Is model risk worth worrying about? An empirical analysis of model risk

Is model risk worth worrying about? An empirical analysis of model risk

Added in AFIR / ERM / RISK

Thanks! Share it with your friends!


You disliked this video. Thanks for the feedback!

Sorry, only registred users can create playlists.


Speaker(s): David Shkel (University of Hagen), Rainer Baule (University of Hagen)

Structured financial products as a combination of classical investments such as stocks or bonds and derivatives provide investors with the possibility to buy a portfolio with a complex payoff profile through a single product. If the embedded derivative is not liquidly traded, it is inevitable to use financial models to determine such a product's fair value. Therefore, the valuation of such products is subject to model risk.

Investors should include this risk as a factor in their investment decisions since we found evidence that margins of issuing banks tend to rise if the model risk is high. For a company using fair value accounting, following the mark-to-model approach to determine a product's fair value, the model risk induced by illiquid derivatives is incorporated in a company's balance sheet and therefore should be monitored carefully. Our contribution analyzes model risk in the case of bonus certificates which incorporate illiquid path dependent options and are publicly traded. The existing literature on structured financial products focusses on issuer margins and their determinants, e.g., Baule and Tallau (2011) or Schertler (2016).

The paper extends the literature by quantifying model risk and analyzing the model risk's determinants. We investigate the products over their whole life cycle, thus analyzing model risk in many potential market conditions, and establish a connection between model risk and issuers' pricing policy. We show that the famous Black-Scholes model is no reasonable choice regarding the pricing of exotic products since model risk could be arbitrarily high if the volatility parameter is misspecified. We further analyze the model risk by using the Heston (1993) model, the Bates (1996) model, and the local volatility model of Dupire (1994), and find that model risk can reach substantial levels even if sophisticated valuation models are used.

Post your comment

Sign in or sign up to post comments.
Be the first to comment