Drawdowns due to COVID-19 happened very quickly compared to major events like the Great Depression. Advanced risk management and proper portfolio allocation are becoming more important.
By Professor LI Yingying,
Department of Finance & Department of Information Systems, Business Statistics and Operations Management, HKUST Business School
The COVID-19 pandemic is having a dramatic impact on the global stock market. Take the major stock indices, for example. Figure 1 below shows the cumulative returns in 2020 of seven major indices: SP500, FTSE100, DAX, CAC40 (left), HSI, SSE Composite Index, and Nikkei 225 (right). The drawdown periods are shown in bold. For all these markets, major drawdowns occurred in March. In terms of magnitude, European markets experienced the largest declines, which was almost 1.5 times as large as the Hong Kong market.
Figure 1: Maximum drawdowns for global stock indices in the first two quarters of 2020 (left for US & European markets and right for Asian markets)
To get a bigger picture of the current COVID-19 shock, we must employ history and compare the 2020 drawdown with major historical drawdowns. Focusing on the SP500, we can calculate its annual maximum drawdowns since 1928, totally 93 years.
We see that, compared with the major drawdowns of the past, the drawdown this year occurred within the shortest period of time, which denotes the severity of the crisis.
Meanwhile, the volatility index VIX saw a rapid spike in March. Although the index dropped significantly afterwards, as of June 30, 2020, the VIX value still stands at a high level.
Besides the stock markets, COVID-19 also significantly impacted the commodity and cryptocurrency markets. Figure 3 shows the cumulative returns in the first two quarters of this year for WTI crude oil, gold, and bitcoin (BTC). All of them suffer huge declines.
COVID-19, along with many other major events, can disrupt the entire financial market. When such events happen, portfolios which lack proper risk management are very vulnerable.
Mean-variance optimization
In the following discussion, we re-visit one fundamental risk management and portfolio optimization framework, the mean-variance optimization framework, and introduce one of the latest developments.
The Markowitz Mean-Variance optimization intends to maximize portfolio return at a given risk constraint. The risk constraint level varies among different investors according to their risk preferences.
One unique feature of the modern mean-variance optimization is that it is convoluted with “big data” challenges, simply due to the fact that there are a large number of stocks/assets available for investing.
In the article "Approaching mean-variance efficiency for large portfolios" by Ao, Li, and Zheng (2019, The Review of Financial Studies), the authors discussed how to find the mean-variance efficient portfolios of a large number of assets. The traditional method would use the historical sample mean and sample covariance matrix for portfolio construction. The resulting portfolio had been found to perform poorly in practice. Ao, Li, and Zheng (2019) provided a theoretical explanation to such a “puzzle”. The fundamental reason is high-dimensionality, which means that the number of assets is not small compared with the sample size. To resolve the challenge, the authors provide an innovative method, MAXSER, that can approximately attain the maximum expected return and also keep risk below the desired level.
The key ideas include transforming the original Markowitz optimization into an equivalent unconstrained optimization problem and applying the LASSO regression. The method has been found to give a superior performance in extensive empirical studies. With the risk constraint set to the same level as the risk level of SP500, we build a portfolio with MAXSER based on the US stock market. Figure 4 shows the comparison between SP500 and MAXSER in the first two quarters of this year. We see that MAXSER performs much better than SP500. It achieved a substantially higher total return and incurred a much smaller drawdown. In terms of long-term performance, during the period 2000 – 2020, MAXSER and SP500 had a similar (annualized) risk of 18%. But MAXSER delivers a much higher Sharpe ratio (0.6 vs 0.2, after deducting transaction costs).
To summarize, we witnessed the historical COVID-19 drawdown. The decline appeared strikingly quickly, even compared to devastating historical events such as the Great Depression and World War II. Furthermore, as modern portfolios involve large numbers of assets, advanced risk management and proper portfolio allocation become ever more important.
“The COVID-19 drawdown which triggers the decline appeared incredibly quickly, even compared to the devastating historical events such as the Great Depression and World War II.”
Reference
Ao, Li, and Zheng. Approaching mean-variance efficiency for large portfolios. The Review of Financial Studies, 2019.