The market cap-based weighting methodology of the S&P 500 index has long been the source of the index's popularity as a benchmark among market participants. Indeed, the ability to passively track this index on huge scales with comparatively lower transaction costs to other approaches have attracted tens, if not hundreds, of billions of dollars in assets indexed to this behemoth of a benchmark. Yet for all that money tracking the S&P 500, it contains inherent weaknesses for the scores of investors who don't have multi-billion dollar portfolios and aren't constrained by portfolio scalability like large mutual funds, hedge funds, and other large institutional investors.
One of the main problems of the cap-weighting approach is how top-heavy the index can become when a handful of individual members perform well and begin to dominate the index's return (a la Apple and Exxon Mobil). Another issue for the cap-weighting methodology is the tendency for certain industries to outperform on a market-cap basis over time, thus leading the index to be overly concentrated in one or a few sectors of the total market. Technology stocks dominated the S&P 500 in the late 90′s leading up into the tech bubble and financial stocks were a large chunk of the index during the mid-2000′s leading up the financial crisis in 2008.
By investing in a cap-weighted index (or ETF like SPY), investors are invariably exposing themselves to these single name and industry over-concentration effects and end up owning equity portfolios that can be highly imbalanced, creating a drag on portfolio performance over long periods of time. To demonstrate what I mean, take a look at the chart below which compares the cap-weighted S&P 500 that we all know and love (and which makes and breaks portfolio manager careers) in comparison to two alternative weighting methodologies … equal-weighting and volatility-weighting the same basket of stocks.
Before we get to results, let's go over the equal-weight and volatility-weight approaches first. Simply put, the equal-weight approach is exactly that ... equally weight all 500 index members and re-balance quarterly. For the volatility-weighted index above, I looked at multiple time frames of standard deviation of daily price returns for both the 10 GICS sectors in the S&P 500 index and also for each individual stock in the index. Weights are assigned by an inverse function of price volatility with lower weights given to high volatility stocks and sectors and higher weights to low volatility stocks and sectors. This portfolio from the data above is rebalanced semiannually. It's a multi-tier weighting approach that first targets what the weights will be for each individual sector, and then what each individual security weight will be within that sector's targeted allocation.
Looking at the performance chart above you can see that equal-weighting significantly improves the return profile of the cap-weighted S&P 500. While this represents a risk-adjusted improvement over the S&P 500, the equal-weight comes at the cost of a higher risk profile. The smart money investor could always adjust by taking a slightly unlevered position in the equal-weight (back of the envelope 85% of your target equity risk allocation) to get a similar risk level as the cap-weight with significantly higher expected return (and have that residual 15% available as opportunistic capital).
Despite its improvement upon cap-weighting on a risk-adjusted basis, however, the equal weight index still comes with its own shortfalls. For one you are still exposed to the problem of industry concentration depending on the vagaries of the S&P 500 index committee. The equal weight's industry allocation is essentially a slave to whatever companies the index committee decides should be in the S&P 500. I'd like to think the committee factors in how much one sector is represented in the S&P 500, but the problem is you really have no guarantees. So while the equal-weight was not as heavily allocated to tech stocks compared to the cap-weight during the NASDAQ evisceration from 2000-2002, it did suffer from over-allocation to financial names during the '08 chaos. Additionally, the equal-weight is naturally exposed more heavily to the smaller names in the S&P 500, leading to an allocation that is more tilted to mid-cap style names (explaining in part the plus-sized beta on the equal-weight…see below).
On a returns-based analysis for the Guggenheim S&P 500 Equal Weight ETF (NYSEARCA:RSP) over the last 5 years, you could argue that it's a 50/50 equity portfolio of large-cap and mid-cap stocks. This could be more of a philosophical point simply because the S&P 500 is so large that it's really a mega cap index and most other "large-cap" alternatives statistically will look more like mid-cap simply because the benchmark is so far above "large".
When you start talking about volatility-weighting, the general idea behind it is to create a balanced portfolio as it pertains to risk and return. When you break down the cap-weighted S&P 500, you will discover that the largest 25% of the companies in the index routinely account for more than 75% of the total market cap of the index. This also means that the largest 25% also accounts for at least 75% of the risk/return profile of the index as well. Not exactly what I would consider a balanced equity portfolio when three-quarters of your holdings make up less than one-quarter of your returns. If the concept of equal-weighting is to be agnostic in position sizing, then the core concept of volatility-weighting is to be agnostic in risk assumption.
Take two stocks in the S&P 500 as an example, Apple (NASDAQ:AAPL) and Clorox (NYSE:CLX). Apple's weight in the cap-weighted index runs at roughly 4.50% while Clorox measures a measly 0.06%. If this two-stock portfolio were itself cap-weighted, Apple would comprise 97.4% of the market value of the portfolio and Clorox the other 2.6%. Given that Apple's current 6-month daily price return standard deviation is roughly 3 times larger than Clorox's (37.9% vs. 12.3%), Apple represents well over 99% of the risk/return of the portfolio. If you equal-weight this two-stock portfolio, you obviously introduce more balance. However, the risk-return in this portfolio is still dominated by Apple, with the forbidden fruit representing over 75% of the risk/return of the current portfolio (based on current 6-month price return deviations). By comparison, a volatility-weighted portfolio between these two stocks would allocate 75.4% to Clorox and just 24.6% to Apple. This makes sense when you consider the aforementioned greater risk of holding Apple stock in relation to a position in Clorox.
Below I have supplied three charts that visually demonstrate the daily return impact that Apple and Clorox have had on the three aforementioned weighting schemes for the S&P 500 over the last 12 months. I used an average weight estimate for the twelve-month period for the cap-weighted numbers using the weights mentioned in the above paragraph. For the volatility-weighting the numbers were an average of 0.18% for Apple and 0.38% for Clorox.
It's pretty incredible how little impact that Clorox has on the cap-weighted index when you compare it to Apple's daily impact. It's like Clorox is not even there. And if you were to look at Clorox's price action over the period, you really wish it were "there" for your portfolio. To be fair, using Apple as a comparison is somewhat of an extreme example, if only to highlight the issues of cap weighting. It's incredible the damage that Apple did to the S&P 500 on January 24 of this year, when the stock dropped by more than 12%. Apple single-handedly subtracted an estimated 0.5% from the S&P's return that day. Incredible that one stock did that when you consider the S&P 500 was essentially unchanged that day.
By comparison, equal-weighting does a much more effective job at bringing the portfolio impact between Clorox and Apple into line. However, there is still a noticeable difference between the daily return impacts. From my perspective equal-weighting, while an improvement on cap-weighting, still introduces a bias to riskier stocks in your portfolio by not accounting for positions that carry inherently different levels of risk. Not all stocks are created equal. Therefore, they should not be weighted as if they were.
Volatility-weighting yields the most stable results from the idea of having Apple and Clorox making equivalent contributions to the index's performance. I think the point is obvious that you should expect each member in your portfolio to do roughly the same amount of work. When you buy a cap-weighted portfolio like the S&P 500, you are essentially throwing away the bottom half of the index out without getting any real benefit from owning those names. Equal-weighting does address this problem somewhat, but volatility-weighting really maximizes what you can expect out of each of your holdings.
What I also found interesting is that on a returns analysis, the volatility weighted S&P 500 is more loaded to the S&P 500 itself, reducing the mid-cap exposure inherent to the equal-weighting approach. The return numbers also bear this out as you see the volatility-weighted returns are not only larger, but also carry lower levels of absolute risk. Volatility-weighting is not only superior to cap-weighting and equal-weighting on both an absolute and risk-adjusted basis, it also carries a below-average beta. The second chart below gives a risk-adjusted comparison of the volatility weighted and equal-weighted indexes, as measured by a cumulative rolling 24-month alpha relative to the cap-weighted index. Both methodologies generate pretty good positive alpha against the cap-weighted index, but not only does the volatility-weighted approach produce more alpha, it does so more consistently.
So in conclusion, I think investors who are currently sticking to the traditional cap-weighted approach in their equities really need to consider at a minimum moving out into the equally-weighted products like RSP. The adventurous and those so ambitiously inclined really should look at volatility-weighting methodologies if they really want to maximize their portfolio's return-generating efficiency. There are also fundamental-based indexes and ETF products (such as PRF and PRFZ) which I did not mention in this article from Rob Arnott and the folks at Research Affiliates. These have also generated superior returns relative to the cap-weighting approach and also make good candidates to replace cap-weighted holdings.
And please note that volatility-weighting is not to be confused with low volatility or minimum volatility products. Those are methodologies that SELECT assets by volatility (or seek a statistically optimized portfolio which minimizes volatility), not necessarily WEIGHTING them by volatility. As such, they are heavily overexposed to low beta, high dividend paying stocks that dominate their selection universes. A weighting-based volatility approach will not introduce that type of low beta selection bias.