active passive and smart beta part 4 systematic factors and risk premium
This is Part 4 of the 5-part series on Active, Passive and Smart Beta strategies.
In Part 3 (From Assets to Factors), we discussed multi-factor models beyond CAPM and the notion of moving from assets to risk factors.
We also introduced “systematic” factors that cannot be diversified or arbitraged away and hence must offer a positive risk premium in the long run.
Dynamic systematic factors are the cornerstone of Smart Beta strategies (as we will see in Part 5). In this blog, we will discuss some well-known dynamic systematic factors and try to answer the following questions:
Some well-know dynamic factors that we will analyse in this blog are:
While there are many other factors in both academic literature and the industry, we will restrict our focus to the above-mentioned 4 factors.
Note that we have so far described factors as “phenomena”. But for the purpose of analysis and investment, we must describe these factors as an investment portfolio. There are 2 approaches:
1. Market neutral long-short portfolios
These are factor portfolios that are designed to be market neutral by taking long-short positions. This style was made famous by the seminal Fama and French Model (1993)[4].
The idea is to create a portfolio that is not impacted by the overall movement of the broader market (market neutral). Hence, the returns are purely driven by the factor.
An example of such a portfolio is the HML portfolio of Fama and French Model which goes long on top-30% of stocks with highest book-to-price and goes short on bottom-30% book-to-price stocks. Similarly, they have constructed portfolios for size (SMB – Small minus Big) and momentum (WML – Winners minus Losers).
These factor portfolios are “pure” factor portfolios since they extract out any impact of the market returns and are purely driven by the underlying factor dynamics.
2. Long-only portfolio
A lot of investors face long only constraint and hence cannot replicate the market-neutral factor portfolios. Long-only factor portfolio is constructed by just going long on stocks that have higher factor scores. The exact construction will depend on the practitioner.
A long-only momentum portfolio can be constructed by going long on top-30% of stocks with highest momentum with either equal weights to each stock or market-cap weights. For example, Nifty 200 Momentum 30 Index is a portfolio of top-30 stocks (from the universe of top-200 stocks by market cap) based on a certain momentum score. These stocks are then weighted based on a combination of the momentum score (signal strength) and market capitalization.
For Nifty 200 Momentum 30 Index, momentum scores (pure returns) are normalized by 1-year realized volatility of individual components. Since, final weights are a combination of momentum score and market-cap, the index also has a positive tilt to the “low-vol” factor and a negative tilt to the “size” factor.
We note that long-only factor portfolios are not “pure” factor portfolios because their returns are driven by the underlying factor dynamics as well as the overall movement of the broader market. Buying Nifty 200 Momentum 30 Index gives you exposure to the market risk as well as the factor risk.
Now let’s analyse these factors using 2 parameters – historical risk premium and future risk premium.
In this section, we will study the historical risk premium of factors mentioned above. We will first look at the US market and then analyse the Indian market.
For the US market, we use data from Ken French Data Library. Note that this gives us access to market-neutral factor portfolios (refer to previous section) and hence we can extract pure factor premiums.
For India, we do not have access to pure factor portfolios but we do have long-only indices. We rely on CAPM regression to analyse the premium.
Here we analyse last 50 years of data (from January 1972) and look at the following factors:
The table below summarizes the return characteristics of these 4 factors:
As we can see, all these 4 factors have earned a positive risk premium with Market and Momentum leading the pack (albeit with much higher volatility). Size has the lowest premium. Value, although has lower volatility, experienced the worst drawdown of 58.5%.
Looking at these numbers, one might ask – why bother with Factors at all? Why not just invest in the market (passive indexing) and simply earn the market risk premium. One can do that and a lot of people actually do it (passive index investors). But it is not optimal.
The idea is to find factors that exhibit a very low correlation with the market and, at the same time, earn a positive premium. This enables you to earn a risk premium which is over and above the market risk premium (adjusted for your market beta). Finding such systematic factors allows you to optimally allocate factor exposures.
This is what we discuss next. We show how HML and WML portfolios exhibit negative correlation with the market and deliver a statistically significant alpha. Size on the other hand exhibits positive correlation with the market and the alpha is statistically insignificant.
Note that although these factors have been constructed to be market neutral, they do have some exposure (even negative) to the market. It would be interesting to check the “alpha” by controlling for the market exposure. Therefore, we run a CAPM regression for each factor on the market factor (for more details on this technique, refer Part 2 of this series - Active vs Passive Investing).
Table below summarizes the results of CAPM regression:
Both Value (HML) and Momentum (WML) delivered statistically significant alpha while Size (SMB) alpha was statistically insignificant. More importantly, HML and WML actually had negative exposure to the market (thus magnifying the CAPM alpha).
Finally, we look at correlation of returns over the last 50 years
As expected, both HML and WML were negatively correlated with the market while SMB exhibited a positive correlation. Also, factors within themselves show zero to negative correlation.
For Indian market, we rely on long-only Nifty indices (we are in the process of constructing our in-house market neutral factor replicating portfolios).
We consider the following Factors:
First, we plot the growth of Rs 1,000 invested on 1st Jan 2009.
Next, we show a table of summary statistics:
As we can see, barring the size factor (Nifty Smallcap 50), all other factors have significantly outperformed the market with Momentum being a standout outperformer.
But remember, these are not market-neutral factor portfolios. Therefore, in order to extract the real factor premium, we need to adjust for the market. To do this, we run a CAPM regression for each index and use liquid fund returns as a proxy for risk free rate.
Table below summarizes the results:
Barring the small cap index, all other factor indices have delivered statistically significant alpha. This is consistent with our analysis for the US markets and can be summarized as follows:
Our finding on “size” premium (or lack of it) is consistent with the academic literature. Size premium was discovered in 1981, but since then, empirical studies (across markets) have shown no significant size premium. Fama and French, who actually created the SMB portfolio in 1993 found no evidence of size premium in 2012 [5].
So far, we have restricted our analysis to historical risk premium – risk premium delivered by the factors historically. As investors, what we are really interested in is the future risk premium. Specifically, we are interested in knowing whether these factors will continue to earn a positive risk premium in the future.
This is same as asking the following question – are these factors systematic i.e. cannot be diversified or arbitraged away? In simpler words, will the risk premium persist even when more and more investors start investing in these factors? We have already seen evidence of “size” premium vanishing since 1980s. Will the same happen to other factors?
Answer to this cannot be just empirical evidence (although it is a starting point). Rather, the risk premium must be justified by some fundamental, rational or behavioural reason (or a combination of them).
Factors that we have discussed above – Value, Momentum and Low Volatility are backed by strong academic research. Multiple research papers have demonstrated the underlying rational or behavioural logic that justifies the risk premium. We will not go into the details of these research but interested readers can refer to this book by Andrew Ang - Asset Management – A systematic approach to Factor investing.
Systematic factor risk premium is the genesis of Smart Beta strategies. In the next and final part of this series, we will introduce Smart Beta strategies and explore multiple flavours of Smart Beta.
[1] Rolf W. Banz, The relationship between return and market value of common stocks,
Journal of Financial Economics (1981)
[2] N Jegadeesh and S Titman. “Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency.” The Journal of Finance Vol. 48, No. 1, 1993, Pages 65-91
[3] The Cross-Section of Volatility and Expected Returns. Ang et al. (2006)
[4] Eugene F. Fama and Kenneth R. French. "Multifactor Explanations of Asset Pricing Anomalies." The Journal of Finance, Volume 51, No. 1, 1996.
[5] Fama, Eugene F. and French, Kenneth R., Size, Value, and Momentum in International Stock Returns (September 2012). Journal of Financial Economics