Gordon is a recently converted passive investor who has just realized a $5,000 capital gain while implementing his new portfolio. The next month, his $100,000 investment in the iShares S&P/TSX Capped Composite Index Fund (XIC) drops by $5,000 – making him regret the day he first stumbled across the Canadian Couch Potato website.
After he has finished sulking, he considers realizing the loss by selling XIC, and replacing it with the Vanguard MSCI Canada Index ETF (VCE) (a similar but not identical security). This would offset the $1,150 of tax he would be on the hook for (assuming a marginal tax rate of 46%). His initial concern is the potential opportunity cost of holding VCE instead of XIC over the next 30 days (he must hold VCE for at least 30 calendar days in order to avoid the superficial loss rule). If VCE underperforms XIC by a wide enough margin, his attempt at tax management may prove to be futile. This is one of the risks to consider before engaging in tax-loss harvesting strategies.
Gordon knows that it makes little sense to save $1,150 in tax if he ends up missing out on $1,150 of market growth. To get an idea of how much VCE could be expected to underperform XIC in any given month, we could examine the monthly return differences between 1970 and 2011 for the S&P/TSX Composite Index and the MSCI Canada Index (which are reasonable proxies for XIC and VCE). As we can see in the graph below, there has been a wide range of monthly return differences. However, over 90% of the monthly observations were above -1.00%. He feels comfortable with this probability of underperformance, but wants to have a better understanding of the costs of buying and selling these securities.
Sources: Dimensional Returns 2.0, MSCI
Bid-ask spreads can be a significant cost but often take a back seat in the media relative to trading commissions. Both costs can derail the benefits of tax-loss harvesting. Gordon estimates the bid-ask spread for these transactions to be around 0.17% (or about $160 in his case). He adds to this amount four trading commissions of $10 each, resulting in total implementation costs of $200.
For the example below, we will assume that XIC returned 1% over the 30 day holding period while VCE returned 0% (for an opportunity cost of $950). We’ll also assume the implementation costs amount to $200.
The example above provides a useful “break-even” scenario. Even with a relatively bad outcome over the month (missing out on 1% growth), Gordon’s overall situation is arguably not any worse than if he had done nothing. His decision on whether to harvest his losses could have also been reached by using Larry Swedroe’s rule of thumb (where both conditions listed below must first be met):
Marginal Tax Rates
Keep in mind that Swedroe’s rule of thumb appears to be a reasonable one only if the investor is in the top tax bracket. If the investor is in a much lower tax bracket (such as 23%), they may want to use a more conservative rule of thumb, such as:
The decision on whether to harvest your losses is not as straight-forward as it may seem – it will depend on a number of factors, including:
Some of the lowest-cost ETFs trade on the U.S. stock exchange, forcing cost-conscious Canadian investors to bite the bullet and convert their loonies to dollars (most likely at a less than optimal rate). There are now ETFs and trading strategies available that can help mitigate a portion of the hidden costs of converting Canadian dollars to U.S. dollars.
The Horizons U.S. Dollar Currency ETF (DLR) is an investment vehicle that holds cash and cash equivalents that are denominated in U.S. dollars. It is traded on the TSX and priced and transacted in Canadian dollars.
The Horizons U.S. Dollar Currency ETF (DLR.U) is the exact same investment vehicle as DLR. It holds cash and cash equivalents that are denominated in U.S. dollars and also trades on the TSX. The only difference is that it is priced and transacted in U.S. dollars.
Both DLR and DLR.U have the same International Securities Identification Numbers (ISIN), which is basically the “DNA” of the security. The first two characters “CA” are the country code, the next nine characters “44049C104” are the Committee on Uniform Securities Identification Procedures (CUSIP) number, and the final character “1” is the check digit.
The process of using these two identical securities to convert Canadian dollars to U.S. dollars more cost-efficiently may seem like a daunting task at first, but once you get the hang of it, it’s a cinch.
An investor has $58,140 Canadian dollars that they would like to convert to U.S. dollars in order to purchase the Vanguard Total Stock Market ETF (VTI) within their brokerage account. Their foreign exchange department has just offered them $1.0161 U.S. dollars for each of their Canadian dollars (for a total of $59,076 USD). The investor is not entirely convinced that this is such a good deal for them, and proceeds to use DLR and DLR.U to get a better currency conversion rate. The four steps they would need to take are listed below (*Remember to always use limit orders when placing ETF trades):
Step 1: Buy 6,000 shares of DLR in the Canadian Dollar Account
Since DLR is priced and transacted in Canadian dollars, there will be no forced currency conversion for the purchase of 6,000 shares of DLR on the TSX. If the quoted ask price is $9.69 per share, the total cost in Canadian dollars (excluding commissions) would be $58,140.
Step 2: Sell 6,000 shares of DLR.U in the U.S. dollar account
It may appear as though you are selling a security you don’t have, but remember that DLR and DLR.U are identical securities – the only difference is the currency that their price is quoted and transacted in. It is extremely important that you sell 6,000 shares of DLR.U in your U.S. dollar account (to avoid unintentional currency conversions). DLR.U will show as a short position (i.e. negative 6,000 shares) until the dust settles (approx. trade date + 4 days). If the quoted bid price is $9.91 per share, you should end up with cash proceeds of $59,460 USD (excluding commissions).
By using this method, the investor ends up with $384 additional U.S. dollars in their account ($59,460 - $59,076).
Step 3: Buy 800 shares of VTI in the U.S. dollar account
Since the proceeds from the DLR.U sale will settle on T+3, and the purchase of VTI will also settle on T+3, the trade can be initiated immediately. If the quoted ask price for VTI is $74.325 per share, the total cost would be $59,460 USD (excluding commissions).
Step 4 (T+3): Journal 6,000 shares of DLR in the Canadian dollar account to the U.S. dollar account
Recall that we sold 6,000 shares of DLR.U that the buyer now wants delivered to them – although we don’t own DLR.U, we have something equally as good! We “journal” 6,000 shares of DLR from the Canadian dollar account to the U.S. dollar account in order to settle the sale of DLR.U, keeping the account in good order.
Perhaps as the Canadian ETF market develops, there will be equally comparable products both here and south of the border. Until that time, using DLR and DLR.U is one potential method for lessening the burden of currency conversion costs. For a more intuitive explanation of the process, please refer to the flow chart below.
Dan Bortolotti (a.k.a. The Canadian Couch Potato) posted an incredible blog yesterday that will no doubt become the go-to resource for many investors (and advisors alike) who want to wrap their heads around the withholding tax implications of owning foreign ETFs and mutual funds. Although the concept is hard to digest for most people, those who battled through the article will no doubt be better investors for having done so.
One issue I see popping up for investors is basing their asset location decisions solely on foreign withholding tax implications. Although these can certainly be a drag on returns, we must remember that there are other costs to consider. In this post, I’m going to highlight one of the most obvious ones: foreign income tax.
In the examples that follow, we’ll see how an investor who makes their asset location decisions based on foreign income tax and foreign withholding tax implications can come out ahead of the game. We’ll assume the investor has a $600,000 RRSP and a $400,000 non-registered account (invested in a balanced Couch Potato portfolio). We’ll also assume:
In Option 1, the investor is of the opinion that by avoiding foreign withholding taxes, they will reduce their overall foreign tax liability. Since the Vanguard S&P 500 ETF (VOO) avoids foreign withholding tax when held in an RRSP, and the TD International Index Fund e-Series (TDB911) can recoup the foreign withholding taxes levied when held in a non-registered account, this seems like the most tax-efficient asset location choice.
Taxable Foreign Dividends: $7,380 ($200,000 × 3.69%)
Income Tax on Foreign Dividends: $3,395 ($7,380 × 46%)
Non-reclaimable Foreign Withholding Tax: $0
Total Foreign Tax Liability: $3,395
In Option 2, the investor is aware of foreign withholding tax implications, but also knows that the TD International Index Fund e-Series (TDB911) has a higher dividend yield and therefore a higher foreign tax liability, relative to the Vanguard S&P 500 ETF (VOO). They decide to take the hit on the 12% foreign withholding tax by holding the TD International Index Fund e-Series in their RRSP, thereby shielding their net dividend (approx. 3.25%) from immediate tax at their 46% marginal tax rate. The Vanguard S&P 500 ETF (VOO) is held in their non-registered account and the investor recoups the foreign tax withheld.
Taxable Foreign Dividends: $4,340 ($200,000 × 2.17%)
Income Tax on Foreign Dividends: $1,996 ($4,340 × 46%)
Non-reclaimable Foreign Withholding Tax: $886 ($7,380 × 12%)
Total Foreign Tax Liability: $2,882
As we can see, the optimal asset location choice when only considering foreign withholding taxes turned out to be the least optimal after we considered the overall tax consequences of holding higher dividend paying foreign securities in a non-registered account. The results would also change if we lowered the marginal tax of the investor, indicating that the optimal asset location could be different for each investor.
With government bond yields near record lows, investors have been looking to corporate bonds to make up some of the shortfall. As credit spreads (the difference between corporate bond yields and government bond yields) hover at around 150 basis points, I can’t blame yield-starved investors for wanting to make the switch.
Source: Bank of America Merrill Lynch Global Index System
The strategy of increasing a portfolio’s exposure to corporate bonds when spreads are wide (and vice-versa) is called the variable credit approach to fixed income. Without trying to sound like a fortune-teller (or worse yet, an active bond manager), research does appear to suggest that there has historically been a reliable relationship between current credit spreads and future return differences. In other words, if most investors are fearful, it might be the right time to be a bit greedy.
If we run a regression analysis using Canadian credit spreads at the end of the year as the independent variable and the annual return difference between corporate and government bonds in the following year as the dependent variable, we can determine if a reliable relationship exists between current credit spreads and future return differences.
If you thought you’d be able to get through one of my blogs without having to sit through another boring regression analysis, think again. To make things less painful, however, I’ve included only a few of the main points below:
Source: Bank of America Merrill Lynch Global Index System
Yields: 1996-2010; Returns: 1997-2011
Simulating a Variable Credit Strategy using Canadian Data
Now I know performance junkies are jonesing for a back-test of this strategy, so here it is:
Although the results do not seem to be impressive, realize that the intention of the variable credit strategy is not to knock the ball out of the park. It is a disciplined, rules-based tilt towards the credit factor only when the risk is reasonably expected to be compensated for. As you can also see, the variable credit simulation had a higher return, similar standard deviation, as well as a higher average allocation to government bonds over the period studied, relative to the benchmark.
Sources: Dimensional Returns 2.0, Bank of America Merrill Lynch Global Index System
Although a low-cost, passively managed bond portfolio is arguably one of the best choices for individual investors, a variable credit strategy may prove to add-value in an already low-interest rate environment.
In a past blog, we discussed the many intricacies of the variable maturity approach to fixed income investing. Recall that the main idea is to extend the maturities of your bond portfolio when yield curves steepen and reduce the maturities when yield curves flatten. In this blog, we will look at how the average investor could implement a similar strategy with their own portfolio, using commonly traded ETFs.
For this example, we will use the iShares DEX Short Term Bond Index Fund (XSB) as our default ETF, and the iShares DEX Universe Bond Index Fund (XBB) as our replacement ETF. Our assumption will be that XSB will be held continuously unless we can receive at least 0.15% of additional expected yield per year for taking on each additional year of term risk (this could always be adjusted, depending on your risk appetite).
Using data that is readily available on the iShares site, we can determine that XBB has an additional 6.85 years of term risk relative to XSB, and this is compensated for with an additional 0.65% of annual expected yield (after MER). Since we require at least 1.03% of additional yield per year (6.85 years × 0.15%), we will continue to hold XSB and no changes will be made to our fixed income portfolio.
Variable Maturity Simulation
Using historical ETF data kindly provided by Rahim Surani and Marie Amilcar of BlackRock Canada, we can create a simulation of this strategy to gain a better understanding of the process.
In our analysis, quarterly yield differences (after MER) and maturity differences between XSB and XBB were examined to determine when a switch from one to the other would have made sense (i.e. would have given us an additional 0.15% of annual expected yield for each additional year of term risk). Interestingly, the only time period when XSB was swapped for XBB was between April 2009 and June 2010.
In the table above, we notice that the variable maturity simulation resulted in higher returns than XSB, as well as higher risk-adjusted returns than both XSB and XBB.
Although there are many other factors to consider when deciding on any investment strategy (your willingness to take risk would be at the top of the list), the variable maturity approach to fixed income investing is based on the sound investment philosophy that investors should take risks that they are expected to be compensated for in the long term.