In this episode we take the opportunity to talk to one of the most credit risk experienced men in Fintech who has ranged from a dozen years at seminal Capital One through managing Barclay’s £40 bn portfolio, being CRO at Funding Circle to now founding his own Fintech to mechanise further short term SME liquidity financing, Jerome’s stats are that banks automate roughly 5% of this type of lending whereas he believes the vast majority to be automatable.

We discuss what credit risk is in the age of data, automation and Fintechs and dive into the multiplicity of models used in a credit pipeline not just for risk but for wider business management and portfolio management purposes as well as looking at the parameters of automation versus human input. We also look at Jerome’s experience of what is stable about the models value and what is not over significant economic dislocations.

Importantly there are a huge range of uses of credit models above and beyond the simple case of credit risk and Jerome gives us a flavour of these across his own career – a fascinating stat being the incarnation which had 25 models in operation :-!

Topics discussed include:

• Jerome’s background in the Art world as well as Banking world
  
• Jerome’s career journey through credit in various incarnations
  
• running a team of 1,000 risk metrics folk
  
• developing an interest in SMEs
  
• 7 years at Funding Circle
  
• defining risk, uncertainty and so forth
  
• the many different types of risk through the whole lending soup to nuts process
  
• the balancing act on decisions at each stage of this process
  
• using multiple models and cross testing them against alternatives to gain greater insight – challenger models
  
• Barclaycard had about 25 different models for different stages of the credit card process
  
• keeping the models fresh and tuned
  
• not using “black box” models
  
• challenges of models coping with continuous and discontinuous phenomena
  
• practical approaches to life’s discontinuities in the lending world
  
• lending versus trading risk in this context
  
• ranking of risk tends to be stable across perturbations even if the level of risk gets well off
  
• the commercial importance of  the relative ranking in itself to make the lending decisions within a given risk appetite at any point
  
• experience at Barclays and Funding Circle over significant economic dislocations
  
• requirements to create a model in the first place
  
• the role of human creativity and comparison with human decisions making versus a “computer”
  
• Jerome’s many tests over his career with automated versus underwriters credit decisions and the results
  
• the requriem3ents for automatable credit decisions in the SME sector – where it can and cannot be done well
  
• the irony that larger companies have more data but smaller sample size
  
• Jerome’s motivation in establishing his own firm to do short-term SME cashflow management
  
• the business model – API- and partner- centric
  
• leveraging external creativity in this model – cf ChatGPT
  
• the differing business model viability compared to banks
  
• decisions in 2.5 minutes
  

And much much more