There was a query in yesterday’s comments about the difference in yield between the OpRet and SplitShare Indices.
Well – some of the difference is due to the nature of the calculation, which includes the negative yields-to-worst for the Operating Retractible issues CM.PR.A (-5.44%) and PWF.PR.D (-10.30%). These two issues have the combined effect of bringing the average yield of the entire index down by 180bp … but that’s just the way the mean works.
Assiduous readers will note that in the final version of the indices, I am using median yield rather than mean, which helps a lot but has problems of its own (choppiness in an index comprised of two issues of roughly equal weights and greatly different yields, for instance).
realBoomer continues:
It seems to me that Op. Retract and Split-Shares are essentially similar investments – both pay a fixed dividend and will be redeemed at par at some date in the future. In fact, I would think a Split Share issue should be preferable to an Op. Retract issue of the same credit quality, since the Split Shares dividends and unit values are protected to some extent by the Capital Shares. Am I missing something, or is the market just irrational for Split Shares?
Well … for an introduction to the similarities and differences between the two issues, I can suggest Retractible Preferreds and Bonds and SplitShares.
The investments are in principle the same, having cash-flows that are analyzed the same way.
As far as credit quality is concerned … my article Are Floating Rate Prefs Money Market Vehicles contains a highly unscientific and subjective ratings migration table, showing the downgrades from Investment Grade to … er … not investment grade by DBRS. Be careful with the table and read the errata given in the post about the article! SplitShares are worse than the table makes them.
SplitShare ratings have historically been more volatile than operating company ratings, basically for the same reason as CDOs have more volatile ratings than regular bonds. They are dependent upon a mark to market of their underlying assets … exchange traded underlying assets. It’s not like you have a regular company, for instance, that through good years and bad owns a factory worth $100-million. DBRS is making an attempt to tighten up their standards, as I noted in a post about SBN.PR.A, which will help a lot, but basically it’s a question of visibility … it’s much easier to see that asset coverage has declined to 1.1:1 than it is to see that Weston / Loblaws is having real difficulties as opposed to a mere bad year.
Aside from rating volatility, there’s the question of hidden resources. Let’s say Quebecor gets into even more trouble than it’s in now. There’s a chance – just a chance, mind you, but it’s there – that a strategic buyer will step up to the plate and buy it for a song, sticking it to Quebecor’s common shareholders, but having to bail out the preferred shareholders because they don’t have to agree to nothing. I’m thinking of a situation, for instance, where the rational price of a Quebecor common share goes to ten cents, or something like.
If a split share corporation gets that close to the line, there ain’t no deus ex machina coming.
Even with all this, Assiduous Readers will note that Malachite Fund often owns SplitShares and much less often Operating Retractibles. Even after correcting for the funny averages, there is a very real yield difference between the two classes.
It is my unsupported, deniable, and thoroughly irrelevant belief that this yield difference is due to the nature of the market. Corporations often own preferred shares. As we learned in the ABCP fiasco, many companies – even those with nine-figure investment portfolios – do not seek professional investment advice. It’s money, right? Who does money around here? The CFO and Treasurer, right? Get cracking!
So, you have guys buying prefs without much knowledge of the market. If they squeeze out half a point more return, nobody’s going to thank them. If they have to ‘fess up that something tanked, they might lose the keys to the executive washroom. They might get fired and find that the company’s statement of defense against unjust dismissal is full of quotation marks and “Structure Investment Vehicles” and “Complex Investment Strategies” and “Covered Call Writing” using “Derivatives”.
It should also be noted that most, if not all, split share issues are rated only by one credit rating agency – DBRS. That might run afoul of generalized investment guidelines. I’m not aware of any reason why S&P wouldn’t rate SplitShares – I suspect it’s just a question of rating fees and issuers not wanting to pay them. Operating companies will generally have two ratings on their prefs (there’s something of a fad for getting three, lately); makes sense, given that they pretty well have to have at least two ratings to sell their bonds: the marginal cost is, well, marginal.
One way or another, it’s a lot safer, career-wise to buy an issue with a recognizable name on it. Nobody ever got fired for buying IBM. So – I suspect – Operating Retractible issues will trade with lower yields simply because there are more potential buyers.
I agree with James’ comments but would add one important proviso.
If split shares don’t get common dividends from the underlying, the prefs won’t pay dividends, irrespective of the asset coverage. AND, split share prefs include a short put option that has increased in price recently, reducing the split share pref price more than OpRets.
Past Example: Bombardier cut its common dividend several years ago, but continued to pay dividends on its corporate prefs. Had there been a split share based on Bombardier, which depended on the Bombardier common dividends, those dividends could not have been paid.
Even companies with very shaky finances, such as Nortel and now Quebecor IQW) are paying their pref share dividends.
Current Example: BAM and BPO prefs are direct obligations of Brookfield (at two different corporate levels, but that distinction may not matter too much). BNA prefs A, B and C are obligations of the Split Share company holding Brookfield common shares. BNA prefs are dependent upon Brookfield to be paying sufficient common dividends, but BAM and BPO prefs are not. Furthermore, BAM and BPO prefs always represent a $25 claim on the hard assets of Brookfield, whereas BNA prefs only represent a claim on the first $25 of the (paper; common share) assets of the split share company.
Now, BPO is in the real estate business and uses a lot of leverage. If Brookfield common shares lose 75% of their value, the BAM and BPO prefs still have a claim of $25 on the company (and the OpRets among them could well be converted into common to realize a $26.20 maturity value), but BNA split share prefs will be underwater (Asset Coverage
Oops, it looks like I have a character limit and the last portion of the previous comment was deleted (JAMES — can you increase?). It went something like:
(Asset Coverage
Now I am getting pissed off. I think the “less than” sign breaks the reply! JAMES — Please fix! To continue without using the dreaded sign:
(Asset Coverage less than 1 vs 3.8 recently), and will have a very uncertain maturity value. It would not surprise me if a 75% fall in Brookfield common triggered a reduced dividend, which could flow through to the BNA prefs.
Some observers think that commercial real estate is the next implosion due in the US financial crisis. Real asset prices and REITs have been bid up quite substantially, and Brookfield common has a lowish dividend yield. It would not be impossible for the common to fall 75%.
To the extent professional investors recognize this, perhaps they have been avoiding buying splitshare prefs, leaving a few retail investors to sell when there are no bidders. Witness the BNA.PR.C pref absence of bids yesterday, even at 8-10% yields to normal maturity.
Finally, the split share pref is short a put option on the common (the capital shares are long a put and call). These days market and implied option volatility is 2-3X higher than six months ago, so splitshare pref prices should have fallen more than OpRets.
In times of market trouble, the dividend continuity and maturity provisions of OpRets are much more valuable than those for splitshares. This is why the OpRet yields are much lower today. No way these are similar investments!
The assertions regarding income coverage are not correct. The BNA.PR.C prospectus, for instance, is available on SEDAR:
In my article, Split Shares, you will find many examples of split share corporations with an income coverage ratio of less than unity.
I do agree, however, that while the BAM and BNA prefs are both (for all intents and purposes) backed by the same underlying security (BAM.A), the BAM prefs have a direct claim on the company. So yes, it is possible that, for instance, BAM could go bankrupt with BAM.A getting wiped out, BNA getting wiped out along with it, while the BAM prefholders got paid in full. Remote, but possible.
It is for this reason that DBRS will not (generally) rate a split share based on a single name higher than prefs issued directly by that name, regardless of other metrics. From their publication “Split Share Issuers: A Performance Overview” (difficult to find on their site. Go to Advanced Search and search for the phrase “Split Share”. This paper was released Jan 31, 2007 and is classified under “Methodology”):
It was in accordance with this revised policy that the BNA issues were downgraded from Pfd-2 to Pfd-2(low) on May 4, 2007.
Sorry about the mechanical difficulties. At some point I will be upgrading the software used to run this blog … but I have not yet become sufficiently annoyed with the glitches to do all that.
Anyway, the idea that creditors are short a put on the common is referred to as the Structural Model of Default Risk. Here’s a link to a randomly chosen paper on the topic.
If Company A has both common and prefs, the common being priced at $100, then the prefs are, according to the structural model, short a put with a zero strike price. If there is a split share company that buys a share of Company A at $100, financing with capital stock of $75 and pref stock of $25, the pref holders are short a put on Company A stock with strike price of $25.00.
It certainly may be that professional investors other than myself recognize the difference of strike prices and factor them into their valuation models … I doubt it, frankly, but I’m willing to be convinced!
Another problem with split shares that restricts the pool of potential buyers is liquidity. I have a buddy who has direction over a pool of about $50-million in prefs. The underlying accounts are segregated. His minimum issue size is about $5-million. If he wants to swap issues, he’s got to get a dealer to cross 200,000 shares (given a $25 price). That’s rarely going to be possible for a split issue. His other choice is to get the dealer to work the order for a few days, get … oh, say, 68,700 shares and then figure out how to allocate them to his underlying segregated accounts. The mechanical difficulties are daunting.
1. Liquidity of BNA.PR.C does not apply. There are $200M outstanding, and liquidity is comparable to BAM.PR.M or N but BNA.PR.C has a YTM of 8% vs 6.8% for the BAM prefs.
2. Option prices are based on a log scale — not the linear scale that makes 0% and 25% strike prices look similar. I can estimate from exchange traded BAM options, that an 11 year put at 25% strike price is worth $3.63 per BNA.PR.C pref, while an 11 year put at $1.00 on the common (not zero, which is inaccessible on a log scale, but getting close) is worth 6 cents. Math can be supplied on request (Black Scholes; put implied volatility = 0.50). There is a huge difference. Just try eliminating this risk with exchange traded options!
1. I am not sure how liquidity the market is pricing in – BNA.PR.C is trading well now (in volume terms!) but this will not necessarily last into the future. PIC.PR.A, for instance, has a current market cap of $279-million and HIMIPref™ calculates the “Average Trading Value” to be $145,000.
I suspect that at this time next year, BNA.PR.C will be about half as liquid (measured by Average Trading Value) as the BAM perps. I’ll bet the price of lunch on it, if you like!
2. (A) We are fortunate to have BAM.PR.J around, with a current pre-tax bid-YTW of 4.82% based on a bid of 26.45 and a softMaturity 2018-3-30 at 25.00, compared with BNA.PR.C’s pre-tax bid-YTW of 8.27% based on a bid of 18.00 and a softMaturity 2019-1-10 at 25.00. What should the spread be, based on your calculation?
(B) I confess to being a little dubious about the application of volatilities extracted from exchange-traded options to long-term embedded options. I see BMO 4.66% Deposit Notes maturing 2009-3-31 quoted at 89-84bp over Canada 4.25%/08; the BMO 4.55% DNs maturing 2017-8-1 are quoted at 139-134 bp over Canada 4%/17. You can use 3.825% and 4.003% for the respective Canada yields. What happens when you plug these numbers into your formulae?
[…] Update 2007-11-22: See Split Share and OpRet Yields for some updated yield numbers and a fascinating discussion in the comments. […]
Oh, yeah.
2. (C) How well does your method fit S&P Default Data? Note that Table 12 of the study shows that default rates for A/BBB issuers have an accellerating pace up to about five years, but are then more linear with time.
I love a good discussion, which is what the 2-way interaction of blogs encourages.
1. Liquidity — I’m sure you are right and let’s not wait a year for a lunch on me.
Anyway, I would guess that a liquidity premium is seldom over 10-20 bp (but you have the data).
2A. Lots of questions there. The easiest to answer is that, using my $3.63 put option value, I would say that comparing BAM.PR.J ($1.35 divs) to BNA.PR.C (1.09 Div) would go roughly as follows:
Value of BAM.PR.J = 26.45 with YTM of 4.82%
Comparable Value of BNA.PR.C without the short put = $18 + 3.63 = $21.63 for a YTM of 6.0%. This is more in line with market yields, with some of the difference with 4.82% representing interest rate protection on the BAM.PR.J.
Of course, I am also dubious about estimates of implied volatility, but I think 0.50 assumption is quite conservative for something so much out of the money. Higher assumed implied volatility would raise option value. Anyway, the point of the calculation of the option values is simply to indicate that the magnitude of the expected value is significant to split share prefs, and perhaps meaningful.
2B. Not sure what this is about. My assumed risk-free rate in the option calculations was 4.5% for 2-11 years maturity. They were US options. My brief experience with bond spreads as a function of maturity is that the spread is narrowest for short term and grows with increasing term (more opportunity for downgrade – risk).
2C. The Black Scholes option pricing formula is not dependent on S&P Data. A random walk would result in a SQRT(Time) relationship which can look like two linear portions. I have used S&P default data to estimate “Fair Spreads” for bonds that compensate for default and recovery. Interestingly, in the market place, the observed spread seldom gets less than 2X the “Fair Spread” calculated in this way and is often many-fold higher (at least for investment grade issues). I use the 2X spread as a “sell signal” and attribute it and higher observed factors to compensation for greater risk.
Anyway, even if the apparent embedded short put option value in split share prefs is a correct semi-quantitative model, BNA.PR.C may still be an interesting buy, in part because of the pref purchaser’s desire to sell an “Overvalued” put option based on temporarily high implied volatility. Furthermore, an arbitrage play short BNA.PR.B and long BNA.PR.C tends to remove a lot of exposure (though not all) to BAM credit or stock price.
2A. Shall I take this as agreement that the BNA.PR.C split share is cheap compared to the comparable direct pref?
2B. What I’m interested in is the evaluation by your methodology of these two discount notes of different tenors. By examining spreads to Canadas we can (at least partially) cancel the term premium; since they are parri-passu any credit difference is due to time. So, I am assuming, your methodology (using data for BMO, of course) will allow for an estimate of what the spreads should be; and I am most curious as to what they are.
2C. Sure, Black Scholes is not dependent upon S&P data; but I’m pretty sure that the extrapolation of short term volatilities that your methodology uses will greatly overestimate the chance of default in the long term. Does your model explain the data?
2D. I realized late, late last night what I don’t like about your methodology. Most structural models do not involve looking at the stock and estimating the chance it will go to zero … after all, we’re interested in credit risk and default probability, which imply a negative value for the common (and whatever else is junior).
Instead, structural models say that the common shareholders own all the assets of the company and are long a put on them with a strike price equal to total liabilities less their equity. So if a company has total assets of $100, shareholders equity of $30, debt of $30 and other liabilities of $40, then you analyze the credit risk to the debt holders by estimating the value of a put on the assets for $70.
2Di : OK, let’s set this up so that we can compare split share prefs with direct prefs.
Company A has Assets worth $100, trade payables of $40, debt of $30, prefs of $15, shareholders equity of $15.
Company S owns shares of A worth $100, financed by capital shares of $80 and prefs of $20.
I claim that a more accurate method of using the structural model is to say that the direct pref holders are short a put on the assets at an exercise price $85.
The split share pref holders are short a put on the assets at the point where they will have direct exposure to the equity. This will happen when Company S’s assets are worth $20, an 80% decline in the value of company A’s equity. Company A’s equity will have fallen by 80% when Company A has shareholders’ equity of $3 (an 80% decline from $15) and therefore the Company S pref holders are short a put on the assets of Company A at $88.
I will further claim that the ratio of prices attributable to these puts will be nowhere near the 3.63:0.06 that you estimated above.
OK, we’ve set a record for the number of comments on one item.
I can’t accept your logic that the put strike on a conventional pref would be common share book value today. Book value is easily frittered away and the losses go to the common shareholders. The put sold by the pref buyer is such that they lose principal when the common equity goes negative — which matches up with a very low but positive share price (usually less than $1.00).
As to 2A, I’m just saying there is a model which says BNA.PR.C could be appropriately priced because of the short put option. BNA.PR.B would (probably) be overpriced on the same basis, though the put would be somewhat lower in value (9 vs 11 years). It is just a model — one among many — but the point is we can’t say “there is no justification for BNA.PR.C to be so low”.
2B. I’m not sure about the relevance of the bond question. BMO is going to have much lower implied volatility than BAM, and if it is a bond, then the put strike is very low (pennies to $1.00) so the put has nearly zero value in a random walk stock option model. Of course the spread increases because of higher probability of downgrade or default. Both Bonds and normal Prefs have a negligible value short equity put for investment grade firms, but split share prefs have a much higher (in log scale terms) strike for their short put.
Not sure if we are getting closer here, but I have to go check my trade,
Cheers
I certainly agree that the common equity can go negative; but this is precisely the problem with your model that I am trying to correct. You are, essentially, assigning a value of zero to a put with a strike price of zero, which implies that investments senior to the common bear no default risk – I will remind you that your model shows a six-cent discount due to default risk for the BAM direct preferreds; I claim that this is far too small even for Royal Bank!
The more standard model that I have proposed does allow for senior note-holders to be short an in-the-money put, which is not possible with your model.
Readers should note that there is now an entire thread devoted to the particular question of : What’s Up with BNA.PR.C? Yield!
2A. Yes, you can come up with models whereby BNA.PR.C is fairly priced (although I will note that this is not what your model has done). I claim, however, that any such model will either (i) show that BNA and BAM comparables are enormously expensive, or (ii) be self-evidently absurd.
2B. With the bond question, I am simply trying to show that your methodology of using short-term volatilities to price long-term options doesn’t work. I don’t think you will be able to account for the yields of the BMO discount notes using your model and the volatility of the BMO common.
This is the same thing I am trying to show with 2C, except that with 2C I am trying to get you to explain historical default rates rather than current market prices.