S&P 500: Market Cap Vs. Equal Weighting

|
Includes: RSP, SPY
by: Barron Gati

Summary

At first glance, the equally weighted S&P 500 index might appear to have outperformed the market capitalization weighted S&P 500 index over the past 15 years.

Another article on Seeking Alpha used annual returns and index values instead of daily returns and ETF values.

After correcting these two aspects, the outperformance nearly completely disappears.

Here’s a quick look at how one’s approach to data analysis can yield significantly different results. A broker friend of mine approached me with an idea for an article. He was interested in explaining why the Guggenheim S&P 500 Equal Weight ETF (NYSEARCA:RSP) outperformed the SPDR S&P 500 Trust ETF (NYSEARCA:SPY). He’d invested in that and suggested to his clients that they do likewise, and, as it turns out, RSP has outperformed SPY.

Indeed, upon researching an answer to this question, which eventually led to this article’s construction, I found this post by somebody on Seeking Alpha: Equal Weighted Over Value Weighted: RSP Vs. SPY

In it, the author takes yearly data looking at the two indices (S&P 500 equal and market cap weightings), not the ETFs, and concludes that there is a large difference in performance when comparing the two investments.

At first glance, the analysis might appear to be straightforward, especially since the author notes the most obvious difference between market cap and equal weighting: that riskier stocks are more heavily weighted in an equal weighting scheme than they would be when weighted by market cap (i.e. a smaller company that represents only 0.04% of the total market cap would get a 0.20% weight in an equal weighting scheme of the S&P 500). As a result, we do expect that the riskier allocation (equal weighting) will return more than the less risky allocation (market cap weighting). But, the author then corrects for risk-adjusted return and concludes that the Sharpe Ratio of RSP was more than 50% higher than that of SPY (0.21 vs. 0.13, respectively).

If you look closer, however, you’ll see that the article’s findings were built on a suboptimal approach to the data analysis. First, I noticed that the author used the indices rather than the ETFs. So I tested the SPY vs. the S&P 500 over the relevant time frame (1/29/1993–10/20/2017) and it's important to note that the raw index doesn't include dividends. So while there is little tracking error for the dividend-corrected ETF and the S&P 500 index, if the author didn't use the non-dividend-corrected values for both then it would be an inconsistent analysis.

Unfortunately, the author didn't note which was used, but to provide an overview of how large the difference could be, Here’s the annualized return, standard deviation, and post expense Sharpe Ratio for the raw SPY vs. the dividend-included SPY, and the raw S&P 500 index (data sourced from Yahoo finance and are from 1/29/1993).

Now, in terms of why the author chose to use the index instead of the ETF, that was most likely because the ETF RSP didn’t exist back in 1993. That made it impossible to compare indices; however, when you’re looking at gauging real world returns, it’s best to use the closest approximation to what an investor might earn. And to do that, you want to compare ETFs to ETFs.

In addition to not specifying exactly what was done with dividends, the author also only used annual returns. Those are fine when trying to ascertain general performance, but when comparing the specifics of two return streams, it’s best to use as granular a price series as possible.

To see what the difference would have been, I used the daily return stream of RSP and SPY since 5/1/2003 (when RSP was launched) and the outperformance all but disappeared. Here’s the same results for RSP, SPY, and 10 year treasuries (I used modified duration to match price moves based on constant maturity yield changes on daily yield data from the Federal Reserve Board):

Putting these data on the Capital Market Line, we see that there’s a very small remaining unexplained difference between SPY and RSP:

So overall, we see that once we take a closer look at the data, the original article’s claim that the equally weighted S&P 500 significantly outperformed the market capitalization weighted S&P 500 was the result of suboptimal data analysis.

In reality, while there still remains a small difference in risk-adjusted return performance between RSP and SPY, that difference can most likely be explained by chance, or through the RSP's rebalancing since the SPY doesn't need to rebalance.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article.