Archive for the ‘Reader Initiated Comments’ Category

ytc_resets.xlsx : Slight Modification

Wednesday, September 27th, 2023

I have recently been discussing the question of yield and forecast income from Malachite Aggressive Preferred Fund with a client, and as part of that referred him to the Yield Calculator for Resets so he could see for himself why the projected income from the fund was so much higher than the current income.

As part of that, I had to explain that HIMIPref™, my analytical software, uses semi-annual compounded yield, which is a higher number than the quarterly compounded yield calculated by the spreadsheet. And my income projections use HIMIPref™ calculations. The more I looked at my explanation, the more it looked like bafflegab and handwaving.

So, in order to reduce the complexity of this explanation in the future, I have added a display field on the spreadsheet showing the yield as the semi-annual compounded value (for comparability with bonds) as well as the quarterly compounded value (applicable only to instruments that pay quarterly).

AX Preferred Unit Conversion is (Probably!) a Taxable Event

Wednesday, September 27th, 2017

Assiduous Reader JB writes in and brings to my attention a nuance in the conversion of Artis preferred units that had previously escaped me:

You have probably dealt with this question, so I apologize if that is the case.
I own a small position in Artis Preferred Series C. It is a $US pref. (I own lots of other preferreds.)
In the prospectus it states (I’m paraphrasing) that the CRA would consider conversion to the D shares a taxable event. I’m at a loss to determine why this is the case, since in another 5 years, I could convert back, should that be my choice.
Is this just an anomaly, or is this really the CRA’s position?

Well! First of all, let’s have a look at the prospectuses (not directly linked because rights are owned by the Canadian Securities Administrators, and why would they allow convenient access to public documents of interest to investors?)

Preferred Units, Series A (AX.PR.A) Prospectus on SEDAR under “Artis Real Estate Investment Trust Jul 25 2012 20:10:04 ET Prospectus supplement – English PDF 288 K”

In general, a disposition or deemed disposition of a Series A or Series B Unit will give rise to a capital gain (or a capital loss) equal to the amount by which the proceeds of disposition, net of any reasonable costs of disposition, exceed (or are exceeded by) the adjusted cost base of the Series A or Series B Unit, as the case may be, to the Preferred Unitholder. In the Ruling, the CRA expresses the preliminary view that the reclassification of Series A Units as Series B Units (or Series B Units as Series A Units) would likely result in a taxable disposition at that time. In such circumstances, a Preferred Unitholder will generally be considered to have disposed of the reclassified Preferred Units for proceeds of disposition equal to the fair market value of the Preferred Units into which such units are reclassified.

Preferred Units, Series C (AX.PR.U) Prospectus on SEDAR under “Artis Real Estate Investment Trust Sep 11 2012 16:24:11 ET Prospectus supplement – English PDF 287 K”

In general, a disposition or deemed disposition of a Series C or Series D Unit will give rise to a capital gain (or a capital loss) equal to the amount by which the proceeds of disposition, net of any reasonable costs of disposition, exceed (or are exceeded by) the adjusted cost base of the Series C or Series D Unit, as the case may be, to the Preferred Unitholder. In the Ruling, the CRA expresses the preliminary view that the reclassification of Series A Units as Series B Units (or Series B Units as Series A Units) would likely result in a taxable disposition at that time and the same consideration will apply on a reclassification of Series C Units as Series D Units (or Series D Units as Series C Units). In such circumstances, a Preferred Unitholder will generally be considered to have disposed of the reclassified Preferred Units for proceeds of disposition equal to the fair market value of the Preferred Units into which such units are reclassified.

Preferred Units, Series E (AX.PR.E) Prospectus on SEDAR under “Artis Real Estate Investment Trust Mar 14 2013 12:50:30 ET Prospectus supplement – English PDF 298 K”

In general, a disposition or deemed disposition of a Series E or Series F Unit will give rise to a capital gain (or a capital loss) equal to the amount by which the proceeds of disposition, net of any reasonable costs of disposition, exceed (or are exceeded by) the adjusted cost base of the Series E or Series F Unit, as the case may be, to the Preferred Unitholder. In the Ruling, the CRA expresses the preliminary view that the reclassification of Series A Units as Series B Units (or Series B Units as Series A Units) would likely result in a taxable disposition at that time and the same consideration will apply on a reclassification of Series E Units as Series F Units (or Series F Units as Series E Units). In such circumstances, a Preferred Unitholder will generally be considered to have disposed of the reclassified Preferred Units for proceeds of disposition equal to the fair market value of the Preferred Units into which such units are reclassified.

Preferred Units, Series G (AX.PR.G) Prospectus on SEDAR under “Artis Real Estate Investment Trust Jul 22 2013 13:34:56 ET Prospectus supplement – English PDF 304 K”

In general, a disposition or deemed disposition of a Series G or Series H Unit will give rise to a capital gain (or a capital loss) equal to the amount by which the proceeds of disposition, net of any reasonable costs of disposition, exceed (or are exceeded by) the adjusted cost base of the Series G or Series H Unit, as the case may be, to the Preferred Unitholder. In the Ruling, the CRA expresses the preliminary view that the reclassification of Series A Units as Series B Units (or Series B Units as Series A Units) would likely result in a taxable disposition at that time and the same consideration will apply on a reclassification of Series G Units as Series H Units (or Series H Units as Series G Units). In such circumstances, a Preferred Unitholder will generally be considered to have disposed of the reclassified Preferred Units for proceeds of disposition equal to the fair market value of the Preferred Units into which such units are reclassified.

So in each case the company has warned of a preliminary view by the CRA that conversion is a taxable event, which all appears to be based on the view they took when examining the first issue. Of course, it’s only preliminary, but to those of us who are unwilling to spend six figures discussing the matter in tax court, that counts as definitive.

As to why this should be the case … I simply don’t know. I suspect it has a lot to do with the idea that (from the AX.PR.G prospectus):

The Canadian federal income tax considerations that may arise in connection with the acquisition, holding, disposition or reclassification of preferred units of a trust are, in some respects, materially different from the acquisition, holding, disposition or exchange of preferred shares of a corporation.

“REIT Exception” means the exception from the SIFT Rules available to a SIFT trust which satisfies a series of conditions relating to the nature of a SIFT’s revenue and property, as more particularly described below under “Principal Canadian Federal Income Tax Considerations – SIFT Rules and REIT Exception”;

“SIFT Rules” means the amendments to provisions of the Tax Act proclaimed in force on June 22, 2007, as amended, that implement the changes announced as part of the Tax Fairness Plan proposed by the Minister of Finance (Canada) on October 31, 2006 which modify the tax treatment of “specified investment flow-throughs”, including publicly traded income trusts and limited partnerships, and the tax treatment of their unitholders in the manner described below under “Principal Canadian Federal Income Tax Considerations – SIFT Rules and REIT Exception”;

The balance of this summary assumes that Artis qualifies as a mutual fund trust and will continue to so qualify at all material times. If Artis were not to qualify as a mutual fund trust, the income tax considerations described below would, in some respects, be materially different.

… but this is getting into arcane interpretations of tax law in which a simple Portfolio Manager such as myself should take the view that anything he says will be wrong. However, I must say that I am surprised that Artis did not highlight this unusual nuance in its notice of extension for AX.PR.A.

Calculator: FixedResetPremium Tax Effects

Tuesday, April 11th, 2017

Assiduous Reader prefhound recently commented:

With recent strength in the Pref market, some Fixed Resets are priced north of $27 with YTW of 2-4%. What is your take on how sustainable that is and how far up they could go – north of $28?? Negative YTW??.

Two Examples are:
BPO.PR.C $27.35 YTW (first call) of 3.73% when the other BPO fixed resets average 4.86% (including BPO.PR.E, which is also likely to be called).
MFC.PR.O $27.61 YTW (first call) of 2.4% when the other MFC fixed resets average 3.99% (including MFC.PR.R, which is also likely to be called).

I had been thinking of highlighting this, but it took the comment to rouse me from my lethargy.

The interesting thing about FixedResets with very large premia is that there will be some investors who should definitely not hold them in taxable accounts due to differential tax rates. For most taxable investors a normal yield calculation will be just fine, since tax payments on larger-than-normal dividends will be offset by a recovery of taxes on the capital loss on the (presumed) call date – but this approximation is not exact and at worst can be completely wrong.

Some investors might be sitting on massive capital losses; an additional capital loss expected in the future might not be claimable immediately or, in the worst case scenario, at all. These problems were discussed in the post Tax Impact on FixedResetPremium Yields; and John Heinzl was kind enough to quote me in the Globe in his article Beware the tax trap of these tempting preferreds.

A long time ago I published a spreadsheet automating the calculation of tax effects on these issues; I’m pretty sure I noted the link in PrefLetter, but I don’t believe I ever posted about it on PrefBlog.

The calculator is an Excel Spreadsheet and is linked in the right-hand navigation panel under the heading “Calculators”.

So let’s look at four issues – the two highlighted by Prefhound and the two highest priced FixedResets:

Attribute
Attribute BPO.PR.C MFC.PR.O RY.PR.R CWB.PR.C
Bid Price 27.30 27.26 27.50 27.45
Call Price 25.00
Settle Date 2017-4-11
End Date 2021-6-30 2021-6-19 2021-8-24 2021-7-31
Quarterly
Dividend
0.375 0.35 0.34375 0.390625
Cycle 3 3 2 1
Pay Date 30 19 24 30
Include first div? Yes Yes Yes Yes
Reset Date 2021-6-30 2021-6-19 2021-8-24 2021-07-31
Q’ly Div after reset 0.39125 0.378125 0.3675 0.409375
Marginal Div Tax 29.52%
Marginal Cap Gain Tax 23.20%
Results  
Non-Taxable 3.68% 3.36% 3.21% 4.07%
TaxableClaimLoss 2.44% 2.22% 2.10% 2.70%
TaxableNoClaim 1.98% 1.76% 1.62% 2.22%

Tax Data is from Ernst & Young’s calculator, Ontario, 2017, taxable income of $150,000. “Dividend Rate after reset” has been input according to a constant GOC-5 yield of 1.08%, but is irrelevant to the calculation.

So to get back to Prefhound‘s questions: is this sustainable? Well not in the medium- to long-term, obviously, because one must assume that these high-spread, high-price issues are going to be called at the first opportunity. And one must also anticipate the price dropping towards 25.00 with every dividend paid. But the yields are probably sustainable – there are some investors who view issues of this type as substitutes for GICs, given the high call probability, and they’re just fine with 2%+ yields. Could these issues go over $28? Well, I won’t say anything’s impossible, but I consider it unlikely. A lot of people really don’t like paying such a high premium.

eMail To A Client

Saturday, August 1st, 2015

There has been a very steep decline in the Canadian preferred share index in 2015 – so steep, in fact, that some investors are selling simply because their investment has lost value, which has to be one of the worst trade techniques ever (it imposes a form of negative convexity on your portfolio, among other bad things).

Still, it is unnerving. Look at the graph of the value of an investment in CPD, as published by Blackrock:

prefsYTD_150731
Click for big

This isn’t the smooth ride that some were expecting! The broad TXPR index was down 4.10% on the month and is down 11.47% over the past year. The FixedReset TXPL index has fared even worse, down 5.31% on the month and a horrific 17.26% on the year. I don’t have figures for the BMO-CM 50 at this time, but if I plug in the TXPR results for July, I can draw the following graph, which shows the rolling twelve month and twenty four month total returns from December 31, 1992:

prefIndexReturns_150731
Click for Big

So both the one- and two-year returns for the index now show losses exceeded only by the depths of the Credit Crunch in the 20+ years of data I have available. And, I will note, the four year total return for TXPR is now negative – in fact, you have to go back to January, 2011, to find a starting point that will give you a better than zero return through the period.

So I received an eMail from a client that said, in part:

But my real problem is that in trying to decide whether to stay in your fund or pull out, I do not know what I am betting on. The prospect of rising CDN interest rates (seems unlikely that would help), the overall Cdn economy? Something else?

What is your take on what it would take for preferred values to start moving in the right direction?

What follows is my answer, with minor edits to ensure anonymity and to reflect the medium of the message.

I can appreciate your concern.

Your first investment was valued on 2012-11-19; the second on 2013-1-21.

From the end of November, 2012, to June, 2015, the fund’s total return (reinvesting dividends, before fees) was -0.35%, compared to the BMO-CM “50” index return of -3.64%. TXPR (the broad S&P/TSX Preferred Share index) returned -4.04%, while TXPL (S&P/TSX, FixedResets only) returned -9.65%.

For the period beginning 2013-1-31 I find: Fund, -1.95%; BMO, -4.86%; TXPR, -5.56%; TXPL, -11.42%.

So the problem is not with the fund so much as it is with the market.

The indices are currently comprised of about 1/3 Straight Perpetuals, 2/3 FixedResets. For an idea of what has happened to Straights, see the attached Chart #22 from the July PrefLetter, which shows the interest-equivalent spread between Straight Perpetuals and long-term Corporate bonds (the “Seniority Spread”).

PL_150710_Body_Chart_22
Click for Big

Market Yields changed as follows, from November 28, 2012 to June 30 , 2015
Five Year Canadas: 1.31% … 0.81%
Long Canadas: 2.38% … 2.37%
Long Corporates: 4.2% … 4.0%
Straight Perpetuals: 4.88% … 5.20%
Interest-Equivalent Straight Perpetuals: 6.35% … 6.76%

These changes have had the effect of widening the Seniority Spread from 215bp to 276bp. I can think of two rationales for this widening:
i) the retail investors who dominate the preferred share space are demanding a higher spread to compensate for perceived risks of losses once “interest rates start to rise”; that is, they are reacting more than the institutional investors in the bond market to risks of loss. This could be due to higher risk-aversion (defining “risk” as chance of loss), less binding duration constraints on the portfolio, simple lack of sophistication, or any combination of these three considerations. Note that I have not made a formal study of the subject and there may be other factors, but those are the ones that occur to me through my experience talking to investors.
ii) Straight Perpetual yields are being pushed up (or at least supported) by FixedReset yields (see chart FR-44, below, from the extract from the July PrefLetter ). This would be due to a perception amongst investors that Straight Perpetuals are more “risky” (whatever that means!) than FixedResets and hence deserving of a positive spread; note that this effect is not observed when comparing sovereign inflation-indexed bonds to nominals (the Inflation Risk Premium).

PL_150710_App_FR_Chart_44
Click for Big
These spreads use Yield-To-Worst, not Current Yield
This is Chart FR-44

With respect to FixedResets, it is clear from the horrible performance of TXPL referenced above relative to that of the broader TXPR (which one can approximate as being comprised of about 2/3 TXPL throughout the period of interest, although it has, of course, varied, with FixedReset issuance slightly overcompensating for capital losses) that FixedResets have been whacked.

I have hypothesized a rationale for this underperformance in the attached extract from PrefLetter under the heading “An Experimental Data Series”, to wit: in the face of declines in the Five-Year Canada yield (which is the basis for the resets of of this type of preferred share), investors are attempting to maintain a constant yield irregardless of what is happening with other yields. This is hard to justify on rational grounds, but there has always been an element of irrationality in preferred share pricing! Thus, declines in the GOC-5 yield have been 100% compensated for by declines in price, without referencing yields of comparable long-term instruments; this contradicts one of the features of FixedResets that was used (perhaps inadvertently through indiscriminate use of the term “interest rates”) to help sell the issues when they were developed – that price would remain constant given parallel shifts in the yield curve (with credit spreads assumed, again implicitly, to be constant).

PL_150710_App_FR_Chart_48
Click for Big

This 100% dependence of FixedReset price on GOC-5 has a very large effect, as derived in the last equation on page 3 of the extract:
i) The base Modified Duration of FixedResets is equal to (1 / EFCY). The term EFCY (“Expected Future Current Yield”) is about 3.75%, implying a Modified Duration of about 27 – not only far higher than long bonds, but dependent upon more volatile five-year yields to boot!
ii) The term (25/P) in the equation implies negative convexity

So to summarize, I feel that the poor performance of the market since your initial investment is due to:
i) very high dependence of FixedReset prices on GOC-5 levels, which has contradicted prior assumptions of an equal and opposite co-dependence on long-term yield levels.
ii) maintenance of a spread to PerpetualDiscounts, which has prevented Straight Perpetuals from participating in price increases due to declines in long-term corporate yields.

PL_150710_App_FR_Chart_43
Click for Big
The “Bozo Spread” is the Current Yield of PerpetualDiscounts less the Current Yield of FixedResets
It is not yet clear whether the market pays more attention to these Current Yields, or to the Yields-to-Worst, when relating FixedResets to PerpetualDiscounts

I will also note that to a certain extent, we’ve seen this movie before: during the Credit Crunch Floating Rate issues performed appallingly poorly, since their dividends were linked to contemporary (as opposed to expected!) Canada Prime while their yields were linked to PerpetualDiscounts (see my contemporary article and the next chart)

floaterTotalReturn
Click for Big
Negative Total Return Over Fifteen Years!

So, while I can appreciate your dismay regarding the performance of your investment, I will point out that:
i) the key consideration is not past performance but how the characteristics of the asset class may be expected to fit into your portfolio requirements going forward.

ii) Expected income per unit in the fund has actually increased over the period, from $0.4643 in December 2012 to $0.5217 in June 2015 (see MAPF Performance: June 2015 ). This calculation is dependent upon various assumptions which you may or may not accept, but it represents my best guess!

iii) The increase in spreads over the period implies a significant reduction in expected income should you switch to another Fixed Income type of investment at this time.

iv) Expected future performance of FixedResets is highly geared to GOC-5, insofar as we can accept that the last equation on page 3 of the July PrefLetter extract reflects market reality. While I agree that we might be waiting a while for GOC-5 to increase substantially, I will suggest that current levels must be at or near a bottom. Mind you, I’ve been suggesting that continually for several years now and been wrong every time, so you may wish to disregard that particular exercise in market timing!

v) Expected future performance of Straights should be better than that of corporate long bonds over the medium term; and corporate long bonds should in turn outperform long Canadas; in both cases due to moderation of current high (by historical standards) spreads

I hope all this helps. I realize that I have used a fair bit of jargon in this eMail (and, what’s worse, jargon that I’ve developed myself!) so if there is anything in the above that makes no sense, feel free to ask for clarification. And, of course, if you would like to discuss this further prior to making an investment decision, that’s fine too – whether by eMail or telephone.

Sincerely,

Market Inefficiency: AIM.PR.A vs. AIM.PR.C

Saturday, March 28th, 2015

I was challenged on Financial Wisdom Forum to opine on preferred share market inefficiency:

I do wonder if this market is as inefficient as you suggest. It seems to me when inefficiencies exist (that is to say easy money to be made) in capital markets such inefficiencies don’t last long as smart money rushes in to scoop up the cash and thus eliminating the inefficiency . Perhaps Mr Hymas would care to offer an expert opinion on the preferred share market with respect to its inefficiencies or lack there of.

Fortunately enough, there’s an example of inefficient pricing noted just above:

Take AIM.PR.A (which is one of my holdings and is definitely not investment grade). It was yielding 7.3% (current) when I started buying in February. I knew it was resetting end of March at around 4.5% of the redemption price (for a current yield of 5.1% at my ACB), but thought: ‘It must be safe to buy because the market must have priced in whatever.s going to happen’. Well, in early March they announced the details of the reset and then it sunk like a stone! So a week or so after the announcement, I thought: ‘Now the market must really have priced in everything, so now I can buy some more at a bargain’. Which I did. But it kept sinking and sinking and still is sinking!

Let’s look at the yield of AIM.PR.A using the new and improved yield calculator for FixedResets. Assumptions are always necessary when making yield calculations so assume

  • The bid in thirty years will be the same as the close on Friday, 19.31
  • Constant 5-year Canada yield of Friday’s close of 0.79%

and combine that with what we know about the issue

  • Resets 2020-3-31
  • Reset yield is GOC-5 +375bp
  • Paydates are quarterly from June 30

We calculate the yield as 5.85%.

Now look at AIM.PR.C, which closed on Friday at 24.91, currently pays $1.5625 p.a. and resets at five year intervals commencing 2019-3-31 at GOC-5 + 420bp. Make the same assumption of a constant price. The yield is 5.33%.

This relative valuation makes no sense. AIM.PR.C should yield more than AIM.PR.A, since it has greater call risk.

Some people may tell you that the differential makes sense, because when AIM.PR.C resets in four years it is GUARANTEED!!! to reset at a higher level since GOC-5 will DEFINITELY!!! be higher at that time … to which we may retort that in that case, AIM.PR.A will also reset higher since it will reset again one year later.

In order for AIM.PR.C to achieve the 5.85% yield offered by AIM.PR.A, we must assume a constant GOC-5 yield of 1.49%. If we assume that GOC-5 will reach 1.49% and stay there forever, then AIM.PR.A will then yield 6.45% (n.b. a greater increase since the lower price of AIM.PR.A means it is more levered to the GOC-5 rate). So for the realized yields to be equal, we must assume that GOC-5 will increase to 1.49% by the time AIM.PR.C resets in four years, but return to 0.79% when AIM.PR.A resets a year later (there will be infinite equivalent paths for equality of yield, but they will all look more or less like that) and that this zig-zag will continue forever. This seems like a rather complex path to be betting on.

And the above ignores call risk, i.e., assumes that the Volatility of the Market Reset Spread for AIM is zero and that neither issue will be called with 100% certainty. This is another rather aggressive assumption.

If we turn the question around a little, we can determine that, in order for the yield on AIM.PR.A to be equal to the yield on AIM.PR.C (again, with zero allowance made for call risk), then we may say that the constant price of AIM.PR.A should be $21.20, given a constant GOC-5 yield of 0.79 (for both issues!). Thus, we may conclude to a first approximation that AIM.PR.A is about 9% undervalued relative to AIM.PR.C at current prices.

It is not at all unusual to conclude that cheaper issues are unduly cheap relative to their more expensive siblings. I believe that this is due to some feeling among preferred share investors as a group that:

  • Anything priced at around par will always be priced near par, because, dammit, they’re PREFERRED SHARES
  • Anything priced significantly below par is a speculative piece of shit

Regrettably, this hypothesis would be very difficult to prove. And, as the regrettable timing of MAPF’s move into low-spread FixedResets demonstrates, just because something is probably mostly true most of the time doesn’t mean it’s always true all of the time. But … if the odds are with you on all your decisions and you take care that an unlucky streak won’t wipe you out … you’ll do fine.

What’s The Benchmark Five-Year?

Tuesday, February 24th, 2015

Assiduous Reader gsp of the Financial Wisdom Forum writes in and says:

as I posted on the Preferreds thread on FWF (LINK) I am having a hard time understanding which source to best trust when trying to figure out the GOC 5 year benchmark that resets are based on.

The site you link on prefblog(LINK) says the closing 5 year on Feb 20th was 0.72 while investing.com(LINK) says 0.79. The definitive source(BOC) today posted it as 0.80. I’m confused by the variance in all these quotes, especially for a closing rate.

I like to be as precise as possible when using your YTC resets spreadsheet, what’s the best source for intraday BOC 5 year quotes that I can access for free? I have no real use for real time quotes but prefer not to be out to lunch when the rate moves considerably intraday.

Using today’s quotes, we see that CBID’s site (which is the one I use) shows a “Closing Markets as of: 4:00 PM EST 23-Feb-15” yield of 0.66% for the “Canada 5 year”, while the invest.com GOC-5 yield list yield of 0.741% for February 23 for “Canada 5-Year Bond Yield Historical Data”.

That’s a big difference for a five year! So what’s a five-year bond, anyway? Is it the same one today as it was yesterday? Just what exactly is a “five year bond”?

According to the BoC Benchmark definition:

Selected benchmark bond yields are based on mid-market closing yields of selected Government of Canada bond issues that mature approximately in the indicated terms. The bond issues used are not necessarily the ones with the remaining time to maturity that is the closest to the indicated term and may differ from other sources. The selected 2-, 5-, 10-, or 30-year issues are generally changed when a building benchmark bond is adopted by financial markets as a benchmark, typically after the last auction for that bond. The selected 3-year issue is usually updated at approximately the same time as changes are made to the 2-year, and sometimes with the 5-year. The selected 7-year issue is typically updated at approximately the same time as the 5- or 10-year benchmarks are changed. The current benchmark bond issues and their effective dates, shown in brackets, are as follows.
•2 year – 2017.02.01, 1.50% (2014.11.21);
•3 year – 2017.08.01, 1.25% (2014.10.09);
•5 year – 2020.03.01, 1.50% (2015.02.20);
•7 year – 2022.06.01, 2.75% (2015.01.26);
•10 year – 2025.06.01, 2.25% (2015.01.26);
•Long – 2045.12.01, 3.50% (2014.02.21);
•RRB – 2041.12.01, 2.00% (2010.10.21);

So that’s pretty cool! The “Five Year Benchmark”, as defined by the Bank of Canada, changed last Friday, February 20;

From their page BOC Bond Auction information, we see that their Excel spreadsheet (updated to 2015-1-31) lists two prior auctions (of $3.4-billion a pop) of the 1.5% March 1, 2020, bond, on 2014-11-26 and 2014-10-08. The three prior five year auctions were for the 1.75% September 1, 2019, issue, on 2014-8-6, 2014-5-7 and 2014-4-9, each of which also had $3.4-billion size. And we also see that there was another “five year” auction February 18 for delivery February 23. So, it would seem, that they changed their official benchmark as of the day prior to delivery of the third and final auction of the issue.

We can go back to the CBID page: at the bottom, there are quotes for individual issues and we see:

Canada 1.750 2019-Sep-01 104.85 0.66
Canada 1.500 2020-Mar-01 103.72 0.74

So – while it’s not absolutely definitive, it would appear that investing.com is quoting the yield on the 1.5% of March 2020, while CBID is quoting the 1.75% of September 2019 as the “Five Year”.

Who’s right? Who’s wrong? It’s a meaningless question: virtually everything in the bond market is quoted in terms of convention, which is often highly exasperating when discussing yields.

How does the US Treasury do it? They provide a Constant Maturity Yield:

Treasury Yield Curve Rates. These rates are commonly referred to as “Constant Maturity Treasury” rates, or CMTs. Yields are interpolated by the Treasury from the daily yield curve. This curve, which relates the yield on a security to its time to maturity is based on the closing market bid yields on actively traded Treasury securities in the over-the-counter market. These market yields are calculated from composites of quotations obtained by the Federal Reserve Bank of New York. The yield values are read from the yield curve at fixed maturities, currently 1, 3 and 6 months and 1, 2, 3, 5, 7, 10, 20, and 30 years. This method provides a yield for a 10 year maturity, for example, even if no outstanding security has exactly 10 years remaining to maturity.

Treasury Yield Curve Methodology. The Treasury yield curve is estimated daily using a cubic spline model. Inputs to the model are primarily bid-side yields for on-the-run Treasury securities. See our Treasury Yield Curve Methodology page for details.

… and on the Treasury Yield Curve Methodology Page it states:

The Treasury’s yield curve is derived using a quasi-cubic hermite spline function. Our inputs are the Close of Business (COB) bid yields for the on-the-run securities. Because the on-the-run securities typically trade close to par, those securities are designated as the knot points in the quasi-cubic hermite spline algorithm and the resulting yield curve is considered a par curve. However, Treasury reserves the option to input additional bid yields if there is no on-the-run security available for a given maturity range that we deem necessary for deriving a good fit for the quasi-cubic hermite spline curve. For example, we are using composites of off-the-run bonds in the 20-year range reflecting market yields available in that time tranche. Previously, a rolled-down 10-year note with a remaining maturity nearest to 7 years was also used as an additional input. That input was discontinued on May 26, 2005.

More specifically, the current inputs are the most recently auctioned 4-, 13-, 26-, and 52-week bills, plus the most recently auctioned 2-, 3-, 5-, 7-, and 10-year notes and the most recently auctioned 30-year bond, plus the composite rate in the 20-year maturity range. The quotes for these securities are obtained at or near the 3:30 PM close each trading day. The inputs for the four bills are their bond equivalent yields.

Between August 6, 2004 and June 2, 2008, to reduce volatility in the 1-year Treasury Constant Maturity (CMT) rate, and due to the fact that there were no on-the-run issues between 6-months and 2-years, Treasury used an additional input to insure that the 1-year CMT rate was consistent with on-the-run yields on either side of it’s maturity range. Thus, Treasury interpolated between the secondary bond equivalent yield on the most recently auctioned 26-week bill and the secondary market yield on the most recently auctioned 2-year note and inputted the resulting yield as an additional knot point for the derivation of the daily Treasury Yield Curve. The result of that step was that the 1-year CMT was generally the same as the interpolated rate during that time period. As of June 3, 2008, the interpolated yield was dropped as a yield curve input and the on-the-run 52-week bill was added as an input knot point in the quasi-cubic hermite spline algorithm and resulting yield curve.

Between December 3, 2007 and November 7, 2008, due to Treasury’s discontinuance of 3-year notes, we added a composite rate in the 3-year range based on an average of off-the-run securities in that time tranche. This composite was replaced on November 10, 2008 with the on-the-run 3-year note upon its reintroduction.

Treasury does not provide the computer formulation of our quasi-cubic hermite spline yield curve derivation program. However, we have found that most researchers have been able to reasonably match our results using alternative cubic spline formulas.

Treasury reviews its yield curve derivation methodology on a regular basis and reserves the right to modify, adjust or improve the methodology at its option. If Treasury determines that the methodology needs to be changed or updated, Treasury will revise the above description to reflect such changes.

Yield curve rates are usually available at Treasury’s interest rate web sites by 6:00 PM Eastern Time each trading day, but may be delayed due to system problems or other issues. Every attempt is made to make this data available as soon as possible.

This is a much more sensible way to estimate what a reasonable person might call a “Five Year Yield”, with the reservation that I have always been deeply suspicious of the cubic spline curve fitting methodology. It is too abstract for me and there are mathematical problems at the knot points. But I can’t deny that it fits the data well.

While all of this may be considered illuminating, it still doesn’t really answer Assiduous Reader gsp-from-FWF’s problem: what number should he plug into his calculation in order to estimate a projected future dividend rate for FixedResets? Because the following definitions from the prospectus for RY.PR.J are pretty typical:

“Annual Fixed Dividend Rate” means, for any Subsequent Fixed Rate Period, the rate (expressed as a percentage rounded to the nearest one hundred–thousandth of one percent (with 0.000005% being rounded up)) equal to the Government of Canada Yield on the applicable Fixed Rate Calculation Date plus 2.74%.

“Bloomberg Screen GCAN5YR Page” means the display designated on page “GCAN5YR” on the Bloomberg Financial L.P. service (or such other page as may replace the GCAN5YR page on that service for purposes of displaying Government of Canada bond yields).

“Fixed Rate Calculation Date” means, for any Subsequent Fixed Rate Period, the 30th day prior to the first day of such Subsequent Fixed Rate Period.

“Government of Canada Yield” on any date means the yield to maturity on such date (assuming semi-annual compounding) of a Canadian dollar denominated non-callable Government of Canada bond with a term to maturity of five years as quoted as of 10:00 a.m. (Toronto time) on such date and which appears on the Bloomberg Screen GCAN5YR Page on such date; provided that, if such rate does not appear on the Bloomberg Screen GCAN5YR Page on such date, the Government of Canada Yield will mean the arithmetic average of the yields quoted to the Bank by two registered Canadian investment dealers selected by the Bank as being the annual yield to maturity on such date, compounded semi-annually, which a noncallable Government of Canada bond would carry if issued, in Canadian dollars in Canada, at 100% of its principal amount on such date with a term to maturity of five years.

And we don’t know how the GCAN5YR page is calculated (because it’s Bloomberg), although we can guess that it’s more akin to the US Treasury interpolation-on-a-fitted=curve method than it is to the Canadian pick-a-bond method because of the way the alternative calculation is stated. But that’s not a guarantee! Don’t bother calling your salesman to find out: if there’s one thing I have learnt over the course of my career, it’s that front-line staff don’t have a clue how their software works and wouldn’t understand it if they were told. They’re bankers, the sweet little dears, it’s their job to say “0.74 per cent” in a sincere voice, not to have a clue.

And, what’s more, we can’t even look up (for free) just what the GCAN5YR page might be saying at any particular point in time because fuck you, that’s why.

I don’t have Bloomberg – it’s incredibly expensive, it’s completely useless for serious work and it rots the brain – so I can’t provide any clues as to how the number might be calculated. Perhaps if some kind reader who does have access could provide a screenshot or two taken at around 4pm we can examine the matter more closely.

Effect of Varying GOC-5 Rate On Implied Volatility Rich/Cheap Analysis

Tuesday, February 3rd, 2015

Assiduous Reader Prefhound can always be relied upon for detailed analysis and he has not disappointed in his comment on the February 2 Market Report:

For the Jan 23 FTS series, the lowest reset spread was said to be “cheap”, but its return would only be higher than a higher reset spread if long run GOC-5 rose to an equilibrium around 3%. Current price and reset spreads made sense if the long run equilibrium GOC-5 yield were in the 1-1.5% range (vs 0.85% at the time). Only if the long run equilibrium GOC-5 Yield were 0-0.50% would the original rich/cheap analysis produce substantially different long run returns. This suggested to me that rich/cheap was fairly sensitive to long run GOC-5, so arbitrage returns would depend on changes in (and perception of) that benchmark. As you often note, perception can differ enormously from reality, so fixed reset arbitrage appears to have a substantial element of added GOC-5 risk.

It will be recalled that in my original essay on Implied Volatility for FixedResets I made the point that both the “Pure” price (that is, the price of a non-callable annuity) with any given spread would approach par as GOC-5 increased, while the option value would approach zero; thus, we may conclude that an increase in GOC-5 will cause all issues to move closer to their par value (and contrariwise!) regardless of whether they are at a premium or a discount.

As Prefhound has focussed on the January 23 analysis of the FTS FixedResets, I will show their data for that day to make it easier for Assiduous Readers to replicate and extend the analysis. My findings are at variance with Prefhound‘s conclusions, but I’m sure a bit more methodological detail will sort out a difference in assumptions:

FTS FixedResets: Characteristics
Ticker Current
Dividend
Issue
Reset
Spread
Next
Exchange
Date
Bid
Price
2015-1-23
FTS.PR.G 0.9708 +213 2018-9-1 24.70
FTS.PR.H 1.0625 +145 2015-6-1 18.28
FTS.PR.K 1.00 +205 2019-3-1 25.15
FTS.PR.M 1.025 +248 2019-12-1 25.58

So first we will perform a series of computations using the January 23 bids, but varying GOC-5; we come up with the following table:

  Rich / (Cheap)
GOC5 ImpVol Spread FTS.PR.H FTS.PR.G FTS.PR.K FTS.PR.M
5% 1% 247 -3.31 0.84 1.56 0.55
4% 1% 241 -2.98 0.79 1.55 0.3
3% 3% 234 -2.55 0.68 1.51 -0.02
2% 4% 227 -1.92 0.57 1.46 -0.31
1% 5% 217 -1.04 0.24 1.22 -0.7
0% 11% 196 -0.17 -0.17 0.8 -0.7

… which may be graphed as:

impVol_FTS_150123_varyGOC
Click for Big

Further, we can use the Yield Calculator for Resets, which was given a thorough explanation in early December to determine the 25-year yield expected for each of the GOC-5 levels – note that no prior call is assumed in any of these calculations and that the end-price is set equal to current price. We derive the following table (nb: incorrect figures from the original post have been struck out and replaced with corrected figures 2015-2-4).

GOC5 FTS.PR.H FTS.PR.G FTS.PR.K FTS.PR.M
5% 8.80% 6.41% 6.19% 5.62% 6.26%
4% 7.47% 5.69% 5.50% 4.97% 5.64%
3% 6.14% 4.94% 4.79% 4.29% 4.99%
2% 4.80% 4.16% 4.05% 3.58% 4.31%
1% 3.45% 3.35% 3.28% 2.83% 3.60%
0% 2.09% 2.51% 2.47% 2.06% 2.86%

… and plotted as:

yields_FTS_150123_varyGOC_CORRECTED
Click for Big
Corrected 2015-2-4

What makes this chart particularly fascinating is that the minimal difference between the four calculated yields is found at a value for GOC-5 that is very close to the actual GOC-5 rate of 0.78% at the close of that day:

yields_FTS_150123_varyGOC_detail_CORRECTED
Click for Big
Corrected 2015-2-4

This bears investigating … one might almost wonder if there isn’t some market making going on that has the effect of grouping these yields together …

Update, 2015-02-04: Prefhound wants to see the prices for the Implied Volatility fitting adjusted to reflect the period until the next Exchange Date. OK, here goes!

  FTS.PR.H FTS.PR.G FTS.PR.K FTS.PR.M
  Spread 145 213 205 248
  Exchange
Date
2015-6-1 2018-9-1 2019-3-1 2019-12-1
  Dividends
Until
Exchange
Date
2 15 17 20
  Current
Dividend
1.0625 0.9708 1.00 1.025
Future Dividends
GOC5 5% 1.6125 1.7825 1.7625 1.87
4% 1.3625 1.5325 1.5125 1.62
3% 1.1125 1.2825 1.2625 1.37
2% 0.8625 1.0325 1.0125 1.12
1% 0.6125 0.7825 0.7625 0.87
0% 0.3625 0.5325 0.5125 0.62
Price Adjustment
GOC5 5% -0.35 -1.64 -2.07 -2.03
4% 0.15 2.11 2.18 2.98
3% 0.03 1.17 1.12 1.73
2% -0.10 0.23 0.05 0.48%
1% -0.23 -0.71 -1.01 -0.78
0% -0.35 -1.64 -2.07 -2.03
Effective Price
GOC5 5% 18.56 27.74 28.39 29.81
4% 18.43 26.81 27.33 28.56
3% 18.31 25.87 26.27 27.31
2% 18.18 24.93 25.20 26.06
1% 18.06 23.99 24.14 24.81
0% 17.93 23.06 23.08 23.56

And now we will perform a series of computations using the January 23 bids as adjusted in the above table, using the appropriate GOC-5:

  Rich / (Cheap)
GOC5 ImpVol Spread FTS.PR.H FTS.PR.G FTS.PR.K FTS.PR.M
5% 1% 193 -4.71 1.65 2.58 3.25
4% 1% 194 -4.51 0.50 1.52 2.27
3% 1% 216 -3.25 1.02 1.80 0.75
2% 3% 225 -2.11 0.65 1.38 -0.12
1% 7% 226 -0.73 0.29 0.95 -0.74
0% 26% 184 -0.23 0.23 0.67 -0.79

This allows the following chart to be drawn:

impVol_FTS_150123_varyGOC_adjPx
Click for Big

The price adjustments, of course, are very large, but it doesn’t make any difference to the fitting, which uses only prices. The Expected Future Current Yields are calculated only for display purposes. At any rate, while there are significant differences, the qualitative conclusions are the same – this chart looks pretty much the same as the one with unadjusted prices, although there’s a curious jog in the ‘Adjusted Price’ one.

What Is The Yield Of HSE.PR.A?

Friday, December 5th, 2014

Assiduous Reader B writes in and says:

I am a subscriber to your monthly newsletter but haven’t notice anything recent on this issue

My question is why would investors embrace the new issue at a yield of 4.50% while selling down the existing A issue which is now paying a yield that is a full percentage point higher.

I recognize the higher reset rate but the yield spread still seems excessive.

Thanks for your assistance

So, since he’s a customer I answered; and I said:

I will address your question in a post on prefblog.com tonight, but in the meantime can you tell me why you believe that HSE.PR.A is yielding a full percentage point higher?

… and he responded:

Thanks James for getting back to me – according to my screen on TD, HSE PR A is yielding 5.68 – the new issue is yielding 4.50% – I know there is something to be said for the extra reset pickup but the difference in current yield seems excessive

… and he included a picture:

HSEPRAquote
Click for Big

OK, so his first mistake is getting advice – even advice on such a simple thing as yield – from a bank. You should never seek advice or analysis from a bank, because they’re all domeless wonderboys, with about enough brains to say “We’re big!” and not much else.

In this particular case, TD has told him that the yield on HSE.PR.A, when quoted at 19.56-59, is 5.6789%, which a little experimentation tells us, is the Current Yield Ask, that is to say, the Current Dividend, 1.1125, divided by the Ask Price of 19.59, equals 5.6789178%, where I have tacked another three decimal places on to their reported figure just to sneer at the bank and their precious four decimal places of meaningless precision.

Never Use Current Yield When Analyzing Preferreds

It isn’t even accurate when evaluating Straight Perpetuals (since the relationship between the calculation date and the next payment date is a significant source of error), and is absolutely hopeless when evaluating something that may be called (which is not important in this case) or which is expected to experience a change in dividend (which is very important in this case).

Assiduous Reader B has made the mistake of assuming that the Issue Reset Spread is of minor importance, a mere adjustment to Current Yield, but in this case the projected dividend is so different from the current dividend that he’s wrong.

HSE.PR.A is a FixedReset, 4.45%+173, that commenced trading 2011-3-18 after being announced 2011-3-10. It resets in March, 2016, and if the GOC-5 yield continues to be at its yield of 1.45%, the reset rate will be 3.18%, a 29% drop from current levels.

One chart I am particularly fond of illustrates the relative importance of the Current Dividend vs. the Issue Reset Spread for FixedResets that may be assumed to be perpetual (which is a pretty good bet in this case):

PL_141114_App_FR_Chart_17
Click For Big

Given that HSE.PR.A resets in a little over one year, we see that the headline figure, 4.45%, contributes less than 10% of the valuation of the instrument – all the rest is entirely up to the Issue Reset Spread.

So, given that we know the importance of the Issue Reset Spread, how can we work out the all in yield of the issue in order to allow us to compare HSE.PR.A to the new issue, which is a FixedReset, 4.50%+313?

The answer is to use the Yield Calculator for Resets, which is an Excel Spreadsheet I have made available to the public, linked on the Right-Hand Navigation Panel under the heading “Calculators”. [Update: Note that this calculator has been improved since this post was written; the input of the data has been simplified. … JH 2015-8-7] [and a very minor modification added 2023-9-27 … JH 2023-9-27] It should be noticed that this is not a magic black box, nor is it particularly sophisticated. It’s simply a tool to allow a schedule of cash flows to be input into a spreadsheet easily. So to use the tool, we input our data into the yellow boxes. We’ll get the results of the calculation in the green boxes and the calculation is performed in the turquoise boxes;; we don’t touch them. Only touch the yellow boxes:

  • Current Price: we’ll put in 19.59, because that’s what the bank used.
  • Call Price: You can put in the call price here, but we’re not expecting the issue to be called – we expect it to remain outstanding in 25 years. So what will the price be in 25 years? There are various approaches to this, one of which is discussed in PrefLetter, but it’s reasonable to assume that in 25 years it will be priced the same as it is now, so we’ll put in 19.59. If you don’t like 19.59, put in some other number. It’s not magic. The Yield Calculation Police won’t take you away if you put in some other number. But your calculation is only as good as your assumptions, so if you calculate a very high yield by inputting some silly price – like $50.00 – as the end-price, well, your calculation is only as good as your assumptions.
  • Settlement Date: Strictly speaking, we should put in the date that a trade executed today will settle (2014-12-9), but I usually use the Trade Date, on the grounds that the bank won’t even let you enter the order unless you’ve got money available RIGHT NOW to pay for it. So I’ll input 2014-12-4.
  • Call Date: If it was priced at $26.00 and I was expecting it to be called, I would put in the call date. But I expect it to be around in twenty-five years (the maximum allowable in this spreadsheet) so I’ll put in 2039-12-4. Again, it’s up to you. If detailed examination of the numerological code embedded in The Gospel According To St. Mark has convinced you that it will be priced at 21.13 on 2028-7-8, go ahead and put in that call price and that date. Don’t worry about the Yield Calculation Police, I’ve paid them off.
  • Quarterly Dividend: So what dividend does it pay right now, expressed as a quarterly amount? I hate using a calculator to calculate six decimal places, so I will input a tiny Excel formula “=25 * 0.0445 / 4”, that is, “equal to the par value times the annual coupon rate divided by four”.
  • Cycle: This gets a little tricky, because we need to know the pay-date of each dividend. A little research tells us it’s paid on the last day of each quarter, March / June / September / December, which is cycle 3. So plug in “3”
  • Pay Date: So what day of these months? It’s the last day, so plug in “31”. In the cash flow schedule, the calculated date “June 31” will be transformed to “July 1”, as you can see in the turquoise area to the right of the data input area. This is a bit of an error, but a very tiny one.
  • Include First Dividend: This is quite important. As the spreadsheet tells you, the next dividend payment is December 31, based on the information you’ve input above. If you buy it today, will you earn that dividend? You’ll have to look up the ex-dividend date for the issue; in this case the ex-dividend date was 2014-11-25, which is now in the past, so you WON’T get the next dividend, so input “0”
  • First Dividend Value: For most issues, the first dividend payment is for a different amount from the others, since it’s adjusted to reflect the time from the security’s issue to the pay date, rather than pay-date to pay-date. HSE.PR.A has been around for a long time, so this does not apply and we’re not even earning the next dividend anyway, so it doubly doesn’t matter. Leave this field blank.
  • Reset Date: The issue resets 2016-3-31. Plug in this date
  • Quarterly Dividend After Reset: This is the moment we’ve all been waiting for! We have to estimate what the dividend will be after the reset, while bearing in mind that the yield we calculate will only be as good as our estimates. It’s generally best to assume that major market yields will not change; that on the reset calculation date the 5-year GOC yield will be the same as it is today, 1.45%. But if you feel this is unreasonable, put in another number you’re more comfortable with. If you think that 1.45% is ridiculous and that GOC-5 will be 2.00% on recalculation day, use 2.00%. You have to use some kind of assumption, there’s no way around that. We will note that TD’s calculation, in using Current Yield, assumed the dividend would not change; i.e., that the dividend would reset to be equal to the 4.45% it is currently, i.e., that GOC-5 on reset calculation date will be 2.72%. Well, if that’s the number you want to use, go ahead. It’s a free country and you can assume anything you like. Just remember that the quality of your yield estimate will reflect the quality of your assumptions; and also remember that consistency is a virtue, so if two issues are resetting at the same time, you should use the same estimate for GOC-5. But I will assume a future GOC-5 rate of 1.45%, so I’ll input the Excel formula “=25 * (0.0145 + 0.0173) / 4” = par value * (sum of assumed GOC-5 rate and Issue Reset Spread [expressed annually]) divided by 4 [quarters per year]. We should also note that the spreadsheet makes no provision for changes in GOC-5, so if you feel that GOC-5 will be 2.00% on the 2016 reset calculation date, but 3.00% on the 2021 reset calculation date, you’ll have to develop your own elaboration of this spreadsheet.

And that’s the end of our input and our answer pops up in the green boxes! Current Yield, 5.68%, just as advertised by TD, but an Annualized Quarterly Yield To Call of 4.17%. That’s quite a difference! And, I will note, it is substantially less than the New Issue FixedReset, 4.50%+313. Implied Volatility Theory tells us to expect less for a deeply discounted issue compared to one at or near par value, but just how much less it should be is a whole ‘nuther issue.

And, I suggest, you should always think of this number as “4.17%, assuming an end-price of 19.59 and a constant GOC-5 of 1.45%”, just to remind yourself of the two critical assumptions you made.

So there you have it. I suggest that those interested in using this spreadsheet as an adjunct to their trading first check all the calculations – Trust But Verify! – and second, play around with it a bit. Change the assumptions of end-price and GOC-5 estimate on reset, see how sensitive the answers are to the inputs. The better you understand your data, the better an investor you’ll be.

Market-Based Bank Capital Regulation

Wednesday, March 5th, 2014

Assiduous Reader DR sent me the following query:

Today’s Financial Posts has an article “A better Basel mousetrap to protect taxpayers”, by Finn Poschmann regarding NVCC.

What is your opinion?

A short search brought up the article in question, A Better Basel Mousetrap to Protect Taxpayers, which in turn led me to the proposal by Jeremy Bulow and Paul Klemperer titled Market-Based Bank Capital Regulation:

Today’s regulatory rules, especially the easily-manipulated measures of regulatory capital, have led to costly bank failures. We design a robust regulatory system such that (i) bank losses are credibly borne by the private sector (ii) systemically important institutions cannot collapse suddenly; (iii) bank investment is counter-cyclical; and (iv) regulatory actions depend upon market signals (because the simplicity and clarity of such rules prevents gaming by firms, and forbearance by regulators, as well as because of the efficiency role of prices). One key innovation is “ERNs” (equity recourse notes — superficially similar to, but importantly distinct from, “cocos”) which gradually “bail in” equity when needed. Importantly, although our system uses market information, it does not rely on markets being “right.”

Our solution is based on two rules. First, any systemically important financial institution (SIFI) that cannot be quickly wound down must limit the recourse of non-guaranteed creditors to assets posted as collateral plus equity plus unsecured debt that can itself be converted into equity–so these creditors have some recourse but cannot force the institution into re-organization. Second, any debt guaranteed by the government, such as deposit accounts, must be backed by government-guaranteed securities. This second rule can only realistically be thought of as a very long-run ambition – our interim objective would involve a tight ring-fence of government-guaranteed deposits collateralized by assets that are haircut at rates similar to those applied by lenders (including central banks3 and the commercial banks themselves!) to secured borrowers.

Specifically: first, we would have banks replace all (non-deposit) existing unsecured debt with “equity recourse notes” (ERNs). ERNs are superficially similar to contingent convertible debt (“cocos”) but have important differences. ERNs would be long-term bonds, subject to certain term-structure requirements, with the feature that any interest or principal payments payable on a date when the stock price is lower than a pre-specified price would be paid in stock at that pre-specified price. The pre-specified price would be required to be no less than (say) 25 percent of the share price on the date the bond was issued. For example, if the stock were selling at $100 on the day a bond was issued and then fell below $25 by the time a payment of $1000 was due, the firm would be required to pay the creditor (1000/25) = 40 shares of stock in lieu of the payment. If the stock rebounded in price, future payments could again be in cash.

Crucially, for ERNs, unlike cocos:

  • any payments in shares are at a pre-set share price, not at the current share price or at a discount to it—so ERNs are stabilizing because that price will always be at a premium to the market
  • conversion is triggered by market prices, not regulatory values—removing incentives to manipulate regulatory measures, and making it harder for regulators to relax requirements
  • conversion is payment-at-a-time, not the entire bond at once (because ERNs become equity in the states that matter to taxpayers, they are, for regulatory purposes, like equity from their date of issuance so there is no reason for faster conversion)–further reducing pressures for “regulatory forbearance” and also largely solving a “multiple equilibria” problem raised in the academic literature
  • we would replace all existing unsecured debt with ERNs, not merely a fraction of it—ensuring, as we show below, that ERNs become cheaper to issue when the stock price falls, creating counter-cyclical investment incentives when they are most needed.

OK, so I have difficulties with all this. Their first point is that non-guaranteed creditors “cannot force the institution into re-organization.” Obviously there are many differences of opinion in this, but I take the view that being able to force a company into re-organization – which may include bankruptcy – is one of the hallmarks of a bond. For example, I consider preferred shares to be fixed income – as they have a cap on their total return and they have first-loss protection – but I do not consider them bonds – as they cannot force bankruptcy. The elimination of bankruptcy, although very popular among politicians (who refer to bankruptcy as a form of terrorism) is a very big step; bankruptcy is a very big stick that serves to concentrate the minds of management and directors.

Secondly, they want insured deposits to be offset by government securities. There’s an immediate problem about this in Canada, because insured deposits total $646-billion while government of Canada marketable debt totals $639-billion. You could get around this by saying the CMHC-guaranteed mortgages are OK, but even after years of Spend-Every-Penny pouring fuel on the housing fire, CMHC insurance totals only $559.8-billion (out of a total of $915-billion. At present, Canadian Chartered Banks hold only about $160-billion of government debt. So it would appear that, at the very least, this part of the plan would essentially force the government to continue to insure a ridiculous proportion of Canadian residential mortgages.

And, specifically, they want all (non-deposit) existing unsecured debt with “equity recourse notes”. OK, so how much is that? Looking at recent figures from RBC:

RBCBalanceSheet
Click for Big

So roughly a quarter of Royal Bank’s liabilities would become ERNs …. and who’s going to buy it? It’s forcibly convertible into equity long before the point of non-viability – that’s the whole point – so for risk management purposes it is equity. If held by another bank, it will attract a whopping capital charge (or if it doesn’t, it should) and it can’t be held by institutional bond portfolios (or if it is, it shouldn’t be). I have real problems with this.

The paper makes several entertaining points about bank regulation:

The regulatory system distorts incentives in several ways. One of the motivations for Citigroup to sell out of Smith, Barney at what was generally believed to be a low price, was that it allowed Citi to book an increase in regulatory capital. Conversely, selling risky “toxic assets” with a regulatory value greater than market is discouraged because doing so raises capital requirements even while reducing risk.[footnote].

[Footnote reads] : Liquidity reduction is another consequence of the current regulatory system, as firms will avoid price-discovery by avoiding buying as well as selling over-marked assets. For example, Goldman Sachs stood ready to sell assets at marks that AIG protested were too low, but AIG did not take up these offers. See Goldman Sachs (2009). For an example of traders not buying even though they claimed the price was too low, see the FCIC transcript of a July 30, 2007 telephone call between AIG executives. “We can’t mark any of our positions, and obviously that’s what saves us having this enormous mark to market. If we start buying the physical bonds back … then any accountant is going to turn around and say, well, John, you know you traded at 90, you must be able to mark your bonds then.” Duarte (2012) discusses the recent trend of European banks to meet their requirements to raise regulatory capital by repurchasing their own junk bonds, arguably increasing the exposure of government insurers.

However, don’t get me wrong on this: the basic idea – of conversion to a pre-set value of stock once the market breaches that pre-set value – is one that I’ve been advocating for a long time. They are similar in spirit to McDonald CoCos, which were first discussed on PrefBlog under the heading Contingent Capital with a Dual Price Trigger (regrettably, the authors did not discuss McDonald’s proposal in their paper). ERNs are ‘high-trigger’ instruments, and therefore will help serve to avert a crisis, rather than merely mitigate one, as is the case with OSFI’s NVCC rules; I have long advocated high triggers.

My basic problem is simply that the authors:

  • Require too many ERNs as a proportion of capital, and
  • Seek to Ban the Bond

However, it may easily be argued that these objections are mere matters of detail.

Calculation of RatchetRate Dividend Yield

Wednesday, February 12th, 2014

Assiduous Reader DT writes in and says:

I have been following your blog for quite some time but I have a question that I can not find a clear answer to….
Can you explain how an issuer calculates the ‘Ratchet Rate’ of their preferred shares on a given reset date?

The prospectus for BCE.PR.S / BCE.PR.T provides an archetypal example:

The annual floating dividend rate for the first month will be equal to 80% of Prime. The dividend rate will float in relation to changes in Prime and will be adjusted upwards or downwards on a monthly basis by an adjustment factor whenever the Calculated Trading Price of the Series S Preferred Shares is $24.875 or less or $25.125 or more respectively.

The maximum monthly adjustment for changes in the Calculated Trading Price will be ±4.00% of Prime. The annual floating dividend rate applicable for a month will in no event be less than 50% of Prime or greater than Prime.

The Adjustment Factor for a month will be based on the Calculated Trading Price of the Series S Preferred Shares for the preceding month determined in accordance with the following table:

If the Calculated Trading Price for the Preceding Month is The Adjustment Factor as a
Percentage of Prime shall be
$25.50 or more -4.00%
$25.375 and less than $25.50 -3.00%
$25.25 and less than $25.375 -2.00%
$25.125 and less than $25.25 -1.00%
Greater than $24.875 and less than $25.125 nil
Greater than $24.75 to $24.875 1.00%
Greater than $24.625 to $24.75 2.00%
Greater than $24.50 to $24.625 3.00%
$24.50 or less 4.00%

The maximum Adjustment Factor for any month will be ±4.00% of Prime.

This mechanism is very briefly summarized in my article Preferred Pairs.

All RatchetRate issues will be paired with a FixedFloater, but both elements will not necessarily be trading at the same time.

The Pairs Equivalency Calculator takes advantage of the known time before conversion opportunity and the fact that all these are now paying 100% of prime (and are more likely than not to continue at this rate until this time) to calculate an implied average prime rate that makes the two series equivalent. This relative value analysis can be useful; if you are enamoured of this type of share, it may turn out that your best bet is to buy the FixedFloater with the intent of converting.

The pairs currently are:

FixedFloater RatchetRate
BAM.PR.G BAM.PR.E
BBD.PR.D BBD.PR.B
BCE.PR.T BCE.PR.S
BCE.PR.Z BCE.PR.Y
BCE.PR.A BCE.PR.B
BCE.PR.C BCE.PR.D
BCE.PR.F BCE.PR.E
BCE.PR.G BCE.PR.H
BCE.PR.R Not trading
BCE.PR.I Not trading

It is the adjustment to the RatchetRate that makes these unsuitable for banks – in order to qualify at Tier 1 Capital, preferred shares must not have any provisions that provide compensation for loss of credit quality.

For those seeking to compare RatchetRates with FloatingResets, note that Prime is usually 3-Month Bills + 200bp. For this reason, we can reasonably expect that the RatchetRates currently extant will (a) trade below $25 forever and (b) remain outstanding forever and (c) that we could be wrong about (a) and (b), so don’t mortgage the house.