expectile-based Value-at-Risk (EVaR)

Expectile-based Value-at-Risk (EVaR) is an advanced market risk measure derived from the concept of expectiles, introduced by Newey & Powell (1987). Unlike traditional quantile-based VaR, EVaR is defined by minimizing an asymmetric squared loss function, making it sensitive not only to the frequency of losses but also to their magnitude beyond the threshold. This property makes it an 'elicitable' risk measure and, under certain conditions, a 'coherent' one. While not specified in standards like ISO 31000, its philosophy aligns with the Basel Committee on Banking Supervision's (BCBS) Fundamental Review of the Trading Book (FRTB, standard d457), which advocates moving from VaR to Expected Shortfall (ES) to better capture tail risk. EVaR serves as a more conservative and sensitive alternative to VaR, addressing VaR's inability to quantify the severity of extreme losses.

In enterprise risk management, particularly for financial institutions, EVaR significantly enhances the accuracy of market risk capital calculations. The implementation involves three key steps: 1. **Data Collection & Model Selection**: Gather high-quality historical financial data spanning at least five years. Select an appropriate EVaR model, such as a dynamic EVaR-GARCH model, to capture market volatility clustering. 2. **Parameter Estimation & Calculation**: Use Asymmetric Least Squares (ALS) regression on the historical data to estimate model parameters for a specific confidence level τ (e.g., τ=0.995). Calculate the one-day or ten-day EVaR value based on the calibrated model. 3. **Validation & Backtesting**: As required by regulatory frameworks like the Basel Accords, conduct rigorous backtesting by comparing past EVaR forecasts with actual profit and loss (P&L). If the number of exceptions (actual losses exceeding EVaR) is too high, the model must be recalibrated. Global banks have reported a 15% reduction in backtesting exceptions during stress periods after adopting such advanced models.

Taiwanese enterprises face three main challenges when implementing EVaR: 1. **Data Quality and Availability**: Insufficient length or granularity of historical data for specific local financial products can impair model accuracy. The solution is to use data augmentation techniques or proxy data from highly correlated markets, with all assumptions clearly documented. 2. **Model Complexity and Talent Gap**: EVaR models require sophisticated quantitative and programming skills, which are scarce. A phased implementation, starting with core portfolios, is recommended. Partnering with expert consultants like Winners Consulting can bridge the talent gap and facilitate knowledge transfer. 3. **Regulatory Justification**: Demonstrating the new model's validity to local regulators, who may be more familiar with traditional VaR, requires significant validation and documentation efforts. The strategy is to run the EVaR model in parallel with the existing VaR model and present comparative backtesting results that prove EVaR's superior performance in capturing local tail risks.

Winners Consulting specializes in expectile-based Value-at-Risk (EVaR) for Taiwan enterprises, delivering compliant management systems within 90 days. Free consultation: https://winners.com.tw/contact