Now before you stop reading, all else being equal fees do matter. But rarely if ever is all else equal. Fees for ETFs have already been reduced to such an extent that a difference in costs of a few basis points between funds is likely to be overwhelmed by the difference in performance of the underlying investments, making fees a secondary and far less important consideration.
For example, among several large cap domestic stock ETFs, the S&P 500 SPDR (NYSEARCA:SPY), with an annual fee of 10 basis points, returned 15.8% in 2006, while the Dow Industrials DIAMONDS (NYSEARCA:DIA), with an annual fee of 18 basis points, returned 18.9%. The large-cap iShares Russell 1000 (NYSEARCA:IWB), at 15 basis points, was up 15.4% last year and the NASDAQ-100 Tracking stock (QQQQ), at 20 basis points, gained only 7.1%.
Between all those large-cap domestic stock funds there is a spread of no more than 10 basis points in fees, yet a difference in performance of 11.8 percentage points. And don’t assume that the difference evens out in the long term: over the past five years, the spread in cumulative performance is 23.4 percentage points!
It would be absurd to assume that over the next five years the cumulative performance of these funds will be within 0.5 percentage points of each other — yet that is all the impact that the difference in fees between the most expensive fund and the least expensive fund will have.
You’d find similar divergence in returns if you examine small caps ETFs, ETFs within a certain geographic category, or even ETFs from the same fund company targeting Growth stocks! In fact, of the 74 iShares ETFs from Barclays Global Investors with at least five years of trading history there is only minimal correlation between expense and performance (“r-squared” = 15%). And for what it is worth, the relationship is positive, i.e., higher fee funds had better total returns than lower fee funds—the opposite of what conventional wisdom would predict.
Further examination of the data would reveal that the relative outperformance of international-stock funds, which also tend to have higher fees, was the cause. Comparing fees and returns within each category found that the correlation between the two is statistically insignificant (Figure 1). In other words, fees are largely irrelevant!
Instead, each of the large-cap domestic funds mentioned above have notably different profiles in terms of sector balance, expected earnings growth, historical profitability of constituent firms, and valuation measures such as the price-to-earnings ratio and price-to-book value, despite targeting the same category. It is these differences, not fees or past performance, that are likely to determine by and large how the funds perform in the future.
Critics will argue that fees for some ETFs are much higher, and therefore hardly insignificant. The popular iShares MSCI Emerging Markets fund (NYSEARCA:EEM), for example, has an annual fee of 75 basis points which, while far cheaper than the average emerging markets mutual fund, is not something to be ignored.
True enough. And in fact in my own firm’s ETF rating system, which tries to measure an ETF’s investment merit, does take fees into account for the sake of accuracy. But that still misses the point. EEM may or may not turn out to be a good investment going forward, but the answer to that question is unlikely to be determined by fees. The difference in returns between EEM and say, iShares MSCI EAFE fund (NYSEARCA:EFA) of developed-market foreign stocks, with a fee of 35 basis points, is most likely going to be far larger than 40 basis points per year!
For ETFs, fund fees often amount to a rounding error. So the obsession many industry analysts and investors have with fees focuses on the wrong question. Instead of a race to the find the cheapest expense ratios, investors need to try to answer the far more important question of whether the basket of stocks held by the ETF offers attractive investment potential—because the answer to that question is what is going to largely determine an ETF’s ultimate performance.