IV. Should the Collateralized Money Market be Stabilized or Euthanized?
This is Part IV of a lengthy critique of Bagehot was a Shadow Banker by Perry Mehrling, Zoltan Pozsar, James Sweeney, and Daniel Neilson. The authors of Bagehot was a Shadow Banker equate the shadow banking system with what they call the “market-based credit system” (at 2). To be clear, the authors focus specifically on a market-based short-term credit system or on money markets. In this Part I ask what does it mean to call a credit system “market-based” and whether such a system exists, then I discuss the consequences of moving from an unsecured money market to a collateralized money market, and finally I evaluate the likely effectiveness of the solution proposed by the authors of Bagehot was a Shadow Banker in stabilizing the collateralized money market.
A. Does a “market-based” credit system exist?
The “market-based” credit system is often contrasted with the “bank-based” credit system to distinguish environments where firms raise funds by issuing securities on markets from those where firms raise funds by borrowing from banks. This distinction is clear when we focus on long-term capital markets, such as bond markets where established companies can and do raise money on a regular basis.
When it comes to money markets, however, the line between market-based and bank-based systems cannot be clearly drawn, because the so-called market-based systems rely heavily on guarantees provided by the banking system. The principle instruments used to borrow short-term on markets are commercial paper (the short-term version of a bond) and repurchase agreements, in which the borrower simultaneously sells an asset and agrees to repurchase it at a fixed future date and price, effectively creating a collateralized loan. As was discussed in Part II, as a rule non-financial companies can only borrow on these markets if they have liquidity support from a bank: Because commercial paper typically needs to be rolled over at maturity, almost all non-financial issuers including asset-backed commercial paper conduits must have a liquidity facility, or a bank’s promise to retire the paper if the issuer is unable to do so. Similarly, the most important repo market is backstopped by guarantees provided by the tri-party clearing banks, which bear the credit risk of the borrowers during the day. In short, in the money markets the “market-based” credit system might as well be called the “bank-guaranteed” credit system.
Furthermore, because so-called “market-based” money market instruments require bank guarantees and those guarantees are most likely to be called upon in a crisis, the “market-based credit system” insulates banks from the credit risk of borrowers in normal times, but not in crises. Because such instruments are always designed in normal times when the likelihood that the bank will be obliged to make good on its contingent liability is deemed minimal, these instruments create a form of bank risk that typically carries lower capital requirements than alternatives. For these reasons, the “market-based” short-term credit system is best viewed as a form of bank lending that is designed to minimize capital requirements, and it should be categorized as lying within the “bank-based” credit system.
Another sense in which money market instruments are only nominally “market-based” is that these assets do not trade on secondary markets. Commercial paper is placed and almost never resold before maturity. Repo obligations, similarly, are not traded actively, but held until maturity. By contrast, in 19th c. London, the archetype of a traditional bank-based credit system, there was an active secondary market in the bills that were the primary tool by which central bank policy was implemented.
Repurchase agreements and asset-backed commercial paper (but not commercial paper more generally or 19th c. bills) are both collateralized forms of “market-based” money market instruments. Such collateralized money market instruments can also be considered “market-based” in the sense that they are subject to market risk; that is, if the value of the collateral falls below the value of the debt, the borrower will have to post additional collateral. Market risk is a key concern of the authors of Bagehot was a Shadow Banker, and it is possible that when the authors use the term “market-based credit system” they mean only to refer to collateralized money market instruments that are subject to market risk; if this is the case, in my view, the term, collateralized money market is more clear.
It is not clear whether the authors of Bagehot was a Shadow Banker imagine a world in which so-called “market-based” money market instruments exist without the support of bank guarantees. If so, it should be noted that their solution is tailored to market risk, whereas the bank guarantees are needed to address funding risk. That is, even if there were no market risk and the collateral’s value could not fall, the possibility that the borrower would find itself illiquid and unable to retire or to roll over the debt (for reasons specific to the borrower or to the money market, not to the collateral) would almost certainly mean that funding guarantees were still necessary to support the market.
To summarize, because the finance of longer-term assets requires that these short-term instruments be rolled over, funding risk is always a concern in the so-called “market-based” short-term credit system, and this almost certainly means that this “market-based” credit system cannot exist except when it is backstopped by the banking system. Thus, what is commonly known as the “market based” short-term credit system – including most of the shadow banking system – should properly be understood to lie within the “bank-based” credit system.
The rest of this Part will assume that the authors use the term “market-based” credit system only for the purpose of focusing attention on the market risk inherent in the collateralized portions of the money market. Thus, I will focus on what I will call the collateralized money market.
B. Is transforming credit risk into market risk a good idea?
In the collateralized money market, borrowing is collateralized in order to provide additional security to the lender, and market risk substitutes for a portion of the credit risk that lenders traditionally face. From the borrowers’ point of view less-creditworthy borrowers have access to credit, but this access comes at the expense of raising the costs of borrowing for borrowers, who now have to worry not only about having the resources to pay the debt at maturity, but also about maintaining sufficient collateral to back the loan throughout the life of the loan. Thus, borrowers will elect to use this system if they are not sufficiently creditworthy to borrow unsecured or they have easy access to collateral.
The authors of Bagehot was a Shadow Banker are correct to emphasize that the most important difference between 19th c. British money markets and modern money markets is the role that market risk plays in modern markets. 19th c. money markets focused on managing credit risk exclusively – so much so that Thornton viewed the “science” of credit as the great innovation of the British banking system and a driving force of the economy’s growth. The careful management of credit risk in the 19th c. is illustrated both by the fact that every bill discounted by the Bank of England was guaranteed to be paid in full by at least three different parties, the issuer, the acceptor, and the discounter, and by the fact that the Bank had negligible losses on its discount portfolio, even after a crisis.
The authors of Bagehot was a Shadow Banker treat the growth of market risk and collateralized money market assets as a demand-side phenomenon (at 12). They reference Pozsar’s work which emphasizes that the largest lenders in modern markets, asset managers such as mutual and pension funds, are reluctant to extend unsecured credit to the banks in the form of uninsured deposits and prefer to lend via repos or asset backed commercial paper. (Note that this point should not be overemphasized, because banks are able to raise significant funds, unsecured, by issuing commercial paper.) The second demand-side explanation for the growth of the collateralized money market is the modern asset management practice of using derivatives markets to take on the risks of investing while holding invested funds in monetary assets. Presumably, supply-side effects may also have played a role in the growth of this market, as investment banks found the market both convenient for financing inventories, and possessed of the useful property that in normal times the market was not very sensitive to changes in the credit quality of the borrower.
The authors view this transformation of the money market from one in which credit risk was carefully managed to one where market risk substitutes for credit risk as a benign, if not beneficial, development. There are, however, many concerns that this development should raise.
First are the implications the movement towards collateralized short-term lending may have for the credit quality of our financial institutions. It is possible that this movement reflects declining credit quality among financial institutions that makes it difficult for them to borrow on an unsecured basis. The fact that this change took place alongside the transformation of the investment banking industry from unlimited liability partnerships to limited liability corporations may be an indicator that declining credit quality is an important driving force behind this change. Another potential concern is that the movement to collateralized short-term lending aggravates declining credit quality among financial institutions. Research has shown that repo lending terms are principally determined by the quality of the collateral posted and do not tend to reflect incremental changes in the credit quality of the borrower. For this reason, it is possible that the movement towards collateralized borrowing makes borrowers less concerned about whether or not they are viewed as good quality borrowers.
Second, as was discussed in Part I, early monetary theorists such as Henry Thornton believed that banking contributed to economic growth because it allowed the money supply to expand based on the needs of the economy and that the “science” of credit facilitated this expansion. Collateralized money markets, however, substitute market risk for credit risk, and as techniques for issuing unsecured money market instruments fall into disuse may have the effect of limiting the use of unsecured credit more than the principles of managing credit risk would require. If unsecured credit falls into disuse, the growth of the money supply, and arguably of the economy itself, may be limited by a deficit of collateral. In short, it is not clear that collateralized money market instruments can play the same role in expanding the money supply and in economic growth as that played by unsecured money market instruments.
Finally, repo markets are likely to be even less stable sources of funding for financial institutions than deposits, and thus even more prone to fire sales. Because the realization of market risk in collateralized lending markets can force immediate deleveraging, the availability of funding on repo markets can disappear just as quickly as deposits can be withdrawn – and even more quickly than credit based on unsecured term instruments which can only be withdrawn as the instruments mature. In addition, however, the fact that the collateral on repo markets is funded on a leveraged basis means that small changes in the market prices of assets can result in the need to sell off a large fraction of assets. Because of leverage, repo markets are probably less stable than deposit-based funding.
An example (drawn from a Fitch Ratings report) will make the instability inherent in repo market finance more clear. Consider a borrower with a $5 million equity stake, which uses repo markets to finance the purchase of a $105 million portfolio of corporate bonds on which the lender imposes a 5% haircut, so that $1 can be borrowed for every $1.05 in collateral repo’d. The borrower will therefore have a leverage ratio of 21 to 1. A 2% decline in the value of the portfolio would reduce the total portfolio value to $102.9 million, reducing the equity in the portfolio to $2.9 million. If we assume that the borrower has no additional equity to contribute, the borrower can now only finance a $60.9 million portfolio at a 5% haircut. In short, because of the leverage inherent in using repo markets to finance assets, a 2% drop in portfolio value can force a sale of 42% of the assets held. Note that this example doesn’t take into account the possibility that the lender increases the haircut on the repo, which would mean that even more of the assets had to be sold. In short, once a borrower has maximized the use of leverage on repo markets – whether the borrower does this intentionally in order to “maximize” returns or simply ends up in this situation after the collateral has declined in price – very small declines in price can force the borrower to sell a significant fraction of the assets. If the borrower is a large market participant, such as an investment bank, this is likely to be the first step in a liquidity spiral, where asset sales further reduce the value of the collateral and trigger additional assets sales.
C. Should collateralized money markets be stabilized by protecting leveraged dealers from losses?
The authors of Bagehot was a Shadow Banker recognize that collateralized money market instruments are prone to fire sales and liquidity spirals, but they do not appear to realize that, due to the leverage that is used in repo markets, very small price declines can have very large effects. They argue that a dealer of last resort is all that is needed to stabilize these markets, and that the dealer can do this by putting a price floor on the assets used as collateral. Specifically they write:
just as in Bagehot’s day, the critical infrastructure is an interconnected system of dealers, backstopped by a central bank. Just as in Bagehot’s day, the required backstop may involve commitment to outright purchase of some well‐defined set of prime securities (such as Treasury securities). But it must also involve commitment to accept as collateral a significantly larger set of securities, in order indirectly to put a floor on their price in times of crisis. (at 9).
Two points should be emphasized with respect to this proposal. First, it is important to understand that despite the authors’ focus on the support of asset prices, the key innovation in the “dealer of last resort” proposal is the extension of central bank support to dealers. Secondly, because dealers are very different from banks, the extent of the support provided by the central bank to dealers is likely to be much greater than that provided to banks.
As for the first point, the Federal Reserve has had the ability to accept virtually any security as collateral since 1932 – as long as it was collateral for a loan to a commercial bank. The problem in 2008 was not the nature of the collateral that could be used, but the fact that investment banks didn’t have access to the central bank. Thus, the key innovation of the “dealer of last resort” proposal is not the type of collateral that can be used for a loan, but the fact that dealers – or investment banks – are able to borrow from the central bank using that collateral.
For non-prime assets, the proposal is only that the central bank accept them as collateral, not that it buys them outright. One potential advantage of accepting assets as collateral, rather than buying them outright, is that the central bank may not need to determine their value, but can choose to rely on the valuation of the collateral given by the borrower, because the borrower will be liable for any shortfalls if the collateral is worth less than the loan. In this case, the purpose of accepting an asset as collateral is not to “put a floor” on its price, but to prevent the borrower from being forced to sell it in a fire sale. This is an important distinction to make, because the central bank is not, in fact, intervening in the market pricing mechanism in the way that the phrase “put a floor on the price” implies. Instead, the central bank is making it possible for the borrower to carry the asset at the valuation at which the borrower is willing to buy the asset back in the future. The level of the asset’s market price is affected, but not determined, by this policy.
Under this interpretation of the proposal, the dealer of last resort should not be viewed as supporting the price of assets (as the authors claim in many places), but as supporting the dealer system. While this understanding is not consistent with “putting a floor” on asset prices, it is more consistent with the author’s claim that when the central bank intervenes, the price decline “is not so much halted or reversed as it is contained and allowed to proceed in a more orderly fashion” (at 14). After all, if the goal is to put a floor on the price of the assets, it doesn’t make much sense to claim that these prices will continue to decline in an “orderly fashion.”
Secondly, the function of dealers in supporting market liquidity is very different from the function of banks in honoring deposits and funding guarantees; as a result the nature of a central bank backstop, and indeed the degree of reliance on the central bank is likely to be very different for a dealer of last resort as compared to a lender of last resort. In order to address the problem of liquidity spirals, the authors of Bagehot was a Shadow Banker argue that:
what is clearly needed is some entity that is willing and able to use its own balance sheet to provide the necessary funding. … what we need is a dealer system that offers market liquidity by offering to buy whatever the market is selling. Only in crisis time does the central bank backstop become the market; in normal times, the central bank backstop merely operates to support the market. (at 9).
Because the primary work of a dealer is that of smoothing the market’s movement to a new price, and not of setting a floor on asset prices, it’s far from clear that dealers can ever be expected to the play the role that the authors of Bagehot was a Shadow Banker seem to expect them to play in supporting prices on markets. The traditional constraints on the behavior of a dealer are described by Jack Treynor, the source of the authors’ model of dealer pricing: “the dealer has very limited capital with which to absorb an adverse move in the value of the asset. Furthermore, the dealer’s spread is too modest to compensate him for getting bagged.” Treynor contrasts the role played by a dealer with that of an investor who has the capital to hold positions over a longer term. Thus, a traditional dealer does not “use its own balance sheet to provide the necessary funding” except over very short time horizons. On the other hand, it is true that large-scale proprietary trading and the management of balance sheet exposure to related risks has become a core function of dealers in recent decades. (This is almost certainly related to the role played in the market by limited liability investment banks that have access to vast capital resources the use of which is monitored, not necessarily effectively, by boards of directors.)
Because the traditional function of a dealer is not to support asset prices, however, it seems likely that even dealers engaged in large-scale proprietary trading will let a liquidity spiral run before stepping in to buy in significant quantities – and then they may well allow the price to continue falling as they build up a significant stake in the assets. After all, if sellers want to sell at unreasonably low prices, this will only add to the dealers’ proprietary trading profits in the end.
Indeed, this is what we witnessed in 2008 when Bear Stearns and Lehman were failing: for the most part, the dealers instead of using their balance sheets to support prices on the market sought to avoid being caught holding assets that are falling in value. Thus, one would expect the burden of establishing a price floor for assets to fall heavily on the dealer of last resort or central bank, just as it fell heavily on the Federal Reserve which had to jerry rig facilities such as the Primary Dealer Credit Facility and the Term Securities Lending Facility in order to take on hundreds of billions of dollars of the asset risk of the investment banks in 2008. At the start of October 2008, these two facilities accounted for 60% of the massive expansion of the Federal Reserve’s balance sheet as compared to the previous year. In short, because providing a price floor for assets is not the economic function of a dealer, a central bank that acts as a “dealer” of last resort must be prepared to purchase assets on this scale – and effectively to become the market – in every crisis.
These two points indicate that the value of the dealer of last resort policy is that a troubled dealer can use the assets as collateral to borrow from the central bank, and doesn’t need to sell them at all. The fact that the largest market participants are protected from ever finding themselves forced to sell their assets will undoubtedly be very effective in protecting asset prices from instability due to massive fire sales.
But the proposed policy would also introduces a very troubling asymmetry into our markets. Who has access to central bank’s discount window? Retail investors clearly do not, whereas “dealers,” however they end up being defined, do. The privileged dealers effectively have access to an unlimited balance sheet and can employ leverage without worrying about being forced into a fire sale – and no longer face the traditional constraints that govern dealers’ activities. By contrast, those without this privilege are limited by their capital position in the degree to which they can increase their profits using leverage.
In short, this policy is likely to have the effect of protecting the privileged dealers from losses due to market risk, while other market participants do not receive similar protection. By reducing the costs of leverage to the privileged dealers it is also likely to increase their use of leverage. If the central bank does not monitor the behavior of the privileged dealers vigilantly, it could end up making financial markets more risky, by making the largest financial market participants believe that it is “safe” for them to take on more risk.
These dangers are offset in part by the fact that the dealer of last resort’s actions in a crisis will forestall an immediate collapse and the economic repercussions of such a collapse. And this fact almost certainly justifies the Federal Reserve’s actions in 2008. It is less clear that this fact is enough to justify embracing the “dealer of last resort” proposal as a standing policy.
The reason that the lender of last resort is good policy, whereas the dealer of last resort probably is not good policy, is that banks are different. Their value lies in making possible money markets based on unsecured credit that is extended broadly across the business community and makes modern economic growth possible. The banking system merits the protection of a lender of last resort, because it provides such broad benefits to the community at large, and there is good reason to believe that without a lender of last resort a banking system will collapse entirely.
There is, by contrast, a long history of dealer systems that support trade on financial markets and are not at risk of collapse in the absence of a dealer of last resort. Furthermore, there is little or no evidence that collateralized money market instruments play the same role as uncollateralized money market instruments in economic growth, and thus little or no evidence that collateralized money markets are necessary to the community at large. In fact, the growth of collateralized money markets may undermine traditional unsecured money markets, where a financial institution’s ability to borrow depends on its credit quality, and thereby undermine the market forces that promote high credit quality in the financial industry. For this reason, the collateralized money markets may be destabilizing the financial industry. While the temporary support of these markets in 2008 was well justified, much more evidence of the value of collateralized money markets to the process of economic growth needs to be presented before dealers and investment banks are given privileges similar to those of commercial banks.
 See, e.g., Michael Woodford, Financial Intermediation and Macroeconomic Analysis, 24 J. Econ. Perspectives 21, 21 (2010).
 The collateral posted against a repo can often be rehypothecated, but this is very different from the resale of the debtor’s obligation that takes place in secondary markets.
 Vincent Bignon, Marc Flandreau, & Stefano Ugolini, Bagehot for beginners: the making of lender-of-last-resort operations in the mid-nineteenth century, 65 Econ. Hist. Rev. 580, 602 (2012).
 Zoltan Pozsar, Institutional Cash Pools and the Triffin Dilemma of the U.S. Banking System 10-11 (IMF Working Paper No. 11/190, Aug. 2011).
 Fitch Ratings, Repo Emerges from the “Shadow” 3 (Feb. 3, 2012).
 Critics complained that “any cat and dog” could be used as collateral at the Fed, after the Federal Reserve Act was amended in 1932. David McKinley, The Discount Rate and Rediscount Policy 97, quoted in David Small and James Clause, The Scope of Monetary Policy Actions Authorized Under the Federal Reserve Act 10 n.22 (FEDS Research Paper 2004-40, 2004).
 Jack Treynor, The Economics of the Dealer Function, 43 Fin. Analysts J. 27, 27 (1987).
 Federal Reserve Board of Governors, H.4.1 Release: Factors affecting reserve balances, Oct. 2, 2008.
 For the role that capital constraints typically play in risk-taking, see Andrei Schleifer & Robert Vishny, The Limits of Arbitrage, 52 J. Fin. 35 (1997).