First, thanks to Giuliano Bologna for providing the analysis at my request. Second, as I’ve stated in other forums, I respect Giuliano’s capabilities as an analyst and the work that he does, and his diligence in extensively showing how he arrives at his conclusions.

Few companies disclose everything we would like to know as analysts because at some point too much disclosure starts to give away too much information to competitors. So, we analysts are forced to read between the lines, use market-based data, comps from comps, and, as I’ve done many times in my career, securitization data from loan-selling lenders. Giuliano is skilled in these techniques.

However, in his latest work, he’s missed a couple of things that SoFi has disclosed and has filled in alternative assumptions that overstate his case. Also, in cases where he’s had to make assumptions, he has filled in worse assumptions than I think are justified.

If he is correct and if there is an actual catalyst that would force SoFi’s mark close to his (two pretty big “ifs” especially when compounded), he correctly identifies the consequences as significant: SoFi would miss guidance for the first time in its history.

First, and foremost of the assumptions I must question, is Giuliano’s assumed loss rate. At 5%, this estimate significantly influences calculations of ultimate investor returns because it drives the risk-adjusted yield, which then drives the magnitude of the leverage necessary to achieve an “acceptable” levered yield.

Giuliano gets this every bit as well as I do:

Note, SOFI has said that the company has significantly tightened underwriting on more recent origination vintages and that should lead to lower annualized NCOs for the loans sold vs. retained. For every 50 bps reduction in annualized NCOs for the 2023-1 scenario, we estimate the implied FV would increase by 75 bps.

Recent Securitization Data Adds to List of Questions
About SOFI’s FV Marks/Assumptions by Giuliano Bologna

The losses in personal loans were running at 4% in 4Q23. As Giuliano points out, SoFi has tightened underwriting standards. Chris Lapointe pointed out in a recent investor conference that 15% of the losses in the quarter came from 3% of the loan book that represents the loans that the tightened underwriting has taken out of the system.

Another component of Giuliano’s loss rate assumption that is questionable is that he assumes a weighted average seasoning of SoFi’s fair value loan book of ~6-8 months and extrapolates from this that cash flows are lower in the on-balance sheet FV loan book. When in fact, this average represents a combination of older loans that have seen their peak losses and new loans that are in their high-cash flow period and were underwritten to tightened standards. Focusing on the average makes it seem as though the entirety of SoFi’s on balance sheet personal loans are only now ready to start their high credit loss phase.

The final major assumption that Giuliano must intuit (since it isn’t disclosed by SoFi), is the rate that loan purchasers are getting to lever their purchases to achieve mid-single digit returns, which he puts at ~6.18% (which he backs into assuming what he views as comparable tranches of other securitizations at the analogous tranche level). However, this assumption doesn’t represent SoFi’s loan pricing dynamics at all. In theory, SoFi could make these loans at rates as low as there deposit cost and have make all the sense in the world to them from a capital management perspective.

I trust that if you were to make these two adjustments, as well as others that actually reflect SoFi’s actual transaction dynamics (like SoFi’s underwriting is more trustworthy than its competitors even at similar FICOs) and that don’t rely on extrapolated/estimated assumptions that don’t match reality, you’d get far more comfortable SoFI’s loan mark and its, so-far, completely reliable guidance.