This was brought home recently when I was doing some research on iShares exchange-traded funds (ETFs). Now, I don’t mean to single out iShares in this regard. In fact, I like their ETFs and use them in my portfolio. But I find one has to be careful when gathering information from their website.

For example, some of their ETFs are shown as having tracking errors less than their management expense ratios (MER). Of note, the iShares CDN Completion Index fund (XMD) is reported as having an annualized tracking error since inception of 0.44%, less than the MER of 0.55%.

This might be possible if the ETFs were based on “optimized portfolios” (i.e. use samples). But these cases are not. So, their tracking error should be larger since the MER is just one of its components.

It turns out that the ETFs’ underlying indexes were changed in recent years. So when one calculates an annualized tracking error since inception, it’s really an amalgam of two different indexes. That’s not of much use. At least there should be an warning asterisk?

Second, staying with tracking errors, the iShares CDN S&P 500 Currency-Hedged Index is shown as having an annualized tracking error of just over 1 per cent. However, as mentioned elsewhere, that figure is determined with reference to a currency-hedged version of the S&P 500 index. If the ETF is compared to the unhedged S&P 500 (which would be of more interest to many DIYers), the annual divergence averages more than 2 percentage points over the past three years.

A third challenge is trying to find out what the yield is on an ETF. This is not calculated on the site. What the visitor gets is the distribution history — so it is up to them to estimate a yield themselves. But using these tables can be a little tricky.

The novice might go to the table for 2008 and then pick the “total distribution for the year” to estimate the yield at current unit prices. But this total includes “Reinvested distributions per unit,” which is a distribution of capital gains at year end. This is a non-cash, irregular item (and doesn’t result in extra units as in the case of mutual funds). A better number to use for estimating the yield would seem to be the “cash distribution per unit” entry on the table.

This can be significant if the index basket has changed and the ETF manager had to sell and buy securities to keep up. See, for example, the iShares CDN Large Cap Index fund (XIU). Its total distribution in 2008 was $0.62 per unit, which at current prices would generate an estimated yield of 4%. But the reinvested distribution was $0.146 per unit, leaving a “cash distribution per unit” of $0.474. This generates a yield of 3.1%, a lot lower that the first estimate of 4%.

A number of empirical studies have looked at this issue. From their backtests of hedged/unhedged globally diversified portfolios, they have found that hedging was unnecessary for long-term investors. Two examples of the studies, covering the 1975 to 2003 period, are the Thomas paper, ‘Currency Risks in International Equity Portfolios’ and the Statman and Fisher paper, ‘Hedging Currencies with Hindsight and Regret.’

If currency hedging is unnecessary for long-term investors, that would seem to be good news. It would spare them the costs of hedging. And this cost can be noteworthy for those people buying foreign exchange-traded funds (ETFs) that include currency hedging. Costs take the form of higher MERs and tracking errors.

Yet, one risk with the unhedged view is the “fat tail.” Most of the time, as recent history points out, currencies will tend to fluctuate in ways that average out. But once in a while, a currency may collapse or go into a long-term decline against others. Latin American currencies suffered this fate in the past.

The U.S. dollar may well avoid such a fate given its central role in the world economy. But many articles and books nonetheless have been written warning that accumulating financial and trade imbalances could some day result in a currency crisis or flight from the U.S. dollar.

U.S. investors in foreign assets would thus seem to have even less need to hedge than what the empirical studies suggest – at least if they are long-term investors. True, the U.S. dollar could rise as the fiscal deficit widens and pushes up interest rates. This is what happened under Reaganomics in the 1980s. But that appreciation lasted only a few years.

Foreign investors in U.S. assets may see currency hedging as worthwhile – especially if they have a high level of risk aversion based on “fat tail” outcomes. They may end up with a lower net return than an unhedged investor if the fat tail does not occur, but that would be acceptable as the premium on an insurance policy. Investing choices should, after all, be made on a risk-adjusted basis. Currency hedgers may want, however, to find the lowest cost way of hedging as opposed to convenient choices such as currency-hedged ETFs.

Take the example of a Canadian wishing to buy into the S&P 500 Index: they can get the currency-hedged iShares CDN S&P 500 Index Fund (XSP) or the unhedged iShares S&P 500 Index Fund (IVV). The MER for XSP is 0.15% higher, which doesn’t look too bad. But then, as Canadian Capitalist calculated, the XSP trailed IVV by -1.73%, -2.30% and -3.5% in 2006, 2007, and 2008, respectively.

The iShares.ca website reports a smaller “tracking error” of 0.34%, 0.56% and 1.31% for those years. But their benchmark was a currency-hedged version of the S&P 500. The difference in Canadian Capitalist’s and iShares.ca estimates mostly represents the tracking error of the latter’s currency-hedged benchmark against the actual S&P 500 Index. As iShares Canada Head of Business Development, Heather Pelant, coveyed in an email:

“An important thing to consider however, is that the benchmark of XSP is not the basic S&P 500, but a currency hedged version of this index which is calculated by S&P. That hedged index has in fact has somewhat underperformed the basic S&P 500 over the last couple of years

The difference between the basic and hedged indices results from a few factors – one is the cost of implementing the currency hedge…. Another is interest rate differentials, which impact on the pricing and performance of forward contracts used in hedging. Finally, market volatility affects the way currency hedged investments compare to un-hedged alternatives.

The fund’s tracking against the hedged index is affected by transaction costs for trading the currency hedges, management/trustee fees, and in 2008, market volatility. The unprecedented volatility of markets in 2008 significantly affected the management of currency-hedged strategies.

The IVV itself tracking the basic S&P 500 index almost perfectly, with all tracking error essentially the result of management fees.”

So that explains the difference in tracking error estimates. For the index investor, the tracking error relevant to their experience will be the one estimated by Canadian Capitalist. If currency fluctuations average out over the long run, as he says, currency hedging is not likely the way for most long-term investors to go when diversifying into foreign markets. The cost can be high. Check the tracking error.

Serious students of investing, however, know there are a number of empirical studies from academia and elsewhere that conclude there is no need to hedge currencies if the investing horizon is long term. Going by the historical data (for U.S. investors venturing into multiple foreign stock market), returns are nearly the same — as noted in this post last May.

Yet, structural imbalances (e.g. trade and fiscal deficits) in the U.S. continue to accumulate beyond thresholds that typically have triggered sustained currency depreciation in other countries. Perhaps past empirical studies, covering periods when U.S. imbalances were less extreme, need to be taken with a grain of salt.

Even if currency fluctuations do average out over the long run, such an outcome provides little solace to investors who do not have a long time before they need the funds, such as persons who will be retiring in less than 15 years. Unhedged foreign funds are more suitable for younger people — although even here there seems to be a major caveat.

What’s the caveat? While stocks can be expected to return 6% to 10% annually over the long run (going from the past record), there is no expectation of a similar long-run, positive return for currencies. Currencies are only expected to cycle in long swings around the investor’s breakeven point, which means there is a risk the cycle may not be at, or above, breakeven when the time for withdrawal comes.

Perhaps, then, the decision to hedge depends on one’s risk tolerance. Some may see the extra 15 basis points or so in the fund’s annual fees (plus any tracking error) as an acceptable insurance premium to pay for a more certain outcome. Others will see the premium as too expensive, especially when the currency contribution can be positive just as well as negative.

Canadian investors often ask whether or not they should hedge the currency when investing in funds that track U.S. stocks. This is a different situation from what most empirical studies examine (i.e. impact of several fluctuating currencies in terms of the U.S. dollar).

There seems to be, to a greater extent, certain regularities in the Canadian and U.S. dollar exchange rate that open the door to an active approach. Specifically, the loonie has a lengthy history of trading in a range roughly between $0.70 (U.S.) and $1.00 (U.S.), so one may benefit from overweighting currency-neutral funds when the loonie descends toward the lower boundary (as it is now) and shifting to an overweight in unhedged funds when the loonie approaches its upper boundary.

When the loonie is getting close to its lower boundary, history says it is more likely to go up than fall further, so a currency-hedged foreign fund may be the answer for the Canadian investor. Conversely, when the loonie approaches its upper boundary, it is more likely to fall than rise, so an unhedged fund may be more appropriate.