Can Cash Flow Underwriting Create More Equitable Lending?

For lenders trying to separate transient setbacks from true repayment risk, how a borrower manages money today is more revealing than what happened years ago. 

‍

That’s the core insight behind cash flow underwriting, and it’s grounded in both behavioral economics and a growing body of empirical evidence.
‍

The behavioral economics case: recent habits beat old history
‍

Behavioral economics teaches that people exhibit present bias and adjust behavior as circumstances change. In credit, that means a borrower’s current habits, such as income regularity, spend patterns, cash buffers and bill repayment behavior, can signal future performance more reliably than an outdated negative mark.

‍

Two practical points follow:
‍

• Recency matters.
  
  Even in traditional scoring, recent negative information weighs more than older events. This means relying on a 10‑year‑old credit event can be economically misleading.
  
  
• Liquidity and volatility drive outcomes.
  
  Real‑time checking‑account data captures whether households can absorb shocks. For example, households typically need about six weeks of take‑home income in liquid assets to weather a simultaneous income dip and expense spike; 65% don’t have that buffer, according to this research by JPMorgan Chase. 

What the evidence says (and why it matters now)
‍

Independent research has also found that bank‑transaction (“cash flow”) features are predictive of default and complement bureau scores:

‍

FinRegLab examined data from six lenders and found that cash‑flow variables on their own performed about as well as traditional scores. It also demonstrated improvements to automated underwriting for its ability to provide a fuller view of how applicants manage their finances. improved prediction when combined with bureau data.

‍

FinRegLab later built head‑to‑head models and published in their 2025 research findings that the most predictive model combined cash flow data and bureau data using machine learning (ML).

Additionally, the National Bureau of Economic Research (NBER) conducted research that revealed augmenting FICO with cash flow metrics raises approval odds for younger entrepreneurs without worsening defaults, because cash flow features (e.g. balance volatility, overdrafts, revenue inflows) carry information that long credit histories don’t. 

‍

Together, the evidence shows that cash flow data can reliably predict risk and, paired with bureau scores, expands access without raising defaults.
‍

What real‑time spending and saving patterns reveal
‍

Cash flow‑based features provide live measures that make behavioral credit risk observable. Some examples of this include: 

‍

• Income regularity & concentration: Is pay stable (e.g. steady payroll deposits) or gig‑heavy with high variance?
  
  
• Buffer strength: Average balances, days‑cash‑on‑hand, and drawdowns can indicate resilience; thin buffers can precede delinquency spikes.
  
  
• Obligations vs. inflows: A cash flow debt-to-income (DTI) (recurring outflows divided by recurring inflows) can highlight affordability more precisely than statement DTIs for variable‑income borrowers.
  
  
• Payment discipline: On‑time utilities/rent, subscription churn, and overdraft/NSF incidence can act as operational measures of self‑control and shortfall risk visible in transactions that are largely absent from bureaus.
  
  

These signals demonstrate the potential for reduced discriminatory power in some segments, making fresh data even more valuable.
‍

Compliance, fairness, and governance aren’t afterthoughts
‍

Cash flow underwriting intersects naturally with compliance and fair lending when grounded in transparency and explainability. 

‍

Under the Equal Credit Opportunity Act (ECOA) and Regulation B, lenders must base decisions on permissible, explainable factors and provide clear adverse action reasons when credit is declined or terms are unfavourable. Cash flow underwriting creates the ability to support these obligations when features are defined in plain language (e.g. income regularity, cash buffers, recent payment performance) and can be mapped directly to adverse action reasons.

This enhances auditability, reduces reliance on opaque proxies, and helps ensure that notices accurately reflect the factors driving a decision, aligning modern data use with ECOA’s transparency and fairness requirements.

How to put cash flow underwriting into production 
‍

Every lender starts from a different place, whether it’s portfolio mix, data coverage, governance appetite, and systems constraints. When it comes to putting cash flow underwriting into production, the goal is to sequence sensible steps that fit your risk objectives and policy framework. 

‍

You could use the below as a basic guide:

1. Define the economic objective up front. Be explicit about the trade-off you’re targeting (e.g., a measured lift in approvals at flat expected loss) and the guardrails that matter.
   
   
2. Engineer a transparent feature set.  Prioritize plain-language signals (e.g. income regularity, deposit volatility, days cash on hand, cash flow DTI, housing/utility payment cadence, overdraft/NSF incidence) built on stable windows.
   
   
3. Test before you scale. Run champion-challenger evaluations: shadow first, then stage exposure with expand/stop rules tied to loss and accept rates. Keep adverse action reason codes consistent across iterations.
   
   
4. Blend data where it helps. Combining bureau and cash flow data has been shown to outperform either alone. Apply ML where your governance and explainability standards can support it.
   
   
5. Monitor fairness and drift continuously. Track performance, drift, and fairness across segments (e.g. thin file, variable income) and refresh thresholds and models as conditions change.
   
   

Want to move faster? Carrington Labs’ Cashflow Score and Cashflow Servicing help lenders stand up governed cash flow underwriting quickly so you can test, learn, and scale with confidence. Contact us to find out how.
‍

Key takeaways

1. Behavior today predicts behavior tomorrow. Real‑time income, spend, and savings patterns provide a sharper, fairer view of credit risk.
2. Combining bureau and cash‑flow data can be more predictive and increase approvals at constant risk, particularly helping young/thin-‑file borrowers.
3. Compliance isn’t a trade-off: explainable cash flow features map to ECOA adverse action reasons, improving transparency and auditability.
   ‍