Before we jump into this week’s conclusion to our three-part beta series, I want to briefly veer back to an earlier topic I had covered during our recent macro series, which generated many positive comments and feedback from our readers. For those of you who would like to continue your macro explorations, my colleague Kevin Wahlberg will be co-presenting at a joint webinar with Qontigo on June 23 alongside Qontigo’s Melissa Brown titled: Macro Matters, Even for Fundamental Managers. We anticipate a timely, deep-dive discussion with extensive Q&A, and we hope to see many of you there.
Our recent issues of Factor Spotlight have centered on the deteriorating efficacy of beta as a measure of market risk. One possible culprit is the increasing concentration in market-cap-weighted universes that represent key ingredients when deriving beta. The evidence showed a downward trend in the correlation of beta to the market that coincides very closely with market cap concentration. The charts below illustrate that as beta has decoupled from the market, the effective number of assets (our measure of concentration) has decreased as well, signaling that we may be able to tie increased market concentration to beta’s breakdown.
This week, we wrap up our current beta investigation by proposing alternative benchmarks to use in beta calculation that may be more representative of the actual investable market.
Alternatives to Cap-Weighted Benchmarks
For our analysis, we endeavored to create new benchmarks that will be more illustrative of today’s broad market and prove more effective for market risk measurement. To accomplish this, we started with the Russell 3000 (proxied by IWV) and the MSCI All Country World (proxied using ACWI) and investigated three alternative weighting schemes:
- Square root of market capitalization - this weighting scheme takes is a commonly used quantitative technique to diminish the influence of the largest companies and elevate the smallest companies.
- Market capitalization, with securities greater than 1% removed - this weighting scheme removes any security that has greater than 1% weight in the standard cap-weighted benchmark, and then pro-rata re-weights the remaining securities.
- Market capitalization, with securities greater than 1% capped - this weighting scheme caps security weight at 1%, and then pro-rata re-weights the securities that have less than 1% in the standard cap-weighted benchmark.
A benefit of the first approach is that it is less discretionary than the other two approaches, both of which require some pre-defined threshold to determine whether a security should be excluded or reduced. However, the downside of the first approach can result in overcorrecting the problem and providing too much representation of small-cap names. Our second and third approaches benefit from giving an intuitive and straightforward mechanism for reframing market benchmarks.
Reducing Benchmark Concentration
To evaluate the effectiveness of our proposed benchmark weighting schemes, we recalculated the benchmark concentration for each scheme. As a refresher, our measure of concentration is the effective number of assets in the universe, calculated as the inverse of the Herfindahl-Hirschman Index (sum of squared security weights). A lower effective number of assets conveys that the universe is becoming more highly concentrated, whereas a higher number suggests that the universe is becoming more diversified.
The table below summarizes the distribution over time for the effective number of assets.
Immediately, we can see that the alternative weighting schemes improve the concentration issue and generally have a higher number of effective assets. Because the range of effective number of assets differed heavily between the root-cap weighting scheme and the other two methods, we indexed the metric to 100 at the beginning of the period to more easily compare the trends over time.
Interestingly, the alternative schemes have varying impacts depending on the universe.
For the US universe, all three schemes create much more consistency around the effective number of assets over time, with the “> 1% capped” weighting resulting in the most stable level of effective assets while avoiding the massive increase in concentration that we’ve seen over the past several years.
For the global universe, the results of the alternative schemes were mixed. It appears that the root-cap scheme may have overcorrected the problem, whereas the other two schemes still saw some increase in concentration, particularly in the 2019 time frame. However, it is clear that the “> 1% capped” and the “> 1% removed” weightings were improvements over the standard market-cap weighting.
Alternative Benchmarks Reduce Sector Concentration
For those investors who don’t hold any mega-cap technology stocks, the current market-cap-weighted benchmarks are almost nonsensical for comparison purposes. But do our new proposed benchmark weighting schemes provide improvement?
A review of the industry distribution of the new benchmarks vs. the standard cap-weighted benchmark shows lower industry concentration across all three new measures.
To take an extreme example, we considered the industry and industry group breakdown for August 2020 when the Russell 3000 and MSCI ACWI hit low points of effective number of assets (90 effective assets and 207 effective assets, respectively).
Across both the US and global universes, the new weighting schemes reduced the concentration of the top 5 industry groups. In particular, the impact of Software & Services is lower, and the overall distribution of the top 5 industry groups is much more even compared to the traditional market cap-weighted indices. The results encourage that alternative weighting schemes may be invaluable to many investors, especially those who have a broader view of the markets beyond mega-cap tech stocks.
Time to Overthrow Mega-Cap Tech in Market Benchmarks?
The conversation around benchmark concentration is not a new one; yet, the alternatives noted above, along with others, have struggled to become adopted within the investment community. Though further analysis is required to definitively prove that the key to fixing beta is linked to reducing concentration, our cursory investigation illustrates that a more holistic approach to benchmark measurement could be a step in the right direction. Investors must begin considering new ways to look at beta to capture market risk in their portfolios appropriately.
If you are interested in discussing ways to evaluate your portfolio’s beta using benchmarks with custom weighting schemes, please feel free to reach out to us.
US & Global Market Summary
US Market: 06/07/21 - 06/11/21
- The market ended the week slightly higher, with the Nasdaq handily outperforming the other major indices - up 1.9% on the week vs. the S&P 500 which saw a +0.4% return.
- Investors seemed to agree with the Fed’s thesis that the current spike in inflation is only transitory, and summarily dismissed a report showing May CPI was up 5% YoY (its steepest increase since 2008).
- The bond market corroborated this agreement as yield on 10Y Treasuries saw the largest one-week drop in nearly a year, down to 1.46%.
- Job openings in the US increased by almost 1mm in April (now totaling 9.3mm), which is the highest overall number since the BLS began tracking this data at the turn of the century.
Normalized Factor Returns: Axioma US Equity Risk Model (AXUS4-MH)
Methodology for normalized factor returns
- Volatility continued to soar for another week, up another 0.7 standard deviations as it shot towards Extremely Overbought space (now at +1.92 SD above the mean).
- Market Sensitivity saw a similar boost as it vaulted out of negative normalized territory and landed at +0.15 SD above the mean.
- Earnings Yield continued to head deeper into Oversold territory, currently sitting at -1.53 SD below the mean.
- Value drifted deeper into Oversold territory, now at -1.52 SD below the mean.
- Momentum moved lower by 0.3 standard deviations after exiting Overbought territory last week.
- Profitability was again the biggest loser, down 0.34 standard deviations and landing at -1.48 SD below the mean.
- US Total Risk (using the Russell 3000 as proxy) decreased by 15 basis points after seeing a slight uptick last week.
Normalized Factor Returns: Axioma Worldwide Equity Risk Model (AXWW4-MH)
Methodology for normalized factor returns
- Mirroring the US, global Volatility continued to slingshot upwards by +0.6 standard deviations, in knocking distance of an Extremely Overbought designation at +1.82 SD above the mean.
- Market Sensitivity crossed over the mean as it headed into positive normalized space, up another +0.33 standard deviations.
- Exchange Rate Sensitivity saw a slight increase that left it on the threshold of Overbought space at +0.97 SD above the mean.
- Earnings Yield appears to be recovering from a recent bottom of -1.71 SD below the mean on 6/4, with a slight uptick that leaves it at -1.65 SD below the mean.
- The free fall in Profitability continued, as it dropped by another 0.85 standard deviations and landed in Oversold territory at -1.22 SD below the mean. If we look at the trailing 12 months for this factor, we can see a) how sharply it’s fallen since late May, and b) that it is nearing it’s 1-year low of -1.5 SD below the mean on 6/24/20.
Profitability: 6/11/20 - 6/11/2021
- Global Risk (using the ACWI as proxy) decreased by 12bps.