Investment Themes

Sign up to receive Indexology® Blog email updates

In This List

The Skew Is Not New

Can High Concentrations Lead to Equal-Weight Outperformance?

Decomposing Recent Volatility Events Part 1

Do Earnings Revisions Matter in India?

The Difference a Few Days Make

The Skew Is Not New

Contributor Image
Craig Lazzara

Managing Director and Global Head of Index Investment Strategy

S&P Dow Jones Indices

two

Market observers have noted that the S&P 500’s performance so far this year has been dominated by a small number of technology stocks.  This observation is certainly correct, although it’s fair to question the relevance of a statistic based on fewer than two months’ data.  What’s more important is to bear in mind that this is not unusual.  For most equity indices, skewed returns are the rule, not the exception.

One technical measurement of skewness is that a distribution’s mean is greater than its median.  Unlike in Lake Wobegon, where all the children are above average, in the real world of positively skewed returns, most stocks are below average.  Graphically, there’s a long tail to the right, as pictured here:

The chart covers a 20 year period, but we don’t need long time horizons to detect skewed results.  For the S&P 500, e.g., returns have been positively skewed in 23 of the past 27 years.  Results are similar for other markets.

Why should investors care about skewness?

If skewed returns continue for the balance of 2018, it would be unsurprising to see active underperformance and equal-weight outperformance continue as well.

The posts on this blog are opinions, not advice. Please read our Disclaimers.

Can High Concentrations Lead to Equal-Weight Outperformance?

Contributor Image
Tim Edwards

Managing Director, Index Investment Strategy

S&P Dow Jones Indices

two

Assigning equal weights to each constituent, such as in the S&P 500 Equal Weight Index, historically would have offered material outperformance over capitalization-weighted benchmarks across a range of markets.  Earlier this year, we published a paper examining how this occurred.

There are several perspectives one can take, ranging from factor or sector exposures all the way down to the pattern of contributions from each individual constituent stock.  However, one of the most interesting aspects of equal weight outperformance is that it arises primarily not through the differences between equal and cap-weighted sector exposures, but rather through the consequences of equally weighting within each sector.

So, what is going on?  We suggest that successful competition within each industry provides a mechanism for outperformance within equal weight indices.

Within a competitive industry setting, if a firm secures an advantage and begins to outperform, then three things will likely follow in consequence.  The first is that market capitalization of that firm will likely increase, relative to its peers.  The second is that – supposing such outperformance continues – the related sector index will become more concentrated into the outperforming company.  The third and final consequence – assuming a high level of competitiveness – is that the peers of that company will find ways to catch up with the prior leader, and reverse the prior concentration.

These consequences should be observable through the relative performance of equal weight indices.  If the largest companies in a sector outperform their peers, then a capitalization weighted index will outperform an equal weight index.  If the smaller companies in a sector outperform, then the equal weight index will outperform.

High levels of concentration might therefore provide a potential indicator for the future outperformance of equal weight indices in highly competitive sectors, since unusually high concentration in a sector may indicate significant potential for smaller firms to catch up – either through restrictions (regulatory or otherwise) placed on the larger firms, or organically as those smaller firms act catch up with the leaders.

We test this idea by comparing the relative level of historical concentration in each sector to the subsequent performance of equal weight indices.  In particular, for every month from December 1995 to December 2016, we first calculate the percentile rank in HHI concentration for each of 10 S&P 500 sectors compared to its (monthly) history since 1990.   We then examine the relative total return of the respective equal weight sector index, relative to its capitalization-weight sector counterpart, over the next 12 months.  The chart shows the results, split by quartile of relative concentration:

Concentration Versus Subsequent Equal Weight Outperformance in S&P 500 Sectors, 1995-2017

As the chart shows, on average, the strategy of equally weighting within a sector appeared to have been most attractive when that sector was unusually concentrated.  With the increasing availability of products linked to equal weight sector or market indices, investors might therefore consider “backing competitiveness” by adopting equal weight strategies in sectors or industries that appear to be more concentrated than usual.

The posts on this blog are opinions, not advice. Please read our Disclaimers.

Decomposing Recent Volatility Events Part 1

Contributor Image
Berlinda Liu

Director, Global Research & Design

S&P Dow Jones Indices

two

After a strong January 2018, the U.S. equity market started February with a roller coaster ride. The CBOE Volatility Index® (VIX®), which has been relatively quiet over the past couple years, spiked up and crossed the 50 mark intraday on Feb. 5, 2018. On the same day, the S&P 500® VIX Short Term Futures Inverse Daily Index and its linked exchange-traded products (ETPs) lost more than 90% in value. In light of this significant drop, it is worthwhile to review the mechanics of an inverse volatility index, and how the index value is established. In this two-part blog, we examine factors potentially contributing to the recent spike in volatility and significant negative performance in the S&P 500 VIX Short Term Futures Inverse Daily Index. In this first part of the blog series, we demonstrate that an inverse volatility index is not a “true short” of the underlying reference index.

An Inverse Index Is Not a “True Short”

Before we look into the S&P 500 VIX Short Term Futures Inverse Daily Index, we need to understand that an inverse index is designed to provide daily inverse return to a benchmark. In other words, an inverse index is designed to deliver the opposite of the performance of the benchmark on a daily basis. Its performance over longer periods of time can differ significantly from the “true short” of the underlying benchmark during the same period of time. This effect can be magnified in volatile markets.

Exhibit 1 illustrates the performance of a hypothetical benchmark index and its inverse over a two-day period. As the examples demonstrate, an inverse index that is set up to deliver the inverse of the performance of a benchmark every day will not necessarily achieve that goal over weeks, months, or years due to the compounding effect. If the benchmark index moves around a lot and then ends up in the same place, an inverse index will lose value while a “true short” would not. However, an inverse index does not always underperform. If the benchmark index is trending down, an inverse index can possibly deliver better performance than -1x cumulative performance.

Due to the power of compounding as illustrated in Exhibit 1 and the low volatility regime we have had in the past a couple of years, the S&P 500 VIX Short Term Futures Inverse Daily Index has outperformed the “true short” of its benchmark, the S&P 500 VIX Short-Term Futures Index. In 2017 alone, the S&P VIX Short Term Futures Inverse Daily Index returned 186%, while the S&P VIX Short-Term Futures Index lost 72%. The underlying mechanism for an inverse index (not limited to this example) to achieve this outperformance over a “true short” is to increase its position while it’s rising and decrease its position while it’s dropping. In the case of the S&P VIX Short Term Futures Inverse Daily Index, this would theoretically translate to increasing the number of VIX futures to short, while its benchmark is dropping. Given that each VIX futures contract has a constant vega exposure of 1000, the S&P VIX Short Term Futures Inverse Daily Index has been gradually increasing its vega exposure over the past couple of years. Many market participants may not necessarily be aware of the jump in the vega exposure of the ETPs linked to this index. The size of the index-linked, short-volatility ETP market (which stood around USD 2.7 billion at the peak[1]) may call for even more hedging in light of this increased vega exposure should another VIX jump happen.

In the next blog, we will analyze how the mechanics of a VIX futures index, as well as hedging by market participants, may have contributed to the Feb. 5, 2018, after-hour spike of VIX futures.

[1] Source: Bloomberg.

The posts on this blog are opinions, not advice. Please read our Disclaimers.

Do Earnings Revisions Matter in India?

Contributor Image
Akash Jain

Associate Director, Global Research & Design

S&P BSE Indices

two

Market participants have used common risk factors such as value, momentum, dividends, low volatility, quality, and size (small cap) to construct portfolios historically. Our recently published research paper, “Do Earnings Revisions Matter in Asia?,” explored the performance of earnings revision strategies as an alternative source of return drivers across seven markets in Pan Asia, including Australia, China, Hong Kong, India, Japan, South Korea, and Taiwan. In this article, we will take a closer look at the performance and characteristics of this strategy in India.

The study examined the EPS revision strategies based on two different measures: the first is based on the three-month change in the consensus EPS estimate[1] (EPS change), and the second is the three-month diffusion of the EPS estimates (EPS diffusion.)[2] The study concluded that stock prices tended to move in the same direction as their earnings revisions in the majority of Pan Asian markets. In India, an equal-weighted top quintile portfolio of EPS diffusion and EPS change strategy generated an excess return (against the base universe[3]) of over 8% and 4.5% (CAGR) over the 11 years ending December 2016 (see Exhibit 1).

Exhibit 1: Performance Characteristics of Equal-Weighted Earning Revision Portfolios Versus Base Universe in India
CHARACTERISTIC S&P INDIA BMI BASE UNIVERSE3 EPS DIFFUSION STRATEGY EPS CHANGE STRATEGY
Return (%) 11.9 12.4 20.5 17.0
Risk (%) 30.0 28.8 25.4 27.5
Return/Risk 0.40 0.43 0.81 0.62
Excess Return[4] (%) 0.5 8.1 4.6
Tracking Error (%) 2.5 9.7 8.3
Information Ratio 0.20 0.83 0.55

Source: S&P Dow Jones Indices LLC. All portfolios shown are hypothetical. Performance based on local currency total return from Dec. 31, 2005, to Dec. 31, 2016. Table is provided for illustrative purposes and reflects hypothetical historical performance. Please see the Performance Disclosure at the end of “Do Earnings Revisions Matter in Asia?” for more information regarding the inherent limitations associated with back-tested performance.

Some factor strategies such as value and low volatility witness pronounced sector biases, which could be the dominant driver for factor performance. For most markets in Pan Asia, sector compositions for EPS revision factor portfolios were correlated with the base universe, and therefore one may be able to conclude that sector revision strategies were free from sector allocation bias. Within these markets, however, India had a relatively higher sector bias, with the correlation between sector representation in the factor portfolio and base universe dropping below 70% in the majority of the periods covered in the study. In India, industrials was the most under-represented sector in the EPS revision portfolios compared to the base universe. Although some sector bias was observed in India, it was also observed that sector allocation bias wasn’t a dominant driver of alpha, as can be seen in the performance of sector-neutral stock selection quintiles. For example, the top quintile equal-weighted portfolios of EPS change and EPS diffusion delivered a monthly excess return of 0.31% and 0.52%, respectively, over the base universe. Sector-neutral, stock-selection-based EPS change and EPS diffusion strategies delivered monthly excess returns of 0.30% and 0.47%, respectively, over the base universe.[5]

The study also notes that size bias was not an important driver of excess returns for the earnings revision strategies in India. Over the back-tested period, the average weight of the small-cap factor in the equal-weighted base universe was 58.6%, whereas the factor’s weight in equal-weighted top quintiles of the EPS change strategy and the EPS diffusion strategy was 60.4% and 55.8%, respectively.

In India, market participants tended to penalize stocks with poor earnings revisions more during downward and neutral market periods.[6] In contrast, the portfolio of stocks with poor EPS change outperformed the base universe with significant excess return during upward market trends, indicating that market participants in India tended to disregard earnings downgrades when the market was bullish (see Exhibit 2).

Exhibit 2: Average Monthly Excess Returns of the Earnings Revision Portfolios Versus Their Respective Universes in Different Market Trends (%)
MARKET EPS DIFFUSION STRATEGY EPS CHANGE STRATEGY
TOP QUINTILE BOTTOM QUINTILE TOP QUINTILE BOTTOM QUINTILE
Up -0.09 0.24 0.11 0.75*
Neutral 0.63* -0.69 0.19 -0.96*
Down 1.44** -1.30** 0.67** -2.07**

Source: S&P Dow Jones Indices LLC. All portfolios shown are hypothetical. *Denotes statistical significance at 5% using one-tailed t-test. **Denotes statistical significance at 1% using one-tailed t-test. Figures based on monthly local currency total return from Dec. 31, 2005, to Dec. 31, 2016. Table is provided for illustrative purposes and reflects hypothetical historical performance. Please see the Performance Disclosure at the end of “Do Earnings Revisions Matter in Asia?” for more information regarding the inherent limitations associated with back-tested performance.

A potential challenge to implement this strategy is the relatively high average annualized turnover of 257% for the EPS diffusion top quintile portfolio and 263% for the EPS change top quintile portfolio over the entire back-tested period in India. The turnover for India was at the lower end of the spectrum when compared with other markets studied. A combination of an EPS revision strategy with other fundamental factors may help lower turnover and provide a practical approach to passively implement this strategy.

 

[1]   The three-month EPS change was calculated as the current median EPS estimate minus the prior three-month median EPS estimate, divided by the absolute value of the prior three-month median EPS estimate. EPS estimates are in INR.

[2]   EPS diffusion was computed as the number of upward revisions minus the number of downward revisions of EPS estimates divided by the total number of EPS estimates.

[3]   Constituents from the S&P India BMI with at least three analyst estimates at the time of rebalancing formed the base universe.

[4]   Excess returns and tracking error for the base universe was calculated against the S&P India BMI. Excess returns and tracking error for EPS diffusion and EPS change strategies were calculated against the base universe.

[5]   For a full comparison versus a sector-neutral portfolio, please refer to Exhibit 4 and 9 in the report.

[6]   Market trend was calculated based on S&P India BMI performance, with periods of monthly return greater than 1% representing up markets, periods of monthly return less than -1% representing down markets, and the rest of the periods representing neutral markets.

The posts on this blog are opinions, not advice. Please read our Disclaimers.

The Difference a Few Days Make

Contributor Image
Fei Mei Chan

Director, Index Investment Strategy

S&P Dow Jones Indices

two

For investors, things looked very different between the end of January and the first part of February. Following a few days of market turmoil in February, volatility jumped to levels where it is once again at the forefront of investors’ consciousness. Volatility based on a 252-day lookback generally declined for S&P 500 sectors (Telecom excluded) through January 31 compared to three months prior.

252-Day Volatility Declined Across All S&P 500 Sectors Except Telecom Through January 31, 2018

Fast forward a little more than a week later and we see a very different picture. The same metric through February 9 shows that volatility has increased for all S&P 500 sectors, and by generally similar amounts.

252-Day Volatility Jumped Across All S&P 500 Sectors Through February 9, 2018 

These two charts provide some context around the latest rebalance for the S&P 500® Low Volatility Index, which tracks the 100 least volatile stocks in the S&P 500. Given recent market turbulence, one might be surprised to find that the latest rebalance has not shifted the composition much in terms of sector allocation within the S&P 500 Low Volatility Index. The biggest change (effective at the close today) was a 2% shift to Real Estate. Weight changes in other sectors were negligible.

The index methodology calls for the rebalance to take place on the third Friday of every third month (February, May, August and November); the reference date (the date from which we measure historical volatility) is the last trading day of the month prior. In this case, constituents’ volatility was measured as of January 31, before the recent spike in market volatility occurred. February’s volatility spike, along with whatever else happens between now and April 30, will help drive the next rebalance. It will be interesting to see what the next two and a half months bring.

Latest Rebalance for the S&P 500 Low Volatility Index

The posts on this blog are opinions, not advice. Please read our Disclaimers.