Traditional Cost-Cutting Peaking, AI Becomes New Engine for Mid-Term Efficiency in Asian Banks

Overview

This report deeply analyzes the structural revenue pressures faced by Asian banking over the past two decades and the effectiveness of cost control, exploring the potential of artificial intelligence (AI) as the next-generation efficiency driver. The report points out that affected by population aging, Asian banks' net interest income and fee income both face long-term downward pressure. Although banks significantly reduced operating costs through consolidation, branch optimization, and technology investment, these measures only partially offset revenue declines. As traditional cost-cutting leverage matures, AI is seen as the most important structural efficiency improvement tool in the medium term, but its short-term financial contribution remains insignificant, and the market has not yet fully priced in the differences in banks' AI readiness.

Core views

Revenue side faces structural headwinds. Over the past 20 years, the average total revenue/assets ratio for Asian banks declined by 60 basis points, among which net interest income/assets ratio declined 45 bps, and fee income/assets ratio declined 15 bps. This trend is mainly driven by population aging, leading to slower savings accumulation, decelerated loan growth, and sustained pressure on net interest margins. Meanwhile, due to regulatory constraints and market structure factors, banks' efforts to transition to non-interest income did not fully make up for the loss of interest income. Cost control effects were significant but approaching limits. To cope with revenue pressure, Asian banks reduced the average operating expense/assets ratio by 40 bps over the past 20 years. However, this cost saving only offset about one-third of the revenue decline. Notably, emerging market banks fully offset their smaller revenue drop (25 bps) through larger cost cuts (operating expense/assets ratio down 45 bps), while developed market banks offset only about half of the revenue drop. Cost-income ratios (CIR) have converged between developed and emerging markets, both around 45-50%, breaking the stereotype that emerging markets have lower efficiency. Analysis of traditional cost sources. Past efficiency improvements mainly benefited from bank consolidations, branch rationalization, and personnel restructuring. Although branch numbers dropped significantly, many banks' total employee counts continued to rise due to shifts towards technology and back-office functions. Technology spending continued to rise (accounting for 4-5% of revenue) but did not bring significant step-wise cost savings at the system level, more so used to maintain competitiveness and operational stability. AI: From Narrative to Delivery Obligation. AI adoption is accelerating, but currently focuses mainly on isolated scenarios such as customer service, compliance monitoring, and internal productivity tools. Financially speaking, early AI applications more reflect capacity release (same staff handling more business) and cost avoidance (avoiding new hiring), rather than direct layoffs or explicit cost cuts. Additionally, saved resources are often reinvested into data infrastructure and network security. Therefore, short-term positive impacts of AI on income statements are limited and uneven. Medium-term Outlook: AI as the Largest Remaining Leverage. Given that operations and labor costs account for 60-70% of total bank costs, end-to-end AI integration is expected to achieve true structural cost cutting in the medium term. As banks pass the high-cost infrastructure construction period, 'digital labor' will expand with extremely low marginal costs, allowing revenue growth no longer to linearly depend on increases in manpower and physical infrastructure. The report believes AI readiness and execution capability will become key drivers of future bank valuation divergence, with the market shifting from focusing on AI narratives to focusing on quantifiable financial results.

Analysis framework

The report adopted a method combining long-cycle historical data analysis with structural driver breakdown. First, by reviewing financial data over the past 20 years (e.g., revenue/assets, operating expenses/assets, cost-income ratio, etc.), it quantified the severity of revenue headwinds and the hedging effect of cost control. Second, using volume-price split and structural decomposition logic, it broke down operating costs into labor costs and other operating expenses, comparing different driving factors between developed and emerging markets. Finally, combined with industry common sense (e.g., structural impact of population aging on financial demand) and technology adoption curve theory, it assessed the financial conversion path of AI from pilot to scaled application, emphasizing the time-lag effect between 'capacity release' and 'explicit cost reduction'.

Methodology notes

• Industry/Industrial Analysis FrameworkSupply-demand framework

  Structural impact of population aging on bank deposit and lending businesses

  The report uses a supply-demand framework to analyze demographic changes: aging leads to reduced savings supply and weakened loan demand, thereby structurally compressing banks' net interest margins and revenue base. This is not merely cyclical fluctuation, but a long-term trend.

• Company Fundamentals and Financial FrameworkOperation/Financial Leverage Analysis

  Cost-Income Ratio (CIR) and Operating Leverage

  By analyzing the cost-income ratio and its components (labor vs. other expenses), assess the bank's profit protection ability under revenue decline conditions. The report points out that although CIR converges, different markets reach balance through different paths (emerging markets rely on substantial cost cuts, developed markets rely on relatively stable revenue).

• Industry/Industrial Analysis FrameworkIndustry Chain Upstream-Midstream-Downstream Transmission

  Time Lag in Transmission from Technology Investment to Financial Results

  The report analyzes how technology investment (upstream) eventually reflects as cost savings on financial statements (downstream) by changing workflows (midstream). It notes that current AI is in the early stage, benefits mainly manifest as 'capacity release' and 'cost avoidance', rather than direct income statement improvement, exhibiting obvious transmission time lags.

• Event Dynamics and Behavioral FinanceExpectation Gap/Expectation Management

  Valuation of AI Readiness Not Priced In

  The report points out that although the potential impact of AI varies greatly across banks, the market does not distinguish this in valuation currently. As AI moves from concept to implementation, differences in execution capabilities will create huge expectation gaps, thereby driving valuation divergence.

Key data

• Change in Total Revenue/Assets Ratio Over Past 20 Years-60bpsAverage for Asian banks, indicating structural revenue pressure
• Change in Total Operating Expense/Assets Ratio Over Past 20 Years-40bpsCost savings only offset about 2/3 of the revenue decline
• 2025 Average Cost-Income Ratio (CIR) for Asian Banks~46%Developed and emerging markets tending to converge
• Proportion of Operating and Labor Costs60-70%Account for bank total cost base, main space for potential AI cost reduction
• Average Technology Spending Proportion for Major Asian Banks4-5%Percentage of revenue, Singapore, Australia, Taiwan as high as 6-10%
• Chinese Banks 2025 CIR~38%Significantly below emerging market average, benefiting from strong digitalization capabilities
• Vietnam Banks 2025 CIR~30%Mainly benefiting from strong revenue growth rather than simple cost control

Impact & implications

The report considers Asian banking currently at the critical node transitioning from 'traditional cost discipline' to 'AI-driven efficiency'. For investors, this means re-evaluating banks' value drivers: simply looking at past cost cuts is insufficient to predict the future; strategic deployment and execution capability of AI become crucial. Banks that successfully integrate AI into core operations and achieve end-to-end automation will gain significant cost advantages and valuation premiums in the medium term. Conversely, banks staying at the pilot stage or lacking data foundation may face risks of declining competitiveness. Moreover, since AI selection projects are not yet fully priced by the market, this provides potential alpha opportunities for identifying banks with high AI readiness.

Risks

• Inflation risk in countries with current account deficits and low foreign exchange reserves, which may lead to rising interest rates, damaging profit margins of banks with high loan-to-deposit ratios and deteriorating credit quality.
• Asymmetric rise in deposit rates with less rise in loan rates may squeeze currently higher loan spreads.
• Banks with equity/assets ratio below 5% and ROA below 1%, if loan annual growth exceeds 15%, will face economic and regulatory risks of capital shortage.
• AI Execution Risk: Technology investment fails to convert into expected financial benefits or risk mitigation outcomes.

What to watch

• Progress in banks' AI deployment from pilot to enterprise-scale application.
• Quantifiable income statement (P&L) impact brought by AI-related investments, especially appearance of explicit cost savings.
• Differences in AI readiness across different markets and banks and their impact on valuation.
• Extent to which the long-term structural impact of population aging on net interest margin and fee income continues.