In past posts, I’ve pointed out that currency-hedged funds such as the iShares S&P 500 (CAD-Hedged) ETF (XSP) and TD US Index Currency Neutral e-Series Fund (TDB904) exhibit large tracking errors that cannot be explained by their higher MER alone. I’ve never had a good explanation for why is so. And more importantly, I couldn’t answer with conviction whether large tracking errors are simply a statistical anamoly or whether negative tracking errors are likely to persist in the future. A recent paper by Raymond Kerzérho of PWL Capital (thanks to Larry MacDonald for pointing to this in a recent post) titled Currency-Hedged S&P500 Funds: The Unsuspected Challenges provides a conclusive answer to both these questions.
The paper breaks down tracking error into four components: (1) Difference in management fees (2) Transaction costs such as commissions, bid-ask spreads and other administrative costs (3) Interest-rate differential between Canadian and US exchange rates and (4) Residual-Currency Effects (RCE).
The last of these — RCE — is the most interesting because it accounts for the bulk of the observed tracking errors. RCE results from fluctuations in stock prices. Here’s one example from the report: A S&P 500 currency-hedged index fund with $100 million in assets starts off with a $100 long position in the S&P 500 index and a $100 million short position in US dollar forward contracts (all US dollars). If the S&P500 index advances 3%, the fund will be long $103 million in the S&P 500 but the forward contracts cover only $100 million. The fund is underhedged by $3 million. If the US dollar falls by 3% during the same time period, the fund will have a tracking error of +0.09%. If the US dollar increases by 3%, the fund will have a tracking error of -0.09% (the fund outperforms the index).
RCE will give rise to a positive tracking error (fund underperforms the index) when stocks and currency moves in opposite directions and negative tracking error (fund outperforms the index) when stocks and currency moves in the same direction. If stock prices and currencies move randomly with respect to each other, you would expect currency-hedged funds to underperform at certain time periods and outperform in other time periods (reader Avon Barksdale made this point in Why Currency Hedging is Necessary).
It turns out currency and equities more often than not move in opposing directions. This won’t be a surprise to investors who have watched US equities zig when the US dollar zags and vice-versa in the recent past. But this observation is not limited to the 2006-09 time period. Mr. Kerzérho estimates that the -0.38 correlation between the S&P500 and US dollar in the 1980 to 2009 period would have resulted in a RCE hit of 0.40% per annum for a currency-hedged portfolio. In fact, Mr. Kerzérho found that 60-month rolling correlation between US stocks and the US dollar never moved into positive territory in the 1980 to 2009 period. Such a long history suggests that the 2006-09 period isn’t simply an anomaly. Mr. Kerzérho concludes [emphasis his]:
Based on this evidence, I believe that a negative correlation is likely to persist rather than revert towards zero, and that this therefore imposes a large inefficiency cost on the unitholders of S&P500 currency-neutral funds.