TWSE Market Insights

＜中文版＞

Under Taiwan’s T+2 settlement cycle, securities firms are exposed to price volatility and performance uncertainty during the interval from trade execution to final settlement. As market trading volume expands and trading strategies diversify, managing "pending settlement positions" has become central to ensuring the integrity of the clearing and settlement ecosystem.

To mitigate systemic risks, the Taiwan Stock Exchange (TWSE) employs a multi-layered defense strategy. Daily surveillance is conducted through cross-departmental meetings, monitoring liquidity, exposure concentration, and market anomalies in real-time. On the financial resource front, the Joint Clearing and Settlement Fund (JCSF) serves as a backstop. Stress tests are regularly performed to ensure the fund satisfies the "Cover 2" standard—the ability to withstand the simultaneous default of the two largest clearing members. Critically, individual firm contributions to this fund are dynamically adjusted based on the daily average Value at Risk (VaR) from the previous month, ensuring that financial buffers are commensurate with actual market exposure.

Functionally, this mechanism mirrors the "Initial Margin" (IM) requirements mandated by international Central Counterparties (CCPs). Both seek to quantify market risk as a prerequisite for institutional stability. Accordingly, this analysis utilizes global IM standards as a benchmark to evaluate the efficacy and future evolution of Taiwan’s risk measurement framework.

VaR quantifies the maximum potential loss of a portfolio over a specific horizon at a predefined confidence level. In the context of pending settlement, market risk—driven by price fluctuations—is the most immediate threat to position value.

TWSE currently adopts the Variance-Covariance Method (a parametric approach). This model is favored for its transparency, computational speed, and consistency, making it ideal for high-volume, daily risk monitoring in markets dominated by linear cash products.

However, the parametric approach has its limitations:

1. Distributional Assumptions: The method typically assumes that asset returns follow a specific probability distribution. If actual market returns exhibit skewness or fat tails, the model may underestimate tail risks.
2. Stability of Parameters: It often assumes that volatility and asset correlations remain relatively stable during the observation period. However, during periods of market stress, correlations between assets may rise simultaneously, diminishing the intended diversification effects and impacting the model's risk estimation for extreme market conditions.
3. Information Beyond the Threshold: VaR only defines a threshold below which losses are expected to stay. It offers no further information regarding the magnitude of losses that exceed that threshold.

Despite theoretical limitations, the parametric method remains highly robust when applied to Taiwan’s specific market architecture:

1. Market Instrument Structure: Over 99% of Taiwan’s centralized market trading value is concentrated in stocks and ETFs. The absence of widespread, non-linear derivatives simplifies the return structure, validating the use of parametric models.
2. Regulatory Volatility Controls: The 10% daily price limit serves as an institutional constraint on extreme volatility. Historical data (2021–2025) shows that single-day fluctuations exceeding 5% occur in only 0.4% of trading sessions. While the TAIEX does exhibit "leptokurtic" (fat-tail) characteristics, the magnitude of these tails is significantly constrained by regulatory price ceilings.
3. Operational Efficiency: Real-time risk management requires models that can process vast datasets within tight windows. The efficiency of the parametric method ensures that surveillance indicators are available for immediate decision-making.

International clearing standards, guided by the CPMI-IOSCO Principles for Financial Market Infrastructures (PFMI), emphasize risk sensitivity and the need for addressing procyclicality. In Europe, the European Market Infrastructure Regulation (EMIR) and its subsequent amendments also provide stringent standards for Initial Margin (IM) models, confidence levels, and model validation standards. Under these global regulatory frameworks, major Central Counterparties (CCPs) such as the US NSCC, Japan JSCC, Korea KRX, Hong Kong HKSCC, and EU Cboe Clear Europe have developed IM systems that, while distinct, share a comparable multi-layered risk management structure.

This framework is generally organized across four functional layers, beginning with asset segmentation to determine the applicability of core risk models based on liquidity profiles or specific product characteristics. This is followed by core risk models where market risk is estimated using tools such as Value at Risk (VaR) or Expected Shortfall (ES) to capture potential losses under normal market conditions. Additionally, risk add-ons are integrated to address exposures that core models may not fully encompass, such as market illiquidity, position concentration, or specific credit risks. Finally, anti-procyclicality (APC) arrangements—including margin floors, buffer factors, or stress scenarios—are implemented to maintain stable margin levels across varying market environments and mitigate excessive procyclical effects. By adopting this tiered approach, international CCPs ensure that their margin systems are resilient enough to cover both systemic price volatility and idiosyncratic risks.

Based on Taiwan’s market characteristics and international trends, the current parametric-based risk measurement framework remains fundamentally robust. However, as market scale grows, return distributions exhibit fat-tail characteristics, and international regulatory standards are progressively strengthened, the TWSE could pursue strategic refinements in two key areas:

1. Evaluating Multi-layered Risk Frameworks: The TWSE may evaluate the feasibility of differentiated treatment for positions with low liquidity, unique price volatility, or high concentration. By incrementally incorporating simplified liquidity or concentration add-ons, the exchange can effectively mitigate residual exposures that a single VaR estimate may overlook. Furthermore, referencing international practices to operationalize anti-procyclicality (APC) frameworks could assist in dampening the volatility of monthly fund call requirements.
2. Operationalizing Parallel Simulation Mechanisms: While maintaining operational efficiency, the TWSE could conduct parallel simulations utilizing Historical Simulation or Expected Shortfall methodologies alongside the existing parametric approach. Such long-term data aggregation would facilitate the observation of model variance under extreme scenarios and serve as a quantitative baseline for future parameter recalibrations, the introduction of add-ons, or iterative improvements to core risk models.

Overall, Taiwan’s current risk measurement framework for pending settlement positions—anchored by parametric VaR and supplemented by multi-layered monitoring, stress testing, and the JCSF dynamic contribution mechanism—remains highly effective. Given the market’s concentration in cash products, simplified asset structures, and existing regulatory price limits, the system offers significant operational advantages.

However, international trends indicate that risk management frameworks are increasingly prioritizing tail risks, risk add-ons, and anti-procyclicality designs. Looking ahead, strategic refinements to Taiwan’s models will enhance sensitivity to market dynamics, ensuring that the clearing and settlement system remains robust and internationally competitive while further fortifying the overall stability of market operations.