Can We Find Smarter Beta From 2 Factor Portfolios?
Summary Smart beta ETFs offer a rich source of data for factor-based investing. I use ETF focused on low volatility, momentum and quality factors as sources for mining these data. Here, I look at the stocks that share positions in two of the ETFs with an objective of identifying stocks that rank positively for two of the factors. Smarter Beta? Maybe. In the first article in this series ( A Quest for the Smartest Beta ), I dissected three Blackrock iShares smart beta ETFs. Each of these is designed to exploit a single risk-premium factor: low volatility, momentum or quality. The three ETFs are: iShares MSCI USA Minimum Volatility ETF (NYSEARCA: USMV ), iShares MSCI USA Momentum Factor ETF (NYSEARCA: MTUM ) iShares MSCI USA Quality Factor ETF (NYSEARCA: QUAL ) As we saw a portfolio equal-weighting the three ETFs handily beats the broader market as represented by the SPDR S&P 500 Trust ETF ( SPY), and can provide an excellent entry into factor-based investing. In that first article went on to look at the full portfolios for the three funds and determined which of the holdings overlapped more than one fund. Fourteen of a total of 336 equity positions are currently held by all three. I analyzed those 14 as an equal-weighted portfolio I called MQLV. That analysis formed the focus of the first article. This is the Venn diagram showing the full overlap for the three funds’ holdings. (click to enlarge) As we saw previously the MQLV portfolio, comprising the 14 positions shown in the box on the right of the figure, turned in an exceptionally strong performance since its inception in June. The obvious follow-up question is to ask about the two-factor overlaps. How well have they performed and how do they compare to one another? As you can see these are interesting complexes of securities. Let’s begin by consider what is in each of the three clusters. The LVxQ cluster is the largest, containing 32 positions. This is not a complete surprise as quality and low volatility do tend to track together. Quality, as defined for the purposes of QUAL’s index includes three fundamental variables: Return on Equity, Debt to Equity and Earnings Variability. I discussed QUAL and its index in detail previous ( here ). It’s not unreasonable to expect that stocks from companies that rank highly for those metrics would also exhibit lower volatility. The 32 LVxQ stocks are: Apple Inc (NASDAQ: AAPL ), Ace Ltd (NYSE: ACE ), Automatic Data Processing Inc (NASDAQ: ADP ), Berkshire Hathaway Inc Class B (BRKB), Costco Wholesale Corp (NASDAQ: COST ), Campbell Soup (NYSE: CPB ), Chevron Corp (NYSE: CVX ), Dollar Tree Inc (NASDAQ: DLTR ), Henry Schein Inc (NASDAQ: HSIC ), Hershey Foods (NYSE: HSY ), Intuit Inc (NASDAQ: INTU ), Gartner Inc. (NYSE: IT ), Johnson & Johnson (NYSE: JNJ ), Lockheed Martin Corp (NYSE: LMT ), Mastercard Inc Class A (NYSE: MA ), Mcdonalds Corp (NYSE: MCD ), Marsh & Mclennan Inc (NYSE: MMC ), 3M Co (NYSE: MMM ), Monsanto (NYSE: MON ), Microsoft Corp (NASDAQ: MSFT ), Paychex Inc (NASDAQ: PAYX ), Public Storage Reit (NYSE: PSA ), Qualcomm Inc (NASDAQ: QCOM ), Ross Stores Inc (NASDAQ: ROST ), Sherwin Williams (NYSE: SHW ), At&T Inc (NYSE: T ), l TJX Inc (NYSE: TJX ), Travelers Companies Inc (NYSE: TRV ), Varian Medical Systems Inc (NYSE: VAR ), VF Corp (NYSE: VFC ), Exxon Mobil Corp (NYSE: XOM ), and Yum Brands Inc (NYSE: YUM ). Sector allocations are led by Information Technology, Consumer Discretionary and Financials. (click to enlarge) The MxLV cluster holds 21 positions. These are: Allergan (NYSE: AGN ), Autozone Inc (NYSE: AZO ), C R Bard Inc (NYSE: BCR ), Church And Dwight Inc (NYSE: CHD ), Dollar General Corp (NYSE: DG ), Ebay Inc (NASDAQ: EBAY ), Facebook Class A Inc (NASDAQ: FB ), Fiserv Inc (NASDAQ: FISV ), General Mills Inc (NYSE: GIS ), Alphabet Inc Class C (NASDAQ: GOOG ), Alphabet Inc Class A (NASDAQ: GOOGL ), Mondelez International Inc Class A (NASDAQ: MDLZ ), Mccormick & Co Non-Voting Inc (NYSE: MKC ), Partnerre Ltd (NYSE: PRE ), Synopsys Inc (NASDAQ: SNPS ), Stericycle Inc (NASDAQ: SRCL ), Target Corp (NYSE: TGT ), UDR Inc. (NYSE: UDR ), Unitedhealth Group Inc (NYSE: UNH ), Vantiv Inc Class A (NYSE: VNTV ), Water Corp Corp (NYSE: WAT ). Sector allocations are led by Information Technology and Consumer Staples. (click to enlarge) The QxM cluster with seven positions is the smallest. I find it interesting that momentum correlates poorly with quality using the definitions of these ETFs. With the 14 stocks included in the 3-ETF overlap cluster, there are only 21 stocks that meet the index criteria for both quality and momentum. The seven stocks in this cluster are: Assurant Inc (NYSE: AIZ ), Brown Forman Corp Class B (NYSE: BF.B ), CDK Global Inc (NASDAQ: CDK ), Edwards Lifesciences Corp (NYSE: EW ), Progressive Corp (NYSE: PGR ), SEI Investments (NASDAQ: SEIC ), Torchmark Corp (NYSE: TMK ). More than half (4 of 7 positions) of the sector allocation for this cluster is to financials. Consider that financials was a dominant sector in the MQLV cluster as well, where it accounts for four of the 14 positions. (click to enlarge) Portfolio Performances What happens when we try to create portfolios from each of the 3 clusters? Ideally, we’d have the data to track changes as the ETFs indexes rebalanced. But I’m unaware of any publicly available sources for historical portfolio compositions for either the ETFs or the Indexes. So we’re restricted to current holdings. Each of the three indexes are rebalanced semi-annually at the end of May and the end of November. The current clusters have been in place since the last rebalancings implemented on June 1. What I’ll do is compare how equal weighted portfolios for each of the clusters compare in performance and risk metrics since June 1. My plan is to come back to this at the end of this month and see how the portfolios have changed. My expectation is that USMV will have changed the least, closely followed by QUAL. MTUM will have changed the most; such is the nature of momentum-it’s transient. I’ve used Portfolio Analyzer to track portfolio performances for the 14 positions in each of the 3 ETFs and SPY for reference standards. The results are quite interesting. (click to enlarge) MQLV is the clear standout here. It is followed by QxM and MxLV. The third two-factor cluster (LVxQ) underperforms everything but SPY. This tends to suggest that momentum was the key factor for this five-month period. But, let’s look at the ETFs. Each beats SPY but none stands out as having been exceptionally better than the other two. QUAL is the best performer of the three but only by a slim margin, and USMV is the worst, but again only by a slim margin. The previous indication that momentum was the key to performance over this time span is not borne out by the full portfolio performance records. I suspect an important driver for these results is the size of the portfolios. The smaller portfolios are more highly selected for the factors under consideration. The 32 position LVxQ portfolio comprises some excellent holdings, many of which I have in my own portfolio. But a critical look at that list makes clear that this is not a group of stocks one would target for short-term outperformance. I don’t own the ones I do for that purpose and I doubt many do. Another driver is the stability of the portfolios. I expect that LVxQ will be the most stable of the four. As I said, momentum is transient and it is the momentum factor that is going to most strongly affect how the various models change at rebalancing. Obviously, what we have here is a single data point. It is impossible to draw any conclusions from these results. But the fact remains that they are intriguing and suggest that this approach may have merit in pulling out attractive opportunities for stock picking on a semi-annual basis. Investors with longer term perspective can use the LVxQ cluster as a resource for portfolio constructions. Those more willing to trade regularly may be more attracted to the MQLV group, but they should be prepared to rebalance, perhaps extensively, at 6 month intervals. I will certainly be interesting to see what the month-end restructuring of the indexes brings. I’ll be on it and I’ll try to get a report out here as quickly as I can get it done. Before closing it should add that there are many other ETFs one can choose from to exploit the various risk-premia factors that have been identified. I’ve selected these three because I’m familiar with them (I hold all three), I considered that their approaches complemented rather than duplicated one another, and because I’ve found that iShares and MCSI, the index provider for these funds, tends to provide accessible and transparent data for my research. It also helps that they all have the same sources because starting with the data all in the same format makes for much more efficient use of my time. As readers commented, I’ve not included two of the best-documented factors: value and size. This was an intentional choice. Size was excluded because it made more sense to me to restrict myself to large- to mid-caps. That was an easy call. Excluding value was less obvious. I wanted to limit the analysis to three funds which I think is the sweet spot for this sort of thing. More than three gets unwieldy. I felt these three factors — low volatility, momentum and quality — had minimal overlap but two of the three had some overlap with value. I also felt adding value as a factor would have de-emphasized momentum to a greater extent than I wanted. I have no real evidence for this point of view, but it made intuitive sense to me. As it happens, one value factor counterpart of these funds, the iShares MSCI USA Value Factor ETF (NYSEARCA: VLUE ), has 21 positions in common with MTUM, as many as USMV. Regardless, I did not want to replace either QUAL or USMV with VLUE. Might be grist for another go-round however.