Skip to main content
F I N T R A D E

Loading

Back to Commentaries

When A 25-Cent Error Becomes A$1.73 Million Accountability Failure - By Riaz Patel, FSCL Director

When A 25-Cent Error Becomes A$1.73 Million Accountability Failure

There is a tendency in financial services to measure the seriousness of a problem by the amount involved in an individual transaction. That approach can be dangerously misleading. A fee of 25 cents may appear insignificant in isolation, but the New Zealand Financial Markets Authority’s (FMA) public warning to TSB Bank Limited on 21 September 2026 demonstrates how a small charge can expose much larger questions about institutional controls, customer fairness and management accountability.

Between 30 November 2017 and 20 November 2023, TSB charged a 25-cent fee on certain business cheque transactions that, under its own terms and conditions, should not have attracted the charge. The error affected 7,309 customer accounts, and the bank’s total remediation amounted to NZ$1,734,800.

For me, the central issue is the gap between what a financial institution tells customers and what its systems actually do. Customers are entitled to expect that charges applied to their accounts reflect published terms. They cannot reasonably be expected to audit transaction-processing systems, inspect software rules or reconstruct individual charges. Where the institution controls the information, systems and calculations, it must also accept responsibility for ensuring that every charge is contractually justified.

The TSB error followed a change to the wording of its Business Cheque terms and conditions in November 2017. Although the revision was intended to clarify fees, certain transactions were omitted from the revised wording while the bank’s systems continued charging for them. The result was a prolonged mismatch between the information provided to customers and the fees actually collected.

The public record does not establish that TSB deliberately intended to overcharge customers, and I would not suggest otherwise. The documented failure is serious enough without attributing motives that the available evidence does not establish. An absence of evidence of deliberate conduct, however, does not make a systems failure lasting almost six years inconsequential.

Financial institutions increasingly depend on interconnected technology platforms. Customer terms, product configuration, pricing engines, transaction-processing systems and accounting records may operate across different platforms, often modified incrementally over several years. A change to one component does not necessarily update every other component. Without proper testing and reconciliation, systems can continue functioning technically while producing outcomes inconsistent with contractual commitments.

I believe financial institutions should therefore treat changes to customer terms as operational control events, not merely legal or communications exercises. Whenever fees, exemptions, interest calculations or eligibility conditions change, institutions should identify the affected transactions and verify that corresponding systems have been updated. Testing must cover transactions that attract charges as well as those that should be exempt.

A successful software deployment does not necessarily produce a successful customer outcome. A system may process millions of transactions without interruption while consistently applying the wrong charging rule. Conventional technology monitoring establishes whether a platform is available and processing transactions; it does not necessarily establish whether those transactions comply with contractual obligations.

This distinction separates operational performance from conduct performance. The former examines whether systems function; the latter examines whether they treat customers in accordance with applicable terms, representations and requirements. Both require attention, but conduct performance demands coordination between legal, compliance, technology, product design and operational risk functions.

Legacy technology presents an additional challenge. Established financial businesses often operate systems that have evolved through successive modifications. Such technology is not inherently defective, nor is wholesale replacement always practical. The essential requirement is that institutions understand system dependencies, maintain clear ownership and establish controls capable of detecting when changes fail to produce their intended outcomes.

The value of an individual transaction is only one measure of financial impact. Frequency, duration, the number of customers affected and the difficulty of detecting an error are equally important. In the TSB case, the FMA noted that customers had no practical way of checking the relevant fees themselves. Reviewing a bank statement is not equivalent to having the information necessary to establish whether every charge complies with the applicable terms.

The cumulative impact is evident in the remediation figures. TSB refunded NZ$1,425,127 in fees, paid NZ$10,404 in related interest and fees, and provided NZ$299,269 in compensation for the use of customers’ money. Together, these payments amounted to NZ$1,734,800.

Small charges may also escape ordinary customer scrutiny. A substantial unexpected deduction is more likely to attract immediate attention than a 25-cent fee appearing among numerous routine transactions. This makes internal monitoring particularly important where customers cannot readily identify discrepancies.

Materiality should therefore be assessed from the perspective of the affected customer population and the institution’s obligations, not solely by the value of an individual transaction. A modest charge remains incorrect when the applicable terms do not authorise it. Repetition across thousands of accounts can turn an apparently insignificant error into a substantial institutional and financial liability.

The chronology disclosed by the FMA deserves attention. TSB identified the issue internally in August 2023 and stopped charging the fees on 20 November 2023. It notified the FMA verbally in December 2024 and in writing in February 2025. The regulator issued its public warning on 21 September 2026.

The FMA acknowledged that TSB had identified and reported the issue itself, cooperated constructively and had largely completed reimbursing customers. It nevertheless cited the duration of the overcharging, the number of affected accounts and delays in reporting the issue and refunding customers as reasons for issuing the warning.

I would not dismiss the importance of internal detection and self-reporting. Identifying a problem and stopping the incorrect charges demonstrate important elements of institutional control. However, detection is the beginning of the response, not its conclusion. Once an institution discovers that customers may have been charged incorrectly, it must establish the full scope, prevent recurrence, identify affected customers, determine appropriate remediation and assess its regulatory notification obligations promptly.

Delays in notifying regulators can matter because supervisory authorities may need information to assess risks, determine whether intervention is necessary and establish whether the problem extends beyond its initially identified scope. The precise notification obligations depend on the jurisdiction and applicable rules. The broader governance requirement is clear: institutions need defined escalation criteria, responsibility for regulatory notifications and documented decisions explaining the action taken.

Customer remediation requires similar discipline. Stopping an incorrect charge prevents further losses but does not restore amounts already collected. Institutions should use reliable transaction records to identify affected customers rather than relying exclusively on complaints. Refunds must be accurate and timely, with clear explanations. Where circumstances justify additional compensation, returning the original fee may not be sufficient.

TSB’s compensation for the use of customers’ money recognises that the consequences of an incorrect charge extend beyond the original amount. The period during which customers were deprived of those funds is also relevant to remediation.

The FMA’s action should be described accurately. It issued a public warning under its statutory powers; the warning is neither a court finding nor a penalty. That qualification deserves respect, but it does not diminish the value of examining the underlying facts as a governance lesson for the wider financial services industry.

The risk identified is not unique to cheque accounts or to one institution. Any financial business automating fees, commissions, interest calculations, penalties or other contractual charges must ensure that its systems accurately reflect the rules communicated to customers.

For institutions operating across multiple jurisdictions, the challenge can be more complex. Different products, customer categories and regulatory regimes may carry distinct contractual and operational requirements. A group-wide policy does not, by itself, establish that every local implementation is correct. Institutions need clear ownership, jurisdiction-specific mapping where necessary and evidence that actual systems reflect applicable commitments.

The appropriate response to the TSB case is to examine whether relevant controls exist, whether they have been tested and whether there is evidence that they work. Financial institutions should review how contractual changes translate into technology updates, how exemptions are tested, how customer outcomes are monitored and how errors are escalated. Internal audit, compliance and operational risk functions should be able to examine the relationship between published terms and actual charges.

I believe customer fairness is strongest when embedded in routine business processes. It cannot depend entirely on customers noticing errors, raising complaints and persisting until an institution investigates. Nor can it depend solely on periodic regulatory scrutiny. The primary responsibility rests with the institution that designs the product, defines the fee and collects the money.

Changes to customer terms should have an identified business owner and a documented implementation process. Technology changes must be tested against the contractual requirements they are intended to implement. Institutions should retain evidence of testing, monitor transactions for anomalies and establish clear escalation procedures. When errors emerge, the response must cover containment, investigation, customer identification, remediation, appropriate regulatory notification and examination of the underlying cause.

Senior management should be able to answer straightforward questions. Who ensures that published fee schedules match system configurations? How are changes independently checked? What evidence demonstrates that exempt transactions are actually exempt? How quickly can every affected customer be identified? What triggers escalation to compliance, senior management and the regulator?

These are practical governance questions. Their answers should be supported by records rather than assurances.

There is also a cultural dimension. Where teams are rewarded primarily for completing product launches, implementing changes quickly or reducing operating costs, control testing can become an administrative afterthought. A product is not ready simply because software has been deployed. The institution must have reasonable evidence that it behaves as promised and that customers can obtain redress when it does not.

The TSB case offers a clear lesson for financial services. Customers should be charged only what applicable terms permit. Financial institutions must demonstrate that their systems implement those terms correctly, that changes are tested and monitored, and that errors are addressed promptly and comprehensively.

The NZ$1.73 million remediation figure is striking, but the more important question is what allowed an incorrect fee to continue for almost six years following a change in the wording of the bank’s terms. The publicly reported facts establish a prolonged mismatch between contractual terms and the charging system, followed by internal identification, cessation of the charge, remediation and regulatory warning. They also establish that the FMA recognised TSB’s self-reporting and cooperation while expressing concern about the duration of the problem and delays in reporting and refunding customers.

For me, the standard of good financial conduct is ultimately measured in ordinary transactions. Customers rarely see an institution’s internal control framework, technology architecture or compliance reports. They experience the result when a fee is deducted, interest is calculated or a contractual promise is put into practice. Trust depends on those outcomes being accurate and on mistakes being identified and corrected without placing an unreasonable burden on customers.

A 25-cent fee can be small in value and significant in principle. The lesson is not that every technical error signals misconduct, or that every discrepancy should be judged without regard to its circumstances. It is that financial institutions must take ownership of the systems through which they fulfil their commitments. Accountability begins long before a regulator issues a warning. It begins when an institution makes a promise to a customer and ensures that its own systems keep it.