Research — March 12, 2026
Macro-to-Micro Simulation Nankai Trough Earthquake Scenario
By Takei Ryo, Yan Jiang, and Min Jiang
As we enter 2026, financial institutions face a wide range of macro risks. Recent geopolitical tensions—including conflicts in the Middle East and shifts in U.S. policy—have heightened concerns around global instability. Geopolitical instability, and trade and tariff policy changes remain top-of-mind1, while issues such as affordability pressures, persistent inflation, and the economic spillover effects of rapid AI adoption on labor markets and the energy sector continue to shape the risk landscape2.
Against this backdrop, it is unusual to encounter a macro risk that is (1) highly likely to occur, (2) expected within a defined time window, and (3) potentially catastrophic. In Japan, the Nankai Trough mega earthquake represents precisely such a risk. Professor Hiroki Kamata of Kyoto University, a leading authority on seismic and volcanic hazards, has warned:
A huge Nankai Trough earthquake is expected to occur in the 2030s, affecting 68 million people, half of the total population. The Great West Japan Earthquake [another name for the Nankai Trough earthquake] will cause 10 times more damage than the Great East Japan Earthquake, and before and after it, there is a possibility that an earthquake directly hitting the Tokyo metropolitan area will cause five times more damage than the Great East Japan Earthquake, and that it will also trigger an eruption of Mt. Fuji. Both are severe disasters that shake the very foundations of Japan and require urgent countermeasures.3
Based on Professor Kamata’s assessment, the Japanese government and the private sector may have only 5–15 years to prepare for and mitigate the potential impacts of such an event.
Macro-to-Micro Simulation Framework
For financial institutions with asset or loan portfolios, identifying macro risks on the horizon is essential, but equally important is understanding how those risks translate into specific exposures. To support this, we introduce a framework that maps macro‑level shocks into portfolio‑ or risk‑factor‑level “micro” risks.
The transition from a macro event to granular risk measures can be distilled into three steps:
- Generate macro shocks. Interpret the macro event and translate it into shocks on key macro risk factors. This step typically relies on expert judgment and historical analogues of similar crises to produce plausible shock values across variables such as equity indices, GDP, interest rates, currency pairs, and commodity prices.
- Translate macro shocks into micro-level shocks. Using historical correlations between macro and micro risk factors, project the macro shocks onto portfolio‑level drivers to produce consistent micro‑level risk‑factor shocks.
- Compute granular risk measures. Apply the micro‑level shocks within valuation models to derive portfolio P&L impacts or to estimate firm‑level credit risk metrics.
The main challenge in implementing these steps is ensuring the framework remains both historically grounded and scalable. Steps 1 and 2 must be anchored to historically observed covariance relationships, while steps 2 and 3 must be capable of handling a large number of risk factors without sacrificing consistency or tractability.
To demonstrate the framework, we apply it to the Nankai Trough mega earthquake scenario for a Japanese Yen (JPY)‑denominated asset portfolio. In addition, to highlight its relevance for credit analysis, we extend the illustration to major constituents of the iTraxx Japan index.
Macro Shock Generation – Global Link Model
The S&P Global Market Intelligence Global Link Model4, designed for forecasting and scenario planning, links 70 individual country models with each other and with key global drivers of performance. The model accounts for 95% of global gross domestic product (GDP), with 250 to 500 time series per country, and is updated quarterly. The model provides baseline and bespoke 30-year forecasts and incorporates detailed expert assessments.
In this scenario, the earthquake is assumed to occur at the start of 2026, with shocks evaluated over four quarters. These shocks include movements in Japanese equities, interest rates, credit spreads, and USDJPY, as shown in the table below. The result indicates that the earthquake‑driven shock occurs in the first quarter—most visibly through Japanese equity and credit spread movements—while the following quarters are characterized by post‑shock recovery dynamics.
Inference Model
The inference model, a component of the S&P Global Market Intelligence Buy-Side Risk Solution, enables users to derive micro‑level shocks from a selected set of macro shocks. At its core, the model applies a multivariate regression framework in which the sensitivities—or “betas”—of micro risk factors to macro shocks are estimated using historical covariance, similar to the Capital Asset Pricing Model. The historical covariance structure is determined primarily by the chosen calibration window and other model parameters5. In practice, this covariance structure is as influential as the macro-shock inputs themselves in shaping the resulting micro‑level shock predictions.
To model the Nankai Trough earthquake scenario, we use two distinct covariance windows:
- First‑quarter covariance window (Feb 1–Mar 17, 2020): This period captures the initial market shock associated with the onset of the COVID‑19 pandemic and is used to calibrate the behavior of risk factors during the immediate disruption.
- Subsequent‑quarter covariance window (Mar 18–Jun 1, 2020): This window reflects the early recovery phase in financial markets and is applied to shocks in the remaining three quarters.
These windows are selected to align the modeled dynamics with the scenario assumptions.
Portfolio Profit and Loss
We consider an asset portfolio, denominated in JPY, composed of four asset types:
- Japanese stocks
- Japanese government bonds (JGB)
- Foreign (non-Japanese) stocks
- Foreign (non-Japanese) bonds
Bond positions are valued using risk‑free yield curves and z‑spreads, while equity positions are valued using spot equity prices. Foreign asset valuations also incorporate the relevant exchange rate between JPY and the assets’ currency. All asset values are computed under both the Nankai Trough earthquake scenario and the baseline (no‑earthquake) scenario.
Figure 1 shows the portfolio value for both the Nankai-Trough (red) and the baseline (black) scenarios. The dotted lines show the portfolio value should the assets be rebalanced at the end of each quarter.
Figures 2–5 show the evolution of the mean and distribution of the four asset types. These plots show that the initial rise in the portfolio is driven primarily by the JGBs and the sharp rise in USDJPY on the foreign assets, which overwhelm the drop in the Japanese stocks. However, by the second quarter, the reversal of USDJPY pulls back the portfolio gains. By the final quarter, the reversion of USDJPY and the overall weakness of the Japanese and foreign stocks push the portfolio lower, despite positive returns on the JGBs and foreign bonds.
Figure 1: Evolution of the portfolio value for the Nankai-Trough earthquake and baseline scenarios. Dotted lines show the evolution with end-of-quarter rebalancing.
Figure 2: Evolution and distribution of Japanese stock positions. Left: Baseline scenario. Right: Nankai-Trough scenario.
Figure 3: Evolution and distribution of Japanese bond positions. Left: Baseline scenario. Right: Nankai-Trough scenario.
Figure 4: Evolution and distribution of non-Japanese stock positions. Left: Baseline scenario. Right: Nankai-Trough scenario.
Figure 5: Evolution and distribution of foreign bond positions. Left: Baseline scenario. Right: Nankai-Trough scenario.
Credit Risk Analysis
Beyond market valuations, the Nankai Trough event also has meaningful implications for both default risk and credit transition risk. Micro‑level analysis can be applied to specific risk factors and then be post‑processed to generate informative credit‑risk insights. In the sections that follow, we illustrate this using two examples: single‑name default probabilities and credit transition probabilities.
Default Probability Impact
Five-year CDS credit-spread shocks were used to estimate default probabilities6 across major Japanese firms. The results highlight broad deterioration, with lower-rated firms experiencing sharper increases in default risk.
Credit Rating Deterioration Probability
The outputs of the inference model can be used to feed into the S&P Market Intelligence Credit Transition Product (CTP). CTP takes in, among other factors, CDS par credit spreads (single-name and sector curves), z-spreads, equity index prices, and GDP7 to estimate the likelihood of a credit deterioration. Below is the 1-year credit deterioration probability for the same set of firms.
Similar to the default probability impact, findings show that lower-rated firms experience substantial jumps, with some seeing 10 percentage point increases in 1-year credit deterioration likelihood. For several entities, this probability rises from 6%–8% to well above 10%, underscoring the significant impact of the earthquake shock.
Conclusion
An effective risk management framework integrates expert judgment, scalable software solutions, and advanced analytical models into a seamless, transparent, and flexible workflow. The framework presented here demonstrates how a macro‑level event can be translated into granular outputs, such as trade‑level P&L impacts and single‑name credit‑risk metrics. Although the case study focuses on a specific macro event and portfolio, the methodology is broadly applicable to both historical and hypothetical scenarios.
Content Type
Segment