Gilt fantasyland

[peterbirks]

I don't normally focus on the gilts market. I leave it to the likes of the latter-day Lord Oakeses of Waterloo to spot that an issue should be priced at 97 and 5/8 rather than 97 and 9/16, before lumping in 200 million quid of an investment bank's money for a quick spin and a 200 grand profit overnight. (Incidentally, did I tell you the Oakes tale of when he went for an interview at an investment bank for the job of head of gilt trading. He was asked what the maximum position was that he would be willing to leave open overnight. He replied "About a hundred million", and the bank decided that they did not want someone willing to take such big risks. The name of the bank? Barings.)

But all of a sudden the gilts futures market has got a bit interesting. It revolved around the hitherto uncommented-upon fact that CBOT's 10-year bond futures are derivatives based, in the main, on a non-existent reality. Now, provided the contracts are rolled over, this doesn't cause a problem. But when some large holders demand delivery, well, Chicago, we have a problem.

In a sense we almost had a Nelson Bunker Hunt silver situation. People who had sold the contracts to Pimco (for it was they who were demanding delivery) were scrambling around everywhere to actually get hold of the bonds on which they had happily sold futures. One can imagine the panicked board meetings.

"What do you mean, they've demanded delivery? Are they allowed to do that?"
"Well, yes sir, that's what a futures option is -- a right to buy at a future date."
"So how many of these options have we sold?"
"About 2 million sir."
"And how many of the actual items exist?"
"About 1 million, sir".
"So, we've promised to deliver 2 million of something and the government has only issued 1 million of them?"
"Well, actually they haven't actually issued any of them, not for that precise date, but they have issued a million of somethiing that looks a bit like them, if you view them in the right lighting. But it's worse than that."
"How can it be worse than that? How can it possibly be worse than that?"
"Citadel has about 800,000 of those 1 million, and it isn't selling".
"Shit, you're right, it's worse than that."

CBOT's 10-year T-Bond futures contract is in a sense a laugh and a half, because the actual note doesn't exist at all. Or, if it does, that is purely fortuitous. CBOT has a formula that converts existing T-Bonds into theoretical ones. There are currently about eight times as many open contracts on 10-year T-bond futures as there are "real" futures (say, the September 2012 expiry issue) that could be used to cover them, if everyone suddenly decided to go physical.

This is representative of a big potential problem. Denis Mahoney at Aon asked in speeches for many years "where has all the risk gone?" Company after company will show you complex derivatives transactions which, in theory, magic away interest-rate risk, catastrophe risk, political risk, reputational risk, operational risk and, for all I know, Risk The Board Game. But risk is a constant. It's still there, somewhere. The use of the derivatives market (and collateralized debt obligations are my big worry here, rather than 10-year T-bond futures) makes it look as if the risk has disappeared. But all that is smoke and mirrors. As LTCM found to their cost, when liquidity vanishes, all that risk turns out not to have vanished at all.

Comments

Just Curious...

[jellymillion.livejournal.com]

...what are you concerned about with CDOs? I ask because I'm somewhat involved with CDO risk systems just now. For sure, they're complicated little buggers, but I don't see any liquidity issues as in the Treasury Futures story. I don't really see CDOs and their simple brethren default swaps as securities in the normal sense anyway - they're more like insurance policies.

Re: Just Curious...

[peterbirks.livejournal.com]

The key phrase is the one you use at the end -- "they are just like insurance policies". However, I take your point that I was wrong to really call them derivatives. "Synthetics" is probably a better term, although in the technical sense they are a product "derived" from something else.

Anyway, here's my take on them and I'm quite happy to be corrected if my interpretation is incorrect. You will be pleased to know, Mike, that my worry is systemic rather than micro in nature. And if I had to phrase my worries briefly, it would be "moral hazard".

Suppose you are a bank, or any lender. In the old days, you took deposits, and you lent money. You had an interest in both ends of the market.

But then the market got more sophisticated. You found that after you had lent money on, say a million dollars of mortgages, you could slice and dice that book of business and sell it on according to investors' risk aversions. And I will admit that on the surface this looks very appealing.

However, once the bank has done this, it is once again sitting on some capital that it wants to use. So it lends more money, and slices and dices that and sells it on. In fact, the bank no longer has any real interest in whether a loan it puts out is going to be repaid or not, except to the extent that it can parcel it up and sell it on. In theory, you could have a two-bit bank in Oklahoma (a state I mention with memory of the 1980s) lending billions of dollars in mortgages, parcelling them all up and selling them on in various tranches of CDOs.

My contention is, I guess, that the risk being accepted by the buyers of these CDOs is not compensated for by the real risk of default. There might be, say, a 2% risk of default on one tranche, but the investor is accepting 1% over base rate on that risk. I think that there is an imbalance because lenders are now less concerned about whether or not a loan defaults. Also, investors are chasing that "little bit extra" of return, without allowing for how much "little bit extra" risk is associated with that better return.

Now, if you get a mismatch between real risk and perceived risk you have a time bomb. Given the massive expansion of CDOs in recent years, this could mean (and I am not screaming inevitability here) that you have a bigger time bomb. The thing about insurance policies is that you and I can assume that the person who assumes the risk will have the funds to pay up if something goes wrong. But if CDOs are being sold on to investors and these investors are not always putting up 100% of the funds to pay for them, then to assign a 100% value to those funds is potentially dangerous.

Now I strongly suspect that a lot of the sellers of risk are not taking 100% cash up front when they transfer the risk off their books. But they are assigning a 100% certainty to that cash being paid. Now, suppose there is a mass default on these loans, that not just the Z tranches, but the medium "very safe" tranches suddenly head into junk bond territory. What we have to worry about now is the liquidity of the market and the extent to which some investors "walk away". At this point, something given a 100% value by the sellers of the risk turns out not to have that value. The sellers of the CDOs suddenly find that the risk has come back to them.

The reason that I am most worried about CDOs is not that I think they are the most likely instrument to cause a meltdown, but that, if a problem occurred, then CDOs could cause a much bigger meltdown than people realize.

Re: Just Curious...

[jellymillion.livejournal.com]

Ah. We have a terminology difference, rather than anything serious. The instrument you're referring to as a CDO I call a Collateralised Mortgage Obligation, the remainder of that story is well-told in Liar's Poker, for a start. I have a feeling that CDO was quite possibly a generic term for securitised debt at the time, though. The Savings & Loans of the time were particularly suckered into securitising their debt at fire-sale prices, but securitisation generally is still cheerfully going on, still largely mortgages and credit cards. The difference is that the issuers are rather better informed these days, probably because competition between the large number of institutions offering the service has removed the opportunity for scamming the naive.

CDOs as traded today are rather different. These days the term refers to a package of credit default swaps. A CDS is very much an insurance policy. Say you hold a corporate bond. It will trade at some premium to the risk-free (government) rate due to the possibility of the issuer (borrower) defaulting on its obligation. So as the holder you have a small chance of a big loss. But you can purchase protection against the Bad Thing happening by taking out insurance against it: you pay a regular premium and your counterparty pays out the insured amount in the event of any agreed "credit event" occurring. That event could be anything from a total collapse to a stipulated decline in credit rating, it depends what sort of protection you want. If the event happens, you deliver the asset (which may well have some residual value for the counterparty to pursue) and get your cash. You've converted a risky asset into a risk-free (or risk-reduced) one for giving up some of the return.

So now consider a portfolio of risky assets. A CDO packages up protection in agreed amounts across a set of issuers (names, or reference entities) and will pay out on one (first-to-default) or more (first N to default) credit events. All negotiable and tailorable. The premium is assessed according to the bank's estimation of the likelihood of payout over the life of the deal. And of course they can hedge away some or all of the package (re-insurance). A quantity of CDOs can be similarly packaged as a "CDO-squared" and there are even CDO-cubeds. A lot of attention is paid to the likelihood of a systemic failure (correlation risk) and valuing these beasts is tremendously good news for blade server manufacturers: Monte Carlo simulation is the only tool available to date and the bigger banks have literally thousands of grid-computing machines grinding away every night.

Of course, it all getsa more "interesting" when one considers that there's no actual need to be exposed to a particular name's credit - one can simply desire to acquire exposure because one considers that the relevant price is out-of-line. So the inevitable secondary market, arbitrage opportunities and even derivatives (there are now a raft of credit indices for hedging and outright speculation) follow hard on the original instruments' heels. Bags of fun for everyone.

And of course, post-Barings, risk has to be calculated every which way, sliced, diced, reported and managed. More Monte Carlo, more blade servers.

I've gone into this a little more than is probably necessary for you - but who knows who else might have stayed the distance...

Re: Just Curious...

[peterbirks.livejournal.com]

Yes, sorry, I should have said collateralized mortgage obligations. And I could also have had happy times talking about MBNA and how it "eliminated the risk" of default on its credit card lending. Bermuda enters into play here, as does a specially set up reinsurance company, several SPVs, and quite possibly an AIG affiliate somewhere along the line.

However, if we are taking the area of credit default swaps, also great fun, then I think there could also be trouble down the line. In particular the more risky "any one of these to default" tranches strike me as pretty horribly priced (for the buyer of the risk). Or, they were when I was looking at these things last year (it's not a field I deal with that much any more).

Although you might think that the naive buyer is a thing of the past, there are rumours that a lot of the mid-sized German savings banks have amassed acres and acres of risk at the wrong price.

As we know from Taleb, people underestimate the likelihood of unlikely events -- kind of the law of big numbers in reverse. If you have a tranche of any one of 10 triple A-rated companies to default within five years, well, I have no idea what the assigned probability of this in the market is, but I would have a hunch that it is too low and that the real risk of default is higher. The trouble is, we are probably talking about 0.4% instead of 0.1% assumed. So, most of the time, the event doesn't happen.

Incidentally, have you noticed the volatility in the spreads over the past couple of years? These are a bit like bankroll requirements in the Malmuth calculator. I'm sure that the likelihood of a Ford default kind of more than doubled and then halved, all within a couple of months about a year or so ago. Now, I don't care what you say, there is no way that the "real" risk doubled and then halved. What happened was that the perceived risk doubled and halved. Clear evidence, if evidence be needed, that the market swings too far in the world of perceived risk.

Re: Sorry, that should have been MBIA

[peterbirks.livejournal.com]

On the Bermuda thing and the SPV, I was thinking of the bond insurer MBIA rather than the credit card company MBNA. Brain not functioning properly after a hard day!

Re: Just Curious...

[jellymillion.livejournal.com]

As we know from Taleb, people underestimate the likelihood of unlikely events

Donning my cynical, "Grumpy Old Man" hat for a moment (I lie, it's always on), I'd suggest that people don't really underestimate the likelihood at all. Well, there are exceptions, such as the assumption in Black-Scholes that sufficient liquidity always exists in the market (LTCM - oops). I think they simply consider that the chance is sufficiently small that the event can be disregarded in the short term. In other words, there is a strong likelihood that they'll clean up - personally - through high salaries and higher bonuses for long enough that when the proverbial hits the fan they'll already have a nice pile of cash stashed, so it won't matter. Even better, if the event is broadly spread then they won't smell any worse than anyone else and they'll be able to move to another institution and do it all again.

Possible case in point - about 20 yards from me sits the guy who was head of Market Risk at Barings. I had dealings with him a few years back when building a risk model for a moderately complicated Japanese index futures & options portfolio. With no discernable irony, he said something like "I have quite a bit of experience with Nikkei 225 options from my time at Barings".

The spread volatility thing is interesting - with the typical geek's lack of in-depth knowledge, I'd suggest that part of it is a sign of a still relatively immature market (by comparison with fixed income and equities) where the secondary market is as yet failing to provide consistent liquidity and where the arb opportunities aren't yet being taken out fully by the prop desks and hedge funds. So the market tends to react in an overly cautious manner to unexpected news or credit-affecting corporate actions.

Also now curious...

[iadams.livejournal.com]

I've always used 'CDO' as 'collateralised Debt Obligation', a subset of which is 'CMO', the Collateralised Mortgage Obligation'; these forming a subset of 'ABS' or 'MBS', 'Asset/Mortgage Backed Security'. CDOs are a diiferent beast from CDSs. Back in the day (mid 90's), it started becoming fashionable for banks, and some large manufacturers, to collateralise future cash flows from loans, even accounts receivable. It's definitiely not just linked to mortgages.

Even David Bowie had a bond issue, backed by the future receipts of royalties from his back catalogue. It was brought to market by a boutique NY investment house, and almost the entire issue ended up in the portfolio of a US (Pennsylvanian?) insurance company. In fact, as a ten year issue, it should exist.

Pete's concerns about the system risks of CDOs are well made. What has happened on the surface is that much financial risk has been transferrred from medium/small banks to insurance companies. If the default rate on the underlying Obligations should rise sharply (and let's face it, the average default rate may be reasonable, but the actaul events tend to be bunched), the banks who have sold on the debt may find that they don't actually have the cash available to meet their own liabilities.

Not following these matters as closely as I used to, I don't know what the World Bank is saying about this market - it certainly used to take a strong line on the potential moral hazard of repaying investors who lost money on emerging market debt. Of course, when it comes to systemic risk to Western banking, the money seems always to be found (vide LTCM).

There was a case in the mid 1990s of a firm who created a short-squeeze in the CBOT 2year future, by buying lots of contracts, and then buying 80%+ of the (one) devliverable issue by recycling their purchases in the repo market. The regulator (CFTC, IIRC) fined them their profit (some hundreds of millions of USD), plus two or three times the same on top as a punitive measure.

Sorry to be so late into the discussion, I've been away.

Re: Also now curious...

[jellymillion.livejournal.com]

The biggest imponderable in the calculation of risk with these mixed bags (OK, baskets) of debt seems to be correlation. With CDOs (as defined by the credit derivs industry, in which I find myself mired at present from the Risk systems perspective) the terms of the contracts vary considerably, which also introduces a rather more difficult degree of - quite possibly hidden - risk. I suspect that on the other hand, the benefits - at least assuming there's no upcoming catastrophic event - are substantial. I'm thinking of liquidity here: any instrument that allows institutions to separate elements of risk and trade away the ones they're uncomfortable with should ultimately feed back in part to the borrower on the street in the form of slightly cheaper money.

At least, that's the argument I usually trot out when people ask me to justify the apparently parasitical business on which I leech in turn...

Mike

Re: Also now curious...

[peterbirks.livejournal.com]

Here I have to enter the world of reinsurance and mention the word "receivables". There is nothing wrong with instruments that separate out the constituent parts of a financial instrument and enable trading of those parts separately. Where the danger arises is if it gives a feeling of safety that may be illusional. So, an insurer passes on some of the risk to a reinsurer. This is fine if he can be 100%-sure that the reinsurer will pay. And, usually, that is what the insurer assumes. So the "receivable" is booked at 100%. But suppose the reinsurer has reinsured some of this risk. He too books it at 100%. You can get into a death spiral of reinsurance here (this is what nearly killed Lloyd's in the early 1990s) where what is counted as "certain" money is not certain. In other words, a financial party might have thought that it had traded away all the elements of risk, when it hadn't. This is what Mahoney meant when he asked "where has all the risk gone?"

Life assurers are a good example of where this splitting up of matters works well. Life asssurers aren't life assurers any more. They are asset managers who have stripped out the "insurance" side by passing on the life risk element in any variable annuity life contract (or any other type of life contract with a savings element) to a life reinsurer such as RGA or Scottish Re.

And the system works very well, even though it makes the nomenclature of some so-called "insurance" companies completely incorrect.

But life reinsurance is not catastrophe reinsurance or, more worryingly, debt insurance. If the systemic worry lies anywhere, it is probably with MBIA, or Ambac (or one of the other major bond insurers). The upshot of these companies seeing their triple A status taken away would be very interesting. If any of them (the bond insurers) saw their rating downgraded to junk (or even somewhere in the Bs) then I could see a credit crisis appearing pretty quickly.

[(Anonymous)]

Pete

What is risk to one person/institution is normal business to another. For example, why should a construction company that takes a loan out a libor + xxbp be subject to the vagaries of future interest rates if it does not have a natural offset in the normal course of its business? It makes much more sense to implement an interest rate swap and fix those liabilities, getting rid of the interest rate risk and allowing it to concentrate on its construction business without worrying about the future cost of its financing. Of course, it has instead the risk of default from the bank but so does the bank on the corporate.

The rates risk has been transferred to a bank but this is the normal part of a banks' business. The quantum of risk remains the same but it is more effectively managed within the bank that already has large fixed and floating rate exposures.

You can take this to a slightly higher level and prove that adding individual risk, if it is uncorrelated to the current exposures you have, will actually reduce the overall risk profile (see Markovitz' work on the efficient frontier).

Your point about liquidity is a good one. However, in times of financial crises or when there is systemic as opposed to idiosyncratic market movements, it is the derivatives market that always has greater liquidity than the cash market. This is one of the reasons why the CDS market has grown so fast since 2000 and has actually been strengthened by every credit event we have seen over the last few years.

With respect to the Pimco story the summary is that they owned 45% (approx $11bn) of the cheapest-to-deliver bond and $10bn of futures contracts, so the bond went special in the repo market...of course there are a lot more US T-bonds than that outstanding (over $5 trillion) so there is no danger of bonds not being available to deliver. But, this could be considered illegal market manipulation. You see, the legal risk has precedent over all others...just ask Ken Lay. Please see the Bloomberg story below:

Aug. 19 (Bloomberg) -- Pacific Investment Management Co.,
manager of the world's biggest bond fund, was sued by an
investor alleging the firm manipulated the price of June 10-year
Treasury futures contracts on the Chicago Board of Trade.
Raymond Chiu claims Pimco violated the Commodity Exchange
Act through ``manipulative conduct'' that created ``artificially
high prices'' in the futures market to its benefit. The suit,
filed on Aug. 16 in U.S. District Court in Chicago, seeks class-
action status.
``We strongly believe the complaint is without merit and we
intend to vigorously defend ourselves,'' said Mark Porterfield,
a spokesman for Newport Beach, California-based Pimco. The firm,
whose chief investment officer is Bill Gross, manages more than
$400 billion.
Gross last week gave an ``unequivocal'' denial to financial
news network CNBC that his firm was involved in a shortage, or
``squeeze'' of Treasuries related to the contract on which Chiu
based his complaint. Futures are agreements to buy or sell
securities at a set price and time.
Chiu said Pimco owned about $10 billion of contracts and
``billions of dollars worth'' of the most-sought after notes
that could be used to fulfill the futures contract.
U.S. Securities and Exchange Commission filings show that
on June 30 Pimco funds held more than $11 billion, or about 45
percent, of the 4 7/8 percent February 2012 Treasury note, which
was the so-called cheapest to deliver.
It is illegal to manipulate prices in the futures market by
attempting to corner supply of a deliverable security, according
to the Commodity Exchange Act.
Ryan Long, one of Chiu's lawyers at Lovell, Stewart
Halebian LLP, declined to comment. Calls to Chiu's lawyers at
Miller Faucher and Cafferty LLP weren't returned. Christopher
Lovell was the lead attorney in a 1998 suit alleging commodity
trader Phibro Inc. manipulated the silver market, before Warren
Buffett said he was the one buying up silver and Phibro was just
his broker.
The case is Raymond Chiu v. Pacific Investment Management
Company LLC, U.S. District Court for the Northern District of
Illinois Eastern Division, 05C-4681.

I am out of time and will have to save discussing CDOs for later
All the best
Lloyd