Unusual market volatility challenges risk models
Recent market behavior is showing signs of unusual volatility, which may challenge existing risk analytics models. In a recent podcast, analyst Brent Kochuba noted that in the last eight sessions, 115 S&P 500 stocks fell over 7% in a single day, while the index itself only declined 1.5%. Historically, such widespread single-stock declines have coincided with market drawdowns of around 34%, suggesting current models may be underestimating tail risk.
- Tail risk, also known as "fat tail risk," refers to the financial risk of an asset's price moving more than three standard deviations from its average, an outcome considered highly improbable in traditional models that assume a normal distribution of returns. These rare "black swan" events can have a disproportionately large impact on a portfolio. - Standard risk metrics like Value-at-Risk (VaR) have limitations in volatile conditions; while VaR estimates the maximum potential loss at a certain confidence level, it does not provide insight into the potential severity of losses that exceed this threshold. More advanced metrics like Conditional Value at Risk (CVaR) are used to estimate the average loss in worst-case scenarios. - For the insurance industry, sustained market volatility directly impacts investment income, a significant contributor to profits, and complicates the pricing and hedging of products with long-term guarantees, such as variable annuities. The European Insurance and Occupational Pensions Authority (EIOPA) has flagged market risks as a key vulnerability for the sector, noting that lapse rates have also increased. - The CBOE Volatility Index (VIX), a key measure of market expectation of near-term volatility, surged to its highest level since early 2023 during recent market drawdowns. While volatility was generally lower in the first half of 2024 compared to 2022 and 2023, some forecasts expect a return to a medium-to-high volatility range. - The challenges of modeling in volatile markets are not just theoretical; in 2020, Zillow's automated home-price valuation model failed to predict market shifts accurately, leading to over $420 million in losses and the shutdown of its home-buying division. - In response to the limitations of static models, the field of Model Risk Management (MRM) is increasingly focused on continuous, real-time monitoring and integrated stress testing of models in production. This is especially critical for newer AI and machine learning models, where issues of interpretability and potential for bias can introduce additional layers of risk. - From an actuarial viewpoint, market stability can be understood through the lens of social networks and the spread of ideas. A tipping point for a major market swing can occur when a new belief about the market infects as little as 25% of the tightly-knit group of influential participants, a much lower threshold than a simple majority.
