Did duration predict it? Bond ETFs through a three-week rate swing
Treasury yields jumped in late September 2026 and then partly fell back. We checked how large bond ETFs actually moved against what their durations implied—and what to look at when the simple rule misses.
Our guide to what duration tells you about a bond fund gives a rule of thumb: a fund's duration is roughly the percentage it loses if interest rates rise by one percentage point. A fund with a six-year duration would lose about 6%.
Between September 22 and October 15, 2026, Treasury yields rose sharply for two weeks and then partly fell back. That makes a good real-world test of the rule.
The rule of thumb, on one fund
The iShares 20+ Year Treasury Bond ETF (TLT) reported a duration of 15.1 years. From September 22 to October 5, the 10-year Treasury yield rose 0.35 percentage point.
15.1 × 0.35 ≈ 5.3% expected loss. TLT lost 5.30%.
When the 10-year yield then fell 0.21 point by October 15, the same arithmetic pointed to a gain of about 3.2%. TLT gained 3.7%.
The match in the first period is closer than a rule of thumb deserves. It leaves out the interest the fund earned, and TLT's bonds mature well beyond 10 years, where yields rose a little more. Here those two simplifications happened to cancel out. A fuller estimate that accounts for both was a 5.19% loss: still close.
What rates did
| Sep 22 | Oct 5 | Oct 15 | |
|---|---|---|---|
| 2-year Treasury yield | 4.71% | 4.84% | 4.77% |
| 10-year Treasury yield | 4.96% | 5.31% | 5.10% |
| 30-year Treasury yield | 5.29% | 5.66% | 5.43% |
| High-yield credit spread | 2.68 | 3.12 | 3.23 |
The credit spread is the extra yield, in percentage points, that high-yield corporate bonds paid over Treasuries.
Two things to notice. Long-term yields moved more than short-term ones. And high-yield spreads kept widening even after Treasury yields started to fall.
Treasury funds: longer duration, bigger move
For each fund, the estimate combines three things: the fund's duration, the change in Treasury yields at the maturities its bonds actually hold, and the interest it earned over the period. The actual figure is the ETF's total return, including distributions.
| Fund | Duration (years) | Rates rising: estimate | Rates rising: actual | Rates falling: estimate | Rates falling: actual |
|---|---|---|---|---|---|
| SHY | 1.9 | −0.04% | 0.00% | +0.17% | +0.26% |
| IEI | 4.3 | −0.82% | −0.82% | +0.74% | +0.81% |
| IEF | 7.0 | −1.81% | −2.12% | +1.34% | +1.61% |
| TLH | 11.8 | −3.87% | −4.29% | +2.59% | +2.93% |
| TLT | 15.1 | −5.19% | −5.30% | +3.45% | +3.71% |
Rates rising: September 22 to October 5. Rates falling: October 5 to October 15. All five are iShares Treasury Bond ETFs, holding Treasuries that mature in 1–3 years (SHY), 3–7 (IEI), 7–10 (IEF), 10–20 (TLH) and 20+ years (TLT).
The funds lined up in duration order: the longer the duration, the bigger the move, in both directions. The largest miss was about 0.4 percentage point, on a move of more than 4%. Duration gave the right direction and roughly the right size. It is an estimate, not a precise forecast.
The broad iShares Core U.S. Aggregate Bond ETF (AGG), with a 5.8-year duration, behaved the same way: it lost 1.86% against an estimate of 1.73%, then gained 1.36% against 1.14%.
Lesson 1: use the rate that matches the fund
The quick version of the rule multiplies duration by the change in the 10-year yield. That worked for the intermediate and long-term funds. It did not work for the short-term one.
The iShares 1-3 Year Treasury Bond ETF (SHY) has a 1.9-year duration. Using the 10-year yield's 0.35-point rise, it should have lost about 0.5% after income. It was flat.
SHY holds bonds that mature within three years, and the 2-year yield rose only 0.13 point. Using that rate instead points to a loss of about 0.1% after income, much closer to what happened. The same duration means different things depending on which part of the yield curve moves.
Lesson 2: for corporate bond funds, check credit-spread duration too
The iShares iBoxx $ High Yield Corporate Bond ETF (HYG) has an interest-rate duration of only 1.7 years. On rates alone, it should have lost about 0.3% while yields rose. It lost 1.71%.
The rest came from credit spreads, which widened 0.44 point over those two weeks. HYG's credit-spread duration was about 2.9 years. Adding that effect, plus the interest earned, brings the estimate to a 1.37% loss, close to the actual result.
When Treasury yields fell, spreads kept widening, and HYG gained only 0.19%. Its rate duration alone made it look like a fairly stable fund. Its spread duration told the rest of the story.
Lesson 3: mortgage funds can move more than their duration says
The iShares MBS ETF (MBB) has a 5.4-year duration. The estimate was a 1.42% loss as rates rose and a 1.07% gain as they fell. MBB actually lost 2.18% and then gained 1.68%, about one and a half times the estimate in both directions.
Homeowners can refinance or pay off their mortgages early, so how sensitive these bonds are to rates shifts as rates move. For mortgage funds, treat the reported duration as a low estimate.
Two funds that didn't follow Treasuries
- Municipal bonds have their own supply and demand. The iShares National Muni Bond ETF (MUB) lost 1.32% as rates rose, less than the 1.72% a Treasury-based estimate implied, then gained 1.34% against an estimate of 1.25%.
- Floating-rate funds have almost no duration, because their interest payments reset with rates. The WisdomTree Floating Rate Treasury Fund (USFR) gained about 0.14% in each period, roughly its income, whichever way yields moved.
Try it on a fund you own
- Find the durations. A bond fund's page on this site shows its interest-rate duration and its credit-spread duration side by side. TLT's fund page shows 15.1 years.
- Pick a move. For example, rates rising by half a percentage point.
- Multiply. A fund with a six-year duration would lose roughly 6 × 0.5 = 3%, less whatever interest it earns in the meantime.
- Ask three questions before trusting the number:
- Which rates does it hold? Short-term funds follow short-term yields, not the 10-year.
- Does it own corporate bonds? Then spreads matter too, so check credit-spread duration.
- Does it own mortgages? Then expect larger moves than its duration suggests.
Three weeks is a short window, and the next rate move will not look exactly like this one. But it shows what duration can tell you, and where you need to look further.
How we did this
Estimate. Estimated return ≈ income − (rate duration × change in Treasury yields) − (credit-spread duration × change in credit spreads). The rate term uses each fund's reported sensitivity at five points on the yield curve (3 months and 1, 5, 10 and 30 years), matched to the Treasury yield change at each point. The spread term uses the ICE BofA US Corporate spread for investment-grade sensitivity and the US High Yield spread for high-yield sensitivity. It is applied to corporate and core bond funds only, not to Treasury, mortgage or municipal funds. Income is each portfolio's yield to maturity times the days in the period divided by 365. It ignores fund expenses. USFR has no computable yield in our data, so we assumed 4.2%.
Data. Durations come from Item B.3 of each fund's most recent Form N-PORT report in our data: May 31, 2026 for every fund shown. Treasury par yields are from the U.S. Department of the Treasury, and credit spreads are from the Federal Reserve Bank of St. Louis (FRED). Total returns are from dividend-adjusted closing prices (Sharadar). The funds shown are a selection of the largest bond ETFs in our data across fund types. Funds driven mainly by non-US rates, convertibles or loans are not discussed.
Limits. ETF prices and Treasury yields are recorded at slightly different times of day, and an ETF's price can drift from the value of its holdings, especially in credit funds. Durations describe each portfolio on its report date and may have changed since. These are illustrations of how the measures behaved, not forecasts or recommendations.