Tuesday, 28 April 2009

Easing understanding


1.     As the financial crisis has developed and increasingly affected real economic activity, the US Federal Reserve and the Bank of England (BoE) have resorted to progressively more aggressive and unconventional easing measures. While these measures have been described in the media as "quantitative easing" and even "printing money", these terms have not been used rigorously or even consistently. Commentators often misunderstand the significance of issues such as the payment of interest on reserves and the growth in base (central bank) money. It seems that the confusion is at least partly explained by the fairly superficial coverage of monetary policy implementation in economics textbooks, which abstract from certain details that are key to appreciating the differences between conventional easing, the pre-crisis idea of quantitative easing (QE) as applied by the Bank of Japan (BoJ) from 2001 to 2006, and the unconventional measures currently being applied by the Fed and the BoE.

2.     This post is an ambitious attempt to provide an explanation of monetary policy easing that covers all the aspects necessary to understand both conventional and unconventional easing measures. Although there has been some such explanation in the media, RebelEconomist has not found it entirely satisfactory, and offers his own here. It is hoped that his combination of experience of central bank market operations and academic monetary economics might just produce an account of monetary policy that does not leave vital questions unanswered and expresses some ideas in a slightly different way that can provide fresh insight. The post classifies Fed and BoE easing policies as they have evolved through the crisis. It is argued that while both central banks' unconventional measures have always been specified in quantitative terms, they are more accurately described as quantitative liquidity, credit and term easing rather than quantitative monetary easing. Neither central bank has yet begun to print money.

3.     Given its objective, the post is long, so readers who are mainly interested in present policy may wish to fast-forward to the discussion of unconventional easing measures in the last section. And in view of the complexity of the issue, assessment of the effectiveness and merit of the various policy options is mostly left for subsequent posts. Therefore, in the style of the central banker he once was, RebelEconomist has numbered the paragraphs for ease of reference here and in subsequent posts on monetary policy.

Fundamentals of easing

4.     Modern money (as opposed to commodity money such as gold coins) is debt; a readily transferable and widely accepted type of continuously redeemable debt. Typically money is non-interest-bearing debt owed by a highly creditworthy borrower such as a central bank in the case of banknotes or, in the case of deposit money, a commercial bank backed by state deposit insurance.

5.     To avoid going into the complexity and mystery of the monetary policy transmission mechanism, suffice it to say that, other things equal, an increase in the stock of money (also known as the money "supply", which perhaps confusingly implies a flow) has a stimulating effect on economic activity for a few months at least, and that short-term interest rates play a key role in the process. Because money is a substitute for interest-bearing short-term debt (ie money available with a short delay), an increase in the money supply tends to increase the price of all types of short-term debt, and since the price of a unit of debt (ie a unit of money due to be repaid when the loan matures) is conventionally expressed in terms of a rate of interest, increasing the money supply lowers short-term interest rates. Naturally, the influence tends to be greatest on the shortest and most creditworthy forms of debt that are money's nearest substitutes. The relationship between the money supply and the interest rate on such debt is normally described as a money demand curve, but is equivalently a debt market supply curve (Figure 1), and the short-term debt market is known as the "money market". Perhaps because the transactions-facilitating function of money is often likened to the lubricating oil in a machine, increasing the money supply or lowering interest rates is described as easing, while contracting money supply is called tightening. The degree of easing or tightening is called the stance of monetary policy. In the longer term, say about two or three years, the interest-rate-lowering and activity-stimulating effect of easing dissipates as the additional money becomes distributed in a pattern similar, in the absence of changes in the structure of the economy, to that of the original stock of money, at which point the residual effect of the ease will be an equi-proportional rise in prices – in short, a completed round of inflation.

6.     The supply of money is, of course, regulated by the central bank. Although there is more than one type of money, elementary explanations of monetary policy often abstract from this complication by assuming that they are held in stable proportions defined by "multipliers", meaning that the central bank need only regulate the supply of one kind of money to control them all.

7.     The form of money that the central bank directly influences is commercial banks' current account balances at the central bank. Although central banks typically pay little if any interest on positive balances in these accounts, commercial banks nevertheless tend to keep their central bank current accounts in credit, for at least two reasons. First, these accounts are used to settle any transaction that the banks undertake with the central bank, and sometimes also to settle inter-bank payments and transfers between the government and the banks, mostly on behalf of the banks' customers. Since these payments are not entirely predictable and the central bank normally heavily penalises overdrafts, the banks maintain a positive current account balance to provide some margin for error. Second, as a liability of the currency issuer itself (Figure 2), credit in a bank's current account balance at the central bank is the safest of all monetary assets, so banks may use their central bank current accounts as a store of value if the extra return available on alternative investments is not enough to compensate for their greater risk. In view of the functions that they serve, banks' current account balances are also called "reserves".

8.     Banks' holdings of reserves for transactions purposes can be expected to be proportional to the size of their customers' demand deposits. An archetypal transaction that commercial banks undertake with the central bank is to buy and sell banknotes as their customers respectively withdraw and deposit currency (currency includes coins as well as banknotes, but coins are relatively insignificant by value and are not always a liability of the central bank). Transactions between bank customers generate inter-bank payments. In either case, it seems likely that the greater the value of customer deposits, the larger the maximum outflow that customer activity is liable to generate, which their bank must be prepared to cope with. The amount of reserves that a bank would choose to hold for investment purposes depends on the trade-off between the risk and returns on reserves and other available assets, but normally, the return on reserves is so much less than even other credit-risk-free assets such as treasury bills that commercial banks hold practically no reserves as a store of value. Left to its own devices, a bank will determine its holding of reserves according to the trade-off between the likelihood and cost of overdrafts and the relative risk-adjusted return on reserves. However, in many countries, the regulatory authorities mandate a minimum fraction of deposits that must be held as reserves, partly for prudential reasons, and partly to control deposit money growth via the effective tax on deposits imposed by the need to hold zero or low interest reserves. In such cases the regulatory reserve requirement is usually the binding constraint. Anyway, whether voluntary or mandatory, the reserve ratio means that central banks' control over the quantity of banking system reserves also implies some control over the stock of deposits and hence of broader monetary aggregates.

9.     Although reserves represent the interface of the central bank and the rest of the economy, the stock of banknotes is normally a far larger liability of the central bank. Commercial banks may exchange reserves for banknotes at par, freely and on demand, and together, reserves and banknotes comprise what is called base money. The central bank would be unwise to try to control the money supply via banknotes – restricting the supply of banknotes could trigger a panic run on any bank apparently having difficulty meeting customer withdrawals. Like reserves, however, the value of banknotes in circulation is related to the size of demand deposits. Just as reserves provide the basic type of money for banks, banknotes provide the most widely accepted and dependable form of money for individuals. The transfer of banknotes represents immediate, often unrecorded, settlement of a transaction, which makes banknotes most suitable for occasional, relatively small transactions outside of committed business relationships, and for this purpose individuals hold a buffer stock of banknotes representing some fraction of their deposits. This means that the central bank has some indirect control of banknote circulation via reserves.

10.     While the banking system as a whole can ultimately only acquire reserves from the central bank, commercial banks trade reserves between themselves by borrowing and lending for settlement in their central bank current accounts, generally from one day to the next. The interest rate in this inter-bank market for overnight reserves loans indicates whether the supply of reserves is adequate for the banks' collective needs. Assuming that no interest is paid on positive reserve balances and that the central bank's overdraft charges, or the penalties for insufficient reserves in a regime of statutory reserve requirements, begin with a heavy cost for a shortfall of any size, the short-term market demand curve for reserves can be expected to resemble that shown in Figure 1. When the banks are close to their collective reserve requirement, overnight reserves loans will trade at a high interest rate as almost as many banks are in danger of incurring shortfall charges as being left with excess reserves. If, however, reserves are even slightly in surplus the interest rate will drop steeply, until as the interest rate approaches zero, it becomes worth borrowing reserves for the security they provide, for example against accidental overdrafts arising from operational errors. Such surpluses, however, do not persist for long, because banks can make more profitable use of excess reserves by expanding their lending to non-bank counterparties, so the inter-bank market for reserves normally operates on the steep part of the demand curve. The practical implication is that the demand for reserves is normally highly inelastic with respect to interest rates, with the result that small changes in the supply of reserves lead to large changes in short-term inter-bank interest rates.

11.     Since it is a point that seems to cause some confusion, note that, although an individual bank does expend reserves when it lends to a customer, reserves are not consumed or transferred outside the banking system in the process, even if the borrower is a non-bank customer. When the borrower draws on the loan, for example to purchase a car, reserves are transferred from the borrower's bank to the car seller's bank. Although it is conceivable that the car buyer might withdraw banknotes to pay the car seller, generating a fall in reserves as the car buyer's bank purchases banknotes from the central bank (Figure 2), these would normally be quickly paid into the car seller's bank account and sold back to the central bank for reserves. A banking system surplus of reserves is absorbed by an increase in the value of bank lending up to the point that the banks are generally content with the allocation of their assets between reserves and loans, not because the reserves are used up in some way, apart from a slight drain of reserves to pay for the increased circulation of banknotes commensurate with the growth of bank deposits.

12.     Central banks manage the stock of reserves in pursuit of both macroeconomic and microeconomic objectives. Their macroeconomic task is to use the influence of reserves to adjust the money supply and short-term interest rates to regulate economic activity, subject to the constraint of holding inflation close to some target rate. The trick is to exploit the medium-term, activity-altering phase of the money transmission mechanism without allowing its long-run outcome to generate unacceptable results in terms of inflation. Central banks' microeconomic task is to offset unwanted variations in the stock of reserves as it is impacted by flows associated with events like tax payment deadlines and holiday surges in banknote circulation, while accommodating structural changes arising from economic growth and evolving payments technology. Although in principle a central bank could attempt to determine the size of the adjustment required and arrange one or more transactions to produce exactly that change in reserves, in practice, central banks calibrate the supply of reserves according to short-term interest rates. They operate in this way because of the key role played by interest rates in the money transmission mechanism, and because the inelasticity of the demand for reserves means that a small mistake in the central bank's estimate of the need for reserves can drive interest rates to extremes. Most commonly, the target is specified in terms of the overnight interest rate in the inter-bank market for loanable reserves (ie "Fed funds" in the USA).

13.     Note that interest rate targeting means that the textbook account of deposit money creation, in which the central bank supplies base money followed by an iterative process of bank lending and customers re-depositing base money that creates some multiple of the initial injection of base money, is unrealistic. Actually, money creation is generally initiated by bank lending, with lending generating deposits and deposits in turn generating a need for reserves which are reactively supplied by the central bank in order to fix its targeted interest rate. Nevertheless, the fact that banks are obliged to hold some fraction of their deposits as low interest reserves, plus an additional fraction as non-interest-bearing banknotes in the bank's safes and tills to cover withdrawals during the interval before its branches can be re-supplied, allows the central bank to exert some influence over deposit money creation. By restricting the supply of reserves to raise the (opportunity) cost of commercial banks' base money requirement, the central bank can curtail the marginally profitable expansion of banks' balance sheets. And the much-discussed difference between targeting interest rates and targeting money supply should not be exaggerated. Nearly all central banks maintain some economic model that links the interest rate they target with their macroeconomic objective variables like economic activity (as represented by GDP) and inflation. The relationships included in the model incorporate, implicitly if not explicitly, the role of the money supply in the transmission mechanism.

14.     In principle, the central bank could fix its targeted interest rate by adding or subtracting reserves through any type of transaction that settles in commercial banks' current accounts, relying on the substitutability of reserves and short-term debt to affect the relevant interest rate. However, as it may be necessary to withdraw as well as inject reserves, central banks typically buy assets with relatively stable value and which can be liquidated (ie sold or lent in return for money) with minimal transactions costs. Even if the bulk of the assets are unlikely to ever need to be sold because contractions of the base money supply by more than a small proportion are rare, the assets are normally retained by the central bank to ensure its solvency and hence public confidence in its currency, so central banks are famously conservative investors across their entire portfolio. Traditionally, central banks traded gold, but since convertibility has been abolished, they have been able to buy income-generating financial assets instead. Given the importance of security, central banks favour fixed income debt assets, and normally only debt of the most creditworthy borrowers or short-term secured loans to banks in the form of repurchase ("repo"; strictly, from the money lender's point of view, "reverse repo") agreements collateralised by creditworthy bonds. Since nearly all investors are willing to hold very safe debt, such debt is also highly liquid, and short-term debt is quickly self-liquidating anyway as it matures. In practice, therefore, debt has now become the standard tool for monetary policy adjustment. Conventional easing involves creating reserves to make repo loans or to buy government or quasi-government bonds (Figure 2).

From interest rate targeting to paying interest on reserves

15.     Naturally, when debt is used as a monetary asset, there is potential for confusion between the interest rate on the central bank's target debt instrument, such as overnight inter-bank loans, and on the debt it trades in pursuit of that target in its monetary policy operations. Moreover, using debt allows the central bank to influence interest rates through its assets as well as its liabilities, an idea which is key to understanding and classifying the unconventional easing policies being applied during the present financial crisis. It is therefore unfortunate that few textbooks make this potential distinction between target and operational interest rates entirely clear.

16.     Not all central banks buy long-term debt in their monetary policy operations, and even those that do, notably the Fed and the Bank of Japan, tend to use short-term debt to make their marginal adjustments to the supply of reserves and buy long-term debt occasionally for the stable core of their asset portfolio. In fact, as RebelEconomist noted in a previous post, it is not clear why a central bank that targets short-term interest rates only should buy long-term debt at all. In developed countries in normal economic conditions, the balance sheet of the central bank tends to be small compared with that of the whole banking system, so it would seem best to concentrate the influence of the central bank's asset purchases on debt that, if not actually the target debt instrument itself, is as similar to it as possible. It may be necessary to use a close substitute rather than the target debt instrument itself because the target debt instrument is not considered a suitable investment for the central bank. Usually, this is because the target debt type is unsecured, such as inter-bank deposits, and the central bank wants to avoid managing the credit risk involved in buying such debt (which besides running some risk of loss, may require the central bank to favour dealing with some banks over others). Also, if an overnight interest rate is targeted, the central bank may prefer not to renew its entire stock of assets every day, in which case it may choose to deal in, say, one-week debt instead. Another advantage for the central bank of specialising in short-term debt is that its short maturity produces a strong natural drain of reserves as the debt is repaid, which puts the central bank in a strong position to reset the stock of reserves and interest rates frequently.

17.     For an interest rate targeting central bank, its operational objective is to hold the rate of interest on its target debt instrument close to the target with minimal variability. To this end, central banks use two basic trading strategies. Their primary strategy is to attempt to estimate the change in reserves necessary to guide the interest rate to the target and arrange one or more transactions to engineer that change. The estimation problem is more one of allowing for scheduled flows like maturing central bank loans and anticipating the actions of institutions and people, known as autonomous factors, rather than identifying and using the money demand curve. Usually, the adjustment required is an injection of reserves to replace maturing debt held by the central bank. The central bank then establishes transactions to make that adjustment by offering to buy or sell as necessary the corresponding amount of debt, either at a fixed interest rate or by auction. Although the central bank usually deals with a limited set of counterparties, these may include non-banks, since transactions with them are settled using their correspondent bank's reserve account. This procedure is called an open market operation (OMO). Central banks' secondary strategy is to provide standing facilities which offer to take deposits at an interest rate some spread below the target and to lend reserves at an interest rate some margin above the target, in practically unlimited size. These standing facilities provide respectively a floor and ceiling to the targeted interest rate in case, even after OMO adjustment, it would otherwise miss the target by an intolerably large amount. Since the standing facilities are designed to be used only in exceptional circumstances, the spreads are typically prohibitively wide.

18.     In theory, the interest rate inelasticity of the demand for reserves means that a change in the target interest rate can be accomplished with a small change in the size of the next OMO, and also that a failure to make the correct size change could result in the new target interest rate being missed badly. In practice, because the central bank's money market counterparties know that the central bank has the market power to correct any persistent miss and is able to impose sanctions on any of them considered to be behaving unhelpfully, the process of setting interest rates is facilitated by a degree of cooperation. Normally, the mere announcement of a change in the target interest rate is sufficient to shift market interest rates to the new target.

19.     Note that the fact that the target and operational debt instruments may differ means that the bounds on the target interest rate may not exactly correspond to the rates at which the standing facilities are offered. In particular, because central banks generally only lend on a secured basis, unsecured inter-bank deposits such as Fed funds could conceivably trade at a slightly higher rate than the standing lending facility. Adding to the difficulty of summarising the stance of monetary policy in terms of a single target interest rate, the debt instruments used in OMOs and standing facilities may well differ, with the standing facilities normally being provided only for the shortest possible term of overnight, while the instruments used for the standing facilities themselves usually differ, if only because the central bank does not give collateral when it takes deposits but requires collateral when lending. Nevertheless, as they have refined their operational techniques to achieve greater control of interest rates, central banks have increasingly emphasised the interest rates they set directly rather than market rates, especially the interest rate in the short-term OMOs they use to make marginal adjustments. This key interest rate is generally referred to as the "bank rate", or "repo rate" if that is how the OMOs are conducted. The standing facilities are typically offered at an equal and rarely-adjusted spread above and below the bank rate to define a symmetric interest rate corridor that moves up and down with the bank rate. In fact, monetary policy is now routinely set in terms of the central operational bank rate rather than some target rate, with market interest rates serving merely as indicators of the effect of the policy and of market expectations of policy changes.

20.     As central banks have placed increasing importance on controlling, and being seen to control, interest rates, a natural development of their strategy has been to change the nature of base money itself to make their job easier. Since it is the interest rate inelasticity of the demand for reserves that makes market interest rates inherently volatile, and since it is the sharp difference between the return on adequate reserves – avoiding a penalty of some size – and on surplus reserves – zero – that causes the inelasticity, an obvious way to reduce the volatility of interest rates is to smooth this disparity. The opportunity cost of holding excess reserves can be reduced by paying interest on reserves. Effectively, this raises the dashed line in Figure 1, so that the money demand curve – or to be more precise, since the return on reserves now differs from the zero return on banknotes, the reserves demand curve – is flatter in the normal range of interest rates. Where there are mandatory reserve requirements, the sharpness of the shortfall penalty can be blunted by applying it to the average level of reserves over some maintenance period, so that a shortfall on one day can be offset against a surplus on another day. Assuming that an overdraft, even for one day, is still unacceptable to the central bank, reserves averaging works better when the reserves requirement is larger and the level of reserves varies around a level well above zero. Paying interest on reserves makes such a larger reserve requirement more acceptable to the commercial banks. For these reasons the European Central Bank and the Bank of England remunerate reserves, and the Fed had been asking the US Congress for permission to pay interest on reserves long before the financial crisis.

21.     To stabilise money market interest rates with maximum effectiveness, the reserves demand curve should be practically flat at the policy rate, which suggests that the interest rate paid on reserves should be close to the policy rate – but certainly not above it, which would establish an arbitrage opportunity. However, paying any interest rate approaching the policy rate encourages the use of reserves as an unbeatably secure investment. The solution adopted by the ECB and the BoE is to remunerate reserves holdings of up to the required size at their policy rate, but pay no interest on any amount above this.

22.     It is important to realise that, because, other things equal, banks can be expected to hold more reserves when they are remunerated, paying interest on reserves changes the relationship between the stock of base money and other economic variables, notably inflation. The introduction of voluntary reserve requirements and remunerated reserves by the BoE, which previously had no reserves requirements, provides a revealing case study of the effect of paying interest on reserves. Prior to the change in regime on May 18th 2006, banks' current account balances at the BoE were normally much less than £1bn. When the BoE began to pay interest on reserves at their policy repo rate, banks' current account balances increased to around £20bn almost immediately – the BoE simply bought more repo debt in its OMOs, on which it charged its repo rate, to accommodate the notified increase in demand for reserves remunerated at the same rate. Since the increase in the stock of reserves was equal to about half the value of the stock of sterling currency in circulation at the time, the "velocity" of base money (essentially the number of times in a year that the money stock would need to change hands to sell annual economic output at prevailing prices) fell by a third, yet the regime change had no noticeable effect on inflation or real output. Analysts accustomed to using base money growth to assess the monetary discipline of a central bank, assuming that the ability to freely switch from reserves to banknotes and vice versa makes them macroeconomically indistinguishable, need to revise their approach when interest is paid on reserves.

23.     Paying interest on reserves represents the apotheosis of interest rate targeting. Market interest rates on short-term, highly creditworthy debt are closely shepherded towards the target by the existence of a large volume of similar assets in the form of central bank liabilities bearing a near-target rate of interest as well as the availability of a large volume of secured loans at a near-target interest rate. To see how remunerated reserves hold up market interest rates, consider the example of a large retailer flush with currency after an unexpectedly successful January sale. The retailer deposits the currency with its bank at its stipulated deposit rate, and the bank pays the currency into the central bank in return for an increase in its reserve balance. The retailer's bank then has the problem of somehow earning as much interest as possible on its additional assets in the interval before it can arrange a loan to another customer. In the absence of interest on reserves, the bank may have to offer to lend at a very low interest rate in order to induce another bank to hold the reserves, which one might just be willing to do if they are aware of the possibility of a large withdrawal by one of their customers and the opportunity cost of holding the reserves is sufficiently low. With interest paid on reserves however, even if the retailer's bank has excess reserves, it can either afford to hold the excess itself and wait for a shortfall later in the reserve maintenance period, or lend the reserves to another bank waiting for a shortfall, at a higher interest rate reflecting the lower opportunity cost of holding remunerated reserves. If a shortfall never emerges, at least one bank may be forced to use the standing deposit facility, perhaps at a modest loss. If the reserves requirements are voluntary, in view of the recent use of the standing deposit facility some banks may specify an increased holding of reserves for the next reserve maintenance period, in which case even a permanent increase in reserves can be accommodated without lower interest rates.

24.     The efforts made by central banks to strengthen their grip on short-term interest rates reflect what has become the mainstream approach of calibrating monetary policy according to short-term interest rates on highly creditworthy debt, rather than some measure of money supply or exchange rate for example. In normal economic conditions, central banks leave interest rates on longer-term debt to be determined by market expectations of the future path of short-term interest rates and the profile across the range of debt terms (ie the term structure) of market term risk premia. Although central banks may try to shape market interest rate expectations by committing to adjust short-term interest rates in pursuit of certain policy objectives and by giving their view of the macroeconomic situation and outlook, they do not normally use their monetary policy market operations to influence longer-term interest rates directly. Similarly, in normal economic conditions, central banks do not attempt to influence the liquidity and credit risk premia, and hence the interest rates, on debt that is less liquid and less creditworthy than the debt they ordinarily buy.

The zero lower bound and quantitative easing in Japan

25.     If the policy rate is a target for some money market interest rate such as the interest rate on overnight inter-bank reserves loans, there will exist some finite amount of zero or low interest reserves that the central bank can supply to bring down this market rate to a positive target, even if the autonomous factors make it hard to specify exactly what that supply of reserves will be. And if the policy interest rate is the central bank's own OMO repo rate, the central bank can of course establish any particular positive interest rate of its choice on that type of debt, provided that it is prepared to lend or borrow in sufficient size at that rate. It is, however, conceivable that in severe economic downturns, the prospects for economic activity and inflation become so weak that the central bank's macroeconomic model suggests that a zero or even negative interest rate is needed to meet its objectives, and in this case it is theoretically impossible for a central bank to engineer a precisely zero (nominal) interest rate using market operations alone. If reserves are available to borrow at zero interest either from the central bank or other market counterparties, a bank's demand for reserves will in theory be unlimited – if it costs nothing to borrow reserves, a bank might as well have a huge holding, just in case an opportunity arises to purchase an asset that offers a greater than zero risk-adjusted return over the same period as the reserves loan. In mathematical terms, in Figure 1, the dashed line is an asymptote, and zero provides a lower bound for the policy interest rate.

26.     In practice, central banks can succeed in lowering their policy rate to zero or even slightly less, partly because of real world limitations such as transactions costs, the convention of dealing in rounded interest rates and the finite capacity of the banking system, and partly because of the banks' willingness to cooperate with the central bank within reason. It might even be possible for a central bank to force its policy rate significantly below zero by imposing restrictions such as charging interest on reserves, but this has not so far been attempted, probably because it would generate undesirable distortions, such as driving banks to use banknotes instead of reserves wherever possible and to settle inter-bank payments outside the central bank.

27.     Once the policy interest rate reaches zero, however, it no longer provides a sufficient measure of the monetary policy stance, because the practicalities of fixing interest rates mentioned in the preceding paragraph combined with the flatness of the reserves demand curve at a zero interest rate mean that a zero policy rate is compatible with a range of sizes of the stock of reserves. This matters because any amount of reserves in excess of the supply necessary to sustain a zero interest rate on the policy debt instrument influences the interest rates on other debt types that are near substitutes for reserves. For example, if the policy rate is a target for the overnight interest rate in the inter-bank market for loanable reserves, when the stock of reserves has increased to the point that the overnight interest rate has been driven to zero (ie reserves available tomorrow trade at parity with reserves available today), a further increase in reserves supply affects the price of the next nearest substitute (eg two-day loans) entirely directly (ie no longer via the overnight interest rate). Similarly, if the policy rate is defined as the central bank's operating interest rate (eg an overnight repo rate), the central bank is able to add to its easing effort without relaxing its grip on the policy rate by, for example, undertaking supplementary OMOs which lend for a slightly longer term or accept lower quality collateral than usual. And when the interest rate on the nearest substitute to the policy debt instrument reaches zero, the central bank can turn to the next nearest substitute and so on, in a cascade of easing. Ultimately, the central bank can buy real assets and even goods and services with base money, so there is no question that a central bank can create inflation if that is considered necessary. As Fed Chairman Bernanke explained in his well-known speech of 21 November 2002, a central bank has not "run out of ammunition" when its policy interest rate has been reduced to zero. On the contrary, it seems reasonable to believe that easing is no less effective at boosting activity and prices after the interest rate on the central bank's primary target debt instrument has reached zero.

28.     Clearly, a precise description of the monetary policy stance when the policy rate reaches the zero lower bound could comprise a vector of interest rates on debt types that are progressively more distant substitutes of the policy debt instrument. A more straightforward approach, which is also consistent with a policy of leaving the market to determine how easing spreads beyond the policy debt instrument, is to express the monetary policy stance in terms of the quantity of reserves. The best-known example of such an easing campaign is from Japan, where between 2001 and 2006 monetary policy was (besides the virtually zero interest rate prevailing in the overnight inter-bank debt market – known in Japan as the overnight call rate) specified in terms of a series of increasingly large targets for the banks' current account balances at the BoJ, culminating in a target of ¥30-35tn (equivalent to about 8% of Japanese GDP). Monetary easing that is expressed in terms of the quantity of reserves held by the banks is, in RebelEconomist's opinion, the most appropriate definition of quantitative easing.

Unconventional policy and the financial crisis

29.     Although the unconventional measures now being applied by the Fed are often described as quantitative easing, Fed policy developed from the opposite direction to the BoJ's QE policy in the sense that the Fed policy started with the aim of easing risk premia and evolved into monetary easing, whereas the BoJ policy began as monetary easing and compressed risk premia as it extended. Fed and BoE unconventional easing began as a response to the financial crisis, at a time when they were holding their interest rates well above zero because of the potential inflationary threat posed by the high commodity prices sustained into last year. In fact, it is questionable whether Fed and BoE unconventional easing could until recently even be properly described as monetary policy, although of course, as the financial crisis has increasingly restrained the real economy, both central banks have also been undertaking conventional monetary easing by cutting interest rates.

30.     At first, the financial problem was seen as one of a lack of liquidity (liquidity being defined here as ease of transacting rather than as, for example, abundance of money). As the US housing bubble burst and it became clear that many marginal borrowers were likely to default, investors shunned mortgage-backed securities (MBS), especially complex structures including sub-prime mortgages, and the debt of any borrower believed to have a significant fraction of its (book) net worth tied up in them. Banks and money market funds that had funded holdings of MBS with short-term borrowing were faced with either paying ruinous interest rates to roll over their debt, especially for any term longer than a day or so, or else selling their MBS at such low prices that they would have become insolvent. The Fed took the view that market prices for MBS and for the debt of their holders were depressed more by an unwarranted rise in investor risk aversion leading to larger liquidity, credit and term risk premia than by a realistic assessment of greater risk, and acted to help the systemically important holders of MBS obtain funding through what was expected to be a temporary crisis. For this purpose, on December 12th 2007 the Fed introduced the term auction facility (TAF), which provided funding for a longer period than the normal Fed OMOs, on the security of lower quality collateral, including certain types of MBS.

31.     The idea was that the TAF would reduce the spread between the interest rates on highly liquid debt such as treasury bills and on bank debt collateralised by lower quality bonds. Besides narrowing spreads on short-term debt, by providing a way of raising liquidity from MBS the TAF could also reduce the spread between MBS and treasuries of similar maturity. However, the Fed set no target for any spread, but instead announced a schedule to auction a fixed quantity of loans. In that way the TAF was a quantitative easing policy.

32.     Note that the TAF was not supposed to ease the credit risk premium. Contrary to what at least one academic has written, a repo loan is not free of credit risk just because it is collateralised by government bonds, and is not necessarily more risky when less creditworthy bonds are accepted as collateral. A repo loan suffers credit loss if the borrower defaults and the collateral is then worth less than the value of the loan, regardless of whether this shortfall arises from interest rate or credit risk. To ensure robust security, an initial margin or haircut is subtracted from the value of the bonds pledged as collateral in determining the amount of collateral required. When less creditworthy bonds are pledged, a larger haircut is applied to cover the expected value of the additional losses associated with credit impairment, so provided that the haircut is assessed correctly, a repo loan collateralised by mortgage-backed securities is no more risky than a repo collateralised by government bonds. Note further that the TAF funding was not ultimately provided by the Fed itself in the form of base money, because the Fed raised the money it lent by either selling or not rolling over maturing treasury bills from the assets it had bought in supplying the existing stock of base money. In other words more commonly associated with central bank intervention in currency markets, the operation was sterilised. In fact, some of its spread narrowing effect was due to the additional supply of treasury bills keeping treasury bill yields higher than otherwise as the Fed reduced its holding. The Fed effectively substituted one asset on its balance sheet for another less liquid asset of similar maturity and credit risk, without changing the size of its balance sheet. The operation might be most appropriately described as substitutional quantitative liquidity easing.

33.     As the financial crisis grew through 2008, with the Fed increasing the size and maturity of the TAF and adding new programmes like the Primary Dealer Credit Facility lending against a wider variety of collateral, the Fed reached the point where it was running out of treasury bills. The Fed could no longer continue to increase its lending against illiquid securities without either selling longer-term treasuries, and thereby probably raising longer-term interest rates including the economically important mortgage rates, or expanding its liabilities and balance sheet. At the time, however, persistently high commodity prices made the Fed reluctant to deliberately expand its non-interest-bearing, monetary liabilities (expansion on the scale required would have driven the Fed funds rate to zero) and limited borrowing powers prevented the Fed issuing securities without permission from Congress. The solution adopted was for the US Treasury to issue and sell new treasury bills and deposit the proceeds at the Fed to fund further lending, in a scheme called the Supplementary Financing Program introduced on September 17th. Now, extra lending by the Fed was matched and funded by a liability in the form of a Treasury deposit, and the Fed had progressed to what may be termed expansional quantitative liquidity easing. As unconventional easing continued apace, the Fed's balance sheet began to grow much faster than previously.

34.     A similar programme, called the Special Liquidity Scheme, had already been introduced by the UK authorities on April 21st 2008. This allowed banks to exchange existing mortgage-backed securities for UK treasury bills with the BoE, although the treasury bills were not owned by the BoE but were specially issued by HM Treasury and loaned to the BoE for the purpose.

35.     Following its discussion with Congress mentioned in paragraph 20, the Fed had in 2006 been granted authority to pay interest on reserves from 1 October 2011. This effectively allowed the Fed to borrow money by issuing interest-bearing liabilities, which offered a more straightforward way to raise funding for liquidity easing, since it avoided the need to continuously coordinate operations with the Treasury. After the demise of Lehman and AIG intensified the financial crisis, Congress allowed the Fed to bring the date from which it could pay interest on reserves forward to October 6th 2008. From then on, Fed purchases of less liquid assets were increasingly unsterilised with the stock of reserves allowed to grow. This new approach could be described as money-financed quantitative liquidity easing.

36.     Previously, creating a large amount of base money to fund expanded asset purchases would have been regarded as dangerously inflationary, but with interest paid on reserves, whose risk-free nature was even more highly valued during a time of financial turmoil, the banks were happy to hold the reserves. The interest-bearing reserves were effectively self-sterilising and providing the funding for the Fed asset purchases that injected them. Since Fed credit is practically the same as US government credit, paying interest on reserves is economically like selling a daily-resetting floating rate note instead of a treasury bill. In short, extra reserves are being created to allow easing rather than to cause it.

37.     Some analysts define quantitative easing as something like "an increase in the size of the central bank's balance sheet through an increase in its monetary liabilities". Under this definition, the Fed had begun quantitative easing once it began to use remunerated reserves to fund its continued balance sheet expansion. But this would be a poor definition of quantitative easing anyway, because in theory any easing involves some expansion of the central bank balance sheet and increase in monetary liabilities, even if the inelasticity of the demand for unremunerated reserves allows this change to be small.

38.     The next easing step taken by the Fed was to start buying lower quality securities outright (ie beyond only taking them as repo collateral), albeit cautiously at first. On September 19th 2008, the Fed announced that it would begin buying short-term debt of the Federal housing agencies. Initially, the Fed limited itself to only indirect exposure to the debt of less creditworthy issuers. The Commercial Paper Funding Facility was established on October 14th 2008 to buy commercial paper via a special purpose vehicle funded and owned by the Fed. On November 25th 2008, the Fed presented a plan for a Term Asset-Backed Securities Loan Facility (TALF) to lend on a collateralised but non-recourse basis to buyers of asset-backed securities comprising student, car, credit card and small business loans – the non-recourse funding effectively involving selling a credit guarantee. With these programmes, the Fed had graduated to credit easing.

39.     Also on November 25th 2008, the Fed announced that it would begin outright purchases of $100bn of longer-term agency securities and $500bn of agency-guaranteed MBS, specifically mentioning its desire to lower mortgage interest rates by this action. At its monetary policy meeting on December 16th, the Fed cut the Fed funds target to 0-¼%, meaning that it had practically exhausted its ability to ease conventionally anyway, and the post-meeting statement said that purchases of longer-term treasuries would be considered. A programme to buy $300bn of longer-term treasuries as well as a further $200bn of agencies and $750bn of agency MBS was duly announced on March 18th 2009. The Fed's easing efforts now included term easing, and were specified in quantitative terms, so it does not seem unreasonable to describe Fed policy as quantitative easing. Although the Fed did not state an intention to deliberately increase the quantity of reserves, the relatively stable demand from the public for banknotes means that, in practice, the proceeds of these asset purchases as they are made through 2009 can be expected to increase reserves by a similar amount. Fed policy has now reached the same point as BoJ QE a few years previously, with the quantity of reserves held at the Fed standing at $925bn (about 6% of US GDP) at the time of writing.

40.     The BoE meanwhile had already adopted money-financed quantitative easing on March 5th 2009, when it cut its repo rate to ½% and undertook to purchase £75bn of longer-term UK government bonds (gilts) and corporate bonds financed by an increased stock of reserves. The BoE suspended reserve requirements and undertook to manage its OMOs so that the stock of reserves would increase above the aggregate of the banks pre-existing voluntary reserve requirements by an amount corresponding to the proceeds of its asset purchases. It is interesting to note that the policy rate was not cut to zero despite the BoE's commitment to increase the stock of reserves. This suggests either that the BoE will accept market interest rates significantly below ½% if necessary (perhaps moving the repo rate down for appearances sake) or that the BoE believes that are sufficient rigidities in the sterling money market to allow the stock of reserves to be increased without OMOs at ½% becoming redundant. The BoE evidently prefers to keep its repo rate sufficiently far above zero to allow some inter-bank lending to continue – if the central bank lends at a rate that does not cover administrative costs and expected credit losses, it would not be viable for a private sector bank to lend surplus reserves.

41.     Although it might seem that every conceivable easing measure has by now been brought to bear on the financial and economic crisis, one, arguably desperate, tool remains. This is the convergence of monetary and fiscal policy, involving the central bank lending directly to the government to fund public expenditure on goods and services. Base money is created when the government draws down the loan. Monetary financing of government expenditure is the traditional definition, and RebelEconomist's preferred definition, of the term "printing money". It is unfortunate that this term has been devalued by being used during the easing campaign to express how readily central banks seem able to create resources to buy valuable assets in general, which is actually no more than the normal function of a monetary authority. Printing money to finance government expenditure would, however, be a big step, because it would mean that the central bank had given up its monetary policy independence and ability to prevent inflation when the need for easing has passed. When a central bank buys assets in the markets, because it pays the market price in base money, it should be possible for a well-capitalised central bank to reverse this position by re-selling these assets if and when it considers this necessary to bolster the value of its currency. When the central bank prints money, however, the exchange of base money for government debt is typically off-market, meaning that the debt assigned to the central bank is unlikely to be marketable and is not guaranteed to have a value as large as the base money credited to the government. So far, during this crisis, no developed country central bank has yet been obliged to print money.


Anonymous said...

Glad to see Buiter's plagiarism hasn't deterred you from posting!

Devin Finbarr said...

Awesome post, this was really informative.

Quantitative easing has never quite made sense to me. If I understand it correctly, the goal of quantitative easing is to reflate and increase aggregate demand, thus preventing a sharp recession or depression.

But for an individual or a company, spending is determined by the balance sheet. When I log into my investment accounts, and its higher than expected, I can go on a nice vacation. If it's too low, I have to cut back spending. This is how every company and individual operates.

Thus if a central bank buys up assets at market prices, it is not reflating. If I am a retiree, and the Fed buys up my house at market price, I cannot suddenly start spending more money. The market price of the house was already factored into my long term spending plans. Selling the house at market price has not changed my balance sheet, and so I cannot suddenly increase my spending.

The only way quantitative easing can reflate is if the Fed buys at above market prices, or buys enough to push market prices up. But if it does this, the Fed has engaged in a massive transfer of wealth to the owners of these assets. In my opinion, financial crises should not be used to transfer wealth and real resources from one person or institution to another. A reflation should be as neutral as possible.

Modern money (as opposed to commodity money such as gold coins) is debt; a readily transferable and widely accepted type of continuously redeemable debt.Debt means an obligation to pay a quantity of good X at time Y. If money is the debt obligation, what is the good X?

In my understanding, the dollar is not debt, it is simply a fiat currency. A dollar is a collectible, an intermediary good.

Base money is created when the government draws down the loan.Do you have any idea what the legal limitations are on the Fed and the Treasury with regards to print money? If the Fed wishes to create base money and buy assets, can it do so at will? Or is that considered an extraordinary measure that requires Congressional approval? And when a treasury bill matures, does it simply become base money? Or does the money come out of the Treasury's account at the Fed (with risk of overdraft if the necessary money is not there)?

The reason I ask these questions is that a lot of the Fed's actions seem like elaborate schemes to print money without actually printing money. I was wondering if this is because the Fed is pushing the legal limits of what it is allowed to do.

RebelEconomist said...

Thanks Anon,

I must admit I was knocked back by Willem Buiter’s refusal to acknowledge my early quite developed proposal of the good bank spin-off idea. My nom de plume reflects a history of making myself unpopular as a central banker by questioning comfortable thinking (eg about the danger posed by rising asset prices) and as an academic by preferring an accessible style of analysis and teaching (eg minimal use of equations). I therefore write a blog as a way of presenting, free of the constraints of these two fields, challenging but hopefully reasonable economic ideas in a style that tries to be rigorous yet readable. I don’t expect to attract a large audience – I lack the productivity to sustain their attention anyway – but I would hope that enough interested readers see my ideas to give them a chance of being taken up, and that my posts at least provide a public account of the ideas that I can refer to in comments on more famous blogs. So it was demoralising to find that, when one of my ideas did become popular, established commentators, even if they arrived independently at the same conclusion as me, would not acknowledge that an unrecognised outsider had travelled the path well before them. I particularly resent such behaviour from Buiter, because he is often critical of what he sees as insider privilege and, I note in his recent post on Google, breaches of copyright. Moreover, occasional misconceptions in Buiter’s writing – for example, that mentioned in paragraph 32 here – suggest that he can learn from a wider range of contributors.

RebelEconomist said...

Congratulations on getting through the post, Devin, assuming that you read the whole thing!

One should be careful with the use of the word "reflate". To me, to reflate means to increase the amount of money circulating to raise a wide range of prices – if not the general price level – that have previously fallen. In other words, re-inflation. If this was what the Fed was trying to do, the key would be to increase the stock of money rather than targeting particular types of asset. Given that the base money supply directly controlled by the Fed is small in relation to the total stock of US assets, the Fed could supply more than enough money to reflate by spreading their purchases over such a wide variety of assets (eg stocks, real estate, public infrastructure etc) that the impact of Fed buying on their relative prices would be negligible. Given that the stock of currency in circulation is about $900bn, one might expect for example, that, creating another $900bn of currency might double the general price level, which would probably be more than enough to ensure that practically all prices rise in nominal terms and that the toxic asset problem would disappear. I know it is not as simple as that, and that the demand for base money might change, but the point is that, if they really want, the Fed can blow up the general price level without an unfeasibly large amount of asset purchases. It would be a case of pressing the inflationary reset button.

While I would not be surprised if the Fed is eventually forced to press the inflationary reset button, this would create other problems, so it is not what the Fed is trying to do now. What the Fed is trying to do now is to raise certain economically influential asset prices such as the price of mortgage debt, without, officially at least, raising the general price level. For this purpose, the Fed does want to buy enough of these assets to raise their market prices in both absolute and relative terms. Ideally this would boost aggregate demand via reviving depressed parts of the economy such as construction. While you are right that this benefits the owners of mortgage debt and houses, the Fed would no doubt say that (a) the prices of these particular assets are unrealistically depressed so that it is case of mitigating unfortunate losses rather giving their owners a windfall and (b) that any distortion is a price worth paying to avoid depression. While I am wary about expressing controversial opinions without supporting analysis, my inclination is to agree with you – the Fed is bailing out reckless investors at the expense of those who did the right thing, and this represents the continuation of a pattern of bailouts sustaining a moral hazard culture that will eventually make the US economic situation worse. Personally, I think that the option of a Mellon-style liquidation, albeit followed by aggressive monetary and fiscal easing if necessary, was dismissed too readily. But then I do sometimes say that I was too conservative to be a central banker!

You are right that the dollar is not defined in real terms, but that does not mean that the dollar does not have real value. The dollar has real value because it can be used to pay US taxes, which entitles the citizen to various government services. In theory, a fiat currency should be superior to, say, a gold standard, because it does not depend on the relative value of a single unproductive commodity. But it is essential for a fiat currency to be controlled by highly principled people, and that, in practice, is the Achilles heel of fiat currencies.

I am not an expert in the legal powers of the Fed, but I believe that the Fed has no constraints on its power to issue base money, other than its statutory objectives of low inflation, unemployment and long-term interest rates. A treasury bill does not imply the forward creation of base money; in fact its sale and maturity proceeds need not even be paid in base money if the government banks outside the central bank.

I do not think that the numerous Fed easing programmes are stealth schemes to print money, but I do believe that they could make printing money inevitable, and I do agree that the Fed is getting involved with the government in ways that it should not do. If the elected government believes that public support for, say, the housing market, is desirable, it should either raise taxes or borrow to pay for that support. For example, assuming that a tax rise to fund the purchase of MBS to drive down mortgage rates would not be considered appropriate in a recession, the government should sell treasuries and buy MBS on its own account. This presents two problems. One problem is that the need to raise the debt ceiling would force the government to convince Congress to approve the scheme, and its approval might be withheld or come with strings. The second problem is that, if the scheme makes a loss, the government would be forced to either raise taxes or borrow more to cover the cost. Getting the Fed to buy the MBS funded by interest-bearing reserves circumvents the debt ceiling and allows any losses to be stealthily absorbed in reduced seigniorage payments from the Fed to the US Treasury. Nevertheless, any losses do ultimately make the government's fiscal situation worse, which increases the temptation to press the inflationary reset button referred to above.

By the way, I checked out your blog, and you have a dubious link in the most recent comment!

Devin Finbarr said...


Given that the stock of currency in circulation is about $900bn, one might expect for example, that, creating another $900bn of currency might double the general price level, which would probably be more than enough to ensure that practically all prices rise in nominal terms and that the toxic asset problem would disappearI am a little unclear how creating $900bn of base currency would double the price level. For instance, let's say that the Federal Reserve creates $900 billion out of thin air, and then mails every adult American a check for $4,000. I can see how this would increase the price level, but not double it. Four thousand dollars is still just a fraction of average annual income. People would not be able to suddenly double their spending.

If everyone immediately deposited the checks, bank reserves would double. But in order maintain the Federal Funds rate at the desired level, wouldn't the Fed then have to engage in open market operations to remove the excess reserves? Or - in other words - is increasing the base to stimulate inflation the exact same operation as lowering the Federal Funds rate?

And even, if the Fed did not immediately pull out the reserves, aren't the banks still constrained by capital requirements and the lack of credit worthy borrowers? So wouldn't the reserves just pile up and sit there?

RebelEconomist said...


You are picking the nits that I wanted you to avoid (by saying "I know it is not as simple as that")!

My point was just that the stock of assets that the Fed could buy to supply base money is so large that the Fed could expand the base money stock many fold without having a huge impact on the (relative) market price of those assets (eg they could spread their purchases over stocks, foreign assets etc). In other words, a policy of general reflation can be distinguished from a policy of lowering interest rates on particular types of debt. Arguably, the BoJ tried the former and the Fed has so far been trying the latter.

Could a policy of general reflation work? I think it could, but as you say, not unless the Fed set its policy rate to practically zero - which they now have. You cite the extreme case of what I call printing money (paragraph 41) - the Fed lends to the government - in which case the base money supply can be expanded without limit. Even if the initial inclination of people receiving cheques from the government is to simply hold their additional dollar wealth (meaning that their bank holds extra reserves and the Fed funds rate is allowed to fall to practically zero), eventually, people will receive enough money that they will start to spend. Note that it is not necessary for the banks to lend for this money to be used - it can either pass from person to person via bank accounts (in which case the banks involved transfer reserves when the payments are settled) or it can be withdrawn as banknotes (as in Zimbabwe). Nor is bank capital a constraint when (zero risk-weighted) reserves are the addition to banks' assets. I have no doubt that the Fed can inflate if they want to.

But let me repeat, I think the Fed would be unwise to inflate. It may mitigate the immediate economic problems, but in the medium run, inflation would make matters worse. First, overseas lenders will demand a higher interest rate to lend in dollars, so unless the US is prepared to stop borrowing (including rolling over existing debt), Americans will pay the price for inflationary repudiation of their debts (in fact this was the subject of my first ever post). Second, as I said in my previous comment, pressing the inflationary reset button would extend the moral hazard that, in my view, is the major cause of the present crisis. I lived through such a period in Britain in the 1970s, and my conclusion was that it was worse than futile.

Devin Finbarr said...

The reason I was picking at the nits, is that some monetarists seem to think that creating a small amount of base will lead to a much bigger inflationary impact. This did not seem right to me, so I was trying to ask you because you seem to know more about the mechanics of the reserve system.

But let me repeat, I think the Fed would be unwise to inflate. It may mitigate the immediate economic problems, but in the medium run, inflation would make matters worse. The logic behind reflation is pretty clear cut. There have been a cascade of collapsing bubbles ( debt, housing, and stocks) wiping out 20-50% of paper wealth. As a result, demand for dollars has increased dramatically as people repair their expenditures to paper wealth ratio. But doing so requires cutting spending on luxuries and durables leading to massive unemployment. The problem is purely nominal, the Fed can simply directly alter people's balance sheets to fix the ratio without any need for unemployment.

The cleanest way to inflate would simply be to make an accounting of all paper backed by the full faith and credit of USG ( reserves, FDIC insured accounts, treasury bills, perhaps money market funds), and then increase the face value by 20-30%. I wake up tomorrow and my CD says $13K instead of $10K.

In this way you can restore the paper wealth to expenditures ratio, without the side effect of destroying demand for dollars or T-Bonds.

JKH said...


You’ve produced a very good post here; coherently structured, informative, and well written. The blogosphere is weak on central bank operations; a lot of money theory with tidal waves of policy criticism and prescription flowing from flimsy operational grasp. By contrast, this is an excellent reference piece.

Since I agree with most everything you’ve written, I have just a couple of brief comments.

First, I think it’s important to be clear when one is referring to the banking system versus individual banks within it. Nothing specific here in terms of what you’ve written, but in my own case I often think/write about system dynamics implicitly, assuming away individual bank dynamics. That can be confusing to others. But the distinction is crucial in understanding how banks respond individually and collectively to central bank policy.

Second, I believe that there is broad failure to differentiate properly among different elements of the monetary base. You’ve done a good job, although I would push the point further. The various pieces behave in different ways that are critical to understanding monetary policy at both the operational and strategic levels. E.g. it is impossible to understand properly what’s going on in the current environment without rigorously differentiating between excess reserves, required reserves, and currency. I’ve given up pointing this out to people, particularly when they insist that such differentiation is not important. Since you understand how a central bank balance sheet works, you probably know what I’m driving at.

As a minor observation, I find the definition of “quantitative easing” to be a mug’s game. The varied application of the term is symptomatic of superficial treatment of central bank operations. There aren’t any shortcuts in describing what they do and how they do it. Witness the length of your post. For what it’s worth, my preferred definition is any increase in excess reserves greater than or equal to one cent. But that’s as much an expression of disdain for the phrase, including its original application in the case of Japan, as it is a preference for unambiguous meaning. The term itself is an insult to the true integrated nature of central bank asset-liability management.

Finally, I was pleased to see your interpretation of the reserve multiplier:

“Note that interest rate targeting means that the textbook account of deposit money creation, in which the central bank supplies base money followed by an iterative process of bank lending and customers re-depositing base money that creates some multiple of the initial injection of base money, is unrealistic. Actually, money creation is generally initiated by bank lending, with lending generating deposits and deposits in turn generating a need for reserves which are reactively supplied by the central bank in order to fix its targeted interest rate.”

Perhaps my memory is wrong, but I didn’t think you held such a view at the time of our long discussion on Setser’s blog. In any event, it doesn’t matter. What you’ve written on this issue and in total is very good.


RebelEconomist said...

I am intrigued by your comment Devin, but not quite sure what to make of it. If the nominal value of all paper backed by the US government was increased, then who would bear the cost? This is a particularly important question in the case of an FDIC-backed deposit. Presumably you do not intend the bank or its borrowers to have to come up with 30% greater redemption proceeds? But if not the bank or its borrowers, do you mean the central bank, backed by the government? I dare say that such a scheme would involve creating a lot of money and would certainly generate inflation, but many creditors (eg China) would come out at least whole.

RebelEconomist said...

I am grateful for your attention and appreciation, JKH.

I tried here to come up with an explanation of easing that worked whether the lower constraint on a bank's holding of reserves was provided by a regulatory reserve requirement or a ban on end-of-day overdrafts; maybe that was an abstraction that did not work well.

To me, the definition of quantitative easing ought to justify the use of the word "quantitative", so I do not like your definition (surely even under conventional monetary policy there is some relationship between the volume of excess reserves and the target interest rate, which would, by your definition, make any easing quantitative) but I do agree that the term has not been carefully used. I think the term applies best to the BoJ policy. Incidentally, watching that policy evolve as an investor in the JGB market got me interested in the details of monetary policy operations.

I have not forgotten our long discussion about how central banks control interest rates on Brad Setser's blog in May 2007 and my thinking is still much the same. I made the mistake of mentioning the money multiplier in explaining my understanding to you though! The interpretation of the multiplier that you liked here was deliberately written to be acceptable to you! If you recall, the main point I was trying to make then was that I could not see how central banks could control interest rates without (a) becoming one side of the market, and (b) losing money in the process. And I would say that what subsequently happened - a massive expansion of the Fed's balance sheet as it tried to drive down interest rates and an unresolved problem of how to emerge from QE - has confirmed my thinking. But that is a subject for another post, for which this one is meant to provide some preparation.

Devin Finbarr said...

If the nominal value of all paper backed by the US government was increased, then who would bear the cost? This is a particularly important question in the case of an FDIC-backed deposit. Presumably you do not intend the bank or its borrowers to have to come up with 30% greater redemption proceeds?

No, the Fed would print enough money to back the 30% increase in deposits and proceeds.

I dare say that such a scheme would involve creating a lot of money and would certainly generate inflation, but many creditors (eg China) would come out at least whole.

The point is to counter the existing deflation. The existing deflation is caused by shock in the demand for money, and that must be met with a corresponding increase in the supply of money. Note that inflation is usually bad for two reasons:

1) It transfers wealth from savers to the inflater
2) It destroys demand for the dollar, which creates bubbles, financial instability, and potentially a currency run

The plan I have posited has neither drawback. Because the money is injected into the hands of current holders of USG liabilities, in direct proportion to their current holdings, it neither transfers wealth nor penalizes dollar holders. Thus the Fed can counter the deflationary spiral, without creating nasty side effects.

RebelEconomist said...


This makes a lot of sense to me. In fact, I had a similar idea as an alternative to the UK's VAT cut - cut the withholding tax on savings interest instead, to narrow the gap between deposit and lending interest rates. An interest subsidy as you propose is probably a better idea, because its value is not less (in fact, proportionately more) at lower interest rates. A withholding tax cut would only benefit domestic taxpayers, which may or may not be an advantage.

Adam P said...

Rebel, I left you a couple of responses at WCI if you're interested:



Salmo Trutta said...

I've have a couple of minor criticisms. "money demand curve" This is old school. It's more accurate to talk of the demand for and the supply of money, because in your sense, the meaning is just the opposite.

I.e, Alfred Marshall: money thus is truly a paradox - by wanting more, the public ends up with less, and by wanting less, it ends up with more. All motives which induce the holding of a larger volume of money will tend to increase the demand for money - and reduce its velocity.

Interest rates are the price of loan-funds. Interest rates are not the price of money. The “price of money is the reciprocal of the price level.

It was apparently this confusion (Keynesian training), that led to the assumption that interest rate manipulation could achieve the proper level and rate of growth in the money supply.

Salmo Trutta said...

The money supply can never be managed by any attempt to control the cost of credit.

RebelEconomist said...

Thanks for the comments Flow5 - I was surprised to find someone reading this post after so long!

I sometimes find your comments a bit cryptic, but I think I see what you mean here. Base money might be supplied by trading it against one particular type of debt, but the terms of that exchange will be affected by the the exchange rates between money and all other items that might be traded against money. I do agree with you that the "price of money" strictly ought to be expressed in terms of the prices of all these items.

RebelEconomist said...

Note to self:

The Bank of Japan began to pay interest on non-required (ie excess) reserves following an announcement on 31 October 2008. When introduced, the scheme, called the "complimentary deposit facility", was described as temporary and an expiry date of 16 March 2009 was set, but the life of the scheme was subsequently extended repeatedly and it remains in operation at the time of writing.

guitry@ca.rr.com said...

I am sorry to be so stupid but please someone explain to me just one thing: What EXACTY does the Treasury receive when they SELL bonds to the Federal Reserve Bank that is different than what it receives when I buy a Treasury bond?

RebelEconomist said...

Not stupid at all. In both cases the Treasury receives credit in their current account at the Fed.

Although you, as, I assume, a retail investor, do not have a current account at the Fed, your bank probably does (and if not, it will have a current account at a correspondent bank that does). When you buy a bond from the Treasury, you write an instruction (eg a cheque) to your bank to pay the Treasury, and, in turn, your bank instructs the Fed to pay the Treasury (ie to move credit from the bank's current account at the Fed to the Treasury's), while debiting your current account with your bank for the value of your bond purchase.

What can get confusing is that credit in a bank's Fed current account counts as reserves, whereas credit in the Treasury's Fed current account does not. The idea is that banks' Fed current account credit is effectively money which the banks and their customers can choose to circulate, and may therefore have an influence on prices generally. However, I find it hard to imagine why the Treasury would be selling bonds to obtain reserves that it does not intend to spend, and thereby pass on to, the non-government sector, so I would say that the distinction between Fed current account credit held by banks and by the Treasury is of limited economic meaning. (One other minor difference is that the Treasury does not receive interest on its Fed current account credit, whereas the banks do).

So, hopefully I have explained that there is no difference in the nature of what the Treasury receives in either case. I suspect, however, that you are really wondering why it matters whether the Treasury sells a bond to the Fed rather than to the private sector. It matters because when the Fed itself pays, it typically creates current account credit that did not previously exist ("typically", because an alternative possibility is that the Fed could sell its own bonds to obtain current account credit from the private sector), and hence (as soon as the Treasury spends that current account credit) reserves and effectively money. And, other things equal, more money chasing the same number of items means higher prices - ie inflation. Naturally, other things are not equal. Given the present lack of economic confidence, the public are more than usually inclined to hold money than to circulate it, so that Treasury borrowing from the Fed may not be inflationary for now, but if the economy improves, to avoid inflation, the Fed would probably have to sell the bonds it bought from the Treasury to withdraw reserves.

guitry said...

Thank you, Rebel, for your clarifying reply. That leads me to think that BB is being disengeneous when he states that he is not printing money but merely adding to the Fed's reserve.

RebelEconomist said...

Actually, in this case, I agree with Bernanke. I discuss and define "printing money" in paragraph 41 of the post, and it remains the case that no major central bank has yet consented to print money according to my definition. Because the demand for safe assets like reserves is high due to financial uncertainty and because reserves bear some interest, the "moneyness" of reserves is presently quite low, so QE has not obviously been inflationary. QE can be seen as an exchange of assets, in which the Fed exploits the elevated demand for reserves to buy longer term and less creditworthy bonds, in the hope that lowering their yields will prompt more issuance of such bonds by the private sector and hence boost economic activity. That is a reasonable policy for now, but Bernanke's real test comes when QE is no longer needed, and reversing the asset swap is necessary to avoid inflation, but probably costly.

guitry said...

I see what you mean but the definition of not printing money assumes that the MBS's which the FED bought are "remarketable" on at least a break-even basis. I believe that this is the concensus. Let us hope so. Thanks again for your answers.

Anonymous said...

You say "money supply" is confused with a flow. But this is exactly what it is. The monetary base, if called "money supply" has been inaccurately named. It is not a "supply".

RebelEconomist said...


If you read paragraph 5 again, you will see that I am making a generic point about all definitions of money stock. They are often called money supply, but none of them are flows. It is just an attempt to dispel all possible confusion before moving on in what is supposed to be an explanation from first principles.

genauer said...

off topic, but I should have asked this like half a year ago.

Reading economics papers, often more for implementing, I came repeatedly across severe errors, not just typos, missing footnotes ....,
but lethal construction problems, IMHO.

I would be very interested, if somebody versed in the field would take a look at my argument, and maybe tell me what I got wrong.

1. Cobb-Douglas 1928
2. Solow 1957
3. reception of that by Barro, Krugman 1992, in references to an intermediate paper, I have to find

Interested ?

Or any suggestion, where, who might be helpful with that?