Risk.net
Latest on Risk management
Singapore banks tighten ML governance amid regulatory scrutiny
DBS, StanChart and Deutsche build model inventories and draw up standards around use cases
EU banks rue SA-CCR mismatch with US
European clearers are stuck with CEM until 2021, but some US banks are reluctant to switch early
CECL models may leave banks ill-prepared for next downturn
Mortgage backtest study shows some loan-loss models miss the mark
Latest
Quotes
The leverage ratio offset lowers the bar a bit for new entrants. But there is a fundamental profitability issue within the clearing ecosystem. It’s a business of scale, where you have a very high fixed cost base and any incremental revenue is dropped to the bottom line. I think many firms are failing to make a profit, never mind making a good return on capital
John Dlubac, Emea head of prime derivative services at Credit Suisse
We still believe the SA-CCR framework is too conservative and does not accurately reflect the market
Panayiotis Dionysopoulos, the head of capital at Isda, highlights that despite tweaks to SA-CCR in the US implementation, the trade body would like to see additional changes made.
What we increasingly do is engage with customers to find out what’s creepy and what is not
Paul Cobban, chief transformation officer at DBS, says the bank is seeking input from its customers on its use of machine learning models
Editor's Choice
BNP leads a comeback for Europe’s clearers
Brexit, leverage ratio tweaks and concentration fears could help European banks compete with US FCMs
Big Figure
Weathering the storm(s)
The difficulty of modelling natural catastrophes is getting more extreme with the march of climate change. Across the five largest loss events from 2018, average losses estimated by the market’s two largest providers of cat models – AIR ($15.75 billion) and RMS ($12.75 billion) – were roughly 65% below the true average loss figure of $40.3 billion.
Read the full article7 days in 60 seconds
SA-CCR, IM relief and the fuzziness of good behaviour
The week on Risk.net, February 15-21, 2020
Receive this by email
Comment
When the data’s not there, expert-led models could help
The end of the fintech gold rush?
Our Take
No silver bullet for AI explainability
Grand designs? Time to rein in the Pillar 2 project