Prudent Valuation: The Challenge of Model Risk Quantification for Equity Barrier Options

The sub-prime crisis that took place in 2007-08 revealed several vulnerabilities in the global system such as the lack of liquidity and failures in accounting reporting standards. The new regulation on Prudent Valuation complies banks to hold sufficient high-quality capital reserves to resiliently face risks coming from different market factors. To this extent, financial institutions have to calculate Additional Valuation Adjustments for financial positions accounted at fair value, to be destined at reserve in the CET1. The purpose is to provide the market with most truthful view of the balance sheet. The scope of this thesis is to compute the Model Risk AVAs for exotic options, in compliance with the current regulation on Prudent Valuation. The first step is gathering a set of candidate models and proceeding with an accurate calibration implemented via MSE minimization and implied volatility matching, across different datasets. Due to the fast computational power, the pricing method used in the process is based on the Fast Fourier Transform. In order to build a price distribution from which deriving the prudent value for exotic options, a probability weight is assigned to each model based on the Akaike Information Criterion, which provides a measure of fit. Afterwards, Equity Barrier Options prices are obtained using Montecarlo algorithm, and the corresponding AVAs are derived in a way that is compliant with the new legislation. The work concludes with an insight on parameter analysis conducted by using the re-sampling technique. The latter methodology is a powerful tool to check parameters consistency and model reliability.