4
$\begingroup$

What are some quantitative approaches to estimating credit risk for investments that aren't publicly traded, such as private equity and direct real estate? I'm particularly interested in estimating incremental changes in credit quality and probability of bankruptcy, and its implications for issuer risk. While I am familiar with approaches to this for publicly traded assets, such as structural models and stress tests, I haven't come across any literature applying these quantitative techniques to private investments.

How is this done in actual practice? I would be interested in any papers or literature reviews you've come across that tackle this problem.

$\endgroup$
1
$\begingroup$

You could calculate Z-score of a privately owned company and compare it with its publicly traded peers. Z-score is based on the financial ratios of the company and you will have to apply some sort of weighting. You can take a look at here to get a basic idea about this technique.

$\endgroup$
  • $\begingroup$ Thank you. I think I can see how this might be useful for predicting bankruptcy. But do you have any suggestions for similar metrics that would predict a change in non-default credit quality, analogous to a rating downgrade (ex downgrade from A to BBB) for public debt for instance? $\endgroup$ – beeba May 1 '17 at 15:56
1
$\begingroup$

For publicly traded instruments market-based models are indeed favoured.

For non-publicly traded instruments, such as leveraged finance loans to private equity owned companies, industry practice is to use fundamentals-based models. This is often most simply done by looking at a wide variety of company financials/fundamentals (and macro-economic variables if you would also like to include the state of the macro-economy in the analysis) and running a regression model with realised defaults to calibrate a probability of default. One can then bucket the PD's into rating grades analogous to AA, A, BB, ... for public debt for instance. The Z-score mentioned by AK88 is a simple fundamentals-based model, but the score would need to be calibrated to PD to be used in the manner in which you require. An article by the Dutch Central Bank summarises this quite nicely.

If too few data are available to accurately calibrate regression models then experts based models are often used, with credit scoring based on fundamental factors.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.