Barry Gordon admits he was surprised when he first read Justin Bender’s entry in First Asset’s Search for Canada’s Next Top ETF contest, which I introduced in my previous post. “It runs against the grain of everything we thought we knew about strip bonds,” he says. But Gordon’s firm turned the idea into the First Asset DEX 1-5 Year Laddered Government Strip Bond Index ETF (BXF), which begins trading next Tuesday. Here’s an overview of this innovative new index fund, as well as an explanation of how it might be used in a portfolio.
Inside the ETF
First Asset tasked PC-Bond with creating the index for their new strip bond ETF. Here’s the basic methodology:
- the ladder will have “term buckets” with bonds of approximately one, two, three, four and five years to maturity
- each bucket will include five individual strip bonds: four provincial (mostly issued by Ontario and Quebec) and one federal (or federal agency)
- the bonds will be selected with liquidity in mind: the issues must be at least $50 million and will be screened for maximum trading volume
- the index will rebalance annually in June: bonds with less than one year to maturity will be sold and the proceeds used to purchase new bonds for the five-year bucket
No surprises so far. But you’ll recall that one of the key characteristics of strip bonds—and the main reason why conventional wisdom says you should not hold them in taxable accounts—is they don’t generate any income. That’s where this ETF is different: it will make quarterly distributions based on the fund’s average yield to maturity, which is expected to be about 1.6%. That means investors can expect a quarterly payout of about $40 on every $10,000 invested.
Where are these distributions coming from if the holdings don’t actually pay interest? Gordon explains that “for the foreseeable future we’ll hold sufficient cash” to fund the payouts. The drag from this uninvested cash should be trivial, since the amounts are so small.
The cost of the new ETF is also a pleasant surprise. First Asset is offering a fee holiday for the first 12 months: that means a 0% MER until July 2014, after which the management fee will be a modest 0.20%.
A premium problem
Justin’s eureka moment came while working with a client who has a seven-figure fixed-income portfolio, mostly in non-registered accounts. GICs are usually the best vehicle for investors in this situation, but they have a few limitations. First, wealthy investors may find it difficult to stay within the CDIC limit of $100,000 per issuer, especially if their brokerage offers a limited menu. The second problem—more relevant to those of us unburdened by millions of dollars—is that they’re illiquid. Because you can’t sell them before maturity (unless you’re using cashable GICs), they’re not helpful if you want to rebalance your portfolio after a downturn in the equity markets.
In theory, a short-term government bond ETF would solve both these problems, but traditional bond ETFs are terribly tax-inefficient. The reason is that virtually all bonds now trade at a premium: they were issued when interest rates were higher, so they’re priced above face value. That’s a bad combination for taxable investors, because it means you pay tax on a high coupon and then get stuck with a capital loss when the bond matures.
An example: the iShares 1-5 Year Laddered Government Bond (CLF) would have provided similar credit risk and better liquidity than a GIC ladder, but Justin’s analysis showed it would have a negative after-tax return. CLF pays out about 4.2% in fully taxable interest, and since its yield to maturity is just 1.4%, you can expect it to suffer a significant capital loss every year.
There must be hundreds of millions invested in this tax-inefficient manner. CLF has more than $1 billion in assets, and the iShares DEX Short Term Bond (XSB) is the second-largest ETF in the country with more than $2.2 billion. Surely not all of that is held in registered accounts, which must be making the Canada Revenue Agency dance a little jig.
More tax-friendly than you think
Faced with this problem, Justin thought of strip bonds, which always trade at a discount. With strips you pay tax only on an amount equal to the yield to maturity, not on an inflated coupon. And you won’t suffer a capital loss unless interest rates spike and the bond is sold before maturity.
These tax advantages are overlooked by folks who say strip bonds should never be held in a non-registered account. It’s true they don’t generate income, but if you weren’t planning to spend the interest payments, that’s irrelevant. In any case, First Asset’s new ETF is structured to pay distributions, so you end up with the best of both worlds: tax-efficiency and cash flow.
To sum up, BXF is designed for non-registered accounts where the investor wants a fixed-income product combining the tax-efficiency of GICs and the liquidity and security of government bonds.
The only trade-off is that strip bonds have a longer duration than traditional bonds of the same maturity, so BXF (with a duration of about 3.6) will be somewhat more sensitive to interest rate movements than CLF (duration 2.5) and XSB (duration 2.8). That means a little more volatility, but if the idea is take advantage of rebalancing opportunities, that’s not necessarily a bad thing.
A strip bond ETF is a useful and innovative product, I’d say. Makes you wonder why no one thought of it before.
With an average yield to maturity of 1.6%, is there any reason for an investor to choose this fund for holding cash (for instance prior to a rebalancing) over a high interest savings mutual fund paying 1.2% over shorter periods of 3-9 months.
I can clearly see that a conservative portfolio with maxed RRSP and TFSA contributions might benefit from this product for longer term investing of fixed income assets outside of registered accounts.
Unburdened by millions of dollars,
JAH
@JAH: I’d say this product makes no sense as a place to park short-term cash. Indeed, no bond fund is useful for that purpose, because bond funds are vulnerable to short-term losses. This would be a core holding for the long term.
The ETF has not started trading yet, but do you have any concerns about bid/ask spreads or tracking errors for this new ETF?
@KayT: Yes, bid-ask spreads and tracking error are always a potential problem and something to keep an eye on. That said, if an investor is currently holding bond ETFs in a taxable account, the savings would likely be significant even with a little slippage for trading costs and tracking error.
I don’t understand the supposed magical tax friendliness.
The way I understand a strip bond, it is like a bond with an automatically reinvested (phantom) coupon at the original yield. So, if rates go down, you win a higher coupon on reinvested money, and if rates go up, you lose with a lower coupon relative to current rates.
Now, this changes nothing for taxes. The reinvested phantom coupon is taxed as interest income. This coupon is determined at the time the (strip) bond is issued and does not vary based on the market value of the bond. So, in the end, the ETF buyer is NOT taxed on the average yield to maturity of the ETF, but on the phantom reinvested coupon of the ETF’s strip bonds.
So, this ETF will be as tax innefficient as any other bond ETF for (strip) bonds issued at higher rates than current rates.
What am I missing?
@ccpfan: Lots to unpack here. First, the strip bond has no coupon, and there is no reinvestment. It’s not enough to call it a “phantom coupon” and then assume it acts like an actual coupon on a regular bond. The structure is fundamentally different. A strip bond has only a face value and a market price, and the difference between those two values is then amortized from the purchase date until maturity. So yes, the investor is taxed on the yield to maturity of the strip bond, not on any “phantom coupon.”
The other fundamental difference is that premium bonds will incur a capital loss when held to maturity, while strip bonds will not. (Bond ETFs typically sell bonds when they have one year left to maturity, but this too usually results in a capital loss.) Although the higher coupon offsets this loss on a pre-tax basis (making CLF and similar ETFs just fine in an RRSP or TFSA), interest is taxed at twice the rate of capital gains, so even if you can use that loss to offset gains elsewhere you are still way behind on an after-tax basis.
Hope this makes sense. When BXF has a full year under its belt I will run a side-by-side comparison of their after-tax returns and the results will be obvious.
This may help, too:
http://www.taxtips.ca/personaltax/investing/taxtreatment/stripbonds.htm
Thanks for the reply, but I still don’t get it. Let me illustrate with an example.
I’ll reuse your last article’s 10k at maturity strip bond. The ETF buys it on year 1 for 9,057.00. Person A buys a share of it in year 1 for 90.57, and gets taxed on 1.81 (notional interest). In year 2, interest rates go down to 1%. Accordingly, the NAV changes to 96.09 (instead of 92.38 in a 2% environment). Person B buy a share for 96.09.
If I understand correctly, persons A and B will have to pay taxes on the same amount in year 2: 1.85 (notional interest). Yet, the ETF’s yield to maturity, in year 2, is 1%. So, person B has is paying taxes on an amount that is double the yield to maturity (which is 0.96).
So, if the above holds, it seems to me that the ETF wrapper, around the strip bond, scraps its tax advantage for person B.
What did I miss or get wrong?
@ccpfan: Now I understand your question: I thought you were asking about the tax treatment of strip bonds in general, rather than about the ETF wrapper specifically. I will have to take your question to First Asset and ask them for an answer, since I don’t know enough about its structure to answer accurately. Stay tuned.
@ccpfan – I’ll take a shot at it, and I’ve also asked my accounting group on the point. I think the missing link here is the unit price in your calculation. When strips are purchased by the ETF, each strip has a yield to maturity, and the tax is accrued at that rate. The tax is calculated on the yield to maturity at the time you buy it, not on the “original” YTM of the strip. And we’re assuming that the new purchaser, Buyer B, is not buying into a pool that has had subsequent subscriptions at higher or lower YTM, in the interim. And one year later, 20% of the portfolio has matured and been reinvested in the lower rate environment you have used. So, to use your math, Buyer A invests $9,057 and receives 100 units. Buyer B, when buying it at a different yield to maturity, using the same amount of money $9,057, will only get 94.25 units. So while the amount of the distribution per unit remains unchanged, Buyer B is not paying a disproportionate amount of tax. He/she bought at a higher price, so is receiving a lower distribution amount (and lower taxable amount). The distribution per unit has remained the same, but Buyer B has fewer units. In essence, it is self correcting. I hope I’ve understood your questions correctly
Thanks
Barry
As a follow up on the last, my accounting group informs me that BXF will operate exactly the same as any other bond ETF in regards to distributions and tax. In the case of BXF, it will calculate the amount of the taxable interest accrual, make cash distributions equal to that amount (on a quarterly basis), and a buy of BXF will receive their proportionate distribution (and thus taxable income) based on the number of units they buy. Cheers
@Barry: So, that would mean my example was correct, right? If so, where is the tax advantage?
@ccpfan: When Barry says BXF is “the same as any other bond ETF in regards to distributions and tax,” he didn’t mean there’s no tax advantage with the strip bonds. He meant all individual investors in the ETF will receive the same distribution per share, just as they do with any other bond ETF.
I believe the people at First Asset are preparing some material to clarify this idea for investors. I’ll follow up as soon as I know more.
@CCP: Thanks a lot. I am sure that I am missing something important in my understanding of this new ETF; it’s just that I don’t know what. I don’t understand all the intricacies of ETFs.
I enjoy learning from your awesome blog and its comments.
@ccpfan: First off, I don’t think I fully understood your question and your example, so my apologies. In essence, I think that if BXF were to only hold a bond or bonds with one maturity and that rates declined as you state, then BXF would also develop the exact same issue that Dan is pointing out with respect to “premium” bonds. By unitizing the pool as an ETF, the subsequent purchaser would be embedding a capital loss just like any current ETF that has a portfolio of premium-to-par bonds with the attendant erosion of after tax returns.
However, I think Dan’s original point, with which I concur, is that because the strips are all being purchased at below par, BXF will come out of the gate with an inherent advantage in that its yield to maturity will equal its cash yield. And because of the laddered structure, each year 20% of the portfolio is reinvested at the then current 5 year strip rate (also below par). So the scenario you are describing is mitigated/altered by these facts, and the distributions will adjust in the future to reflect the actual portfolio yield to maturity.
Bottom line, individual purchasers of BXF, like any ETF, will have different experiences depending on when they purchase, rates at the time they purchase, when they sell, rates at the time they sell, etc…. There is no tax “magic”, there is just an inherent tax advantage relative to premium-to-par bonds, which advantage, as you correctly point out, can be eroded or eliminated (at least for some period) in a situation where there is a significant rate move and a new purchaser buys into the pool. I hope I’ve addressed your question properly this time, and apologies for the long winded response.
@Barry: Thanks! I see, now. It’s all about the yield of constituent bonds. In the current environment, where older bonds trade at big premium, the strips will have a tax advantage to be enjoyed by BXF share buyers relative to a traditional bond ETF.
No worry about the long response! I actually liked that it made me catch the important element I was missing.
Thinking about it a little more… I guess that, even in an increasing interest rate environment, BXF will continue to buy mostly below par. Wow! Now I’m impressed (and I feel dumb for asking all these questions).
@ccpfan: You absolutely should not feel dumb for asking what were actually insightful questions. Justin, Barry and I spent a lot of time puzzling this out, and I think we all learned a few things. BXF is an innovative product (as far as I know there is only one other strip bond ETF, and it’s in the US), so I’m not sure anyone can honestly say they know exactly how it would behave in an a volatile interest rate environment. So thanks for raising the issue—and thanks to Barry for replying so thoroughly.
@ccpfan: NOT a dumb question. Very insightful and something that it took me all night to get my head wrapped around. Glad I was able to make sense of it. Cheers
Fearing that my questions will seem awfully simplistic relative to ccpfan’s…
First, all else being equal, the ability to claim a capital loss is a Good Thing, provided you also have capital gains that year. So I presume the core problem with premium bonds is more the fact that you have to pay actual tax now in exchange for a “capital loss” which is merely a potential tax future savings. Do I understand this right?
If so, my main question is: Would there by any merit in buying both regular and strip bonds in such a proportion that the capital gains of the strip equals the capital loss on the regular “premium” bond? In that case, the capital loss is no longer a future potentiality, but an actual immediate tax benefit. Capital gains vanish from the equation, and you’re left just paying the tax on the interest you earned (which, granted, is still not ideal).
I hope I haven’t missed the whole point here.
@Patrick: It’s great to offset capital gains with losses, but the problem with premium bonds is that interest is taxed at twice the rate of capital gains. Let’s use one-year bonds to keep it simple:
– Bond A is sold at par ($100) with a 5% coupon, so it pays you $5 in interest and there is no capital gain or loss at maturity. If you’re in the 40% tax bracket, your tax bill is $2.
– Bond B has a 6% coupon and is sold at a premium ($101) so it yields 5%. After one year you collect $6 in interest and the bond matures with a capital loss of $1, so your total return is $5. That’s the same as the first bond, but only before taxes. With Bond B you would pay $2.40 in tax on the interest. And even if you could use that capital loss to offset a gain somewhere else, you’d only save $0.20 in tax, because your capital gains are only taxed at 20%. So you’re after-tax return is $0.20 lower than with Bond A.
So that’s why capital losses on bonds are not a good thing. (Note that in an RRSP or TFSA Bonds A and B are identical.)
As for your second question, that wouldn’t work. Strip bonds don’t earn capital gains as a rule: they just mature at par. (They could earn capital gains if interest rates fell and you sold them before maturity, but that’s true of any bond.) It is possible to harvest losses opportunistically and carry them forward to offset future gains: indeed, this is smart portfolio management. But it takes constant monitoring: you can’t just by two offsetting securities.
Hope this helps.
@Dan: Ah I see what you mean.
Really, bonds are fairly straightforward math, but I find it hard to develop any intuition for them somehow. You just have to crunch the numbers.
@CCP: My apologies for returning to an old topic, but I need a product “designed for non-registered accounts where the investor wants a fixed-income product combining the tax-efficiency of GICs and the liquidity and security of government bonds” and went through the mental gymnastics again of figuring out how BXF undertakes to fulfill the above mandate. It certainly is complicated to parse.
I have 2 questions: The first is how the quarterly distributions are treated tax-wise. As you explained it, the distributions are the amortized amounts of the yield, and…
“Their yield is calculated by comparing the bond’s purchase price and its future value”.
The difference between purchase price and future value would seem to me a capital gain; so why is an amortized quarterly payment of the initial sum treated as interest at full marginal rate? Or is there no answer — that’s just what the CRA demands?
The second question is that if the whole point of having this vehicle is to have a tax efficient (in a non registered account) method of having an income generating asset allocation that may be tapped to provide funds for rebalancing at appropriate times, does that not also require the ability to be efficient when treated as a short term source of funds? But you have stated:
“@JAH: I’d say this product makes no sense as a place to park short-term cash. Indeed, no bond fund is useful for that purpose, because bond funds are vulnerable to short-term losses. This would be a core holding for the long term.”
This appears to contradict the short term requirement I was looking for — or am I taking your comment out of context, and there is no contradiction?
Still on my steep learning curve, I was wondering if “Oldie”‘s 11 Feb 2014 query was ever answered, for it seems that I have the exact same requirement as he: a product “designed for non-registered accounts where the investor wants a fixed-income product combining the tax-efficiency of GICs and the liquidity and security of government bonds” ?
Regards
Donihue
@Donihue: The product “designed for non-registered accounts where the investor wants a fixed-income product combining the tax-efficiency of GICs and the liquidity and security of government bonds” is the strip bond ETF described in the post. Since the launch of BXF, two other options have appeared:
https://canadiancouchpotato.com/2014/02/13/new-tax-efficient-etfs-from-bmo/
https://canadiancouchpotato.com/2014/05/08/a-tax-friendly-bond-etf-on-the-horizon/
@CCP: Many thanks for your very prompt reply. I will study the links you mention. I am inching my way towards understanding what products meet my needs.
In the meantime, for what it is worth, I believe that @Oldie’s points about tax treatment of the FirstAsset strip bond ETF BXF are implicitly answered by FirstAsset itself via a spreadsheet it produces which shows that for BXF.A distributions are approx 75% income and 25% return of capital, whilst for BXF the ratios are approx 84% and 16% respectively.
Regards
Donihue
@Donihue: I actually was unclear in two areas: Firstly regarding the tax treatment of the returns (and also, regarding the approximate range of expectations of the returns). This is gradually becoming more clear with the 2013 distributions spreadsheet.
The other point I was unclear on was in light of the comment:
“@JAH: I’d say this product makes no sense as a place to park short-term cash. Indeed, no bond fund is useful for that purpose, because bond funds are vulnerable to short-term losses. This would be a core holding for the long term.”
I was initially a little puzzled by this, but I think I have sorted it out. By short term cash, I think he means if you have X thousand dollars and you know that you will make a down payment on a house purchase within the next 3 years, say, then a short term investment in a bond fund is unwise because in that time frame the value of your investment is vulnerable to losses if prevailing interest rates rise in the interim; the projected value in a GIC or a high interest bearing account, on the other hand is guaranteed.
My initial puzzlement was why, if it was vulnerable to short term losses, it would work as a component of a diversified portfolio such as the recommended Global Couch Potato Portfolio (40% Canadian Bond ETF, 60% Canadian, US and World Equities ETFs in equal portions) which may very well be aimed at a 15 year or more horizon, but would possibly require cash coming out of the bond portion in the short term for rebalancing. I think the answer is that as part of a larger portfolio, under the usual circumstances in which you need to purchase more equities because their value has dropped, most often bond values have risen. And even if they haven’t actually risen, the range of expected fluctuation of bond values is much smaller than the fluctuations possible in equity values, thus fulfilling a “flywheel” function in maintaining the overall value of the portfolio on a tighter range of limits than if there were no bond component. @CCP: am I on track in my reasoning?