Skip to content
All apps

CreditMetrics — credit portfolio model

About this app

Monte-Carlo simulation in single steps: systematic factor, correlated returns, rating migration and loss distribution with EL, VaR and Expected Shortfall.

The idea

CreditMetrics is a model for the credit risk of a portfolio. It considers not only defaults but also rating migrations, that is, downgrades or upgrades that change the value of a bond. Each obligor’s creditworthiness depends on a latent asset return Zᵢ = √ρ · M + √(1−ρ) · εᵢ: M is a common market factor, εᵢ the idiosyncratic risk and ρ the asset correlation. Thresholds on Zᵢ translate the transition matrix into new ratings. A Monte Carlo simulation then produces the portfolio loss distribution with expected loss, value at risk and expected shortfall.

What the app shows

A deck of six cards to flip through: an introduction and five steps. The example portfolio consists of 24 BBB bonds, with the asset correlation set to ρ = 0.20. On each card you carry out one step: draw the market factor M, draw the asset returns Z, perform the migration, compute the loss of the path and finally simulate 1,000 trials. Prices per rating are simplified (BBB ≈ 1.00, BB ≈ 0.93, CCC ≈ 0.55, default with about 0.40 recovery). The last card shows the histogram with EL, VaR₉₉ and ES₉₉.

Try it yourself

Draw M several times and notice how a bad economic year (negative M) pulls all 24 returns down together. This shows why correlated defaults drive the risk in the tail. After the simulation, look at the shape: the loss distribution is right-skewed, and VaR₉₉ and ES₉₉ lie far to the right of the mean. On every card you can ask the AI tutor to explain the step.

Frequently asked questions

What is CreditMetrics?
CreditMetrics is a credit portfolio model that simulates the value of a bond or loan portfolio over a time horizon. It accounts for defaults and rating migrations and links obligors through correlated asset returns. The result is a distribution of possible portfolio losses from which risk measures such as VaR are read off.
What does the formula Zᵢ = √ρ · M + √(1−ρ) · εᵢ mean?
It splits an obligor’s standardised asset return into a common part M that affects everyone and an idiosyncratic part εᵢ. Both are standard normal, so Zᵢ is standard normal too. The correlation between two obligors is ρ; with ρ = 0 they are independent, with ρ = 1 they move together.
What is the difference between VaR and expected shortfall?
Value at risk at the 99 % level is the loss that is exceeded in only 1 % of cases. Expected shortfall is the average loss within exactly those worst 1 % of cases. ES therefore also reflects how large losses become beyond the VaR.
How is the CreditMetrics app structured?
The app is a card deck with an introduction and five steps in which you run the simulation yourself: market factor, correlated returns, migration, loss, and the loss distribution from 1,000 trials. Each card comes with a prepared question for the AI tutor.

Subject: IRBA / banking supervision

More from IRBA / banking supervision

A VisuApp by heyprof: interactive, free in your browser, no sign-up. What is a VisuApp? · All apps