Honey, I Shrunk the Beta
Investors often look to good ol' beta as the tried and true mechanism for hedging market risk in their portfolios. However, many have noted that lately, beta feels “broken”.
Hedging beta no longer seems to have the desired effect of mitigating risk in portfolios, and in fact, investors who have tried to hedge beta in 2020 and 2021 will have found major headwinds as the beta factor took off on a wild ride in the post-COVID era.
A look at the Beta factor from the MSCI Barra Global Equity Model illustrates this well.
While the chart above shows that there was a similar behavior in the performance of the Beta factor coming out of the Global Financial Crisis in 2008 and 2009, investors will tell you that beta “feels” different compared to any time in recent history.
To investigate this, we went back to basics to answer the very simple question: does beta still explain market movements?
Does Beta Explain the Market?
For our analysis, we will use the following risk models: Axioma Worldwide 4 Medium Horizon (“Axioma Worldwide”), Axioma US 4 Medium Horizon (“Axioma US”), MSCI Barra Global Equity Model (“Barra Global”), and MSCI Barra US Equity Model (“Barra US”). Rather then rely on analysis from a single model, we are leveraging the views of multiple risk models to further confirm whether or not the changing relationship of beta to the market is prevalent across universes and methodologies.
To answer our question, we calculated a 6-month rolling correlation between the Beta and Market factor returns from each of the risk models. For the global models, the Market factor represents the return to a global market (think roughly the ACWI); for the US models, the Market factor represents the return to a US market (think roughly the Russell 3000). If beta is a good indicator of market returns and volatility, then at a minimum, the correlation between the Beta and Market factors should be positive and relatively close to 1.
A chart of the Beta vs Market factor correlation for the global models shows us that historically, our expected relationship has held, with correlations between 0.8 and 1 for much of the period since 2008. However, the past few years have shown a major decoupling of beta from the market. The events of 2020 and 2021 have only proven to accelerate that decoupling, with the correlation from the Axioma Worldwide model currently at 0.56 and the Barra Global model at 0.68.
A chart of the correlations for the US models shows an even steeper unraveling for the Beta vs Market factors.
Though beta is not intended to be a perfect indicator for market volatility, we see that it is even farther from perfect in today’s market environment where many other systematic effects are at play.
Is There Any Hope for Beta?
So, what does the above tell us? In short, if historical beta no longer correlates to the market, it may no longer be an effective tool for reducing market risk in your portfolio. One silver lining that we can offer is that all correlations presented above appear to be showing a small uptick as of the past few weeks. Perhaps this is a sign that market dynamics might be marching back to “normal” as the global reopening accelerates. However, it remains to be seen which direction beta will ultimately go, and it is also possible that the current relationship of beta to the market will persist. If that proves to be the case, investors need to be prepared to look for alternative tools to more adequately mitigate risk.
In future editions of Factor Spotlight, we will continue our investigation into the efficacy of beta as a market risk hedging tool and offer potential alternatives to the traditional beta-hedging strategy.
US & Global Market Summary
US Market: 05/17/21 - 05/21/21
- It was a mixed week across the major market averages, with the S&P 500 and DJIA ending the week lower and the Nasdaq slightly positive after tech shares rebounded on Thursday.
- Inflation fears drove weakness on Tues and Wed, as the minutes from the Fed’s April meeting suggested that some members are inclined to reduce asset purchases in future meetings.
- Earnings season is drawing to a close, with over 95% of S&P 500 companies having already reported Q1 results. According to Factset, a record 86% of those companies beat EPS estimates.
- Crypto was front and center this week after regulatory fears (primarily around China) prompted sharp volatility across all digital currencies - with Bitcoin falling by 25%.
Normalized Factor Returns: Axioma US Equity Risk Model (AXUS4-MH)
- Volatility saw the strongest positive move this week, moving more than 1/2 of a standard deviation back towards the mean and exiting Oversold territory.
- Market Sensitivity has started to rebound off of a recent bottom of -1.9 on 5/13 as it moved up +0.17 SD towards the mean.
- Value continued to bounce off a recent trough of -1.76 SD below the mean on 5/4.
- Momentum remains Overbought at +1.14 standard deviations above the mean after a relatively flat week.
- The decline in Earnings Yield continued, as it nears Oversold territory at -0.91 SD below the mean.
- Size fell out of Overbought space with a -0.33 standard deviation move, and now sits at +0.74 SD above the mean.
- Profitability, the week’s biggest loser, declined by -0.35 SD as it now hovers just above its historical mean.
- US Total Risk (using the Russell 3000 as proxy) declined by 22 basis points this week.
Normalized Factor Returns: Axioma Worldwide Equity Risk Model (AXWW4-MH)
- Just as we saw in the US, global Volatility rallied on a normalized basis and has shed its Oversold label at -0.74 SD below the mean.
- Exchange Rate Sensitivity saw continued strength after entering positive normalized territory last week.
- Momentum dropped by 0.19 standard deviations, and is now on the cusp of exiting Overbought space.
- Earnings Yield continued to decline, garnering an Oversold designation at -1.23 SD below the mean.
- Profitability was the biggest loser worldwide as well, tumbling out of Extremely Overbought space after recently peaking at +2.41 SD above the mean on 4/21.
- Global Risk (using the ACWI as proxy) declined by 25bps.