Earlier this week I discussed the promise of fundamental indexing. This strategy takes aim at the shortcomings of traditional cap-weighted indexes, which overweight growth stocks and are prone to bubbles. In 2005, Rob Arnott and his colleagues unveiled the Research Affiliates Fundamental Indexes (RAFI) and produced data showing that they would have outperformed cap-weighted indexes by about 2% to 3% per year over the long term in almost every market.
Any strategy that can outperform by a couple of percentage points is certainly worth a look. However, the proof is in real-world performance, not in backtested computer models. The question for investors, then, is whether the ETFs and mutual funds based on RAFI indexes have lived up to their billing.
One crucial point before we look at the data. I wasn’t interested in whether the fundamental ETFs have outperformed their cap-weighted counterparts: over a period of five years, this is virtually meaningless. Lots of actively managed funds can make the same claim, but over longer periods, that outperformance generally disappears. So I was more interested in whether the RAFI funds are tracking their own indexes closely.
Cap-weighted index funds are not perfect, but they do have several things going for them: they are cheap and usually have very low tracking errors. The iShares S&P 500 (IVV) has tracked its benchmark within six basis points (0.06%) since its inception more than 11 years ago. In Canada, the iShares S&P/TSX 60 (XIU) has a tracking error of just 0.19% over the same period.
I wanted to know if ETFs based on fundamental indexes could make the same claim. Remember, the data suggest that the RAFI indexes can deliver excess returns of about 2% to 3%. However, higher fees and larger tracking error could easily wipe out that advantage. And fundamental ETFs typically have management fees 20 to 50 basis points higher than their cap-weighted counterparts, and their indexes may be harder for a manager to follow.
The real-world performance
Let’s take a look at the most popular funds based on the RAFI indexes:
- Since its launch in February 2006, the Claymore Canadian Fundamental ETF (CRQ) has returned 6.37% annualized, versus 7.01% for its index, a difference that’s actually a bit lower than the fund’s MER of 0.69%. Incidentally, it has also outperformed the S&P/TSX 60 over that period and was ranked second out of 93 Canadian Equity funds by Morningstar during its first five years.
- The Claymore US Fundamental ETF (CLU) and International Fundamental ETF (CIE) have fared much worse. Both lagged their own indexes by huge margins in years past. Part of the reason is that these funds relied on representative sampling rather than owning all 1,000 stocks in the RAFI indexes. The ETFs now own all the stocks in their indexes and they improved their tracking errors in 2010, but both still lagged their benchmarks by more than two percentage points.
- In the US, the RAFI indexes are licensed by the PowerShares family of ETFs. Since December 2005, the PowerShares FTSE RAFI US 1000 (PRF) has returned 5.03% annually, compared with 5.54% for the index. That shortfall is entirely accounted for by its 0.45% MER. Another good result.
- The PowerShares FTSE RAFI Developed Markets ex-U.S. Portfolio (PXF), which tracks the same index as Claymore’s CIE, has a tracking error of –1.13% since its launch in June 2007.
Canadian investors can also get access to the RAFI indexes through mutual funds from Invesco PowerShares. However, these funds are designed to be sold through advisors, so they may have front-end loads and trailer fees added. They typically have MERs in the range of 1.6% to 1.8%. That guarantees that these funds will lag their benchmarks.
Another issue to consider is taxes. If you’re not investing in an RRSP, you’ll receive a tax bill for any capital gains your fund incurred during the year. Cap-weighted ETFs tend to be extremely tax-efficient: the iShares S&P 500 (IVV) has never distributed a capital gain in its 11 years of existence. In Canada, XIU and XIC have done so in the past, although not since 2008. In 2010, both CRQ and CLU also distributed significant capital gains that would have lowered returns for investors holding these funds in a taxable account.
The jury is still out
Fundamental indexing is one of the most important innovations to come along in passive investing. If it were possible to create a mutual fund or ETF that tracked the RAFI indexes within a few basis points, I have little doubt that its long-term performance would be superior to that of a traditional index fund.
But in the end, fundamental indexing must overcome the same hurdles as active management. That is, it must add value after accounting for fees and taxes. So far the results have been encouraging, but it would be hard to argue that investors should expect a 2% to 3% outperformance over the long term. I would expect any outperformance to be less than that because of the added costs of administering the funds. And a poorly run fund would forfeit any advantage at all.
Cap-weighted index funds may be flawed, but they do have at least one thing that fundamental indexes don’t yet have: a long track record of delivering on their promises.
What value is there in considering tracking error in isolation? What affects an investor’s bottom line is whether the net return after tracking error from a RAFI strategy exceeds the net return after tracking error for a comparable cap-weighted strategy. All six of the Powershares ETFs have done so dramatically, even if one is charitable and assumes a tracking error of zero for the cap-weighted strategy.
@Chris: In the short-term, you’re absolutely right. But over the long-term, any value added by the strategy may be eroded by the persistent drag of higher cost and tracking error. That is, unless you believe that the fundamental indexes will outperform cap-weighting every single year, which is clearly impossible.
One extreme example: CLU returned 28.5% in 2009, which was much higher than XSP’s return of 23%. However, the RAFI index returned 38.2%, for a tracking error of –9.7%. If you look at 2009 in isolation, you did well with CLU. But you can’t possibly expect a fund with a tracking error like that to outperform over the long term.
This is why I make the parallel with active management. Its proponents often say, “the fees don’t matter if the net returns are higher.” That’s true in theory, but expenses are relentless: they reduce returns during good years and during bad years. Over periods of 10 or 15 years those obstacles become too high for most to overcome.
Some RAFI-based funds have delivered on their promises, such as CRQ and PRF. Kudos to them. Some, however, have a ways to go. And mutual funds with more than one layer of fees are almost sure to wipe out any 2% or 3% advantage the strategy might claim.
@Potato
CLU is a side issue — any ETF or fund that uses representative sampling is going to have a larger tracking error than an ETF that buys the index. This is not an issue specific to RAFI funds; you see the same kinds of errors for cap-weighted international funds that use representative sampling.
I’m not sure what you mean by: “But over the long-term, any value added by the strategy may be eroded by the persistent drag of higher cost and tracking error. That is, unless you believe that the fundamental indexes will outperform cap-weighting every single year, which is clearly impossible.” Clearly impossible in what sense? This result is in fact not statistically improbable.
The only way I can see that your argument makes sense is if you accept the Efficient Markets Hypothesis, in which case cap-weighted funds are optimal and your reasoning is sound. But it’s been increasingly clear for the last 30 years that the EMH is not valid. The reality that the simplest and most brainless strategy one can imagine–equal-weight indexing–generally outperforms cap-weighted indexing, is a pretty clear indicator that the EMH adds no (or negative) predictive value. Even in theory, it is perfectly possible for fundamental indexing to outperform cap-weighted indexing every year on a long-term basis, after tracking error and management fees.
Arnott’s backtesting suggests, statistically, a highly probable likelihood of consistent outperformance over time in the 1.5% to 4% range, negatively correlated to the efficiency of a given market. This kind of result is not true of active management. Although one is right to be skeptical of backtests, even the youngest of the Powershares RAFI funds has almost 5 years of real-world results, and the *worst*-performing (relatively speaking), PXH, has still delivered roughly 0.9% annualized alpha after fees, etc. over the corresponding cap-weighted index. Given that the typical MER of an ETF that tracks the MSCI Emerging Markets index has a MER in the 0.6%-0.7% range, this means the worst-performing Powershares RAFI ETF has still delivered roughly 1.5% net excess returns. Effectively, real-world results have been validating Arnott’s extensive backtesting.
@Chris: You’ve raised a lot of issues here. I’ll just try to make my point as simply as possible. If I have a strategy that adds an average of 2% alpha a year in theory, but it costs me 2% more to implement because of fees and other costs, then the alpha will disappear over the long term.
It may not disappear in the short term: one year I might add 8% in alpha, which is reduced to 6% because of costs, and that still looks good. But perhaps another year it will add 1% alpha, reduced to -1% after costs. During still other years, the RAFI strategy will underperform: even Arnott acknowledges this. Of course, during these years, the underperformance will be magnified by the higher costs. Remember, costs compound, too.
Don’t get me wrong: I like the RAFI indexes a lot. All I’m arguing is that strategy will only add value in the long run if the funds track their indexes tightly. CLU is not a side issue, it’s exactly the issue I want to emphasize to investors. Unless that tracking error gets consistently below -2%, there is no way the fund can add value in the long term.
Dan, an excellent series! Seems to me (albeit a simplistic view of the issue) that Fundamental Indexing is trying to beat the market, by repositioning percentages. But the stats in your article seem to show that beating the market, even with a re-distribution of weighting – doesn’t in fact beat the market. There is the issue of continuous buying and selling to ensure those percentages, and the eventual tracking error.
The obvious answer would be although certain stocks create unequal weighing in the index, and therefore creates more volatility, that is also the reason why the index has the performance it does. By taking out the weighting of certain stocks (although mathematically) Fundamental Indexing is not only taking out the laggards, but also the winners. In a way its averaging down the performers. So I’m not surprised it doesn’t beat the index in the long run.
However I’ll go out on a limb and suggest that Fundamental Indexing might perform better in a bear market over a bull market – since it reduces volatility. Did you find that in the research? Seems every few years something comes out claiming to beat the market, but never actually does. Interesting…
@Ninja: I should clarify: fundamental indexing actually has beaten cap-weighting counterparts over the last five years in most cases. My concern is that this outperformance is not likely to endure unless the funds track their indexes closely.
If Claymore is now buying the entire FTSE RAFI US 1000 Index why wouldn’t it’s tracking error match it’s stated expense ratio of ~0.7% from now on? In any case this is a US index so we aren’t stuck with Claymore. Dan’s uber tuber model portfolio uses Powershares’ PRF to track the same index and he has indicated again recently that it tracks the index well, plus it’s stated expense ratio is lower.
As a reminder, Claymore has hedged (CLU) and non-hedged (CLU.C) versions. The hedged version isn’t going to track the USD index. It’s not supposed to.
If you want hedging for some reason CLU is probably the best option. If you don’t want hedging PRF is probably better than CLU.C *if* you can get a good enough exchange rate to buy and sell it.
@Charles: One would hope that CLU’s tracking error would be about 0.7% from now on, but managing an index fund isn’t as easy as many people think. There are all kinds of reasons why tracking error can be higher than a fund’s MER. It happens all the time.
RE: “The hedged version isn’t going to track the USD index. It’s not supposed to.” Actually, the hedged version is designed to give Canadians the same return as US investors would get from PRF. So in that sense it does track the USD index. But you’re right, if you don’t want to hedge the currency you are likely to be better off with PRF than with CLU.C.
re CLU.C’s tracking error, I called Claymore this past week, spoke to an analyst there who claimed the tracking error up to April 30th was down to 1.55%. That’s definitely moving in the right direction.
correct me if I am wrong, but fundamental indexing is actually quite a step from traditional indexing as it starts to almost “actively” decide which companies are safe bets and which are getting dangerously “hot” due to their unproven market cap; I don’t think the concept is that innovative and I would agree that long term the difference in fees and tracking error could prove to be quite a punch;
however as an alternative if one is concerned with overvalued companies yet still wants to maintain a passive index approach, why not consider capped indexes? much less work to arrive at the final composition of the index and maybe more rewarding; I am actually referring to one of your posts regarding the REIT etfs in Canada the ZRE from BMO vs its competitors; it’s a simple analogy but I think it basically addresses the same point
Hi Dan, I was wondering if you or anyone else here had revisited the idea of using fundamental indexing? Although I will always utilize traditional indexes in my portfolio, I looked at ishares fundamental indexes and their results have been quite impressive. Moreover, the MER’s are low at 0.72%. I was just wondering if any new consensus had been made since you posted the above article.
@Ross: My thoughts on this subject haven’t really changed much. I think fundamental indexing has a lot of merit: it’s really just one of several ways of accessing the value premium. With our clients at PWL Capital we combine plain-vanilla ETFs with funds from Dimensional Fund Advisors, which use a different methodology to screen for value stocks, but the principle is similar.
It’s interesting that you mention the MERs are low. I would argue that the recent price drops in traditional index ETFs (often 0.05% to 0.10%) have actually made fundamental indexes relatively more expensive. They now have to outperform by at least 0.60% before costs just to match a traditional index fund, which may be hard to do over the long term.
Hi dan, any updates on the fundamental indexes now that we have more than a decade of real world performance?