On the full endogeneity of high-powered money:
Lessons from the Canadian case



Marc Lavoie*
Full Professor
Department of Economics
University of Ottawa


First draft: October 2003

Paper prepared for the Conference in honour of Augusto Graziani,
to be held in Benevento, Italy, December 5-6, 2003.


* This paper has benefited from the financial assistance of C-FEPS, the Center for Full Employment and Price Stability, located at the University of Missouri in Kansas City. I wish to thank Kevin Clinton and Warren Mosler for their comments.




On the full endogeneity of high-powered money: Lessons from the Canadian case

Marc Lavoie




Although Augusto Graziani is clearly a proponent of endogenous money, he paid little attention to the relationship between commercial banks and the central bank. This may due to the fact that Italy was an overdraft economy, where the endogeneity of high-powered money is quite obvious. The purpose of the paper is to show that even asset-based financial systems rely on a fully endogenous supply of high-powered money. This is demonstrated with an analysis of the monetary process tied to the new procedures of large value transfers in Canada.

In Canada, there are no reserve requirements any more, and the Bank of Canada does not enter into outright open market operations. This makes the monetary operations particularly transparent. The overnight rate, which is now the main concern of Canadian monetary policy, gravitates around the target overnight rate, set by the Bank of Canada. The discrepancy between the two rates is seldom more than one basis point. This is compared to the situation in the United States, with the federal funds rate and its official target, where discrepancies on average are about ten basis points. This is explained by the fact that, in contrast to the Fed, the Bank of Canada knows with perfect certainty both its supply of settlement balances and the demand for settlement balances. In addition, the Bank has a symmetric system, whereby excess balances are remunerated at 25 basis points below the overnight rate target, while deficit balances are costed at 25 basis points above the target rate. The institutional and technical analysis shows that central banks engage essentially into ‘defensive’ operations, as argued by post-Keynesian authors from the neo-chartalist school.   


On the full endogeneity of high-powered money: Lessons from the Canadian case

Although Augusto Graziani has devoted a large number of his works to the notion of endogenous money, there is little to be found about the behaviour of the central bank. This in itself is not surprising, since most writers of the circuit school have paid no attention to the role of central banks. Most analyses of the monetary circuit are set within a kind of pure Wicksellian economy, closed and without a government. The links between commercial banks and the central bank, as well as the interdependence between the decisions of the central bank and the non-financial sectors, are thus left aside.

By contrast, post-Keynesian writers, in particular American ones, have devoted a great deal of attention to the relationships between the central bank and the commercial banks, in particular to the question of reserves – compulsory reserves and reserve-constrained economies. One could speculate at length on the causes of this divergence, but I will bring forth here what seems to me an obvious explanation.

Most circuit writers come from two countries that, according to the distinction made by Hicks (1974), are based on overdraft financial systems – Italy and France.[1] “The overdraft economy is defined by a double level of indebtedness: that of the firms to the banks and of the banks to the central bank” (Renversez 1996: 475). We are concerned here about the latter link, whereas Graziani and circuit writers in general are concerned with the former. A large proportion of the assets of these overdraft central banks are made up of claims over their domestic banking sector – the advances made to commercial banks. Before the advent of the Euro and the European Central Bank, claims on banks by the Bank of France represented about 70% of its total domestic claims, while the claims on banks by the Bank of Italy represented about 15% of its total domestic claims. In terms of overall central bank assets, claims on banks represented 25% and 10% respectively. As one would expect, advances to credit institutions still play a large role in Euroland. Lending to banks in the consolidated financial statement of the Eurosystem still represents between 25 and 30% of its total assets.

When circuit theorists were developing their ideas, it was quite obvious to all that overdraft financial systems imply that the supply of high-powered money is endogenous, and responds to the needs of demand. In France commercial banks hardly held any Treasury bills. The only means to acquire banknotes or compulsory reserves was to borrow them from the central bank, as had been shown in detail by Le Bourva (1992 [1962]). The central bank thus had to provide the required high-powered money through the discount window, but at the interest cost of its choice. Since the supply of high-powered money is so obviously endogenous and demand-led under these conditions, this would explain why circuit authors paid little attention to it.

I have argued previously that the supply of high-powered money in the so-called financial market economy (or asset-based financial system)  is just as fully endogenous as it is under overdraft systems (Lavoie 1992: 169). I made the claim that “the logic of a monetary production economy is such that the consequences of an overdraft economy also apply to an economy with open market operations.... whatever the actual financial institutions” (Lavoie 1992: 179).  Others have made similar claims, such as Thomas (1981), Goodhart (1987) or Mosler (1997-98), expressing the view that the supply of high-powered money is just as endogenous in the UK or in the USA, where financial markets appear to be of great importance. My claim then was that these Anglo-saxon monetary systems, dominated by asset-based financial systems, had more complex institutional features that hid the reversed causality and the essential mechanism of the endogenous supply of high-powered money.

           The prevailing view, however, among mainstream ‘overdraft’ economists, and implictly among some members of the so-called post-Keynesian structuralist approach, is that while the supply of high-powered money is endogenous in an overdraft economy, it is not the case in an asset-based economy. Renversez (1996: 475) for instance, claims that the functioning of the monetary system in an overdraft economy is “different from that achieved under the control of the central bank in the financial market economy. ... The intervention of the central bank is discretionary in the financial markets economy, but it is obligatory in the overdraft economy”. Among post-Keynesian authors, support for this kind of proposition has been offered by a string of authors, notably Cottrell, Pollin, Dymski and Palley. These authors somehow refer to reserve-constrained banks or claim that “central bank efforts to control the growth of non-borrowed reserves through open market restrictiveness exert significant quantity constraints on reserve availability” (Pollin 1996: 495). These views have been criticized, and rightly so, elsewhere (Rochon 1999, ch. 6).

The purpose of the present paper is to take advantage of the recent institutional changes that have arisen within the North-American financial systems in the 1990s, which help to cut through the complexities of asset-based financial systems. The new procedures put in place by the Federal Reserve System and the Bank of Canada illustrate quite clearly, to those who are not blinded by mainstream textbook presentations of central banking, the endogenous nature of the supply of high-powered money. They show, as emphasized by horizontalist authors such as Basil Moore (1988), that short-term interest rates are the exogenous variable under the control of central banks. The central banks do not, nor can they, control any monetary aggregate. The supply of high-powered money, even in Anglo-saxon countries with asset-based financial systems, is fully endogenous, in the revolutionary sense defined by Rochon (1999: 63).

I start by describing the Canadian monetary system. This will then be compared to the situation occurring in the United States. The comparison will allow us to understand why some American post-Keynesian authors came to believe that the federal funds rate was not really under the control of the Federal Reserve, and why they thought the central bank could somehow constrain the amount of reserves.


Zero-reserve requirements at the Bank of Canada

Prior to 1991, Canada had a textbook monetary system: Commercial banks faced reserve ratios on their deposits, advances to banks at the discount window were strongly discouraged, and open market operations by the Bank of Canada were frequent. The discount rate (the Bank rate) was set as a mark-up over the Treasury bill rate, giving the illusion that financial markets, not the central bank, were responsible for the high interest rates that were then prevailing. The only peculiar characteristic was the use of transfers of government deposits, from the accounts of the central bank to those of commercial banks (or vice-versa) to increase (or decrease) the amount of high-powered money.

 Discussions on the possibility of implementing monetary policy with highly reduced reserve and even zero-reserve requirements started in September 1987 (Bank of Canada 1987, 1991). A first step towards this process was implemented in 1991, when the frequency of advances or loans to commercial banks became unrestricted (with the appropriate collateral) and left to a new price mechanism designed by the staff at the Bank of Canada. While bank deposits at the Bank still did not pay interest, advances that lasted through the averaging period were costed at twice the discount rate. The purpose of such a move was to insure that the opportunity cost of holding excess reserves was about equal to the opportunity cost of central bank advances (relative to returns on alternatives).

It was known however that this new system was only a transitory one. Compulsory reserve requirements were progressively diminished, until they were completely dismantled in mid-1994 (Clinton 1997: 14).  The focus of monetary policy moved away from the Treasury bill rate, towards the overnight rate. A 50 basis points operating band for the overnight rate was put in place in 1994, and in 1996 the Bank rate was set at the upper end of the operating band, to provide more clarity as to the intentions of the Bank (Lundrigan and Toll 1997-98: 36). A second round of discussion took place in 1995, when the present system, dealing with electronic large-value payments, was designed (Bank of Canada, 1995). It was implemented in 1999. The term ‘reserves’ was struck out and replaced by the expression ‘settlement balances’ (in the US, they are called ‘clearing balances’). An official target overnight rate was put in place. This rate is in the middle of the operating band. Its upper limit is the Bank rate (the discount rate), at which commercial banks can borrow settlement balances (reserves); its lower limit is the rate on positive settlement balances – the rate paid on bank deposits at the central bank.[2] All this is illustrated in Figure 1, adapted from Bank of Canada (1995) and Clinton (1997). 


In Canada, as in New Zealand, the amount of high-powered money is now limited to the amount of banknotes held by the general public or in the vaults of commercial banks. There are no compulsory reserves. In addition, one can say that there are virtually no reserves of any kind. Bank deposits at the Bank of Canada are normally zero.[3] This in itself should help to demonstrate that the supply of high-powered money is fully endogenous. High-powered money in these countries is only made up of banknotes (issued by the central bank). Besides extraordinary situations as the one that occurred in Argentina in 2001-2002, it is difficult to imagine that the supply of bank notes through automatic teller machines would be restricted by the central bank. Whenever commercial banks need banknotes to feed their machines, as a result of the demand for banknotes arising from their customers, they are being provided by the central bank. Indeed, as noted by a researcher at the Bank of Canada, “ withdrawals of bank notes from the central bank are made as needed by the clearing institutions” (Clinton 1991: 7).


The operation of the settlements system

          In Canada, as in many other countries, banks and other direct clearers are required by law to settle their payment obligations on accounts at the Bank of Canada (Goodlet 1997). If there were no transactions with the public sector, or with the foreign exchange fund, the level of net settlement balances would always be zero. Since any debit for a bank corresponds to a credit for some other bank,  the net amount of settlement balances in this pure credit economy cannot be any different from zero. By contrast the gross amount of settlement balances would vary according to the dispersion in incoming and outgoing payment flows between banks. A given amount of transactions can give rise to widely different amounts of gross settlement balances.

However, as has been emphasized recently by members of the neo-chartalist school, the situation is modified when government transactions are entered into the clearing system, or when the central bank intervenes on foreign exchange markets (Mosler 1997-98; Wray 1998; Bell 2000; Bell and Wray 2002-3). As is well known, when the central bank purchases foreign currency to keep the exchange rate fixed, this adds to the reserves or the settlement balances of commercial banks. Similarly, when governments pay for their expenditures, by making cheques on their central bank account, which are later deposited at banks, these transfers add to reserves. By contrast, when private agents pay their taxes by writing a cheque to the government, this transaction withdraws reserves or settlement balances from the financial system once the cheque is deposited in the government account at the central bank. Similarly, when banks acquire banknotes, this reduces their settlement balances.

The Bank of Canada normally acts in such a way that the level of settlement balances in the financial system by the end of the day is exactly equal to zero. “To maintain the level of settlement balances at zero, the Bank must neutralize the net impact of any public sector flows between the Bank of Canada’s balance sheet and that of the financial system” (Howard 1998: 59). To achieve this, the Bank transfers government deposits in and out of its own accounts, towards or from, government deposit accounts held at various commercial banks.

The Bank effects such neutralization late in the afternoon, after all settlement transactions with the government are completed. When the Bank makes its final cash management decisions, it knows with perfect certainty the amounts that need to be transferred between government accounts at the Bank and government accounts at commercial banks to achieve complete neutralization of the public sector flows. In addition, early in the morning, when most of the clearing transactions occur, the Bank offers open-market operations (in the form of overnight repos or reverse repos, called sale and repurchase agreements and special purchase and resale agreements in Canada), at the target rate, to keep the market overnight rate on target.[4] This often has the effect of promptly neutralizing government flows. For instance, on a day when tax receipts are high (a drain on the system liquidity), the Bank will be providing central bank credit from the outset.

           In terms of standard terminology, one could say that these transfers of government deposits and open-market operations are part of the ‘defensive’ operations of the Bank of Canada. As Eichner, Forman and Groves (1985: 101) put it, “this is the neutralizing component of a fully accommodating policy”. I argue that it cannot be otherwise. There is an overall demand for high-powered money, exactly equal to the demand for banknotes, to which the central bank responds by providing the precise amount being demanded. The extent and the importance of these ‘defensive’ operations are nothing new.  Prior to the implementation of the new procedures, the Bank already knew with a good degree of  accuracy the amount of defensive operations that were required (Clinton 1991: 7-8).


The determination of the overnight interest rate

          There are two substantial changes between the new procedures and the previous ones. First, banks are given the opportunity to get rid of their surplus settlement balances, or to wipe out their negative settlement balances (their day overdraft at the Bank of Canada), by being able to have a last go on the overnight market in the evening, when they know with certainty what their clearing balances are (this is the so-called pre-settlement period). This allows banks to have day overdrafts only, and to avoid the discount window. Secondly, the central bank now knows with perfect certainty not only the amount of settlement balances being supplied but also its demand. Previously, when compulsory reserves were still required and averaged through the month, the daily demand for settlement balances by the banks could vary, with the Bank being unable to predict the changes. This was because the daily demand for reserves was responsive to interest rates.

This reflected mainly two factors: “First, that uncertainty about the results of the clearings creates a precautionary demand to hold reserves in excess of minimum requirements; and second that reserve averaging allows the banks some flexibility in respond to expected changes in overnight rates” (Clinton 1991: 9). If overnight rates were expected to move up in the future, the demand for reserves would move up, in an attempt to accumulate excess reserves that could be depleted at a later stage of the averaging period, when their cost would be higher. This made the overnight rate fluctuate, as demand moved around the demand for settlement balances expected by the Bank of Canada, and hence the amount of non-borrowed reserves supplied by the Bank. Any adjustment had to be carried through either by changes in borrowed reserves or in changes in the overnight interest rate.

None of this, or very little of it, occurs with the new rules. The new procedures ensure a determinate demand for settlement balances. First, banks need not play any games about  expected clearing positions or about future expected overnight rates as there are no averaging provisions anymore, since no amount of reserves need to be held. Secondly, the Bank has put in place “incentives that motivate the banking system to target zero settlement balances at the central bank” (Clinton 1997: 4). As already pointed out, there is symmetry in the opportunity cost of being in an overdraft position vis-à-vis the central bank and in holding a reserve deposit at the Bank. Overnight rates, repos rates and Treasury bill rates are  normally in the mid range between the Bank rate on overdrafts and the rate paid on deposits at the central bank. This mid-range is the target overnight rate, publicly announced by the central bank. This encourages banks to rely on the overnight market to obtain or get rid of their excess settlement balances.

The overall demand for settlement balances is thus equal to zero, in normal circumstances, since no surplus-clearing bank will desire to keep its surplus balances as deposits at the central bank, while no deficit-clearing bank will rely on advances that can be granted on demand by the Bank of Canada, since settlement balances can be borrowed or lent at a rate which is somewhat half-way in between the rates that could be obtained from the Bank. In the worst of circumstances, the overnight rate cannot be any higher than the Bank rate, for otherwise deficit banks would prefer to get central bank advances. Similarly, the overnight rate cannot fall any lower than the rate on deposits at the central bank, for otherwise, surplus banks would all put their surplus balances on the accounts of the Bank of Canada. Supply of and demand for settlement balances would readjust to each other.

Going back to Figure 1, Clinton (1997: 11) argues that in normal times both the supply and the demand for settlement balances are given by the vertical line arising from the zero level of settlement balances. “Since equality of demand and supply is represented by the intersection of two vertical lines (at zero quantity), on any given day the precise overnight rate at which the market settles is indeterminate within the 50-basis-point operating band. The actual rate will be influenced by a variety of technical factors, such as the size and distribution of clearing imbalances among the banks”.

           The overnight rate of interest could thus be any rate within the operating band. In a truly competitive market, however, one would expect the overnight rate to be right in the middle of the operating band. If the target overnight rate is set as the mid point of the operating band, there is thus some likelihood that it will be exactly realized. Under non competitive conditions, or if some banks are viewed as less credit-worthy than others, the overnight rate might be different from the target set by the Bank of Canada. For instance, if deficit-clearing banks happen to be the less credit-worthy, there is a chance that the overnight rate would exceed the mid-range point. Also, if a single bank holds positive settlement balances, while all others are in a negative position, the surplus-clearing bank may take advantage of its monopoly status and the overnight rate could be higher than the target overnight rate.

In reality, it turns out that the overnight market rate is systematically equal to the target rate set by the Bank of Canada. With the new procedures, tied to zero-reserve requirements, near-perfect certainty on the demand for settlement balances and absolute control over the supply of settlement balances, the Bank is able to control the overnight rate to the tune of one basis point. Over the last sixty days preceding the writing of the current paper, the overnight rate was either exactly equal to its target, or one basis point below it, except one day, when if was off by four basis points. When target rates are changed, overnight rates move instantaneously to their new position. For instance, on Septermber 3rd, 2003, the target rate was moved down from 3.00% to 2.75%. On the same day, the actual overnight rate dropped to 2.75%.[5]

          Another feature that is worth noting is that overnight rates change in response to target rates without central banks having to add or subtract any amount of settlement balances. This has been noted for other countries as well, under the name of ‘open-mouth operations’ (Guthrie and Wright 2000). In the Canadian case, the Bank of Canada keeps targeting zero settlement balances, even when a new rate is announced. The associated changes in the rates of the overnight credit and deposit facilities at the central bank are sufficient to enforce the new rate on the interbank money market. In general, there is no specific need to intervene on the repos market. The target rate set by the central bank, with its operating band, provides an anchor to the financial system. The anchor is credible because the Bank of Canada has the capacity to enforce it. If the overnight rate were to wander away from the target, the Bank could get it back on track.

The above analysis clearly shows that reserves are fully endogenous. The Bank of Canada supplies high-powered money by fully responding to the demand for it, i.e., by providing banknotes whenever banks require them. The fact that no reserves are required anymore, and that cost incentives have been put in place that encourage banks to hold neither positive nor negative settlement balances, makes the endogeneity of high-powered money very clear. In addition, it is quite clear that the control variable of central banks is the overnight rate of interest. The Bank of Canada sets the target overnight rate, and the actual overnight rate adjusts to it within the day, either right on the dot, or one or two basis points above or below it. As Wray (1998: 107) correctly concludes, “ the Canadian system makes central bank operations more transparent – reserves are not a lever to be used to control the money supply. The Bank of Canada intervenes to keep net settlement balances at zero, an operation that by its very nature must be defensive”.

This is precisely the argument that I wish to make. In the case of the overdraft economy, it is quite obvious that reserves are being provided on demand by the central bank. It is not so obvious in an asset-based financial system. But in systems such as the Canadian one, which is an asset-based financial system, the veil of open market operations is superseded by the transparence of the zero-reserve requirements. It becomes nearly as obvious that the day-to-day role of the central bank is to provide on demand the required level of high-powered money. It becomes obvious, once again, that high-powered money is a fully endogenous variable, while the overnight rate is the exogenous interest rate, determined by the target rate set by the Bank of Canada.

Within such as system it becomes obvious that commercial banks cannot be reserve-constrained. This part of the structuralist story just does not hold up, and is clearly a remnant of neoclassical analysis. In addition, it contradicts the views espoused by Minsky himself  (Wray 1989: 154).


The case of the American financial system

           The argument that I wish to make here is that the American financial system obeys to the same logical requirements that rule overdraft economies or financial systems with zero-reserve requirements.[6] Some post-Keynesians have pointed out long ago, that open market operations had little or nothing to do with monetary policy. For instance, Eichner et al. (1985: 100) start their article by making the following statement: “It is usually assumed that a change in the Fed’s holdings of government securities will lead to a change, with the same sign attached, in the reserves of the commercial banking system. It was the failure to observe this relationship empirically which led us, in constructing the monetary-financial block of our model, to try to find some other way of representing the effect of the Fed’s open market operations on the banking system”. That other way is that “the Fed’s purchases or sales of government securities are intended primarily to offset the flows into or out of the domestic monetary-financial system” (Eichner 1987: 849).


Throughout most of its history, the Federal Reserve System has acted on the premise that its main role in the financial system is to conduct ‘defensive’ operations, since the monetary base is an endogenous variable beyond its direct control. At times, the Fed has attempted to be non-accommodating, by restricting the amount of non-borrowed reserves; but this policy, besides pushing up interest rates, has had little impact on total reserves. For instance, between December 1979 and February 1980, the Fed sold for over three billion dollars worth of securities on the open market, thus reducing non-borrowed reserves by this amount; but borrowed reserves increased by that very amount during that period (Thomas 1981: 960). In addition, as is well-known, even when the Fed had monetary targets, these targets were implemented through the estimation of a money demand function; this estimate led the Fed to target unannounced federal funds rates, and the game was to guess the Fed’s target overnight rate. In 1987, the Fed reverted to official federal funds rate targeting, and that rate became publicly announced in 1994. In the USA, as in Canada, there has been a move towards greater transparency, removing the scaffolding that hid the true monetary operations of the central bank. As Mosler (2002: 419) points out, “the Federal Open Market Committee’s target has been the focus of activity under previous Fed policies as well, and the difference is that prior to 1994 the target rate was known only within the Fed, whereas currently it is disclosed to the general public”.

It is now much more obvious that the Fed is mainly pursuing ‘interest maintenance operations’ (Mosler 1997-98: 170; Wray 1998: 87). Again, neo-chartalist post-Keynesians have made this quite clear over the last years. For instance, Wray (1998: 115) claims that “Fed actions with regards to quantities of reserves are necessarily defensive. The only discretion the Fed has is in interest rate determination”. Similarly, Mosler (1997-98) writes that “as a practical matter, the Fed can only react to required legal reserve imbalances that threaten to alter the targeted federal funds rate. The Fed does not have the option to act proactively to add or drain reserves to directly alter the monetary base ....”.

Still, in the United States there have been important fluctuations in the overnight rate, relative to the federal funds rate target. Taylor (2001: 36) reports that the standard deviation of the spread between the federal funds rate and its target has been 18 basis points over the 1998-2000 period. The maximum deviation was -150 basis points, on December 31, 1999  – the result of a mistaken evaluation of the fears generated by a possible Y2K mishap. Similar deviations between the target rate and the actual overnight rate can be observed with the new European Central Bank. Nonetheless, over the last years, the average federal funds rate is virtually equal to its average target rate.

In view of these results, it is easier to understand why some American post-Keynesians are reluctant to recognize that reserves are fully endogenous and that interest rates are set exogenously by central banks. In the States, as in Europe, the central banks do not appear to have full control over the shortest of the rates –  the overnight rate. Interest rates under the control of the central bank do not appear to be truly exogenous. Their levels seem to depend on the interaction between the demand for and the supply of reserves. It should be noted that this feature of the American system was underlined by the major proponent of exogenous interest rates. In his book, Moore (1988: 124) wrote that “the federal funds rate is predetermined within a small range, ordinarily within fifty or sixty basis points.... It is not directly set by the Fed. .... It is ... disingenuous and misleading to declare that the funds rate is now ‘market-determined’. Market forces are really attempting to forecast the behavior of the Fed itself”.


 Pollin (1996) is quite aware Moore’s analysis, but, on the basis of his Granger-Sims causality tests, he argues that there is a two-way causality between interest rates controlled or nearly controlled by the central bank (such as the discount rate and the federal funds rate) and other market rates. Pollin (1996: 511) rejects the horizontalist view that the Fed independently administers short interest rates and that it is unable to constrain reserves.

More recent evidence enlightens this debate.  Taylor (2001) points out that, when examining recent episodes of changes to the official target rate, that changes in the federal funds rate often precede the change in the target rate, and that this is confirmed by causality tests based on daily data. Atesoglu (forthcoming), based on monthly data, shows that there is a two-way causality between federal funds rates and the prime rate between 1987 and 1994, whereas causality is uni-directional between 1994 and 2002, running from the federal funds rate to the prime rate. In addition, there is a nearly complete pass-through in the latter period.

My interpretation of all this evidence is the following. The Fed is pursuing essentially defensive operations, just like the Bank of Canada. The difference is that the Fed does not have perfect information about the drains on reserves that must be compensated for, nor does it have perfect information about the daily or even hourly demand for free reserves or for discount window borrowing; as a result, the Fed cannot perfectly equate supply to demand at the target funds rate (or at the actual rate). As Sellon and Weiner (1997: 18) put it, “the size of a daily surplus or shortage in the settlement system depends, in large part, on the central bank’s ability to estimate settlement bank demand for settlement balances.”.


One cause of this are the averaging provisions, that encourage banks to speculate about daily or even hourly evolutions of the federal funds rate, by modifying their demand for reserves. The markets try to anticipate changes in the target rate, and they try to anticipate the evolution of the federal funds rate around the target rate. This is why the prime rate does not seem to ‘cause’ the federal funds rate any more, ever since the target rate is being publicly announced. When setting the prime rate, banks don’t need to second-guess the evolution of the federal funds rate; the target acts as the anchor. As pointed out by Mosler (2002: 420), “this is in sharp contrast to the notion often supported by the media that market rates, rather than anticipating Fed action, contain information as to where the Fed should target the federal funds rate”, the media view being precisely that of Pollin (1996).[7] In the USA, over the reserve-averaging period,  the Fed supplies high-powered money on demand, as in overdraft economies or in zero-reserve financial systems, but it is unable to do so perfectly on a day-to-day basis. In other words, the apparent ‘non-defensive” operations arise inadvertently.



        In the conclusion of his 1996 paper, Pollin (1996: 511) asserts aggressively that the horizontalist position is steering in the wrong direction and that it is wrong to start from the “absolutist and unsustainable assertions that central banks have no power to constrain reserve levels but complete control over interest rates”. The present paper has shown that it is the “reserve-constraining view” which is contrary to the hard technical facts. If we understand complete control over interest rates as meaning control over the overnight interest rate within a tight operating band, one or two basis points for Canada, and a dozen basis points for the American system, then it is clear that short-term interest rates are exogenous in that sense. Indeed this is the point of view adopted by central bankers and some New Keynesians, through the so-called ‘new consensus view’ (Fontana and Palacio-Vera 2002).


As to the supply of high-powered money, after a long intermission, central bankers are coming back to the view that movements in money aggregates or in the monetary base contain no useful information for monetary policy; they are a sideshow – “a meaningless abstraction” as Albert Wojnilower (1980: 324) once put it. The new procedures put in place in Canada are particularly enlightening. Central banks do not attempt to control the monetary base. The latter is entirely demand-determined. The monetary operations of central banks are entirely defensive. Their purpose is precisely to ensure that the supply of high-powered money is exactly equal to its demand, at the target interest rate of their choice. The central bank may also intervene in specific markets, besides the repos market, to make sure that interest rates in certain specific markets are in line with the target overnight rate. Monetary operations are always interest rate maintenance operations.

Thus one must distinguish between ‘defensive’ and ‘accommodating’ behaviour. In my opinion, central banks pursue defensive operations at all times, as emphasized by the neo-chartalist authors. High-powered money is thus always fully endogenous. On the other hand, central banks can be accommodating or not. When they are, they will peg the interest rate, whatever the economic conditions (or they might reduce it). When they are not accommodating, i.e., when they are pursuing ‘dynamic’ operations as Victoria Chick (1977: 89) calls them, central banks will increase interest rates. As shown above, to do so, they need simply announce a new higher target overnight rate. The actual overnight rate will gravitate towards this new anchor within the day of the announcement. No change whatsoever in the supply of high-powered money is required. “Money is in some sense endogenous whether central banks are dynamic or not” (Lavoie 1984: 778).




Atesoglu, S. (forthcoming) “Monetary transmission mechanism – Federal funds rate and prime rate”, Journal of Post Keynesian Economics.

Bank of Canada (1987) “Discussion paper on the implementation of monetary policy in the absence of reserve requirements”, September 29, Bank of Canada.

Bank of Canada (1991) “The implementation of monetary policy in a system with zero reserve requirements”, Bank of Canada Review, May, 23-31.

Bank of Canada (1995) “A proposed framework for the implementation of monetary policy in the Large Value Transfer System environment”, Discussion Paper I, Bank of Canada.

Bank of Canada (2001) “The framework for the implementation of monetary policy in the Large Value Transfer System environment: Addendum III”, www.bankofcanada,ca./en/lvtsad-iii.htm

Bell, S. (2000) “Do taxes and bonds finance government spending?”, Journal of Economic Issues, Seoptermber, 34 (3), 603-620.

Bell, S. and Wray, L.R. (2002-3) “Fiscal effects on reserves and the independence of the Fed”,  Journal of Post Keynesian Economics, Winter, 25 (2), 263-272

Chick, V. (1977) The Theory of Monetary Policy, Oxford, Parkgate Books.

Clinton, K. (1991) “Bank of Canada cash management: the main technique for implementing monetary policy”, Bank of Canada Review, January, 3-32.

Clinton, K. (1997), “Implementation of monetary policy in a regime with zero reserve requirements”, Working Paper 97-8, Bank of Canada.

Eichner, A.S. (1987) The Macrodynamics of Advanced Market Economies, Armonk, M.E. Sharpe.

Eichner, A.S., Forman, L. and Groves, M. (1985), “The demand for money further reconsidered”,

in A.S. Eichner (ed.), Toward a New Economics: Essays in Post-Keynesian and Institutionalist Theory, London, Macmillan, pp. 98-112.

Fontana, G. and Palacio-Vera, A. (2002), “Monetary policy rules; what are we learning?”,  Journal of Post Keynesian Economics, Summer, 24 (4), 547-568.

Goodlet, C. (1997) “Clearing and settlement systems and the Bank of Canada”, Bank of Canada Review, Autumn, 49-64.

Goodhart, C. (1987) “Monetary base”, in J. Eatwell, M. Milgate and P. Newman (eds), The New Palgrave, vol. 3, 500-502.

Guthrie, G. and Wright, J. (2000), “Open mouth operations”, Journal of Monetary Economics, 46 (2), 489-516.

Hicks, J. (1974) The Crisis in Keynesian Economics, Oxford, Blackwell.

Howard, D. (1998) “A primer on the implementation of monetary policy in the LVTS environment”, Bank of Canada Review, Fall, 57-66.

Lavoie, M. (1984) “The endogenous flow of credit and the Post Keynesian theory of money”, Journal of Economic Issues, September, 18 (3), 771-798.

Lavoie, M. (1992) Foundations of Post-Keynesian Economic Analysis, Aldershot, Edward Elgar.

Le Bourva, J. (1992) “Money creation and money multipliers”, Review of Political Economy, 4 (4), 447-466.

Lundrigan, E. and Toll, S. (1997-98) “The overnight market in Canada”, Bank of Canada Review, Winter, 27-42.

Moore, B.J. (1988) Horizontalists and Verticalists: The Macroeconomics of Credit Money, Cambridge, Cambridge University Press.

Moore, B.J. (1991) “Money supply endogeneity: ‘reserve price setting’ or ‘reserve quantity setting’ “, Journal of Post Keynesian Economics, Spring, 13 (3), 404-413.

Mosler, W. (1997-98), “Full employment and price stability”, Journal of Post Keynesian Economics, Winter, 20 (2), 167-182.

Mosler, W. (2002) “A critique of John B. Taylor’s ‘Expectations, open market operations, and changes in the federal funds rate’ ”, Journal of Post Keynesian Economics, Spring, 24 (3), 419-422.

Palley, T.I. (1991) “The endogenous money supply: consensus and disagreement, Journal of Post Keynesian Economics, Spring, 13 (3), 397-403.

Pollin, R. (1996) “Money supply endogeneity: what are the questions and why they matter?”, in G. Deleplace and E.J. Nell (eds), Money in Motion: The Post Keynesian and Circulation Approaches, London, Macmillan, pp. 490-515.

Renversez, F. (1996), “Monetary circulation and overdraft economies”,  in G. Deleplace and E.J. Nell (eds), Money in Motion: The Post Keynesian and Circulation Approaches, London, Macmillan, pp. 465-488.

Rochon, L.P. (1999) Credit, Money and Production: An Alternative Post-Keynesian Approach, Cheltenham, Edward Elgar.

Sellon, G.H. and Weiner, S.E. (1997) “Monetary policy without reserve requirements: case studies and options for the United States”, Federal Reserve Bank of Kansas City Economic Review, 82 (2), 5-30.

Taylor, J.B. (2001) “Expectations, open market operations and changes in the federal funds rate”,  Federal Reserve Bank of St. Louis Review, July-August, 83 (4), 33-48.

Thomas, J.G. (1981) “Illusion monétaire et illusions monétaristes”, Revue d’économie politique, 91 (6), 947-965.

Wojnilower, A.M. (1980) “The central role of credit crunches in recent financial history”, Brookings Papers on Economic Activity, (2), 277-326.

Wray, L.R” (1989) “A Keynesian theory of banking: a comment on Dymski”,  Journal of Post Keynesian Economics, Fall, 12 (1), 152-156.

Wray, L.R. (1998) Understanding Modern Money: The Key to Full Employment and Price Stability, Cheltenham, Edward Elgar.


[1] Keynes (1930, ch.32) made a similar distinction, when he compared the British and the Continental central banking systems.

[2] A similar system, with identical rules, but a wider operating band, was also put in place for paper-based payments (cheques). See Howard (1998). Although most transactions are made by cheques, these constitute only a small fraction of the value of payments that go through the electronic large-value transfer system. As a result, monetary policy is only concerned with the latter.

[3] Usually, the Bank of Canada leaves in settlement balances of $50 million, to reduce frictions in the system (Bank of Canada, Addendum III, 2001). But this amount  pales compared to the daily amounts that are transacted through the large-value payments system.

[4] Since the Bank is not targeting the Treasury bill rate any more, there is no need for intervention on that market. Indeed, the Bank of Canada has not performed any outright open market operation since 1995 (Lundrigan and Toll 1997-98: 36 ).

[5] Even on September 11, 2001, and its aftermath, the Bank of Canada was able to keep the overnight rate right on target, as the rate hovered between 3.98% and 4.00%. It is true that to achieve this, the Bank did not target zero settlement balance, but rather a large amount of surplus balances. When, on Monday, September 17, the target rate was dropped from 4.00% down to 3.50%, the overnight rate fell to 3.54%, and the next days it stood at 3.48% or 3.49%.

[6] Compulsory reserves in the USA account for less than 2% of the monetary base.

[7] This also explains why long-term interest rates always seem to Granger-cause overnight rates or Treasury bill rates. Long-term rates in part reflect anticipations of future short-term rates. I believe the new evidence substantiates the position taken by Moore (1988: 286; 1991) and Palley (1991) regarding the meaning of Granger-causality tests when expected variables are involved.


posted 10-16-2003