A comparison of the Basel III capital requirement models for financial institutions

Abstract: The purpose of this report is to implement and compare the two Basel III standard methods on how to calculate the capital requirement for finan- cial institutions, related to counterparty credit risk. The models being the Standardized Approach for Counterparty Credit Risk (SA-CCR) and the Internal Model Method (IMM). The SA-CCR model is a simpler and more standardized model with prescribed methods while the IMM model is a more flexible model optimized to the specific portfolio. Because of this, the IMM model requires more work to implement. The comparison of these two models is done by looking at a small number of transactions from the Bank’s trading book and computing the Exposure At Default (EAD) that these would give. Both models are used to compute this, and these results are compared. To obtain EAD the transactions need to be priced and their Net Present Value (NPV) needs to be calculated. One need simulated interest rates to do so, which is done using Monte Carlo simulations. For this, a Hull-White process is used to simulate the interest rates and the parameters of this process is calibrated using market data. Out of the two models, the IMM model is the more complex one. It requires both normal and stressed data as input parameters and it also needs to be validated. The validation of the model is done by doing some- thing called ”backtesting” on it, which investigates if the model created does give the expected results. Backtesting is performed by taking the interest rate for one day and then creating a confidence interval using this date, predicting where the rate will be 10 days into the future. This confidence interval is then compared with the true value 10 days into the future to see if the prediction does in fact cover the actual value. If so, the prediction does give us a credible result, so this is a way of checking how good the model is. The result was that the usage of the IMM model, instead of the SA- CCR, would lead to the institute being required to hold 11 % less capital. However, this result is based on only one type of instrument and tested for only a fraction of the institute’s total trading book, so it would re- quire some testing on a bigger scale to really ensure this result. Another thing that could improve this work is if a different simulation process, other than Hull-White, would be used. Some alternative methods that could potentially give a better result and other future improvements are discussed.