Curve Bootstrapping: What the Equations Can and Cannot Identify

Start with payments rather than yield labels

The basic valuation equation is price equals the sum of each cash flow multiplied by its date-specific discount factor. A bootstrap reverses that relationship. Given prices and payment schedules, it solves for factors that make the equations hold. The input must therefore describe amounts and dates, not just maturity labels. A coupon bond's yield to maturity summarizes a price but is not itself a spot discount rate.

For a clean sequential solution, the next instrument introduces one unknown factor after earlier factors have been established. When it introduces several unknown payment dates, the problem needs interpolation or another restriction. This is an identification issue, not a reason to increase numerical precision. More decimal places cannot turn one independent equation into enough information to determine several independently varying discount factors.

Solve the first hypothetical discount factor

Assume a deliberately simplified annual payment grid with no accrued interest, settlement adjustment, default or embedded option. A hypothetical one-year zero-coupon claim pays USD 100 and is assigned a present value of USD 96. These numbers define a teaching exercise, not a market quote. Its equation is 96 = 100 * D(1), giving the one-year discount factor D(1) = 0.96.

That factor means one dollar payable in a year has a present value of 96 cents under this example's assumptions. It is dimensionless; USD 96 is the value of the USD 100 claim, not the factor itself. Keeping the distinction visible prevents a common scaling error when a later instrument is quoted per 100 of face but its cash flows are entered in full currency amounts.

Use the known first payment to solve the second

Now assign a two-year annual-coupon bond face value USD 100, annual coupon rate 4% and present value USD 100. It pays USD 4 after one year and USD 104 after two years. Its equation is 100 = 4 * 0.96 + 104 * D(2). Subtract the first coupon's USD 3.84 present value before solving for the final factor.

The result is D(2) = 96.16 / 104 = 0.9246153846. Substitution independently gives 3.84 + 96.16 = USD 100. Converting that factor to an annual-compounded two-year spot yield gives D(2)^(-1/2) - 1, approximately 3.996672%. The bond's 4% coupon and par price do not require its two-year spot yield to be exactly 4%, because the earlier coupon uses a separate factor.

Recognize what an exact fit leaves unresolved

The two equations above identify factors at two payment dates. They do not uniquely identify the factor halfway between those dates or any factor beyond the second year. Many curve shapes could pass through the same solved points. Reporting a complete monthly curve would therefore add a construction choice. It would not reveal information that was secretly contained in the two original input prices.

An exact fit can also preserve bad data exactly. A stale price, mistaken coupon or incorrectly scaled face value may produce a solvable equation with an economically misleading result. Repricing the same erroneous inputs is not an independent data-quality test. Review the input evidence separately from the algebra, and distinguish a successful equation solve from a conclusion that the resulting curve is appropriate for another instrument.

Avoid mixing incompatible claims into one curve

Combining bonds with different credit exposure or embedded options can make the fitted curve absorb differences unrelated to pure time value. A callable security's cash flows are not fixed in the same way as those in the example. An inflation-linked principal is not a known nominal amount. Treating either as an ordinary bullet can contaminate factors that will subsequently be applied to unrelated payments.

The ECB's technical paper emphasizes bond-sample quality as part of yield-curve estimation. The practical implication is to maintain an eligibility rule before fitting anything: currency, instrument structure, usable dates and consistent conventions. An unexplained outlier should remain an explicit review item. Deleting it solely because the curve looks smoother afterward substitutes visual preference for a documented assessment of why the observation is unsuitable.

Read official curves as constructed datasets

Treasury's methodology states that its inputs are indicative bid-side quotations, not actual transactions, and describes a bootstrap followed by interpolation. This distinction matters when reusing the output: a published par curve is not a list of executable bonds. Reconstructing its points also requires attention to the methodology version and available input information, rather than assuming that matching the table recreates Treasury's underlying process.

For a reproducible educational comparison, preserve the source date, input units and exact construction assumptions. Separate price residuals at calibration instruments from valuation differences at other payment schedules. If the method changes, compare both on unchanged inputs before attributing any resulting movement to economic conditions. A method revision and a source-data revision answer different questions, even when their effects appear in the same published series.

Keep the scope narrower than a market-pricing engine

A regular-bond calculator can illustrate how scheduled coupons and principal respond to one selected yield. It does not thereby solve a system of instrument prices for a term structure. The parent tools accept integer years from one through thirty and coupon frequencies of one, two or four, at a coupon date. They do not resolve irregular settlement dates or infer missing discount factors from live securities.

The appropriate conclusion from this guide is conditional: given these hypothetical payments and prices, these factors solve these equations. It is not a certification of a real curve, a trading opportunity or a forecast of future rates. Keeping that conditional statement intact makes the exercise useful. It tells a reviewer exactly which evidence would be needed before extending the calculation to a broader valuation purpose.

Primary references

1. U.S. Treasury: Treasury Yield Curve Methodology
2. ECB Statistics Paper 27: Yield Curve Modelling and a Conceptual Framework for Estimating Yield Curves