Page 503 - ISC PROCEEDINGS 21.4

four bank clusters defined by IT expenditure intensity and business model orientation; (ii)
content analysis of annual reports and secondary data sources to classify banks and
extract investment indicators; and (iii) documentary analysis of the regulatory and policy
framework. The dataset covers 27 joint-stock commercial banks listed on HOSE and HNX
over 2018–2024 (189 bank-year observations), drawn from: audited annual reports and
FiinGroup database (financial indicators); Vietnam ICT Index (MIC, 2018–2024) and SBV
reports (digital infrastructure indicators). Data construction followed a systematic three-
stage procedure. In Stage 1 (Source Collection), annual reports for all 27 banks were
retrieved from official bank websites and the FiinGroup database for each year from 2018
to 2024; SBV and MIC official statistical publications were collected for sector-level
indicators. In Stage 2 (Variable Extraction), IT expenditure was extracted from the notes
to financial statements under “technology and digital investment” line items,
standardised as a percentage of total operating costs to enable cross-bank comparability.
Financial performance variables - Return on Assets (ROA), Return on Equity (ROE), Cost-
to-Income Ratio (CIR), and Non-Performing Loan ratio (NPL) - were taken directly from
audited consolidated financial statements. Digital transaction share was sourced from
SBV’s annual payment activity reports cross-validated with bank-disclosed figures. IT
workforce ratios were drawn from MIC’s Vietnam ICT Index and supplemented by publicly
available industry labour market surveys. In Stage 3 (Data Validation), all variables were
cross-checked across at least two independent sources where available; observations
with missing IT expenditure data for more than two consecutive years were flagged and
excluded from cluster assignment. It is acknowledged that IT expenditure disclosure
formats are not yet standardised across Vietnamese banks, and that some figures
represent estimates derived from technology-related cost line items rather than explicit
“IT budget” disclosures. This constitutes a recognised limitation of the dataset and is
discussed further in Section 4.
Given the objective of mapping sector-wide digital infrastructure development
under heterogeneous data disclosure conditions, a descriptive panel approach is
considered methodologically appropriate. The absence of standardised IT investment
reporting across banks limits the feasibility of constructing consistent continuous
variables required for causal econometric modelling. Therefore, the study prioritises
cross-cluster comparative analysis to identify structural patterns and stylised facts,
providing a foundation for future causal research.
Bank cluster classification follows a two-step procedure grounded entirely in
publicly available secondary data. First, each bank is assigned to one of four clusters
based on business model orientation and IT expenditure intensity as reported in audited
annual reports and the Vietnam ICT Index: (C1) state-owned banks (BID, CTG, VCB, AGR),
characterised by large branch networks and operational safety priorities; (C2) enterprise-
oriented banks (TCB, MBB, VPB, MSB and peers), characterised by above-average IT
expenditure intensity (≥18% of operating costs) and AI/data-platform investment; (C3)
retail-oriented banks (ACB, STB, VIB, TPB, HDBank and peers), characterised by mobile-
first product strategies; and (C4) small and medium-sized banks, characterised by limited
IT budgets and outsourced infrastructure. Second, financial performance indicators (ROA,
ROE, CIR, NPL ratio, digital transaction share) are compared across clusters both cross-
sectionally (2024) and longitudinally (2018–2024), enabling evidence-based descriptive
associations between IT investment profile and financial outcomes. The four-cluster
typology is grounded in two theoretical and empirical rationales. Theoretically, the

502