Last Week in Lending: Tuesday, 28 November

?? A lot of incredibly valuable research has been released in the past month on trends relevant to FinTech and banking… here are 3 pieces I think every credit and risk expert should read:

1. ‘Our Money in Data’: A deep dive into US income statistics ??

Francisco Javier Arceo spent six months pulling, cleaning, and revising a breakdown of US income data – something that should be very insightful for anyone working in consumer finance.

While I recommend taking an in-depth read of the article, some key highlights:

• Many Americans lack significant savings and don’t have a FICO score exceeding 750.
• The largest employer in the United States is the government.
• Individuals with less than a high school diploma typically have zero investment income – compared to those with a graduate degree who have an average investment income of approximately $7,200.
• Approximately 25% of Americans receive some form of government-provided income, and for those individuals, it makes up an average of around 65% of their total income.

Francisco has also published an open-source Python library (PyIPUMS) for people to work with the data, which is incredible.

2. MIT x UBS report: Finding value in generative AI for financial services ??

The TL;DR, according to MIT & UBS: Generative AI is a valuable tool, but yet to be disruptive.

With ChatGPT reaching 100 million users in record time this year, it’s sparked interest in every sector of the world. But as this latest report has reiterated, Generative AI tools aren’t quite ready to transform entire industries yet – particularly financial services.?

Deploying generative AI in financial services poses challenges from both a technological and regulatory standpoint.

This report highlights how FinTechs and banks wanting to leverage generative AI in high-risk activities like lending face major hurdles – as even “pre-packaged” tools require a ton of time-intensive customization (for example, training on proprietary data) to be relevant for their specific use cases.

Plus, generative AI models can introduce a variety of regulatory issues, including biases, lack of accountability, and the difficulty of validating complex output. So, even though the financial sector is gradually adopting generative AI, its current applications are still restricted to automating lower-value activities under close human supervision to mitigate risks.

In terms of credit underwriting specific applications of generative AI, I spoke about this on a recent Money/2020 panel – so if you’d like to dive more into that topic, I recommend checking out the recap.?

3. Experian’s State of Consumer Credit Report for Q2 2023 ??

Experian’s latest consumer credit report also features some useful consumer finance statistics – painting a picture of what lenders could expect from the US consumer credit market.

Although employment levels and credit scores remain relatively stable, incomes aren’t keeping up with inflation. And due to a higher interest rate environment, monthly debt repayments continue to increase.

Some key insights from the report:

• Looking at various debts like auto loans, personal loans, and credit cards (excluding mortgages), the average amount people owe went up by 4.7% to $23,317. During the same time, prices, in general, went up by 3%, according to the consumer price index.
• In Q2 2023, the average amount owed on credit cards by consumers was $6,365, which is 11.7% more than it was in Q2 2022.
• At the same time, the average percentage of credit used by people went up from 27% to 28% in 2023.
• The overall average credit score (FICO Score) in Q2 2023 improved a bit, going from 715 to 716.
• On average, people had to pay 8.5% more each month to cover their debts in the year leading up to Q2 2023.
• With credit card interest rates at 22% in mid-2023, people who have balances on their cards might feel more constrained than usual – which could limit further spending.