2.6 The department has worked with the Bank to develop analytical and forward-looking expected credit loss models that are compliant with International Financial Reporting Standards (IFRS 9). These models provide a sophisticated approach to forecasting expected credit losses across the COVID-19 loan guarantee schemes, utilising granular data from lenders and other sources. 2.7 For BBLS, the models include a specific input to reflect the probability that a loan facility may have been contracted fraudulently by the borrower. The fraud input is based on the results of sampling work completed by PwC in 2021. The department will continue to work with the Bank and Cabinet Office to better understand the links between fraud risk indicators, fraud occurrence and fraud loss, which will help to refine our methodology. 2.8 The accuracy of the models will improve over time as more information becomes available. However, there will continue to be a high degree of uncertainty and actual losses will depend to a significant extent on the wider performance of the economy over the next decade. The department is committed to providing regular updates on these estimates via its Annual Report and Accounts and supplies estimates to the National Audit Office to inform their regular Covid Cost Tracker publications.