The Many Problems with Collateralized Interbank Lending #1-10

In addition to creating systemic risk and obfuscating financial reports, collateralized derivatives are not a good tool for hedging risk.  Let me be more specific:  derivative contracts can be perfect hedges as long as there is no risk that either counterparty will default.  Counterparty risk means that the derivative contract may be worthless and undermines its role as a hedge.  Collateralized derivatives are designed to address counterparty risk.  The problem with this “solution” is that collateralizing a derivative contract can undermine its use as a hedge, too.

First note that the standard analytic framework used to explain how derivatives serve to hedge risk assumes that both counterparties are money good.  In particular whenever a proponent of derivatives claims that they can be distinguished from other securities by the fact that they are “zero sum” and not subject to the losses that we see in asset markets, that individual is assuming that derivative contracts are not subject to counterparty risk.[1] It is true that, if one assumes away counterparty risk, derivative contracts can offset certain types of economic risk perfectly. An example is an airline that enters into a futures contract where the airline contracts to buy fuel three months in the future at today’s market price.  This airline has a perfect hedge against the risk that fuel prices will rise – as long as the airline’s counterparty honors his contractual obligations.[2]

One reason that the financial community developed the habit of assuming away counterparty risk when discussing derivatives is that, from the early years of the 20th century through approximately 1990, the vast majority of derivative contracts were exchange-traded.  When a derivative is traded on an exchange, the exchange is the counterparty to every transaction.  Because the exchanges put in place safeguards protecting them from losses, they are stable organizations and in the United States none has failed.  Thus, it is not unreasonable to make the assumption that there is no counterparty risk when discussing exchange-traded derivatives.

Over the counter derivatives differ from exchange traded derivatives because they are simply private contracts.  As a practical matter the only restriction on an over the counter derivative contract is that the two contract participants accept each other as counterparties.  Clearly counterparty risk can be a factor when dealing with over the counter derivatives.

In the event that a derivative counterparty defaults – and no collateral has been posted – the derivative contract no longer serves as a hedge.  Returning to our airline example, if fuel prices rise and the counterparty declares bankruptcy, the airline will be an unsecured creditor for the value of the contract at the date of bankruptcy.  The airline will be unhedged if fuel prices continue to rise after bankruptcy is declared and, like all creditors, the airline may receive partial or no recovery on its claim.  For this reason the credit risk of an uncollateralized over the counter derivative is more comparable to a financial asset such as a bond than to an exchange traded derivative – and uncollateralized over the counter derivatives are subject to losses due to default just like any other financial asset.

One way to mitigate the credit risk inherent in an over the counter derivative is to include a clause in the contract that requires the counterparty who owes money to post collateral.  This is comparable to the margin requirements used by exchanges, except that the terms of an over the counter contract are not standardized.  As late as 2003, the ISDA Margin Survey reported that only 30% of over the counter derivative contracts required that collateral be posted. Only three quarters of these collateralized contracts had terms that required both parties to post collateral.  Over time collateralized contracts have become the norm.  The 2009 survey reported that 65% of contracts included clauses requiring collateral.[3]

To understand how collateral works let’s use our airline example again and recall that the price fixed for the contract was the market price on the day the contract was signed.  If the price of fuel has fallen, so has the value of the airline’s contract – if the contract expired today the airline would lose money on it by paying a higher price for fuel than the market price.  In this case, the airline is “out of the money” and the airline’s counterparty is “in the money”.  (By contrast, if the price of fuel rises, the airline is “in the money” and the counterparty is “out of the money”.)  The purpose of collateral is to protect the “in the money” counterparty from the risk that the “out of the money” counterparty defaults on the derivative contract.

If the airline’s contract required collateral posting and the price of fuel fell steadily for the first month, then every day the airline would be required to post more collateral to cover the difference between the market price and the contracted price.  By contrast, when the price of fuel starts to rise, collateral will be returned to the airline.  And if the price of fuel exceeds the contracted price, the counterparty will post collateral to the airline.[4]

When there are no collateral requirements, the “in the money” counterparty to a futures contract faces credit risk.  When margin is posted, the “in the money” counterparty is protected from credit risk, and the “out of the money” counterparty must be prepared to post collateral well before actual payment on the contract is due.  Thus, a collateralized derivative contract protects against credit risk at the cost of creating liquidity risk (that is, the danger that liquid assets are not available to use as collateral) for the “out of the money” counterparty.  Futures exchanges have imposed margin requirements for over a century and, thus, have demonstrated that, despite the increase in liquidity risk, collateral is an effective way to protect the futures market from credit risk and guarantee its stability.

Over the counter derivatives, however, are sometimes very different from futures contracts.  Swaps in particular involve not just a single future payment, but repeated payments over an extended period of time.  Because the collateral posting requirements for a swap can involve summing over twenty or more separate payments, in some cases collateral becomes an onerous obligation.  The effect of a collateralized swap that by chance moves dramatically against one of the counterparties is similar to someone with a mortgage suddenly being called upon to have the full value of the house deposited at the bank.  Liquidity risk, when it involves large sums, can be a serious danger for derivatives investors.  This liquidity risk may undermine the swap’s effectiveness as a hedge.

To examine in more detail how collateral can undermine the use of a swap as a hedge, consider the example of a “plain vanilla” interest rate swap.  A university endowment has $10 million of debt on which it must pay the money market rate plus 3% for the next five years.  When the money market rate is 3%, the endowment will pay $50,000 per month and when the money market rate is 2% the endowment will pay $41,667 per month, and so on.[5] In order to protect itself from the possibility that interest rates rise, the endowment enters into an interest rate swap where it pays out a fixed rate of 6% per year (or $50,000 per month) and receives the money market rate plus 3% for five years.  Because the endowment receives from the swap exactly the amount that it needs to pay out on its debt, the swap is a perfect hedge for the endowment – if it is uncollateralized and the endowment’s counterparty does not default.

In the absence of a collateral agreement, the endowment has converted its adjustable rate debt into fixed rate debt – its liability is just $50,000 per month for five years.  If, however, the swap includes a collateral agreement and the money market rate falls and is expected to stay low for years, the endowment will be “out of the money”.  To illustrate what can happen to an “out of the money” swap counterparty, let’s look at the worst case scenario, where the money market rate falls to zero and is expected to stay there for five years:  in this example, the endowment pays $50,000 and receives $25,000 every month for five years.  Then the collateral the endowment can be asked to post is the present value of $25,000 a month for five years; when interest rates are close to zero, this approaches $1,500,000.  Thus, when a collateral regime is combined with a swap contract, it can have the effect of requiring a counterparty to have the means to prepay its obligations before they are due.

Because collateralized swap contracts involve large-scale liquidity risk, the only sense in which the endowment is hedged is on its balance sheet.  For the balance sheet the timing of the payments is irrelevant – all that matters is the total value of the firm’s claims and obligations.  On the balance sheet posting cash collateral is just a matter of moving the amount in question from cash assets to receivables.  Both of these items are assets so this change has a neutral effect on the financial report.

By any other standard, however, the endowment is not hedged.  In particular, there’s always the possibility that “cash assets” aren’t large enough to sustain the withdrawal of the cash needed for collateral.  In this situation, the endowment has a cash flow problem – just as a homeowner would have a cash flow problem if the mortgage lender had the right to demand that the full value of the mortgage be deposited at the bank.  “Liquidity risk” is a term that refers specifically to this problem.

The role of collateral is therefore to protect the “in the money” counterparty from credit risk at the cost of exposing the “out of the money” counterparty to liquidity risk.  For over the counter derivatives like swaps, it is far from clear that the gains from collateralizing the swap outweigh the costs.  Our example of an endowment was not entirely hypothetical.  Harvard University’s endowment faced precisely this situation – and, because cash flow was unavailable when needed, the University was forced to terminate the swaps, when their value was close to a nadir.  The University posted huge losses, but more importantly lost the hedge it had invested in over a period of years.  If the contract had not been collateralized, the University would have continued to make payments and in the event that interest rates rose again in the future, the University would have been protected.  Instead, the collateral terms of the derivative agreement were extremely risky – and cost the University its hedge.[6]

Interest rate swaps are not the only contracts that can require large amounts of collateral.  The terms of a credit default swap require the counterparty known as the protection buyer to make regular payments, and the other counterparty (called the protection seller) to make a much larger payment only if a specific firm defaults on its debt.  As the likelihood of a default by the underlying firm increases, the protection seller must post collateral.  Because the promised payment is large, the collateral requirement may also be large.  AIG is an example of a firm that failed due to collateral calls on credit default swaps – if the government had not taken over its obligations, it would have been forced to declare bankruptcy.

Not only do collateral requirements in some cases undermine the effectiveness of the contract as a hedge, but collateralized over the counter derivatives don’t necessarily provide effective protection against credit risk – despite safe harbor privileges.  The collateral necessary to cover the “out of the money” counterparty’s obligations changes every day with changes in the value of the asset or rate underlying the derivative.  In some cases these changes can be dramatic resulting in extremely large collateral calls.  Of course, it is precisely when price changes are dramatic that market turmoil and collateral calls are likely to cause the bankruptcy of a derivative counterparty.  Naturally, when a firm is bankrupted by collateral calls, then the firm’s counterparties will find that they are not holding enough collateral to cover the full amount due – and they will become unsecured claimants in bankruptcy court for the remainder.  In short, collateral requirements may not be an effective way to protect firms from losses due to market turmoil and sudden price movements.

Collateral requirements on credit default swaps are particularly difficult to manage.  In a credit default swap the protection seller owes nothing until a credit event occurs and, once the credit event occurs, the seller may owe tens of millions of dollars.  Thus, by their nature default swaps involve sudden changes in value.  Since protection buyers want to protect themselves from the credit risk inherent in such contracts, they use the cost of replacing the swap or the price of the referenced bond to estimate the likelihood that a credit event will occur.  Unfortunately, credit default swaps and the bonds they are written on often trade infrequently and may be difficult to price.  For example, when AIG faced collateral calls in late 2007, Joseph Cassano complained in a memo summarizing those calls:

[T]he prices we have received are all over the place and everyone we talk to has openly admitted that the bonds we are referencing have not, and do not, trade … As you can see where we do have more than one level they are never that close.  As a few examples, Goldman priced Dunhill at 75 and Merrill priced it at 95:  Independence V is subject to collateral call from both ML and GS, but the former calculates a price of 90 and the latter is using 67.5.[7]

As the AIG memo indicates, when there is no objective way to determine a fair price for a credit default swap, collateral calls are made without a clear foundation.  Thus, credit swaps have the particular problem that it is often difficult to determine how much collateral should be posted until a credit event actually takes place and full payment is due.

These two factors, that sudden price changes can render collateral inadequate and that for some derivatives it is very difficult to determine appropriate collateral requirements, mean that collateral may fail to protect the “in the money” counterparty from losses.  Thus, yet another problem with collateralized derivatives is that they can create the illusion of protecting a firm from credit risk without actually protecting the firm – particularly in a tumultuous market when protection is most needed.

Finally it is possible that, in the event of the failure of a large financial firm, the safe harbor provisions will actually serve to increase credit losses.  At the moment that a large firm declares bankruptcy many counterparties will simultaneously seize and sell collateral.  This process of exercising safe harbor privileges can generate the sudden sale of a large number of assets and have the effect of reducing the value of the collateral that was posted.  It may even be the case that these losses are just as large, if not larger, than the losses from having the collateral tied up in bankruptcy court for a few weeks.

In 1998 Long Term Capital Management, a hedge fund with a massive balance sheet, was heading towards bankruptcy and a fire sale of collateral loomed.  At the behest of the Federal Reserve Bank of New York, the major investment banks – who were also the hedge fund’s largest creditors – intervened, effectively taking over the hedge fund.  While a fire sale of assets was averted, the danger was acknowledged by everyone in the financial community.  In 1999 the ISDA published its first collateral review and found that cash collateral was “less commonly used”.[8] From that date forward the survey clearly documented increasing use of cash collateral in derivatives contracts.  In 2009 this trend culminated in cash accounting for 84% of all derivative collateral. [9] The reasoning behind this trend seems obvious:  cash collateral is the only form of margin that cannot lose value in the event that there is a fire sale.

Nowadays, it is repurchase agreement counterparties who worry that safe harbor provisions will result in a fire sale of collateral.  This is, in part, due to the nature of repos – they are always secured by a financial asset that can be sold – but also due to the fact that in 2005 safe harbor was granted to repos of less liquid securities like investment grade debt.  Of course, less liquid securities are far more likely to lose value dramatically in a fire sale, than, for example, government bonds.  For this reason repo markets that (i) trade in securities of limited liquidity and (ii) are granted safe harbor are inherently unstable.

[1] See for example the testimony of Richard R. Lindsay before the Senate Agriculture Committee on October 14, 2008. This is a commonly held view of derivatives.

[2] Of course, if fuel prices fall, the airline is locked into the higher price – this is the nature of a futures contract.  An airline that wishes to protect against a possible rise in prices without giving up the benefits of a fall in prices must pay an upfront fee for an option contract.

[3] ISDA Margin Survey, 2009,
ISDA Margin Survey 2003,
In 2009 one quarter of collateralized contracts only required one side to post collateral.

[4] For simplicity of exposition, I am assuming that the collateral agreement is bilateral and that both parties have a zero threshold for posting collateral.

[5] When the money market rate is 3%, (3% + 3%)/ 12 months = 0.5%.  Thus, each monthly payment is 0.5% of the principal or $50,000.  When the money market rate is 2%, (2% + 5%)/12 months = 0.4167%.

[6] Richard Bradley, 2009, “Drew Gilpin Faust and the Incredible Shrinking Harvard”, Boston Magazine.


[8] 1999 ISDA Collateral Review, p. 9

[9] 2009 ISDA Margin Survey.  While cash collateral has increased in recent years, even in 2003 70% of collateral was cash.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s