Indexology^® Blog

Factor-based strategies have been regularly used by market participants in their portfolio construction process. Apart from the established factors like value, size, volatility, etc., research on alternative factors has remained important to explain sources of alpha. One such alternative factor is consensus analysts’ earnings forecasts. Ample empirical research exists that explains the market’s reaction to analysts’ earnings forecasts. Most of the research focuses on the change in the consensus estimate or the number of upgrades or downgrades in the estimates over short-term periods.

Compared with the U.S. and the European markets, the Asian market is more distributed and each individual market in Asia has its own characteristics. In our research paper Do Earnings Revisions Matter in Asia, we tested earnings revision strategies across seven Pan Asian markets—Australia, China, Hong Kong, India, Japan, South Korea, and Taiwan. We examined the three-month change in the consensus estimate and the three-month diffusion of analysts’ earnings forecasts from Dec. 31, 2005, to Dec. 31, 2016.

The key findings were as follows.

1. Stock prices tended to move in the same direction as their earnings revisions in the majority of Pan Asian markets. Earnings revision strategies delivered the most significant excess returns in South Korea, India, and Taiwan, but they did not work in Japan (see Exhibit 1).
2. Market participants generally had stronger reactions to the net percentage of upward and downward revisions in earnings estimates rather than the percentage change of the consensus estimate figures.
3. Companies with poor earnings revisions tended to have more volatile returns, as market participants reacted negatively to companies with downward revisions in estimates.
4. Earnings revision strategies tended to generate more alpha in the small-cap universe than in the large-mid-cap universe, although there was no strong sector or size bias.

Like other momentum strategies, the earnings revision strategies also had high turnover. In addition, the market-cap effect disrupted the signal from the earnings revision strategies. Therefore, implementing this strategy in combination with other fundamental factors and alternative weighting schemes would be essential to capture alpha. To find out more, please see the full report.

There may be a legitimate concern about rising inflation since inflation has been under the Fed’s target for years and there is now very low unemployment with signs of accelerating economic growth.  The low unemployment rate and strong growth are probably positive for the stock market and the tax cuts may provide even more upside.  However, if inflation upticks and the Fed raises rates to slow it, the increased cost of capital and likely drop in consumer spending may put pressure on stocks.

Understanding more about how to position equities in the event of rising inflation may be useful in this uncertain time.  In conjunction with the impacts on stocks from accelerating growth and the falling dollar, a more complete picture of investment strategy inside an equity allocation can be formed.  Small-caps benefit most from accelerating growth, mid-caps do best from a falling dollar, and what will be shown in this analysis is that mid-caps are most sensitive to inflation, and energy is by far the most sensitive sector.  Also what is interesting is that though crude oil itself is more sensitive to inflation than the energy equities, once the roll yield of the futures contracts is considered, all the benefit is lost.

Let’s start with the 10 year correlation using monthly year-over-year data of index levels and CPI (Consumer Price Index.)  Overall, equities are not that correlated with inflation at just over 0.2.  However, the energy sector correlation to inflation of 0.6 is triple the correlation of the broad equities to inflation.  The crude oil futures are even more correlated to inflation, measuring over 0.8, that is considered highly correlated.  This is because energy is the most volatile component of CPI, essentially driving it.

Another common measure used to evaluate the sensitivity of assets to inflation is called inflation beta.  It means, “if there was a 1% change in CPI, then there was an x% change in the index.”  For example, in the chart below, for every 1% rise in inflation there was a 17% increase on average for small-cap energy stocks.  The S&P MidCap 400 is the most sensitive broad-based index, with an inflation beta of 3.3, only slightly higher than the 3.1 inflation beta of the S&P 600, but markedly higher than the 2.6 inflation beta of the S&P 500.  Again, the crude oil futures have an even higher inflation beta than the equities at about 20.  Since the small-caps are least likely to hedge, they may benefit most with rising oil and inflation.

One might conclude that for inflation protection, crude oil is the best bet.  Though the problem with getting inflation protection with crude oil futures is that market participants need to pay storage costs, reflected through the roll yield when there is excess inventory.  In a decade plagued with high inventory, this has cost crude oil futures investors an additional 48% beyond the -37% lost in the spot market.  The result has been a negative up market capture ratio of -109.7, meaning for every 1% rise in inflation, the S&P GSCI Crude Oil Excess Return actually fell by 1.1% on average in the past 10 years.

Given equities are better than futures for inflation protection, after the roll yield is included, and that small-caps and mid-caps have done best with accelerating growth, the falling dollar and rising inflation, perhaps the strategy might be to underweight large caps.

The investment community routinely uses historical realized volatility as an indicator of future volatility. One prominent example of this is the construction of ranking-based, low volatility equity strategies where realized volatility is used to form low volatility portfolios. Can the realized volatility of an asset indicate its future volatility? We extend the low volatility analysis to U.S. preferred stocks to see if the low volatility effect is present in other asset classes. For our analysis, the S&P U.S. Preferred Stock Index is used to represent the investment universe of U.S. preferred stocks.

To construct a low volatility factor portfolio, it is common to select securities that had low realized volatility over a pre-specified period and hold the portfolio for the subsequent n months. Accordingly, in our analysis, we measure the relationship between the realized volatility over the previous one-year period and that of the subsequent three months across all preferred securities in the universe for each quarterly rebalance date.

Exhibit 1 shows the volatility for the next three months versus the realized volatility for the past year in scatter plots. The top plot shows a positive relationship between these two sets of realized volatility. Moreover, the relationship seems to be exponentially related, rather than linearly. The bottom graph plots the two sets of volatility in natural logarithm terms and shows a more linear relationship than the top plot.

In Exhibit 2, we show the rolling correlation, where at each time point the correlation is calculated between the realized volatility over the past year and the realized volatility over the next three months for all the preferred stocks in the universe as of each quarterly rebalance date. Since low volatility factor portfolios typically select securities using volatility rank, we also calculated the correlation of volatility ranks based on these two sets of realized volatility.

The top panel shows that the correlation of the logarithmic volatilities seems to be more stable than that of the volatilities themselves for each point in time. The bottom panel shows that the correlation of volatility rank and the logarithmic volatility rank are identical. Since October 2003, the average correlation of the rank of the two sets of volatility is 0.66.

The analysis shows that, on average, a preferred stock that ranks low or high based on its realized volatility over the past year is likely to continue to rank low or high for the volatility exhibited in the next three months. This lends support to arguments for using realized volatility to construct a low volatility factor portfolio for preferred stocks.

Until 2013, the exchange-traded fund (ETF) industry in India was in a nascent stage, with negligible assets under manager (AUM). As of Dec. 31, 2013, the total AUM for ETFs was INR 8,000 crores (or USD 1.2 billion), out of which commodities-based ETFs tracking gold noted the largest share, with total AUM of INR 6,500 crores (USD 1 billion). Starting with such a small base, the ETF industry noted exponential growth over the past four years, primarily driven by a decision from the Employee Provident Fund Organization (EPFO, one of the largest pension houses in India) to start investing in equities-based ETFs, including those tracking the S&P BSE SENSEX, as well as the Government of India’s decision to disinvest its holdings via ETFs.

As of Dec 31, 2017, the total AUM of the ETF industry stood at INR 78,000 crores (USD 12 billion), with an annualized growth rate of 76.6% during the past four years. The industry saw four new product launches in 2017, including the Bharat 22 ETF, which tracks the S&P BSE Bharat 22 Index and had the largest (by AUM) single ETF launch ever worldwide. This ETF was part of the disinvestment of listed government-owned companies in India.

Exhibit 1: Growth of ETF AUM – India 

Among the three asset classes, equities noted the highest increase, with an annualized growth of 213% over the past four years. The exponential growth was also supported by a sustained bull run in Indian and global markets, with the S&P BSE SENSEX, S&P BSE MidCap Select, and S&P BSE SmallCap Select increasing 29.56%, 53.23%, and 51.13%, respectively over calendar year 2017.

In December 2015, S&P Dow Jones Indices launched the S&P Real Assets Index, the first index of its kind, which is designed to measure global property, infrastructure, commodities, and inflation-linked bonds, using liquid and investable component indices that track public equities, fixed income, and futures.

The S&P Real Assets Index includes global infrastructure (35%), property (25%), natural resources (35%), and inflation-linked bonds (5%), using stocks (50%), bonds (40%), and futures (10%). It is constructed as an index of indices, using the components and weights shown in Exhibit 1.

Why Use Real Assets?

The two main reasons market participants might consider using real assets are diversification and inflation protection. The S&P Real Assets Index design incorporates both equities and fixed income to more fully represent companies, and it adds commodity futures for more direct exposure to natural resources. The result is that market participants may achieve more diversification and inflation protection than with just equities or with any single asset inside the index.

As shown in Exhibit 2, the S&P Real Assets Index has provided relatively strong inflation protection, with an inflation beta of 4.46, as measured by monthly and year-over-year returns of the index and the CPI, compared with 2.4 for the S&P 500^® and the negligible inflation protection of the S&P U.S. Aggregate Bond Index. Inflation beta can be interpreted as a 1% increase in inflation resulting in a 4.46% increase in the return of the S&P Real Assets Index.

Real assets are moderately correlated to each other, since their underlying characteristics are alike in many ways, and, while they have a strong correlation to the S&P 500 (83.0%), they have a more modest correlation to the S&P U.S. Aggregate Bond Index (28.1%). In addition, including real assets in a portfolio offers access to a larger spectrum of assets.

Exhibit 3 depicts two hypothetical portfolios—one with an allocation of 60% to equity and 40% to bonds, and the second with a 50/40/10 allocation to stocks, bonds, and real assets, respectively.

While the annualized return for the one-year period dropped slightly from 14.0% for the equity/bond portfolio to 12.9% for the portfolio that includes real assets, the annualized risk declined as well. The numbers are not mediocre, considering the historical drawdown natural resources have suffered over the past 10 years and the rally of U.S. equities in 2017. It is important to note that analysis up until the end of 2015 showed that the Sharpe ratio increased from 0.47 for the equities benchmark to 0.68 for the equity/bond portfolio and to 0.7 for the portfolio that included real assets.