Modeling Credit Spreads with Negative Stock Return Correlation

Summary

The document asks how to simulate a credit spread series that tends to tighten when stock returns exceed their average and widen when returns fall below it. It proposes a linear model: a positive baseline spread minus a positive coefficient times the deviation of stock return from its average, plus Gaussian noise. This captures the desired negative relationship in a simple form.

The author identifies a key limitation: because the model is linear, sufficiently large stock returns can produce negative spreads. The document does not offer a revised specification, tests, or empirical evidence; it is a modeling question rather than a worked solution. Any implementation would need to address the nonnegative-spread constraint while preserving the intended dependence and random variation.

Key ideas

• A linear model can link spread changes to deviations of stock returns from their average.
• The proposed coefficient makes spreads move inversely with stock returns.
• Gaussian noise adds randomness around the modeled relationship.
• The linear form can generate negative spreads when returns are sufficiently large.

Tags

# Simulate correlated credit spread


# Simulate correlated credit spread












I want to simulate a credit spread index which is negatively correlated to a given random walk of a stock index. They should be correlated in such a way that larger than average stock growth tend to tighten spreads, and lower than average growth tends to increase spreads.

One idea I have is \begin{equation} \text{spread}(t) = \alpha - \beta*(R(t) - R) + \epsilon(t), \end{equation} with $\alpha>0$, $\beta>0$, $\epsilon(t)$ a gaussian distribution, $R(t)$ the stock return (or drift) and $R$ the average stock return. This relation is obviously inspired by CAPM, but with some important modifications. One problem with this model is negative spreads when the return is of order $\alpha / \beta$.

I am wondering whether this is the right approach, or whether another approach is adviced. To emphasize: I want to keep it simple. Of course I know the relation between stock prices and credits is more complex, but the negative correlation with a random element suits my purpose.

Thank you in advances!

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.