In this series of posts, we’ve seen several reasons why factor returns found in academia might not translate into the real-world performance of smart beta funds. To bring the series together, I’ve provided a few examples below of how the various topics we’ve discussed have impacted the smart beta landscape, and what this means for investors.
Lettau, Ludvigson and Manoel
The first, and probably best known, paper on the subject of factor returns and smart beta comes from US researchers Martin Lettau, Sydney C. Ludvigson, and Paulo Manoel in their paper “Characteristics of Mutual Fund Portfolios: Where Are the Value Funds?”.
To summarise their findings, they concluded that:
- Funds do not systematically tilt their portfolios toward profitable factors, such as high book-to-market (BM) ratios, high momentum, small size, high profitability and low investment growth.
- There are virtually no high BM funds in the sample, while there are many low BM “growth” funds. For example, only seven out of 2,657 funds in their sample have a BM score in the fourth quintile or above (while 18% of S&P 500 stocks have a BM score above 4). The bulk of “value” funds (mutual funds, ETFs and hedge funds) have BM scores between 2 and 3.5. Further, value funds hold a larger portion of their portfolio in stocks in the lowest BM quintile (24%) than in stocks in the highest BM quintile (13%).
- Portfolios of “growth” funds are concentrated in low BM stocks, but “value” funds hold stocks across the entire BM spectrum—more than half of all “value” funds hold a larger share of low BM stocks than high BM stocks, and only 7% hold more than 25% of their portfolio in high BM stocks.
The researchers’ findings were not merely a function of managers avoiding the price-to-book ratio as a proxy for value (the drawbacks of which I wrote about in this post). They reached a nearly identical conclusion when using a number of alternative criteria of value, such as price-to-earnings ratios, price-to-cash-flow ratios and dividend yield.
Robeco Asset Management
Research from David Blitz of Robeco Asset Management, the Dutch asset management firm, has similarly shown that factor ETFs don’t always have high exposure to the ‘traditional’ definition of the factors. The chart below shows the author’s universe of ETFs, and how much exposure the ETFs actually have to the traditional “HML” definition of value (as defined by Fama French):
Source: David Blitz
It’s pretty clear from the chart that the vast majority of ETFs classified as “value” do not have high loadings to the HML factor. In fact, most of the ETFs have a factor loading of 0.20 or less, and a surprising number of ETFs actually have negative value exposure.
Morningstar
Using a universe of 18 index-tracking funds within the large-value Morningstar Category, Daniel Sotiroff of Morningstar ran a four-factor regression using data from Ken French’s data library to find out how many of these value funds actually have high loadings to the traditional value factor, and how they had performed from March 2009 to June 2018. Those with the largest loadings to HML were classified as ‘deep value’, and those without were classified as ‘mild value’.
Factor loadings
Source: Morningstar
Deep value vs mild value – performance
Source: Morningstar
The results are interesting for a few reasons:
- All funds studied called themselves “value” funds, yet the HML exposure between funds ranged from 0.19 to 0.66. The name of the fund alone gave no indication of how much value factor exposure the investor was getting.
- Similarly, the drawdowns of the funds ranged from -43.25% to -70.3%. Funds that appeared to do similar things on the face of it produced significantly different performances.
- Deep value funds outperformed mild value funds over the time period – no doubt a result of the funds overweighting cheap stocks in the aftermath of the financial crisis and benefitting from the huge subsequent rally. But this outperformance came at a cost – both standard deviations and maximum drawdowns were higher for the deep value funds. No pain, no premium.
Factor Research
Nicolas Rabener of Factor Research is a prolific blogger on the topic of factor investing and smart beta, and has unsurprisingly produced a number of excellent articles on the topic.
The chart below shows that the average factor exposure of a smart beta ETF is relatively low:
Source: Factor Research
Looking specifically at “value” smart beta ETFs, the research shows that there’s a big divergence between those value ETFs with top quartile value factor exposure, and those with bottom quartile exposure. In addition, those with the largest value factor exposure also have a considerable loading to the size factor – which is explained by many products focusing on value in small cap indices like the Russell 2000:
Source: Factor Research
In addition, the correlations of the largest smart beta ETFs are far more correlated to the S&P 500 than the factors themselves:
Source: Factor Research
The resulting performance of smart beta ETFs in the past 18 years has been nowhere near the performance of factor returns:
Source: Factor Research
Honing in on the value factor and comparing it against the largest “value” smart beta ETFs, the value factor (in black) has significantly outperformed the funds attempting to capture it:
Source: Factor Research
Conclusion
Hopefully it’s pretty clear by now that returns from smart beta products aren’t likely to produce the same returns as the factor returns they’re trying to capture. The examples above show that smart beta products often produce returns that fall well short of what investors might expect if they’d been hoping for similar returns to what factors had produced in academia.
What the examples also do a good job in showing is that you can’t rely on a fund’s name to determine whether or not it captures the factor exposure you might be expecting. There’s a huge amount of difference between funds with similar-sounding names, and understanding how a portfolio is constructed, and the exposure that it provides, is an incredibly important exercise to do before investing.
Interestingly for those who are willing to incur the higher tracking error and wish to capture “purer” factor exposure by selecting funds with higher factor loadings, the research has shown that because most ETFs don’t really capture factor exposure, actual factor exposure is less at risk of being arbitraged away.
Returning to the very first article discussed in this post, the authors Lettau, Ludvigson and Manoel concluded, “Our results suggest that active mutual funds do not systematically hold the stocks with characteristics associated with high returns and thus are unlikely to contribute to any shrinking of factor premia during the sample period.”
As a more general point, despite their difficulties, smart beta products remain a useful addition to the investment universe as they’re forcing active fund managers to reduce fees. More and more people are realising that what was previously though to be a fund manager’s alpha, is in fact no more than factor exposure – which can be bought for basis points. Costs have a significant impact on returns, especially over the long term, and from that perspective, smart beta ETFs are continuing to benefit investors.