AI and Proximity in U.S. Bank Small-Business Lending

Overview

Using the FDIC’s 2022 survey and a February 2025 working paper on 2017–2019 data, this series asks whether AI can reduce the distance penalty in U.S. bank small-business lending. It explains soft information, maps automation, tests lending geography and outcomes, and examines start-ups, trust, monitoring, and flexibility. The evidence suggests AI may extend selected distant lending; studied SBA 7(a) results are consistent with better screening, while branches and relationships remain important.

Learning outcomes

AI may lower the cost of using hard information by extracting, integrating, and interpreting messy records.
In 2017–2019 data, higher AI use was linked to a weaker distance penalty, clearest beyond 1,000 miles—not to more lending everywhere.
In the studied SBA 7(a) sample, higher AI use was associated with lower charge-offs and interest spreads for distant borrowers, consistent with better screening.
AI can stretch informational proximity, but it does not supply trust, local knowledge, monitoring, or flexible judgment.

Keywords

AIProximitySmall-business lendingUnderwritingSoft informationHard informationDistanceFinTechStart-upsSBA 7(a)

Chapters

Can AI Replace Proximity in U.S. Bank Small-Business Lending?

How to distinguish stretched reach from true replacement

10 min · How to distinguish stretched reach from true replacement

10 min

Why U.S. Bank Small-Business Lending Stays Local

How local relationships solve an information problem

12 min · How local relationships solve an information problem

12 min

What U.S. Banks Automate in Small-Business Lending

Separating workflow support from AI underwriting

15 min · Separating workflow support from AI underwriting

15 min

Testing AI’s Distance Effect in U.S. Bank Small-Business Lending

The proxy, data, and limits behind the test

11 min · The proxy, data, and limits behind the test

11 min

AI, Distance, and Screening in U.S. Bank Small-Business Lending

What the 2017–2019 working paper can support

13 min · What the 2017–2019 working paper can support

13 min

Where Proximity Still Matters in U.S. Bank Small-Business Lending

Start-ups, thin files, and the case for hybrid review

11 min · Start-ups, thin files, and the case for hybrid review

11 min

AI Can Stretch Proximity—Not Erase It—in U.S. Bank Small-Business Lending

How to judge reach, trust, and residual risk

13 min · How to judge reach, trust, and residual risk

13 min

Preview every chapter · Unlock full text, audio, and video from chapter 4

Plans

Ch.1-3 FreeCh.4-7 Premium

FAQ

Can AI replace proximity in U.S. bank small-business lending?

Not completely. In U.S. bank small-business lending, AI may make hard information more usable and extend selected distant loans, but the evidence does not show that branches, relationships, local knowledge, monitoring, or flexible judgment are unnecessary.

Why has U.S. bank small-business lending stayed local?

Because local interaction helps banks gather soft information, verify unusual facts, monitor borrowers, build trust, and find referrals; distance makes those activities more costly. Digital tools can reduce travel or paperwork without recreating all of that context.

What does the evidence show about AI and distant small-business loans?

Evidence from 2017–2019 bank data shows higher AI use was associated with a weaker distance penalty, clearest beyond 1,000 miles. In the studied SBA 7(a) sample, distant loans also had lower charge-offs and interest spreads, consistent with better screening rather than simple risk-taking.

Which small businesses are least suited to fully automated lending?

Start-ups, thin-file firms, and unusual businesses are less suited when key evidence is contextual or missing. Human review, a relationship, or an SBA guarantee may help, but the evidence does not show that any one path works for every borrower.