Why Small-Business Lending Has Been Local
U.S. bank small-business lending has remained local because proximity helps solve an information problem, not merely because borrowers like visiting a branch. The useful question is how local interaction, standardized records, and bank scale combine to produce a lending decision. Understanding that combination lets us see which part of proximity AI might eventually reduce—and which parts are still tied to people, place, and judgment.
Soft information is not just missing data
A small-business loan begins with uncertainty on both sides. The business needs capital to buy equipment, hire staff, manage a seasonal gap, or survive an unexpected shock. The bank needs to estimate whether the business can repay, what could go wrong, and how much it would lose if repayment fails. A large company may provide long financial histories, audited statements, and market disclosures that make this assessment relatively portable. A small firm may have useful evidence, but in forms that are incomplete, inconsistent, or meaningful only in context.
Banks commonly describe the evidence in two broad categories. Hard information is numerical and standardized: a credit score, a debt-service ratio, a financial statement, a record of delinquency, or the value of offered collateral. It can be entered into a system and compared across applicants without preserving every detail of the setting in which it was produced. Hard information is not necessarily more truthful. It is simply more portable and easier to transmit through a multi-level organization.
Soft information is qualitative, contextual, and harder to verify or reduce to a single number. It can include a loan officer’s assessment of the owner or management team, the plausibility of a business plan, the experience of the people running the firm, the condition of a local market, or an explanation for an unusual financial result. A business with a temporarily weak margin may be a poor credit risk, or it may have made a deliberate investment that a knowledgeable reviewer can understand. The distinction is not between objective facts and subjective feelings. It is between information that travels cleanly and information whose meaning depends on context.
That difference creates a distance penalty. A bank can gather soft information by meeting a borrower repeatedly, visiting the business, observing local customers and suppliers, and speaking with people who know the firm. Those activities become more expensive as the borrower moves farther from the bank’s offices. Monitoring also becomes harder: a distant bank may have to rely more heavily on documents, scheduled calls, or third-party reports when it wants to know whether a business is operating as expected.
The result is not that a distant business cannot be creditworthy. It is that the bank has less of the information it normally uses to distinguish a strong but unusual applicant from a weak one with attractive numbers. When the cost of collecting that context is high and the potential loan is small, the bank may decide that the expected return does not justify the additional work. Distance can therefore restrict credit even when the physical act of transferring money is effortless.
The FDIC’s report uses the language of relationship and transactional lending to describe different ways of handling this problem. Small banks are generally associated with relationship lending because they gather and use more soft information. Their loan officers may have fewer layers of management between themselves and senior decision-makers, and their local presence can reduce the cost of knowing a borrower and the surrounding market. Large banks are generally associated with transactional lending because their scale makes it worthwhile to standardize large amounts of hard information and move it through repeatable processes.
These are useful tendencies, not fixed identities. A small bank can use credit scores, and a large bank can maintain personal relationships. The important question is which information carries the most weight when a case reaches an approval decision, especially when the loan is small enough that the bank cannot justify unlimited investigation.
The local bank is an information system
It is easy to treat a branch as a distribution point: a place where a borrower hands over forms and a bank hands over funds. In small-business lending, the branch can be part of the bank’s information system. It gives a borrower a place to ask which product fits, gives staff a reason to learn about the business before an application arrives, and creates a recurring channel for questions after the loan is made. A local bank may also hear about a potential borrower through an existing customer, an accountant, a supplier, or a community organization. These connections are not automatically reliable, but they widen the bank’s access to context at a cost that is often lower than searching for distant applicants one by one.
A meeting with a decision-maker has a more direct underwriting role. The FDIC survey finds that nearly 90% of small banks have decision-makers meet with applicants at some level of approval, compared with 39% of large banks. For a young firm or a business with unusual records, the meeting can expose information that a form does not contain: how the owner understands the business, whether the proposed use of funds is coherent, and whether an apparent weakness has a credible explanation. It can also allow the banker to ask follow-up questions rather than rejecting an application because one field is missing.
The meeting is not a guarantee of approval. It can reveal adverse information as easily as favorable information, and personal judgment can be biased. But it changes the bank’s information set. It also gives the borrower a chance to explain a situation in which standardized evidence is incomplete. That is why the FDIC report connects meetings with flexibility and with the ability to lend to some marginal borrowers, including start-ups.
Proximity continues after approval. A business loan is not only an origination decision; it is a relationship that produces feedback over time. Staff may notice whether the business is expanding, whether local demand has changed, or whether the borrower is struggling before a formal delinquency appears. The report does not claim that every branch employee performs sophisticated monitoring. It does show that banks see on-site visits, staff continuity, community participation, referrals, and direct relationship-building as central practices rather than obsolete rituals.
The survey’s broader market results make the same point from another direction. About 80% of banks define their small-business lending market by their branch footprint. Among all banks, 36% report that their small-business borrowers are generally within 20 miles of a branch, and two-thirds place them within 60 miles. For banks that define their market by branches, the average reported distance is about 40 miles. These figures describe the center of bank lending, not an absolute boundary. Some banks operate without a branch-based market, and some lend hundreds or thousands of miles away. But the dominant pattern is that bank relationships form within a relatively close operating area.
The reasons banks give for staying within that area reveal what “local” is doing. The most common reason is that borrowers outside the bank’s market prefer to borrow from a closer bank. That is a borrower-side constraint: a bank cannot win every customer simply by making an application technically available. Banks also cite difficulty developing a deposit relationship, finding referrals, visiting distant borrowers, monitoring loans, understanding local economic conditions, and verifying applicant information. These are competitive, informational, and operational constraints at once.
A digital connection addresses only some of them. A video call can reduce the cost of a conversation, but it does not automatically create local market knowledge or a trusted referral. A remote document upload can reduce travel, but it does not prove that the records are complete or that the business’s explanation is credible. A centralized monitoring dashboard may flag an anomaly, but staff still need to decide whether it reflects fraud, a temporary disruption, or a reasonable business choice. Proximity is valuable partly because it bundles several capabilities together.
Bank size changes the way those capabilities are assembled. In the FDIC report, “small banks” have less than $10 billion in assets and “large banks” have at least $10 billion. For small loans, small banks use more soft information across every measured category. A loan officer’s assessment is used for most or all small loans by 81% of small banks, compared with 44% of large banks. Large banks, by contrast, are much more likely to emphasize business credit scores: 62% use them for most or all small loans, compared with 18% of small banks. Small banks also evaluate more information overall, while large banks’ advantage lies in standardization and scale rather than in considering no information at all.
Automation follows the same pattern. Only 9% of small banks use a credit-scoring model for small-business lending, versus 46% of large banks. Among large banks, 28% can automatically approve a small loan of $25,000, but only 11% can automatically approve a $250,000 loan and 1% can automatically approve a $1 million loan. Automated underwriting is therefore concentrated at the smaller end of the loan range, where standardized evidence is easier to use and the bank can process a high volume of similar cases.
Speed is not a simple proxy for which bank is better. Slightly more than half of large banks can approve a small and simple loan in one business day or less, compared with 29% of small banks. But 76% of small banks and 81% of large banks can approve such a loan within five business days. The difference at the one-day margin may reflect large banks’ greater use of automated lending and credit-scoring information. It may also reflect the kinds of applications each bank receives. A system can be faster for the cases it recognizes while being less flexible with cases that do not fit its template.
This is the central reason not to collapse relationship lending and customer interaction into one scale. The FDIC finds that nearly all banks use high-touch practices to generate and maintain small-business relationships. Customer referrals and walk-ins are each used by 99% of banks. Long-term consistency in staff assignments is used by 97%, and asking staff to actively develop relationships is used by 94%. These practices remain common even among banks that do not typically meet applicants during approval.
In other words, a bank can separate the place where it builds a relationship from the way it evaluates a particular loan. A large bank may use a standardized credit model for a small application and still want staff to maintain the customer relationship. A small bank may know a borrower well but still require financial statements and collateral for a larger exposure. Underwriting style and customer-facing style interact, but one does not mechanically determine the other.
The borrower’s perspective reinforces the distinction. Banks say that small-business customers highly value customer service and speed of service more than any other surveyed areas: 97% and 93% of banks, respectively, identify them as highly important. Lending flexibility and pricing are also important, at 82% each, followed by convenience at 79%. Customer-facing technology matters, but only 49% of banks classify it as highly important. These are banks’ perceptions rather than a revealed-preference experiment, yet they explain why a local relationship can remain competitive even when an online lender offers a faster interface.
The local model is therefore not simply a refusal to modernize. It is a way of combining information production, risk assessment, monitoring, and customer value. It has costs: branches require staff and physical investment, local coverage can limit scale, and relationships do not guarantee accurate judgment. Consolidation and branch closures can weaken the network through which soft information is gathered. But replacing the branch would require replacing all of the functions that currently travel through it, not just the paper form.
That is the baseline against which AI must be judged. If AI makes hard information cheaper to collect and more useful to interpret, it may lower the informational cost of serving a distant borrower. It does not follow that it lowers the cost of trust, local knowledge, monitoring, or flexible judgment by the same amount. The next question is therefore not whether banks use technology, but where technology enters the lending process and what banks still keep human.