last month.

The September tape in one paragraph

The US 10-year Treasury yield pushed above 5.2% — levels not seen since 2007 — and the 30-year traded toward 5.5% after hot inflation data and hawkish Fed commentary (the Fed raised rates a quarter point mid-month, with projections signaling more). Canada's 5-year yield, the benchmark that anchors reset arithmetic, printed as high as 3.68% on September 23. Against that backdrop, the longest-duration corner of the preferred market repriced fastest; names anchored by an upcoming reset barely moved.

Duration, quickly

Duration measures how much a price moves per unit change in yield: a modified duration of 8 means roughly an 8% price move per 100bp change in yield.

For a fixed-coupon bond, the formula is tidy. For a perpetual — a coupon with no maturity — the textbook approximations are the perpetuity formula (1+y)/y applied to the yield: at a 6.1% yield that works out to roughly 17 years. Real perpetuals print somewhat shorter than the textbook figure because of the par-call option and other features, and modified durations commonly estimated for the perpetual cohort land in the low-to-mid teens. Either way: the largest duration in the fixed-income complex, outside of very-long-dated bonds.

The September arithmetic illustrates: a 40–50bp rise in perpetual yield-to-worst through the month translates to roughly a 5–7% price drawdown on a 14-year-duration instrument — close to what below-par perpetuals actually did.

Why reset dates cap the damage

A FixedReset has a different anchor. Within a few years of its reset date, its price is pulled toward par by the reset: when the five-year term rolls over, the coupon reprices to GoC-5y + the issue's spread, and the holder's downside case converges to "collect par at reset or hold a repriced instrument." That structural pull caps how far a near-reset issue can de-rate — which is why premium FixedResets sitting above $25 with a reset inside two years trade like short-dated bonds, not like 14-year duration instruments.

Where the market priced exactly that at September month-end (trailing-indicated yields from our database):

Symbol	Structure	Price	Trailing yield	Duration story
BN.PF.D	Perpetual	$20.10	6.10%	Deep below par; no reset anchor; longest rate exposure
POW.PR.A	Perpetual	$23.97	5.84%	Near par but perpetual; highly rate-sensitive
NA.PR.K	FixedReset (premium)	$28.37	6.74%	Reset anchor holds it above par; acted short-dated
ENB.PF.V	FixedReset (discount)	$24.98	9.44%	Yield-dominated; recovery-driven, less rate-anchored

The pattern in the drawdown matches the structure. The names that fell hardest on the yield spike were the ones whose coupons don't reprice — and the ones that got bought hardest into quarter-end were the same cohort, because a 40bp yield move in two weeks is exactly what long-duration instruments do.

The after-tax lens

Dividend taxation changes the comparison set. For taxable Canadian accounts, the dividend tax credit means a preferred dividend yield compares against a pre-tax bond yield that is materially higher — a commonly cited working rule of thumb is that dividend income needs roughly a 1.3x gross-up to compare against interest income at the top marginal rate. At a 5.85% median perpetual trailing yield, that works out to roughly 7.6% interest-equivalent — against long Canadian corporates yielding around 5.4% late in the month. The spread is compensation for duration and call features, not a free lunch — but it is a real number worth carrying into any comparison.

What this means in practice

Rate-market volatility of the kind seen in September is, structurally, a duration event. Practices commonly discussed for managing it:

Know each holding's reset date. A premium reset six months away behaves nothing like a perpetual, even at the same headline yield.

Use yield-to-worst, not current yield, when comparing — the trailing number hides how much duration risk sits behind a big coupon.

Size for the drawdown, not the average. Long-duration names tack 10%-plus drawdowns onto 100bp of yield moves in stress episodes like September's; position sizes should reflect that arithmetic.

Ladder structures — mixing reset vintages spreads single-date rate exposure across the calendar.

These are structural observations about how the instruments behave, not a view on any particular issue. Our FixedResets vs Perpetuals deep dive covers the yield-gap mechanics, the yield-to-worst explainer covers the measurement, and the rankings page sorts the whole universe by the yield measure of your choice.

---

This article is for general information only and does not constitute financial advice. It is not investment advice and is not a recommendation to buy or sell any security. Preferred shares carry credit, interest-rate, and liquidity risk; past performance is not indicative of future results. Do your own research and consult a licensed professional before acting.