Incumbents DBS, OCBC, and UOB possess decades of proprietary data that digital banks cannot legally replicate, creating an unassailable advantage in credit modeling.

Singapore's longest-established banks maintain customer transaction histories spanning up to 40 years—eight times the regulatory minimum of five years—creating a proprietary data moat that underpins their credit models and structurally excludes digital newcomers from competing on equal terms. This gap, rooted in decades of institutional continuity and reinforced by data retention and deletion rules, means that even the most well-capitalized digital bank cannot simply purchase or accumulate its way to comparable credit assessment capabilities without regulatory intervention.

Regulatory Foundation: Five-Year Minimum vs. Business Retention

Monetary Authority of Singapore (MAS) Notice FSM-N02 sets the baseline: prescribed financial institutions must retain records relating to business relations, transactions, and disclosures for "a period of at least 5 years" from initiation, receipt, or termination of the relevant activity (Source 4, para 9.3). The same notice permits banks to retain data longer, "subject to any other requirements imposed by law" (Source 4, para 9.3). However, upon termination of services, MAS requires banks to delete or destroy records "as soon as possible" (Source 1), creating a hard stop for transaction histories tied to former customers.

Bank of Singapore's data protection policy illustrates how incumbents leverage this latitude: the bank explicitly reserves the right to use "transaction pattern and behaviour, financial background and demographic data" for direct marketing (Source 2). It further retains records beyond account termination for "archival management" and "legal proceedings" (Source 2), indicating that commercial incentives—not regulatory fiat—are the primary driver of extended retention. The gap between a 5-year minimum and a 40-year maximum is thus a deliberate strategic choice, not a compliance requirement.

Years (Years)Source: Orionmano Industries

Incumbent Banks' Historical Data Depth: Decades of Transaction Records

The three dominant incumbents—DBS, OCBC, and UOB—have accumulated transaction data through organic growth, mergers, and acquisitions spanning generations. OCBC has operated continuously since 1932, giving it an institutional presence of over 90 years (Source 5). This longevity directly translates into deep transaction archives: industry reporting indicates that Singapore's longest-established banks store customer transaction histories for up to 40 years (Source 1), meaning records can span from the mid-1980s to the present.

Mergers and acquisitions have further concentrated historical data. DBS acquired the Post Office Savings Bank (POSB) in 1998 (Source 6), inheriting POSB's decades of deposit records from a bank that historically served the mass retail market. In 1999, Keppel Bank and Tat Lee Bank merged (Source 6), consolidating their customer bases and transaction archives; OCBC subsequently acquired Keppel TatLee Bank in 2001 (Source 5). Each consolidation deepened and broadened the pooled transaction histories available to the surviving entity, creating data sets that cannot be replicated by any institution founded in the 21st century.

Proprietary Credit Scoring: Leveraging Long Data Histories

The strategic value of these archives lies in credit modeling. Industry analysis explicitly links long transaction histories to "proprietary credit scoring models" that incumbents have developed (Source 1). These models draw on decades of customer transaction patterns, repayment behaviour, and financial background data—the same categories that Bank of Singapore's policy identifies for marketing and credit assessment (Source 2).

Longitudinal transaction data enables incumbents to train models on behaviour across multiple economic cycles, including the 1997 Asian Financial Crisis, the 2008 Global Financial Crisis, and the COVID-19 pandemic. A customer's spending, saving, and repayment patterns over 20 or 30 years provide a richer signal of creditworthiness than the two-to-five-year window available to a digital bank. This allows incumbents to price risk more accurately, approve credit with greater confidence, and detect early signs of distress that short-window models might miss. The ability to retain records "for archival management" (Source 2) after account closure further allows incumbents to maintain reference data sets for model validation and stress testing.

Competitive Barrier: Why New Entrants Cannot Replicate Incumbent Models

Digital banks operating in Singapore face a structural barrier: they lack comparable historical data, and they cannot legally accumulate it through time. The ecosystem-based digital banks licensed in Singapore—GXS (a Grab-Singtel joint venture) and SeaBank (the digital arm of Shopee parent Sea Limited)—were founded after 2020 (Source 7). Their transaction histories span at most five to six years, and even that data is fragmented across customer relationships that may terminate under MAS deletion rules (Source 1).

The AI summary of industry research states plainly that new entrants "cannot replicate" incumbents' proprietary models without comparable data (Source 1). This is not merely a matter of time: MAS's requirement to delete records upon termination of services (Source 1) prevents digital banks from retaining transaction histories of customers who close accounts or switch banks. Incumbents, by contrast, have held a stable, growing customer base for decades, and their retention practices keep historical records even for accounts no longer active.

The gap is self-reinforcing. Without deep credit data, digital banks face higher default rates or must be more conservative in lending, limiting revenue and customer acquisition. With higher risk costs, they find it harder to invest in data accumulation or attract customers with competitive rates. This creates an equilibrium in which incumbents' advantage compounds over time.

Outlook

The structural data advantage of DBS, OCBC, and UOB will persist unless regulators change the rules. No digital bank operating today can match 40-year transaction histories, and the five-year regulatory minimum ensures that new entrants cannot be forced to share data they do not have. Two potential catalysts could shift this dynamic: a data-sharing mandate requiring incumbents to make anonymized historical data available to competitors, or a relaxation of MAS's deletion rules to allow digital banks to retain terminated-account records. Neither appears imminent. In the absence of such regulatory change, the "big three" incumbents will maintain their dominance in credit risk assessment for at least another decade, with digital banks confined to niche segments where long-history credit models are less critical.