Why Rapid Loan Growth Can Look Better Before the Risk Arrives

A new loan earns before it proves

When a bank makes a loan, the asset begins accruing interest according to the contract, subject to accounting treatment.

The borrower may not encounter the first serious stress for months or years. A mortgage can perform through a stable employment period and fail only after a recession. A business loan can look healthy until a large customer disappears. A commercial property loan can remain current until refinancing is required.

origination → early interest income → seasoning → stress event → arrears or default → recovery → final credit outcome.

Seasoning is the time required for credit behaviour to become informative

A portfolio originated three months ago has had little opportunity to show long-term repayment quality.

Comparing its low arrears rate with a mature five-year portfolio can therefore be misleading.

Good analysis compares loans by origination vintage so each cohort can be observed at similar ages.

Vintage analysis separates age from quality

Suppose the 2024 mortgage vintage had 1 per cent serious arrears after twenty-four months. The 2026 vintage currently has only 0.2 per cent arrears—but it is only six months old.

The correct question is not “Which current arrears rate is lower?” It is “How did each vintage perform at six months, twelve months, twenty-four months and through comparable economic conditions?”

Fast growth can mathematically dilute the bad-loan ratio

A bank has S$100 million of loans and S$5 million of non-performing loans: a 5 per cent NPL ratio.

It originates S$100 million of new loans that are still current. NPLs remain S$5 million while total loans rise to S$200 million.

The reported NPL ratio falls to 2.5 per cent—even though the bank did not cure one old bad loan.

That arithmetic is not manipulation. It is why analysts should examine absolute problem assets and vintage performance as well as ratios.

Growth can improve earnings before provisions catch up

Expected credit-loss accounting is designed to recognise expected deterioration before final default, but estimates still depend on data, models and forward-looking assumptions.

If underwriting standards weakened in a way the model has not yet recognised, provisions can lag the true risk even when the accounting process is operating in good faith.

Read Expected Credit Loss and Loan-Loss Provisions.

Growth changes the denominator faster than the evidence

Revenue, total assets and customer numbers can expand immediately.

The evidence about lifetime default, recovery and fraud arrives slowly.

This creates an information asymmetry through time: management knows how much it originated long before it knows how good those originations ultimately were.

Rapid growth can stretch underwriters

A credit team designed to review 1,000 applications a month is suddenly asked to process 3,000.

Management can respond by adding people, technology and controls—or by allowing shortcuts, higher approval limits and weaker challenge.

The growth rate therefore creates operational pressure on the credit process itself.

Experienced staff can become the bottleneck

Hiring more analysts does not instantly create more experienced judgement.

New markets, industries and complex borrowers can require knowledge that develops slowly. A rapidly expanded team can have more headcount and less average experience.

Automation can scale both good and bad underwriting

An automated scoring system can process ten times more applications without proportional staffing growth.

If the model is well calibrated, that can improve efficiency and consistency. If the model has a hidden bias, data error or out-of-range assumption, automation scales the mistake at machine speed.

Current US interagency model-risk guidance issued in April 2026 reinforces a risk-based approach to model governance. The durable lesson is independent of jurisdiction: scale increases the consequence of model error.

Competition can weaken lending standards invisibly

A bank targets rapid market-share growth. Competitors offer lower rates and looser terms. To keep winning customers, the bank can gradually:

• accept higher LTV;
• require weaker covenants;
• reduce borrower equity;
• accept lower debt-service coverage;
• price risk more cheaply;
• approve longer tenures;
• rely more heavily on collateral;
• enter unfamiliar segments.

No single change may look dramatic. The portfolio can drift materially across thousands of loans.

Underwriting drift is visible only if the bank stores the original decision variables

The bank needs to compare new vintages on:

• credit scores or ratings;
• LTV;
• debt-service ratios;
• covenant strength;
• collateral type;
• loan tenor;
• pricing spread;
• exceptions and overrides;
• geography and sector;
• approval authority.

Without the origination coordinates, “credit quality remained stable” can become an unsupported impression.

Exception rates are an early-warning signal

A bank’s formal policy can remain unchanged while more loans are approved as exceptions.

Growth therefore should be analysed alongside:

• policy exception frequency;
• override rates;
• approval-level escalation;
• documentation defects;
• post-approval conditions outstanding.

Documentation quality can deteriorate before default rates do

Missing financial statements, incomplete collateral registration or weak verification may not cause a loan to miss its first payment.

They can make future monitoring and recovery much worse.

The credit file can therefore weaken before the borrower visibly weakens.

Collateral appreciation can hide weaker underwriting

During a property boom, rising collateral values reduce current LTV and make recoveries look strong.

The bank can then believe its underwriting is excellent when the real support came from a rising market.

When property prices stop rising, the hidden difference between borrower quality and collateral appreciation becomes visible.

Refinancing can postpone the loss signal

A weak borrower can remain current by refinancing or borrowing elsewhere.

The loan appears to perform, but the repayment source is new debt rather than the cash flow originally expected.

Read Rollover Risk.

Rapid growth can create concentration without looking concentrated at first

A bank grows in one fashionable sector because demand is high and losses are low.

Each loan may look acceptable individually. Together they can depend on the same property market, commodity price, employer base or refinancing channel.

Portfolio risk emerges from common drivers, not only borrower names.

Growth into a new geography creates information risk

A bank can enter a region where property law, borrower behaviour, data quality or economic cycles differ from its home market.

Historical models and policies may not transport cleanly.

Growth into a new product creates operational risk

A bank launches auto finance or buy-now-pay-later lending. The customer acquisition engine works immediately.

Collections, fraud controls, disputes, repossession or hardship processes may not yet have been tested at scale.

The credit product can grow faster than the operating system needed when borrowers stop paying.

Growth needs funding

Loans stay on the asset side after the deposits created at origination can move elsewhere.

The bank therefore needs stable deposits, wholesale funding, secured funding or another sustainable liability structure to carry the enlarged loan book.

Read Why Banks Still Need Funding Even When They Can Create Deposits.

Fast asset growth funded by unstable liabilities compounds risk

A bank expands long-term loans using confidence-sensitive wholesale funding.

If lenders become worried about the new loan book, funding can leave before loans mature.

Credit uncertainty becomes liquidity pressure.

Growth consumes capital

New loans generally add risk-weighted assets and expected loss.

If assets grow faster than retained earnings or new equity, capital ratios can decline.

Read Cost of Capital.

Growth can make capital look temporarily efficient

New loans add interest income quickly while credit losses remain low because the portfolio is young.

Return on equity can therefore improve early.

If later losses rise, the same growth can damage capital sharply.

Growth targets can distort incentives

If management compensation depends heavily on loan volume, originations or market share, staff can be rewarded for creating assets before the losses associated with those assets have matured.

Strong governance therefore balances growth goals with credit quality, risk-adjusted return and later vintage performance.

Clawbacks and deferred compensation can reconnect incentives to later outcomes

If all rewards are paid at origination, the person making the decision can leave before the loan’s true quality becomes visible.

Deferred compensation and clawback mechanisms can align part of the reward with later performance, subject to the institution’s framework and applicable rules.

The bank should ask whether control capacity grew with the book

The OCC’s current lending and problem-bank supervision materials emphasise this question directly: did risk-management systems, management expertise and resources keep pace with growth?

Useful capacity measures include:

• underwriter caseload;
• collections staffing;
• model validation;
• fraud-control capacity;
• collateral-review backlog;
• audit coverage;
• exception queues;
• funding and liquidity capacity;
• capital headroom.

Loan growth should be compared with capital growth

The OCC specifically lists asset growth compared with capital growth as one useful measure when assessing bank growth.

A bank that doubles its loan book while capital barely moves has changed its leverage and loss-absorption profile even before one new borrower defaults.

Loan growth should be compared with deposit and funding growth

A bank can grow loans faster than core deposits and fill the gap with wholesale or brokered funding.

That can be legitimate. It also changes funding stability and liquidity sensitivity.

Loan growth should be compared with economic growth

If one bank grows 40 per cent in a market growing 5 per cent, it is winning share rapidly.

Possible explanations include superior distribution, acquisitions, new products or substantially looser underwriting.

The growth difference generates a question; evidence determines the answer.

Margins can reveal whether the bank is buying growth

A bank reports rapid volume growth while loan spreads fall significantly.

That can reflect lower market rates or improved funding. It can also mean the bank is pricing risk too cheaply to win customers.

Growth should therefore be read alongside margin and underwriting quality.

Early arrears are more informative than waiting for final default

First-payment default, thirty-day delinquency, repeated overdraft, covenant breach and restructuring can reveal stress before a loan becomes non-performing.

Leading indicators matter particularly when the portfolio is young.

Fraud can emerge early in high-growth portfolios

When approval speed becomes a competitive advantage, identity and document controls can be attacked.

Higher early fraud or misrepresentation rates can be a sign the acquisition system scaled faster than verification.

Collections behaviour can expose underwriting drift

A bank can maintain low reported NPLs while call-centre contacts, hardship requests and payment arrangements rise sharply.

Those softer signals can reveal borrower stress before formal default classification changes.

Stress testing should apply to the new book, not only the old book

A bank can stress-test its historical portfolio using assumptions from borrowers underwritten years ago.

If the new growth cohort has higher LTV, longer tenure or weaker covenants, historical loss behaviour may understate its vulnerability.

Rapid growth can be good when controls scale first

A bank can grow safely when:

• credit standards remain clear;
• exception rates stay controlled;
• underwriter capacity expands;
• models are validated for the new population;
• funding remains durable;
• capital grows with risk;
• portfolio concentration stays within appetite;
• servicing and collections can handle stress;
• management monitors vintage performance;
• pricing still covers risk.

Growth is therefore an outcome to govern, not a condition to fear.

The OCC’s 2026 risk perspective shows the other side of the story

The OCC reported in May 2026 that bank earnings improved in 2025 with support from loan growth and lower funding costs, while balance sheets remained strong and overall credit risk manageable in the federal banking system it supervises.

That current example is important because it prevents a false rule: loan growth does not imply bad lending. The warning is about growth combined with weak controls, concentration, funding strain or unseasoned risk.

A worked denominator example

A bank has S$5 million of NPLs on S$100 million of loans: 5 per cent.

It originates S$100 million of new current loans. NPLs stay at S$5 million. The ratio falls to 2.5 per cent.

No asset-quality repair occurred. The denominator doubled.

A worked capital example

A bank grows risk-weighted assets by 25 per cent and common equity by only 5 per cent.

Unless the initial buffer was large or the risk mix changed materially, its CET1 ratio will come under pressure.

A worked vintage example

The 2025 unsecured-loan vintage shows 2 per cent serious arrears after twelve months.

The 2026 vintage has only 0.4 per cent current arrears at four months.

The bank should not declare the 2026 vintage five times better. It needs to observe the cohort at comparable seasoning points.

A worked underwriting-drift example

Mortgage originations rise 40 per cent. Average LTV rises from 60 to 72 per cent, debt-service ratios rise, and policy exceptions double.

Current arrears remain low because the loans are new.

The deterioration is already visible in the origination coordinates even before arrears arrive.

A worked funding example

A bank grows loans by S$10 billion while stable customer deposits grow by only S$2 billion.

The remaining funding comes from shorter-term wholesale markets.

Credit growth has changed the liability side of the balance sheet as much as the asset side.

The durability test: wait for the future to answer the origination claim

Credit models, products and reporting rules will change. The enduring growth questions remain:

• How fast are loans growing relative to capital?
• How fast are they growing relative to stable funding?
• Are underwriting standards changing?
• Are policy exceptions rising?
• How mature are the new vintages?
• What leading stress signals exist before NPL classification?
• Are concentrations increasing?
• Can operations and collections handle the larger book?
• Does pricing still cover risk?
• What happens when the benign economic environment ends?

The World Return: growth is validated only when the loans come back

A loan origination is a claim on future household or business cash flow. The bank has not proved the claim when it books the asset.

new loan → early interest → borrower use → real economic performance → repayment or stress → recovery or loss → revised underwriting.

Rapid growth passes the test only when later vintages show that the bank did not buy short-term earnings by selling away future asset quality.

Twelve misconceptions to remove

Misconception	Better model
“Rapid loan growth is always bad.”	Growth can be healthy when underwriting, capital, funding and controls scale with it.
“Low current arrears prove new loans are high quality.”	Young portfolios may simply be unseasoned.
“A falling NPL ratio always means bad loans were cured.”	Rapid new lending can dilute the ratio’s denominator.
“Interest income proves the loan is profitable.”	Lifetime credit loss and capital cost arrive across the full loan horizon.
“Policy is unchanged, so standards are unchanged.”	Exception and override rates can loosen effective underwriting.
“Automation solves capacity constraints.”	It can scale model error and weak data as quickly as good decisions.
“Rising collateral values prove underwriting quality.”	A strong market can hide weak borrower fundamentals.
“Refinancing proves the original repayment source worked.”	New debt can postpone recognition of weak operating cash flow.
“Loan growth affects only assets.”	Funding, liquidity and capital must expand to carry the assets.
“More market share means stronger franchise quality.”	Share can be purchased through underpricing and looser terms.
“Credit losses begin when NPLs are reported.”	Underwriting drift, early arrears and hardship can appear earlier.
“Growth is complete at origination.”	Credit quality is validated only as borrowers repay through real economic conditions.

Observable mastery

1. Why does a new loan earn before its full risk is known?
2. What is seasoning?
3. Why does vintage analysis improve comparison?
4. How can rapid growth reduce the NPL ratio without curing old problem loans?
5. How can effective underwriting weaken while written policy stays unchanged?
6. Why do funding and capital have to scale with loan growth?
7. What leading indicators can appear before formal default?
8. How can collateral appreciation hide weak credit quality?
9. Why can automation amplify growth risk?
10. When is loan growth healthy?
11. Why does current OCC evidence prevent a simplistic “growth is bad” conclusion?
12. What event finally validates the origination decision?

If those answers connect, rapid loan growth becomes visible as a timing problem as much as a volume problem: the income arrives early, the evidence arrives slowly, and the bank earns the right to call the growth good only after the loans survive the future they were written against.

---