2. Collateral Velocity

Manmohan Singh
Published Date:
October 2016
  • ShareShare
Show Summary Details

A great deal of short-term financing is generally extended by private agents against financial collateral. The collateral intermediation function is likely to become more important over time. This chapter looks at a new concept: collateral reuse (or velocity) in the market. Although there is large issuance of good collateral, very little reaches the market. We describe how to measure this reuse rate and why this metric is increasingly important for policymakers to understand, especially when there is a shortage of collateral.1

Sources of Collateral

In the global financial system, the non-banks generally allow reuse of their collateral in lieu of other considerations. The key providers of (primary) collateral to the “street” (or large banks/dealers) are:

  • ❑ hedge funds (HFs);

  • ❑ custodians on behalf of pension, insurers, official sector accounts and so forth; and

  • ❑ commercial banks that liaise with dealers (this is relatively small compared with the supply from HFs and custodians).

Typically, HFs are suppliers of collateral while money-market funds are users, in that they supply funds to the market in exchange for collateral. HFs via their prime brokers allow for collateral reuse as a quid pro quo for the leverage/funding they receive from large banks. The other non-bank providers of collateral generally loan collateral for various tenors to optimise their asset management mandates.

The supply of pledged collateral is typically handled by the central collateral desk of dealers, who reuse the collateral to meet the demand from the financial system. Such securities serve as collateral against margin loans, securities borrowing, reverse-repo transactions and OTC derivatives. This collateral is secured funding for the dealers and is received in lieu of borrowing and/or other securities given to a client. Major dealers active in the collateral industry include Goldman Sachs, Morgan Stanley, JPMorgan, Bank of America/Merrill and Citibank in the US. In Europe and elsewhere, important collateral dealers are Deutsche Bank, UBS, Barclays, Credit Suisse, Société Générale, BNP Paribas, HSBC, Royal Bank of Scotland and Nomura.

Hedge Funds

HFs largely finance their positions in two ways: (i) loans made under prime-broker agreements with their prime brokers (PBs) and (ii) repurchase agreements (repos), generally with other banks that are not their PBs.

HFs usually pledge their securities as collateral for reuse to their PB in exchange for cash borrowing from the PB (a process also known as rehypothecation). There are limits to the degree of reuse, however. In the US, for example, Regulation T and the SEC’s Rule 15c3 limits PBs’ use of rehypothecated collateral from a clients. This means that any excess collateral of an HF cannot be used by the PB in the US for their own use, and thus remains “locked”. Regulation T limits debt to 50%, or a leverage factor of 2. With portfolio margining (ie, after netting positions), HFs can increase leverage beyond the factor of 2. However, to have more unconstrained leverage, aggressive strategies are booked offshore (eg, UK).

Typically, equity-related strategies such as equity long/short, quant-driven, event-driven and so forth are funded via PBs. Similarly, fixed-income arbitrage – global macro strategies that seek higher leverage – is done via repo financing.

Let us suppose we wish to know how much collateral was sourced from HFs (end-2007 and end-2015). This information is not readily available and needs to be estimated, since the hedge fund industry is not subject to regulatory requirements that warrant financial statement disclosures, as is the case with the banks.

Collateral Released From Equity Strategies

Intuitively, the more long positions relative to short positions there are, the more collateral is released to the market. Let us now look at the equity strategies and the associated arithmetic. As shown in Table 2.1, HFs generally borrow from PBs for equity long/short and event-driven strategies. The share of these two strategies in the mark-to-market value of collateral was 50% as of end-2007. Event-driven strategies are usually of two types: credit/distressed and merger arbitrage, which are equally split. Only merger arbitrage uses PB funding, since credit/distressed strategies do not employ leverage; so we adjust and reduce the event-driven strategies by half. Thus, for 2007, equity strategies were 36% of mark-to-market value of collateral. Based on available data, the HF industry estimates assets under management (AUM) to be at US$2.0 trillion for end-2007.

Table 2.1Hedge Fund Strategies
Market share of various HF strategies after leverage in 2007 (%)
Convert arbitrageEmerging marketsEvent-drivenFixed-income arbitrageGlobal macroLong/short equityManaged futures
Source: CS Hedge Index
Source: CS Hedge Index

The mark-to-mark value of collateral generally equates to AUM times relevant leverage in these strategies. This is the sum of long-market-value (LMV) positions and the absolute value of short-market-value (SMV) positions. Intuitively, the long positions are indicative of collateral released by HFs to the “street”. Figure 2.1 gives delta bias on the left axis. Delta bias captures the ratio of LMV/SMV. This ratio is a very useful indicator to gauge PB borrowing for HFs’ equity long/short strategies. For example, as of end-2007, the delta bias was about 50%, which means an LMV/SMV ratio of 150/100 or 3:2 (ie, collateral to PB was 3/5 of total positions). Arithmetically, delta bias equals total LMV/total SMV, minus 1. For end-2007, with AUM with HFs at US$2 trillion and equity leverage of 2.0, and a 36% share of relevant strategies, and adjusting for long/short ratio, the borrowing from PBs was about US$2 trillion (AUM) x 2.0 (leverage) x 0.36 (share of equity strategies) x 0.6 (delta bias), or about US$850 billion. Similar calculation for end-2015, gives PB borrowing of US$3 trillion x 2.0 x 0.4 x 0.55, or about US$1,350 billion. Since we separately show securities that come via custodians for sec-lending, when a PB exchanges clients’ “shorts” with custodians, we avoid the double counting.

Figure 2.1Equity Long/Short Hedge Fund Position (ie, Delta Bias)

Collateral Released From Repo Strategies

The non-equity-type strategies and non-prime-brokerage funding are largely via repos. HFs used more repo-related strategies than pre-Lehman, from about 27% in 2007 to over 40% in 2013, as rehypothecation via PB-related borrowings lost favour relative to other types of funding and the prudence for HFs to engage with more than one PB. As a rule of thumb, the repo strategies of HFs are roughly 60–70% back-to-back; so only about one-third of pledged collateral that comes in from hedge funds is free with rights to sell onward or repledge. When the rate cycle is high (as at end-2007), the 60–70% threshold may be smaller (so more pledged collateral can then be reused). However, when the rate cycle bottoms out (as at end-2013), the 60–70% threshold may be more like 80% with more back-to-back hedges, and thus less pledged collateral is released for reuse.

Again, the more long positions relative to short positions, the more collateral is released to the market. Let us now look at the nonequity strategies and the associated arithmetic for strategies that involve repo and derivatives. To estimate repo-related collateral from HFs for 2007, we take the AUM of US$2 trillion and the 27% share of strategies that used repo (see Figure 2.2).

Figure 2.2Share of Repo Strategies (Without Derivatives)

Source: FSA HFS

Including the use of derivatives, aggregate leverage in 2007 (in fixed income and global macro strategies that are funded via repo) was higher relative to equity-type strategies – at around 4 (see Figure 2.3). When rates are high, the “carry” is higher and shows a bias towards long strategies relative to short strategies; relatively more long strategies imply more collateral released to the banks for reuse. Now some associated arithmetic. Generally, about 60–70% of nonequity strategies are hedged simultaneously, so only roughly one-third of collateral is free to be onwardly repledged. So for end-2007, about US$750 billion pledged collateral came to the banks for reuse (arithmetically, 2 trillion AUM x 27% non-equity strategies x 4 (leverage including derivatives) x one-third (via rate cycle hedging threshold). In general, leverage in the UK is higher than elsewhere and so any extrapolation of leverage levels from the FSA hedge fund data (eg, Figure 2.3) needs to be trimmed. The FSA’s semi-annual hedge fund survey is now replaced with its annual hedge fund survey (by Financial Conduct Authority, UK). Since there are some changes, we rely on market information to estimate the 2015 data and not Figures 2.2 and 2.3.

Figure 2.3Representative Leverage Levels in Some HF Strategies

Source: FSA HFS

Doing similar arithmetic for 2015, with aggregate leverage a bit lower relative to 2007 at 3.5, AUM higher at US$3 trillion and share of HF strategies using repo (around 30%) and back-to-back “threshold” closer to 80% due to the bottom of the rate cycle, we would estimate that US$650 billion collateral from HFs was pledged for reuse to the banks via repos (arithmetically, 3.0 trillion AUM x 30% in repo strategy x 3.5 (leverage including derivatives) x 0.20 (via rate cycle hedging threshold).

Strategies that do not Involve Borrowing/Leverage

Note that a managed futures strategy is via cash that goes to an exchange such as the Chicago Mercantile Exchange (CME), and thus is not a collateral/leverage-based strategy; also, emerging markets or distressed strategies do not generally require leverage via PBs or repo via non-PBs. Some HFs hold AUM in cash. Thus, the total PB and non-PB strategies (with leverage) do not entail that all the total AUM x leverage will hit the street.

To be technical, if about two-thirds of strategies are hedged, the collateral from the remaining one-third may not all be turned to cash by the banks – it depends on their balance-sheet space, and this issue began to get more traction in 2015. Also, banks can be very different, with UBS curtailing balance-sheet activities in the pledged-collateral area, while some others try to enter this market.

So, in summary the total collateral from HFs that came to the large dealers (and hit the street) is estimated to have been about US$1.6 trillion as of end-2007, with US$850 billion to have come via PB funding and US$750 billion from repo funding outside the PBs. Similarly, arithmetic for 2015 suggests that US$2.0 trillion of collateral from HFs came to the large dealers. Leverage became lower (after the collapse in 2008–9) but slowly inching higher in more recent years; however, AUM with HFs are higher as of end-2015 at US$3 trillion.

We now look at the other source of collateral that comes to banks: non-HF sources.

Securities Lending – Another Primary Source of Collateral

Securities lending provides collateralised short-term funding, just like repo. In a repo there is an outright sale of the securities accompanied by a specific price and date at which the securities will be bought back. On the other hand, securities lending transactions generally have no set end date and no set price. The beneficial owner can recall the shares on loan at any time and the borrower can return the shares at any time. Thus, securities lending transactions are much more flexible than repos and thus are more conducive to covering shorts where the position’s profitability relies on exact timing/tenor matching. Furthermore, with respect to legal rights, securities lending is effectively identical to repo. For example, both transactions include full transfer of title. The asset-management complex, which includes pension, insurers and official sector accounts such as sovereign wealth funds and central banks, is a rich source of collateral deposits. The securities they hold are continuously reinvested to maximise returns over their maturity tenor.

We use the Risk Management Association (RMA) as the main data source (see Table 2.2), which includes only primary sources of securities lending from clients such as pension funds, insurers, official sector accounts and some corporate/money funds. The RMA’s data includes the largest custodians such as the BoNY, State Street and JPMorgan. (Another data source, Data Explorers, shows larger numbers, as it includes a significant part of the secondary market activity also. A Bank of England paper (2011) using Data Explorers states that about US$2 trillion of securities were on loan, but includes secondary holdings or collateral reuse rate also, ie, it also counts the bank-to-bank holdings of primary sources.)

Table 2.2Securities Lending 2007–2015 (US$ Billion)
Securities lending versus cash collateral1,209935875818687620669701644
Securities lending versus noncash collateral486251270301370378338425454
Total securities lending1,6951,1871,1461,1191,0589981,0081,1371098
Source: RMANote: Collateral received from pension funds, insurers, official accounts, etc (USD, billions)
Source: RMANote: Collateral received from pension funds, insurers, official accounts, etc (USD, billions)

The risk aversion due to counterparty risk since Lehman has led many pension and insurance funds’ official accounts not to let go their collateral for incremental returns. These figures are not rebounding as per end-2011 financial statements of banks, and anecdotal evidence suggests even more collateral constraints since.

The decline in the first row of Table 2.2 needs some explanation. The US regulatory rules that guide borrowers permit only cash, and certain government securities. Hence, the US developed as a cash collateral business, where the lending agent lends client assets versus cash and then reinvests the cash, according to the client’s instructions, in very short-term reinvestments. Outside the US (in the UK, for instance), regulatory rules permit certain types of noncash collateral that are readily available (such as FTSE equities). In the aftermath of Lehman and the liquidity crisis, borrowers in the US borrowed more hard-to-borrow stocks (specials), and less general collateral; this explains the decline evident in the table. Noncash collateral deals (ie, collateral for collateral) effectively provide the lenders with a hard fee for the deal, and it does not give temporary cash to generate excess returns by creating a short-term money-market book (also see Panel 8.1 on the future of securities lending).

Bank–Dealer Collateral

Dealers occasionally receive requests from commercial banks for collateral swaps. In such a transaction, typically the collateral posted by the commercial bank may need an “upgrade”. Discussions with dealers suggest that such requests are generally minimal and thus insignificant relative to the collateral flows from the key clients (such as HFs, pension funds, insurers and official accounts). We acknowledge such flows in Figure 2.4 with a de minimis, but do not consider these flows to impact on the arithmetic for the velocity of pledged collateral. Other sources of collateral are not material, since we consider only that collateral that has no legal constraints on reuse.

Figure 2.4The Sources and Uses of Collateral – Summary (2007, 2010-2015)

Figure 2.4 shows the sources of collateral (in the circles) and overall collateral received by the banks (in the rectangle) for 2007, 2010, 2011, 2012, 2013, 2014 and 2015. The years 2008 and 2009 were in flux with several key banks in the pledged-collateral market disappeared (eg, Lehman) or merged with other banks (eg, Bear Stearns, Merrill Lynch). Even for the year 2010, this data has been verified with investor-relations groups wherever banks were merged or absorbed.

Methodology for Calculating the Velocity of Collateral

Our understanding is that there are 10–15 large banks active in collateral management globally. We may have missed a couple of banks but believe we have picked up over 90% of the pledged collateral that is received from primary sources such as hedge funds, pension funds and insurers, and official accounts.

We compare data between 2007 and 2015 to see how this market has changed from before Lehman’s bankruptcy through the financial crisis, which straddles monetary policy experiments. As a starting point, we take the total collateral received by the banks as of end-2007 (almost US$10 trillion) and compare it to the primary sources of collateral (the two primary-source buckets identified in Figure 2.4, namely HFs and security lenders on behalf of pension, insurers, official accounts etc). The ratio of the total collateral received/primary sources of collateral is the velocity of collateral due to the intermediation by the dealers:

Collateral Sources as of End-2015

Similarly, for 2015, total collateral from primary sources that could be repledged by the large dealers from hedge funds was US$2.0 trillion, plus US$1.1 trillion via security lending operations of custodians on behalf of pension funds, insurers and official sector accounts, for a total of US$3.1 trillion. The total collateral received by the 10–15 large banks was US$5.6 trillion as of end-2015 (still sharply lower than the US$10 trillion peak as of end-2007, but bouncing back from the trough of US$5.0 trillion as of end-2009).

Table 2.3 provides a summary of the sources of collateral, the total volume received by the large banks and the resultant velocity. The velocity is not an exact metric, but gives an idea of the length of the collateral chains in that year. So we can infer that, on average, the collateral chains were longer in 2007 than in 2015. The intuition is that counterparty risk before Lehman was minimal but has changed since then (due to some central bank’s quantitative-easing policies, the ongoing European crisis, etc). With fewer trusted counterparties in the market owing to elevated counterparty risk, this leads to stranded liquidity pools, incomplete markets, idle collateral and shorter collateral chains, missed trades and deleveraging.

Table 2.3Sources of Pledged Collateral, Velocity, and Collateral, 2007, 2010–15 (In US$ Trillions; Velocity in Units)
YearHedge fundsOthersTotalVolume of secured operationsVelocity
Sources: Risk Management Association; IMF Working Paper: Velocity of Pledged Collateral (Singh, 2011)
Sources: Risk Management Association; IMF Working Paper: Velocity of Pledged Collateral (Singh, 2011)

Panel 2.1:Augmenting Rate of Return on Security by Pledging IT for Reuse

The “supply” of pledged collateral comes from non-banks. This is received by the central collateral desk of banks that reuse the collateral to meet the “demand” from other intermediaries – bank or non-bank – in the financial system. This collateral primarily moves to augment returns (ie, return enhancement, not risk transformation). Thus a US Treasury that matures in 30 years that has a coupon of 4% does not, over its lifespan, yield 4% to the owner. Aside from the fluctuating market pricing/yield of this security, the return due to reuse in the collateral space will typically provide an extra return to the owner of the security over its tenor (t0 to t30). Mathematically, if x is the 30-year US Treasury with 4% coupon, then total returns to the owner if the security is not siloed is 4% plus x. The source collateral may include AAA securities such as US Treasuries, German bunds (ie, German government debt issuance), or CCC bonds or equities. Thus, this collateral market moves securities that may not be “safe” or AAA/AA as long as the security is liquid and has a market clearing price.

Collateral Velocity and Deleveraging

Large dealers are incredibly adept at moving collateral they receive that is pledged for reuse. The interconnections nexus has become considerably more complex over the past two decades. The reuse rate of collateral – analogous to the concept of the “velocity of money” – indicates the liquidity impact of collateral. A security that is owned by an economic agent and can be pledged as reusable collateral leads to chains. Thus, a shortage of acceptable collateral would have a negative cascading impact on lending similar to the impact on the money supply of a reduction in the monetary base. Thus the first-round impact on the real economy would be from the reduction in the primary-source collateral pools in the asset-management complex (hedge funds, pensions, insurers, etc), due to averseness to counterparty risk etc; such collateral remains idle and does not contribute in completing markets. The second-round impact is from shorter “chains” – from constraining the collateral moves – and the higher cost of capital resulting from a decrease in global financial lubrication (see Figure 2.5).

Figure 2.5Deleveraging Components – Balance Sheet and Interconnectedness

Source: FSA HFS

Panel 2.2:The 10–15 Banks at the Core for Global Financial Plumbing

Let the financial system that includes banks, hedge funds, pension funds, insurers, SWFs (sovereign wealth funds), etc, be represented by A to Z. Only a handful (say XYZ) can move financial collateral across borders. XYZ also happen to be the large 10–15 banks discussed earlier. The rest of the financial system from A to W that demand and supply collateral need to connect with each other via XYZ. Entry into this market is not prohibited but extremely expensive and difficult, as we need a global footprint and global clients (and the acumen and sophistication to move and price liquid securities very quickly – in seconds sometimes).

For example, a Chilean pension fund may want Indonesian bonds for six months, and W (a hedge fundor a security lender in Hong Kong) may be holding these bonds and willing to rent out to A for six months for a small fee.

But W does not know there is demand from A. Only via XYZ can A connect to W. Since XYZ sit in the middle of the web, they have the ability to optimise in ways that give them an advantage – the Indonesian bonds may come into their possession because they’ve loaned W money, or because they have a derivative with W, or through a security lending agreement.

Such securities that need to move cross border under a “repo” or “security lending” or related transaction need to be legally perfected (and herein legal perfection entails rules such as title transfer and rehypothecation). Similarly for OTC derivative margins, there is an International Swaps and Derivatives Association master agreement. For prime-brokerage/HF collateral, there is a similar master agreement that resonates easily between XYZ.

Thus it is not easy for all real-economy collateral to be able to move across borders. This market for bilateral pledged collateral is the only true market that prices at mark-to-market all liquid securities (bonds + equities).

Given that collateral is in short supply (as reflected by repo rates), either of two things can happen.

  • Velocity of collateral comes back – this is a task that only XYZ can handle in bulk if more good collateral is sourced through them. However, regulatory proposals such as leverage and liquidity ratio may result in balance-sheet constraints for XYZ to do collateral transformation. Or central banks can make balance-sheet room for XYZ (as with the Fed’s reverse repo programme since September 2013 – see Chapters 4 and 11).

  • Or – like the Reserve Bank of Australia (RBA), which provide good collateral to meet the increase demand when regulations kick in – this will be market-based. The RBA will not issue new debt to meet this demand (unlike proposals in academic circles – Gourinchas and Jeanne 2012). The European Central Bank (ECB) type of approach also helps, but collateral pricing may not be market-based. However, QE or purchase of good collateral that is taken from the market and silo-ed at a central bank will only lower velocity of collateral.

The first (and more familiar) round involves the shrinking of balance sheets. The other is a reduction in the interconnectedness of the financial system. The reduction in debt (or deleveraging) has two components (see the last equation in Annex 2.1). Most recent researchers have focused on shrinking balance sheets (by shedding assets), overlooking this other deleveraging resulting from reduced interconnectedness. Yet, as the current crisis unfolded, key actors in the global financial system seemed to be ring-fencing themselves owing to heightened counterparty risk. While this seems rational from an individual perspective, such behaviour may have unintended consequences for the financial markets. Deleveraging from the shrinking of bank balance sheets is not (yet) taking place; however, we still find the financial system imploding as collateral chains shorten.

The balance-sheet shrinking due to “price decline” (ie, increased haircuts) has been studied extensively, including in the April 2012 Global Financial Stability Report of the IMF and the European Banking Association recapitalisation study of 2011. Some of the academic literature on this issue spans the work initiated by Geanakoplos (2003).

However, deleveraging of the financial system due to the shortening of “repledging chains” has not (at the time of writing) received attention. This deleveraging is taking place despite official sector support. This second component of deleveraging is contributing to the higher credit cost to the real economy. In fact, relative to 2006, the borrowing costs (in spread terms, after adjusting for rate cuts) have been lowered – see Figure 2.6, first two charts; however, most of the world outside US is still dominated by banks, as the last chart of Figure 2.6 shows, for the past three decades, the cost of borrowing for financials has been below nonfinancials; however, this has changed post-Lehman. Since much of the real economy resorts to banks for borrowing (aside from the large industrials), the higher borrowing cost for banks is then passed on to the real economy.

Figure 2.6Average Cost of Borrowing for the Real Economy (US and Europe Indexes)

Source: BoA-ML indexes; Barclays Intermediate

Markets will first use money or collateral, whichever is “cheapest-to-deliver” when settling debits/credits. Markets will maximize and hold money or collateral, whichever has a better rate of return. Thus it is difficult to quantify what fraction of pledged collateral is used by the “real economy”; both money and collateral are pooled together.

As the “other” deleveraging continues, the financial system remains short of high-grade collateral that can be repledged. Recent official sector efforts – such as the ECB’s “flexibility” and the ELA (or emergency liquidity assistance) programmes of national central banks in the eurozone – in accepting “bad” collateral attempt to keep the good–bad collateral ratio in the market higher than otherwise. The ECB’s acceptance of good collateral and bad collateral (which during the crisis was at nonmarket prices) brings into play Gresham’s Law, which, briefly put, states that bad money drives out good. But, if such moves become an integral part of the central banker’s standard toolkit, the fiscal aspects and risks associated with such policies cannot be ignored. By using this aspect of the standard toolkit, the central banks have interposed themselves as risk-taking intermediaries with the potential to bring unintended consequences: front-loading consumption by sizable quantitative easing and (in some cases) interfering in markets’ plumbing is all unchartered territory.

With new Regulations, Collateral Velocity Calculation May Get Difficult

So far, the demand and supply for financial collateral by non-banks (and other commercial banks) is intermediated by the large 10–15 banks/dealers that have a niche in this cross-border collateral market (we met this group under “Methodology for calculating the velocity of collateral” above). However, as regulations kick in, some of the non-banks can develop in-house teams to deal with central counterparties, or CCPs, directly: Allianz, La Mondiale, Scottish Widows, Friends Life, VPV, Sun Life, etc. These may consider liaising directly with banks (and not via agents/custodians).

Similarly, central banks may become large conduits and alleviate collateral shortage to non-banks (see Chapter 4 for the Fed’s reverse repo with non-banks).

The dangers of the possible effects of weakening collateral chains were discussed in a column in the Financial Times in 2012, written by this author, which we have reproduced here as Panel 2.3.

Panel 2.3:Beware Effects of Weakening Collateral Chains

Financial Times column by Manmohan Singh, June 27, 2012

The past decade’s build-up of debt capacity in the global financial system needs to unwind. Bloated banks are desperate to shed assets but there are very few buyers; so, for now, the official sector is keeping afloat their balance sheets.

The reduction in debt (or deleveraging) has two components. The first (and more familiar) involves the shrinking of balance sheets. The other is a reduction in the interconnectedness of the financial system. Most recent researchers have focused on the impact of smaller balance sheets, overlooking this “other” deleveraging resulting from reduced interconnectedness. Yet, as the current crisis unfolds, key actors in the global financial system seem to be “ring fencing” themselves owing to heightened counterparty risk. While “rational” from an individual perspective, this behaviour may have unintended consequences for the ecology of the market.

The interconnectedness of the financial system may be viewed from the lens of collateral chains. Typically, collateral from, among others, hedge funds, pension funds, insurers and central banks, is intermediated by the large global banks. So a Hong Kong hedge fund may get financing from a Swiss bank secured by its collateral, say, Indonesian bonds which will be pledged to the Swiss bank’s UK affiliate for reuse. There may be demand for such bonds from, for instance, a pension fund in South America which may have a Spanish bank as its global bank. However, due to heightened counterparty risk, the Swiss bank may not want to onward pledge to the Spanish bank; such collateral thus remains idle with the Swiss bank. With fewer trusted counterparties in the market owing to elevated counterparty risk, this leads to stranded liquidity pools, incomplete markets, idle collateral and shorter collateral chains, missed trades and deleveraging.

At the end of 2007, the large banks received about $10tn in pledged collateral. For the same period, the primary sources of collateral (via hedge funds and custodians) that was intermediated by the banks was about $3.4tn. So the ratio (or reuse rate of collateral) was about 3 times as of end-2007. A recent IMF paper shows that this ratio decreased to about 2.4 as of end 2010 (and 2.2 as of end-2012), largely due to counterparty risk within the financial system in the present environment. These figures are not rebounding as per end 2011 financial statements of banks (indeed anecdotal evidence suggests even more collateral constraints recently).

Such reduced market interconnectedness, or the trend toward “fortress” balance sheets, may be viewed positively from a financial stability perspective if one simply views each institution in isolation. However, the vulnerabilities that have resulted from the weakened fabric of the market may yet have to become fully evident. Since the end of 2007, the loss in collateral flow is estimated at $4tn–$5tn, stemming from shorter collateral chains and increased “idle” collateral due to institutional ringfencing; the knock-on impact is higher credit costs for the economy. In recent months, European banks have been given some breathing room due to the European Central Bank’s longer-term refinancing operations (LTROs); the deleveraging from the shrinking of balance sheets may therefore not be immediately evident – except in some obvious cases. Yet the financial system ecology is being transformed by ring fencing, with potential significant unintended consequences.

Overall financial lubrication in the US, UK and eurozone exceeded $30tn before Lehman’s bankruptcy (of which one-third came via pledged collateral and the rest from money, measured by M2). Certain central bank actions, such as the ECB’s LTRO, the US Federal Reserve’s easing and the Bank of England’s asset purchase facility have been effective in alleviating collateral constraints. However, these “conventional” actions, to the extent they merely exchange bank reserves for collateral of prime standing (such as US Treasuries), do not address the issue at hand. A rebound in the pledged-collateral market would be better served by further easing of liquidity constraints than by more QE. It would also more likely restore bank lending to the overall economy.

As the deleveraging continues, the financial system remains short of high-grade collateral that can be repledged. Recent efforts such as ECB’s flexibility in accepting “bad” collateral may partly alleviate this. But, if such moves become part of the central banker’s standard toolkit, the fiscal aspects and risks cannot be ignored. By so doing, the central banks have interposed themselves as risk-taking intermediaries with the potential to unleash significant unintended consequences.

Annex 2.1: Deleveraging Components – Balance Sheet and Interconnectedness

The purpose of this annex is to provide a mathematical framework developed by Hyun S. Shin (2009). In summary, this shows how the unwinding of systemic leverage can be separated into two components: balance-sheet shrinking (due to haircuts/shedding of assets) and reduced interconnectedness within the financial system (due to shorter collateral chains).

xi = market value of bank i’s total liabilities

yi = market value of bank i’s assets that can be pledged as collateral

ei = market value of bank i’s equity

ai = market value of bank i’s assets

πji = proportion of j’s liabilities held by i

di=1(eiai) is the ratio of debt to total assets

Noting that the total assets of bank i are given by ai=yi+jxjπji and from a simple accounting identity, it follows that the total debt can be computed by multiplying the totals assets with the leverage ratio:

Let x = [x1xn], y = [y1yn], and Δ = diag[d1 …, dn] and rewriting the previous equation in vector form:

Solving for x and using Taylor series expansion,2

The matrix ΠΔ is given by:

The interaction between institutions and the system is elegantly captured by the above matrix notation. While we often talk about systemic leverage and systemic risks, the notation captures a very subtle issue in that it makes a distinction between the impact of systemic leverage on an institution and the impact of the institution on the remaining system. This distinction between the two concepts is essential to breaking down endogenous systemic leverage into two exogenous variables, which provide additional insight into the economics of building leverage through collateral. The sum of the elements of the i-th row of ΠΔ represents the net impact of bank i’s leverage of the remaining system. The sum of the elements of the i-th column represents the net impact of systemic leverage on bank i. Note that the powered matrices (ΠΔ)t indicate the collateral value of the asset in the t-th link of the repledging chain.

Using the matrix, the change in deleveraging can be decomposed into two effects: the price decline on balance-sheet assets and the decline in the interconnectedness factor, independent of the price decline of assets. Assume there is a parameter _ that captures measured risks, which affects both the price of marketable assets (y) and the haircuts (which determines the debt ratios and consequently Δ). Denote Δ(σ) as the diagonal debt ratio matrix, and y(σ) as the market value of marketable securities as function(s) of σ. (note (y) is defined here as price of marketable assets on the balance sheet and off the balance sheet, ie, pledged assets).


Suppose σ < σ’, then the decline in debt is given by:

Rewrite this as follows:

This identifies two parts: the balance-sheet shrinking (via price declines/haircuts on the balance sheet) and the reduced interconnectedness (due to shorter collateral chains). The first has been studied extensively. The second term represents the deleveraging in the financial system and could be significantly larger than the on-balance-sheet (first term).


Note that the sum of the elements of the rows of is always strictly less than 1. This means that the infinite Taylor series converges and hence, has a well-defined inverse.

    Other Resources Citing This Publication