Contagion depends critically on short-term funding, without such funding creditors cannot run. They can, however, refuse to lend more, and this was a major problem in the crisis—institutions that were dependent on short-term funding, like broker-dealers or corporate commercial paper issuers, could not get such funding. If one could design a financial system that was not critically dependent on short-term funding, the possibilities of contagion would be greatly diminished. In the extreme, if there were no private short-term funding, contagion would be impossible. This, of course, depends on one’s definition of short-term funding—if short-term funding was defined as 30 days or less, and such funding was replaced by 31 day funding or even 60 day funding, there would still be a contagion concern, albeit it would take more time to unfold. This chapter and the next one explore how short-term funding could be sufficiently limited to control contagion to the point where the need for lender of last resort or guarantees was virtually eliminated, and concludes this is not feasible. It also explores the downside of the government replacing the private sector as supplier of short-term funding.
Intermediation chain in repo transaction
Financial institutions have increasingly relied on short-term instruments to meet their funding needs.1 In quantifying an appropriate measure of this reliance on short-term funding, two measures are relevant: a gross amount and a net amount. We estimate both the gross amount, which is all “runnable debt” in the United States, and the net amount, in which we eliminate the double counting of liabilities that arise from financial intermediaries (namely, money market mutual funds and securities lenders). The gross amount is relevant since it is a measure of how vulnerable the financial system is to panicked runs as a result of short-term debt issuance, while the net amount is relevant to the chapter 21 discussion of how much debt the Treasury or the Fed would have to incur to crowd out private short-term debt with public short-term debt.
Consider the following example that illustrates the difference between the gross and net amounts. Suppose, as illustrated in figure 20.1, that a MMF wishes to lend $1.00 against $1.02 of collateral in a repo transaction, and that a hedge fund wishes to borrow $1.00 against $1.02 of collateral. If the two funds have institutional relationships with different banks (bank A and bank B), their single “net” trade could give rise to three “gross trades.” If we are interested in crowding out, one would net the transactions, as the entire chain could be replaced by a single transaction with the Fed, so the net amount is the relevant measure. However, if we are interested in financial fragility, and where runs could start, we should consider each runnable step of the chain and the gross amount is the relevant measure.
Summarized in table 20.1 are estimates of the $7.4 to $8.2 trillion in gross runnable short-term debt (with maturities of 30 days or less) at the end of 2014, based on the inclusion of repurchase agreements backed by US treasuries. A recent Federal Reserve note places the gross amount of runnable debt at a higher figure based on an inclusion of certain uninsured liabilities that may or may not be runnable. Either way, for contagion purposes, there is still a substantial amount of runnable debt.2 As previously discussed in chapter 12, the size of the entire private uninsured short-term funding market is approximately $10.6 trillion, or 18.6 percent of the $57 trillion total US credit market, but this higher number is not restricted to 30 day or less maturities.3
Level of runnable short-term debt
Our estimate of gross short-term funding includes five components: commercial paper, repurchase agreements, uninsured deposits, securities lending, and certain money market fund shares. We adjust these components to more narrowly identify debt of 30 days or less. First, we estimate the total amount of commercial paper with a maturity of thirty days or less to be $788 billion.4 Second, we estimate the size of the market for repurchase agreements with maturities thirty days or less to be $1,983 billion to $2,743 billion. While the maturity structure of all repo is not available in public filings,5 it is possible to roughly estimate this figure. Using money market fund repo holdings as proxy for the overall repo market, we estimate that 86 percent of term repo has a maturity of thirty days or less. Since the overall repo market (excluding repo purchased from the Fed through its reverse repo program (“RRP”), which is not runnable) is $3,190 billion, there is then roughly $2,743 billion of runnable repo.6 However, that figure includes repo collaterized by treasuries. Exlcuding treasury-backed repos would drop the amount to $1,983 billion.7 The inclusion of repos backed by treasuries depends on whether one thinks such repos are “runnable.” Recent scholarship suggests that repo collateralized by treasuries is just as vulnerable to runs as other forms of repo, so we include both amounts.8 Net repo lending from outside the financial sector tends to take place through tri-party repo. Copeland, Martin, and Walker (2014) provide evidence that Lehman Treasury tri-party repo experienced runs as severely as its other forms of tri-party repo. Third, runnable uninsured deposits are $2,537 billion.9 Fourth, net securities lending liabilities amounts to $721 billion.10 However, as noted by the New York Fed, most securities lending is conducted by asset managers, pension funds, or insurance companies that engage in the practice to increase yield, not to fund their operations.11 As a result a run on securities lending will affect returns for these institutions, but not necessarily lead to solvency problems. Finally, we assume that only prime money market mutual funds are at risk of runs, which amount to $1,403 billion.12 Therefore we estimate that the gross size of runnable private debt is around $7.4 to $8.2 trillion.13
We are also interested in estimating a measure of net runnable debt, since that is how much debt the Treasury or the Fed would have to incur to crowd out private short-term debt with public short-term debt. Since the gross amount double counts many liabilities, as illustrated above in the example of MMFs, the net figure should remove the double-counted liabilities. Therefore our estimate of the net amount of runnable debt does not include certain financial intermediaries in the calculation, namely money market mutual funds, and does not include certain financial intermediation transactions, namely securities lending, which are double counted in the gross figure. Further we consider only repo transactions with funding sourced from outside the banking sector since bank-to-bank repo transactions are simply a lengthening of the intermediation chain and are therefore double-counted in the gross amount (similar to MMFs). Therefore, to obtain the net amount we only include commercial paper, uninsured deposits, and net repurchase agreements funded outside the banking system (i.e., excluding intra-banking system repo).
As above, the total amount of commercial paper with a maturity of thirty days or less is estimated to be $788 billion and runnable uninsured deposits amount to $2,537 billion. For repos we are interested in the amount of short-term financing that originates from the nonbanking sector, thus excluding the repo transactions that occur entirely among banks. We estimate this amount by determining the quantity of repo assets held by major cash investors, namely money market funds, mutual funds, securities lenders’ cash collateral reinvestment, GSEs, and the domestic nonfinancial sector. We then exclude the amounts of the holdings held in Fed RRPs, since those liabilities should not be considered “runnable.” Our total estimate amounts to $1,172 billion.14 However, some of the repo transactions in this figure have maturities greater than thirty days. Therefore, using our prior estimate that 86 percent of term repo has a maturity of 30 days or less, we calculate the runnable repo market to be $1,008 billion, including treasury-backed repo. Excluding treasury-backed repo reduces the amount to $729 billion.15 In total, we estimate that the net size of runnable private debt is around $4.0 to $4.3 trillion.
The scale of the gross runnable short-term funding market of $8.2 trillion is comparable to the total US government debt of $13 trillion.16 However, it is nearly six times larger than the $1.4 trillion of US government debt issued with a maturity of less than one year.17 It is also nearly twice the size of the Federal Reserve’s $4.5 trillion balance sheet.18
Increased reliance on private short-term funding makes the financial system more susceptible to contagious panics, as illustrated by the run on money market mutual funds during the financial crisis. As noted in the 2014 International Monetary Fund’s Global Financial Stability report, the growth in private short-term debt issuance (i.e., growth in the supply of short-term debt) may be driven by increased demand for money-like securities from large institutional cash pools.19 These institutional cash pools include the cash balances of large nonfinancial corporations, large institutional investors and asset managers, and large sovereign wealth funds.20 These cash pools have increased from $2 trillion in 1997 to approximately $6 trillion at the end of 2013.21 These pools have large cash-management needs, and therefore represent a large and consistent source of demand for safe and liquid money-like claims.22
Consistent with this increased demand, recent research has found that the yield on safe short-term debt instruments is lower than would otherwise be expected by the added safety and liquidity of these instruments. They define and quantify this yield differential as the “money premium.”23 Indeed the money premium presumably reflects the fact that the recent increase in demand from these pools has not yet been met by a commensurate increase in supply of safe short-term debt instruments. The presence of a money premium therefore incentivizes private-sector firms to fund themselves by issuing short-term money-like claims to capture this premium. Issuers do not fully, if at all, internalize the increased risk of contagion created by this increased issuance.24
Although total short term funding has not decreased, as set forth in table 12.1 in the chapter 12 section on insuring short-term liabilities, US banks have generally become less reliant on wholesale short-term funding since the peak of the financial crisis. The composition of bank short-term funding in the United States has shifted increasingly from wholesale short-term funding to retail deposits. Wholesale short-term funding as a percentage of retail deposits has declined from roughly 140 percent in 2008 to 62 percent at the end of 2012.25 While retail funding may not be as likely to disappear in a contagion situation, as reflected in the liquidity coverage ratio, it is still vulnerable to runs. And concern continues about the level of wholesale funding, however reduced. Further the shift of bank short-term funding to the nonbank sector, over which there is less control, arguably makes the situation not better but worse.
Federal Reserve Governor Daniel Tarullo has noted the “considerable conceptual appeal” in proposals to actually cap short-term funding, although he also points out the problems with a cap and, therefore, has not endorsed this approach.26 A short-term funding cap could be instituted through an aggregate industry cap, which would require a consideration of the appropriate threshold consideration what the cap should be, such as some specified percentage of GDP.27 This approach would also require an analysis of the potential social cost from lost economies of scale and scope if financial institutions are forced to shrink to meet the cap.28 Regulators must also consider the effects on the stability of the financial system if short-term funding were to switch to the less regulated shadow banking system.29
Another approach to limiting short-term funding, endorsed by the UK Independent Commission on Banking in its September 2011, the “Vickers report,” would place a cap on the portion of a bank’s balance sheet that may be funded with short-term liabilities. While the recently enacted UK Financial Services (Banking Reform) Act of 2013 does not include an explicit short-term funding cap, it does grant regulators authority to require banks to issue “debt instruments of a particular kind,”30 leaving open the flexibility to limit short-term funding.
The focus on limiting wholesale short-term funding of financial institutions is supported by a number of recent academic studies which show that banks reliant on such funding are more likely to suffer distress.31 The IMF has found that commercial and investment banks that required government assistance during the 2008 financial crisis held significantly higher ratios of short-term debt to total debt than did banks that did not require assistance.32 Short-term wholesale funding has also been found to be the best predictor of a bank’s contribution to systemic risk.33
A key question with regard to capping short-term funding is whether such funding could be limited sufficiently to appreciably reduce contagion risk. Limits on the use of short-term debt could be implemented in conjunction with an extended insurance regime for short-term liabilities, but there could be a practical inability to lower short-term debt to a level that would appreciably reduce the probability of contagion remains.
Various academic studies have suggested indirect ways in which restrictions on short-term funding might be implemented. Many of these proposals are aimed at reforming the use of repurchase agreements and money market funds, which dominate the short-term financing markets, as asset backed commercial paper has largely disappeared as a funding instrument for financial institutions.34
One proposal is to reduce banks’ reliance on short term funding by rolling back 2005 amendments to the Bankruptcy Code that added repurchase agreements on mortgage-related assets to the exemptions from the automatic stay, the normal bankruptcy rules preventing the seizure of collateral.35 Prior to 2005, collateral in repurchase agreement transactions eligible for the automatic stay was limited to US government and agency securities, bank certificates of deposits, and bankers’ acceptances. Rolling back this new exemption would arguably increase counterparty risks on these repurchase agreements thereby reducing the willingness of financial institutions to enter into them for short-term funding. Moreover, narrowing the safe harbors to only the most predictably liquid securities can help prevent these exemptions from exacerbating a liquidity crisis.36 Although safe harbors are designed to insulate counterparties from default at the individual level, they can actually function to aggravate liquidity problems, as they facilitate “panic selling.”37
Other proposals also focus on repurchase agreements, including calling for strict regulation of securities used as collateral in repurchase agreements, limiting such collateral to only the highest quality securities for banks.38 One study has found that during the 2008 financial crisis, the contraction in repurchase agreements played a significant role for systemically important dealer banks. For example, nearly half of the repurchase agreements of Merrill Lynch, Goldman Sachs, and Citigroup with money market funds were backed by nonagency MBS/ABS and corporate debt, and almost all of this financing disappeared during the crisis.39 The FSB has recommended that regulators consider mandatory haircuts on collateral for all repurchase agreements.40
Another future Fed action that could potentially affect the short-term funding market is the Fed’s impending implementation of restrictive margin requirements for securities financing transactions.41 While the Fed’s main objective with the restrictions would likely be to address asset bubbles, higher margin requirements would directly reduce the amount of short-term funding that can be obtained with a given amount of collateral. The Office of Financial Research estimates this market to be $4.4 trillion.42 However, the effect of higher margin requirements on short-term funding markets will be dampened because the new rules will likely exempt Treasurys and agency securities, which make up two-thirds of the market.43 Regardless, as the increased restrictions limit the use of a portion of the securities financing market, the effect may be an overall increase in unsecured borrowing. Substituting unsecured borrowing for secured borrowing will consequently expose the financial system to greater risk, not less, which should be seriously considered as an unintended consequence in the wrong direction.