Asset correlation in 3D
About this app
Rotate a bivariate normal density, adjust correlation and default thresholds, and see how joint default probabilities change.
Joint defaults
In the Vasicek one-factor model each borrower has a standardised credit factor X = √ρ · Z + √(1 − ρ) · ε, with a common factor Z (the economy) and an idiosyncratic part ε. A borrower defaults when X falls below the threshold c = Φ⁻¹(PD). Two such factors are bivariate normal with correlation ρ. The probability that both default equals PD_A · PD_B only when ρ = 0. With positive correlation it is a multiple of that, and joint defaults drive portfolio risk.
What you see
The rotatable 3D surface is the joint density of the two credit factors over the range −3 to +3. Two cutting planes mark the default thresholds c_A and c_B, and a coloured rectangle on the floor marks the zone where both default. The values panel lists P(A defaults), P(B defaults), P(both) at the chosen correlation, P(both) at ρ = 0 and the ratio between them. P(both) is computed exactly by numerically integrating the bivariate normal distribution.
Try it
Set both PDs to 1 % and ρ to 0: the density is a round hill and P(both) is 0.01 %. Raise ρ and the hill turns into a ridge along the diagonal while the factor in the values panel climbs. Negative values down to −0.9 turn the ridge onto the other diagonal. Vary the PDs up to 20 % and toggle the threshold planes to see the default zone more clearly.
Frequently asked questions
- What is asset correlation in credit risk?
- Asset correlation measures how strongly the asset values or credit quality of two borrowers move together, usually because both depend on the same economic factor. It is not the same as default correlation, but together with the PDs it determines it. The higher it is, the more often borrowers default together.
- Is the probability of a joint default simply PD times PD?
- Only if the two borrowers are independent, i.e. at zero correlation. With positive asset correlation the joint default probability exceeds the product of the individual PDs, often by a large multiple. Assuming independence therefore understates portfolio risk.
- Which asset correlation does the IRB formula use?
- The supervisory risk-weight formula in Art. 153 CRR uses a correlation R between 12 % and 24 % for corporates, decreasing as PD rises. For retail exposures it is fixed, for example 15 % for residential mortgages and 4 % for qualifying revolving exposures. A higher correlation means more capital for the same loan.
- What can I adjust in the 3D app?
- You set the asset correlation ρ between −0.9 and 0.9 and the PDs of both borrowers between 0.1 % and 20 %, and you can show or hide the default thresholds. The surface can be rotated; the values panel compares the joint default probability with the case ρ = 0.
Subject: Financial Risk & Analytics
More from Financial Risk & Analytics
- PD calibration and reliability
- Risk-bearing capacity: loss distribution
- ROC curves and discrimination
- ROC model comparison
A VisuApp by heyprof: interactive, free in your browser, no sign-up. What is a VisuApp? · All apps