How do credit rating agencies work and why do they matter?

The short answer

The three dominant agencies — S&P Global, Moody’s, and Fitch — produce letter grades on a comparable scale. AAA/Aaa is the strongest category, BBB-/Baa3 is the lower bound of investment grade, and ratings descend through speculative grade down to default categories like D or C.

Issuers pay agencies for their ratings. Critics argue this creates incentives for grade inflation. The defenders argue that reputation across many issuances anchors agencies to honest ratings on average. Both sides have evidence; the truth is regime-dependent.

For practical purposes, ratings function as coordination devices: pension fund mandates, insurance capital charges, central bank collateral rules, and ETF eligibility all reference the IG/HY boundary. The rating is therefore a structural input to how capital flows in the global system.

→ New to credit? Financial education framework

What the data shows

Long-run data on rating accuracy comes primarily from the agencies themselves and from independent academic studies (Moody’s, S&P, BIS research, 1981-2024):

• Cumulative 5-year default rates: roughly 0.5% for AAA issuers, 2% for BBB, 18-20% for B, 30-40% for CCC and below
• Order preservation: in the long run, lower-rated issuers default at higher rates than higher-rated issuers, with very few exceptions
• Big Three market share: S&P, Moody’s and Fitch issue roughly 95% of all global ratings (ESMA data)
• Structured product downgrades during 2007-2009: a substantial share of AAA-rated subprime CDO tranches were downgraded by 6+ notches, against an expected loss profile of near zero
• Sovereign rating changes: roughly 100 sovereign actions per year across the three agencies in normal periods, more in crisis years

The exception that matters: ratings have been broadly accurate for traditional corporate and sovereign debt over many decades, but failed catastrophically on structured finance products before 2008. The cause was modeling assumptions on correlation, not the corporate rating process.

→ Dataset: US investment grade credit spread

Why it happens — the macro mechanism

To understand why ratings matter even when they are imperfect, it helps to separate three layers.

The information layer. Agencies produce credit research that complements but does not replace internal investor analysis. They have access to non-public information from issuers and apply standardized methodologies. Their value-added on individual issuer forecasts is debated, but on portfolio-level default ranking it is well documented.

The institutional layer. This is the source of structural power. Most institutional capital — pension funds, insurance reserves, central bank repo collateral, money market funds — operates under mandates and capital rules that reference ratings explicitly. A downgrade across the IG/HY threshold can force selling regardless of whether the rating is right. Agencies thus act as Schelling points: a coordinating reference everyone uses because everyone else uses it. This is the angle most under-appreciated in standard explanations. The IG/HY split only exists because the rating reference is shared.

The conflict layer. Issuers shop for ratings; agencies compete; methodology choices have material consequences for issuance volumes. The 2007-2009 collapse of structured finance ratings showed what happens when this conflict combines with novel asset classes that lack historical default data — agencies effectively rated correlation assumptions, and those assumptions broke. Corporate ratings, where the underlying business has decades of data, have not failed at this scale.

Synthesis by regime: in stable corporate cycles, rating accuracy at the portfolio level holds up well — defaults happen but mostly in the lower rating buckets as expected. In structural shocks affecting whole sectors (energy 2014-2016, hospitality 2020), rating actions lag the market and downgrades cluster after spreads have already widened. In genuinely new asset classes without history (subprime 2005-2007, possibly some private credit segments today), ratings can fail catastrophically because the inputs are theoretical rather than empirical.

Ratings are not principally forecasts — they are coordination devices the financial system uses to allocate capital, and that is what makes them powerful even when they are wrong.

→ Framework: Systemic fragilities and debt

What it means for different economic actors

Savers. Most exposure to rated debt is indirect, through bond funds and life insurance. The rating universe shapes what these vehicles can hold, so rating shifts affect prices that savers experience even without ever looking at a rating themselves.

Investors. Sophisticated investors typically combine ratings with their own credit analysis. The rating is a starting point, not an endpoint. Where the market reacts mechanically to rating changes — fallen angel events, IG/HY transitions, ETF inclusion or exclusion — there is potential value for those willing to hold across the boundary.

Sovereigns. Sovereign rating changes have grown more politically charged since 2010. The downgrade of US Treasury debt by S&P in 2011 and France by S&P/Fitch in subsequent years generated political reactions and methodology debates. Sovereign ratings increasingly mix economic fundamentals with governance assessments.

A common error is to treat the agencies as oracles or as fraud. They are neither. They are imperfect reference points whose institutional embedding gives them more power than their forecasting accuracy alone would justify.

Practical observation

Go deeper

Related questions

• What is the difference between investment grade and high yield debt?
• What is a fallen angel and why does it matter for markets?
• What is the Z-score and how is corporate distress measured?
• What are CLOs and how did they evolve after 2008?

Frequently asked questions

Are credit ratings a reliable predictor of default?

At the portfolio level, ratings have been a reliable ordinal predictor over four decades — lower-rated issuers default at higher cumulative rates than higher-rated issuers. At the individual issuer level, accuracy is more limited; specific corporate failures often surprise the market and the agencies. The right way to think about ratings is as a useful classification system, not as a precise probability forecast.

Why are ratings considered Schelling points more than risk forecasts?

Because the financial system has built layers of regulation and contracts that explicitly reference rating thresholds. Insurance risk-based capital, central bank collateral haircuts, ETF inclusion criteria, and pension fund mandates all use letter grades as binary triggers. This means a rating change moves capital regardless of whether the change reflects new information — the system pre-committed to act on the label.

How did rating agencies fail in the 2008 crisis?

The failure was concentrated in structured finance, not corporate ratings. Agencies rated mortgage-backed securities and CDOs using correlation assumptions that broke when housing prices fell nationally for the first time in modern history. The combination of the issuer-pays model, intense competition, and modeling of unprecedented assets produced AAA ratings on tranches that lost most of their value. Reforms after 2010 reduced this risk in structured products but did not eliminate it.

Last updated — 8 May 2026