How You Bet Against the Housing Market Before It Fell

The Problem With Shorting a Bond

Shorting a stock is easy. You borrow shares sell them and buy them back later. Bonds are more difficult. Mortgage-backed securities in particular traded infrequently and in large quantities so borrowing short-term was impractical

The solution was the credit default swap. A CDS is a contract in which one of the parties pays a regular premium and the other agrees to compensate it if a specific security defaults. Functionally it is a bond insurance

The distinction that makes this work is that a swap is a contract created on demand rather than an asset that must be located. Two parties who want opposing exposures can enter into one between themselves of any size they agree on without anyone having to find the underlying bond persuade its owner to lend it out or worry about being asked to return it at an inconvenient time

The Feature That Mattered

The critical detail is that it was not necessary to own the bond to purchase protection. That turns an insurance contract into a directional bet

Buying fire insurance on a house you don't own would be absurd and in insurance it is prohibited precisely because it creates an incentive for arson. In credit derivatives it was allowed. Investors who believed that subprime mortgage bonds would default could buy protection on them paying a modest annual premium versus a very large payment if they were right

How the Contract Actually Worked

The mechanics are important because they determine what the position does to you while you wait which turned out to be the most difficult part of the operation

A contract specifies a notional amount a reference obligation that names the exact value and a premium quoted in basis points per year over that notional paid in installments. The protection buyer pays. The seller collects and owes money only if a defined credit event occurs

For contracts on mortgage securities settlement was structured differently than a corporate CDS and the difference is important. A corporate contract typically resolves once in the event of bankruptcy or a missed payment. A mortgage bond does not fail suddenly. It bleeds as the underlying loans default and losses are amortized over months and years

So these contracts were paid as they went. When a loss was charged to the reference bond the protection seller paid that amount and the notional was reduced accordingly. Interest shortfalls were carried forward as they occurred. There was no single dramatic moment of default just a series of payments that came in as the collateral deteriorated which is why the positions took so long to resolve even after the analysis had been justified

The Asymmetry

The deal was attractive because the reward was extremely skewed. The premiums on highly rated securities were small because the market considered default remote. If the bonds defaulted the buyer only lost the accumulated premiums. If they defaulted the payout was many multiples of that

That's the structure of an option and that's why so few people need to be right for the benefit to be huge

Put arithmetic to it. For 100 notional with a premium of 200 basis points the buyer pays 2 a year. I have held the position for two years and 4 have been remunerated. If the bond then amortizes 80 percent of its value the payment is 80 which is twenty times what it cost to maintain

Notice what the same arithmetic does with a slower result. You've been in the job for four years instead of two and you've been paid 8 so the same payout of 80 is ten times the cost instead of twenty. The trade doesn't stop working if you get in early it just gets less and less good and the premium is paid out of a fund whose investors are watching

Why Being Right Early Nearly Failed

The part most often left out is how painful it was to hold the position. Investors who identified the problem were early by a considerable margin and premiums had to be paid every time

Worse as long as the bonds continued to function the market value of the protection fell. Funds holding these positions faced mounting paper losses and had to explain to investors why they were constantly paying for a bet that appeared to be losing. Several of them faced bailout demands that would have forced them to close positions before the reward arrived

This is the recurring lesson about the difference between correct analysis and a profitable trade. Solvency has to outlast the market's willingness to disagree with you and that is a financing issue rather than an analytical one

The Collateral Calls Were the Real Threat

The loss of the paper was not only embarrassing. It generated demands for cash and that is the mechanism that was most likely to end these positions before they paid out

Derivative contracts are collateralized against their current market value. When a position moves against you the counterparty has the right to request collateral to cover the exposure it now has and that collateral must be deposited in cash within a few days

Thus a protection buyer whose thesis had not yet been justified faced two capital outflows at the same time. The premium which was the known cost of the operation. And the collateral calls that grew as the market moved further against the position which were not budgeted for and which came precisely when the operation seemed worst

There was a layer of dispute on top. The value of an illiquid contract over an illiquid bond is a matter of opinion and the party asking for collateral was frequently the same type of institution that had sold the protection. Disagreements over branding were disagreements over how much cash should be delivered that week

The object lesson generalizes well beyond this episode. A position with asymmetric profitability still requires the ability to meet symmetric funding demands along the way and the size that is analytically correct is often larger than the size that can be survived

The Index Made It a Market

Tailored contracts on individual bonds are difficult to trade and value. What made this something a fund could bring to scale was the advent of standardized indices that referenced baskets of subprime mortgage securities of a given category and rating

An index contract is fungible. It has a published price multiple traders quoting it and standard terms meaning a position can be established quickly sized meaningfully and compared to something other than the counterparty's opinion

The trade-off is precision. An index is a basket so an opinion on a particularly bad set of loans is expressed only in a diluted form and the index may contain securities that behave differently than those identified by the analysis. Investors who had done loan-level work generally preferred single-name contracts over the specific bonds they had studied and accepted worse liquidity for a clearer expression of their view

The index also did something that single-name contracts could not do. It produced a visible continuously quoted price for the market's view on subprime credit so its decline became the visible signal that the position had begun to work

Getting Set Up to Trade It At All

Before any of this could be expressed there was an access issue that is left out of the counts. These contracts were not available to anyone who wanted one

Negotiating them required a framework agreement with each distributor bank negotiated by lawyers regulating how the contracts work what is considered non-compliance and how disputes are handled. Along with it was a guarantee annex that specified how much should be published how often and with respect to whose valuation. Those documents take time to implement and the distributor decides whether it is worth the effort

Then there was size. Traders traded in institutional quantities so a position had to be large enough to merit a seller's attention putting trading out of reach of individuals entirely and in practice out of reach of small funds

And there was the question of whether a broker would sell him protection on securities he had subscribed to and could still hold. Some were happy to do so believing the bonds were solid and the premium was free income. Others were less accommodating once the flow began to seem one-way

None of this is a technical note. The barrier to entry is a substantial part of why so few participants took the position and why the analysis available in public files did not translate into many people acting on it

Who Was on the Other Side

Someone sold all that protection. Much of it was written by insurers and bank trading desks that treated the premiums as almost free income reasoning that highly rated mortgage securities essentially never defaulted

Selling protection on a rare event is profitable almost every year and catastrophic the year it is not. It is the same structure as selling options with a lot of money and accumulates small constant profits in the face of a huge concentrated loss. Companies that had issued large amounts required bailout when payments came due simultaneously

The Version Available to Everyone Else

Anyone who didn't have relationships with dealers and came to the same conclusion had one path left: short the shares of companies exposed to the outcome. The mortgage lenders bond insurers and banks that held the securities were publicly traded and loanable

He expressed the same opinion and behaved much worse. The price of a share reflects the entire company so a correct view of one part of the business can be overshadowed by everything else going on. Borrowing the obvious names became expensive and occasionally unavailable as more people had the same idea which is a cost the exchange did not have

The payout was also limited in a way that derivatives were not. A stock can only fall to zero so the maximum profit is the value of the position while the loss of a position that moves against it has no such limit. This is the opposite of the asymmetry that made the swap attractive and means that the stock version required much better timing to produce a comparable result

What is the general point about expressing an opinion? The analysis identifies what will happen. The instrument determines whether there is any benefit to being right and the two questions are almost completely separate

Being Right and Not Being Paid

The last risk of the structure is the one that turns a winning analysis into nothing and follows directly from the previous section

A credit default swap is a promise from a counterparty. If the event you are insured against is the same event that harms the party that owes you the protection is worth what they can actually pay. The buyers of this protection were in effect betting on a systemic collapse of mortgage credit while depending on institutions highly exposed to mortgage credit to write them a very large check afterwards

That's not a hypothetical concern. It's why bailouts of certain protection sellers were important to the people who had been right and it's why sophisticated versions of the trade paid close attention to the counterparties they were facing how collateral arrangements worked and whether exposure was spread across multiple institutions rather than concentrated in one

The general principle is worth taking away from the episode. Any position whose profit depends on a counterparty acting in a crisis contains a second bet that no one chose to make which is a bet on that counterparty surviving the same crisis

The Bottom Line

Credit default swaps allow investors to short bonds that they cannot borrow and allow others to sell disaster insurance without having reserves to cover them. The winning trade required being right and more difficult maintaining sufficient funds long enough to collect