Here are his key summary points from that article:
First, banks are not just financial intermediaries. The act of saving does not increase deposits in banks. If your employer pays you, the deposit merely shifts from its account to yours. This does not affect the quantity of money; additional money is instead a byproduct of lending. What makes banks special is that their liabilities are money – a universally acceptable IOU. In the UK, 97 per cent of broad money consists of bank deposits mostly created by such bank lending. Banks really do “print” money. But when customers repay, it is torn up.
Second, the “money multiplier” linking lending to bank reserves is a myth. In the past when bank notes could be freely exchanged for gold, that relationship might have been close. Strict reserve ratios could yet re-establish it. But that is not how banking operates today. In a fiat (or government-made) monetary system, the central bank creates reserves at will. It will then supply the banks with the reserves they need (at a price) to settle payments obligations.
Third, expected risks and rewards determine how much banks lend and so how much money they create. They need to consider how much they have to offer to attract deposits and how profitable and risky any additional lending might be. The state of the economy – itself strongly affected by their collective actions – will govern these judgments. Decisions of non-banks also affect banks directly. If the former refuse to borrow and decide to repay, credit and so money will shrink.
The act of saving does not increase deposits. If your employer pays you, the deposit merely shifts from its account to yours
Fourth, the central bank will influence the decisions of banks by adjusting the price it charges (the interest rate) on extra reserves. That is how monetary policy works in normal times. Since it is the monopoly supplier of bank reserves and since the banks need deposits at the central bank to settle with one another, the central bank can in this way determine the short-term interest rate in the economy. No sane bank would lend at a rate lower than it must pay the central bank, which is the banks’ bank.
Fifth, the authorities can also affect the lending decisions of banks by regulatory means – capital requirements, liquidity requirements, funding rules and so forth. The justification for such regulation is that bank lending creates spillovers or “externalities”. Thus, if many banks lend against the same activity – property purchase, for example – they will raise demand, prices and activity, so justifying yet more lending. Such a cycle might lead – indeed often has led – to a market crash, a financial crisis and a deep recession. The justification for systemic regulation is that it will, or at least should, attenuate these risks.
Sixth, banks do not lend out their reserves, nor do they need to. They do not because non-banks cannot hold accounts at the central bank. They need not because they can create loans on their own. Moreover, banks cannot reduce their aggregate reserves. The central bank can do so by selling assets. The public can do so by shifting from deposits into cash, the only form of central bank money the public is able to hold.
Finally, quantitative easing – the purchase of assets by the central bank – will expand the broad money supply. It does so by replacing, say, government bonds held by the public with bank deposits and in the process expands the reserves of the banks at the central bank. This will increase broad money, other things being equal. But since there is no money multiplier, the impact on the money supply can be – and indeed has recently been – modest. The main impact of QE is on the relative prices of assets. In particular, the policy raises the prices of financial assets and lowers their yield. The justification for this is that at the zero lower bound normal monetary policy is no longer effective. So the central bank tries to lower yields on a wider range of assets.
This is not just academic. Understanding the monetary system is essential.
These ideas are not new. But Wolf summarizes them well.
Market monetarists have had some problems with this paper, which is understandable given their emphasis on monetary aggregates rather than interest rates as the appropriate framing for monetary policy and theory. Nick Rowe in particular has written a series of outstanding posts that help connect the market monetarist dots on this issue. Scott Sumner also. And David Glasner opines on aspects relating to the paper and the associated debate. Rather than dwell on the details of what various people have written on this subject (market monetarists and others), I’d like to offer a general impression of the way in which market monetarism and like-minded interpretations respond and object to this sort of characterization of monetary policy.
Market monetarism (or perhaps monetarism more generally) embraces two analytical orientations that inform its worldview. The first is the tendency to view the monetary base as a single entity rather than a complex composition of reserve balances held by banks and currency held mostly by the non-bank public. The second is the tendency to subsume central bank short term interest rate operations within a longer term view of how the (undifferentiated) monetary base affects the broader monetary aggregates, price level, NGDP, and employment.
Market monetarists associate the detailed operational characteristics of central and commercial banking with a myopic perspective on monetary policy. This is the case especially as it relates to the central bank’s role in setting the short term policy interest rate target. They look at the monetary base as a whole and its ultimate effect on the economy, rather than on the nuts and bolts of central banking interest rate mechanics and its transmission through the banking system.
The monetary base is defined as the totality of central bank monetary liabilities issued to the private sector – which includes bank reserve balances and currency. The central bank issues these two different liability types mostly for the benefit of two different institutional stakeholders. Bank behavior in using reserve balances is different from non-bank behavior in using currency – because of institutional differentiation and purpose.
Reserve balances are used by banks to clear customer payments for both asset allocation and GDP expenditure behavior (largely via bank deposit liability activity) across the financial system. Banks also participate as principal actors primarily in the asset allocation process via their reserve accounts. Currency is used almost entirely for GDP expenditure activity – not so much as the medium of exchange in asset allocation decisions. Thus, reserve balances are inextricably linked to financial intermediation and currency is not. That should tell us something. It is notable that the sea change in the relative proportions of reserve balances and currency due to central bank quantitative easing in recent years has not been associated with any change in this normal alignment of functionality.
Market monetarists tend to dismiss the importance of banks in monetary theory. The construct of a seamless “base” becomes a catalyst for this perspective – because it does not differentiate overtly between banks and non-banks. One particular thought experiment often invoked by monetarists is the imagined substitution of currency for electronic reserve balances. But this only begs the question. The issue is not the physical or electronic difference between reserve balances and currency. The issue is the identification of the two different parts of the monetary base held by two different types of users. This is important because supply and demand functions differ as between reserve balances and currency.
For example, in the short term, the central bank has no choice but to supply the quantity of currency demanded by the public. The public acquires currency from the banks by paying for it with deposits. The banks acquire currency from the central bank by paying with reserve balances. The central bank supplies the quantity demanded through this acquisition sequence. This supply function is fully elastic to the prevailing demand. This is a fact of banking operations. In the longer term, market monetarists interpret that the central bank determines this currency demand function as part of total base demand – which results from policy that somehow influences that demand.
Also, in the short term, central banks control policy interest rates. This is the view that is expressed in the Bank of England paper and which seems to be the cause of some consternation among market monetarists. They eschew the description of such a mechanism as a legitimate characterization of monetary policy. Market monetarism interprets that in the longer term the central bank targets things beyond the short rate (e.g. inflation, NGDP) and responds to larger forces which determine the required path of the short rate – which in turn is a function of the path of the monetary base that achieves the desired monetary policy. Conversely, those observing from the operational perspective interpret this longer term connection directly as a succession of short term interest rate control decisions (consistent with inflation targets or other objectives).
The different institutional constituencies come to satisfy their needs for the two components of the monetary base in very different ways. The case of currency was noted. In the case of reserve balances, the central bank has a greater degree of operational and policy freedom in responding to demand than it does for currency. For example, in the case of a central bank channel reserve system, with upper and lower bound standing facilities, the central bank can shift its reserve balance supply to meet a desired policy interest rate target. In the case of quantitative easing, the central bank can set a floor rate for the payment of interest on reserves, enabling it to meet its interest rate target under any quantity of reserves supplied. But there is no comparable short term pricing mechanism in the case of currency, because bank deposits are redeemable at par on demand in exchange for currency.
The robust nature of this monetary base bifurcation is most evident in the extraordinary reversal of the relative shares of currency and reserve balances that has occurred in the US banking system over the period of the financial crisis and the consequent long term asset purchase program (LSAP) of the Federal Reserve. Most observers are familiar with the famous graph that delineates the multi-trillion dollar expansion of excess reserve balances held at the Fed – up from a pre-2008 starting point of only several billion dollars. Over the same period the Fed has transitioned from an asymmetric reserve channel system to an ad hoc floor system where interest is paid on reserve balances in order to manage interest rate target levels while excess reserves have increased a thousand fold.
LSAP (large scale asset purchases) is an unconventional channel for the injection of outsized excess reserve balances into the commercial banking system (otherwise known as quantitative easing or QE). Previously, reserve balances were provided or withdrawn through central bank “open market operations” (OMO), or something similar, in which the central bank transacts directly with a dealer network that in turn works in close proximity with the commercial banks. Indeed, most of these dealers typically are owned by the commercial banks as part of a “universal bank” holding company structure.
The supply/demand configurations for OMO and QE are very different. The shortest term supply/demand configuration for OMO within a channel system of any type comprises a nearly vertical inelastic demand function and a perfectly vertical supply function. Moving beyond that OMO frequency range, between the meetings where policy interest rates are confirmed, the short term policy rate configuration includes a horizontal supply curve at the target interest rate. The QE demand function is different again – horizontal at the support floor rate. The supply function is vertical but with almost unlimited shifting flexibility. These are all very different dynamics for reserve balance supply and demand, depending on technical regime and time horizon.
QE widens the distribution of money connected with reserve balance injections in two important ways. The first is that it expands the distribution system for the immediate effects of money injection to a much broader range of market participants, well beyond that reached by normal OMO, including pension funds, insurance companies, mutual funds, and the like. The second is that the reserve injection has a significant material effect through the “back door” of bank deposit liability expansion. Most of the bonds sold into the central bank as part of QE originate from non-bank portfolios.
Indeed, it is the creation of broad money deposits held by non-banks that is the primary money transmission channel for the eventual QE effect on interest rates. The front door effect of reserve balances held by banks is less important, as explained by the Bank of England paper. The back door channel is the route for the effect on long term interest rates – because that is where most of the portfolio management decisions take place that have to do with this aspect. The front door effect is muted – in accordance with the general discrediting of the ancient “money multiplier” theory.
Thus, the Federal Reserve period of quantitative easing through LSAP has resulted in an extraordinary reversal of the historic quantity relationship that has existed between reserve balances and currency. This phenomenon requires more explanation that can be provided naturally by a monetary theory that prefers to consolidate reserve balances and currency into a monetary base lump and which favors a background role for banks. A long term view should not ignore differentiation in monetary operations and institutions.
Nick Rowe says:
Central banks control their own current and promised future balance sheets. Any influence they have on interest rates, and inflation, and anything else, all ultimately derives from changing the size and composition of their own balance sheets. Targeting interest rates is not what central banks really really do, for you people of the concrete steppes. Interest rate targets are merely a communications device (and a bad one at that), that some central banks sometimes use (see Paul Krugman) as an intermediate step between the machine language of balance sheet quantities and the ultimate target of inflation (or whatever). Nobody should confuse a communications device for ultimate reality.
Central banks control the quantity of reserve balances supplied in the short run but they do not control currency in the same way – because they respond to quantity demanded with a fully elastic supply. Yes, they control reserve balance adjustments they deem appropriate taking into account the quantity of currency they are asked to supply, but interest rate objectives play a critical role in determining just how that control function is executed in the short term.
Also, interest rates are more than a mere communications device in support of quantitative balance sheet operations. The interest rate or yield on a bond or a reserve account is no less important than the quantity of money to which it applies. In monetary policy terms, communication includes interest rate settings as well as inflation or NGDP targets. All of it constitutes communication. The issue is really the interconnectedness and ultimate effect of these various modes of targeting and how well that is communicated and understood.
So while Nick has a point that monetary policy is not just one damn interest rate after another, the framing for the role of interest rates need not be so asymmetric and exclusionary. Moreover, the short term operational perspective on monetary policy has the advantage of being immediately observable and verifiable. The question is how a series of short term operational decisions is best explained by an initial long term view – of which market monetarism is one brand.
Scott Sumner says:
The BoE controls the base in such a way as to target interest rates in such a way as target total spending in such a way as to produce 2% inflation. And yet in that long chain interest rates are singled out as “monetary policy” … Lots of people use the term ‘reserves’ when they would be better off using the term ‘monetary base.’ Back in 2007 the part of the US monetary base that was “coins” was larger than “bank reserves.” So it would have been more accurate to talk about central banks injecting coins into the system. And prior to 2008 new base money mostly flowed out into currency in circulation within a few days, even if the first stop was the banking system. Banks were not important for monetary policy, although of course they were a key part of the financial system.
It’s not clear that an interest rate focus on monetary policy is any more exclusionary in considering broader economic targets such as inflation than is the case with a monetary base focus. Most who describe monetary policy in terms of interest rates are quite aware that economic considerations form the rationale for central bank interest rate adjustment. It just so happens that the market monetarists have hitched their wagon to the targeting of NGDP by monetary base means. But that involves two quite separate things – the choice of NGDP as the ultimate economic target and the choice of the monetary base (instead of interest rates) as the implementation focus. The focus on interest rates is typically associated with a more status quo choice for economic targeting – such as inflation rather than NGDP. So I think the criticism here obscures what really is a comparable relationship between monetary operations and policy targeting, even if the choice of a more status quo economic target tends to be emphasized less frequently than the choice of a more unconventional one such as NGDP.
Market monetarists have been active recently in commenting on the issue of “the money multiplier”. Whatever they are saying about, it certainly doesn’t resemble the money multiplier process described in some of the older economics textbooks. That’s the one where excess reserves suddenly appear and banks proceed to “lend” those reserves in a geometric series that exhausts the surplus by creating deposits which gradually cause the conversion of those excess reserves into a required reserve classification. It depicts a process in which central banks first supply excess reserves and then those excess reserves get multiplied into loans and deposits. That is both empirically wrong in the order in which new deposits are matched up with the first appearance of corresponding reserves and causally wrong in the way in which banks make decisions to lend. The Bank of England and a host of others have explained the nature of this sort of error. The objection is perfectly rational. It simply doesn’t work that way and never has.
The “newer” Market Monetarist rendition of the multiplier seems in general to present it as a sort of long term portfolio balancing of the monetary base against the topology of broad money. This can include intra-base balancing of the currency component of the base with broad money deposits. Whatever. This sort of portfolio balance and asset mix interpretation seems a bit of a stretch in rescuing a discredited idea that originally had to do with how banks make loans at the operational level. And yes folks – that is the way it was first explained in the old textbooks.
Central banks for example no longer embrace the old money multiplier argument in explaining the intended objective of QE as it relates to the creation of extraordinary levels of excess reserves. The normal explanation now relates to the intended effect on the interest rate markets that are involved in selling the bonds to the central bank. QE operates in money terms through bank liability profiles that provide the broad money for institutional portfolio managers to make those decisions. QE is a special case of monetary policy that is quite different than how it has been implemented conventionally through Fed SOMA operations (for example). And while the central bank is managing its balance sheet and base money in both cases, it is also targeting its operational effect on interest rate markets in the process.
In summary, some conceptual friction exists between those favoring an operational perspective on monetary policy and those adhering to a steadfast monetary base view in the long run. Perhaps those coming at the subject from the operational perspective don’t know enough monetary theory. But this is moot if the theorists are under little obligation to present theory that fits nicely with differentiated functionality in real world monetary operations. The important question might be whether such realistic implementation connections are relevant for economics. And so it is unclear why monetary theory should reconcile only reluctantly with the operational details of central banking and commercial banking.