Cash Flow Underwriting Adverse Selection Problem

“Cash flow underwriting has an adverse selection problem. Or not?” The interesting part is that the people saying it and the people pushing back are often both right. If you’re saying it does have adverse selection problems internally, it’s rarely a slam dunk. It has to be a balanced approach. Risk-only data shows one slice of the picture. Not what that same ask did to growth and conversion. And you usually have to justify the trade-off as a business decision, not solely a risk one.

The case for:

• 👉 Usually people with nothing to hide would connect a bank account without hesitation. Makes us think that completion skews toward stronger cash flow
• 👉 Cash flows capturing the current/forward-looking financial picture does add significant prediction value, which can be passed on as higher approvals
• 👉 In thin-files & SMBs willingness to share has become a quality signal in itself
• 👉 Liabilities aren’t captured accurately without cashflow data, causing payments/budgeting to become burdensome

The case against:

• 👉 Super-prime borrowers with small needs do abandon when asked for info they don’t think they need (anything except income & liabilities)
• 👉 It’s product-dependent: high-need products (mortgage) get the connection; low-need ones lose people to friction
• 👉 Banking is often split across accounts, so completing the picture is a genuine effort
• 👉 Some still simply do not trust the data sharing process.

Drop-off is multi-causal, so it’s easy to misread the funnel and over or under-react. Adoption is improving, but it still isn’t free. Imho, it won’t be free until a push mechanism like credit reporting exists for cashflow CRAs.

From a practitioner’s pov question isn’t “does adverse selection exist”, rather it’s “how do I introduce this without denting growth, and lean in harder where the risk case is strong.”

A few practical things I usually suggest worth trying:

• 👉 Treat it from a UX perspective: Make the permissioning feel like part of the flow, rather a hard gate
• 👉 Reframe the value (eg. for super-prime): Tie sharing cashflow data to a benefit like better rate, higher limit, instant decision
• 👉 Try a second look: Although such programs go into high risk territory where risk appetite itself becomes bottleneck, plus ‘completion skewing towards strong cash flow’ case is strong
• 👉 Provide Options: For those still not trusting or preferring to share bank data through aggregators, keep bank statements as an option, document fraud needs to be checked in this case
• 👉 Lean in where confident: Introduce product & segment-wise eg. thin/no/bad credit

This is roughly where CILayer fits. We accurately enrich cash flow data across on-us relationships, off-us aggregators as well as bank statements (with extraction & fraud detection). CILayer also turns enriched transactions into the predictive signals that matter from multiple perspectives in addition to just risk. Glad to discuss your challenges & unique approaches.

#CashFlowUnderwriting #CreditRisk

---

Originally posted on LinkedIn.