Credit risk migration rates modeling as open systems: A micro-simulation approach

The last financial crisis of 2008 stimulated the development of new Regulatory Criteria (commonly known as Basel III) that pushed the banking activity to become more prudential, either in the short and the long run. As well known, in 2014 the International Accounting Standards Board (IASB) promulgated the new International Financial Reporting Standard 9 (IFRS 9) for financial instruments that will become effective in January 2018. Since the delayed recognition of credit losses on loans was identified as a weakness in existing accounting standards, the IASB has introduced an Expected Loss model that requires more timely recognition of credit losses. Specifically, new standards require entities to account both for expected losses from when the impairments are recognized for the first time and for full loan lifetime; moreover, a clear preference toward forward looking models is expressed. In this new framework, it is necessary a re-thinking of the widespread standard theoretical approach on which the well known prudential model is founded. The aim of this paper is then to define an original methodological approach to migration rates modeling for credit risk which is innovative respect to the standard method from the point of view of a bank as well as in a regulatory perspective. Accordingly, the proposed not-standard approach considers a portfolio as an open sample allowing for entries, migrations of stayers and exits as well. While being consistent with the empirical observations, this open-sample approach contrasts with the standard closed-sample method. In particular, this paper offers a methodology to integrate the outcomes of the standard closed-sample method within the open-sample perspective while removing some of the assumptions of the standard method. Three main conclusions can be drawn in terms of economic capital provision: (a) based on the Markovian hypothesis with a-priori absorbing state at default, the standard closed-sample method is to be abandoned for not to predict lenders’ bankruptcy by construction; (b) to meet more reliable estimates along with the new regulatory standards, the sample to estimate migration rates matrices for credit risk should include either entries and exits; (c) the static eigen-decomposition standard procedure to forecast migration rates should be replaced with a stochastic process dynamics methodology while conditioning forecasts to macroeconomic scenarios.