The danger of shallow diagnosis
The collapse of Silicon Valley Bank (SVB) demonstrates how dangerous this gap can become. In March 2023, SVB collapsed after a rapid liquidity crisis, with more than $40 billion withdrawn in a single day. What once appeared to be a successful, fast-growing bank quickly became a financial cautionary tale.
During the technology boom, SVB experienced rapid growth supported by a concentrated base of tech companies and venture capital firms. But beneath that growth were fragile foundations: long-term assets exposed to rising interest rates, a large base of uninsured depositors, and a highly connected customer network.
When concern began to spread, digital banking and social media accelerated panic at a speed traditional risk models had not fully anticipated.
Descriptive analytics could show that deposits were falling, asset values were under pressure, and liquidity was disappearing. But diagnostic analytics would ask deeper questions: Why were depositors so quick to withdraw? Why were assets so vulnerable to rising rates? Why did panic spread so quickly? And most importantly, how did these risks amplify one another?
The failure was not simply that SVB had weak indicators. It was that the connections between them were misunderstood.
Thailand’s household debt problem is not just about debt
While SVB shows how financial stress can accelerate suddenly, Thailand shows how it can compound slowly over time.
Thailand’s household debt remains elevated at 86.8% of GDP, among the highest in Asia. The issue is not only the level of debt, but the conditions driving it. Slow income growth, rising living costs, and reliance on consumption-based borrowing suggest that credit is filling the gap between income and daily expenses.
This strain is unfolding in a weak macroeconomic environment. The IMF points to low investment, weak productivity growth, demographic ageing, and pandemic related household debt as pressures on Thailand’s resilience. With GDP growth slowing to 2.1% in 2025 and projected to fall further to 1.6%, households have less economic capacity to absorb debt burdens.
At the same time, high debt makes recovery harder. It weakens consumption, reduces households’ ability to absorb shocks, and places pressure on banks and credit supply. This creates a self-reinforcing loop: weak growth makes debt harder to manage, while high debt suppresses recovery. As a result, the economy becomes more vulnerable to shifts in financial conditions and traditional monetary policy becomes less effective.
Thailand’s challenge lies in how financial pressure can spread across consumption, lending, and growth, gradually weakening economic resilience. Diagnostic analytics becomes critical to identify where intervention can still restore stability.
How AI can improve diagnosis
AI strengthens diagnostic analytics by helping organizations anticipate how financial stress may spread through a system and intervene before instability accelerates.
Its value lies in combining fragmented signals from financial, behavioral, macroeconomic, and sentiment data into a clearer risk picture. This allows AI to detect patterns and relationships that are too complex, fast-moving, or interconnected to identify manually.
Machine learning can identify drivers strongly linked with repayment stress, while feature importance models separate surface-level correlations from deeper risk factors. Anomaly detection can also flag sudden shifts in repayment behavior or credit usage, giving banks and policymakers earlier warning.
Natural language processing adds a behavioral layer by analyzing customer complaints, public sentiment, social media discussion, and financial anxiety. This can reveal shifts in confidence before they appear in traditional financial indicators.
Scenario simulations could test how rising interest rates, weaker income growth, tighter lending, or falling consumption may affect defaults, resilience or recovery. This shifts analytics from describing the problem to identifying which levers can change its trajectory.
The power of AI-enabled diagnostic analytics lies not in generating more data, but in enabling earlier diagnosis, sharper intervention, and a clearer view of how today’s pressure can become tomorrow’s constraint.
Sources
Bank of Thailand (2024) Financial Stability Review. Available at: https://www.bot.or.th/content/dam/bot/documents/en/research-and-publications/reports/financial-stability-report/FS_Review_2024e.pdf
Casey, A.J. (2025) ‘Silicon Valley Bank Collapse’, Banking and Finance Law Review. Available at: https://chicagounbound.uchicago.edu/cgi/viewcontent.cgi?article=2726&context=law_and_economics
International Monetary Fund (2025) Thailand: Article IV Consultation Report. Available at: https://www.imf.org/-/media/files/publications/cr/2026/english/1thaea2026001-source-pdf.pdf
Klinthanom, T. (2024) Thai Household Debt and Risks to the Economy. Krungsri Research. Available at: https://www.krungsri.com/en/research/research-intelligence/household-debt