RDP 2008-06: Promoting Liquidity: Why and How? 6. The Provision of Systemic Liquidity Services by the Public Sector

To some extent, liquidity can be considered a public good. As discussed above, it is possible that social welfare is improved if financial institutions do not have to fully self-insure against system-wide liquidity problems. Indeed, in some situations it may be almost impossible for them to do so, particularly if there is only a limited supply of outside liquid assets.

If the public sector is going to play a role in providing ‘systemic liquidity services’ to the private sector, there are a number of channels through which this can be done, including:

  • the central bank's open market operations;
  • the outright purchase of assets where liquidity is a problem;
  • the provision of liquidity assistance to an institution experiencing funding difficulties; and
  • assisting with off-market transfers of assets.

Each of these is discussed in turn below. The following section then discusses some of the conditions that might apply to the provision of these services.

6.1 Open Market Operations

A basic function of a central bank is to manage the supply of settlement balances or reserves to ensure that the relevant interest rate (typically, an overnight money market rate) is close to the target level set for the purposes of monetary policy. The way in which this is done can have significant implications for how financial institutions manage their own liquidity, and for the liquidity characteristics of various assets. Through its open market operations, the central bank can create assets with unquestionable liquidity for the financial sector to hold, and by deeming assets as eligible for market operations, it can reduce illiquidity premia that apply to those assets. Market operations can also affect the maturity structure of banks' liabilities and can be used, under some circumstances, as a channel to provide funding to institutions suffering temporary liquidity difficulties.

It has become commonplace for central banks to conduct these operations primarily in repos. Doing so makes it possible to undertake operations in a wide range of assets without taking on a high level of risk, since for a loss to occur, the central bank's counterparty would need to fail, and the value of the underlying asset would need to fall significantly. Many central banks, however, also still use outright transactions to inject or withdraw cash from the system, although these operations are largely restricted to assets of the highest credit quality that trade in very liquid markets, typically government securities.

6.1.1 Accommodating an increase in the demand for liquid assets

As we have seen recently, during a period of strain in financial markets, the demand for assets of unquestionable liquidity increases significantly. The central bank is ideally placed to respond to this increase, as it is in the unique position of being able to create such assets easily. It can do this by buying other assets from the private sector and, in exchange, providing institutions with the most liquid asset of all – a deposit at the central bank. If this is done through a repo, the incremental risk to the central bank need only be small.

In effect, such operations – which involve an expansion of the central bank's balance sheet – allow private institutions to improve the liquidity characteristics of their own portfolios; while the assets that are sold to the central bank may themselves normally be traded in liquid markets, there is always the possibility that some disruption to these markets will reduce their liquidity in times of stress. This possibility does not exist with central bank balances.

This type of expansion in the central bank's balance sheet is more likely if the central bank pays a close-to-market interest rate on deposits. If interest is not paid, there can be a high opportunity cost for financial institutions of holding large balances, so that if the supply of these balances increases significantly, the overnight interest rate is likely to fall below the central bank's target as institutions seek to lend these balances. An expansion is also more likely to occur in countries where the supply of ‘outside’ liquid assets is limited, since if system-wide credit quality concerns emerge, the demand for ‘inside’ assets is likely to decline significantly, with central banks' balances being the main alternative very liquid investment.

The central bank can also accommodate an increase in demand for liquid assets by altering the structure of its own balance sheet (without changing its size). In particular, it can reduce its own holdings of assets that are highly liquid (primarily government securities) and, in exchange, increase its holdings of assets that are less liquid.

Arguably, during periods in which liquid assets are very highly valued (forcing down the relative yields on these assets), it makes little sense for the central bank to hold the most liquid assets in the financial system. Provided the risk issues can be addressed, the central bank can play a type of smoothing role, by being prepared to reduce its own holdings of the most liquid assets at the very time that the private sector most values these assets. It is important to stress that, in playing this role, the central bank is in no sense bailing out banks, or funding the balance sheet expansion of the banking system. It is simply reducing its own call on the assets with the most favourable liquidity characteristics at a time when the private sector most values liquidity. In doing so, it can help reduce the amplitude of swings in the price of liquidity, and it can do so without taking significant risks.

Over the past year or so, many central banks have responded in this way, expanding their balance sheets and/or changing the composition of their assets.[10] The exact details have, to a significant extent, depended on institution-specific factors, including the composition and size of the central bank's balance sheet, the assets accepted in open market operations, and whether interest is paid on balances at the central bank. For example, reserve balances at the Bank of England rose from an average of around £20 billion in the first half of 2007, to an average of around £26 billion over the past six months (Figure 4). Similarly, in Australia, the banking system's balances at the RBA have also risen, from a daily average of around A$0.8 billion in the first half of 2007, to a peak of almost A$7 billion in December 2007 (Figure 5).[11] Early on in the current episode the RBA also reduced its limited holdings of Commonwealth Government securities held on an outright basis, as well as both its government securities held under repo and its US dollar assets held under swap arrangements (Figure 6). At the same time, the RBA increased its holdings of bank-issued paper held under repo. In the United States, there has also been a significant change in the structure of the Fed's balance sheet, with a large decline in the Fed's holdings of government securities held outright and an increase in the value of agency-backed mortgage-backed securities held under repo (Figure 7). With the introduction late last year of the term auction facility (TAF), there has also been a very large increase in the Fed's holdings of the wide range of relatively illiquid assets that banks pledge for use at the discount window. The Fed and the Bank of England also introduced facilities allowing banks to swap assets that were not particularly liquid for highly liquid government securities (the term securities lending facility, TSLF, for the Fed).

Figure 4: Reserve Balances at the Bank of England
Figure 5: RBA Exchange Settlement Balances
Figure 6: RBA Repo Assets
Figure 7: Federal Reserve Assets

6.1.2 The choice of assets eligible for a repurchase agreement

A related issue that has attracted considerable attention is the range of assets that the central bank is prepared to purchase under repo.

As recent experience illustrates, during a period in which conditions are strained, financial institutions have a strong preference to hold assets that can be used in operations with the central bank. This partly reflects a concern that other assets may not be easily sold in the private market if the institution needs funds at short notice. By making an asset eligible for repos, the central bank can reduce the (illiquidity) premium that might otherwise be needed to induce investors to hold that asset. Increasing the range of eligible assets is also likely to give institutions greater confidence that should liquidity pressures emerge, they have appropriate assets to undertake operations with the central bank.

Historically, in many countries, including Australia, the list of eligible assets has been relatively narrow. The logic for this was that the central bank simply did not need to accept a wide range of assets to conduct its markets operations effectively, and/or that accepting assets other than of the highest credit quality exposed the central bank to an unacceptable degree of risk.

One alternative to this historical view is that, in principle, all assets on the balance sheets of financial institutions should be eligible, subject to the risks to the central bank being adequately addressed. By accepting all assets, illiquidity premia that exist because of a lack of market infrastructure or market turmoil would be reduced, and the banking system would be less susceptible to liquidity crises, with both effects potentially increasing welfare. According to this perspective, the risk issue is best addressed, not by the central bank refusing to deal in some asset classes, but by setting appropriate haircuts, advancing fewer funds against more risky assets.

Some central banks have gone a considerable way towards adopting this approach. Since the onset of the turmoil the central banks that had a relatively narrow range of eligible assets for their regular operations, including the Fed, have tended to widen the range, joining the European Central Bank and the Bank of Japan which already had very broad ranges of eligible collateral. At a practical level, one concern with accepting any assets under repo is that it can be very difficult to value illiquid assets, and to determine the true nature of the risks, especially where information asymmetries are acute. This can make setting appropriate haircuts very difficult. One possible response to this uncertainty would be to apply ultra-conservative haircuts to hard-to-evaluate assets, although this may undermine any benefit that might otherwise be gained from making these assets eligible for repos. Furthermore, within a class of illiquid and difficult-to-value assets with idiosyncratic properties – typically non-traded assets such as loans – there is the potential for a ‘lemons’ problem if a common haircut is applied. Within such an asset class, it would be possible that the central bank would only be presented with inferior assets for which a sizeable haircut was effectively less punitive.

A related issue is whether assets that have been either originated or sponsored by an institution (say its housing loans) should be accepted under repo from that institution. The main concern here is that taking such assets as part of normal market operations can increase the risk to the central bank, as the ‘double protection’ that arises from conducting repos in third-party, or non-related, assets is significantly reduced. Doing so may also lead to financial institutions reducing their holdings of other liquid assets, while accepting assets from the institution that originated them may crowd out secondary markets because it reduces the incentive for originators to stimulate markets for those assets.

Again, an in-principle case could be made to take such ‘related’ assets, subject to appropriately calibrated haircuts. Doing so would seem less problematic if the lemons problem could be reduced, say through some combination of credit quality conditions on the loans or the loans being securitised and rated. This approach can be used to overcome, to a significant extent, the information asymmetries that might otherwise arise from taking mortgages directly from an institution, particularly where the central bank does not have the expertise, or in a crisis, time, to evaluate the quality of those mortgages. The RBA has adopted a variant of this approach for banks to have access to additional securities that they can use to obtain liquidity from the RBA in a period of turmoil. Here the RBA will accept only the AAA tranche of a securitisation of an institution's own prime mortgages. The Australian Prudential Regulation Authority has indicated that these so-called ‘self-securitisations’, of which banks have constructed $53 billion in the past six months, should not be substitutes for financial institutions' holdings of more conventional liquid assets.

6.1.3 The maturity of repos

Another aspect of market operations that has drawn attention is the maturity of these operations.

If, at one extreme, the central bank undertakes all its operations in overnight repos, the banking system is required to sell securities to the central bank each and every day, buying them back the next. In this world, an institution that sold securities would get cash only overnight, and would then need to bid again in the open market operations with other institutions the following day. In the event that this institution was unsuccessful on the second day, it might need to arrange a repo (or another transaction) with a private counterparty to obtain the funding it was seeking. If market conditions are unsettled, this may be difficult or costly. To the extent that institutions are concerned about this possibility, they are likely to be less willing than otherwise to provide term funding to their clients.

In contrast, if the central bank conducts longer-term repos, say for a maturity of six months, repo turnover is reduced, but institutions that sell securities to the central bank obtain cash for a longer period, thus reducing their rollover risk. At the margin, this may promote term funding. Similarly, conducting longer-term repos may encourage institutions to purchase longer-term securities in order to repo to the central bank, reducing term premiums. Also, as discussed above, to the extent that repos are used by institutions to substitute less-liquid assets for more-liquid assets, the benefit of doing so is likely to be greater if the substitution is in place for a longer time.

Not surprisingly, given the benefits of undertaking longer-term repos at times when illiquidity premiums are high, most central banks have responded by increasing the maturity of their operations. In Australia, the RBA has long had a flexible approach, and has avoided having fixed maturities. Recently, it has used this flexibility to extend the average maturity of its outstanding repos from around 20 days over the first half of 2007, to around 75 days in May this year (Figure 8).[12] The average maturity of repos in other countries tends to be shorter than that in Australia, although in almost all cases it has increased over recent times (Figure 9).

Figure 8: Average Maturity of RBA Repos
Figure 9: Long-term Repos

6.1.4 Provision of funding to an institution experiencing difficulties

Finally, while a central bank's market operations are typically thought of as dealing with system-wide liquidity issues, they can also address liquidity strains being experienced by an individual institution. In particular, an institution having difficulty funding itself in the market is able to bid aggressively for funds in the central bank's operations, providing it has appropriate assets to repo. It might do this if the private repo or outright markets in the relevant assets have been disrupted or, for some reason, market participants do not want to take any counterparty exposure to a troubled institution, even by way of a well-secured repo. For this to be a practical option, the central bank would have to conduct open market operations frequently, preferably daily, so that a troubled institution does not have to wait to access funding.

There is, however, a limit to the extent to which market operations can be used in this way, as the size of daily operations is often relatively small compared to the funding requirements, particularly of a large bank. Furthermore, an institution that bid very aggressively for large volumes of funds over a number of days might expect to attract follow-up inquiries from the central bank and/or prudential supervisor, and to the extent that its activities become known, this has the potential to heighten market concern.

6.2 Direct Transactions in Markets

A second possible way in which the public sector can address liquidity issues is to purchase assets outright. This can be done by either the central bank or another public sector body.

This idea is sometimes seen as being quite controversial, although it has been applied to the foreign exchange market on numerous occasions. In particular, central banks (including the RBA) have been prepared to intervene in the foreign exchange market to provide two-way liquidity, and have also intervened when the value of the domestic currency was judged to be inconsistent with its fundamental value.[13] Similar intervention in other asset markets is rare, although in Hong Kong the authorities purchased equities during the late-1990s Asian financial crisis.

In principle, the same logic that has been used to justify direct purchases or sales of domestic currency for foreign currency could be used to justify direct purchases of other assets. If an asset market lacks two-way liquidity, or prices have moved far from fundamental value, a case could be made that the public sector should step in. Indeed, in Australia, two proposals have argued recently that an entity sponsored by the Australian Government should be prepared to acquire highly-rated home loans/RMBS if funding conditions in the mortgage market are severely disrupted.[14] Similar arguments have been made by Buiter and Sibert (2007) in an international context.

This type of direct intervention can, however, expose taxpayers to considerable risk, distort the operation of markets in allocating resources, and potentially delay the recovery of the secondary market. Given this, there would seem to be a strong case to consider such intervention only if:

  • the lack of liquidity, or misalignment in prices, was likely to have first-order adverse effects on the macroeconomy;
  • the lack of liquidity, or misalignment in prices, was the result of some clear market failure, and was not likely to be rectified in a timely way; and
  • any intervention was not likely to materially distort the pricing of similar assets or affect the structure of the market in normal times.

If applied, these criteria would significantly restrict the types of assets for which intervention might be considered. They would almost certainly rule out purchases of assets with idiosyncratic features and where there were large information asymmetries. The most likely candidates are perhaps mortgage-backed securities and other high-quality bonds, although even here the likelihood of the above criteria being satisfied would appear to be quite low. Notwithstanding this assessment, it is possible that situations arise where the outright purchase of financial assets is in the public interest. In extremis, the public sector, with its long-time horizon and large balance sheet, may be able to play a role in providing necessary liquidity to key asset markets, and to limit the consequences of severe market disturbances driving asset prices a long way from their fundamental value.

6.3 Emergency Liquidity Assistance

A third possibility is to provide an explicit loan to a solvent, but troubled, institution; this is typically known as ‘emergency liquidity assistance’ or lender-of-last-resort (LOLR) loan. While no such loans have been made in Australia for many decades, emergency liquidity assistance was recently provided by the Bank of England to Northern Rock, and by the Federal Reserve Bank of New York to Bear Stearns/JPMorgan Chase.[15]

This type of liquidity support can expose the public sector to considerably more risk than that incurred through market operations. Not only is the value of any collateral likely to be more uncertain (as the standard assets used for repos will have been exhausted), but the ‘double protection’ offered by repos in third-party assets does not apply. Moreover, liquidity problems will almost certainly reflect market concerns about the ongoing ability of the institution to repay its liabilities. While in some cases such concerns may be unfounded, in others they may have some basis in fact. Finally, as evidenced by Northern Rock, if the liquidity support is extensive, the need to repay the loan can be a major impediment to the institution remaining in the hands of the private sector.

Despite these considerable difficulties, such support might be justified in some circumstances. This is particularly the case if the troubles reflect the breakdown of markets and an extreme increase in risk aversion. If an institution clearly has significant positive net asset value, yet cannot fund its liabilities because of severe dislocation in markets, the central bank can play a stabilising role, preventing a fire sale of assets and perhaps a loss of confidence in the system as a whole.

Under some scenarios, there is likely to be a connection between the degree of flexibility in the central bank's market operations and the probability that a troubled institution will need to seek emergency liquidity support. In particular, the more flexible are market operations – in terms of frequency, volumes, maturities and acceptable assets – the more likely it is that an institution with assets eligible for repos will be able to exchange those assets for liquidity in the course of normal market operations when the need arises. Indeed, an argument for flexibility in regular market operations is that it can avoid the non-linear effects – partly due to adverse effects on public confidence – that can arise when an institution is known to have sought support.

Flexibility in market operations is, however, not without risks. In particular, if the liquidity problems reflect the poor health of the institution, which is seen by the other market participants, then it is possible that flexible market operations might allow the institution to delay the action required to correct its problems, thereby increasing losses if the institution does ultimately fail. This possibility means that in times of strain, close cooperation is required between the central bank and the prudential supervisor.

Finally, given the flexibility that many central banks now have in their market operations, it is highly likely that an institution requiring an emergency loan will have very serious balance sheet problems. Hence, an emergency loan is perhaps best thought of as a bridging loan while new ownership is arranged, or the institution is fundamentally restructured. LOLR might then be thought to stand for ‘lender of last rights’. In today's world it seems unlikely (although not impossible) that an institution would be granted emergency assistance for a short period of time, repay that loan, and then continue as normal. To the extent that emergency assistance is really bridging finance, there is a strong case for it to be accompanied by a credible plan for private-sector support or recapitalisation (Bear Stearns), or some form of government support or recapitalisation (Northern Rock). In either case, the management and shareholders would be expected to incur very significant losses.

6.4 Assisting Off-market Transfers

A fourth way in which liquidity issues could be addressed is through assisting with the off-market transfer of assets.

As noted in Section 3, the failure of a financial firm with extensive activities in financial markets raises concerns, not just because of the direct counterparty exposures, but also because of the potential cascading effects through financial markets. The fear is that many markets are simply not deep enough to deal with the rapid closing-out of positions and the flow-on effects from margin calls that would likely follow a failure. In the event that an institution with extensive operations in markets was forced into liquidation, the potential flow-on effects could undermine the stability of the financial system.

While these concerns are widely held and appear to be soundly based, it is important to note that this scenario has never played out in practice, with no major participant in financial markets having been forced into liquidation. This lack of experience makes it difficult to assess exactly what might happen in the event of such a failure. Notwithstanding this, a reasonable question is how policy-makers should respond to this possibility (over and above providing general liquidity to the market and ensuring that the overall regulatory framework is sound).

It can be argued that these distressed situations are best dealt with by a measured selling-down of positions, rather than an immediate sale in turbulent conditions where information asymmetries are likely to be acute. In some situations such an outcome might be able to be organised by the private sector, either by a single institution, or group of institutions, purchasing the positions off market, at a substantial discount. The public sector may be able to play a useful role here, particularly if coordination issues among the troubled institution's counterparties prevent an effective solution that is in their collective interest.

A more difficult problem emerges if a private buyer cannot be found quickly. One option here would be for the public sector to purchase the assets/positions and then sell them over time when conditions are more settled; the Fed's approach to Bear Stearns can be seen in this light. The argument for such an approach is that it might avoid a fire sale of financial assets that could prejudice the stability of the overall financial system. Furthermore, provided that the assets/positions are bought at a substantial discount to current value, the purchase may deliver a favourable risk-adjusted return to the public sector.

Such actions are, however, not without considerable risks. Not only is there the obvious risk that the assets may ultimately be worth less than the price that the public sector paid, but the possibility of such actions may change the behaviour of the private sector. In addition, when decisions have to be made very quickly, a type of game can develop between the public and private sectors, particularly if the private sector believes that the public sector will go to considerable lengths to protect the stability of the financial system. This game may lead to the public sector paying more for the assets than is desirable.

These are difficult issues to resolve, but as financial markets continue to grow, ways need to be found to allow large participants in these markets to exit without causing instability in the rest of the system. As discussed earlier, improving the financial infrastructure can be helpful here, but mechanisms also need to be found to prevent the fire sale of financial assets and limit the build-up of problems in the first place.


See Borio and Nelson (2008) and CGFS (2008a) for a discussion of recent changes in central bank operations, Debelle (2008) for more detail on Australia, and Hilton (2008) for more detail on the United States. [10]

In both the United Kingdom and Australia, interest is paid on balances at the central bank. In the United Kingdom, it is paid at the policy rate for reserves within the threshold around the reserves target. In Australia, the interest rate paid is 25 basis points below the target cash rate. [11]

The longest single maturity has been 365 days. [12]

For a discussion of the RBA's intervention in the foreign exchange market see Becker and Sinclair (2004). [13]

See Joye and Gans (2008) and Australian Securitisation Forum (2008). [14]

For a history of ‘emergency liquidity assistance’ in Australia see Fitz-Gibbon and Gizycki (2001). For a more recent discussion on the lender of last resort see Stevens (2008). [15]