Unconventional monetary policies in time of crisis

SUERF Annual Lecture delivered by Mr Jaime Caruana, General Manager of the BIS, Brussels, 16 November 2009.


Unconventional monetary policy reshapes the balance sheet of the central bank as a result of efforts to make bank funding and other money markets function better, to restore credit flows, to reduce credit or term spreads or to boost bank reserves. These efforts are distinct from efforts to target a short-term policy rate. Implementing and exiting from such policies poses a number of practical challenges. While central banks are likely to retain some features of recent innovations in their operating procedures associated with unconventional monetary policy, they are likely to use such policies only in exceptional circumstances.

Full speech

1. Introduction

It is a great honour for me to deliver the SUERF Annual Lecture this year, following in the footsteps of such prestigious speakers.

Thankfully, this year has been a bit less eventful than the previous one. These calmer times have allowed deeper reflections among policymakers and academics about a number of fundamental issues, including the appropriate framework for monetary policy.

A key question that has re-emerged is whether it is sufficient for central banks to focus on price stability. Given that the current crisis took place against a backdrop of subdued inflation and well anchored inflation expectations, the answer appears to be "no". And if price stability is not sufficient to ensure financial stability, it is not enough to deliver economic stability either.

This leads to another set of questions. Should central banks better integrate concerns about financial imbalances into policy? At what point do credit growth and asset price booms become excessive and warrant policy action? What additional tools would help central banks in dealing with these developments? Would an explicit financial stability mandate help, particularly in managing the political economy pressures? These are open questions that will be hotly debated.

In light of the theme of this year's conference, however, I would like to concentrate my remarks today on the broad range of responses that central banks have implemented to deal with the current crisis. These have been referred to as unconventional monetary policy, and I have three points to make. First, I will outline how unconventional policies can be viewed as a crisis management tool. Second, I will argue that more attention should be given to the asset side of the central bank balance sheet than the liability side in discussions of unconventional monetary policies. I will question the importance of bank reserves and their relationship to bank lending and inflation. Finally, I will highlight some key practical challenges in implementing such policies, including exiting from them. One conclusion that follows from this discussion is that unconventional monetary policies appear more suited for exceptional circumstances and are unlikely to represent an additional set of tools that central banks can use more generally in their normal day-to-day conduct of policy.

2. Unconventional monetary policies and crisis management

Let me begin by defining unconventional monetary policies as the elevation of liquidity management operations from a passive role in the background, undertaken simply to ensure the attainment of the interest rate target in normal times, to an active role to influence broader financial conditions. Given this definition, I would like to offer some thoughts on unconventional monetary policy from the point of view of crisis management. In particular, I wish to highlight two perspectives from which unconventional monetary policy, as a crisis management tool, can be viewed. From the first perspective, such policies complement the central bank's role as lender of last resort; from the second, they become an extension of monetary policy. Let me discuss each of these in turn.

Apart from conducting monetary policy, a vital responsibility of central banks is to act as lender of last resort. The core objective of this function is to prevent, or at least mitigate, financial instability through the provision of liquidity support either to individual financial institutions or to financial markets.

Traditionally, the lender of last resort function is associated with acute institution-specific shortages of funding liquidity. By funding liquidity, I mean the ability to raise cash or its equivalent in reasonably large quantities, either via asset sales or by borrowing. Typically in such instances, an institution finds itself unable to pay or roll over obligations. Given the institution-specific nature of the intervention, such emergency liquidity assistance can generally be clearly separated from setting the policy interest rate.

In other cases, the situation confronting the central bank is something that can be termed a systemic shortage of both funding and market liquidity. By market liquidity I mean the ability to buy and sell assets in reasonably large quantities and at short notice without significantly affecting their price. Here, the problem involves a breakdown of key financial markets owing to a loss of confidence and coordination failures among market participants. As starkly demonstrated by the current crisis, markets, just like intermediaries, may be subject to "runs". And these runs are driven by fundamentally similar forces. The result is a sudden and prolonged evaporation of both market and funding liquidity, with serious consequences for the stability of both the financial system and the real economy.

From a financial stability perspective, unconventional monetary policy measures can be seen as a lender of last resort response to this second type of crisis. The underlying aim of intervention is to support market functioning by restoring both funding and market liquidity and thereby to shore up confidence in the financial system as a whole. Typically, this will require a broadening of the central bank's provision of liquidity, in terms of both accessibility and structure. From such a viewpoint, targeted interventions in specific market segments are primarily geared to improving market functioning. And while they may exert a beneficial influence on broader economic conditions, such an effect is not viewed as the main objective.

Nonetheless, precisely because these actions typically affect overall financial conditions, it can be difficult to distinguish them from the stance of monetary policy per se. This leads me to the alternative perspective from which unconventional monetary policy can be viewed: namely, as an extension of monetary policy that can be used when interest rate policy alone may not achieve the desired policy objective, perhaps because particular segments of the transmission mechanism fail to work or because of the zero lower bound. Here, central bank operations are aimed at directly affecting broader financial conditions, such as asset prices, yields and funding conditions, over and above the impact of the policy rate.

While the lender of last resort and the monetary policy perspectives are usually distinct from one another, with the former focused on financial stability and the latter on macroeconomic stability, they can become closely intertwined in a crisis. Ensuring continued market functioning as a lender of last resort generally entails interventions that reduce liquidity premia on certain assets. To the extent that the reduction in risk premia translates into easier funding conditions, this adds monetary stimulus. Conversely, insofar as concerted monetary policy interventions to lower risk spreads and ease funding conditions serve to bolster market confidence, they may improve market functioning.

The current episode can be viewed from both perspectives. When the crisis first erupted in August 2007, central bank interventions focused on maintaining liquidity in key markets primarily by supplying central bank liquidity and government securities more flexibly. In this phase, the lender of last resort perspective was clearly dominant. It was reflected in the introduction of various emergence liquidity facilities such as the Term Auction Facility by the Federal Reserve and the Special Liquidity Scheme by the Bank of England.

As the crisis deepened following the failure of Lehman Brothers and spillovers to the real economy intensified, the monetary policy perspective became more important. Interventions were undertaken with the explicit aim of steering broader financial conditions to support central banks' macroeconomic goals. Prominent examples include purchases of government bonds to lower benchmark yields and purchases of mortgage-backed securities to lower mortgage rates.

The defining element that is common to both perspectives is that they involve operations that result in substantial changes in central bank balance sheets - in terms of size, composition and risk profile. On the asset side, the extension of term funding to banks, the purchase of short-term claims on businesses and the purchase of mortgage and government bonds have been termed "credit easing", to highlight the intention to maintain the supply of private credit at reasonable cost. On the liability side, "quantitative easing" refers to policies that emphasise the supply of bank reserves.

To begin with, let me note two important features of such "balance sheet policies".

First, balance sheet policy is not that new or unconventional in its essence. The most familiar form is foreign exchange intervention, whereby the central bank seeks to influence the exchange rate separately from the policy rate. What makes its use in the current crisis novel is the market segments targeted: for example, the long end of the interbank market, long-term government bond yields, private sector risk spreads and the like. The recognition that such interventions do not represent something entirely new facilitates their assessment. Indeed, bearing in mind the parallels with foreign exchange rate intervention helps to provide useful clues about the efficacy and limitations of this broad approach to policy.

The second key feature of balance sheet policies is that they can be decoupled from the level of policy rates. Technically, all that is needed is for the central bank to neutralise the impact that any induced expansion of bank reserves might have on the overnight interest rate.

Let me give an example. Suppose the central bank purchases an asset outright from commercial banks. In the first instance, it pays for this by crediting banks' deposits at the central bank. That is, it creates bank reserves. Now, if the rate of remuneration that the central bank sets on bank reserves is below the market rate, as is typically the case, their expansion will lead to downward pressure on overnight interest rates. This follows because the opportunity cost of holding reserves means that banks will try to lend them out in the interbank market, and in so doing depress the overnight rate. Thus, one way of shielding the overnight rate from the effects of asset purchases is for the central bank to conduct offsetting or sterilising operations, so as to leave the amount of reserves unchanged. There are many ways of doing this, including asset sales, repos, or the issuance of central bank bills. And as clearly demonstrated by many Asian central banks, the scope for this approach is quite large.

Alternatively, if the central bank does not wish to offset the expansion in reserves, perhaps because of limitations in the availability of offsetting instruments, it can still shield overnight rates by paying interest on reserves at the policy rate. This eliminates the opportunity cost of holding reserves, making them, in effect, a close substitute for other short-term liquid assets in banks' portfolios. This is essentially the approach that the Federal Reserve and the Bank of England have followed. Of course, the opportunity cost is also eliminated automatically even if reserves are not remunerated when the policy rate reaches or comes very close to the zero lower bound.

Thus, so long as central banks have sufficient instruments, the size and structure of their balance sheets can be managed separately from the policy rate. One implication of this "decoupling principle" is that exiting from the current very low, or zero, interest rate policies can, at least in principle, be done independently of balance sheet policies. In practice, however, the distinction is unlikely to be as clear cut, especially insofar as the impact on overall financial conditions is concerned. I will return to these issues later. But before I do so, I would like to address the effectiveness of unconventional policies and its relationship to the substantial increases in bank reserves that have taken place.

3. Assessing the effectiveness of unconventional policies

In principle, the effects of balance sheet policy may be transmitted through two main channels. The first is the "signalling channel", whereby central bank actions or their communication influence public expectations about some of the key factors that underpin the market valuation of an asset. These include expectations regarding the future course of policy, inflation, the relative scarcity of different assets or their risk and liquidity profiles. For example, the announcement that the central bank is prepared to engage in operations involving illiquid assets may, by itself, boost investor confidence in them, thereby reducing liquidity premia, stimulating trading activity and improving market functioning.

The second channel of influence is commonly known as the "portfolio balance channel". Here, imperfect substitutability among assets leads to changes in relative yields when central bank operations alter the composition of private sector portfolios. Insofar as shifts in private sector portfolios lead to stronger balance sheets, greater collateral values and higher net worth, they may also help loosen credit constraints, lower external finance premia, and hence boost credit growth. For example, by purchasing risky private securities from banks in exchange for risk-free claims on the public sector, the resultant improvement in the overall risk profile of bank balance sheets may not only enhance their risk appetite but may also increase investors' willingness to lend to them.

The effectiveness of credit easing policies can be seen in credit spreads. Central bank lending and purchases narrowed the spread of term bank funding over expected monetary policy rates, and the spread of mortgage bond over government bond yields. Whether the purchase of government bonds by central banks has had a similarly sustained effect on government bond yields will be debated. In the case of the relatively largest programme of purchases, that of the Bank of England, bond yields responded to surprises in the series of announcements about the initiation and expansion of purchases.

Turning to the liability side, while the central bank has a number of choices in how such operations are funded, a prominent one is to expand bank reserves. Two aspects of the role of bank reserves deserve to be reconsidered. The first is the relationship between reserves and bank lending; the second is the link between reserves and inflation.

Starting with the former, discussions of balance sheet policies often presume a close link between the expansion of reserves and credit creation. The implicit premise is that excess bank reserves induce banks to make loans. Either bank lending is constrained by insufficient access to reserves or more reserves can somehow boost banks' willingness to lend. An extreme version of this view is the notion of a stable money multiplier.

In fact, bank lending is determined by banks' willingness to grant loans, based on perceived risk-return trade-offs, and by the demand for those loans. An expansion of reserves over and above the level demanded for precautionary purposes, and/or to satisfy any reserve requirement, need not give banks more resources to expand lending. Financing the change in the asset side of the central bank balance sheet through reserves rather than some other short-term instrument like central bank or Treasury bills only alters the composition of the liquid assets of the banking system. As noted, the two are very close substitutes. As a result, the impact of variations in this composition on bank behaviour may not be substantial.

This can be seen another way. Recall that in order to finance balance sheet policy through an expansion of reserves the central bank has to eliminate the opportunity cost of holding them. In other words, it must either pay interest on reserves at the positive overnight rate that it wishes to target, or the overnight rate must fall to the deposit facility floor (or zero). In effect, the central bank has to make bank reserves sufficiently attractive compared with other liquid assets. This makes them almost perfect substitutes, in particular for other short-term government paper. Reserves become just another type of liquid asset among many. And because they earn the market return, reserves represent resources that are no more idle than holdings of Treasury bills.

To be clear: this is not to say that central banks are powerless to influence bank lending. If lending is held back by significant funding constraints - because banks are unable to sell illiquid assets or to borrow - interventions that alleviate these constraints will encourage lending. Thus, for example, if banks' access to future funding becomes highly uncertain, central bank operations that supply term funding may allow banks to keep lending. Bank lending may also be encouraged by the financing of such operations with excess reserves or short-term paper that satisfy a demand to hold larger precautionary liquid balances. But the underlying mechanism involves supplying banks with a liquid asset at a time when the access to funding is difficult or becomes uncertain. Reserves simply constitute one possible asset among others that can serve this purpose. Whether a bank holds liquid assets in the form of, say, reserves, one-week Treasury bills or one-month central bank bills will not make a material difference to its willingness and ability to lend. Typically, the main constraint on credit creation, if the demand for credit is there, is bank capital relative to regulatory minimum or market requirements.

What about the concern that large expansions in bank reserves will lead to inflation - the second issue? No doubt more accommodative financial conditions resulting from central bank lending and asset purchases, insofar as they stimulate aggregate demand, can generate inflationary pressures. But the point I would like to make here is that there is no additional inflationary effect coming from an increase in reserves per se. When bank reserves are expanded as part of balance sheet policies, they should be viewed as simply another form of liquid asset that is comparable to short-term government paper. Thus funding balance sheet policies with reserves should be no more inflationary than, for instance, the issuance of short-term central bank bills.

This also suggests that the justification for inflationary fears associated with the notion of "debt monetisation" needs to be qualified. Here, the concern is that purchases of government debt and the associated expansion in bank reserves would lead to inflation. In addressing this issue, it is essential to distinguish the effects that operate through interest rate policy and those that operate through the financing structure of government debt.

If excess reserves are remunerated at a below market rate, their injection would push overnight rates down to the floor established by the remuneration rate (or the deposit facility rate). This is tantamount to an easing of interest rate policy. Any ensuing inflationary pressure can hence be largely attributed to the usual expansion of aggregate demand that accompanies such a move.

In the case where the opportunity cost of reserves has been eliminated, such as by paying interest at the policy rate, their expansion would not affect overnight rates. To the extent that any additional impact on inflation existed, it would result mainly from the effect on aggregate demand of the flatter yield curve that these operations may induce. For example, if the central bank were to inject reserves through the acquisition of long-term government bonds, the net impact on yields and inflation would not be dissimilar to the rebalancing of government financing from long to very short maturities. In fact, such an "operation twist" can be achieved by the fiscal authorities themselves through altered debt management.

Ultimately, any inflationary concerns associated with monetisation should be mainly attributed to the monetary authorities' accommodating fiscal deficits by refraining from raising rates. In other words, it is not so much the financing of government spending per se - be it in the form of bank reserves or short-term sovereign paper - that is inflationary, but its accommodation at inappropriately low interest rates for too long a time. Critically, these two aspects are generally lumped together in policy debates because the prevailing paradigm has failed to distinguish changes in interest rate from changes in the amount of bank reserves in the system. One is seen as the dual of the other: more reserves imply lower interest rates. As I explained earlier, this is not the case. While both the central bank's balance sheet size and the level of reserves will reflect an accommodating policy, neither serves as a summary measure of the stance of policy.

To recap, the focus in assessing the impact on bank lending and inflation should be on how assets taken on by the central bank affect relative yields, and hence aggregate demand, or how they affect market liquidity and access to credit. Balance sheet policies work primarily by changing the composition of private sector balance sheets. Their impact will be greatest when the assets exchanged are imperfect substitutes for each other. Invariably, in such an exchange, the central bank will be providing the private sector with some form of highly liquid, low-risk asset. Such liquid assets tend to be highly substitutable for one another, especially at very low interest rates. Therefore, the specific form chosen, as determined by the central bank's method of funding, will generally be of much less significance than the choice of asset that has been acquired.

4. Practical challenges in implementing balance sheet policy

Let me now move to my next point and highlight some important practical challenges that central banks face in implementing balance sheet policy.

The first challenge is calibrating and communicating the interventions effectively. With little previous experience, with the relevant transmission channels unclear, and in the absence of a shared framework to quantify the various effects, it is very hard to judge the impact of these unconventional policies and hence determine the appropriate amount of intervention. At the same time, central banks have to tread a fine line between acting as a catalyst for private sector activity, on the one hand, and substituting for it, on the other. Moreover, they have to be wary of potential distortions to the level playing field between those receiving and not receiving the support. Finally, they need to take into account what is done in other jurisdictions. Coordination with other central banks can enhance the effectiveness of unconventional policy measures.

Even when policy can be appropriately calibrated, its impact and effectiveness are influenced heavily by how it is communicated. With liquidity management operations being used to affect monetary conditions directly, the official policy stance is no longer summarised by the policy rate. The resulting multidimensionality of policy carries with it the potential for diminished clarity of the policy signal. Communicating the rationale, nature, magnitude and time dimension of unconventional policies can steer expectations effectively, can avoid central bank credibility problems and can mitigate financial market volatility. Indeed, central banks have taken care to explain their unconventional policies with very welcome results.

The second set of practical challenges lies in the effective management of the central bank's relationship with fiscal policy. Balance sheet policy has a large potential overlap with fiscal policy. The clearest example is when central bank purchases of long-term bonds aimed at lowering their yield are counterbalanced by actions of the government's debt management to lock in low yields by issuing more long-term bonds. This hints at a broader point. In principle, almost any balance sheet policy can be undertaken by the government. While the central bank has a monopoly over interest rate policy, the same cannot be said with respect to balance sheet policy.

Moreover, balance sheet policy exposes the central bank to financial risk. Should substantial losses materialise, the central bank's operational autonomy may be threatened in the absence of an explicit or implicit understanding with the fiscal authorities regarding how losses are dealt with. Up to a point, some of the financial risks can be managed, for example, through the restriction of eligible collateral and the use of conservative haircuts. In the end, however, financial risks are part and parcel of balance sheet policy. Therefore, the real issue is how far institutional factors related to the treatment of losses may constrain the willingness and ability of the central bank to engage in such policies.

In this context, perhaps the greatest challenge to sustained utilisation of such policies is of a political economy nature. The more the central bank relies on unconventional policies, the more tricky questions are raised about coordination, operational independence and the division of responsibilities. A case can thus be made for the establishment of clear institutional guidelines to resolve potential conflicts and to enhance clarity in areas where central bank actions may have a large overlap with the fiscal authority, and thereby to preserve the autonomy of monetary policy These include accounting arrangements, rules for the distribution of profits and losses, and also mandates for the scope of actions.

This brings me to the issue of exit strategies.

Since, as I argued earlier, interest rate and balance sheet policies can be decoupled from each other, it is then possible, in principle, to delineate discussions of exit strategies along two separate dimensions: the appropriate level of interest rates, on the one hand, and the desired central bank balance sheet structure, on the other. The former will most likely be dictated by traditional output-inflation considerations; the latter will also be influenced by considerations about market functioning and avoiding financial market stress. In practice, however, this separation may not be so obvious. Since balance sheet policies exert an impact on broader financial conditions, in terms of the overall macroeconomic implications their withdrawal will not be easily distinguishable from a tightening of interest rates. This is part of the broader communication challenges of exit to which I will come back in a minute.

The main challenge is how to properly judge the timing and pace of the exit. This concern is a familiar one, being largely the same as that which applies to interest rate policy. One possibility is that exit occurs too early, hampering an incipient recovery. However, historical experience suggests that the balance of risks is tilted towards exiting too late and too slowly. Political economy pressures tend to go in this direction. At the macro level, the concern is that such a delayed exit may risk accommodating the build-up of a new set of financial imbalances or else lead to inflationary pressures. At the micro level, it may weaken unnecessarily the ability of markets to work effectively without official support and may distort the level playing field.

I should also stress here that, while the principles of exiting from these policies may be apparent, the actual path of exit could prove to be challenging and potentially bumpy. For one, communicating exit can be extremely tricky; there may be knife-edge market reactions to news of withdrawal. Moreover, because considerable uncertainty exists regarding the transmission channels of balance sheet policy, there is a risk that central bank actions will be misinterpreted. For example, a technical liability management operation such as issuing central bank bills to drain bank reserves may be misinterpreted as a tightening of monetary conditions.

The number of potential pitfalls suggests that we should by no means take the scenario of a smooth exit for granted, and here again efforts by central banks to explain are constructive and welcome. No matter how much market conditions have seemingly improved, it is not entirely clear to what extent those improvements are based on the policies in place.

5. Closing remarks

To conclude, central banks' active management of the size and composition of their balance sheets does represent an additional tool to help ease constraints stemming from the zero lower bound and to manage crises. In the current one, it has clearly helped to ease severe liquidity strains and support the rebound in a number of key markets.

Notwithstanding these positive developments, this policy tool is best suited to restoring market functioning and bringing about more accommodative financial conditions. Under current circumstances, it is no substitute for the required fundamental restructuring of private sector balance sheets. In an environment of pervasive uncertainty regarding financial institutions' balance sheets, central bank actions for the most part only ease the problem, alleviating the symptoms rather than addressing the underlying causes. That is not to say that they do not contribute to the balance sheet repair. Indeed, the improvements in asset prices and the boost to bank profitability that have accompanied these policies have certainly helped to shore up balance sheets. Despite this, they cannot replace the forceful implementation of measures that address directly the fundamental weaknesses in private sector balance sheets or the need for better business models.

More generally, sustained reliance on a highly accommodative policy stance with respect to both interest rate and balance sheet policies risks creating a perception that the central bank alone is responsible for generating economic recovery. This could reduce the incentive for market participants and governments to take more fundamental measures. Also, I do not believe that we fully understand what the repercussions would be for asset prices, commodity prices and the global financial system as a whole if the world's major central banks keep policy interest rates very low for an extended period. If recent experience is any guide, we must pay serious attention to the risks that may arise. This is especially so for countries that are not suffering from some kind of economic headwinds, and some that are even benefiting from terms-of-trade gains and resurgent capital inflows.

Finally, there is the question is whether balance sheet policies represent an additional set of tools that can be used not just in crisis management but also in normal times. My own feeling is that the formidable practical challenges and the intense political pressure that inevitably accompany their use suggest that they should be employed only in exceptional circumstances and be withdrawn as soon as economic conditions permit. That said, it will be useful to reflect back and learn how some of these tools can be better designed and deployed in the future. A related question is how central bank operational frameworks should be designed to embed market-stabilising features more systematically and to improve flexibility in response to shocks. For instance, it may be the case that operational frameworks will retain their greater flexibility with regard to collateral requirements, counterparty eligibility and maturity of operations.

We still have a lot to learn from the crisis. Forums such as this one are an essential part of the learning process. It has been a pleasure for me to be here and I thank for your attention.