Index Construction Matters

3) Process--We gauge whether a fund's strategy is prudent, sensible, and repeatable.

4) Performance--We analyze funds' historical performance on an absolute basis, a risk-adjusted basis, relative to peers, and across a variety of market conditions.

5) Price--Fees matter. They are one of the only reliable predictors of future relative performance. As such, they are a vital input into our analysis.

We use this same approach globally, across all kinds of funds (traditional mutual funds, exchange-traded funds, closed-end funds, and so on) and all types of strategies (active, passive, and in-between). Of course, we place more emphasis on some pillars than others depending on the type of fund in question. For example, we will weigh the People Pillar more heavily in the case of a concentrated actively managed stock fund than we will for a total stock market index fund. In other cases, the way we define and dive into a particular pillar may vary. Process is a perfect case in point. Active managers' processes may vary quite a bit. Some may be bottom-up stock-pickers, others may take a top-down view. Some may rely on the skill of a lone portfolio manager, while others may take a team approach. In the case of index funds and ETFs, we take a different line when it comes to defining and analyzing a fund's process. In conducting our due diligence on these funds, we look to answer two key questions:

1) How is the fund's benchmark built?

2) How is the fund built to track its benchmark?

The first question is central to our analysis of index funds and ETFs. Index construction matters. It defines these funds' strategies. We pore over the minutiae of index methodology documents in an effort to understand how the rules governing an index will manifest themselves in a fund's risk/return profile on a stand-alone and peer-group-relative basis. Here, I provide an overview of the basics of index construction and examine a pair of examples that underscore the importance of understanding this essential element of index funds' and ETFs' process.

Index Construction 101
Index construction is a fairly straightforward process. You begin by defining the universe of securities that you will draw from (U.S. stocks, investment-grade corporate bonds, and so on). Next, you apply screens. These screens may vary in type, number, and ultimately their overall impact on the index's portfolio. Preliminary screens may be conducted on the basis of listing exchange, total market capitalization, and trading volumes. These initial screens are intended to insure that the index is ultimately investable. The index may then apply certain rules to select securities from this screened universe. For example, a value index might include only stocks from within the initial universe that demonstrate the strongest value characteristics. Once constituents have been selected, they may be weighted in a variety of ways, based on market cap, some fundamental measure such as book value or earnings, historical volatility, and so on. Finally, there is the matter of rebalancing and reconstituting the index. An index's rebalancing rules will have important implications with respect to the cost of running the portfolio.

Plain Vanilla
If the index's screening and selection process starts and ends with some basic investability screens, no additional selection criteria are applied, and constituents are market-cap-weighted, you have yourself a broad-based, plain-vanilla benchmark. Reconstitution will be driven by corporate actions (mergers and acquisitions, for example) and rebalancing carried out by the natural ebb and flow of the market. In this regard, these indexes are extremely efficient, as they let the market do most of the heavy lifting. They are also macro consistent--that is to say they are the best investable representation of investors' total investment opportunity set within a particular asset class. These are the plainest of all benchmarks and underlie the likes of Vanguard Total Stock Market ETF VTI, iShares Core S&P Total U.S. Stock Market ITOT, and Schwab U.S. Broad Market ETF SCHB. Though each of these funds tracks a bogy from a different index provider (CRSP, S&P, and Dow Jones, respectively), and there are granular differences in the details of their index methodologies, the funds will likely produce substantially similar results. Matters become more complicated as the initial universe narrows, more screening and selection criteria are introduced, the index moves away from market-cap weighting, and systematic rebalancing is put into the mix. These nuances are best illustrated with some examples.

The Case of Small Caps
In Exhibit 2, I have plotted the trailing-three-year annualized returns and standard deviations of four of the most popular small-cap equity ETFs: iShares Russell 2000IWM, iShares Core S&P Small-Cap IJR, Vanguard Small-Cap ETF VB, and Schwab U.S. Small-Cap ETF SCHA. Each of these four funds tracks a distinct small-cap benchmark, and each had a substantially different risk/return profile over the period in question, with the starkest difference evident in the comparison of IWM and IJR.

The case of IWM and IJR is an illustrative example of the importance of an index's initial universe, as well as screening and selection criteria. IWM tracks the widely followed Russell 2000 Index, which is a subset of the Russell 3000E Index. The 4,000 members of the Russell 3000E Index are ranked by market cap, and stocks ranked from 1,001 to 3,000 ultimately form the Russell 2000 Index. No additional screens are applied beyond those used in ensuring the investability of the components of the Russell 3000E Index, constituents are weighted on the basis of free-float-adjusted market cap, and the index rebalances once a year.

The S&P SmallCap 600 Index is born out of a narrower universe, the S&P Composite 1500 Index. It represents the small-cap strata of this universe. Its siblings, the S&P 500 and S&P 400 indexes, represent the large and small-cap segments, respectively. In addition to initial investability screens, the S&P SmallCap 600 Index defines eligibility on the basis of market cap (looking to include companies with an unadjusted market cap of $400 million to $1.8 billion) and financial viability (constituents must have reported positive earnings in the most recent quarter and must have--in sum--positive earnings over the trailing four quarters). Given the narrower universe, the S&P SmallCap 600 Index has a higher median market cap relative to the Russell 2000 Index ($986 million versus $707 million as of the end of May). But perhaps the most important difference between the two benchmarks is the S&P SmallCap 600 Index's financial viability screen. By weeding out stocks with negative earnings, the index introduces a significant loading to quality. This has provided a meaningful performance edge to IJR, as removing low-quality stocks from small-cap portfolios has been documented to breathe life into the size premium.[1] In practice, this has resulted in a substantially superior risk/return profile for IJR relative to IWM.

Shades of Value
Value indexes and indexes that ultimately produce portfolios with significant loadings on value (such as fundamentally weighted indexes) provide a useful illustration of the importance of selection and weighting criteria in driving an index's risk/return profile. In Exhibit 3, I have plotted the trailing 10-year annualized returns and standard deviations of four of the most popular value-oriented ETFs, as well as the largest fundamentally weighted ETF: iShares Russell 1000 Value IWD, Vanguard Value ETF VTV, iShares S&P 500 Value IVE, Guggenheim S&P 500 Pure Value ETF RPV, and PowerShares FTSE RAFI US 1000 PRF.

VTV, IWD, and IVE are more alike than they are different. Each fund's underlying benchmark cuts its beginning universe into a value half and a growth half, based on a number of value and growth factors, and subsequently weights their constituents by market cap. Like their more-broad-based total-market parents, these indexes tend to be very cost-efficient. But weighting on the basis of market cap tends to dilute their weightings on the value factor.

Breaking the link with price when it comes to weighting constituents (that is, doing something other than market-cap weighting) lends itself to a more concentrated value exposure. Straying from market-cap weighting is the common tie between the benchmarks underlying PRF and RPV. The FTSE RAFI U.S. 1000 Index weights its constituents on the basis of current book value and trailing five-year averages of sales, cash flow, and dividends. The S&P 500 Pure Value Index, which underlies RPV, selects and weights its constituents on the basis of the strength of their value characteristics, choosing stocks that represent about a fourth of the value of the S&P 500 (not half, as is the case with IVE's bogy). Both indexes rebalance once a year, shedding stocks whose valuations have risen and re-upping their allocations to those with depressed multiples. In both cases, the net result of moving away from market-cap weighting has been superior long-term performance relative to cap-weighted value indexes with greater levels of risk--far greater in the case of RPV.

The Moral of the Story
Index construction matters. The particulars of an index's makeup will dictate the performance profile of a fund tracking that index. Parsing the details of index methodologies is central to understanding the processes of index funds and ETFs and informs our overall assessment of these funds' investment merit.

[1] Asness, C., Frazzini, A., Israel, R., Moskowitz, T., & Pedersen, L. 2015. "Size Matters, If You Control Your Junk." Fama-Miller Working Paper.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2553889