Duration and credit-spread risk: reading a bond fund's two sensitivities

A bond fund's fact sheet usually gives you a yield and an expense ratio. Neither tells you the thing you most want to know before you buy: when the market moves, how much does this fund move with it? Two numbers answer that — one for interest rates, one for credit — and both now sit in a fund's regulatory filings where you can look them up. This is a guide to what they mean, how to use them, and where they stop being trustworthy.

Where these numbers come from

Every SEC-registered fund with a meaningful bond allocation files a portfolio report, Form N-PORT, and part of that filing (Item B.3) reports the fund's sensitivity to interest rates and to credit spreads. These figures have been part of the public record since funds began reporting on N-PORT in 2019, and the SEC publishes them roughly 60 days after each fiscal quarter ends.

They are worth knowing about precisely because almost nobody reads them. The disclosure is standardized — a Vanguard index fund, a PIMCO active fund, and a two-person boutique all answer the same question in the same units — which is what makes it possible to compare funds honestly. What is relatively new is not the disclosure itself but that it has become accessible in a form you can actually use fund-by-fund. Treat it as a recent and underused addition to the numbers you already check.

One threshold matters up front: a fund only files these figures when at least 25% of its portfolio is in debt. An equity fund, or a fund with a token bond sleeve, files nothing here — and that blank is meaningful, which we return to below.

Interest-rate risk: what duration tells you

The question it answers: how much does this fund's value move when interest rates move?

Funds report the dollar impact of a one-basis-point (0.01%) change in rates. Relative to the fund's size, that translates into its effective duration, expressed in years. The reading is direct: a duration of 6 means a 1% rise in rates costs roughly 6% of the fund's value, and a 1% fall gains roughly 6%.

How to read the number:

• 0–2 — rate moves barely register. Cash-like, floating-rate, or very short-term funds live here.
• 4–7 — a typical core bond fund. A total-bond-market index fund tends to sit around 5–6.
• 10 and up — long-duration funds, which are genuine bets on the direction of rates. A 20-plus-year Treasury fund runs around 15–16.
• Negative is real, not an error. A fund can be positioned to gain when rates rise — some managed-futures and rate-hedged funds carry a negative duration on purpose.

Duration is a single number summarizing a whole curve, and that summary hides something: rates at different maturities do not move together. The short end and the long end regularly move in opposite directions. Two funds with the same total duration can be exposed to entirely different parts of the curve, so the five-point breakdown (3 months, 1 year, 5 years, 10 years, 30 years) is worth looking at when it is available — the total alone can make two very different funds look identical.

Credit-spread risk: the risk duration hides

The question it answers: how much does this fund move when credit spreads move — when the market demands more compensation for lending to riskier borrowers?

A bond's yield has two parts: a risk-free rate, plus a credit spread for the chance the borrower runs into trouble. Duration measures sensitivity to the first part. Credit- spread duration measures sensitivity to the second, and the two can be wildly different for the same fund.

The clearest example is a floating-rate fund. Its bonds reset their interest rate constantly, so its rate duration is close to zero — on a duration-only view it looks almost riskless. But the credit spread on those bonds does not reset, so its spread duration can be that of a full bond fund. Funds built around AAA-rated CLOs are the textbook case: recent filings show one large CLO ETF with a rate duration near 0.03 and a spread duration above 4 — in a credit selloff it behaves like a real bond fund, and its own prospectus says that is exactly what it is designed to do. Duration alone does not just miss this risk; it actively points the wrong way.

The mirror case is two funds that look identical on duration and are not. A total-bond index fund and a multi-sector income fund can carry nearly the same rate duration while one holds almost entirely investment-grade credit and the other holds a large slug of high-yield. In a credit selloff — spreads widening as the market reprices risk — the first is roughly flat and the second falls several percent. Nothing on a standard fund page distinguishes them; the credit-spread number does.

Credit-spread risk is reported with the same five-point maturity breakdown, split further into investment-grade and high-yield exposure. That split is the story: two funds with the same total spread duration can hold it in short-dated investment-grade paper or in long-dated high-yield, and those two portfolios behave nothing alike in a selloff.

Put duration in context: how much of the fund is even bonds?

This is the caveat that matters most, and it is the easiest to forget once you have a tidy number in front of you. Duration is a portfolio-level average, scaled to the whole fund. The bond sensitivity is measured across the fund's debt and then divided by the fund's total size.

The practical consequence: for a fund with only a sliver of its assets in fixed income, none of this really matters. A fund that is 1% bonds and 99% stocks will show a duration near zero — not because its bonds are insensitive to rates, but because they are a rounding error in a fund driven by equities. Reading deeply into a duration of 0.2 for such a fund tells you nothing; the fund's fate is decided elsewhere. (In fact, a fund that far below the 25%-debt threshold files nothing here at all.)

These numbers earn their keep for funds where bonds are the story — core bond funds, credit funds, multi-sector and floating-rate funds, anything where a large share of the portfolio is in fixed income. There, duration and spread duration are close to the whole risk picture. So before you interpret either number, ask what fraction of the fund is actually in bonds. The same duration means something very different in a dedicated bond fund than in a balanced fund that happens to hold some.

These are approximations — and that is fine, within limits

Duration and spread duration are linear approximations to a curved relationship, and knowing that changes how far you should push them.

A bond's price and its yield are not related by a straight line. The true relationship is a curve that bows the right way for an ordinary bond — that curvature is called convexity. Duration is the slope of that curve at today's yield: the first-order, straight-line estimate of how price responds to a small change in rates. For small moves, the straight line hugs the curve and duration is an excellent estimate. For large moves, the line drifts away from the curve.

Because the curve bows in the bond-holder's favor, the straight-line estimate is pessimistic on both sides: it slightly overstates the loss when rates rise a lot and slightly understates the gain when they fall a lot. Convexity is the second-order term that corrects for it. Some bonds bow the other way: mortgage-backed securities and callable bonds have negative convexity — prepayments and calls cap the gain when rates fall and deepen the loss when they rise — so for a fund built around them the straight-line estimate flatters the downside, understating losses rather than overstating them. That caveat lands on exactly the mortgage-heavy funds where these numbers carry the most weight. The takeaway is not that duration is wrong — it is a reasonable and widely used approximation — but that it is most accurate for modest moves and should be read as an estimate, not a promise, for large ones.

This has a concrete effect on scenario numbers. Duration answers a one-basis-point question directly, because that is what the fund files. Any figure for a big move — "what if rates rise 1%?" or "what if spreads widen 1%?" — is that small number scaled up, which is exactly where the linear approximation is weakest. For interest rates, funds also file a separate 100-basis-point figure, so the curvature is at least partly captured. For credit spreads, only the one-basis-point number is filed, so any 1% scenario is a hundred-fold extrapolation and tends to overstate the loss. Read those scenario figures as "roughly 5%," never as a precise "5.07%."

How not to over-rely on these numbers

Used in context, both numbers are genuinely useful. A few honest limits keep them from being used for more than they can bear:

• The data is a few months old. You are seeing a quarter that ended weeks ago. This matters more for some facts than others — a fund's credit-spread duration is very stable from quarter to quarter, so the shape you see is close to the shape you get, but the underlying holdings drift. Always check the "as-of" date, and treat these as the fund's recent posture rather than its position this morning.
• The signs are meaningful — don't discard them. These are signed quantities. A negative duration is a deliberate bet on rising rates; a negative spread duration is a credit hedge. They are not errors to be cleaned up, and taking the absolute value of a fund with offsetting positions overstates its real risk.
• Blank is not zero. A fund with no number here did not report a risk of zero — it is simply below the 25%-debt threshold and files nothing. A reported zero (from a fund that does file) is a real "we measured this and it is negligible." The two look alike and mean opposite things.
• The scenarios assume the curve moves in parallel. They add up the response at each maturity as if rates or spreads shifted by the same amount everywhere. Real curves steepen and flatten, which is why the five-point breakdown is worth more than the single headline when you are trying to understand where a fund's risk sits.
• Filings contain occasional errors. These are real regulatory disclosures filed by thousands of funds, and a handful contain scale typos and implausible values. A duration in the thousands of years is a filer's mistake, not a fund. Sanity-check anything extreme against what the fund actually holds.

None of this argues against using the numbers. It argues for using them the way they are built to be used: as a standardized, comparable read on a bond fund's two main sensitivities, best trusted for small moves, interpreted in light of how much of the fund is bonds, and read alongside the maturity-and-quality breakdown rather than as a single figure. For the funds where fixed income is the story, that read is hard to get anywhere else — and it has been sitting in the filings all along.