Migration Rates Modeling as Open Systems II: Toward an IFRS9 Compliant ex-ante Assessment of Credit Risk with Microsimulation Modeling

Since January 1st 2018 the IFRS9-Financial Instruments (International Financial Reporting Standards; BCBE (2015) and ECB (2017)), issued in 2014 by the International Accounting Standard Board (IASB), substitutes the previous IASB39- Financial Instruments. The IFRS9 introduces new challenges for a more reliable and ex-ante well-balanced and prudential assessment of banking book credit risk, that has to be segmented and forward-looking at estimating bank's potential losses. On one hand, according to the accounting standard provision (IFRS9), the ex-ante assessment concerns expected losses due to impairment events of debtors, or rather significant improvements of their creditworthiness. On the other hand, according to the prudential standard (Credit Risk Directive, CRD (2013), and Credit Risk Regulation CRR (2013)), capital requirements should be consistent with unexpected losses (i.e. the difference between the maximum loss observable with a given confidence level -VaR Approach- and the expected loss). Although within the frame of a simplified microsimulation modeling, this paper proposes a versatile enough approach to account for both expected and unexpected losses with a single methodology, satisfying both accounting and prudential requisites baselines: often, the two types of losses are treated separately with different approaches and this might lead to some inconsistencies. Also, the estimate of potential losses should be based on past and contingent events as well as future macroeconomic scenarios. Moreover, the ex-ante assessment should be segmented to the extent that, based on the risk related to the creditworthiness, each position in the banking portfolio should be classified into appropriated buckets, each specific to the time-regime at which expected defaults may happen in order to assess a reliable economic capital estimate, not depending on their contractual maturity. The main aim of this paper is to involve the IFRS9 standard baseline into a simplified credit risk system migration rate modeling which includes entry (E), stay (S) and leave (L) positions - namely the E-S-L modeling developed by Landini et al. (2018): the paper proposes a microsimulation approach to the balanced estimate of bank’s provision. Without loss of methodological generality, but with the aim to specify a simplified methodological environment for estimating expected (EL, accounting) and unexpected (UL, prudential) credit losses, the paper develops upon a single credit line in the portfolio. Consistently with the IFRS9, differentiated positions are considered with two main maturities: some positions have a short-run maturity while some others have a long-run one. Improving the E-S-L methodology, each position in the banking book is classified according to its own creditworthiness over a master scale of several grades - including the default grade - and migration matrices are obtained accordingly. Period-specific E-S-L matrices, that allow for cure events, are declassified according to the IFRS9 3-buckets model. The paper provides a first level of inclusion of the IFRS9 standard into the E-S-L modeling while satisfying the segmentation criterion and allowing for setting-up the whole modeling to the end of prospective estimate of those parameters that rule the ex-ante evaluation of credit risk (both in terms of EL and UL), either in the short (namely a 1- year) and in the long (i.e. full-loan-lifetime) run expectations. The second level of IFRS9 inclusion consists in completing the forward-looking parameters' estimate conditioned upon future forecasting about a systemic scenario.