The new art of central banking.

Chadha, Jagjit S.

This article outlines some of the intellectual lessons learnt by central bankers during the financial crisis. The key question is whether a broader range of policy options than simple inflation targeting has to be considered in order to limit instability. Interactions with overseas pools of savings, government debt markets and financial risk have all conspired to complicate significantly the task of monetary policymaking. These developments do not mean that the target for inflation has to be modified or dropped but that setting policy will be a more complex task and require more explanation than it has in the recent past.

Keywords: monetary policy; quantitative easing; macro-prudential instruments

JEL Classifications: E52; E58; N1

I. Confronting risks

Both our notion of money and of monetary policymaking has continued to evolve significantly. The original function of money was to facilitate trade with a standardised unit of account. A monetary policy would then have implied simply some arrangement of institutional practice so that the amount of commodity-based money was appropriate for the level of trade. It is probably the case, as is still the case in many parts of the world, that large amounts of trade stood outside the monetary system and relied on barter or non-pecuniary grace and favours. Even standardisation was and remains no easy matter as it is no simple task to set the correct relative prices between various types of monies and goods, ensure the absence of counterfeiting or clipping and decide on the right amount of money in circulation. We continue to debate whether notes should be limited in denomination and whether monetary policy will have to be re-thought should all money become electronic.

The experiences of the late 18th and 19th centuries involved both a recognition that the rate of exchange between currency and gold might be changed in the event of crises but also that the banking and financial system required regular bouts of support because of an inherent fragility. The guiding principles were framed by Bagehot and the evolution of monetary orthodoxy, or sound money, was evident. Monetary orthodoxy here involved adherence as well to some form of the Real Bills Doctrine, involving the discounting of all commerical bills presented at the Bank of England, and an understanding that the Bank would not buy government debt. This orthodoxy also suggested some adherence to low levels of public debt in peacetime, a gold standard and circumspect choices in Bank Rate. The suspension of the Gold Standard associated with World War I, the interwar boom and bust and the Great Depression provided an incentive and a 'Keynes-inspired' blueprint for the operation of counter-cyclical monetary and fiscal policy. Whilst it is not clear whether these policies were responsible for the economic recovery prior to World War II, it is clear that there had been a profound change in the responsibilities of government. Hawtrey (1932) had summed up the interwar mood well: "[T]he result has been not merely that the world has been insufficiently prepared to deal with the new problems of Central Banking which have arisen in the years since the War, but that it has failed even to attain the standard of wisdom and foresight that prevailed in the nineteenth century. Moreover, they should endeavour to adapt their measures of credit regulation, as far as their domestic position permits, to any tendency towards an undue change in the state of general business activity. An expansion of general business activity of a kind which clearly cannot be permanently maintained should lead Central Banks to introduce a bias towards credit restriction into the credit policy which they think fit to adopt, having regard to internal conditions in their own countries. On the other hand, an undue decline in general business activity in the world at large should lead them to introduce a bias towards relaxation. In pursuing such a policy the Central Banks will have done what is in their power to reduce fluctuations in business activity ... ". (1) From that point on, the rate of inflation and economic growth would continue as the government's problem and an important backdrop to the assessment of the performance of political leadership--which at some level is rather odd given that the dominant models of economic fluctuations do not predict a permanent impact on output from monetary policy. And as figure 1 shows, monetary policy, as tracked by Bank Rate, seems to have become more active over time both in the level and the frequency of its changes. Note also the close relationship between Bank Rate and a long-term interest rate, in this case, on UK bonds where the latter looks very much like a moving average of the former.

[FIGURE 1 OMITTED]

Accordingly, in the postwar period, there was an incredible intellectual effort to understand not only the mechanical interplay between monetary policy and the real economy, via its impact on market interest rates and asset prices, but also how monetary policy effectiveness was a function of its interplay with private agents' plans and expectations. The elegant models developed allowed the study of optimal monetary policy and the development of strategies to minimise inefficient fluctuations in output, particularly in the aftermath of the end of Bretton Woods and the subsequent costly inflation and disinflation. The great mirage of the Long Expansion was that whilst it appeared that business cycle risks had been eliminated they were, in fact, increasing rapidly (Chadha et al, 2016) Once the risks had become apparent, the economy quickly jumped to a world of profound financial constraints that acted to bear down on activity in a persistent manner. Interest rates hit the zero lower bound and public debt was stoked up to precarious levels by the historical standards of peacetime.

Practically speaking two issues were then exposed, which have occupied much of the debate on the setting of monetary policy. How should policymakers deal with a response to large, negative economic shocks that seemed to threaten to exhaust policy ammunition? What kind of defences should be put up so that such shocks could not build up in the same way, or that stocks of policy ammunition would still be available? The former problem led to the re-discovery of open market operations as a way of influencing longer-term interest rates, (2) and the latter problem led to the re-discovery of tools that act to constrain financial intermediation. But there is another issue that has become more relevant. As well as thinking in terms of normal times--with small changes from the steady-state--and abnormal times, with low growth and extraordinary policies, it is becoming increasingly clear that there is a transitional state to work through. Debt does not disappear, default notwithstanding, overnight and so balance sheet repair is a tricky and time consuming business. Table 1 shows the average contribution to the fall in public debt in the US and the UK after WWII. Both countries were able to engineer large average falls annually of over 4.5 per cent and 7.6 per cent respectively. In the UK low interest rates of 2 per cent played an important role in ensuring that the primary deficit did not have to be any higher, allowing nominal GDP growth to bring down the deficit. (3) So public debt takes time to get back to pre-crisis levels but, at the same time, policy has to deal with nursing a sick economy rather than licking a healthy one into shape.

I will examine in section 2 the main policy lever used in the financial crisis, quantitative easing. We shall go on to examine the role for macro-prudential instruments. The case for considering the policy nexus as some point in monetary-financial-fiscal space is still being explored but in this triplet lies a generalised way of thinking about policy and its transmission. Finally I conclude.

2. Quantitative easing

When policy rates hit their effective zero lower bound central banks began to consider how they might influence market interest rates that typically followed movements in Bank Rate quite closely. These market interest rates form part of what central bankers call the monetary transmission mechanism and typically include short-term money market rates that forecast Bank Rate for one to three months, as well as government bonds of varying maturities. As we have already seen, the long-term bond rate is essentially a moving average of current and expected short-term Bank Rate and so if market participants could be persuaded that policy rates would stay at low levels for a long time then these long-term rates would fall and so provide some succour to an economy in financial crisis. One way to persuade market participants is to do a lot of talking that central banks will "do whatever it takes" to stabilise the economy, which can be an effective strategy if credible, as the Governor of the ECB discovered in the summer of 2012. When the economy hits a nadir something sometimes needs to be done and more importantly may need to be seen to be done.

We can calculate a model of the term structure in which a key determinant of the long-term rate is the speed at which market participants believe that the rate will return to steady state; it is possible to knock some 250 basis points off the 10-year rate simply by convincing market participants that the half-life of the rate of return slows from five to ten years. (4) One metric of the problem in 2008/9 was the large fall in nominal GDP growth below Bank Rate. The former combines real GDP growth and inflation and the latter aggregates the real rate and short-run inflation expectations, with both rates in normal times basically expected to be something like 4-5 per cent. It is the large triangle below the Policy Rate in 2008 and 2009 which creates the motivation for Quantitative Easing (QE), or monetary policy by other avenues.

[FIGURE 2 OMITTED]

The signalling game can go quite far but central banks found that actual purchases of assets were required, which lowered long-term interest rates by one of two mechanisms. The first is that following a large increase in fiscal deficits, which acts to stimulate demand, market clearing in the money market will require higher interest rates in order to allow for a substitution from speculative to transaction demand for money and this increase in rates will tend to reduce the impact of the fiscal expansion. If, on the other hand, money underpinning transactions is expanded by some type of QE, interest rates need not rise and the full force of the stabilising effort will reach the economy. The second is more subtle and relies on the argument that in buying bonds, which are high in price at a low interest rate, the central bank is signalling that it will not raise interest rates which would lead to an immediate lowering of bond prices and, perhaps a loss on the trading operation. (5) Personally, I put more weight on the former as central banks seem more likely than not to make money from these trades because they were providing market participants with much-needed liquidity in the form of central bank money in exchange for less liquid bonds. So following the financial crisis of 2008, QE--which I define as large-scale purchases of financial assets in return for central bank reserves--became a key element of monetary policy for a number of major central banks whose interest rates were at, or close to, the zero lower bound. But despite its widespread use, the question of the effectiveness of QE remains highly controversial.

QE as an open market operation

Generally speaking, quantitative easing is really just an extended open market operation involving the unsterilised swap of central bank money for privately held assets. Let me break up these ideas. An open market operation involves the swap of central bank money (a liability on its balance sheet) for the purchase of an asset. If the operation is sterilised, the resulting increase in central bank money is mopped up by the sale of some other asset on the central bank balance sheet, which means that the central bank money is returned to the central bank, so that the overall impact is a change in the type of asset held. If, on the other hand, the supply of the new central bank money is not immediately mopped up by central bank asset sales, then the market operation is said to be unsterilised and thus adds to the quantity of money in the economy, until it is reversed.

The key innovation that QE represents, as separate from a short-term open market operation, is that the duration of the swap is both intended to be long-term and of uncertain length. As explained, an open market operation, if unsterilised, leads to an increase in the quantity of base or outside money. This money represents claims on the public sector and will not be neutral with respect to any given expenditure plan if there is a real balance effect that induces a fall in interest rates. This is because the increase in money changes the wealth position. If, however, the private sector fully discounts the present value of taxes that will need to be paid to meet these obligations, then these bonds will not represent net wealth and the operation will be neutral. The debate on the efficacy of such operations hinges on the question of whether the supply of outside money changed the wealth position of the private sector (see Gale, 1982).

The analysis of such operations lies outside the remit of the workhorse New Keynesian (NK) Model in which the evolution of monetary aggregates, which were simply a veil by which real planned transactions were effected, provided no additional feedback to the state of the economy. These models are highly tractable and were used to develop simple, precise policy prescriptions, even at the zero lower bound of Bank Rate, by influencing expectations of the duration of any given level of Bank Rate in order to induce exchange rate depreciations or positive inflation shocks and so close any given sequence of output gaps in expectation. In these models, open market operations were neutral because at the zero lower bound money and bonds become perfect substitutes and any swap of one for the other does not change the wealth position of the private sector. In fact, in these models QE-type policies are simply forms of commitment strategies that provide signals about the long-term intentions of the central bank to hit a given inflation target.

The NK argument that monetary policy can only work through the management of expectations is not a universal result as it relies on particular assumptions. In these models, financial markets are complete in which a representative agent can spring into life and financial wealth is allocated over an infinite life. Idiosyncratic risk in these economies can be hedged and asset prices depend on state-contingent payoffs. In this case, the prices of financial assets are not influenced by changes in their net supply, as demand is perfectly elastic. It seems quite possible though that demand curves for assets, particularly those which are issued in large quantities, may become downward sloping, in which case changes in net supply can affect their relative prices. This possibility then means that the relative supply of money or credit can influence market interest rates and so impact directly on expenditure paths without having to rely on pure signalling effects. It is this possibility, see figure 3, which gives QE its additional route of influence, particularly in abnormal times.

[FIGURE 3 OMITTED]

Effectiveness

Early work on the impact of large-scale asset purchases as a tool of monetary policy probably began following 'Operation Twist' in the United States in 1961. Although not full QE in the sense of being financed by base money creation, this operation involved Federal Reserve purchases of long-term bonds (financed by sales of short-term Treasury Bills) as well as a change in Treasury issuance with the aim of lowering long-term interest rates. Modigliani and Sutch (1966) found that this operation had no substantive effect on bond yields, though more recent work by Swanson (2011) has found that this operation had some significant market impact. The two studies, separated by over 40 years, agreed on the basic impact in terms of basis points on yields but not on the significance of the operation: 10-20 basis points was not considered a large number in the 1960s but seems to be thought to be significant today.

More recently, the QE programme implemented by the Bank of Japan from 2001 to 2006 generated new interest in unconventional monetary policy implemented through large-scale asset purchases. In a survey of empirical evidence in the Japanese case, Ugai (2007) found mixed evidence. He concluded that the evidence suggested that QE had some signalling impact on market expectations in the sense of confirming that interest rates would remain low for some time, but the evidence on whether the QE operations had any direct effect on bond yields or risk premia was mixed. However, Bernanke et al. (2004), examining the Japanese experience with QE, found little by way of announcement effects but some evidence from a macro-finance yield curve to suggest that Japanese yields were roughly 50 basis points lower than expected during QE. Unsurprisingly perhaps, the QE programmes implemented in the aftermath of the 2008 financial crisis have led to a dramatic increase in research on this topic. Most notably, the Federal Reserve's QE programme has spawned a large and rapidly growing literature.

In the US case, despite a wide range of methodological approaches, there is near-unanimous agreement that the US programme had significant effects on longer-term bond yields, though estimates of the scale of the effect vary considerably. For example Gagnon et al. (2010) find that the $300bn of US bond purchases, which amount to approximately 2 per cent of GDP, resulted in drops of some 90 basis points in US 10-year Treasuries, while Krisnamurthy and Vissing-Jorgensen (2010) find that a reduction in public debt outstanding of around 20 per cent of GDP would reduce yields by between 61 and 115 basis points. So far, the UK's QE programme has attracted less interest. Empirical estimates of the impact of the initial 125bn [pounds sterling] of QE and then the further 200bn [pounds sterling] (14 per cent of GDP) on UK gilt yields by Meier (2009) and then Joyce et al. (2010) suggest that yields are some 40-100 basis points lower than they would have been in the absence of QE. Caglar et al. (2015) do, however, suggest that the event study methodology may have overestimated the effects because of the dominant, possibly exaggerated, impact of the first rather than the subsequent announcements.

Even if a significant impact can be demonstrated on market interest rates along the term structure, the more important question is whether the economy has been better stabilised than it might have been in the absence of QE. Such a counterfactual is rather hard to run but it is the bread and butter of economics. Consider an economy that is suffering a financial shock so that borrowing and lending activities are constrained, which prevents the circulation of capital to new technologies and leads to the increasing prevalence of older technologies. This will tend to hamper the accrual of technological change. If we add in deleveraging by households and firms, which is another way of saying that they start to save, and we add in a multiplier-type story that means lower incomes lead to more savings and a channel of magnification, by which all trade competitors also undergo the same process, we can start to understand both the rapidity and depth of the recession. Much of this response to the financial crisis was thus to do with actual economic capacity, which would fall as the fraction of less productive firms increased, and a change in the perceived 'safe' level of debt in the economy, and would result in a persistent downward response in activity. All monetary policy could seem to do in these circumstances was to offer something of a softer landing than would otherwise have been the case. It ought not to have produced immediate strong growth but simply curtailed the length of the recession, which it most probably did.

Actually there was an even bigger picture being drawn. If even reasonably safe assets like government bonds had risk premia that might be affected by changes in their net supply, it implies that the demand curves for these assets are inelastic. And if demand is inelastic for government bonds at a given price they are likely to be inelastic for other more risky assets. Changes in risk premia can then be thought of as part of the monetary transmission mechanism as they lie within the (at least partial) control of the policymaker. And this meant that the standard picture of the monetary transmission mechanism was a little out of date as it did not explicitly consider risk premia, banks or money supply, and all policy was simply directed through the official interest rate.

3. Incorporating risky borrowing and lending

And so we are about to move away decisively from the model in which the macroeconomy corresponds closely to that of a single household, with income, expenditure and output all determined by the choices of this one representative agent, which has been the basic building block for so much policy thought. But I shall examine what happens when we unbundle the economy into a saver household and a borrower household who each face different interest rates. (6) We can easily also call these asset-rich and debt-poor households but let us keep with savers and borrowers for the moment. We can then examine the equilibrium from an unconstrained, and supply constrained perspective. And consider the case for macro-prudential instruments (MPIs) as a Pigovian tax, one which aims to ensure that the private costs of credit supply do not diverge from the costs likely to be borne by society. In the type of model where there is more than one interest rate, policy may operate either through the standard short rate set by the central bank or through the interest rates available in the bond market. When we consider the standard consumer problem, for a saver, consumption growth is proportional to the deposit (or bond) rate available to savers so that when interest rates go up, consumption is delayed until tomorrow, so that expected consumption growth is positive. Thus the current level of consumption by savers is a negative function of both the deposit or bond rate, and so the pool of savings available today is increasing in these interest rates.

We go on to consider the same problem from the perspective of a borrower-household rather than a saver-household. The consumption of borrowers is tilted by the rate of interest they have to pay on their borrowing, which is the basic deposit rate plus a premium related to the costs of obtaining funds from a financial intermediary, and is normally called the external finance premium. The quantity of borrowing available from a financial intermediary is capped by the present value of collateral and can be further constrained by any limits imposed on the amount that can be lent relative to borrower-household income. Accordingly, the growth in net lending cannot be greater than the growth in the value of collateral offered. Therefore any policy that tightens the borrowing constraint directly will act to reduce net lending. We shall return to this point. The key point here though is that borrower-households are sensitive to changes in lending rates rather than savings rates.

Although saver-households ultimately provide the funds for borrowers, these savings are channelled through a financial intermediary, typically a bank, which will wish to maximise profits by lending subject to the costs of providing funding, which is the interest paid on deposits. In the absence of risk premia, or in the perfectly competitive world of loans supply, the marginal price of a loan will equal the marginal costs of funding and the loan rate will equal the deposit rate. But normally the financial intermediary supplies funds in a costly manner because borrowers need to be screened and monitored. A financial premium arises and there is a wedge between the borrowing and lending rates.

Now, should the supply of funds through financial intermediation not price social welfare accurately, we may have more (or less) lending than is socially optimal. There are a number of possible reasons why intermediation may lead to distortions in asset prices. First, the borrower may walk away from the debt related to an asset purchase and choose not to pay the principal back and leave the bank holding the loss rather than the borrower. In this case, the value of the asset to the borrower will be distorted upwards because he will not pay any losses. A similar phenomenon may be found in the banks themselves, which may feel that they need not make provisions for losses if the state (via taxes on future households) will pay and in this case there will be more lending than when the costs of losses are internalised by the bank. Finally, if asset values underpin lending, household consumption may move sectorally, in the borrower household, or even in aggregate very strongly with asset prices and become rather too volatile. Ultimately any failure or extended 'stop' in the intermediary function will prevent households from borrowing or lending in the presence of temporary shocks in their income and sharing their risks with other households.

Figure 4 illustrates the basic implications of MPIs, which leads from our analysis: the supply of savings by saver-households, and increases in the real interest rate, [R.sub.t], which we can think of as some combination of the deposit rate and the bond rate. The level of consumption is set by the [R.sub.t], which determines the level of consumption by savers and borrowers. At the unconstrained equilibrium, the external finance premium, the wedge between borrower and saver interest rates, is driven to zero and consumption is maximised at C* for borrowers. When we add in an external finance premium, the level of consumption is lower for borrowers and higher for savers, as the latter save less. Indeed, as we move to the left of C*, the consumption of borrowers falls and that of savers increases at time t, and thus the consumption of these two types of households may tend to be negatively correlated. The composition of demand in this economy is determined by the real interest rate, [R.sub.t], as it determines the split between borrower and saver consumption. The market determined external finance premium, [efp.sub.t], reflects the sensitivity of borrower household consumption and I show one possible equilibrium [C.sup.efp], where consumption by borrower households is constrained. Note that the external finance premium falls when real rates rise and increases when it falls. (7) In other words when the economy is expanding and the deposit and bond rates, which are closely tied to the policy rate, tend to rise, lending conditions may actually relax and when the economy is in a rut with low policy rates, lending conditions may well be quite restrictive.

In a general equilibrium, we may expect something like this mechanism to hold because high policy rates, low lending spreads and an economic expansion (and vice versa) all go hand in hand as part of the economic narrative. But if the supply of savings, which is intermediated through banks, does not price the social costs of lending, it may thus be appropriate to place a 'tax' on supply and this will tend to reduce further the consumption of borrowers. The tax can be any policy that reduces the supply of savings at every given interest rate and may include actual taxes on financial transactions, as proposed by Tobin, or simply policies that limit supply by requiring banks to show more circumspection in lending. The basic result would be to further limit the consumption of borrower-households to the point [C.sup.mpi], at which the existing magnitude of the external finance premium is augmented by the [MPI.sub.t]. The lower level of consumption here by borrowers is designed to reduce the build-up in financial risks over the business cycle and can be modified in a manner separate from the policy rate and thus offer policymakers an extra degree of freedom.

[FIGURE 4 OMITTED]

4. Macro-prudential instruments (MPIs)

There is no established workhorse model (yet) for understanding financial frictions in the economy and there are a number of models vying for professional adoption. But Hall (2009) provides a useful taxonomy. In a manner similar to the analysis in the previous section, he reminds us that an increase in any financial friction will tend to increase the interest rate wedge between those who provide capital and the firms and households which demand it. Such a wedge will tend to depress output and employment. The story is similar to the analysis of the inefficiency of taxation of intermediate products, with capital playing the role of an intermediate product. The argument here is that taxing an intermediate good distorts the allocation of factors of production between intermediate and final or consumption goods and so leads to a smaller overall level of income in the economy.

The increase in financial frictions acts to increase the price of capital and so reduce its demand and the lower level of capital induces a fall in output in a standard Cobb-Douglas production function. The argument goes through in the opposite direction with a fall in the size of financial frictions. Indeed under this kind of analysis financial frictions are embedded in the supply side of the economy and may be particularly hard to understand in a NK model, which concentrates on demand and cost-push shocks in the production of goods. When we take financial frictions seriously, it turns out that these frictions will tend to induce changes in output and inflation that look very much like an increase in potential supply and so will tend to be accommodated by monetary policymakers. MPIs are thus any attempt to offset changes in the frictions by offsetting changes in 'taxes', so that a booming economy experiencing a lessening in frictions may be subject to an increase in its MPI tax and an economy in recession and increasing financial frictions may be treated to a fall in the tax in order to attenuate the impact on output.

Monetary and financial stability

The new central banker might take the view that financial and monetary policy should simply run in tandem. Thus managing the latter well also requires attention to be paid and information to be exchanged in pursuit of the two objectives jointly. Indeed the historical record suggests a similar juxtaposition--that the nature and scope of the regulation of financial intermediation was closely linked to the monetary policy regime--so the immediate postwar period with the Bretton Woods system of fixed-but-adjustable exchange rates was associated with both extensive regulation of the financial system and also the virtual elimination of banking crises, apart from in Brazil in 1962. (8)

However, the cost of such extensive supervision may have been that the financial system did not allocate investment particularly efficiently over that period, and momentum for deregulation built up to a considerable degree. In principle, therefore, there is a trade-off between designing instruments to stabilise the financial system and prevent excessively volatile financial outcomes, and ensuring that the financial sector retains the correct incentives to locate investment opportunities and allocate funds accordingly. It is not initially clear that employment of MPIs in a single currency area can work independently of further controls on the movement of capital across currency regions, particularly when financial intermediaries have interests overseas. We are looking for instruments that will work, given some form of monetary policy regime that closely resembles what we currently have in place.

From the perspective of monetary policymakers, the initial debate was on whether inflation targeting could be modified so that an additional instrument could be used to stabilise financial imbalances or directly control the extent of financial intermediation. The answer that emerged before the full force of the financial crisis was understood was that there was limited scope to do very much. Bean (2004) argues that it is optimal under discretion to ignore any asset price boom and only mitigate any fallout on collapse. Under commitment it turns out there is even less incentive to stabilise output when the economy is overheating. Svensson (2009) considers that 'flexible inflation targeting' that stabilises output and inflation may have an occasionally binding constraint to ensure financial stability. Booms (busts) can justify an inflation undershoot (overshoot), as well as an extended period of adjustment back to target. Even if a limited number of modifications to monetary policy operating procedures are sufficient to stabilise macroeconomic outcomes, they may not be enough to realise financial stability. Appropriate supervision and regulation are unlikely to be replaced simply by new instruments.

In fact, in the event, another instrument was developed, as explained previously. This was QE, designed to deal directly with the zero lower bound constraint. It seems to have driven medium-term yields down by the extent to which they might have been expected to fall had short-term interest rates been lowered by some 2-4 per cent below zero. Under risk aversion financial intermediaries cannot create sufficient liquidity and so, in principle, the central bank can regulate the flow of liquidity over the business cycle in order to prevent excessive amplification of the business cycle by financial intermediaries. (9) The swap of reserves for bonds though did not palpably augment bank lending and there seems to be no attempt to consider using this stock of bonds to help regulate the financial system on an ongoing basis.

Loss function

MPIs might involve a large number of possible instruments including capital, margin, liquidity and equity-loan ratios. There is a danger that, given the recent experience of an overextended financial system, the mindset for the pursuit of MPIs implies an asymmetric concern with the stability of the financial system. This is rather like the foundations of a building or the construction of a dam, so that we are in general concerned with reining in excessive intermediation rather than having too little. Put rather bluntly: when we think about financial stability who would lose their job if the financial system were considered to be excessively safe compared to the opposite?

An asymmetric loss function does not necessarily have to be pursued asymmetrically. (10) The policymaker simply has to pursue a slightly different target. This is because the minimum of the loss from a given variance under an asymmetric loss function is not at the turning point of the loss function but at some point in the opposite direction of the steeper asymmetric loss. So if you have an asymmetric target, for example, for catching a train, it does not make sense to plan to arrive at the station just as the train is about to depart but with a safety cushion. In fact, the optimal time to arrive would be given by a term that governs the asymmetry of the function and the likely size of any shocks. So the target for long-run macro-prudential instruments should be driven up above the safe level in some degree proportional to the size of shocks and the extent of asymmetry in the loss function. Once this principle has been established, it makes sense to develop steady state targets that build in a precautionary target for more liquidity, capital and equity to loan ratios than a strict minimum might imply.

Target and instruments

When it comes to MPIs, we want to count, Tinbergen-style, the number of independent instruments and objectives. In the current set-up, the Monetary Policy Committee (MPC) will continue to set Bank Rate to pursue the inflation target and it will be the Financial Policy Committee (FPC) that will have instruments at its disposal to pursue financial stability. To the extent that we cannot be sure about the impact of any instrument, Brainard uncertainty introduces a tradeoff between the achievement of the target and the minimisation of uncertainty induced by the use of an instrument. There are two further problems here in the case of MPIs: (i) there is likely to be considerably more uncertainty with a set of untried instruments that may also have a correlation structure with each other but (ii) also because they may alter the behaviour of the financial system, they will directly affect the impact of any given stance of monetary policy.

On the first point, it might be that we can treat the new MPIs as a portfolio of instruments that jointly will reduce the idiosyncratic risk of using any one new instrument. But without specification, calibration or testing of the impact of any one instrument in combination with the others, we are unlikely to be very sure at all whether such a portfolio of instruments will be effective. Ideally we want to think about which instruments may be used and how they might be used together in a manner that does not induce greater uncertainty into the operation of monetary policy. Figure 5 shows how the use of MPIs may improve the trade-off in uncertainty-space available to policymakers, so that rather than the line ABC the policymaker will be able to choose along AB'C' and would prefer a point fairly near to C', where inflation is at target and an undershoot, such as B, is no longer the preferred point on the trade-off.

To the extent that changing the constraints faced by financial intermediaries will alter the financial conditions, there may not only be an impact on the appropriate stance of monetary policy but also an impact on the appropriate MPIs conditioned on the monetary policy stance. Consider a world in which the monetary policymaker wishes to smooth the response of consumption to a large negative shock to aggregate demand and reduces interest rates faced by collateral-constrained consumers. Simultaneously, financial stability may be considered to be threatened and various MPIs may be tightened, which would act against the interest rate changes made by the monetary policymaker and may need further or extended lower rates of interest rates. If, on the other hand, sufficient precautionary moves had been made by the FPC in advance there may be no immediate conflict.

Operating MPIs

Although MPIs may be used to help stabilise the financial system over the business cycle, there are some separate issues to consider when designing MPIs to help stabilise a reasonably well functioning financial system. This might be thought of as leaning against the wind, and in considering the correct responses for a highly vulnerable and undercapitalised financial system. The former implies the use of cyclical instruments to prevent a problematic build-up of risk and the latter some attention to the superstructure of the financial system, with individual firms and the sector as a whole not only able to withstand shocks but sufficiently robust as not to amplify them.

[FIGURE 5 OMITTED]

Yet the financial system is already undergoing a considerable deleveraging that has involved a build-up in core capital, increased holdings of liquid assets and greater margin requirements. In a sense the financial system is moving from a loose regime to a tighter one, but too fast a transition may have unwanted macroeconomic consequences. The extent to which difficulties in obtaining finance may constrain the investment or consumption plans of some firms and households may imply that, although tougher long-run regulatory targets might be optimal, there may be some sense in thinking about how to allow the divergence from these targets for extended periods. In the same way, a credible fiscal regime that ensures sustainable public finances is more likely to allow the full force of automatic stabilisers to operate. In this sense, if banks are forced to observe a target at all times, this may be counterproductive for the system as a whole. It is an example of Goodhart's (2008) taxi: where a taxi at a railway station at night could not accept a fare because of a regulation that stated at least one taxi had to be at the railway station all the time.

One of the results to emerge from the analysis of monetary policy is that the control of a forward-looking system is best achieved by setting predictable policy that allows forward-looking agents to plan conditional on the likely policy response. There has been considerable work to suggest that the impact of monetary policy is a function of both the level and the path of interest rates, which is likely to be closely related to predictability. As well as thus evaluating instruments, the FPC will have to pay careful attention to how expectations of changes in MPIs are formed and whether partial adjustment towards some intermediate or cyclical target for a given level of capital, liquidity or loan-to-value will be adopted. The alternative of jumping to new requirements may induce large adjustment costs for the financial sector and the use of considerable resources to predict future movements in requirements. The private sector may also be induced to bring forward or delay financial transactions depending on the expectations of collateral requirements. In a slightly different context, the pre-announced abolition of mortgage interest relief at source (MIRAS) may have played a role in stoking the house price boom of the late 1980s (Lawson, 1992). Under some circumstances, such a response reflecting strong intertemporal switching may be entirely what a macro-prudential framework may wish to bring about. More generally, however, when agents are well informed and forward-looking some thought has to be given to developing a framework for understanding agents' responses to any expected or pre-announced changes in the rules governing financial intermediation.

Monetary policy and liquidity

Some recent work on the nexus of MPIs and monetary policy does suggest that there may be a complementarity. The widespread adoption of non-conventional monetary policies has provided some evidence on the efficacy of liquidity and asset purchases for offsetting the zero lower bound. These tools can essentially be thought of as fiscal instruments, as they issue interest rate bearing central bank liabilities. These instruments are placed in the government's present value budget constraint and the consequences of these operations on banking and money can be examined. The supply of reserves as a swap for bonds and capital helped stabilise the economy because such balance sheet operations supply liquidity to a financial market that is otherwise short of liquidity and hence may allow other financial spreads to move less violently over the cycle to compensate. These liquidity operations act like negative taxes on financial intermediation.

There are a number of missing elements to the analysis: (i) the consideration of fiscal policy, which if excessively expansionary may induce increases in liquidity or risk premia, and may not then be in a position to offset liquidity shortages by trading long-run debt for short-run liabilities and (ii) the consideration of non-linearities or discontinuities in responses, e.g. from bankruptcy. That said, there may be some gains from jointly determined MPIs and standard interest rate responses, conditioned on sustainable public finances, leading to welfare gains for households.

Following the financial crisis, and the need to undo the Separation Principle for monetary and financial stability, we can agree there are missing instruments. There has been a hunt to locate ones that can be employed, or suggested for use, by the FPC. I remain concerned as to how long-run targets for capital, liquidity and asset-mix and lending criteria will be set and whether a bias to over-regulation may be set in train. It is not at all clear how many new cyclical MPIs will interact with each other and impact on the setting of monetary policy. A reverse causation is also possible, whereby the stance of monetary policy may have implications for the correct setting of MPIs. The management of expectations over any announcements of changing MPIs will be a crucial area in a modern financial system. It was probably significantly easier in a world of extensive capital and exchange controls that characterised the immediate postwar period. All that said, early results from a new generation of micro-founded macro models do suggest that there may be significant gains from getting the calibration of these new instruments right but much work remains to be done.

5. The interaction of government debt, monetary policy and financial policy

In some sense the classical monetary model places a lot of the action off-stage and so brings into focus the heroic role of the monetary policymaker. Actually there are at least two key interactions that both limit and channel the actions of the monetary policymaker: fiscal policy and the operations of the financial sector. A further interaction concerns that between the financial sector and the fiscal policymaker, which may consider the role that public sector purchases of financial institutions played in stabilising the financial sector and also the extent to which financial sectors' liabilities are hedged with government IOUs of one sort or another. We must have a happy triumvirate.

The fiscal policymaker is typically charged with respecting the government's present value budget constraint, which means establishing plans for expenditure and taxes that mean the level of debt is expected to be (low and) stable under likely states of nature. The financial sector operates to translate savings into stable returns by intermediating between current investors and consumers and future investors and consumers. The stable income streams offered by the government sector may be of value to the private sector as it seeks nominal or real payments that are stable in the face of business cycle shocks. They may also provide a benchmark for the construction of other market interest rates. The monetary policymaker sets the costs of funding for the financial sector and also has a huge influence on the costs of funding government debt. The level of economic activity depends to a large degree on the financial and fiscal sector, so it is an outcome of the central bank's responses to the behaviour of these two sectors.

I am not necessarily arguing that there is a need for explicit co-ordination, but the Nordhaus (1994) example of monetary-fiscal interactions may be instructive if not completely comparable to that of the financial sector and our triumvirate. Consider a y-axis representing economic activity and the x-axis, the policy rate. Also accept that a conservative bank chooses a preferred level of interest rates for every level of activity on a path that will tend to drive the economy to its preferred point, MPR. The financial sector (let us set the fiscal sector aside for the moment), may be stabilising and act to drive up activity when it is below the socially preferred point and help bear down on it when activity is above the socially preferred point, FS. This is because asset prices and market interest rates may act to generate levels of activity that act as a conduit from monetary policy to the overall level of activity back to some notion of long-run equilibrium. (11)

So in normal times, the central bank relies on the financial sector to be stabilising and to carry out a large part of the stabilising response. But in times of boom and bust it may stoke up excessive fluctuations in activity, which begs the question of why any financial agent pursues plans that differ from those the central bank might choose? One, because it has different preferences, two, because it has different information and three because it will not bear the consequences of its choices. As a result, without any co-ordination, the Nash equilibrium may imply high interest rates. But the respective bliss points for monetary policy, MPR, and for the financial sector, FS, imply a contract curve along which losses for each policymaker will be less than under the Nash equilibrium, which means that some form of cooperation is to be preferred. Whether that cooperation can be constructed in a manner that brings us firmly back to our preferred equilibrium is the key question for the economic or monetary settlement after the crisis.

A simple, if not simplistic, reading of the facts, is that monetary, fiscal and financial policy are in need of some coordination not only to provide protection against a repeat of this crisis but also to help us recover normal times. A story of coordination is plausible and attractive but unless the loss functions faced by each type of policymaker lie on the same Bliss Point, there is a danger that the institutional solution will not bear the weight placed upon it. It is also not clear that each arm of policy should be thought of as trying to drive the economy to the same point in levels. An argument can be made that financial policy and perhaps even fiscal policy are about reducing uncertainty or broadly-speaking risks. Financial policy might be best designed to reduce the possibility that the financial sector might be increasing the vulnerability of the economy to shocks and that it can continue uninterrupted in its objective of facilitating risk-sharing. In a similar vein, fiscal policy operates to share risk intertemporally and intergenerationally. A considerable and influential research effort has shown the importance of uncertainty as a driver of the economic cycle and so it might also be the case that monetary policy has to confront higher moments and tackle the question of policy uncertainty directly. One solution suggested by Barwell and Chadha (2013) would be to take ideas about forward guidance, that tie the market participants views about the likely path of Bank Rate to well understood and spelt out conditions, and complete them with conditional paths for the policy instrument which will depend on both economic scenarios and the impact of other arms of policy.

[FIGURE 6 OMITTED]

6. Concluding remarks

I am not sure that wisdom and foresight have necessarily been lost in the search for a simple, credible monetary policy. The ultimate decisions of any policy rely on judgement and that can, unfortunately, remain faulty even in the presence of considerable wisdom and foresight, whilst both judgement and intuition are often formed with reference to the experience of working with models. But because no model can provide a perfect guide to the menu of choices, we must learn not only to choose which ones are useful for policymakers but also to think through the implications of our models being wrong. The robustification of policy may mean working through the implications, however unpalatable, of the unlikely as well as the preferred circumstances.

Some difficult lessons have been learnt over this crisis that bear repeating. First and rather obviously, inflation targeting alone, or as it was practised, cannot prevent boom and bust and needs to be augmented with more instruments and, if possible, better judgement. The operations of the financial sector through the creation of various elements of broad money and also at the zero lower bound, as it changes its demand for central bank money, complicates choices about the path and long-run level of Bank Rate. Policy rates are no longer being ever so slightly perturbed near a well-understood long-run level. Rather they are stuck in the doldrums looking for a chance to escape, but may not return to previous levels, in part because the action of MPIs is raising other market interest rates.

Not only do financial frictions complicate the choices of policymakers because changes in the financial settlement may make the transmission of policy hard to gauge, but they have always acted through the traditional supply and demand sides. This means that they make capacity judgements very hard and it is likely that the key monetary policy judgement involves working out the current and likely future levels of spare capacity in any economy. The sensible application of liquidity and capital targets via macro-prudential policy may seem likely to reduce business cycle variance, albeit at some cost to permanent output. Transitional judgements will have to be even more careful than usual not to treat the permanent as the temporary and vice versa.

The interactions between fiscal, financial and monetary policy notwithstanding, we also now accept that fiscal policy, as well as underpinning aggregate demand, can also provide support to fragile financial institutions, if and only if the private sector wishes to hold government IOUs. This further contingent role for government debt makes the case for slightly more conservative fiscal policy than aggregate demand considerations might themselves imply. During the long and lonely march back to normality, public debt levels are likely to take at least 10-15 years to get back to 'normal' and, as long as demand remains inelastic, positive or negative changes in net supply will impact on price and complicate choices on Bank Rate. So it would seem that plotting the policy path will be considerably more complicated during recovery, even with the resumption of normality, and thus requires significantly more explanation than we have tended to have in the past, rather than requiring a change in nominal target.

NOTES

(1) It seems to me that the continuing reluctance of the FMOC to start the process of interest rate normalisation has echoes of this argument by Hawtrey.

(2) For what follows, it is worth spelling out that a bond price is the present value of the interest payments and principal owed to the holder of the bond by the issuer. The present value calculation incorporates an interest rate forecast over the life of the bond and the price increases as interest rates fall because the present value of far off payments is correspondingly higher. Thus bond prices and interest rates move in opposite directions.

(3) See Allen (2014) for more on this process.

(4) This calculation uses a discrete time Cox-Ingersoll-Ross model of the term structure and is available on request.

(5) Chadha and Meaning (2013) calculated that these QE transactions were unlikely to lead to loss following redemptions and final sale back to the non-bank financial sector see http://www.publications.parliament.uk/pa/cm201314/cmselect/cmtreasy/writev/qe/mO1.htm

(6) This section draws on Chadha, Corrado and Corrado (2013).

(7) Note also that changes in the lending constraint will change the size of the external finance premium

(8) Allen and Gale (2007) make this observation as well.

(9) See Gale (2015) on this point, who also argues that when risk appetite is high, too much liquidity can be created.

(10) These questions are considered in Chadha and Schellekens (1999).

(11) Consider FS when it is steeper than the MPR, this will tend to destabilise the economy.

REFERENCES

Allen, F. and Gale, D. (2007), Understanding Financial Crises, Clarendon Lectures in Finance, Oxford, Oxford University Press.

Allen, W. (2014), Monetary Policy and Financial Repression in Britain, 1951-59, Palgrave Macmillan.

Barwell, R. and Chadha, J.S. (2013), 'Complete forward guidance', in den Haan, W. (ed.), Forward Guidance, VoxEU E-book.

Bean, C.R. (2004), 'Asset prices, financial instability, and monetary policy', American Economic Review, Paper and Proceedings, 94(2), pp. 14-18.

Bernanke, B., Reinhart, V. and Sack, B. (2004), 'Monetary policy alternatives at the zero bound: an empirical assessment', Brookings Papers on Economic Activity, 35, 2, pp. I-100.

Caglar, E., Chadha, J.S., Meaning, J., Warren, J and Waters, A. (2015), 'Central bank balance sheet policies: three views from the DSGE literature' in Chadha, J.S. and Holly, S. (eds), Interest Rates, Prices and Uquidity, Cambridge, Cambridge University Press.

Chadha, J.S., Corrado, G. and Corrado, L. (2013), 'Stabilisation policy in a model of consumption, housing collateral and bank lending', Studies in Economics 1316, School of Economics, University of Kent.

Chadha, J.S., Chrystal, A., Pearlman, J., Smith, P.N. and Wright, S. (eds) (2016), The UK Economy in the Long Expansion and its Aftermath, (forthcoming), Cambridge University Press.

Chadha, J.S. and Schellekens, P. (1999), 'Monetary policy loss functions: two cheers for the quadratic', Bank of England working paper no. 101.

Gagnon, J., Raskin, M., Remache, J. and Sack, B. (2010), 'Large-scale asset purchases by the Federal Reserve: did they work?, Economic Policy Review, May, pp. 41-59.

Gale, D. (1982), Money in Equilibrium, Cambridge University Press.

--(2015), 'Liquidity and monetary policy', in Chadha, J.S. and Holly, S. (eds), Interest Rates, Prices and Liquidity, Cambridge: Cambridge University Press.

Goodhart, C.A.E. (2008), The Future of Finance-And the Theory That Underpins It, London School of Economics (chapter 5).

Hall, R. (2009), 'The high sensitivity of economic activity to financial frictions', NBER mimeo.

Hawtrey, R.G. (1932), The Art of Central Banking, Longmans, Green and Company.

Joyce, M., Lasaosa, A., Stevens, I. and Tong, M. (2010), 'The financial market impact of quantitative easing', Bank of England Working Paper No. 393, pp. 1-44.

Krishnamurthy, A. and Vissing-Jorgensen, A. (2011), 'The effects of quantitative easing on long-term interest rate', Northwestern University Working Paper, pp. 1-47.

Lawson, N. (1992), The View from No. 11: Memoirs of a Tory Radical, Bantam Press.

Meier, A. (2009), 'Panacea, curse, or nonevent: unconventional monetary policy in the United Kingdom', IMF Working Paper No. 09/163, pp. 1-48.

Modigliani, F. and Sutch, R. (1966), 'Innovations in interest rate policy', American Economic Review, 52, pp. 178-97.

Nordhaus, W.D. (1994), 'Marching to different drummers: coordination and independence in monetary and fiscal policies', Cowles Foundation Discussion Papers No. 1067.

Svensson, L.E.O. (2009), 'Flexible inflation targeting - lessons from the financial crisis', comments at Netherlands Bank, Amsterdam 21 September.

Swanson, E. (201 I), 'Let's twist again: a high-frequency event-study analysis of Operation Twist and its implications for QE2', Brookings Papers on Economic Activity.

Ugai, H. (2007), 'Effects of the quantitative easing policy: a survey of empirical analyses', Monetary and Economic Studies, 25, 1, pp. 1-48, Institute for Monetary and Economic Studies, Bank of Japan.

Jagjit S. Chadha, Professor of Economics, University of Kent; Visiting Professor at the Faculty of Economics at the University of Cambridge; The Mercers' School Memorial Professor of Commerce at Gresham College, London. E-mail: J.S.Chadha@kent.ac.uk. This article summarises elements of the Westminster Economics Lecture given in March 2014 at the National Institute of Economic and Social Research, which laid out an agenda for UK monetary policy as we recover from the financial crisis. Some of this article also draws on my Gresham Lecture given in June 2015. I am grateful for comments and conversations with Francis Breedon, Germana Corrado, Luisa Corrado, Mike Dicks, Monique Ebell, Sean Holly. Jack Meaning, James Warren and Alex Waters and any remaining errors are my own.

Table 1. US and UK public debt dynamics
after World War II, annual average 1946-61

      Interest   Inflation      GDP
        rate                  growth
      payments

US      1.8        -2.8        -2.5
UK      4.5        -3.4        -2.2

      Primary    Residual     Average
      balance                change in
                             debt/GDP

US      -2.1        1.0        -4.6
UK      -4.2       -2.3        -7.7