Is there anyone using AI for credit-risk documentation, monitoring. Our view is it creates generic fluff but since doesnt have full access to credit regulations, full set of models, policies, the answers almost always need rewrite again. If someone trains an internal llm on a set of procedures, documents, it might work, but we dont have budget to test this

Are banks considering model risk as a Level 1 risk or as a Level 2 risk within the risk taxonomy?

UK banks are currently re-evaluating their FIRB vs. AIRB decisions. There is not yet clarity around PRA’s views around their acceptance around having “partial FIRB” or if banks would go to FIRB for all Wholesale (across turnover and country segments). This could be a point to highlight to PRA if there is going to be clearer guidance in the near future.

Hi all - not sure what the etiquette is on here, but I have a separate (unrelated) question - I'll just ask it here - but if there is a better place, can the site moderator move it please. I wasn't at the PRA Round Table event, but there was mention of some proposed actions (1 or 2 attendees had a clear view) - were the actions documented anywhere? Or if not, would somebody be able to provide a short summary of the key points we raised with the PRA for action? It sounds like there has been no progress on this.

Hi, PRA refer to Banks as "data rich" and reserve the term "LDP" to portfolios with fewer than 20 defaults; I presume you have more than 20 Bank defaults. For LDPs, PRA expect particular calibration techniques to determine long-run average default rate such as Pluto-Tasche.