Regulation & Unintended Consequences: The Recourse Rule
I recently finished a working paper called “The Recourse Rule, Regulatory Arbitrage and the Financial Crisis.” The paper looks at how an…
I recently finished a working paper called “The Recourse Rule, Regulatory Arbitrage and the Financial Crisis.” The paper looks at how an obscure bank regulation, known as the “Recourse Rule”, might have contributed to the last financial crisis in an unintended way by creating incentives for banks to hold the very assets that lay at the heart of the last crisis. In what follows, I will walk you through a summary of my arguments.
“Recourse Rule” is actually the short name for what the Federal Register reported on November 29, 2001 as final rules on “Risk-Based Capital Guidelines; Capital Adequacy Guidelines; Capital Maintenance: Capital Treatment of Recourse, Direct Credit Substitutes and Residual Interests in Asset Securitizations.” To understand what all that means, I’ll have to explain “rule,” regulatory capital, how that relates to “Recourse” and the securitization of assets, and then how all of this relates to the last crisis.
The Recourse Rule gets no mention in highly publicized explanations of the crisis, such as Inside Job or The Big Short (although if you’re new to this area, and have seen them, they’ll give you a good basic understanding of the products that lay at the heart of the last crisis). I myself, didn’t hear about the rule until 2010, when I read Arnold Kling’s Not What They Had in Mind (he goes into more detail in a more recent contribution). Mike Konzcal had written in the Atlantic about Kling’s view in 2009.
Jeffrey Friedman and Wladimir Kraus subsequently devoted a chapter to the rule in their book Engineering the Crisis, which put forth detailed hypotheses about how the rule could have contributed to the crisis. Kevin Williamson referenced Friedman in a 2009 National Review article. But overall, as Williamson pointed out then, this received little media attention.
Finding data necessary to examine how the Recourse Rule could have contributed to the crisis remains the biggest challenge. Isil Erel, Taylor Nadauld and René Stulz, who also discuss the potential role of the rule in their paper “Why Did Holdings of Highly Rated Securitization Tranches Differ So Much across Banks?” proposed one way to back out estimates of holdings of some of the securities that featured prominently during the last crisis. Although Erel, Nadauld, and Stulz offer no conclusive evidence that the rule might have factored into the crisis, they argue that their results do not rule out that possibility, which is where my recent working paper fits in.
Before I get to the paper, I’ll discuss key concepts such as rule, regulatory capital, the securitization of assets and recourse, and why “structured finance” collateralized debt obligations (CDOs) lay at the heart of the crisis.
Rule
“Rule” refers to a final rule-making or regulation. The Recourse Rule was jointly finalized by the three federal-level bank regulators, the Federal Reserve, the Federal Deposit Insurance Corporation (FDIC) and the Office of the Comptroller of the Currency (OCC), and applied to financial holding companies, the parent corporations of commercial banks. I’ll explain how in more detail shortly, but for now the rule lowered regulatory capital requirements for certain securitized assets.
Definition: Regulatory Capital and Determining Insolvency
I have discussed bank equity capital — the amount by which a commercial bank’s or financial holding company’s assets exceed its liabilities — elsewhere. Regulatory capital concerns the minimum amount of equity or net worth that regulators think a bank or its holding company can operate with (though keep in mind that capital can also include long-term bonds, such as Total Loss Absorbing Capacity, which I’ll ignore here). Regulation concerning bank or bank holding company capital is summarized in the Code of Federal Regulations, Title 12, Chapter I parts 3 and 167 for nationally chartered banks regulated by the OCC, Chapter II, Subchapter A part 217 for bank holding companies or Federal Reserve member state chartered banks regulated by the Federal Reserve, and Chapter III, Subchapter B parts 324 and 325 for non-Federal Reserve member state banks regulated by the FDIC.
Ideally, the value of the entity’s assets exceeds the value of its liabilities, such that the equity is positive; if that’s not the case, it’s insolvent. But how do you measure the value of assets, liabilities and equity?
Liabilities may be the easiest to measure: you can start with the face value of the entity’s debts, including deposits and other short-term debt, which will change frequently. Assets and equity can be measured at market value using market prices, which also change frequently, or by accountants using the concept of “book value”, as current regulations require.
Book value estimates of assets and equity likely differ from market determined values. That’s important to keep in mind because while the market may believe that the value of a bank’s (or its holding company’s) assets and equity changes over time, the accountants’ estimates of the value of assets and equity may change little, as accountants tend to produce smoother estimates than market valuations.
This distinction creates some uncertainty about when an insolvency occurs. For instance, accountants might determine that the book value of a bank’s (or its holding company’s) assets exceed the face value of liabilities, which means it’s technically solvent. However, if investors believe that because of poor investment decisions the market value of the bank’s (or its holding company’s) assets lies below the face value of the liabilities, they’ll think it’s insolvent.
This uncertainty matters because equity investors know they are on the hook to take first losses and are last in line to receive any payment should an insolvent entity’s assets be liquidated. Less funding from equity investors means more funding from investors who expect repayment, like depositors and the bond holders, in case of an insolvency. Of course, the depositors in principle have additional coverage from deposit insurance, and in principle, they should have less concern about bank insolvency — I’ll return to this last point later.
Debate continues among economics and finance academics, and between the banking industry and bank regulators, over how much funding banks or their holding companies should get through equity capital. Some academics and regulators think more capital means a lower likelihood of a system-wide banking crisis (because banks can withstand a larger loss when they invest poorly). However, bankers and their lobbyists complain that equity is an “expensive” funding source, and would prefer to fund their activities with less equity. Anat Admati and Martin Hellwig and more recently John Cochrane, have addressed this fallacy.
Securitization and Recourse
The process of securitizing assets involves buying lots of illiquid assets, such as loans or even other securitized assets, pooling them together in a trust, and then redistributing the underlying assets’ payment flows to investors by selling them a tranche (French for “slice”), a type of bond. Those created by federally controlled agencies, such as Freddie Mac and Fannie Mae, are called “agency” securitizations, while those created by banks are called “private label” securitizations. “Recourse” arises when a bank securitizes assets, and retains some of the credit risk, often known as “skin in the game.”
Specifically, the bank exposes itself to the securitized assets’ risk of default by retaining a part of the entire deal, while selling the remaining tranches to other investors. The other investors might include insurance companies, pension funds, or even other commercial and investment banks not involved in the securitization deal. One reason why other commercial and investment banks might buy tranches is that they may be very familiar with how they’re created, which allows them to more easily assess the underlying risks. The Erel, Nadauld and Stulz study that I mentioned above calls this willingness by securitizing commercial and investment banks to hold securitization tranches the “securitization byproduct” effect.
Mortgage backed securities (MBS), a major class of securitizations, have mortgages as the underlying assets. Whether they’re agency or private label, MBS tranches make up a significant portion of the securitization market. Based on data for mortgage related securities available from the Securities Industry and Financial Markets Association (SIFMA) website, the total volume outstanding for agency MBS tranches in 2016 equaled about $4.76 trillion, up from $3.43 trillion in 2006, and from $2.24 trillion in 2001. The total volume outstanding for private label MBS tranches in 2016 equaled about $1.38 trillion, down from $3.3 trillion in 2006, but up from $1.09 trillion in 2001.
A related class of securities, called private label asset backed securities (ABS), have car notes, credit card debt or even home equity lines as the underlying assets. Based on data for ABS available from SIFMA, in 2016, about $1.39 trillion in ABS tranches were outstanding, down from $1.65 trillion in 2006, but up from $817 billion in 2001.
CDO tranches, a particular kind of private label ABS, have received lots of attention since the crisis, and the Inside Job and The Big Short do offer good intuitive introductions. From the data for ABS, in 2016 there were about $668 billion in CDO tranches outstanding, down from $815 billion in 2006, but up from $221 billion in 2001. A particularly relevant subset of all CDOs here are so-called “structured finance” CDOs, which featured prominently during the last crisis. In 2016, there were $119 billion in structured finance CDO tranches outstanding, down from $437 billion in 2006, but up from only $25.6 billion in 2001.
These figures indicate that the private label market has not recovered from the crisis. Also, in relative terms the MBS market was much larger than the CDO market, where much of the damage occurred during the crisis.
“Structured Finance” CDOs and the Crisis
It remains an ongoing challenge to trace out the effects of CDOs, and in particular the “structured finance” CDOs on the financial system during the last crisis. Given the lack of data availability, a Federal Reserve Bank of Philadelphia study by Larry Cordell, Yilin Huang, and Meredith Williams set out to reconstruct the entire universe of “structured finance” CDOs, from a database created by the notorious firm Intex, described by Michael Osinski in a 2009 article. The study doesn’t identify who held them, but it does show just how damaging they could be to any financial intermediary that did hold them. Out of 727 deals created between 1999 and 2007, which issued a total of $641 billion in tranches, the authors estimate that $420 billion (65%) were written off as a loss. It wasn’t just the fact that they were private label securities, since a more recent working paper by Juan Ospina and Harald Uhlig shows that losses on most private label MBS were not that large (cumulative under 6% through 2013).
Having introduced the basics of bank capital regulation and securitization, I can now discuss how the Recourse Rule could have affected banks, first by creating incentives to originate (and also hold) highly rated tranches that subsequently could have affected bank solvency.
So What Did the Recourse Rule Do?
The Recourse Rule amended Basel I capital adequacy guidelines, which were unveiled in 1988. U.S. banks had to adhere to the guidelines by 1992. The Federal Register notice indicates that the Recourse Rule was introduced in 1994, but not finalized until 2001. As Jeffrey Friedman and Wladimir Kraus point out, the Recourse Rule anticipated changes to the capital treatment of private label, highly rated securitization tranches that would subsequently appear in Basel II guidelines, which were unveiled in 2004. This fact often gets overlooked in debates concerning whether capital requirements could have contributed to the crisis.
To see how, under Basel I guidelines, if a bank held private label MBS, ABS or CDO tranches, it had to have at least 4% or sometimes even 8% capital (about half of it being equity) to fund those asset holdings. In contrast, agency MBS tranches only required 1.6% capital (about half of it being equity). To illustrate, if a bank held $100 million in private label tranches it would have to fund those holdings with either $4 or $8 million in capital, but if it were an agency tranche it would have to fund those with $1.6 million or less. While the agency and private-label tranches were similar in nature, the regulations treated the private label tranches as if they were riskier, which after the fact, seems true.
Both the Recourse Rule and Basel II lowered capital requirements for bank holdings of AAA- or AA-rated tranche holdings from 4% or 8% to 1.6%, effectively equalizing the capital treatment of agency tranches, which tended to get AAA ratings, with the highest rated, private label tranches. A-rated tranches had capital requirements equal to 4% (lower rated tranches had higher capital requirements). All told, with lower capital requirements, a bank may operate with more leverage. If it does, it’s now riskier, because with more debt (less equity) financing, should the bank default, creditor losses will be larger (equity losses will be smaller).
The Recourse Rule applied only to commercial bank holding companies. For the rule to play a role in the last crisis, 1) it would have to give commercial bank holding companies incentives to hold more of these assets, and 2) greater holdings would have to expose the entity to the risk of insolvency, as when the market value of its assets falls below the face value of its liabilities.
No data on bank holdings of “structured finance” CDOs exist (aside from a small amount reported by a few banks in their trading assets between March 2008 and March 2009). All we can do is estimate the amount of private label, highly rated tranche holdings, and the study by Erel, Nadauld and Stulz I previously mentioned showed just how to do that. Importantly, they show that bank holdings of private label, highly rated tranches behaved much like alternative measures they constructed that included estimates of CDO holdings. So even though the measure of estimated holdings does not include CDO holdings, the measure seems to behave as if it did.
What Did I Find?
I find that securitizing banks on average did indeed increase their estimated holdings of the private label, highly rated tranches relative to total assets, just as Erel, Nadauld and Stulz did. Moreover, securitizing banks with at least $50 billion in total assets on average were the ones that increased their holdings the most. On average, they increased their holdings from about 1% of total assets in 2001, just before the rule was finalized, to about 6% of total assets on the eve of the crisis. More importantly, I also find evidence that holdings for the largest securitizing banks on average equaled or exceeded their regulatory capital buffer on the eve of the crisis.
To see why that matters, consider Citigroup. Erel, Nadauld and Stulz show that in 2006 Citigroup had over 10% of its portfolio allocated to highly-rated, private-label tranches of mortgage- and asset-backed securities as well as CDOs. Citigroup also had 6% equity capital. It would take a write-down of 60% of that 10% for its equity capital to be wiped out. While that may sound extreme, the Cordell, Huang and Williams study I mentioned earlier estimated that average write-downs for structured finance CDOs equaled 65% during their sample (they were even higher for tranches originated just before the crisis).
Supporting the view that private label, highly rated tranches could have adversely affected bank solvency, I also find that banks with greater holdings of those assets on average experienced increases in default risk after 2008, but not before then. In contrast, other factors were not associated with a greater vulnerability to distress. For instance, you often hear claims that trading assets caused the crisis, yet I find an economically and statistically weaker relationship between greater trading asset holdings and higher bank default risk. Also, I find short-term funding was not associated with higher bank default risk, which suggests that the last crisis was not like the stylized textbook model of a commercial bank run (although I’ll return to the “crisis was a run” argument later).
Taken together, these findings ultimately point to the importance of understanding the demand for the securities that lay at the heart of the crisis, and also that it helps to view the crisis as a solvency, rather than a liquidity, crisis.
How Capital Requirements Relate to Demand Side Explanations of the Crisis
If changes in capital requirements following the Recourse Rule made holding the securities that subsequently exposed commercial banks to insolvency risk during the crisis more attractive prior to the crisis, then this ultimately offers a demand side explanation of the crisis. Frank Partnoy emphasized the demand side in the Afterword, when his book F.I.A.S.C.O. was re-released in 2009. He wrote concerning the growth in mezzanine (backed by subprime home equity lines of credit) and synthetic CDO originations, which lay at the heart of the crisis:
“When mortgage lenders such as New Century Financial Corporation and Countrywide Financial saw the insatiable demand for risky loans, they began making too-good-to-be-true loan offers to anyone they could find. Many people have criticized these lenders for unscrupulous practices. Others have criticized borrowers for taking loans they couldn’t possibly repay. Much of that criticism is fair, but it ignores the big picture. The driving force behind the explosion of subprime mortgage lending in the U.S. was neither lenders nor borrowers. It was the arrangers of CDOs…
It was hard to believe, but the bankers were running low on risky assets. After low-rated bonds and loans had been pooled once, they were gone, and therefore no longer available to be pooled again.
It was truly incredible, but there simply wasn’t enough crap on Wall Street.
This is where derivatives entered the picture. The major banks recently had begun trading credit default swaps…
Wall Street saw they could use credit default swaps to create an infinite amount of crap. They quickly engineered new repackaging transactions, using credit default swaps to clone risky subprime-mortgage-backed investments that, when pooled, generated more sky-high ratings. These new deals were known as “synthetic” CDOs, because they had been created artificially, through derivatives side bets…
Soon, investors around the world were buying complex subprime backed financial instruments: synthetic CDOs, structured investment vehicles, constant-proportion debt obligations, and even something called CDO-squareds (don’t ask). The demand for these derivatives backed investments was a tail wagging a very large dog.” (pp. 265–266)
While the Recourse Rule changed capital requirements for commercial bank holdings of private label CDOs (as well as ABS and MBS), it would not explain what happened elsewhere in the financial system. Still, capital requirements elsewhere in the financial system may have driven demand for holdings of private label CDOs.
For instance, on June 21, 2004, the SEC finalized the so-called “Alternative net capital requirements,” which subjected investment banks to Basel II capital adequacy standards. Claims that supposedly debunk the myth that the SEC rule contributed to the crisis exist. While they may be correct that the Alternative net capital requirements did not affect leverage the way some initially claimed it did, they also miss the fact that the rule lowered the capital requirements for the highest rated, private label tranches to 1.6%.
Recall, Friedman and Kraus showed in their book that the Recourse Rule and Basel II made the same changes to capital requirements on private label, highly rated tranches. It’s just that the Recourse Rule applied to U.S. commercial banks since 2001, while thanks to the alternative net capital requirements, Basel II applied to U.S. investment banks since 2004.
It remains a challenge to confirm this hypothesis. That said, Taylor Nadauld and Shane Sherlund in a 2013 article show that the top five investment banks (Bear Stearns, Lehman Brothers, Merrill Lynch, Goldman Sachs and Morgan Stanley) had even higher mortgage originations than non-investment banks, after the rule change. They also point out that investment banks funding with repurchasing agreements tended to retain tranches. Since Basel II lowered capital requirements for the highest rated tranches, it’s entirely possible that the investment banks operating under the new rule could have increased their holdings just like the large, securitizing commercial banks seem to have done following the Recourse Rule.
Also, in a related study Craig Merrill, Taylor Nadauld and Philip Strahan examine insurance company holdings of the private label securitization tranches, as insurance companies report much more detail about their securities holdings than do commercial and investment banks. They show that insurance companies, reacting to capital rules on general accounts (as well as a lower interest rate environment in the early 2000s, which had adversely affected their portfolios), increased their demand for private-label securities, which offered high ratings and higher yields than other similarly rated securities. And keep in mind that insurance company capital requirements differ from bank capital requirements.
By examining what happened on the demand side, it’s easy to view the last crisis as a solvency crisis rather than a liquidity crisis.
Solvency Crisis or Liquidity Crisis?
By solvency crisis, I mean that for a large segment of the commercial banking system (a small number of large banks, a large number of small banks, or some combination of the two), banks observe the market value of their assets falling below the face value of their short-term liabilities. Such a view would suggest that the crisis was more about the asset side of bank balance sheets than the liability side. By liquidity crisis, I mean a run. In a run, depositors decide they no long want to fund an intermediary.
In saying so, I do not deny Gary Gorton and Andrew Metrick’s view that the crisis included a run on repurchasing agreements used to fund securitizations (which they suggest reflected uncertainty about investment bank solvency). Gorton and Metrick focus more on the investment banks. Given my focus on commercial banks, I merely suggest that there was also a commercial bank solvency crisis.
Why should we care? If 2008 was a liquidity crisis, then the Fed would have had to take the measures it took to prevent the crisis from getting worse than it was. If on the other hand, 2008 was a solvency crisis, then perhaps some of the Fed’s activities might have violated the spirit of section 13(3) of the Federal Reserve Act, which allows the Fed to lend to solvent institutions in emergencies. I am not suggesting that some kind of conspiracy took place — during a crisis, it may be really difficult to distinguish between insolvency and illiquidity.
But let’s see why the Fed might have intervened. For one, the FDIC’s Deposit Insurance Fund (DIF) might not have been able to cover the insolvency of all the largest securitizing banks at once. The FDIC annual reports from 2007, 2008, and 2009, indicate that in 2007 and 2008, the DIF, which insures deposits, had about $50 billion in total assets, and had positive net worth. However, by 2009, the DIF assets had increased to over $110 billion, but the DIF had a net worth of -$20 billion. Considering that write-downs as of February 2009 totaled $172.4 billion for Bank of America, Citigroup, JPMorgan Chase and Wells Fargo & Company alone, that suggests that the DIF was insufficiently funded.
I can’t imagine many that many people would be happy to discover that their guaranteed deposits were backed by an insolvent DIF. Accordingly, I can also imagine the Federal Reserve stepping in to ensure that the FDIC would not have to resolve the largest securitizing banking entities that were exposed to CDO losses.
John Cochrane suggested in a recent blog post that if banks had 20% equity funding, we wouldn’t have had a crisis. Using the Citigroup example I mentioned above, it’s clear that with 10% of its portfolio allocated to highly rated tranches and 20% equity capital, even 100% write-downs wouldn’t have wiped out Citigroup’s capital. While bankers may not like such high capital levels, for the taxpayer, there’s lots to like about this proposal.