Determining collateralised derivatives haircuts \(-\) the discount applied to the market value of an assets used as collateral \(-\) is becoming an increasingly more important problem. At the same time, the method used today has been found to have significant shortcomings. To sidestep these issues industry is now looking towards a parametric modelling approach to determining haircuts.
In the article The bald truth about collateral haircut modelling at Risk.Net, the author Nazneen Sherif discusses the development of using parametric models when determining the haircut instead of the commonly used historical value-at-risk (VaR) method. As outlined in the article, using a purely data-driven approach, such as historical VaR, you are highly dependent on the availability and quality of market data. This becomes problematic when the considered collateral asset is illiquid or lack reliable data. The parametric model instead uses explanatory factors such as volatility and credit risk measures to reflect specific characteristics of the collateral asset.
More information about the parametric approach to haircut modelling and the full article can be found here.