The advanced IRBA is the most sophisticated approach available under the regulatory framework for credit risk allowing us to make use of our internal rating methodologies as well as internal estimates of specific other risk parameters. These methods and parameters represent long-used key components of the internal risk measurement and management process supporting the credit approval process, the economic capital and expected loss calculation and the internal monitoring and reporting of credit risk. The relevant parameters include the probability of default (“PD”), the loss given default (“LGD”) driving the regulatory risk-weight and the credit conversion factor (“CCF”) as part of the regulatory exposure at default (“EAD”) estimation. For most of our internal rating systems more than seven years of historical information is available to assess these parameters. Our internal rating methodologies reflect a point-in-time rather than a through-the-cycle rating.

The probability of default for customers is derived from our internal rating systems. We assign a probability of default to each relevant counterparty credit exposure as a function of a transparent and consistent 26-grid master rating scale for all of our exposure excluding Postbank. The borrower ratings assigned are derived on the grounds of internally developed rating models which specify consistent and distinct customer-relevant criteria and assign a rating grade based on a specific set of criteria as given for a certain customer. The set of criteria is generated from information sets relevant for the respective customer segments like general customer behavior, financial and external data. The methods in use range from statistical scoring models to expert-based models taking into account the relevant available quantitative and qualitative information. Expert-based models are usually applied for counterparts in the exposure classes “Central governments”, “Institutions” and “Corporates” with the exception of small- and medium-sized entities. For the latter as well as for the retail segment statistical scoring or hybrid models combining both approaches are commonly used. Quantitative rating methodologies are developed based on applicable statistical modeling techniques, such as logistic regression. In line with Section 118 of SolvV, these models are complemented by human judgment and oversight to review model-based assignments and to ensure that the models are used appropriately. When we assign our internal risk ratings, it allows us to compare them with external risk ratings assigned to our counterparties by the major international rating agencies, where possible, as our internal rating scale has been designed to principally correspond to the external rating scales from rating agencies. For quantitative information regarding our advanced and foundation IRBA exposure based on a rating grade granularity which corresponds to the external Standard & Poors rating equivalents please refer to the section “Advanced IRBA Exposure” and “Foundation IRBA Exposure”.

Although different rating methodologies are applied to the various customer segments in order to properly reflect customer-specific characteristics, they all adhere to the same risk management principles. Credit process policies provide guidance on the classification of customers into the various rating systems. For more information regarding the credit process and the respective rating methods used within that process, please refer to Sections “Credit Risk Ratings” and “Rating Governance”.

Postbank also assigns a probability of default to each relevant counterparty credit exposure as a function of an internal rating master scale for its portfolios. The ratings assigned are derived on the grounds of internally developed rating models which specify consistent and distinct customer-relevant criteria. These rating models are internally developed statistical scoring or rating models based on internal and external information relating to the borrower and use statistical procedures to evaluate a probability of default. The resulting score or probability of default is then mapped to Postbank’s internal rating master scale.

We apply internally estimated LGD factors as part of the advanced IRBA capital requirement calculation as approved by the BaFin. LGD is defined as the likely loss intensity in case of a counterparty default. It provides an estimation of the exposure that cannot be recovered in a default event and therefore captures the severity of a loss. Conceptually, LGD estimates are independent of a customer’s probability of default. The LGD models ensure that the main drivers for losses (i.e., different levels and quality of collateralization and customer or product types or seniority of facility) are reflected in specific LGD factors. In our LGD models, except Postbank, we assign collateral type specific LGD parameters to the collateralized exposure (collateral value after application of haircuts). Moreover, the LGD for uncollateralized exposure cannot be below the LGD assigned to collateralized exposure and regulatory floors (10 % for residential mortgage loans) are applied.

As part of the application of the advanced IRBA we apply specific CCFs in order to calculate an EAD value. Conceptually the EAD is defined as the expected amount of the credit exposure to a counterparty at the time of its default. For advanced IRBA calculation purposes we apply the general principles as defined in Section 100 SolvV to determine the EAD of a transaction. In instances, however, where a transaction outside of Postbank involves an unused limit a percentage share of this unused limit is added to the outstanding amount in order to appropriately reflect the expected outstanding amount in case of a counterparty default. This reflects the assumption that for commitments the utilization at the time of default might be higher than the current utilization. When a transaction involves an additional contingent component (i.e., guarantees) a further percentage share (usage factor) is applied as part of the CCF model in order to estimate the amount of guarantees drawn in case of default. Where allowed under the advanced IRBA, the CCFs are internally estimated. The calibrations of such parameters are based on statistical experience as well as internal historical data and consider customer and product type specifics. As part of the approval process, the BaFin assessed our CCF models and stated their appropriateness for use in the process of regulatory capital requirement calculations.

Overall Postbank has similar standards in place to apply the advanced IRBA to its retail portfolios as well as to the advanced IRBA covered institution and corporate portfolios using internally estimated default probabilities, loss rates and conversion factors as the basis for calculating minimum regulatory capital requirements.

For derivative counterparty exposures as well as securities financing transactions (“SFT”) we, excluding Postbank, make use of the internal model method (“IMM”) in accordance with Section 222 et seq. SolvV. In this respect securities financing transactions encompass repurchase transactions, securities or commodities lending and borrowing as well as margin lending transactions (including prime brokerage). The IMM is a more sophisticated approach for calculating EAD for derivatives and SFT, again requiring prior approval from the BaFin before its first application. By applying this approach, we build our EAD calculations on a Monte Carlo simulation of the transactions’ future market values. Within this simulation process, interest and FX rates, credit spreads, equity and commodity prices are modeled by stochastic processes and each derivative and securities financing transaction is revalued at each point of a pre-defined time grid by our internally approved valuation routines. As the result of this process, a distribution of future market values for each transaction at each time grid point is generated. From these distributions, by considering the appropriate netting and collateral agreements, we derive the exposure measures potential future exposure (“PFE”), average expected exposure (“AEE”) and expected positive exposure (“EPE”) mentioned in section “Counterparty Credit Risk from Derivatives”. The EPE measure evaluated on regulatory eligible netting sets defines the EAD for derivative counterparty exposures as well as for securities financing transactions within our regulatory capital calculations for the great majority of our derivative and SFT portfolio, while applying an own calibrated alpha factor in its calculation, floored at the minimum level of 1.2. For December 31, 2013, the alpha factor was calibrated to 1.22 slightly above the floor. For the small population of transactions for which a simulation cannot be computed, the EAD used is derived from the current exposure method.

For our derivative counterparty credit risk resulting from Postbank we apply the current exposure method, i.e., we calculate the EAD as the sum of the net positive fair value of the derivative transactions and the regulatory add-ons. As the EAD derivative position resulting from Postbank is less than 1 % in relation to our overall counterparty credit risk position from derivatives we consider Postbank’s derivative position to be immaterial.