The G-Spread: Better Than Its Reputation, Until the Cash Flows Change
The G-spread is the simplest credit spread measure in the curriculum: a bond’s yield to maturity minus the yield of a government bond of the same maturity. The G is for government.
It is also the one most often dismissed, usually with the claim that it ignores the shape of the yield curve and is therefore crude. That criticism is technically correct and practically misleading. For a plain bullet bond the G-spread lands within a handful of basis points of the more sophisticated Z-spread even on a steep curve, and the numbers below show it. Where it genuinely breaks is somewhere else entirely, and knowing which case you are in is the whole skill.
Key Takeaways
- The G-spread is a bond’s yield to maturity minus an interpolated government yield at the same maturity.
- Interpolation is required because governments do not issue at every maturity, so the benchmark is constructed rather than quoted.
- For a plain bullet bond it lands within roughly 6 basis points of the Z-spread even on a steep curve, so the usual criticism overstates the problem.
- It fails badly when cash flow timing differs from a bullet: on an amortiser it missed a 200 basis point spread by 126 basis points one way and 61 the other.
- Curve shape largely cancels between the two yields. Cash flow timing does not, and that is the real limitation.
- With an embedded option, use the OAS. For a callable bond OAS is below the Z-spread; for a putable bond it is above.
What the G-spread is
Take the yield to maturity of a corporate bond and subtract the yield to maturity of a government bond with the same maturity. The difference, expressed in basis points, is the G-spread.
G-spread = YTMcorporate − YTMgovernment, same maturity
the government yield is interpolated where no bond of exactly that maturity exists
What it is intended to isolate is everything the government bond does not carry: credit risk, liquidity, and whatever else the market charges for holding this issuer rather than the sovereign. The government bond is treated as the risk-free reference, so the spread is read as the compensation for taking on the rest.
Two practical wrinkles sit inside the phrase “the government bond”. The first is that the most recently issued bond at a given maturity, the on-the-run issue, usually trades at a slightly lower yield than older bonds of similar maturity, because it is more liquid. Building the benchmark from on-the-run bonds therefore embeds a small liquidity premium in every spread computed against it. The second is that in some markets the government curve is not a clean risk-free reference at all, which is one reason practitioners in those markets quote spreads against swaps instead.
When the benchmark used is a single actual government bond rather than an interpolated point, the same measure is usually called the nominal spread or benchmark spread. The G-spread is the version that interpolates, which makes it more comparable across issues, and that interpolation is where the arithmetic actually happens.
Computing it, and the interpolation step
Governments do not issue bonds at every maturity, and the ones they have issued age. A corporate bond maturing in six years and two months rarely has a government bond sitting beside it, so the benchmark yield has to be constructed from the two nearest points.
A corporate bond maturing in six years yields 6.85%. The government curve shows a five-year bond at 4.20% and a seven-year bond at 4.60%. What is the G-spread?
- Six years lies between five and seven, so the benchmark must be interpolated.
- The fraction of the way from five to seven is (6 − 5) / (7 − 5) = 0.50.
- Interpolated yield = 4.20 + 0.50 × (4.60 − 4.20) = 4.20 + 0.20 = 4.40%.
- G-spread = 6.85 − 4.40 = 2.45%, that is 245 basis points.
Answer: 245 basis points. Note that the benchmark yield of 4.40% is not a price anyone quoted. It is a linear estimate between two traded points, and on a curved yield curve linear interpolation slightly misstates it. That error is usually a basis point or two and it is the first of several approximations stacked inside this measure.
Where it sits among the spread measures
The G-spread is the first of a family, and each successive measure removes one assumption from the one before it.
| Measure | Benchmark | What it accounts for | What it still assumes |
|---|---|---|---|
| Nominal or benchmark spread | One actual government bond | Nothing beyond the level of yields | A comparable bond exists at that maturity |
| G-spread | Interpolated government yield | Maturity matching, through interpolation | A single yield describes the whole curve |
| I-spread | Interpolated swap rate | The same, against a swap benchmark | The same, plus that swaps are the right reference |
| Z-spread | The government spot curve | The shape of the curve, cash flow by cash flow | Cash flows are certain |
| Option adjusted spread | The spot curve plus an interest rate model | The curve and the embedded option | The model and its volatility assumption |
The I-spread deserves a word because it is the one most often confused with the G-spread. It is computed identically, by subtracting an interpolated benchmark yield from the bond’s yield, and the only difference is that the benchmark comes from the swap curve rather than the government curve. Where the two differ is the swap spread, so a bond’s G-spread and I-spread are not interchangeable and the gap between them moves with conditions in the swap market rather than with anything about the issuer.
The Z-spread is the one worth understanding properly, because it is the direct upgrade. Rather than comparing two summary yields, it asks what constant spread, added to every point on the government spot curve, makes the present value of the bond’s cash flows equal its market price. Each cash flow is discounted at the rate appropriate to its own date, so the shape of the curve is fully respected.
That sounds decisively better, and the natural next question is how much difference it makes. It is worth answering with numbers rather than assertion.
How good an approximation it actually is
Take a government spot curve rising steadily from 2.00% at one year to 6.00% at ten years, which is a steep curve by any standard. A ten-year corporate bond pays an annual coupon of 4% and trades at a Z-spread of exactly 200 basis points. What does the G-spread say?
- Discounting each cash flow at its own spot rate plus 200 basis points gives a price of 75.14.
- Solving for the single yield that reproduces that price gives a corporate yield to maturity of 7.65%.
- The ten-year government par yield implied by the same spot curve is 5.58%.
- G-spread = 7.65 − 5.58 = 2.06%, that is 206 basis points.
Answer: 206 basis points against a true Z-spread of 200. The G-spread overstates by about 6 basis points on a ten-year bond with a 400 basis point curve slope. That is a real error and it is not the order of magnitude the usual criticism implies. For a plain bullet bond the G-spread is a good approximation, and a trader quoting it is not being careless.
The reason the error stays small is worth seeing. Both yields are computed the same way, from the same curve, on cash flows with similar timing. The distortion introduced by collapsing a curve into a single yield affects the corporate bond and the government benchmark in the same direction, and most of it cancels in the subtraction.
That cancellation is the whole basis of the approximation, and it is also the clue to when it fails. It only works while the two sets of cash flows have similar timing.
Where it breaks: cash flows that are not bullet
A bullet bond returns its principal in one payment at maturity, so a ten-year corporate bond and a ten-year government bond genuinely resemble each other. An amortising bond does not. It repays principal throughout its life, so most of its money arrives long before the stated maturity date, and a maturity-matched government bond is no longer a comparable instrument.
On the same curve, take a ten-year corporate bond that repays 10 of principal each year and pays 6% on the outstanding balance, again at a true Z-spread of 200 basis points. Its weighted average life is 5.5 years rather than 10.
- Discounting each cash flow at its own spot rate plus 200 basis points gives a price of 98.61.
- Solving for the single yield gives a corporate yield to maturity of 6.32%.
- Against the ten-year government par yield of 5.58%, the G-spread is 74 basis points.
- Against the five-year government par yield of 3.71%, closer to the bond’s average life, the G-spread is 261 basis points.
Answer: the true spread is 200 basis points and the two obvious benchmark choices give 74 and 261. One understates by 126 and the other overstates by 61. There is no maturity that produces the right answer, because the problem is not the choice of benchmark but the assumption that a single benchmark maturity can represent a stream of cash flows spread across ten years.
| Bond | Z-spread | G-spread | Error |
|---|---|---|---|
| 10-year bullet, 4% coupon | 200 bps | 206 bps | +6 |
| 10-year amortiser, against the 10-year | 200 bps | 74 bps | −126 |
| 10-year amortiser, against the 5-year | 200 bps | 261 bps | +61 |
The usual objection to the G-spread, that it ignores the shape of the curve, is the wrong objection. Curve shape mostly cancels between the two yields. What does not cancel is a difference in the timing of the cash flows, and that is why the G-spread is fine for bullet corporates and useless for amortising securities, mortgage-backed paper, or anything with a sinking fund.
Where it breaks completely: embedded options
Both the G-spread and the Z-spread assume the cash flows are known. A callable bond’s are not.
If rates fall, the issuer calls and the bond ends early. If rates rise, it does not and the bond runs to maturity. The cash flow schedule therefore depends on the future path of interest rates, and no spread computed off a fixed schedule can be right. Both measures respond by inflating: a callable bond looks cheap and shows a wide spread, because part of that apparent compensation is not credit compensation at all. It is the value of the option the investor has written to the issuer.
The option adjusted spread strips that out. It uses an interest rate model to generate paths, values the option along them, and reports the spread that remains once the option has been accounted for.
OAS = Z-spread − option cost
for a callable bond the option cost is positive, so OAS is below the Z-spread; for a putable bond the option benefits the holder, so OAS exceeds the Z-spread
A related detail worth carrying: an OAS can come out negative. That is not necessarily an error. It says the bond is priced richer than the benchmark curve plus the option value justifies, which happens for scarce or specially treated issues and, more often, signals that the model inputs need checking before the number is used. A negative OAS is a prompt to look again rather than a conclusion.
The OAS is not free of assumptions either. It depends on the interest rate model chosen and, in particular, on the volatility assumed, and two analysts using different volatility inputs will produce different OAS figures for the same bond. That is a real limitation, and it is a different kind of limitation from the G-spread’s. One measure is approximate because it simplifies; the other is uncertain because it models.
Which measure to use, and when
The practical rule follows directly from what each measure assumes.
Use the G-spread for option-free bullet bonds when you want a quick, comparable number, which is most corporate bond quoting. It is transparent, anyone can reproduce it from two yields, and the error against the Z-spread is small enough to ignore for that purpose.
Use the Z-spread when cash flow timing differs materially from a bullet, when the curve is unusually shaped, or when you are comparing bonds with different coupons and maturities and want the comparison to mean something.
And be explicit about which measure a quoted number is. Spreads are exchanged in conversation and in research notes without their type attached far more often than they should be, and 245 basis points means three different things depending on whether it is a G-spread, a Z-spread or an OAS. On a bullet bond the difference is small enough to survive the confusion. On a callable bond it is not, and the resulting comparison ranks bonds by how much optionality they have written rather than by credit quality.
Use the OAS whenever there is an embedded option. This is not a refinement in that case. A Z-spread on a callable bond is measuring the wrong thing, and reporting it as a credit spread is an error rather than an approximation.
Two relationships come up repeatedly. For an option-free bond, OAS equals the Z-spread, because there is no option to adjust for. For a callable bond, OAS is less than the Z-spread; for a putable bond, OAS is greater. And when comparing two bonds, the OAS is the only one of the three that is comparable across bonds with different option features, which is exactly why it exists.
The wider habit worth building is to ask what a spread is compensating for before quoting it. A number that is described as a credit spread but contains an option cost, a liquidity premium and a cash flow timing mismatch is not telling you what you think it is, and the G-spread is the measure where that question is easiest to ask and easiest to forget.
The short version
- The G-spread is a bond’s yield to maturity minus an interpolated government yield at the same maturity.
- Interpolation is required because governments do not issue at every maturity, so the benchmark is constructed rather than quoted.
- For a plain bullet bond it lands within roughly 6 basis points of the Z-spread even on a steep curve, so the usual criticism overstates the problem.
- It fails badly when cash flow timing differs from a bullet: on an amortiser it missed a 200 basis point spread by 126 basis points in one direction and 61 in the other.
- Curve shape largely cancels between the two yields. Cash flow timing does not, and that is the real limitation.
- With an embedded option, use the OAS. For a callable bond OAS is below the Z-spread; for a putable bond it is above.
Frequently Asked Questions
What is a G-spread?
The yield to maturity of a bond minus the yield to maturity of a government bond of the same maturity, expressed in basis points. The G is for government. Because a government bond rarely matures on exactly the right date, the benchmark yield is usually interpolated between the two nearest points, which is what distinguishes the G-spread from a plain nominal or benchmark spread.
How do you interpolate the benchmark yield?
Linearly between the two nearest government maturities. For a six-year bond with a five-year government at 4.20% and a seven-year at 4.60%, six years is halfway, so the benchmark is 4.20 plus half of 0.40, which is 4.40%. The resulting figure is a constructed estimate rather than a traded yield, and on a curved yield curve it is slightly off.
Is the G-spread really worse than the Z-spread?
Less than usually claimed, for bullet bonds. On a curve rising 400 basis points across ten years, a ten-year bullet corporate at a true 200 basis point Z-spread shows a G-spread of about 206. The distortion from collapsing a curve into one yield affects the corporate bond and the benchmark in the same direction, so most of it cancels in the subtraction.
When does the G-spread genuinely mislead?
When cash flow timing differs from a bullet. For an amortising bond with a weighted average life of 5.5 years and a true spread of 200 basis points, the G-spread against the ten-year government came to 74 basis points and against the five-year to 261. No benchmark maturity gives the right answer, because a single maturity cannot represent cash flows spread across ten years.
What is the difference between Z-spread and OAS?
The Z-spread is the constant spread added to every point on the government spot curve that makes the present value of the bond’s cash flows equal its price, and it assumes those cash flows are certain. The OAS removes the value of any embedded option first, so OAS equals the Z-spread minus the option cost. For an option-free bond the two are the same.
Why is OAS lower than Z-spread for a callable bond?
Because a callable bond’s wide apparent spread is partly payment for the option the investor has written to the issuer, not compensation for credit risk. Subtracting the option cost leaves the spread that genuinely reflects credit and liquidity. For a putable bond the option favours the holder, so the adjustment runs the other way and the OAS exceeds the Z-spread.
Loading comments...
Add your Thoughts: