Shadow Banking and the Financial Crisis

Not surprisingly, many of the interesting papers that I have been reading in recent months discuss the financial crisis. I have learned much from the work of Gary Gorton, professor of finance at the Yale School of Management. He argues that a vast “shadow” banking system has developed over the last 40 years outside the traditional regulated banking system of governmentally chartered institutions that accept deposits and make loans. As of 2007, two‐​thirds of U.S. lending was in the shadow banking system and only one‐​third in the traditional banking system.

In the shadow banking system (also called the “overnight repurchase” or “repo” market), corporations, governments, and investors “deposited” hundreds of millions of dollars overnight in the equivalent of extremely large money market accounts, invested in securitized loans for cars, houses, and consumer credit. When this shadow banking market operated normally, the trade of cash for securitized assets was automatically renewed every day.

But in the fall of 2008, an old‐​fashioned bank run occurred in the repo market. A bank run is a systematic withdrawal of demand deposits from the banking system as a result of news that the bank is in poor financial health. In this case, instead of renewing the trade of cash for securitized loans, the “depositors” in the repo market changed the terms or simply withdrew their money.

Prior to the 2008 crash, depositors in the shadow banking system traded cash for securitized loans of equivalent face value and required the investment bank to set aside very little or no collateral (cash or capital that is not loaned out). That is, $1 million in cash was traded for $1 million in securitized loans, because depositors thought of the senior tranches of securitized loans as “super safe.” By November 2008, the terms had changed drastically because depositors feared securitized loan losses. That same $1 million in cash was traded for $1 million in securitized loans, but with an additional requirement that $500,000 be set aside in cash or capital. These so‐​called “haircuts” severely reduced the lending ability of the shadow banking system, as did the behavior of other investors who withdrew their money entirely and invested it instead in U.S. treasuries. The result was a severe contraction in lending, which exacerbated the contraction in the real economy that was already occurring because of the collapse of new housing construction and housing‐​backed consumption.

All in all, Sowell offers a number of intriguing ideas in the book, but he leaves too many of them only partly developed and defended. One final example: Sowell claims that those of us who favor free markets are more empirically inclined and have less ego invested in our views than those who advocate central planning and heavy government intervention generally. That is my impression too, and it is one of the most interesting and important ideas in his book. Unfortunately, he gives no evidence for that comparison. The book would have been a great forum in which to do so, but on this idea and too many others, he leaves his readers unsatisfied.

Gorton’s work presents a challenge for libertarians: Unregulated banking is subject to bank runs, but government‐​mandated deposit insurance and regulated banking eliminated them. But large corporate checking accounts were uninsured and received no interest within the post–New Deal regulated banking system. Markets thus responded with the shadow banking system. This provided income for large corporate and investor demand deposits that funded loans for consumers at lower cost than the regulated banking system. For more than 30 years, this market response appeared to be “good” arbitrage around the expensive regulated banking system. But then a large run occurred in the shadow banking system and the Federal Reserve responded by becoming a lender of last resort to all the markets abandoned by “depositors” in the shadow banking system.

Insurance or Narrow Banking? Two polar opposite policy responses have been proposed to avoid future “shadow bank runs.” The first proposed response, called “narrow banking,” has been advocated by Larry Kotlikoff, Oz Shy, and Amar Bhidé, among others. Narrow banking is an attempt to carefully demarcate the difference between the payments system (checking accounts, passbook savings, and other demand deposits) and all other investment. Money in the payments system would be fully backed by cash or treasuries. All other investment would be at‐​risk. The second proposed response is the extension of the current safety net to all demand deposits, along with restrictions on financial activities. That would bring the shadow banking system back within the regulated system, extending deposit insurance to overnight repo and money market mutual funds.

Both of these “solutions” have logical flaws. The problem with narrow banking is the time inconsistency of government policy. Fannie Mae and Freddie Mac were not legally backed by the government, nor were money market funds, nor overnight repo. But when the system came under severe financial stress, the government changed the policy. Likewise, if future stress were to hit at‐​risk deposits, placing the broader economy in jeopardy, the White House and Congress would surely ride to the rescue again — and investors know that. So what would stop investors from taking on higher‐​than‐​appropriate risk if they know Washington stands ready with bailout money?

The second problem with narrow banking is that it would require the legal suppression of what most people now call banking, i.e., financial intermediation in which short‐​term deposits are transformed into longer‐​term investment. Banks arise naturally in a free society, and the narrow banking regime would require the use of the power of the state to suppress the transformation of demand deposits into investment

Morgan Ricks, a U.S. Treasury official, has written an important paper that takes Gorton seriously, recognizes the time inconsistency involved in the narrow banking solution (the state cannot precommit not to help at‐​risk investment), and advocates the second solution: deposit insurance must be provided to all demand deposits and lending entities, subject to capital requirements and activity restrictions. He agrees with Gorton that banks (and shadow banks) strive to create informationally insensitive debt — financial obligations that are unlikely to be upset by news (good or bad). This is in contrast to equity and corporate bonds, which are informationally sensitive — stock and bond prices frequently change as news comes in. Informationally sensitive financial instruments trade in the secondary market, and market actors expend resources to inform their decisions about buying and selling stocks and bonds. In contrast, loans to businesses and individuals from traditional banks and securitized loans in the shadow banking system do not trade in secondary markets, and no one outside the original lender attempts to gather information about loan repayment probability.

When circumstances conspire to make loans in the traditional banking system or securitized loans in the shadow banking system informationally sensitive, depositors and repo investors respond by making a run on the bank. The threat of a run could impose market discipline, incentivizing banks and shadow banks to do a good job in creating informationally insensitive debt, and giving opportunities to entrants to profit from picking up the pieces if a bank fails. But runs would have to be allowed by government in order to enforce that discipline and provide those opportunities. Since the Depression, governments have not been willing to run that experiment. So we do not know whether the Federal Reserve’s creation of all its lending facilities, which replaced repo and supported securitized asset lending during the crisis, occurred more quickly and effectively than private facilities would have. And going forward, government cannot credibly commit to allow the experiment to occur.

Ricks modifies (in my opinion, correctly) the Gorton view that the shadow banking system developed simply to meet the exogenous needs of large demand depositors who wanted to arbitrage around the expensive traditional banking system. The Gorton view may have been true initially, but the increasing flow of funds from the traditional to the shadow banking system was the result of an endogenous feedback loop. The greater the use of short‐​term repo, the more damaging a panic of those creditors would be and the more likely that government would provide a bailout if a panic were to occur. Thus the likelihood of a bailout is an increasing function of the overall quantity of liabilities that are potentially subject to panic. Put simply, shadow banking creates more shadow banking and increases the likelihood of a run and subsequent federal intervention, so one might as well return to the old system and have explicit government assistance ex ante (i.e., deposit insurance) accompanied by rules that constrain lending behavior. The difficulty, of course, with a regulated deposit insurance world for all demand deposits is the mismatch between the potential federal liability of trillions of dollars (the total amount of money in the current banking and shadow banking systems) and the actual federal resources available, as well as the need to suppress arbitrage around the regulated system.

Bankruptcy An important modification to the Gorton‐​Ricks thesis comes from Mark Roe, professor of law at Harvard. He argues that the use of overnight repos and derivatives, the heart of the shadow banking system, was endogenous, as Ricks argues, but not simply because of the inability of the political system to precommit not to bail out short‐​term creditors. Instead, important changes in the bankruptcy code that Congress enacted to prevent systemic risk actually encouraged it. The changes gave priority to repos and derivatives in the bankruptcy process, which encouraged the use of such instruments and led to a lack of concern by repo owners about the quality of assets swapped for overnight cash. That is, the informational insensitivity of the repurchase agreements described by Gorton and Ricks was the result of the privileged position held by such investments at the head of the bankruptcy line, which insured full reimbursement. Thus Roe argues that the ever‐​increasing reliance on short‐​term repurchase agreements was the result of bankruptcy policy changes. And the bankruptcy priority made runs the rational response of participants in the overnight repurchase market.

Roe argues that all creditors should be treated identically in bankruptcy. This would increase market monitoring by the suppliers of deposits on investment and decrease the use of short‐​term funds to back longer‐​term investment.

House Prices At the root of the financial crisis was a steep reduction in the value of homes in many housing markets and the resulting defaults on mortgages whose face value greatly exceeds the reduced market value of the homes on which they are written. Some have argued that the steep rise and fall in prices is sufficient to conclude that housing supply constraints arising from zoning restrictions are the main cause of the housing bubble, and that the current focus of the media and political system on easy mortgage availability ignores the real regulatory problem.

Thomas Davidoff, an economist at the University of British Columbia business school, has written a paper that attempts to determine the importance of demand increases and supply constraints in the recent housing bubble. He does so by comparing house price behavior in what he calls “coastal” metropolitan areas (those areas with low supply growth and large price appreciation in the 1980s, from Boston to New York and California) during the 2004–2007 bubble and the 2007–2009 crash with a “comparison” group of metropolitan areas (those areas with large supply growth and low price appreciation in the 1980s). The appreciation from 2004 to 2007 and depreciation from 2007 to 2009 were not significantly different in coastal versus comparison groups. Thus, metropolitan areas with a history of low price and large supply appreciation in the past had a boom‐​and‐​bust pattern in the 2000s that did not differ from metropolitan areas that had a history of the opposite. This lack of difference in pricing behavior casts doubt on the role that supply constraints (natural or regulatory) played in the housing bubble.

It is possible that the supply in the comparison areas changed from elastic to inelastic between the 1980s and 2000s, throwing off Davidoff’s test. He checked for this by calculating the ratio of housing units permitted in each metro area between 1998 and 2007 with the units permitted between 1980 and 1990, and then determined whether the ratios were different in the coastal and comparison cities. There was no significant supply change in the comparison cities, evidence consistent with a demand‐​driven rather than a supply‐​constrained housing bubble.

A recent paper by the New York Federal Reserve corroborates the role that non‐​traditional mortgage instruments played in augmenting housing demand. Across metropolitan areas, the correlation between nonprime loans per 1,000 housing units and 2000–2006 housing price appreciation was 0.64. And across metropolitan areas the correlation between nonprime loans per 1,000 housing units and 2006–2008 housing price depreciation was 0.75. “Why might this correlation hold?” ask the authors. “It is likely that causation runs in both directions — an increase in nonprime lending led to more significant home price appreciation, and more rapid home price appreciation led to a rise in nonprime lending.”

Foreclosure Assistance Many have argued for foreclosure assistance modeled on the New Deal–era Home Owners Loan Corporation (HOLC), which refinanced short‐​term mortgages into long‐​term, low‐​interest, amortizing mortgages. What effects did HOLC have? In theory, such assistance should have increased demand for and probably the supply of owner‐​occupied housing and reduced demand for and maybe the supply of rental housing. But did that in fact happen?

In an important new paper, Price Fishback and colleagues regress 1940 housing data against 1933–1936 per‐​capita HOLC funding. Their initial regression finds increased home ownership as a result of the program. But the addition of control variables for differences that varied across states and time, as well as the fact that HOLC funding was not random (more funding went to troubled housing markets), reduces the effect of HOLC to zero. The effects were positive in counties with less than 50,000 people, but that result arose from the lack of integration between rural and national lending markets in the 1930s. These results suggest great caution in using HOLC as an example of the positive effects that foreclosure assistance can have on housing markets.

Readings

• “Bankruptcy’s Financial Crisis Accelerator: The Derivatives Players’ Priorities in Chapter 11,” by Mark Roe. March 2010. Available at http://​papers​.ssrn​.com/​s​o​l​3​/​p​a​p​e​r​s.cfm ?abstract_id=1567075.
• “Bypassing the Bust: The Stability of Upstate New York’s Housing Markets during the Recession,” by Jaison R. Abel and Richard Dietz. Current Issues (Federal Reserve Bank of New York) Vol. 16, No. 3 (March 2010).
• “In Praise of More Primitive Finance,” by Amar Bhidé. The Economists’ Voice, Vol. 6, No. 3 (February 2009).
• “Limited Purpose Banking — Putting an End to Financial Crises,” by Christophe Chamley and Laurence J. Kotlikoff. Financial Times, January 27, 2009.
• “Restoring the Banking Social Contract,” by Morgan Ricks. March 2010. Available at http://​papers​.ssrn​.com/​s​o​l​3​/​p​a​p​e​r​s.cfm ?abstract_id=1571290.
• “Rethinking the Roles of Banks: A Call for Narrow Banking,” by Oz Shy and Rune Stenbacka. The Economists’ Voice, Vol. 5, No. 2 (February 2008).
• “Securitized Banking and the Run on Repo,” by Gary Gorton and Andre Metrick. January 2009. Available at http://papers.ssrn .com/sol3/papers.cfm?abstract_id=1440752.
• “Slapped in the Face by the Invisible Hand: Banking and the Panic of 2007,” by Gary Gorton. Presented at the Federal Reserve Bank of Atlanta conference on financial markets, May 11–13, 2009. Available at http://​papers​.ssrn​.com/​s​o​l​3​/​p​a​p​e​r​s.cfm ?abstract_id=1401882.
• “Supply Elasticity and the Housing Cycle of the 2000s,” by Thomas Davidoff. March 2010. Available at http://​ssrn​.com/​a​b​s​tract =1562741.
• “The Influence of the Home Owners’ Loan Corporation on Housing Markets during the 1930s,” by Price V. Fishback, Alfonso Flores‐​Lagunes, William Horrace, Shawn E. Kantor, and Jaret Treber. NBER working paper No. 15824, March 2010.