How Credit Spread Data Is Actually Constructed

Direct Answer

Credit spread data is not a single observed fact — it's the output of a specific index provider's construction rules. The two most-cited families, the ICE BofA indices and the Bloomberg US Corporate/High Yield indices, each set their own minimum issue size, blend rating-agency inputs into a composite rating differently, price constituents on their own schedule, and rebalance on their own calendar. Because "the IG spread" or "the HY spread" is really "one provider's rules applied to one universe on one day," two providers can legitimately print different numbers for the same underlying credit market — and neither is simply wrong.

This page covers how the widely cited benchmarks are actually built, why the option-adjusted spread (OAS) convention exists and what it strips out, how constituent turnover and survivorship effects move the reported number for reasons unrelated to a broad shift in credit conditions, and a worked example contrasting a nominal spread with an OAS for a callable bond. For the underlying concept of what a credit spread measures and how it fits into financial conditions and the credit cycle, see Financial Conditions, Credit Spreads & Liquidity — this page assumes that background and focuses specifically on data construction and cross-provider comparability.

Key Takeaways

• "The credit spread" is a methodology choice, not a fact: Every published spread number is the output of one index provider's eligibility rules, rating source, pricing convention, and rebalancing schedule applied to a specific bond universe.
• Minimum issue size sets the floor for what counts: Most major benchmark indices exclude smaller, less liquid issues (commonly a $250-300 million minimum outstanding face value), which biases the constituent set toward larger, more frequently traded issuers.
• Rating source matters at the margin: Indices differ in whether they use the lowest, middle, or average of Moody's, S&P, and Fitch ratings to classify a bond as investment grade or high yield — a bond near the IG/HY boundary can be included in one provider's IG index and excluded from another's.
• OAS, not nominal spread, is the standard convention: The option-adjusted spread strips out the value of embedded call options so that spreads are comparable across bonds with different redemption features. A nominal spread on a callable bond overstates pure credit compensation.
• Constituents turn over constantly: New issuance enters, matured or called bonds exit, and downgraded issuers migrate from IG to HY indices (or drop out of index eligibility entirely) — each event can move the reported spread independent of any broad credit-market shift.
• Rebalancing frequency creates measurement lag: Most indices rebalance monthly at month-end, so an intra-month downgrade or new issue doesn't affect index composition until the next rebalancing date, even though the bond's own spread has already moved.
• Two providers, two legitimate numbers: A 10-30bp gap between ICE BofA and Bloomberg spread series on the same day is normal and expected — it reflects different but equally defensible construction choices, not a data error.

Core Concepts

How Are Credit Spread Indices Actually Constructed?

A credit spread index is a rules-based basket of corporate bonds, and the reported spread is a weighted average across that basket. Construction starts with eligibility rules: currency (USD-denominated for the most widely cited US series), a minimum time to maturity (typically at least one year remaining), and a minimum amount outstanding — commonly $250-300 million face value for both the ICE BofA and Bloomberg US Corporate and High Yield index families. That minimum-size screen matters more than it looks: it systematically excludes smaller issuers, so the index spread reflects the credit conditions of larger, more liquid borrowers rather than the full universe of outstanding corporate debt.

The next rule is classification: which rating agency input decides whether a bond is investment grade (BBB-/Baa3 or above) or high yield (BB+/Ba1 or below). ICE BofA's US indices generally use a composite of the middle rating among Moody's, S&P, and Fitch when all three rate a bond; Bloomberg's methodology similarly blends multiple agency inputs but with its own tie-breaking rules. For a bond sitting right at the IG/HY boundary, these differences in whose rating (or which blend) governs can place the same bond in different index buckets across providers in the same month.

Pricing is the third construction layer. Index providers need an end-of-day price for every constituent, but most corporate bonds don't trade every day — unlike an equity index, where every member has a continuous market price. Providers use a mix of executed trade prices (from sources like FINRA's TRACE system), dealer-submitted quotes, and evaluated pricing models for bonds without a same-day trade. Two providers pricing the same illiquid bond on a quiet day can reasonably arrive at slightly different marks, and that difference flows straight into the reported index spread.

Finally, indices rebalance on a fixed schedule — typically monthly, at month-end — recalculating weights and applying that period's eligible-universe changes (new issuance added, matured or called bonds removed, rating migrations applied). Between rebalancing dates, the constituent list is generally held fixed even as individual bond prices move freely, so the index's day-to-day spread changes reflect price moves within a fixed basket, while step changes in composition show up at each rebalancing date.

What Is an Option-Adjusted Spread and Why Does It Matter?

Most investment-grade and high-yield corporate bonds are callable — the issuer can redeem the bond early, usually when it can refinance at a lower rate. That call feature has value to the issuer and is a cost to the bondholder, because the investor gives up the bond's upside if rates fall (the issuer calls it away) while still bearing full downside if rates rise or credit quality deteriorates. A simple nominal spread — yield to maturity minus the yield on a comparable-maturity Treasury — doesn't account for that asymmetry at all; it treats a callable bond exactly like a plain bullet bond with no embedded option.

The option-adjusted spread (OAS) corrects for this. It uses an interest-rate model (typically a binomial or lattice-based model of future rate paths) to estimate the value of the embedded call option, then backs that option value out of the spread. What remains is, in principle, the spread an investor earns purely for credit and liquidity risk, stripped of the value given up (or received) through optionality. This is why OAS — not nominal spread — is the standard convention embedded in every major benchmark: ICE BofA's headline series and Bloomberg's headline series are both quoted as OAS, specifically so that a callable bond and a non-callable bond of similar credit quality can be compared on equal footing.

How Do Constituent Turnover and Survivorship Effects Distort Spread Data?

An index's membership is never static. New bonds are added as companies issue debt; existing bonds exit when they mature, are called, or are tendered for. The more consequential turnover, for interpreting spread trends, is rating migration: a bond downgraded from BBB- to BB+ moves from the investment-grade index into the high-yield index (the "fallen angel" case), and a bond downgraded below the high-yield index's rating floor can drop out of index eligibility altogether. Because that bond typically widens sharply in the run-up to a downgrade, its exit from the IG index right as it's near its worst can mechanically tighten the reported IG spread — not because IG credit conditions improved, but because the worst-performing bond just left the basket.

A related effect is survivorship: a spread index reflects only bonds currently meeting eligibility criteria, so defaulted bonds are removed rather than tracked at their post-default (near-zero recovery-adjusted) value. This means a headline HY spread series, by construction, is not a running mark-to-market of "every bond that was ever in the index," but a snapshot of the currently eligible universe at each date. For most short-run interpretation this doesn't matter much, but across a full credit cycle with a wave of defaults, it means the index understates how bad realized outcomes were for investors who held the bonds that exited through default rather than being lifted out through a clean call or maturity.

Why Do Two Data Providers Report Different "IG Spread" or "HY Spread" Numbers?

Put the three effects above together and the answer follows directly: different minimum issue-size thresholds mean the two providers aren't even pricing quite the same universe of bonds; different rating-blend rules mean boundary bonds can sit in different buckets; different pricing sources mean illiquid constituents get different marks; and different rebalancing calendars mean a rating migration or new issue shows up in one provider's index before the other's next rebalancing date. None of these differences implies either provider made an error — they're each internally consistent, published methodologies, just not identical ones. The practical implication for a reader is to track one provider's series consistently over time rather than comparing an absolute level from one provider on one day against another provider's level from a different day, and to treat a modest cross-provider gap as normal rather than as a data-quality alarm.

Worked Example: Nominal Spread vs. OAS on a Callable Bond

1. The bond: A BBB-rated corporate issues a 10-year bond with a 5-year call date (the issuer can redeem it at par starting in year 5). It trades at a price that puts its yield to maturity at 6.20%.
2. Benchmark Treasury: The comparable-maturity (10-year) Treasury yields 3.70%.
3. Nominal spread calculation: 6.20% − 3.70% = 2.50%, or 250bp. This is the entire gap between the bond's yield and the Treasury yield, with no adjustment for the call feature.
4. Estimating the embedded option's value: Because rates could fall over the next five years, the issuer's right to call the bond at par has real value — the model estimates that value at roughly 40bp of spread equivalent, reflecting the probability-weighted benefit to the issuer of refinancing if rates decline.
5. OAS calculation: 250bp (nominal spread) − 40bp (value of the embedded call option) = 210bp OAS.
6. Why it matters: The 250bp nominal spread makes it look like the market is demanding 250bp of compensation purely for this issuer's credit risk. In reality, only about 210bp is credit and liquidity compensation — the other 40bp is the price of the option the bondholder implicitly sold the issuer. Comparing this bond's 250bp nominal spread against a non-callable bond's 210bp nominal spread would wrongly suggest the callable bond carries more credit risk, when their OAS-adjusted credit compensation is actually identical at 210bp.
7. Takeaway for reading index data: Because most corporate bonds carry some call feature, an index built on nominal spreads would systematically overstate credit-risk compensation relative to an OAS-based index — which is exactly why every major benchmark (ICE BofA, Bloomberg) headlines OAS rather than nominal spread.

Measurement Framework

Common Failure Modes

Comparing Two Providers' Levels as If They Were the Same Series

Reading "IG spreads are at 110bp" from one source and "IG spreads are at 95bp" from another on the same day and concluding one of them is wrong misreads how these benchmarks work. The correct response is to check whether both figures are OAS or one is nominal, confirm the underlying index family, and — most usefully — track each series against its own history rather than cross-comparing absolute levels between providers.

When building a chart or model that references a credit spread series, cite the specific index (for example, the ICE BofA US High Yield Index OAS, ticker BAMLH0A0HYM2 on FRED) rather than the generic phrase "HY spreads," and keep using that same series for every subsequent comparison.

Treating a Rating Migration as a Pure Credit-Market Signal

A sudden drop in the reported IG index spread right after a large "fallen angel" downgrade can look like credit conditions improved, when it may simply reflect the worst-spread bond exiting the basket at the next rebalancing date. Before reading a level change as a broad credit-market move, check whether a known large downgrade, default, or unusually large new issue coincided with the move.

Using Nominal Spread When OAS Is Available

Nominal spread is easier to compute by hand but embeds the value of any call feature into what looks like pure credit compensation. For any bond or index with meaningful call optionality — which describes most of the investment-grade and high-yield corporate universe — OAS is the more accurate measure of compensation for credit and liquidity risk alone, and it's what every major published benchmark actually reports.

Assuming "Public Access" Means "Free Redistribution"

ICE BofA and Bloomberg index-level OAS series are published daily on FRED and are free to view and cite for educational purposes, but the underlying constituent-level index data and any commercial redistribution of the index itself typically require a license from the index provider. Being able to read a chart on FRED does not automatically grant the right to redistribute the full dataset.

Frequently Asked Questions

Sources and Further Verification

• Federal Reserve (FRED). ICE BofA US High Yield Index OAS and ICE BofA US Corporate Index OAS — Daily index-level OAS series, free to view and cite.
• FINRA. TRACE (Trade Reporting and Compliance Engine) — Corporate bond transaction reporting used as a pricing input by index providers.
• ICE Data Indices. ICE BofA Index Methodology — Official documentation of eligibility, pricing, and rebalancing rules for the ICE BofA fixed income index family.
• Bloomberg Index Services. Bloomberg Fixed Income Indices — Official methodology documentation for the Bloomberg US Corporate and High Yield index families.

Educational Disclaimer

This guide is for educational purposes only. Index methodologies are set by their respective providers and can change over time. Do not make investment decisions based solely on this content. Trading involves risk of loss.