The problem of implicit recourse
While asset backed commercial paper programs started as a way for firms to borrow against the collateral of their financial assets, such as accounts receivable, it didn’t take long for banks to realize that a very wide variety of assets could be financed using ABCP. All of the loan types that banks held on their balance sheets in the past found their way into asset backed commercial paper conduits: residential mortgages, commercial mortgages, unsecured business loans, corporate bonds, government bonds, etc.
Thus by the late 1980s ABCP conduits had evolved such that banks were deliberately taking loans off their balance sheets and placing them in conduits in order to reduce the amount of capital that regulators required of them. For the most part regulators viewed this phenomenon as a positive development, since it meant that a third party, the conduit, would absorb a large portion of any losses on the loans. To the degree that the conduit would be taking the losses, there was no need for the bank to hold capital against the loans.
While the relationship between conduits and sponsoring financial institutions was subject to steadily increasing regulation over time, the conduits themselves were not regulated. They fall into the same regulatory category as hedge funds: because investments in conduits cannot be marketed or sold to the general public, regulators rely on the self-interest of sophisticated investors to act as a market force that makes government supervision of conduits unnecessary.
Through the 1990s, it became clear that the conduits which were created by banks to hold bank loans were far less independent in practice than they were on paper. While some banks ignored the structured finance revolution and continued to hold loans on their balance sheets as they had in the past, the banks that chose to create conduits developed a new business model. They originated loans with no intention of holding them on balance sheet. Banks that followed the new originate and distribute model of lending were dependent on their conduits in order to continue their loan activities.
Thus, when problematic assets were going to cause a conduit to default, the sponsoring bank suddenly found itself in a position where, if the conduit was allowed to default, the bank would lack the credibility to continue placing assets in any other conduit – and this would be extremely detrimental to its business model. Repeatedly throughout the 1990s regulators agreed that it was in a bank’s interest to support a faltering conduit – despite the fact that on paper the bank had no obligation to do so. In short, it became clear that when a bank sold a loan to a conduit, it was in practice a recourse sale – that is, a sale where the bank can be required to take back the asset when it goes into default.
Here, we run into a problem. The accounting rules for recourse sales are crystal clear: a recourse sale is not a “true sale,” so a loan sold with recourse cannot be removed from the seller’s balance sheet. While implicit recourse was not yet explicitly covered by accounting rules, the violation of accounting principles was evident. Throughout the later years of the 1990s, regulators struggled with the problem of implicit recourse for loans sold to conduits, and by 2002 had a new policy: Banks that provided recourse for assets that had been sold to conduits were threatened with an increase in regulatory capital requirements.
The main effect of stricter regulation was, however, to drive implicit recourse deeper underground. For example in the early years of the current decade there was a noteworthy growth in loans bought back from conduits on the basis of fraud. Under accounting rules a clause requiring the repurchase of fraudulent loans does not make the sale a recourse sale. In 2001, NextBank was forced by regulators to take a conduit on balance sheet, because it was clearly using “fraud losses” to hide the fact that it was buying back bad assets from the conduit. This caused the bank to fail and be seized by the FDIC. NextBank, however, is an exceptional case that did not set a precedent for the banking system as a whole. Researchers have found evidence that other banks were also using “fraud losses” to hide recourse loans and did not suffer regulatory interference.
In short, while it was well understood that banks were using conduits to reduce and even avoid regulatory capital requirements, it is not clear that anyone – the bankers, the regulators or the researchers – was genuinely concerned that the banks were undercapitalized. Many seem to have held the view that capital requirements were too strict and thus that, at least to some degree, avoiding them was in the interests of economic efficiency. (It is worth noting that in precisely these years the Basel II international regulatory standards for banks were being negotiated – and that one goal of Basel II was to reduce the conservative regulatory capital requirements that had been put in place by the Basel I accord.) The regulators, like the banks themselves, could not imagine a real-world scenario in which many conduits would simultaneously draw on their credit and liquidity facilities.
Efforts to regulate liquidity facilities
By March 2004, the Enron scandal had led to the revision of accounting standards. Enron used conduits to hide losses from the public. And the new accounting rule, FIN 46R, stated that, if a single organization was exposed to the majority of a conduit’s expected losses then that organization would have to take the conduit on balance sheet. The new rule also required firms to evaluate the degree to which they were either implicitly or explicitly exposed to losses from off balance sheet conduits on a regular basis. This new rule forced several banks to consolidate asset backed commercial paper conduits onto their balance sheets.
However, since banks were using conduits to fund new loans, not to hide losses, regulators did not view the accounting changes as particularly relevant to banking. The regulators argued that, because the banks’ exposure to losses from conduits was limited to the credit and liquidity enhancement that they were contractually obliged to provide, treating banks as if they were exposed to 100% of the conduits losses by forcing them to take the conduits on balance sheet results in excessive capital requirements. Thus, to this day the calculation of a bank’s risk-based capital requirement does not include any ABCP conduit assets that have been consolidated on to the bank’s balance sheet.
Despite such regulatory indulgence, banks have shown a strong preference for conduits that do not need to be consolidated on balance sheet. Now ABCP conduits are designed so that multiple banks or firms sell assets into the conduit and each is obliged to buy back only a portion of the assets in case of default. In this manner no single entity is exposed to the majority of the conduit’s expected losses. At the end of 2006 about half or $600 billion of the ABCP conduits in the US took this form.
Although the regulators decided to allow banks to ignore ABCP conduits when calculating their risk based capital requirements, they were well aware that implicit recourse could be a problem for conduits. Thus, the regulators also put into place stricter capital requirements for commitments to provide credit and liquidity enhancement.
Since 1992, the capital requirements that banks faced were much higher for credit enhancement than for liquidity support. In fact a bank that provided 100% credit enhancement to a conduit was required to hold capital, just as if the loans were on its balance sheet. When credit enhancement was only for a fraction of the conduit’s assets, the bank was required to treat that fraction of the assets as though they were on its balance sheet. On the other hand up through the middle of 2005 there was no capital requirement for any liquidity facility of less than one year’s duration. It didn’t take banks long to figure out that, by designing a 364 day liquidity facility that also provided credit enhancement, they could minimize regulatory capital requirements.
This practice of combining a liquidity facility with credit enhancement did not escape the notice of regulators. Thus, in the same final ruling that mitigated the effects of post-Enron accounting changes on banks, the capital requirements for short-term liquidity facilities were increased: “ineligible” liquidity facilities would be treated as equivalent to credit enhancement, while “eligible” liquidity facilities would require a tenth of the capital required for credit enhancement. The following scold was published with regulatory guidance on the ruling:
The agencies reiterate their position that the primary function of an eligible ABCP liquidity facility is to provide liquidity – not credit enhancement. Further, the agencies emphasize their view, as stated in the ABCP rule, that an eligible liquidity facility should not be used to purchase or otherwise fund assets with the high degree of credit risk typically associated with seriously delinquent and defaulted assets and assets that are below investment grade. 
An eligible ABCP liquidity facility includes a provision that reduces the funding obligation, before any funds are drawn, by the quantity of assets that are 90 days or more past due, in default, or below investment grade to the degree that these assets are in excess of the credit enhancement available to the conduit. In other words, a liquidity facility is eligible as long as it explicitly precludes the funding of high-risk assets that are not already covered by credit enhancement.,
It is worth observing that the qualification “to the degree that these assets are in excess of the credit enhancement available to the conduit” is not actually part of the final rule issued by regulators on July 28, 2004. In fact, in the background information published with the final rule, the regulators state that commenters had requested that “guarantees providing credit protection” be taken into account and for this reason a clause was added to final rule excepting assets protected by a government or agency guarantee. The relaxation of the final rule as it relates to private sector credit enhancement took the form of interagency guidance published on August 4, 2005. It is not clear to me how this guidance can be considered consistent with the plain text of the final rule.
In order for a conduit to receive an A1/P1/F1 rating from the credit rating agencies, typically a conduit must have 100% liquidity support. Thus, as soon as the assets in default exceed the credit enhancement available to the conduit, this rule creates a very awkward situation for the sponsoring bank. Either the liquidity facility is reduced so that it does not cover the impaired assets and this puts the conduit’s credit rating at risk, or the liquidity facility becomes ineligible and the bank must treat all the assets in the conduit as if they were on balance sheet.
In the context of this regulatory environment, let’s try to make sense of an event that shocked the financial world in November of 2007: $25 billion of new sub-prime CDO exposure suddenly appeared on Citigroup’s third quarter balance sheet. The question everybody was asking was: How could Citi have failed to report this exposure in previous financial statements? The explanation: When Citi sold these CDOs to conduits they were sold with a “liquidity put.” The liquidity put “allowed any buyer of these CDOs who ran into financing problems to sell them back – at original value – to Citi.”
How can a “liquidity put” be consistent with the regulatory standards we’ve just explained? A liquidity put could be part of an eligible liquidity facility as long as financing problems could trigger a repurchase of the asset only if the asset retained its first tier rating. Furthermore, since typically under the law, the reason a recourse sale is not considered a “true sale” is that the seller retains the credit risk of the loan, a liquidity put may not be used to purchase an impaired asset. Thus a liquidity put must be triggered by a financing problem and explicitly rule out repurchase of an asset that has been downgraded or is seriously delinquent.
The example of Citibank indicates, however, that, because delinquencies and downgrades are anticipated before they occur and commercial paper is rolled over on a frequent basis, the expectation of asset impairment precipitates a liquidity crisis for the conduit before the assets are formally recognized as impaired. Thus, in practice, the seller of a liquidity put is exposed to the credit risk of the underlying loans.
As a dramatic overhaul of the treatment of off-balance sheet entities in financial statements is underway, it is appears that regulators and accounting professionals have a clear understanding of the deficiencies of recent industry practices. 
 CREDIT CARD SECURITIZATION AND REGULATORY ARBITRAGE by Charles W. Calomiris and Joseph R. Mason, April 2003, FEDERAL RESERVE BANK OF PHILADELPHIA WORKING PAPER NO. 03-7, p. 9.
 From a final rule published in the Federal Register on July 28, 2004 and available at: http://www.phil.frb.org/publicaffairs/circulars/attachment5631.pdf
“The agencies [The Office of the Comptroller of the Currency, Board of Governors of the Federal Reserve System, Federal Deposit Insurance Corporation, and Office of Thrift Supervision] believe that the consolidation of ABCP program assets generally would result in risk-based capital requirements that do not appropriately reflect the risks faced by banking organizations involved with the programs. Sponsoring banking organizations generally face limited risk exposure to ABCP programs. This risk usually is confined to the credit enhancements and liquidity facility arrangements that sponsoring banking organizations provide to these programs. In addition, operational controls and structural provisions, along with overcollateralization or other credit enhancements provided by the companies that sell assets into ABCP programs, mitigate the risks to which sponsoring banking organizations are exposed. …
The final rule will make permanent the exclusion of ABCP program assets consolidated under FIN 46-R and any associated minority interests from risk-weighted assets and tier 1 capital, respectively, when sponsoring banking organizations calculate their tier 1 and total risk-based capital ratios.”
“Effective September 30,2004, the federal banking agencies amended the risk-based capital standards to permit bank holding companies and other sponsoring banking organizations, when calculating their risk-based capital ratios, to exclude from their risk-weighted asset base those assets in asset-backed commercial paper (ABCP) programs that are consolidated onto the sponsoring banking organizations’ balance sheets.”
Federal Bank Holding Company Law, 1997, by Pauline B. Heller, Melanie Fein p. 5-16.7
 MARKETS AND INVESTING: Guide to the vehicles: conduits, SIVs and SIV-lites
Published: Aug 15, 2007
 SR Letters 2005-13. http://www.federalreserve.gov/boarddocs/SRLETTERS/2005/SR0513a1.pdf .
 http://www.federalreserve.gov/boarddocs/press/bcreg/2004/20040720/attachment.pdf, p. 18. Note that the original rule proposed that eligible liquidity facilities require one fifth of the capital required for credit enhancement and that the liquidity facility be reduced by the quantity of loans that were 60 days or more past due.
 SR Letters 2005-13. http://www.federalreserve.gov/boarddocs/SRLETTERS/2005/SR0513a1.pdf .
Loomis, Carol, November 28 2007 Robert Rubin on the job he never wanted, Fortune Magazine http://money.cnn.com/2007/11/09/news/newsmakers/merrill_rubin.fortune/index.htm?postversion=2007111119
Bank of America also announced $10 billion of CDO exposure due to liquidity puts. http://www.nytimes.com/2007/11/16/business/16norris.html