The Real Business of Banking: Deciding Who Not to Lend To

Every bank has one non-delegable function: to govern what risk does not come onto its balance sheet.

The industry tends to locate the problem at the wrong moment. It talks about arrears, collections, provisions, and write-offs as if deterioration began at default. But by then, the decision that mattered had already been made. Capital was committed, liquidity deployed, management time absorbed, and the bank's optionality reduced.

Default doesn't create the error. It only shows up afterward to put a date on it.

The critical decision happens at underwriting. That's where the bank determines whether a deal justifies consuming scarce capital — or whether it merely found a convenient way to look defensible.

The question that shapes a book isn't only whom to lend to, but whom to decline: the borrower whose capacity to pay doesn't support the risk, whose willingness to pay hasn't been tested with enough rigor, or whose request needs price or collateral to stand in for judgment.

Volume Dressed Up as Strategy

The dominant assumption in banking is as comfortable as it is dangerous: that book growth is a sign of strength.

Under that premise, the organization learns to read volume as evidence of success. The front line celebrates origination. Committees review approvals. Reports show disbursements, market share, and turnaround time. A decline, by contrast, is treated as administrative residue: a deal that stalled, a target missed, business a competitor took instead.

That logic distorts the executive conversation. What matters stops being whether the deal creates real economic value and becomes whether it can be approved without breaking a visible rule. Structures get more “creative,” collateral more central, pricing more flexible, and the exception more routine.

The language shifts too. The risk function becomes a filter. Deals “clear” risk. Deals “fall through.” The committee “resolves.” The entire vocabulary implies that approval is a request's natural destiny and a decline is an interruption.

But growth is not an automatic virtue. It can be real economic expansion or deferred deterioration. The difference is set at underwriting.

A bank that doesn't govern its ability to say no ends up handing its risk profile to commercial pressure, to the mood of the cycle, and to the uneven discipline of its own officers. The organization believes it is building a book. In reality, it is hardening a standard. And sooner or later, the entire balance sheet comes to resemble the standard that senior management was willing to tolerate.

When Price and Collateral Select on Your Behalf

The deepest analytical mistake is assuming that the risk taken at origination can be corrected later.

If the borrower is riskier, charge more. If the information is incomplete, demand more collateral. If the deal is uncomfortable, structure around it. It looks like prudence, but economics warns otherwise: in credit markets with imperfect information, raising the price or requiring more collateral does not necessarily improve selection. It can degrade it.

Stiglitz and Weiss showed in 1981 that raising the rate to compensate for higher risk can change the composition of who accepts the credit. The stronger borrowers may walk away, while those with a higher probability of default remain in proportionally greater numbers. Price stops being mere compensation. It becomes a selection mechanism.¹

Collateral suffers a similar illusion. The same authors showed that demanding more security is no neutral screen either: it can push out sound borrowers with thinner net worth and retain the riskier projects. High collateral does not turn a weak credit thesis into a strong one. Sometimes it only lets a committee approve, with greater peace of mind, a deal that should never have come in.

Here lies the central failure: the bank believes it is managing risk when it is merely justifying admission.

There is also a technical bias. Capacity to pay is modeled with cash flows, coverage ratios, and stress scenarios. Willingness to pay demands less comfortable signals: track record, transparency of information, ownership structure, conduct with other creditors, and consistency between what the borrower declares and what he does. When capacity crowds out willingness, the analysis becomes incomplete even when it looks technically flawless.

Underwriting Is Not a Control

The cost of approving badly does not begin with the accounting provision. It begins with the inefficient occupation of the balance sheet.

A poorly originated deal consumes regulatory capital, liquidity, management attention, monitoring capacity, legal time, workout effort, and risk headroom that could have gone to a better relationship. Part of that cost doesn't show up when the deal is booked. It surfaces later — scattered across renewals, exceptions, partial impairments, and decisions that hijack the executive agenda.

That is why underwriting is not a control applied after the business. It is the point at which the bank decides what kind of institution it will be.

The Basel Committee revised its Principles for the Management of Credit Risk in 2025. The revision reinforces an idea that commercial practice tends to treat as a formality: a bank's quality depends on a sound granting process, well-defined criteria, effective due diligence on the borrower, an explicit risk-return relationship, and remuneration policies that do not incentivize excessive risk-taking.²

That is not supervisory red tape. It is a corporate-governance warning.

As a hypothetical illustration, two officers can coexist for years: one declines 40% of what he reviews; the other, barely 5%. If no one measures or explains that gap, the book stops reflecting an institutional policy and starts reflecting individual preferences. The bank's risk profile is no longer being governed. It is emerging on its own.

No deal should be approved whose primary defense is the collateral, nor one in which willingness to pay has not been assessed with rigor comparable to that applied to capacity.

Collateral is a secondary source of recovery, not a credit thesis. Approving a deal because the collateral makes it defensible is not credit judgment. It is substituting the analysis with the security. And when an institution swaps judgment for collateral, it stops allocating capital and starts accumulating problems.

Governing the No

The divergent decision that preserves value requires treating a decline as an act of corporate governance, not a commercial failure. That demands three breaks with convention.

1. Audit the decline. The bank must measure not only what it originates, but what it declines, why it declines it, who declines it, and what subsequently happens to deals approved by exception versus those that met the full standard.
2. Align incentives to the cycle. Compensating solely on gross disbursement rewards a promise before anyone knows whether it created value. A book should be judged once risk has had time to reveal its true quality — not on the day it is signed. Volume can produce revenue. Only seasoned performance demonstrates RAROC, capital preservation, and risk-adjusted profitability.
3. Elevate the risk function. Risk does not exist to slow the business down. It exists as a steward of capital, protecting the bank's ability to keep making good decisions when the cycle turns. The risk function isn't there to block deals. It's there to keep the pressure to do deals from quietly redefining the institution's standard.

Internal resistance is predictable. It will be said that the bank lost agility, that it turned conservative, that it is ceding ground to more aggressive competitors. Some of that will be true in the short term. Some declined deals will be taken up by others, and some will work out. Underwriting discipline always carries a visible cost before it shows its benefit.

But senior management has to hold that discomfort. It cannot demand book quality and then penalize those who exercise the very judgment that produces it. The tension with the commercial front line is not a flaw in the system. It is one of its most important design features.

The industry tends to applaud the books that grow fastest. The cycle tends to admire others: the ones with the maturity to let pass the business everyone else wanted.

An exceptional bank is not defined by the volume of credit it put on. It is defined by the risk it had the discipline not to bring onto its balance sheet.