To analyze whether or not the safe harbor amendments increase systemic risk, let me present a model of systemic risk that differs from the one presented by the President’s Working Group in the quote above. I assume that the banks are best suited to evaluate each other’s business practices. Then in order to reduce systemic risk, a regulator’s goal should be to take advantage of the banks’ superior knowledge of the business.
When banks lend to other banks on a secured basis, they are protected from losses and may be willing to do business with unsound counterparties. By contrast, when banks lend to other banks on an unsecured basis, they are not protected from losses and will reduce exposure at the first sign that a counterparty is poorly managed. Thus, when banks lend to each other on an unsecured basis, lending by the banking system to unsound banks will be small; and when banks lend to each other on a secured basis, lending by the banking system to unsound banks may be quite large. Systemic risk will be much higher when the activities of banks that mismanage risk are not curtailed by counterparties but are in fact facilitated and allowed to grow. Since the finance of unsound banks facilitates mismanaged risk, my model indicates that systemic risk will be high when secured interbank lending is the norm and will tend to be low when most interbank lending is unsecured.
The exemptions for financial contracts adopted in the 2005 Bankruptcy Act are extremely broad. They guarantee to banks that, as long as they have required sufficient posting of collateral on their contracts, they are fully protected from loss. This law would address the problem of systemic risk, only if it were the case that protecting the banking system from losses was the best way to protect it from systemic risk. In fact, however, banks are specialists in managing credit risk and thus being exposed to losses due to credit risk is an essential aspect of the services a bank provides to the economy. For this reason it is not clear that a policy of protecting banks from losses due to credit risk is consistent with the role played by banks in the economy.
In my model, unsecured interbank lending reduces systemic risk, because it increases the credit risk faced by banks and forces them to monitor their counterparties carefully. Thus, this model is consistent with the view that a principal banking function is to manage credit risk. This approach to banking indicates that the safe harbor exemptions to the bankruptcy code may have increased systemic risk by encouraging banks to lend to each other on a secured basis. To test this model against the data, we would want to know, first, whether we see in the data an increase in secured lending after the safe harbor exemptions are passed and, second, whether there is any evidence to support the view that systemic risk increased after the passage of these laws and the increase in secured interbank lending.
The evidence indicates that the use of both repurchase agreements and of collateralized derivatives has increased dramatically since the passage of the safe harbor amendments. In the 1983 hearings on safe harbor for repurchase agreements the market was estimated to be several hundred billion dollars daily with more than $100 billion in prime dealer repos. More recently the repo market has been estimated to be about $12 trillion with the prime dealers accounting for $4.2 trillion in early 2008. Thus, over the past 25 years the repo market has grown on average about 15% per year.
The swaps market also grew extremely quickly. In 1989 interest rate and currency swaps totaled $2.5 trillion. By the end of 2008 there were $403 trillion of these contracts outstanding. The average annual growth rate over this 19 year period was more than 30%. The growth rate of credit default swaps was, however, most astounding. Data on credit default swaps was first reported in 2001 when they amounted to less than $1 trillion in notional value. By the end of 2007 the market had grown to $62 trillion. The size of the market for credit default swaps almost doubled every year for six years. Given that these contracts barely existed in the 1970s, the rate of growth in the swaps market is truly remarkable.
While repurchase agreements are always collateralized, whether or not collateral is posted on derivatives contracts depends on the details of the contract in question. Some parties have to post collateral to cover the full amount owed – hedge funds generally fall into this category. Others post collateral only if the value of the contract moves against them by more than some threshold amount – the dealer banks are an example. For example in November 2007 AIG had contracts with Merrill Lynch with thresholds of 8%, meaning that the value of the swap had to fall to 92% of the initial value before AIG had to post collateral. At the same time Goldman Sachs had 4% thresholds on their swaps with AIG.
Because the collateral terms of every derivative contract can differ, the use of collateral for derivatives contracts is estimated using surveys. The results from these surveys are published by the ISDA. From 2000 through 2006 collateral use grew along with the derivatives market itself. However in 2007 and 2008 the use of collateral increased even as the derivatives markets themselves finally stopped growing. The increased use of collateral in recent years is due in part to changes in the terms of the contracts themselves and in part to the fact that there have been dramatic shifts in the value of many contracts, thus increasing the number of contracts that have crossed a threshold for posting collateral. We can safely conclude that there was indeed a large increase in secured interbank lending after the safe harbor amendments were passed both because of the growing use of repurchase agreements and because the use of collateral grew along with the derivatives market.
The evidence is also consistent with the view that systemic risk increased after the passage of the bankruptcy amendments: after all the greatest financial crisis in three-quarters of a century took place just a few years after the 2005 amendment widened the scope of the bankruptcy exemptions dramatically. While correlation is not causation, the sequence of events is indeed consistent with the view that the bankruptcy amendments caused a dramatic increase in secured interbank lending, which facilitated the operation of financial institutions that mismanaged risk. It was the failure of these institutions – Bear Stearns, Lehman Brothers, AIG – that caused the crisis and set off the systemic risk event.
 Remarkably, they are protected even if the transfer of collateral was fraudulent as long as the collateral was received in good faith. (President’s Working Group on Fin. Mkts., Hedge Funds, Leverage, and the Lessons of Long-Term Capital Management (1999), p. E-4)
 1983 repo data from Kettering, 2008, note 287. Data on prime dealers from Federal Reserve Bulletin. Available at http://fraser.stlouisfed.org/publications/frb/page/31488.
 Gary Gorton, Dec 31, 2008, “Information Liquidity and the (Ongoing) Panic of 2007,” p. 8 (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1324195) and Federal Reserve Bank of New York.
 Gary Gorton, 2009, “Slapped in the Face by the Invisible Hand: Banking and the Panic of 2007”, p. 11. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1401882&rec=1&srcabs=1404069
 AIG internal collateral memo: http://www.cbsnews.com/htdocs/pdf/collateral_b.pdf
 ISDA http://www.isda.org/c_and_a/pdf/ISDA-Margin-Survey-2009.pdf
In 2003 only 30% of over the counter derivatives contracts required that collateral be posted when the contract moved against the counterparty and by 2009 65% of the contracts were collateralized.