Singularity Bank by Scardovi Claudio

Author:Scardovi, Claudio
Language: eng
Format: epub
Publisher: EGEA Spa - Bocconi University Press

Table 5 How an AI-based credit work out approach works: optimizing recovery value

The best recovery strategy, defined at the level of the single cluster, with the specifics and compositions of the clusters being redefined continuously by the machine as it keeps learning, so as to best fit the available “universes” with the “explainable behaviours”, can then be broken down to the level of the single action, at any point in time, to the name and dossier levels. It follows also that the internal organization and operating model and related key processes of a credit work out unit could be redefined on the basis of this “optimal recovery strategy” and based on an “optimal allocation of clusters/dossiers to recovery teams”. Moreover, optimal selection of third parties (lawyers, for example, matching the best “holder” of a dossier, given the counterpart, collateral, legal stage and location) can be derived as part of optimization across the value chain stages.

Several intelligence explosions and relative competitive advantages could then be derived, in a business notorious for its murkiness and opacity and less for its level of data science optimization and overall value maximization. Principal investors, for example, would be able to know better what maximum price they could bid, given the optimal recovery curve they could attain and their required internal rate of return. And the securitization of NPL portfolios (or of their repossessed collaterals) could also become more efficient and effective in their aggregation of diverse risk profiles, making their successive placement to third-party investors quicker, cheaper and fairer.

Other “deep and narrow” AI applications could then happen in the space of NPL stress testing, simulation analysis and revaluation on an IFRS 9 basis, as this digitized credit work out value chain would allow for quick and periodic “mark to market” of the portfolios’ fair value and for their contingent management under different scenarios. Following a “singularity dream” characterized by an abundance of information on NPL, being continuously refreshed and available for free, and the help of some powerful AGI app, it would not be impossible for this business to reach the transparency of regulated capital markets and the efficiency of algorithmic based trading, with the NPL related to a small business based in the Attica region traded at the same velocity, liquidity, transparency and efficiency/efficacy of the USA Treasury Bill or of the German Bund.

These would be great achievements for the loan component of NPL, but a fascinating intelligence explosion would also be possible regarding their collaterals, starting from the real estate ones that are obviously a key component of the credit underwriting and recovery strategy of a bank. The digital disruption, in this field, would start from the holistic and dynamic valuation of the property, aggregating unstructured data and using AI-based apps and making the entire appraisal process more automated and robotized. This is already happening, with a more and more extensive use of machines used instead of property valuers,10 as they promise to deliver augmented intelligence along all of its three dimensions.

The opportunity

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.