Chalk up another one for ETF innovation in Canada. This week, First Asset launched three new bond ETFs that are the first of their kind in North America. And unlike some innovations in the ETF world, this one doesn’t involve any exotic, expensive and opaque derivatives. On the contrary, it uses a simple fixed-income strategy that’s been around for decades.
The new First Asset funds use what’s called a barbell strategy, which involves holding equal amounts of short-term and long-term bonds, with no allocation to intermediate maturities. According to the index methodology, the ETFs hold 50% of their assets in bonds with maturities between one and two years, and 50% in bonds that mature in 10 to 20 years. The short-term bucket includes both regular fixed-coupon bonds and floating-rate notes, which have coupons tied to prevailing rates. (The benefit of floating-rate notes is that they have extremely low interest rate risk: if rates rise, their prices remain more or less stable.)
Best of both worlds
So why would you use a barbell rather than simply holding a broad-based bond index fund? The goal of the strategy is to balance offense and defense: the long-term bonds give you higher yield, while the short-term bonds protect you from rising interest rates. To get more technical, the barbell strategy lowers the portfolio’s duration without sacrificing yield. In other words, it attempts to deliver better risk-adjusted returns.
Duration is an approximate measure of a bond fund’s sensitivity to changes in interest rates: the longer the duration, the greater the loss in the fund’s value if interest rates rise (and the greater the increase if yields fall). If you’re concerned that rates will move higher, you may want to keep your duration short, but that usually means accepting lower yield. However, with the barbell strategy that isn’t the case. Look at how the new First Asset ETFs stack up against their iShares counterparts:
Yield | Maturity | Duration | |
iShares DEX Universe Bond (XBB) | 2.27% | 9.8 | 6.9 |
First Asset DEX All Canada Bond Barbell (AXF) | 2.20% | 8.4 | 5.4 |
iShares DEX All Government Bond (XGB) | 1.99% | 10.2 | 7.3 |
First Asset DEX Gov’t Bond Barbell (GXF) | 1.98% | 8.3 | 5.3 |
iShares DEX All Corporate Bond (XCB) | 3.02% | 8.8 | 6.1 |
First Asset DEX Corporate Bond Barbell (KXF) | 2.91% | 9.0 | 5.6 |
As you can see, the barbell ETFs are almost identical in yield to maturity, similar or lower in average term to maturity, and significantly shorter in duration. (If you’re inclined to get geeky, this means the barbell portfolio has higher convexity.)
The shape of things to come
We all know there is no free lunch in investing, so what is the trade-off? The important idea is that the relative performance of a barbell strategy will depend on how the shape of the yield curve changes.
Recall that the yield curve describes the difference between short-term, intermediate, and long-term interest rates. Most of the time, the curve slopes upwards, which means that 20-year bonds have higher yields than 10-year bonds, which have higher yields than five-year bonds, and so on. But rates don’t all move in lockstep: sometimes those at the long end rise or fall while the shorter end remains unchanged, or even moves in the opposite direction. The pattern of these changes will determine whether the barbell outperforms or lags a traditional broad-based strategy:
- The best possible scenario for the barbell would see long-term rates fall and short-term rates rise. In other words, the strategy will thrive when the yield curve flattens.
- The worst environment for the barbell is one where the yield curve steepens: that is, long-term rates rise while short-term rates fall.
- If rates rise or fall in parallel (in other words, short and long-term rates move in the same direction by approximately the same amount), then things get a little more complicated. You may be slightly better off with a conventional bond strategy if rate changes are modest—say, no more than one percentage point. However, if yields spike or plummet across the board, then the barbell can be expected to outperform. Remember, however, that if interest rates move sharply higher, then “outperform” simply means “lose a bit less.”
In the end, a barbell strategy is a bet that interest rates will behave in a specific way. But it’s not a particularly large bet: during all but the most extreme periods, the relative performance of the two strategies should not be dramatically different.
If you do decide to use this strategy, the First Asset ETFs are an ideal way to implement it. Sure, you could build your own barbell using existing ETFs, but it makes no sense to do so. The First Asset funds carry fees of just 0.20% to 0.25%, the same or lower than their competitors, and using a single fund means automatic quarterly rebalancing and fewer trading commissions.
Cheaper than XLB and XSB!
One reason you may want to create your own barbell is you could use high interest TFSA yielding up to 3% now which work the same as floating notes as the short end and a XLB or XRB as the long end.
That said with a YTM after MER of 2.82% on XLB I personally would prefer the 3% High Interest Savings and avoid the barbell all together if you had sufficient TFSA room to cover your desired fixed income allocation.
@RJ: What institutions are offering TFSAs at 3%?
Peoples Trust offers a 3% TFSA and they are covered by CDIC insurance.
Interesting post.
A relatives advisor had her in a shorter term ladder of corporate bonds and GICs, 1-7 years, with one what I think I remember he called an “outlier” long bond which was 15-20 years and it was either provincial government or federal.
For simplicity in a higher fixed income portfolio I like a 5 year ladder of maximum interest GICs that are CDIC insured and use XSB and CBO/CLF for liquidity/rebalancing and hold XBB all the time. XRB is long duration and inflation indexed.
Just an overall comment: If you’re looking to target specific spots on the yield curve rather than just holding something like XBB, it’s really important to look at the details of the ETFs. Short, long—these words can be vague.
A combination of XSB and XLB is actually not a good counterpart to the barbell strategy used by the First Asset funds. XSB has an average term to maturity of about three years, whereas the short end of the barbell includes only maturities of one to two years (it also includes floating-rate notes). And more than two-thirds of XLB is in bonds with maturities of more than 20 years, whereas the long end of the barbell is just 10 to 20 years. So you need to be sure you’re really getting the exposure you want.
Another thing worth mentioning: a few people have mentioned XRB on the long end. I think real return bonds are great portfolio diversifier, which is why I recommend them in the Complete Couch Potato. But they have quite a different risk profile compared with nominal long-term bonds. They are not only subject to interest rate risk, but also to changes in inflation. For that reason, the two asset classes may behave quite differently. Again, just make sure you know what to expect.
Great post Dan. Will you be incorporating these in your model portfolio? I am in the process of setting up a couch potato porfolio and have started to buy VSB. The next step would be to buy XBB and XRB. But the reading the above it makes sense to buy the funds from asset class as it will be less costly and easier to implement.
@Karim: I don’t plan to add these ETFs to my model portfolios. Broad-based index funds are still my default choice.
@RJ:
I don’t get it — you can get get 3% if you hold it in a TFSA with these guys? So you would use in in a couch potato type portfolio in the same place as you would otherwise use your Bond ETF (barbell, laddered or otherwise), and they would let you also have proportional Equity ETFs in the TFSA, and annually or whenever you plan to, you can re-balance your portfolio, which includes “selling” by somehow redeeming some of the “savings” for cash to buy more equity or possibly the opposite — “buying” more 3% savings-contract, if it is still available? They would let you do that? You’re not locked in?
@Oldie
You cannot hold your equities at People’s trust. I hold mine at TDW in Vanguard and iShares ETFs (as well as TD eFunds when I am making a smaller contribution).
It is a high interest savings account so you are not locked in. I make most of my contributions in January when I have new RRSP, RESP and TFSA room. So just before December 31 I withdraw cash from my TFSA (that way I get the contribution room back days later) so that I can buy more equities. I then rebuild my cash holding in my TFSA over time. Eventually I hope to be able to instead accumulate my January contributions in a non-registered account and then make them in kind in. But for now I either allow my equity allocation to tick up a couple of percent in January until I can save more to replace the withdrawn to TFSA money or I can hold XSB or TD Bond Index eFund in my RRSP and RESP until my TFSA is replenished.
If my rebalance bands are hit at other times in the year I contribute to my TDW TFSA (I have not yet fully maxed my TFSA room, but RRSP and RESP are maxed at this time).
Alternatively if there was a huge crash in equities People’s trust have committed to out transferring the cash to another institutions TFSA with de-registering it at no cost. This would allow me to re-balance my TFSA to equities at any time of the year due to a market crash.
To be clear my all fixed income is all held in my wife and my TFSA and a whole life policy bought for me when I was a kid that is set to premium offset. My RRSP and RESPs (I have 3 very young kids) are all 100% equities. This leaves me at 85/15 which I am comfortable with given the fact that my pension, human capital, and paid off house are not factored into the 85/15 split.
@Oldie
I should add that by choosing to hold my fixed income in a high interest savings I am trading any chance of capital gains for the fact that the holding is 100% liquid at face value and therefore has a duration of zero and CDIC guarantee.
Given that the yield of the option I have chosen is higher than the YTM of any bond ETF I can find I am very comfortable with that tradeoff because I am getting a higher return (unless interest rates continue to fall at a lower risk)
@RJ
I looked at the Peoples Trust website. Their savings account is 2.1% so it is like they are assuming the savings in TFSA would not be accessed much and therefore are giving it longer duration rates (their 4 year GIC is 3%).
It makes sense given the low real long duration yields to hold such a cash account because it will eventually rise with rising rates. Cash has optionality and interestingly is, for most periods of time, inflation sensitive if it is in an account that floats with short term rates that does not pay tax at all such as TFSA. Cash, as T-bills, can also actually be quite highly correlated with inflation and in fact outperform stocks and bonds (but not necessarily after tax – not an issue in tax reg accts) only when inflation is high which are not insignificant time periods. I always hear the mantra that cash is bad because it is not inflation sensitive but this is a myth if held in a tax free account and if inflation is high, because of how badly stocks and bonds perform during high inflation levels and high CPI rates of change. See here: http://www.fpanet.org/journal/CurrentIssue/TableofContents/TreasuryBillsandInflation/
I did a study once comparing Tbill rates with CPI and found that in Canada during high inflation the T-bills just kept up with inflation, as inflation moderated T-Bills far outperformed and even during low inflation the T-Bills beat inflation, at least as measured by our CPI. Overall Tbills exceeded inflation for most time periods since 1974.
So if the Peoples Trust account rate floats as a function of shorter term government rates there are favourable odds it at least will keep up with inflation if in a tax free account and this account is highly sensitive to short term rates, unless inflation is very high (between 1974 and 1983 the average change in year over year CPI was a whopping 9.43%).
Peoples Trust may offer TFSA at 3% but there’s no guarantee it will be like that in the near future.
I remember ING Direct offering 3% tax-free “high interest” savings account (“high interest” being relative) not long ago. It’s now at 1.4% !
At least, like RJ says, you are not locked in.
Can I play devil’s advocate from the point of view of the statistic nerd in all of us for a moment? First, I admit that I don’t understand how the Maturity and Duration calculations are done for the bond funds, but as I understand it, both calculations involve a weighted average of the individual bonds held in the fund.
The average (or mean) is a very useful statistic for distributions that are normal (as in Gaussian) or approximately normal. This fund has a distribution of bond maturities that is expressly designed NOT to be normal, since it has a large number of very short and very long maturities, with nothing in between. The distribution may be described as bimodal. It’s less appropriate to calculate an average for this distribution, if one is looking for a single meaningful statistic to describe it. The median similarly fails.
I’m not saying the analysis here is wrong, just that I’m not entirely convinced that the calculations of the maturity and duration for the First Asset funds are accurate.
@Andrew D: The figures for the new barbell ETFs were supplied directly by First Asset a couple of weeks ago. (I wanted to wait until the funds were live before I posted this piece.) I know Barry Gordon at First Asset keeps an eye on this blog, so I will let him respond in more detail.
Thx Dan. @Andrew D, you are correct that maturity and duration calculations, as well as yield to maturity, current yield and convexity calculations, are done on a weighted average basis. With respect, I disagree that the bimodal nature of the strategy makes the weighted average stats less meaningful. In fact, they are particularly germane, as they highlight the advantages (and disadvantages) of the strategy over one that, for example, targets a specific maturity band.
A barbell strategy, in its most basic form, uses a short term bond and a long term bond to match the duration of a single medium term bond. The barbell strategy will have the same duration, and slightly lower yield to maturity (the cost of convexity), but will have much higher convexity due to the use of the long bond on one end of the barbell. What we are doing is, instead of using a single bond at each end, to use many bonds to achieve diversification and a scalable ETF. The fact that we use many bonds doesn’t change the analysis.
I totally hear your criticism that in trying to extend the comparison to a more normal distribution may be less relevant. The shape of the yield curve and changes to it, will affect the outcomes as much as individual duration or maturity stats. However, the mathematical calculations are correct. The ETFs will behave the way a barbell portfolio with a weighted average duration of X should behave in a parallel shift, a flattener, or a steepener. Our point is not that the barbells are inherently better, just that they offer a different risk adjusted way to invest in a portfolio of bonds. We believe its a good addition to a portfolio, and with the balanced approach, provides some real advantages in the current environment.
@Barry: Many thanks for taking the time to respond.
Thanks for the responses — I’m completely willing to accept the argument that the barbell type portfolio has been widely used for many years and is known to behave in certain ways.
@CCP:
A question regarding newly launched funds, such as the First Asset Barbells:
I assume that as an ETF you just purchase as usual, through discount broker account etc. Regarding volume, on going to their website, KXF, for instance, one finds listed a total net asset value of about 2.5 Million, and a trading volume today of 1000. Is this number low enough to warrant concern about liquidity, and is this number expected to rise with time?
@CCP:
Further to the above, I was trying to confirm for myself the Yield, Duration and Maturity values you listed above for the various First Asset Barbell Bond ETF’s, but on the First Asset Website, navigating to the individual ETF info page only seems to disclose the current price, the top daily holdings (with the individual yields), but not the info I was searching for. Where would I find the Yield, Duration and Maturity for these Bond ETF’s?
@Oldie: In general, low trading volume is not necessarily a problem with ETFs. What matters more is the liquidity of the underlying holdings. That said, it always makes sense to look closely at the bid-ask spread when you go to trade, and to use a limit order so you won’t be surprised.
The First Asset funds are so new that they may not have all their specs published online yet. (I got them directly from the company.) But in a month or two I would expect that information to appear on the website and fact sheets. In the meantime, I’m sure they will answer your questions if you email them.