Skip to content

FRM Part II · FRM Exam Part II · Financial Correlation Modeling - Bottom-Up Approaches

A risk manager compares a stochastic correlation model with a constant-correlation Gaussian model for a basket of two assets, noting empirical evidence that correlations tend to rise in market downturns and are mean-reverting. Which conclusion is most consistent with this evidence?

A constant-correlation model calibrated to calm periods understates joint extreme losses, because correlations rise in downturns. A mean-reverting stochastic correlation process can produce temporary spikes that cluster with high volatility, though it still needs stress testing and carries model risk.

  1. AA constant-correlation model will tend to understate joint tail losses in stressed markets, while a mean-reverting stochastic correlation can capture the clustering of high correlation with volatilityCorrect
  2. BA constant-correlation model will overstate joint tail losses because correlation averages out over time
  3. CStochastic correlation eliminates the need for stress testing since correlation is then fully modelled
  4. DMean reversion implies correlation cannot exceed its long-run level, so downturn correlation spikes are impossible

Explanation

Empirical correlation is higher in stress and mean-reverts, so a fixed correlation calibrated to normal periods understates joint tail risk. Stochastic correlation can generate such spikes, though it does not remove model risk or the need for stress tests. Mean reversion does not cap correlation at its mean.

Did you get it right without looking?

One question tells you little. A timed set on Financial Correlation Modeling - Bottom-Up Approaches shows your real accuracy, how long you take and where you lose marks.

Practise Financial Correlation Modeling - Bottom-Up ApproachesFree diagnostic mock

More Financial Correlation Modeling - Bottom-Up Approaches questions