Why Index Construction Matters for ETF Investors

Christine Benz: Hi, I'm Christine Benz for Morningstar.com. Indexes tracking the same general market structure may still vary significantly in their construction. Joining me to discuss this issue and why you should care about it is Ben Johnson. He is director of global ETF research with Morningstar.

Ben, thank you so much for being here.

Ben Johnson: Thank for having me, Christine.

Benz: In a recent issue of ETFInvestor you discussed this issue of index construction, and I think it's interesting because for a lot of investors fees are really their main distinguishing factor among ETFs in a given sector or in a given slice of the market. But you say investors really should stay attuned to this issue of index construction. Let's talk about why it matters.

Johnson: Well, index construction matters outside of the context of decisions that are being made among funds tracking like indexes. So, say, for example, I'm assessing the relative merit of the three big S&P 500 ETFs, so the Vanguard S&P 500 ETF; the iShares Core S&P 500 ETF; and the SPDR S&P 500 ETF. And I line those three up and I look at them and I say, OK, one costs 5 basis points, the Vanguard one; one costs 7, the iShares fund; and one costs 9. And looking at fees in isolation can get you a long way in this very specific, very narrow case toward making the best decision among those three.

But once you expand outside of these very narrow cases, we now have in excess of 1,900 exchange-traded products in the U.S. marketplace, you have to look well beyond the scope of fees so you can yield yourself a handful of basis points in these like-for-like comparisons and in incremental returns, but the differences in performance between unlike, though may be similarly labeled indexes, can be measured in full percentage points on an annualized basis over a long time horizon. So, index construction matters. It matters a lot. Index construction essentially defines the Process, which is one of the five pillars that we assess when we're conducting due diligence on active managers, passive strategies, strategies that lie somewhere in between. So we really want to understand how are these indexes built and what might we expect based on their construction with respect to the return and the risk profile of the resulting portfolio.

Benz: So let's take a look at the construction of the most basic type of index that you could find, that's the capitalization-weighted index. How do funds like that put themselves together--indexes like that put themselves together?

Johnson: Broad-based cap-weighted indexes are designed to capture, to the best of their ability, the market. So, generally speaking, if you look at a total stock market index, it will oftentimes capture 90% plus, oftentimes nearly 100% of the opportunity set in, say, the U.S. equity market. It will screen out certain stocks on the basis of investability, so they might not trade enough, they might not be big enough, what have you, but they look to own the entire market and own each of those securities in proportion to their market capitalization. So, it's macro-consistent. What macro-consistent means is, it looks like the market. It represents all of the dollars invested in all of the stocks out there in the marketplace and it weights them accordingly to that level of going market value. So, in that case, the market does a lot of the heavy lifting, prices go up, prices go down, the stocks and their weightings will follow suit. So, it's very efficient in a number of regards, capitalization weighting, with respect to the ongoing maintenance, it's self-rebalancing and as a result, tends to be extremely cost-efficient and tax-efficient.

Benz: So let's discuss another level though. When a fund implements a capitalization-weighted index strategy and say, it's a particularly diffuse index, like maybe the Barclays Aggregate Bond Index or the Russell 2000 Index of small-cap stocks, funds don't necessarily own one-for-one each of the holdings. Some of them do some sampling. Let's talk about that issue because I think some investors might not be aware that that's going on behind the scenes.

Johnson: And that's a great question. With respect to analyzing process, we ask ourselves two big questions. So, the first, which we've already addressed is, how is this benchmark index built? Second and of equal importance is, how is the fund built to track that benchmark index? So, in cases, as those as you've described, like the Bar Cap Ag, like the Russell 2000, going out and trying to build a portfolio that mimics on a one-to-one basis the securities that are represented, that are constituents of those indexes would be cost-prohibitive. You could go out there and you could do it, but the actual portfolio performance would ultimately lag by oftentimes a wide margin the performance of the Russell or the performance of the Ag because you would incur so many costs in trying to locate and add to the portfolio those tiny securities at sort of the bottom rungs of that portfolio that tend to be very illiquid and very costly to trade.

So what portfolio managers will do is they will create a portfolio that represents a sort of best fit representative sample of that index. So I'm not going to go out and get that sort of very tiny basket of illiquid securities. I'm going to leave that out, and I want to make sure that the profile of the portfolio is consistent with the Russell 2000 or the Bar Cap Ag so that it tracks with a high degree of fidelity that index. So it's a trade-off between the tracking, the costs, the sort of representativeness of the sample that you build. So, it's a delicate balance that these index portfolio managers are trying to maintain and one that I think oftentimes goes underappreciated by index fund investors given that most index funds tend to do a very good job of tracking even very broad, very diffuse indexes like the Russell and Bar Cap Ag.

Benz: Are there any cases where a portfolio manager has perhaps tried to be a little heroic in terms of this sampling and maybe tried to pick up a little bit of a return edge or make up for an expense disadvantage by making some bets, if you will?

Johnson: Not outright explicit bets, but I would say more what we've seen anecdotally is, in certain cases, a whiff or two. So, a sample that failed to live up to the goal of continuing to provide high-fidelity tracking with respect to the fund's underlying benchmark index, I think in the realm of ETFs, one of the most prominent examples of that would have been the iShares Emerging Markets Index ETF, so EEM. When that fund emerged through the worst of the financial crisis going from 2008 into 2009, there was a period of time where that fund was lagging the MSCI EM Index by 7 to 8 percentage points at one point in time, just given that this sample didn't adequately capture those smaller higher beta stocks that were sort of spring-loaded and sort higher coming out of the depths of the financial crisis. So when we see events like that oftentimes we get a good sense of the efficacy of these fund managers' sampling techniques.

Benz: So, another issue that you addressed in your ETFInvestor piece was this idea of how these indexes get maintained on an ongoing basis, so how holdings are added to or subtracted from as the years go by. Let's talk about that issue and how investors should think about that issue when they think about their ETF holdings.

Johnson: And it's an important question to ask because it's going to vary quite widely on a case-by-case basis. In the case of broad cap-weighted indexes what you'll see is that additions, deletions, combinations--it's all generally going to be a result of corporate actions, so merger and acquisition activity, bankruptcies, you name it, is you stray sort of more toward the middle ground of the active to passive continuum and get into the realm of strategic beta ETFs and the indexes that track those, selection criteria can become more complex, more nuanced. You'll see portfolio turnover go up not just incrementally but by many orders of magnitude relative to, say, a total stock market index. Whereas in the case of a total stock market index, you would expect to see run rate turnover somewhere in the range of 3% to 5% a year. In the case of some strategic beta ETFs that more regularly rebalance and reconstitute their portfolios, they can see run rate turnover in excess of 100% in some cases.

Benz: So, let's discuss another wrinkle on this. You noted in your article that one style that index providers tend to define very differently is value, how they define value. Can we talk about that and how maybe use some examples of actual ETFs to discuss how that can play out in practice?

Johnson: So not all funds labeled value are created equal and that is obvious, that is apparent. But what we have in the case of exchange-traded funds is a very clean line of sight into the nuanced differences, sometimes nuanced actually, sometimes far less nuanced and sort of sizable differences in respect to that portfolio construction process is defined by the index methodology. So even in the case of cap-weighted value funds, so style funds that first screen the beginning universe, say, the Russell 1000, for certain value characteristics and then buy the half of the Russell 1000 that shows the strongest value characteristics and weights those stocks by market capitalization. That might embed a certain number of value factors that might be distinct from the same ones embedded in the comparable index from, say, S&P.

Now, once you move outside of that context and get into the realm of non-cap-weighted value strategies, what you see are really meaningful differences in portfolio construction by virtue of the selection criteria that are embedded in these index methodologies and more importantly, the weighting criteria. When you get into things like fundamental indexing, so a portfolio like that offered by the PowerShares FTSE RAFI US 1000 ETF, PRF, which weights on the basis of fundamental criteria, or something like the Guggenheim S&P 500 Pure Value ETF. What you see is the resulting risk and return profiles of those value strategies relative to market capitalization-weighted value strategies differ quite widely. So you are absolutely going to get a riskier return profile. You might get more return as a form of compensation for that risk. So, it's important to understand that not all things labeled value are created equal, and is another case in point that underscores the importance of really digging into the index methodology to better understand what is that portfolio going to look like, what is the risk and return profile I might expect.

Benz: So, if you're looking in a given category, don't assume that a large-blend ETF is a large-blend ETF, is a large-blend ETF. These distinctions can really have a significant impact on performance.

Johnson: Absolutely.

Benz: Ben, thank you so much for being here. Always great to hear your insights.

Johnson: Thanks for having me.

Benz: Thanks for watching. I'm Christine Benz for Morningstar.com.