Table 16: If within limit throughout the quarter

Within the limits approach, ADIs may choose one of three approaches to calculate internal limits: the contingent loss method; a VaR model based on the same parameters for the holding period and confidence interval as specified in the internal model approach (i.e. 10 days and 99 per cent (refer paragraph 22 of Guidance Note AGN 113.2 The Internal Model Approach)); or a VaR model based on a different holding period and confidence interval to those specified in the internal model approach.

For ADIs using the contingent loss method or a VaR model based on a 10 day holding period and 99 per cent confidence interval, the resulting limit for each asset class should be entered in the column labelled Internal Limit. The sum of these limits should be entered in the row labelled Total. ADIs using a VaR model based on parameters for the holding period and confidence interval which differ from those required by the internal model approach should first enter these limits for each asset class in the Internal Limits column. The limits should then be re-scaled to obtain 10 day, 99 per cent equivalents. If an ADI specifies internal limits based on a VaR model which uses a 1 day, 95 per cent (two tail) confidence interval, for example, the internal limits should be re-scaled to 10 day, 99 per cent equivalents by multiplying each limit by 10 (to take the holding period up to ten days), dividing the limit by 1.96 (to reverse the 95 per cent scaling effect) and then multiplying the result by 2.33 (to arrive at a 99 per cent confidence interval). The total internal limit is then the sum of the re-scaled limit amounts.

For all three methods, the scaling factor consists of a multiplication factor and a plus factor (refer paragraph 22(j) of Guidance Note AGN 113.2 The Internal Model Approach). The multiplication factor is determined by APRA on a case by case basis.

The capital requirement is the product of the total internal limit across all asset classes and the scaling factor.

Backtesting exceptions

To complete this section the most recent 250 days of aggregated profit and loss and VaR data, up to and including the last day in the quarter, is needed (refer paragraphs 54 to 82 of Guidance Note AGN 113.2 The Internal Model Approach). The aggregated data should be that profit and loss and VaR data that applies to all asset classes for which the internal limit method is being used to calculate the capital charge. Using this data, compare each day’s trading outcome (profit or loss) with the corresponding VaR number. If the trading outcome on a particular day is a loss that exceeds the corresponding VaR number for that day then the result is recorded as an exception. Count the number of exceptions that occur over the 250 days.

The Prudential Standard allows ADIs to perform back testing based on either actual or hypothetical profit and loss. All ADIs using the internal limit method must complete at least one of the two back testing rows in Table 16. Those ADIs that have the capacity to back test using both actual and hypothetical profit and loss must report the results of back testing on both bases.