John Heinzl was kind enough to quote me in his Investors’ Clinic column titled Why only millionaires should invest in bonds directly:
Now, it’s true that bond ETFs typically roll over holdings one year before they mature, because at this point these securities are considered money-market instruments. But in an environment of rising interest rates, a bond ETF that follows a sell-before-maturity policy would buy new, higher-coupon bonds sooner than an identical portfolio of bonds that held to maturity, and the higher income would make up for any capital loss incurred as a result of selling early, said James Hymas, a fixed-income expert and president of Hymas Investment Management.
Bond ETFs have several advantages, he points out. Because ETFs buy in volume, they get much better pricing than retail investors could obtain through their broker, and this pricing advantage for most ETFs will outweigh the management expense ratio. Bond ETFs also provide instant diversification. The notion that bond ETFs don’t mature and should therefore be avoided makes no sense, he said.
“Anybody investing less than $1-million in bonds should do it through ETFs,” Mr. Hymas said. “If you have more than $1-million, then you can talk about buying individual issues, but if you have less than $1-million you’re either going to have poor diversification or poor pricing, perhaps both.”
The Globe’s website shows one comment worth addressing:
I disagree entirely. A bond costs $5K to buy and nothing to hold. For 70K you can set up a 7 year ladder with one bond maturing every 6 months. You hold every bond until it matures, reinvesting each matured bond with all the accumulated interest in the account in a new 7 year bond. When you retire you can use the interest payments as income if you like. You will earn the same as a second OAP, without the clawback.
If you don’t have 70K yet, you buy one $5K 7 year bond every 6 months for the next 7 years to set up.
It is not rocket science, I have been doing it for 15 years. The pros like Hymas hate it because, apart from the small fee when you buy a bond, you pay no fees at all.
A Bond ETF will have a management expense ratio of 25-35bp. The bid-offer spread on seven year bonds purchased in amount of $5,000 will almost certainly exceed this. Additionally, there will be costs associated with further trading, unless you spend amounts exactly equal to your coupon income.
Another commenter suggested:
It is certainly possible to create your own bond ladder as you describe, and there are benefits to that. But the costs are also hidden by the lack of transparency and liquidity in the smaller denominations. Perhaps $1M is overkill, but probably $25,000 is a practical trade-off between price/cost and yield.
I just checked a broker screen and spreads on medium-term corporates are about 35bp for quantities of $1,000. Sorry – I don’t know precisely where the price breaks are, or how much better pricing is at the 25,000 level.
I suggested $1-million because then you can buy 20 bonds in lots of $50,000. The ETF also has the advantage of greater liquidity, as well as freeing you from the tender mercies of your custodial broker’s bond desk should you need to sell, which are often not very tender.
Additionally, note that most retail bond desks will make only a very limited number of names available to investors – proper diversification of a bond portfolio will always be very difficult for retail, even those who do have $1-million.
For more on this theme – which addresses in more detail the ladder / ETF decision – see my March 2010 publication from the Advisors’ Edge Report.
Update, 2011-7-8: One commenter made an excellent point:
And have you ever tried to sell a bond? Sure, if you’ve laddered everything nicely you shouldn’t need to. But sometimes $hit happens and you need money unexpectedly. I’ve tried twice, once through W’house, once through e-trade. It took them days to get back to me with a (horrible) price, by which time I’d raised cash elsewhere. If you’re buying bonds directly, be really, really sure you’ll hold them to maturity.
One common theme in the comment is the view that holding bonds directly is better because “Transaction fees and the spread are a one time cost whereas the MER is forever.” In fact, transaction fees and the spread are a recurring cost, paid anew every time you roll a rung of the ladder. And, as stated in my post above, the spread for medium term corporates in small quantities at one broker is about 35bp per annum – when you express the spread as a difference in yield.
Update, 2011-7-7: See also discussion at Financial Webring Forum.
I missed this when it came out.
The biggest reason NOT to buy bond ETFs is the after-tax return is terrible! The average ETF has premium bonds with average prices in the $110+ range, so will generate capital losses over time and excess income. Of course, non-taxable investors like RRSP/RRIFs don’t need to worry about this, but for taxable investors, the after-tax return can be NEGATIVE.
For TD Waterhouse, the spread on a 8-10 year bond is 35 bp for corporates and about half that for provincials. The cost of purchase is half this, or 9-17 bp if held to maturity. This is for amounts of $5K face or so. At $25K face purchase size, the bid-ask on corporates falls to 28 bp for an annual “cost” of 14 bp. 10 years ago, TDW used to be the best bond game in town. Now the selection is much worse and the benefits to large orders
i-Trade is a lot better, with a quoted bid-ask spread of 8 bp on corporates for a 4 bp cost (plus $20 trading fee — about 5 bp on a $5K order or 1 bp on a $20K order). I-Trade is probably the best game in town right now with a good selection and reasonably narrow spreads for most small investors.
Given the paucity of par or discount bonds around these days, a small taxable investor who intends to hold to maturity is probably best off with STRIPs. They are at least tax neutral (though you pay the tax before you get the interest), and spreads aren’t a lot worse than bonds.
Thus, for many investors, even non-taxable ones, a direct bond purchase at i-Trade is probably superior to an ETF with a 35 to 50 bp MER.
I hadn’t realized iTrade was so good on pricing. What’s their selection like?
Hi James,
(First, sorry again for commenting in old postings, but it wasn’t clear to me what more-recent blog might be more appropriate.)
What do you think of building a investment-grade fixed-income ladder using (mainlly) high-quality split shares with a hard retraction date? CGI.PR.D is rated Pf-1(low) – how does that match up with bond ratings? Does it correspond to A(high)?) What about PVS.PR.xx which are rated Pf-2(low)? Does that correspond to BBB(low)?
Today, my iTrade account indicates that TMX bonds (A(high)) maturing Oct-2023 (close to the June-2023 CGI.PR.D maturity) have a bid-ask of 3.03499 /2.95011. (On top of the spread there is also a flat rate $25 per transaction fee.) On the other hand, at current bid, CGI.PR.D offers a YTM of approximately 3.6%, a nice difference.
I readlize that the selection of investment grade splits is relatively limited, so some bond purchases may be needed to fill in the gaps.
On a related note, you have recently reiterated your view that some lower-rated split shares actually should have a very good recovery rate in the event of default. Wouldn’t this also be true of bonds of equivalent (default) credit rating, or is the recovery generally lower?
Thanks as always for your enlightening articles!
What do you think of building a investment-grade fixed-income ladder using (mainlly) high-quality split shares with a hard retraction date?
I hate ladders. They’re too restrictive, which means that they’re too expensive, since you will have to buy something ‘because it fits’ rather than ‘because it’s cheap’. If you want to target any particular duration, you’re better off just targeting that duration and looking for the cheapest set of prefs that matches that duration while satisfying quality and diversification constraints.
Also if, encouraged by comparisons to bonds, you overweight Splits in general you may experience grief should we see another market crash. My ballpark, approximate, target for Asset Coverage is 1.5:1, which mean that you have direct exposure to equity market losses after a decline of 33%.
CGI.PR.D is rated Pf-1(low) – how does that match up with bond ratings? Does it correspond to A(high)?) What about PVS.PR.xx which are rated Pf-2(low)? Does that correspond to BBB(low)?
P-1(low) = A / A- ; P-2(low) = BBB- (see Table 22 of the link).
I readlize that the selection of investment grade splits is relatively limited, so some bond purchases may be needed to fill in the gaps.
I do not recommend that preferred shares – of any type – comprise more than 50% of a fixed-income portfolio. You need liquidity as well as income (otherwise you wouldn’t own fixed income in the first place) and a liquidity freeze is more likely to hit preferreds than bonds – even with the OSC eagerly preparing to degrade the corporate bond market.
On a related note, you have recently reiterated your view that some lower-rated split shares actually should have a very good recovery rate in the event of default. Wouldn’t this also be true of bonds of equivalent (default) credit rating, or is the recovery generally lower?
Loss Given Default for senior bonds is generally given a ballpark figure of 60% (recovery rate of 40%), but of course it varies all over the place. The LGD of the Agencies (Fannie Mae and Freddie Mac) was tiny, which really screwed investors with CDS recovery locks.
And be careful with that 40% figure!
Pricing Partners Highlights Issues with Recovery Rate Assumptions
CDS Implied Probability of Default – Be Careful
ISDA Standard CDS Converter Specification
I wouldn’t say that “split shares actually should have a very good recovery rate”; I will say they CAN have a very good recovery rate. It will depend, of course, on their underlying portfolio. It is only in contrast to preferred shares of operating companies that I am willing to say “should”, because preferred shares of operating companies will generally recover a big fat zero.