FRM Exam Part II · Structured Credit Risk

Valuation and Rating of Structured Products Explained

Updated 11 October 2026 · Fact-checked

Structured tranches are valued by modelling pool losses, then discounting each tranche's expected cash flows. Ratings give an opinion on credit risk, mostly default probability or expected loss, and they rely on the same models. Both depend on assumptions about default correlation, recovery and housing data, so model risk and ratings flaws can mislead investors.

Understand Valuation and Rating of Structured Products

A structured product pools assets such as mortgages or loans and splits the pool's cash flows into tranches. Losses hit the equity tranche first, then mezzanine, then senior. A tranche is a claim on a slice of pool losses, so its value depends on the loss distribution of the whole pool, not on any single loan.

To value a tranche, you model the pool. You need each asset's default probability, the recovery rate and the default correlation. You simulate or calculate the pool loss distribution, apply the waterfall to find the loss to each tranche, and discount the expected tranche cash flows. The standard market tool was the one-factor Gaussian copula. Correlation is the key input. Higher correlation raises the risk of the equity tranche's opposite: it makes the equity tranche less risky (losses become more bunched) and the senior tranche riskier, because mass moves into the extreme tail.

Ratings agencies rate tranches by estimating either probability of default or expected loss, then mapping the result to a letter grade. A senior tranche can reach AAA because enough subordination sits below it, even if the underlying loans are weak. This is why many AAA tranches looked as safe as AAA corporate bonds, yet behaved very differently in a crisis.

The problems are well documented. Models used short data histories from a period of rising house prices. Correlation was underestimated and often assumed stable. Ratings were designed for single-name bonds, yet tranches are highly sensitive to systematic risk, so a small shift in assumptions can move a rating many notches. Ratings agencies were paid by issuers, which created conflict of interest and rating shopping. Investors and regulators leaned on ratings instead of doing their own analysis.

The result is model risk: the chance that a model is wrong, misused or based on poor inputs. In structured credit it is large because tranches are leveraged on correlation, and because there is little market data to validate the models. Tranche ratings also migrate sharply, with cliff effects, when the collateral deteriorates.

Key formulas to remember

Tranche loss: Tranche loss = min(max(L − A, 0), D − A); L = pool loss, A = attachment point, D = detachment point, all in the same units (for example % of pool). Tranche loss is in the same units as L; divide by (D − A) for the % of tranche notional lost.
Tranche loss as % of tranche: % tranche loss = min(max(L − A, 0), D − A) ÷ (D − A); Shows leverage: a thin tranche loses a large percentage for a small pool loss.
Tranche thickness: Thickness = D − A; Thinner tranches are more sensitive to pool losses.
Expected loss of the pool: EL = Σ (PD × LGD × EAD); Sum over assets. Tranche expected losses add up to the pool expected loss (before fees and excess spread).
Value of tranche: Value = Σ expected cash flow(t) × discount factor(t); Expected cash flows come after applying the waterfall to simulated loss scenarios. Discounting is at risk-neutral or market-consistent rates.
Correlation effect (rule of thumb): Higher default correlation: equity tranche value rises, senior tranche value falls; Holds in a standard copula model; mezzanine can move either way depending on position.

How to solve Valuation and Rating of Structured Products questions

Use this order for any question on valuing, rating or model risk of tranches.

1Identify what is asked: valuation, rating, model risk, or ratings shortcomings.
2Find the attachment and detachment points and compute the tranche loss for the given pool loss using min(max(L − A, 0), D − A).
3If valuing, link value to expected tranche loss and discounting. Note the role of correlation and recovery assumptions.
4If a correlation change is given, decide the direction: higher correlation hurts senior tranches and helps equity tranches.
5If rating is asked, separate the rating target (probability of default versus expected loss) and check how subordination supports it.
6Name the weakness precisely: short data history, correlation or housing assumptions, mapping error, conflict of interest, rating cliffs, or over-reliance.
7State the interpretation: what the number means for the investor or bank, and the remedy such as own due diligence or stress tests.

Quickest way: Attachment-point and correlation shortcut

When to use it: Use for MCQs that give pool losses or ask how a tranche reacts to an assumption change.

Write A and D. Pool loss below A means zero tranche loss. Pool loss above D means total tranche loss.
In between, tranche loss = L − A. Divide by (D − A) for percent.
For correlation questions, remember: equity gains, senior loses as correlation rises.
For ratings questions, pick the answer about model dependence, correlation underestimation, conflicts of interest or cliff risk, not 'ratings measure market price risk'.
Eliminate options that claim ratings measure liquidity or market risk, or that AAA tranches equal AAA bonds in risk behaviour.

Common mistakes in Valuation and Rating of Structured Products

Dividing tranche loss by the pool size instead of the tranche size to get percent loss.
Students stop after L − A.
Fix: Always divide by (D − A) when asked for the percent of tranche notional lost.
Saying higher correlation lowers risk for all tranches.
Diversification intuition from portfolios.
Fix: Higher correlation fattens the tail. Senior tranches become riskier, equity tranches less risky.
Treating an AAA tranche rating as equal to an AAA corporate bond in risk.
Letters look identical.
Fix: Tranche ratings are far more sensitive to systematic risk and model assumptions, and can migrate by many notches quickly.
Blaming ratings failure only on fraud or conflicts of interest.
Popular narrative.
Fix: Include model risk, weak data, correlation assumptions and investor over-reliance. Conflicts were one factor among several.
Confusing the rating target.
Agencies differ in approach.
Fix: State clearly whether the rating reflects probability of default or expected loss, as the question defines it.
Assuming a model is validated because it fits historical data.
Data came from a benign period.
Fix: Model validity needs stress tests and out-of-sample checks. A good fit in calm times says little about tail correlation.

Worked examples

Example 1

A mezzanine tranche attaches at 5% and detaches at 15% of a pool. The pool loses 11%. What percentage of the tranche notional is lost?

Show the solution

A = 5%, D = 15%, L = 11%.
Tranche loss in pool terms = min(max(11 − 5, 0), 15 − 5) = min(6, 10) = 6%.
Tranche thickness = 15 − 5 = 10%.
Percent of tranche lost = 6 ÷ 10 = 60%.

Answer: 60% of the mezzanine tranche notional is lost.

Example 2

A bank values a senior CDO tranche with a Gaussian copula and finds its correlation input was too low. Using a higher, more realistic correlation, what happens to the tranche's value and why does this matter for its AAA rating?

Show the solution

Higher correlation makes defaults more clustered.
Clustering raises the probability of very large pool losses, which reach the senior tranche.
So the expected senior tranche loss rises and its value falls.
A rating built on the low correlation overstated its safety, and the tranche could be downgraded by several notches.

Answer: The senior tranche's value falls and its risk rises. The AAA rating was overstated because it relied on understated correlation, an example of model risk.

Exam tips

Practise the min(max(L − A, 0), D − A) calculation until it takes ten seconds.
For correlation direction, memorise: equity up, senior down when correlation rises.
Ratings questions often reward the answer naming several shortcomings: model dependence, short data, cliff risk, conflicts and over-reliance.
Link every weakness to a remedy: independent valuation, stress testing and reduced mechanical reliance on ratings.
Watch wording: 'expected loss' versus 'probability of default' versus 'market value'.

Practice questions from Structured Credit Risk

Valuation and Rating of Structured Products in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

FRM Part ICredit Rating Agencies and Structured Product Failures

Valuation and Rating of Structured Products: frequently asked questions

Why did credit ratings fail in the subprime crisis?

Ratings relied on models using short, benign data histories and low correlation assumptions. Tranche ratings were highly sensitive to these inputs, issuers paid the agencies, and investors over-relied on the letters. When house prices fell, losses were far more correlated than assumed.

How are CDO tranches valued?

You model the pool loss distribution using default probabilities, recovery and correlation. Then you apply the waterfall to get each tranche's loss, and discount the expected cash flows. Correlation is usually the most influential input.

What is model risk in structured credit?

It is the risk of loss from a model that is wrong, poorly calibrated or misused. In structured credit it is large because tranches are leveraged on correlation and there is little market data to test the models.

Is an AAA structured tranche as safe as an AAA bond?

Not in behaviour. A tranche rating depends heavily on systematic risk and model assumptions, so it can be downgraded many notches quickly. A corporate AAA rating is less sensitive to these factors.