1.38Real lending data from an open research dataset
What each step of risk appetite costs
points of return given up for each point of loss appetite at the tightest setting tested, against 0.37 near the top
- What was observed
- 33,257 finished loans in 12 segments by grade and term, measured on two years of lending, and the 53,367 loans written the year after.
- What a generic approach says
- Lend to the book as it comes. Here that put 43% of the money in 60 month loans and 16% in the riskiest grades, breaking two of the limits a lender would set.
- What the engine read
- Allocated a $148.7m budget with at most 30% in 60 month loans, 15% in grades E to G and 25% in any one segment. The highest return lost 10.29% of the money lent; the lowest loss these limits allow was 5.73%. Between them, each point of loss appetite given up cost 0.37 points of return near the top and 1.38 at the tightest setting, and below what the limits allow it made no recommendation.
- What it meant for the decision
- The appetite has a visible price before the committee sets it, and tighter appetites gave lower losses and lower returns on the next year's loans too. That year every appetite was exceeded by about a fifth, because losses rose about 1.2 points a year across the book, so an appetite set on past loans needs a margin and a check on each new year of lending.