Setting Interest Rates in the Modern Money Era by Scott T. Fullwiler* Working Paper No.

34

July 2004

Electronic copy available at: http://ssrn.com/abstract=1723591

Setting Interest Rates in the Modern Money Era

Scott T. Fullwiler1
Wartburg College and the Center for Full Employment and Price Stability

In 2002, an article by Kansas City Fed Vic e President Gordon Sellon demonstrated that several market interest ratesparticularly banks prime rates, mortgage rates, consumer loan rates, and Treasury rateshave since the 1990s become much more closely linked to movements in the Feds federal funds rate target than in previous decades. In particular, Sellon wrote, changes in the financial system such as financial deregulation (such as removal of deposit rate ceilings and product-line and geographic barriers), growth in capital markets (which enables many borrowers to bypass intermediaries and go directly to financial markets), and financial innovations (securitization in particular) have enabled a faster and larger response of these rates to changes in the federal funds rate. Sellon concluded that these changes in the financial structure appear to have strengthened the interest rate channel of monetary policy (Sellon 2002, 30). Interestingly, Sellons article appeared just as several other economists were expressing concerns that the interest rate channel of monetary transmission might be undercut by some of these same financial market developments. These concerns arrived on the heels of those of others who in the late 1990s noted the decline in required reserves that already had complicated the Feds ability to effectively maintain the federal funds rate target (e.g., Clouse and Elmendorf (1997); Bennett and Hilton (1997)). While variability in the federal funds rate was eventually reduced (the reasons for which are discussed below), the new round of research went a step further. In particular, research by Benjamin Friedman (1999, 2000), Thomas Palley (2000, 2001-2, 2003) and others (e.g., King 1999; Gormez and Capie 2000; Costa et al 2001) suggested technological innovations in the payments system that have already enabled far fewer reserve balances to settle increasing dollar values of payments will continue to reduce the quantity of reserve balances circulating. Eventually, they argued, if reserve balances were to fall due to alternative methods of payment settlement and the continuing decline in required reserves, there might be no reason for banks to hold reserve balances to settle payments; in that scenario, the ability of the Fed to influence interest ratesand through them, the broader economycould come into question. Specifically, if banks have no reason to hold reserve balances, the federal funds rate itself might have no effect upon other interest rates and asset prices. As Friedman and Palley separately have put it,
The threat to monetary policy from the electronic revolution in banking is the possibility of a decoupling of the operations of the central bank from the markets in which financial claims are created and transacted in ways that, at some operative margin, affect the decisions of households and firms . . . . (Friedman 2000, 262; emphasis in original) The challenge to interest rate control stems from the possibility that e-money may diminish the financial systems demand for central bank liabilities, rendering central banks unable to conduct meaningful open market operations. (Palley 2001-2, 217)

Friedman (1999) argued that absent aggressive regulatory actions, continued acceleration in the pace of these changes will eventually result in the central bank being unable to affect the overall level of spending other than by signaling its desires via an interest rate announcement and hoping the markets follow. Palleys policy proposal, asset-based reserve requirements (ABRR), would create a reserve requirement based upon the assets of banks and possibly other financial institutions (as compared to current reserve requirements that are tied to liabilities) that would guarantee a considerable demand for reserve balances

1
Electronic copy available at: http://ssrn.com/abstract=1723591

and thereby reduce variability in the federal funds rate and reaffirm the Feds ability to influence interest rates. This paper argues that much of the research on the electronic money revolution and the Feds ability to set interest rates suffers from a flawed understanding of the process of monetary policy implementation at the Fed. It is shown below that the quantity of reserve balances in circulation is irrelevant to the Feds ability to set the federal funds rate and is similarly irrelevant to whether or not other rates are influenced by the federal funds rate. The fact that banks are obligated to use reserve balances to settle their customers tax liabilities ensures that a non-trivial demand for reserve balances will exist, which itself ensures that the federal funds rate target will remain coupled to other interest rates. Rather than electronic money eras in which the Feds abilities to implement monetary policy are compromised by the proliferation of private settlement systems, what is significant about current and future eras is that they will remain modern money eras as long as taxes must be paid in reserve balances. Use of the term modern money here follows that of L. Randall Wray (1998), who demonstratedusing evidence from history, history of thought, as well as more contemporary experiencethat the imposition by the state of a tax liability payable in its own money is sufficient to generate a demand for the States money. 2 The counterpart is the recognition that such demand is similarly sufficient for the central bank to set interest rates. The organization of this paper is as follows: The first section discusses the Feds daily operating procedures and particularly focuses on how volatility in the federal funds rate both arises and can be reduced. The second section discusses evolutions in the wholesale payments system and explains how these are (or are not) related to the ability of the Fed to set market interest rates through its federal funds rate target. The third section demonstrates that payments to the federal government are the fundamental sources of a non-trivial demand for reserve balances that ensures the Feds ability to set interest rates; the section also demonstrates that the modern money perspective is consistent with a proper understanding of the Feds operations and its ability to set interest rates. The final section offers concluding remarks.

I. OPEN MARKET OPERATIONS AND THE FEDERAL FUNDS RATE

This section discusses in detail the Feds open market operations related to setting and sustaining the FOMCs targeted rate and concludes with application to an era of financial innovation. Though the main points covered here have already been explained in detail in Fullwiler (2003), they deserve repeating and some additional elaboration since it is a misunderstanding of the details of the Feds operations that underlies misinterpretations of the consequences of continued innovations in the payments system for setting interest rates. The most important point to be made in this section is that while the federal funds rate has exhibited substantial variability in recent years as the quantity of reserve balances fell, the Fed could exercise substantially greater direct control over the federal funds rate than it currently does, if so desired, regardless of the quantity of reserve balances in circulation. Proposals, such as ABRR, intended to raise the demand for reserve balances in order to improve the Feds ability to achieve its federal funds rate target thereby misunderstand the nature of the Feds operations. The Feds open market operations are carried out in order to accommodate banks demand for reserve balances; in general, the operations are carried out simply to offset daily, exogenous changes to the Feds balance sheet which left alone would cause the quantity of reserve balances in circulation to deviate from the quantity banks desire to hold. Increases (decreases) in the asset side of the balance sheet (e.g., discount lending, securities owned, float) raise (reduce) reserve balances, while increases (decreases) in the liability side (e.g., currency/vault cash, Treasurys account) reduce (raise) reserve balances. Thus, for example, payments from (to) bank reserve accounts to (from) the Treasury reduce (raise) reserve balances, ceteris paribus. Similarly, bank purchases of vault cash to meet customer demands for currency are paid foron a one-for-one basiswith reserve balances, and similarly reduce

the quantity of reserve balances in circulation. Consequently and contrary to what is taught in economics textbooks (particularly in the money multiplier model), the Feds daily open market operations simply offset current and anticipated changes to the Feds balance sheet in the process of accommodating the demand for reserve balances consistent with the targeted federal funds rate, rather than continuously adding to or subtracting from the quantity of reserve balances (Bell 2000; Fullwiler 2003). These facts are self evident from a careful reading of the Open Market Desks annual reports andthough they had for several years been integrated into the endogenous money literature (e.g., Moore 1988; Wray 1990, 1998)are now recognized by a growing number of neoclassical monetary economists (Hamilton 1997; Clouse and Elemendorf 1997; Furfine 2000; Woodford 2000, 2001). Banks demand for reserve balances arises from two sources: reserve requirements and payment settlement needs. Regulation D requires banks to hold reserve balances during a two-week maintenance period against the difference between deposits and vault cash that were held during the computation period. The computation period ends seventeen days prior to the beginning of the maintenance period. Reserve balances held at the end of most business days count once toward the maintenance-period average; balances held at the close of business on Fridays (which count three times) and on days prior to holidays (which count once plus one more time for each coming holiday) are the typical exceptions. Banks use reserve balances held in reserve accounts at the Fed Banks to settle basic funds transfers, payment for securities transfers, net settlement transfers from private clearinghouses, automated clearinghouse (ACH) transactions, and currency deposits/withdrawals from Fed Banks. While the volume of checks and ACH transactions is 175 times as large as the volume of transfers on Fedwire (the Feds real-time gross settlement payment system), Fedwire transfers make up the vast majority95 percent, in factof dollar-value payments from reserve accounts (Panigay Coleman 2002, 74). In 2003, the total dollar value of Fedwire funds transfers was $437 trillion, or around $1.74 trillion per business day (Board of Governors 2004).3 Beyond these two activities, there is no other reason for banks to hold non-interest bearing reserve balances, and the demand for reserve balances is therefore quite insensitive in the short run to changes in interest rates. What is often misunderstood is that while the quantity of reserve balances demanded is relatively fixed due to reserve requirements and payment needs irrespective of the interest rate, this is not by itself the source of deviations in the federal funds rate from the targeted rate. Banks that fail to meet their reserve requirement for the two-week period have been traditionally assessed a penalty rate of 2 percent plus the discount rate on the deficiency and have also had to hold a higher average level of reserve balances during the next maintenance period. There are further penalties assessed when a banks reserve account falls into overdraft, as the Fed provides intraday credit (daylight overdrafts) or overnight credit (overnight overdrafts). Provision of overnight or intraday credit, or both, at some price is a common characteristic of all central banks, who are charged with ensuring the stability of their nations payments systems (Government Accounting Office 2002). Intraday credit is relatively inexpensive at 36 basis points; however, if an intraday negative balance persists through the end of the day, the penalty for overnight credit is the days federal funds rate plus 400 basis points and carries with it the threat of additional regulatory oversight for repeat offenders, a combination which banks obviously attempt to avoid (Clouse and Elmendorf 1997; McAndrews and Potter 2002). Banks that are not able to secure funding to clear daylight overdrafts by the end of business traditionally have not taken on overnight overdrafts but have rather borrowed at the discount window. However, because the discount rate was set below the federal funds rate, historically the Fed has strongly dissuaded banks from borrowing at the discount window unless all other sources of credit had been exhausted; as a result, banks have been less than eager to borrow at the discount window even as the federal funds rate at time rose well beyond its target (Hakkio and Sellon 2000). Given reserve requirements, the substantial penalty on overnight overdrafts, and the non-monetary costs associated with borrowing at the Feds discount window, the federal funds rate could rise substantially if reserve balances provided were insufficient to accommodate the existing demand for reserve balances. On the other hand, given that beyond settling payments and meeting reserve requirements there is no other reason for banks to hold non-interest bearing reserve balances, should more reserve balances

than banks desire to hold for these purposes be provided (either by the Fed or via exogenous changes to the Feds balance sheet), the federal funds rate could slip well below its targeted rate and may even fall to zero. It is therefore the existing combination of regulations and penalties creating a substantial spread (hereafter, the spread) between the rate paid on reserve balances (zero percent in the U.S.) and the penalty (both monetary and non-monetary) assessed to overnight overdrafts (or, alternatively, the non-monetary costs traditionally associated with borrowing from the discount window in order to avoid an overnight overdraft) that permits wide swings in the rate if too many or too few reserve balances are supplied by the Fed, rather than the fact that banks have no reason to hold more or less than the amount demanded. The two-week maintenance period has traditionally reduced the interest rate variability on individual days within the maintenance period. Reserve requirementsif large enoughcan provide a buffer against overnight overdrafts, while a relatively modest surplus or deficiency of reserve balances supplied on most days (except for the last few days of the maintenance period, at least) could often be offset later in the period. Reserve requirementsparticularly when the maintenance period lasts several days or more and begins after the end of the computation periodalso provide the Fed with a more predictable demand for reserve balances to accommodate as it offsets changes to its own balance sheet (Clouse and Elmendorf 1997; Fullwiler 2003). Consequently, the Fed has traditionally been able to accommodate the demand for reserve balances while limiting itself to one open market operation per day. Beginning in the mid-1990s, banks began using computer software to monitor withdrawal behavior of customers and periodically sweeping unused deposit balances into money market deposit accounts, which are not subject to reserve requirements. Use of these retail sweep accounts for individual depositors rose from near zero in 1994 to over $370 billion by 2000; total checkable deposits fell from $810 billion in 1994 to $595 billion in 2000 (Federal Reserve Bank of St. Louis 2004a, 2004b). The fall in deposits reduced reserve requirements significantly; many banks were then able to meet reserve requirements entirely through vault cash and virtually all banks were able to reduce the quantities of reserve balances held. These banks then became far more likely to incur overnight overdrafts and the penalties associated with them. As the quantity of reserve balances demanded became much more closely tied to daily payment settlement needs of banks rather than the more predictable, bi-weekly demand for reserve requirements, the daily volatility of the federal funds rate increased dramatically and the Feds ability to reliably accommodate the demand came into question. Many, however, recognized that the excessive volatility in the federal funds rate was simply the result of the Feds own penalties on overnight overdrafts in combination with its traditional hesitation to lend from the discount window (Bennett and Hilton 1999; Clouse and Elmendorf 1997; Furfine 2000). By 2000, volatility in the federal funds rate had been reduced to earlier (pre-sweeps era) levels due to a combination of factors including changes to Regulation D (switching from near-contemporaneous accounting of reserve balances to the abovedescribed lagged-accounting method), closer attention to the overnight reserve balance needs of banks in planning operations (whereas without the threat of overnight overdrafts more attention had been paid to the maintenance period average needs of banks), and enhanced monitoring of payment flows by individual banks (Fullwiler 2003).4 The increase in federal funds rate volatility that accompanied the reduced quantity of reserve balances made explicit the connection between the payments system and the Feds implementation of monetary policy. For decades, economists had focused almost exclusively on reserve requirements in modeling the Feds open market operations, rather than on the payments system; nearly every graduate and undergraduate textbook still retains such a focus (e.g., the deposit multiplier). However, analysis of the Feds implementation of monetary policy more appropriately begins with the Feds necessarily automatic provision of sufficient reserve balances for daily settlement of payments. Reserve requirementssince they raise the overall quantity of reserve balances, enable banks to average reserve needs across days, and increase the predictability of the demand for reserve balancescan be seen as simply one possible way of reducing variability in the federal funds rate given the existing regulations and

penalties that have generated the spread (Fullwiler 2003). The more direct method of reducing the volatility of the funds rate is via reduction of the spread itself. Less onerous penalties or conditions on overnight credit extension by the Fed and payment of interest on reserve balances would reduce the range of possible variation in the federal funds rate regardless of how insensitive payment settlement needs and reserve requirements are to changes in interest rates. Many countries without reserve requirementsfor instance, Canada, Great Britain, Norway, Sweden, New Zealand, and Australia have successfully minimized overnight rate volatility simply by paying interest on central bank balances (say, 0.25 percent below the targeted overnight rate) and charging a bit more interest for overnight lending (say, 0.25 percent above the targeted overnight rate) at a Lombard-type lending facility for which there are no non-monetary costs for borrowing banks to consider. In such cases, the overnight rate simply fluctuates between the two rates but does not move outside the range since no bank would borrow for more or lend for less in the federal funds market than the upper and lower bounds set by the central bank. The demand for reserve balancessince it is related to existing payment settlement technologies and payment flowsremains insensitive to changes in the overnight rate even as the overnight rate itself does not move outside the upper and lower bounds (e.g., Woodford 2001). The Feds new primary lending facilitywhich replaced the discount window in January 2003lends to all banks (provided they have submitted appropriate collateral) at a rate of one percent above the targeted federal funds rate (slightly higher rates are required of banks deemed to be greater credit risks); by eliminating the non-monetary costs historically associated with borrowing from the Fed Banks and lending at a slight penalty to all who desire reserve balances, the primary lending facility is consistent with experience in other countries that have chosen to directly limit the volatility of the overnight rate (Sellon and Weiner 1997; Woodford 2001). The Fed has also spoken in favor of legislation permitting it to pay interest on reserve balances, indicating that it desires the ability to reduce the spread still further as in countries without reserve requirements (Meyer 2000). Returning to implications for current operations, since the quantity of reserve balances demanded is related to reserve requirements that have been previously determined and to payment settlement needs that are insensitive to the federal funds rate, when the Fed announces a new target federal funds traders adjust the rate without a change in the quantity of reserve balances being necessary. As Sandra Krieger, head of domestic reserve management and discount operations at the New York Fed, pointed out, the Feds announcement of a new target is sufficient for traders in the federal funds market to adjust their rates since any change in the FOMCs target has virtually no effect on the quantity of reserve balances demanded (Krieger 2002, 74). While the Fed might temporarily change the quantity of balances in order to signal a new rate to traders (which occurred prior to 1994 when official announcements of rate changes were implemented) or to nudge the rate to a newly announced target when traders do not move to the new target quickly enough, such changesunlike a liquidity effectare reversed later in the maintenance period unless they were already consistent with the demand for reserve balances for the entire maintenance period (Krieger 2002, 74).5 The main point here is that while daily deviations of the federal funds rate from its target result from the size of the spread, changes to the federal funds rate target itself can be made without any open market operations regardless of the width of the spread because changes to the target do not alter banks desired holdings of reserve balances. Of course, a sufficiently narrow spread eliminates the need for open market operations to change the federal funds rate target regardless of the sensitivity of banks desired quantities of reserve balances to the federal funds rate and regardless of the quantity of reserve balances in circulation; with a narrow spread, the Fed would simply announce changes to the upper and lower bounds together, changing the rate while the reduced spread limits the size of intraday deviations from the targeted rate.6 To conclude this section, misunderstanding the Feds open market operations can lead to the mistaken view that the Fed might lose the ability to reduce volatility in the federal funds rate should the quantity of reserve balances continue to decline. For instance, Palley suggested that the extreme short run inelasticity of the demand for reserve balances explains why only small changes in the quantity of reserves . . . are needed to make changes in the monetary authoritys target interest rate stick (Palley 2001-2, 227). While conceding that the demand for reserve balances is quite inelastic, Palley mistakenly

asserted that the Fed sets its federal funds rate target through a liquidity effect and thereby confused the regulatory sources of daily swings in the federal funds rate (i.e., the spread) with the reasons why announced changes to the Feds targeted rate can be carried out entirely without changes in reserve balances (i.e., the inelasticity of demand for reserve balances). Although the demand for reserve balances is insensitive to changes in the federal funds rate (i.e., there is no reason to hold any more or less than the amount that settles payments and meets reserve requirements), swings in the federal funds rate are created by the spread and the frequency of the Feds open market operations. Palleys ABRR proposal takes the spread and the Feds operating procedures (i.e., one operation per day) as given and reduces variability in the federal funds rate by raising reserve requirements (which would reduce the likelihood of overnight overdrafts). Compared to ABRR, simply reducing the spread by announcing bid and ask rates for reserve balances would more directly eliminate the potential for wide deviations in the federal funds rate from its target and would not impose additional regulatory burdens on banks.7 Moreoverand in contrast to a liquidity-effect view of the Feds operationsregardless how many or how few reserve balances were circulating, the Feds ability to change the target simply by raising or lowering the upper and lower bounds together would obviously be unaffected and would obviously require no open market operations. In sum, the Feds ability to both set and sustain a federal funds rate target is threatened only by its own (lack of) will to alter penalties and operating procedures to ensure this ability; changes occurring in the payments system will have adverse effects if and only if the Fed enables such effects.

II. INNOVATIONS IN THE PAYMENTS SYSTEM AND THE FEDS ABILITY TO SET INTEREST RATES
While falling reserve requirements and sweep accounts will not affect the Feds ability to set the federal funds rate regardless of the size of the demand for reserve balances, many have subsequently suggested that, even so, other interest rates may not follow the Feds target if the demand for reserve balances is reduced to the point that it is insignificant. As Jerry Jordan and Edward Stevens wrote,
The kernel of the money question emerging on the 21st century horizon is not just about further reduction in demand for central bank money, or even instability [in the federal funds rate] induced by more unpredictable demand [for reserve balances]. Rather, what may distinguish the 21st century is the possibility that central bank money might virtually disappear. Some have posed the theoretical possibility that, in the limit, there will be no appreciable domestic demand for central bank money whether currency or banks balances at Reserve Banks. (2002, 8)

As an initiator of research on the declining use of reserve balances for payment settlement and the possible negative effects upon monetary policy implementation, Friedmans widely discussed papers labeled it a puzzle that central banks were able to have such a large effect upon economies when their securities transactions were so miniscule in comparison to the size of the economies orparticularly in comparison to the value of financial market transactions. Indeed, open market operations by the Fed are rarely more than a few billion dollars on any given dayand often lesswhile U.S. GDP in 2003 was more than $11 trillion; Fed securities transactions are even less significant in comparison to daily financial market transactions that far outnumber the $1.7 trillion in funds transferred daily via Fedwire. As financial innovations continue, the disparity in the size of reserve balances and central bank operations to growing financial market transactions widens. Friedman concluded that in the absence of aggressive regulatory intervention to offset these developments, the central bank of the future would become an army with only a signal corps, able to announce its priorities though unable to affect market interest rates or asset prices. As with sweep accounts and reserve requirements, because reserve balances are non-interest bearing, there is an incentive for banks to minimize their holdings of them as much as possible. However, just as the Feds ability to set the level and manage variability in the federal funds rate does not depend

upon the quantity of reserve balances, one must similarly be careful not to equate a decline in the quantity of reserve balances in circulation or held by banks at the end of the day with a decline in the importance or non-triviality of the demand for reserve balances as a link between the federal funds rate and other short-term rates. Hyman Minsky (1957) recognized long ago that financial innovations (at that time) such as trades in the federal funds market and the repurchase markets would enable a given quantity of reserves to be correlated with far greater economic activity. As electronic payment has become far more prevalent during the last few decades, so too has the quantity of reserve balances necessary to settle a given quantity of payments already been declining significantly. Diana Hancock and James Wilcox note that,
In recent decades, even while the banking industry was growing faster than real economic activity, the dollar value of funds transmitted via large-dollar electronic payments systems was growing relative to the size of banks. . . . Two decades ago, daily transfers were less than one-tenth as large as total bank liabilities. By the mid-1990s, the ratio had risen to seven times its value in the early 1990s. . . . [During the same period] the sum of banks reserves and clearing balances . . . at Federal Reserve Banks relative to their total liabilities fell markedly: After averaging close to 4 percent in the early 1970s, reserve balances as a proportion of liabilities averaged less than 1 percent by the mid-1990s. As a consequence, the value of banks electronic payments relative to their reserve balances increased dramatically: By 1994, the ratio of the value of Fedwire transfers to reserve balances was about forty times its 1973 value. (Hancock and Wilcox 1996, 871)

One of the more important reasons for the decline in reserve balances necessary to settle payments is the fact that clearinghouses net payments and thereby enable their members to settle a small, netted percentage of total transactions via Fedwire. The Fedlike other central bankshas actively promoted netting of transactions in clearinghouses in order to improve payments system efficiency by reducing transactions and transaction exposures of participants (Bank for International Settlements 1989). For example, the Clearing House and Interbank Payments System (CHIPS) clears many international transactions in which dollars are involved and many other payments among large New York banks; gross payments among CHIPS members rival or even exceed those sent daily via Fedwire, while netted and bilateral CHIPS payments are settled via Fedwire throughout the day. Similarly, most small banks use local clearinghouses to clear local checks and other transactions while settling netted obligations using Fedwire. The Fixed Income Clearing Corporation (FICC, which recently merged the Government Securities Clearing Corporation (GSCC) and the Mortgage Backed Securities Clearing Corporation (MSBCC)) does the same for secondary market and repurchase transactions in Treasury securities among its members. The FICC and Depository Trust Clearing Corporation (DTCC, which owns FICC) do the same in the mortgage backed securities markets (including government agency and government sponsored enterprise securities) and several other financial markets (including commercial paper, corporate bond, and equities), respectively, for their members. FICC and DTCC settle netted payment obligations via Fedwire where more than one clearing bank is involved; FICC trades involving Treasuries and many government agency/government sponsored enterprise securities are settled as netted securities transactions versus payment via Fedwire when more than one clearing bank is i volved (i.e., securities n and payment are transferred via Fedwire). As payments system technology has improved and banks have become increasingly concerned with minimizing costs related to holding non-interest-bearing reserve balanceswhile also minimizing the costs of daylight overdraftsthe coordinated timing of outgoing Fedwire payments with incoming payments has further enabled larger dollar-values of payments to be settled with fewer reserve balances. Because payment sent via Fedwire within the same minute that a payment is received to the banks reserve account does not generate a daylight overdraft charge, during periods of normal activity incoming payments used to offset outgoing payments that are entered within the same minute account for 25 percent of the value of these transfers; during peak periods for payment transfers (early in the morning and at the close of business) outgoing payments offset by incoming payments account for as much as 40 percent of total payment activity (McAndrews and Rajan 2000, 18).

Overall, due to sweep accounts, netted settlement, timing of payments, and other payments system innovations, on average banks have payment flows that are now 100 to 200 times larger than their end-of-day reserve balances (Furfine 2000; McAndrews and Potter 2002). There is thus a rather obvious explanation for Friedmans puzzle: given that end-of-day reserve balances desired by banks are in the $10-$20 billion range, the Feds open market can thereby be still smaller in value since they merely offset net changes to the Feds balance sheet that would otherwise force the quantity of overnight reserve balances in circulation to deviate from this desired quantity. The relative size of the Feds own operations compared to transactions occurring in other financial markets is thus completely unrelated to the issue of whether or not the overnight rate set by the Fed influences other interest rates since the Fed obviously is setting the federal funds rate and makes no attempts whatsoever to affect other rates directly. Rather, changes to the federal funds rate target affect other short-term rates because borrowing or lending in the federal funds market can substitute for borrowing or lending in commercial paper, negotiable time deposit, Eurodollar, repurchase agreement, and short-term Treasury markets. Indeed, banks use overnight Eurodollar and repurchase agreement markets as alternatives to overnight trading in the federal funds market for acquiring reserve balances for use in payment settlement and to meet reserve requirements8 (Cyree et al 2003; Demiralp et al 2004). Because financial assets/liabilities created in these markets are in many ways substitutes, rates in each of these markets move together via arbitrage (though there are obviously some differences in rates due to differences in credit risks, collateralization, and liquidity). Long-term rates are known to be based upon both current short-term rates and expectations of future short term ratesonce an additional risk or liquidity premium on long-term rates is accounted forand are thus also influenced by current as well as expected levels of the federal funds rate.9 When intraday credit is considered, the very questions raised by Friedman demonstrate a flawed understanding of Fed operations since the end-of-day (or overnight) quantity of reserve balances targeted through open market operations is not necessarily relevant to the quantity of reserve balances banks desire to use throughout the day. Average daylight overdrafts supplied by the Fed throughout (that is, at each minute of) a typical day are more than $30 billion dollars; peak daylight overdrafts (the largest total or system-wide negative end-of-minute balances) are far higher at $100 billion per day on average (Panigay Coleman 2002, 76). Indeed, the dollar value of intraday credit has actually risen since the mid-1990s even as banks have minimized use of overnight reserve balances through sweeps, netting, and funding outgoing payments with incoming payments (Panigay Coleman 2002, 76). The quantity of reserve balances circulating at any given minute during the day is therefore far larger than the end-of-day quantity, and the total changes made by the Fed to the quantity of reserve balances on a given day are thereby far larger than the quantity of open market operations.10 Consequently, it is only if one considers $100 billion of peak daily credit extensionor $30 billion of credit extended each minute on averageto be a miniscule quantity that one could agree that Friedmans hypothesized puzzle was valid to begin. Nonetheless, ifas has been presumed by many economistsit cannot be relied upon that banks will desire to use reserve balances for payment settlement in the future and recognizing that reserve requirementsif they exist at allcan be avoided, then while the Fed may be able to set the federal funds rate, the federal funds rate may become decoupledto use Friedmans termfrom other short-term rates and to the broader economy at large absent substantial regulatory changes that engender a more robust demand for reserve balances. 11 As Jordan and Stevens implied, several economists have argued that, as payments system innovations continue, banks may no longer have use for reserve balances while settling wholesale payments among themselves. For instance, banks could eventually organize and participate in multilateral clearing and net settlement arrangements for money and securities transfers without using reserve balances (Jordan and Stevens 2002, 10). Mervyn King (1999) suggested that electronic money would eventually eliminate central banks monopoly position as suppliers of the means of payment; Claudia Costa Storti and Paul De Grauwe (2001) also concluded that open market operations and advances to banks would become ineffective as instruments to control interest rates. Friedman agreed that

a private mechanism like CHIPS could evolve into a system of purely bilateral transfers among private banks analogous to the settlement method now used by European countries central banks, which do not maintain clearing balances at the ECB . . . . A quarter century or so into the future . . . it is readily conceivable that one or more of these private clearing mechanisms may sufficiently erode banks need for central bank reserves as to undermine the relevance of the central banks monopoly. (Friedman 1999, 333)

Palley suggested that banks might eventually exchange virtually any private financial asset to settle netted payments.
The key to the emergence of such a system is the ability of banks to value assets to market in real time. The information technology (IT) revolution may be the final development necessary for this. Over the past two decades, the growth of markets for securitized bank loans has meant that bank assets have become much more liquid. Securitization combined with the IT revolution means that banks and financial institutions (FIs) may be approaching the point where the bulk of bank assets can be valued in real time, thereby making it possible to settle debts between banks by transfer of title to these assets. The combination of securitization and IT therefore creates the prospect of a new form of settlement call it mutual fund e-settlement money. Once this happens, bank settlement demand for reserves could decline dramatically, thereby diminishing the ability of central banks to control the overnight rate. Instead rates would be set in a loanable funds-style asset market.12 (Palley 2001-2, 222-3)

However, to recall the earlier quote by Jordan and Stevens, the money question for the Fed in the 21st century is not how many reserve balances banks will desire to hold in their reserve accounts overnight or, alternatively, how much intraday credit they will desire to acquire daily from the Fed. An often overlooked point is that this means that the question is not whether alternative payment settlement systems will ariseas a result of internet banks, internet payment settlement, smart cards, securitization, or other technologiesto compete with or economize upon the use of reserve balances in wholesale payments settlement, since they undoubtedly will. As such, the issue is not whether the Fed will continue to have monopoly power to issue the means of settlement, since it does not now and probably never has had such a power; while settlement with reserve balances (and with currency, for that matter) is large in terms of total dollar value, it has been shrinking in relative size as a result of competing settlement systems and the innovations discussed above. Rather, the money question for the Fed in the 21st century is whether the demand for reserve balances will remain non-trivial in the sense that arbitrage between the federal funds rate and other short-term rates continues. As long as there remains a non-trivial demand for reserve balances, while other retail and wholesale settlement methods might become much more prevalent, the federal funds rate target set by the Fed will still matter. Several reasons for the continued, non-trivial demand for reserve balances have been offered. Regarding the points made by Friedman, King, and others, and as Charles Freedman (2000) noted, Fed District Banks are unlike private clearing banks or private clearinghouses in that they can always provide intraday credit to banks that need it to settle payments without regard to their own profitability or to their own payment commitments (as in lender of last resort actions). Since the central banks of ECB nations are in this respect analogous to Fed District Banks rather than to private clearing corporations or private clearing banks, Friedmans analogy using the ECB nations central banks is not applicable. Also, because private clearinghouses sell their services based upon both efficiency and safety, and because the Fed is widely recognized as the safest counterparty, it is unlikely that all or even most of them would cease using reserve balances to settle net liabilities (Hawkins 2001). Finally, netted or gross settlement via other means exposes parties to the risk of default, and in the case of (netted or gross) settlement using private financial assets predicted by Palley, could expose a payee to the risk that an assets price would fall after the asset was acquired but before the payee could use it to discharge his/her own payment commitments.13 One possibility for ensuring a non-trivial demand for reserve balances that has been suggested is for the Fed to be granted the authority to pay interest on reserve balances (Woodford 2001, 2002;

Goodfriend 2002). With reserve balances as interest bearing assets, the primary motivation behind banks attempts throughout history to minimize their holdings of them would be eliminated (Feinman 1993). In addition to reducing the spread and thereby limiting the range of possible variability in the federal funds rate, the Fed would effectively become a minute-by-minute market maker in the federal funds market as the rate paid on reserve balances and the primary lending rate would become bid and ask rates, respectively. As Michael Woodford argued, since there would be no reason to borrow or lend in other short-term markets for rates outside of the Feds bid-ask rates except for the typical differences in maturity, liquidity, and default risk, There is little reason to fear t at either the development of h electronic cash for retail transactions or of alternative electronic methods of settlement of payments among banks should threaten a central banks ability to control the path of overnight nominal interest rates, and through them spending and pricing decisions in the economy (2000, 250). Friedman noted sympathetically that nobody should doubt that a large enough borrower or lender, willing to enter into transactions in infinite volume, can set market rates (2000, 269). But Friedman then raised concern that a central bank that sets bid-ask rates in the overnight market might be involved in potentially large transactions in order to sustain its role as a market maker. His concern was connected to his earlier puzzle that so few central bank operations are required to set and maintain the federal funds rate at its target. Friedman argued that it must be a central banks credible threat to engage in more operations if necessary that enables it to set rates with so few actual transactions and that such a credible threat would likely be necessary were the Fed to set bid-ask rates and become a market marker in the federal funds market (2000, 271). While the Fed can announce a new rate or reduce the spread through bid-ask rates, Friedman maintained that if the Feds willingness to carry out largescale transactions were ever doubted, In time, the market would cease to do the central banks work for it and the rates set by the Fed would then become decoupled from other interest rates and asset prices (2000, 271). As previously explained, the assumptions about the nature of the Feds activities implicit in Freidmans line of questioning were flawed, which led to his erroneous conclusions. To repeat, because open market operations simply offset net changes to the Feds balance sheet to accommodate the demand for overnight reserve balances and because the total quantity of reserve balances held overnight is relatively small, the Feds operations can be relatively small in value compared to transactions in financial markets. Even as operations are relatively small and while expectations of the Feds actions by federal funds market participants are no doubt importantrather than relying on a credible threatthe Feds almost daily open market operations must nonetheless balance the supply of reserve balances with the demand for them in order to ensure that the federal funds rate stays at its target on average. When deviations from the targeted rate occur, Krieger explained that the Fedinconsistent with the concept of a credible threatsupplies [reserve balances] in such a way that the rate is most likely to return to, and to continue to, trade near the target as soon as possible (2002, 74). The credible threat analogy is an inaccurate depiction of actual or potential Fed operations in part because it fails to recognize that more than being willing to engage in significant operations or to supply significant quantities of credit, the Fed is obligated to do so given the implications for the payments system and the federal funds rate if the Fed were to do otherwise. Recall that the Fed supplies intraday credit in rather substantial volume and value everyday to meet a significant intraday demand for reserve balances. Again inconsistent with the concept of a credible threat, the demand for intraday credit already tests daily the Feds willingness to supply credit on a large scale, to thereby ensure smooth functioning of the payments system, and to avoid sharp deviations from the targeted federal funds rate.14 Similarly, the heightened attention required of the Fed to the settlement needs of banks in planning its daily open market operations in the presence of reduce required reserves shows that the commitments to the payments system and the federal fund rate target are tested daily in the market for overnight reserve balances just as they are by the demand for intraday credit (Fullwiler 2003). Further evidence can be seen in times of recent crisis such as the 1987 stock market crash, the period leading up to the century date change, and the September 2001 terrorist attacks when the Fed relied not upon a credible threat but rather upon its abilitywhich is never in doubtto accommodate a markedly increased demand for

10

reserve balances in each case (Federal Reserve Bank of New York 2001; McAndrews and Potter 2002; Neely 2004). With interest payment on reserve balances and a reduced spread, the Feds obligations to the payments system and to the federal funds rate target would remain unchanged while its commitment to its own bid-ask rates would continue to be tested on a daily basis. The analogy is invalid also because it neglects the fact that other short-term interest rates are linked to the federal funds rate via arbitrage, not as a result of additional Fed operations in those markets or even of the threat of such operations. Recall that the Fed need only sustain its federal funds rate target for rates on similar types of short-term debt to follow.15 Reliance upon a credible threat could only apply where a particular interest rate or asset price is not directly related through arbitrage to the market rate being targeted by the Fed. For instance, in the summer of 2003 traders widely perceived that the Fed would engage in unconventional operations in which it would make markets in the longer term Treasury maturities to further reduce long-term rates while keeping the federal funds rate essentially unchanged. As long as market traders believed unconventional operations were a possibility, the yield curve flattened without any such operations actually occurring. In June 2003 when traders recognized such operations were unlikelythat is, that the Fed would be unwilling to engage in the unconventional operations since it no longer believed the operations would be necessarylonger-term Treasury rates quickly increased to earlier levels. Note that a credible threat in the case of a central bank that is unwilling to engage in large, market-making transactions effectively means that the central bank would not be engaged in targeting at all; indeed, it would be self-defeating for a central bank to desire a particular target and then be unwilling to carry out the market-making operations to ensure the target is sustained (assuming it has the ability to do so). To conclude this section, while the discussion in the previous section demonstrated that the Fed can set the federal funds rate and minimize the rates volatility, whether or not the federal funds rate will be able to affect other interest rates and asset prices has been questioned by many due to expected innovations in the payments system. Friedmans puzzle, which was a starting point for much research into the Feds ability to set rates, arises from a misunderstanding of actual Fed operations and of intraday credit. Rather than relying upon a particular quantity of overnight or intraday reserve balances in circulation or upon the continuation of the Feds supposed monopoly, the Feds ability to affect interest rates and asset prices through its federal funds rate target simply requires a non-trivial demand for reserve balances such that arbitrage between the federal funds rate and other short-term rates continues. If a nontrivial demand can be ensured, either through payment of interest on reserve balances or some other avenue (as discussed below), interest rate targets in the future will not require a credible threat in order to be sustained, just as interest rates are not now sustained via such a threat and current Fed operations are not necessarily applicable to Friedmans puzzle. In sum, concerns that the Fed could become an army with only a signal corps are overstated as long as a non-trivial demand for reserve balances is ensured.

III. THE FUNDAMENTAL SOURCE OF A NON-TRIVIAL DEMAND FOR RESERVE BALANCES

The previous section cited several reasons to suggest that a non-trivial demand for reserve balances will continue in the future even given the electronic money revolution. Note, however, that while the safety of Fed bank settlement and the Feds ability to supply lender of last resort services are significant, if they were the only reasons for use of reserve balances in settlement, one might logically conclude that the use of reserve balances to settle payments might one day be phased outas Friedman, Palley, and others already have suggestedabsent significant regulatory changes in the payments system. Indeed, it is precisely because safety and stability are the only characteristics of reserve balances taken into consideration that their continued use in settlement has been questioned. Similarly, while interest payments on reserve balances might raise the demand for reserve balances if banks have some reason to

11

hold reserve balances, suggesting that interest payments alone will ensure a non-trivial demand for reserve balances does not provide such a reason but rather presupposes one.16 Woodford, perhaps the most vocal proponent of interest payment on reserve balances, admitted as much when he allowed that, even given the possibility of interest payment on reserve balances, A future is conceivable in which improvements in the efficiency of communications and information processing so change the financial landscape that national central banks cease to control anything that matters to national economies (2001, 349). There is, though, a type of payment settlement for which only reserve balances will do, namely the settlement of payments with the federal government. The most fundamental and obvious of these is the payment of federal tax liabilities by corporations and individuals. Jordan and Stevens agreed that even with little public demand to hold central bank liabilities, central banks remain the only source of the national currency units that are required to settle domestic tax obligations (2002, 11). In a short, rather isolated passage that he discussed no further, Friedman also granted that
[a] potential solution that I suspect has a greater likelihood of success [is] requiring all government tax payments to be made in central bank liabilities. Tax payments in most modern economies do not constitute a small, potentially isolated market likely to end up as part of some corner solution [i.e., an interest rate that does not matter]. Most firms and most individuals pay taxes, many in sizeable amounts compared to their incomes or profits. Requiring them to do so in bank checks might go a substantial way toward keeping the demand for [reserve balances] coupled to the expansion or contraction of economic activity. (Friedman 2000, 265)

In the U.S., tax payments from corporations and individuals are already settled through banks via reserve balance debits when a credit is made to the Treasurys account at the Fed (Wray 1998; Bell 2000; Bell and Wray 2002-3). All federal tax payments by businessesincluding employee income tax withholding, Social Security/Medicare (FICA), and corporate income taxeshave traditionally been first credited to Treasury correspondent accounts held at thousands of commercial banks throughout the countrythe remit option type of Treasury Tax and Loan (TT&L) accountsand then transferred to the Treasurys account at the Fed on the next business day. The transfer of funds from a businesss account to a remit option TT&L account merely requires the change of ownership of deposits held at the bank and involves no reserve balances, while the transfer to the Treasurys Fed account debits both the TT&L account and the banks reserve balances. A majority and increasing proportion of taxes paid by businesses have since 1990 been transferred electronically through the Electronic Federal Tax Payment System (EFTPS); with EFTPS, tax payments are electronically transferred to the Treasurys account on the same business day they are received by the bank without first being held in a remit option TT&L account. Rather than having the use of non-interest earning (and reservable) Treasury deposits overnight, participating banks are paid a small fee for each tax payment processed via EFTPS. The exception occurs when a tax payment is processed by one of the 1,500 banks that provide the Treasury with note option TT&L accounts, which are callable balances similar to money market accounts that earn interest for the Treasury at a rate slightly below the federal funds rate. For a bank holding note option TT&L balances, tax payments are transferred to the note option account until called in by the Treasury, at which time a debit is made to the note option account and to the banks reserve balances. Collection of individual income tax payments is done in a similar manner, as the IRS deposits payments received into accounts at various commercial banks; after processing, payments are transferred to the Treasurys account at the Fed or to note option TT&L accounts until called by the Treasury. 17 The Treasurys use of TT&L note option accounts to manage its account balance at the Fed and the resulting simplification of the Feds monetary policy implementation have already been discussed in detail elsewhereincluding Lovett (1978), Hamilton (1997), Wray (1998), Bell (2000), Hillery and Thompson (2000), and Bell and Wray (2002-3)and are therefore summarized only briefly here. Note option accounts are manipulated daily by the Treasury for the purpose of sustaining its account balance at the Fed at a $5 billion targetexcept for periods when tax flows are highest, when the target rises to $7 billion or is ignoredby calling in funds from (adding to funds in) TT&Ls if the Treasury balance is

12

forecast to be below (above) the target. By maintaining a relatively unchanged balance in its account at the Fed through use of such calls and adds, the Treasury reduces the changes to reserve balances caused by daily flows to and from its account at the Fed. Since changes to the Treasurys account are typically the largest and least predictable part of the Feds balance sheet, calls and adds made by the Treasury significantly reduce balance sheet changes that must be offset by the Feds open market operations.18 The current TT&L system has been in place since 1977, though some system of Treasury correspondent accounts in private banks has existed since World War II; the exception was a brief interval during 19741977 for which time the bulk of the Treasurys funds were maintained in its account at the Fed and the Feds daily operations became decidedly more complex (Lovett 1978). Palley recognized that the reality is that taxes are paid using liabilities of the central bank, which creates a demand for reserves for purposes of tax payments (2001-2, 224). He later argued, however, that relying on the demand for tax settlement balances as the means of conducting monetary policy is . . . likely to be associated with increased interest rate volatility. This is because tax payments are highly seasonal, and taxes are also paid in arrears [and as a result the] central bank would have to engage in significant seasonal open-market operations . . .to smooth interest rate spikes . . . (2001-2 229). Here again, Palleys error was in misunderstanding the nature of the Feds daily operations. First, because the vast majority of flows to (from) the Treasury are offset by the Treasury itself through adds to (calls from) the TT&L note option accounts, it is doubtful that there would be a necessity for the Fed to engage in a significantly larger number of offsetting open market operations. Second, and more importantly, the potential for additional volatility in the federal funds rate would not be due to added variability/seasonality in payment flows to the Treasury but rather to the size of the spread between the rate paid on reserve balances and the penalty rate for borrowing reserve balances, as discussed earlier. Whether tax flows to the Treasury are only a few million dollars or less on some days and several billions on others, a demand for reserve balances based solely on tax paymentseven in the absence of offsetting TT&L calls or addswould bring additional volatility in the federal funds rate only if monetary policy operating procedures, regulations, and penalties that enable such volatility were in place. While requiring that tax payments be made in reserve balances is alone sufficient to create a nontrivial demand for reserve balances, the demand for reserve balances arising exclusively from payments that can only be settled via Fedwire is currently more significant than commonly thought. As mentioned, balances in TT&L note option accounts are frequently called in by the Treasury, particularly on days when revenues are smaller than scheduled disbursements; these calls can only be met with reserve balances. Settlement after auctions for Treasury securities are made in reserve balances, as are some securities sold by government sponsored enterprises.19 Further, all Treasury securities and many government agency/enterprise securities can only change ownershipwhether in the primary, secondary, or repurchase marketsthrough use of the Feds Fedwire book-entry securities system, which records changes in ownership against payment sent with reserve balances. (The notable exceptions among government agency/enterprise securities are those issued by GNMA, which do not trade over Fedwire, though FNMA and FHLMC securities do.) Consequently, many of the netted trades settled by FICC must settle through Fedwire.20 Although the majority of such securities (particularly Treasuries) are held on the books of only a few large clearing bankswho are then responsible for maintaining records to identify which securities are h on behalf of individual customerstotal securities trades settled via eld Fedwire still averaged more than $1 trillion per business day in 2003 (Board of Governors 2004). The major implication of the foregoing is that the Feds ability to set interest rates in an age of technological innovation in financial markets and (specifically) in the payments system is not much different from its ability to do so in earlier eras; the issue is not the innovations as much as it is the federal governments acceptation of reserve balances in payment settlement with the private sector. Thus, even as the current era is one of financial innovation, current and future eras nonetheless remain modern money eras (Wray 1998). Much as recent neo-Chartalist research into modern money has demonstrated that the State can create a demand for its own money by levying a tax liability payable in its money, the parallel here is that the central banks ability to set interest rates has similar origins since a non-trivial demand for reserve balances is ensured when reserve balances are necessary to settle tax liabilities. In a well-known

13

passage from A Treatise on Money, Keynes recognized the States role in declaring what qualifies as both a money of account and the thing which is delivered to discharge debts:
The State . . . comes in first of all as the authority of law which enforces the payment of the thing which corresponds to the name or description in the contracts. But it comes in doubly when, in addition, it claims the right to determine and declare what thing corresponds to the name, and to vary its declaration from time to time when, that is to say, it claims the right to re-edit the dictionary. This right is claimed by all modern States and has been so claimed for at least some four thousand years. (Keynes 1930, 4)

Again, the corollary here, as Charles Goodhart explained, is that the ability of the central bank to control interest rates is an issue of political economy. To ignore the role of governments and power would be to miss the key point (Goodhart 2000, 206; italics in original). It is useful to return to the earlier discussion of the Feds open market operations from a modern money perspective. Just as the State has the authority to determine that it will accept its own money in payment for taxes, so does it also have the authority to exercise complete control over the overnight interest rate earned by those lending out its money or paid by those borrowing its money independent of how much of its money is circulating. The rather complex nature of the Feds open market operations has obscured this fact for many economists for many years, leading them to erroneously believe in deposit multipliers and liquidity effects and to then fear that a fall in the quantity of reserve balances would reduce the Feds ability to set and manage volatility in the rate for which its own money is traded. Regardless how the Feds open market operations actually evolve in the future, the Feds ability to have direct control over the federal funds rate is completely unrelated to the quantity of reserve balances and is thus also unrelated to how many deposits are swept into money market accounts, how large reserve requirements are, or evolving payment technologies. While the quantity of reserve balances is endogenously determined and will vary depending upon banks abilities to avoid reserve requirements and technologies in the payments system, the State nonetheless always has within its power the ability to set the price for its own money and to minimize variations in this price regardless of the quantity of reserve balances in circulation, though it may choose not to use this power. From the modern money perspective, even if private clearing and settlement arrangements in the future were to completely eliminate the use of reserve balances to settle private transactions at the wholesale level (or if wholesale settlement itself were somehow eliminated), the fact that banks or other institutions could still be required to deliver reserve balances to the Treasury to settle their customers tax liabilities would be sufficient to ensure a non-trivial demand for reserve balances and the central banks ability to influence market interest rates through arbitrage between the overnight rate and other rates. The ability to set interest rates thus does not rely on the central banks monopoly, the safety of using reserve balances, utilization of the central bank as counter-party in clearing and settlement, or whether the central bank pays interest o reserve balances. Rather, what remains of each of these, as Wray noted, is the n perplexing question of why the non-State sector would accept a fiat moneythe States liabilitieswhen it is not exactly clear for what the State would be liable . By contrast, from a modern money perspective, the State is liable only to accept its fiat money in payments made from the non-State sector to itself (2003, 89). One final issue raised in the literature on central banks and the electronic money revolution is whether or not innovations in retail payments will effectively eliminate or at least severely reduce the quantity of currency in circulation and thereby reduce the governments seigniorage benefits of printing currency. 21 Stevens reminded that earnings on [Treasury] securities are the Federal Reserve Banks dominant source of income, most of which is paid to the U.S. Treasury as seigniorage. If privately issued electronic money someday were to displace todays currency holdings, the Federal Reserve Banks would lose close to $30 billion in annual revenue (2002, 2).22 Palley raised similar concerns that the electronic money revolution would have the effect of burdening taxpayers with the cost of replacing lost revenues from the Fed . . . as economic agents increasingly use newer payment media . . . in place of cash (2000, 9). Loss of seigniorage income has also been at the core of political resistance to the Fed paying interest

14

on reserve balances, since the Feds costs would be increased and fewer profits would be left for the Fed to turn over to the Treasury (Feinman 1993; Meyer 2000). Each of these arguments again misunderstands the Feds implementation of monetary policy and how the Treasurys operations are related to supporting the Feds federal funds rate target. Recall from the offsetting nature of the Feds operations in terms of the effects upon the Feds balance sheet, that in the absence of the Treasurys doing so, the Fed would be forced to engage in the same bond purchases or sales itself in order to sustain the federal funds rate at its target. This demonstrates that just as Treasury TT&L operations simplify the Feds open market operations by offsetting short-term changes to the Treasurys account arising from daily mismatches of revenues and expenditures, Treasury bond sales and purchases do the same operationally in the presence of larger and longer-term deficits or surpluses (Bell and Wray 2002-3). Were the Fed given authority to pay interest on reserve balances, both the Treasury and the Fed could be freed from carrying out bond sales (and TT&L calls) in the presence of a deficit to support the federal funds rate target; absent these bond sales to drain reserve balances, the federal funds rate would simply fall to the rate paid on reserve balances. As such, the interest rate paid on new additions to the national debt would fall to the rate paid on reserve balances; the decline in interest paid wouldif not immediately, then within a short period of timemore than offset any seigniorage loss from reduced Fed profits handed over to the Treasury. From an understanding of modern money, concerns over seigniorage loss were clearly misplaced. Since a sovereign currency issuing State is liable only to accept its fiat money in payments made from the non-State sector to itself, the State does not need nor use its own money received through taxes, bond sales, or profits turned over from the central bank to finance its spending; rather, from a basic understanding of the Feds balance sheet, the Treasurys receipt of reserve balances means that, by definition, reserve balances are taken out of circulation and effectively destroyed (Wray 1998; Bell 2000; Bell and Wray 2002-3).
While bank money (M1) is destroyed when demand deposits are used to pay taxes, the governments money, high-powered money, is destroyed as the funds are placed into the Treasurys account at the Fed. Viewed this way, it can be convincingly argued that the money collected from taxation and bond sales cannot possibly finance the governments spending. This is because in order to get its hands on the proceeds from taxation and bond sales, the government must destroy what it has collected. Clearly, government spending cannot be financed by money that is destroyed when received in payment to the State. (Bell 2000, 615)

Consequently,
The notion of a government budget constraint only applies ex post, as a statement of an identity that has no significance as an economic constraint. When all is said and done, it is certainly true that any increase of government spending will be matched by an increase of taxes, an increase of high powered money (reserves and cash), and/or an increase of sovereign debt held. But this does not mean that taxes or bonds actually financed the government spending. (Wray 2004, 6)

That tax revenues, bond sales, and seigniorage income do not finance government spending but are rather related, like calls from TT&L accounts, to the maintenance of the overnight rate at its target would be self evident were the Fed to pay interest on reserve balances. In the presence of a fiscal deficit, bond sales by the Treasury (or, as would be operationally equivalent, bond sales by the Fed) could be reserved for times when it was preferred that the non-bank public hold government bonds rather than bank liabilities, as in a macroeconomic policy driven by the principles of functional finance rather than by traditional views of sound finance (Lerner 1943; Mosler 1997-8; Nell and Forstater 2003). To conclude, the requirement that reserve balances be used by banks to settle depositors tax liabilities with the federal government is sufficient to ensure a non-trivial demand for reserve balances and the continued arbitrage between the federal funds rate and other short-term interest rates. As such, it is modern money in the sense defined by Wray (1998), not electronic money, that is fundamental to

15

understanding the Feds ability to set interest rates. While researchers have been led astray due to misinterpretations of an admittedly complicated process of monetary policy implementation, were one to begin from an understanding of modern money and the ability of a State issuing a sovereign currency to levy a tax liability in its own money, the ability of the central bank to set the overnight rate while minimizing its variability and to influence other rates through arbitrage would be clear even given the ever-accelerating pace of the electronic money revolution. Similarly, concerns over a loss of seigniorage income due to reduced use of State money in settlement of private payments or due to interest payments on reserve balances suffer from a misunderstanding of monetary policy implementation, a fact that is also easily recognized from a modern money perspective.

IV. CONCLUDING REMARKS

Because economists have historically misunderstood modern money, they have traditionally considered that the S tates money circulates due to the governments monopoly power of the printing press and/or the need to meet reserve requirements. Because they then also misunderstand the Feds implementation of monetary policy, they erroneously fear that the ability to set interest rates would be threatened as alternative, private settlement possibilities proliferate and reserve requirements fall. On the contrary, the States ability to set a tax liability for the public in its own money is the general, fundamental starting point for understanding why a non-trivial demand for the states money exists; the existence in history of reserve requirements and of periods in which the state exercised monopoly power in printing money are simply special cases that have clouded economists understanding of the demand for reserve balances and of the central banks ability to set interest rates. While Palley argued that the e-money revolution fits naturally into the history of money as told by Austrian economists [since it] emphasizes the endogeneity of the form of money, which changes in response to technical innovations and market competition (Palley 2001-2, 217-218), it is on the contrary into a modern money or neo-Chartalist view of moneywhich begins with a demand for the States money to settle tax liabilitiesthat the future of electronic money and the ability to set interest rates fit naturally. What matters is not whether there are other, private methods of settling payments, but whether the States money remains at the top of the economys hierarchy of money (Bell 2001). As Wray noted, it has not been uncommon historically for the States money to not be the primary or even most efficient means of settling private transactions, though in such cases the States money has still mattered given its acceptation of it for settling payments with the State (Wray 1998, Ch. 2). It is important to recognize that the increased variability in the federal funds rate and potential elimination of reserve balances in private settlement have themselves been influenced by the States ability to rewrite the dictionary; in the U.S., both have been heavily dependent upon the Feds own authorization of sweep account technologies, penalties imposed by the Fed on overdrafts (particularly the substantial penalties on overnight overdrafts), the non-monetary costs associated with borrowing at the discount window prior to 2003, legal prohibition of interest payment on reserve balances, and the requirement in the Monetary Control Act of 1980 that the Fed recoup its own imputed costs of capital by charging banks for its settlement services (which is explicitly intended t encourage private competition with the Feds o settlement services). Thus, while the State can ensure its own ability to set interest rates even in an era of revolution in payments technologies, this also means that it can likewise relinquish this ability if it so chooses. To recall an earlier quote from Goodhart, the ability to set interest rates is a matter of political economy. In fairness to Friedman, his ultimate concern was not necessarily how the Fed sets the federal funds rate, short-term rates, or other rates, but rather, whether the central bank can control the rates that matter for influencing economic activity (2000, 268; emphasis in original). Confusion arose when he and others attempted to link this concern with the fundamental source of a demand for reserve balances and the electronic money revolution. This paper has demonstrated that the fundamental source of a non-

16

trivial demand for reserve balances is not threatened by continuing innovations in the payments system; Sellons evidence that financial innovations throughout the last several years have created a closer link between the federal funds rate and other ratessince arbitrage has become more important in these marketssupports this conclusion. However, if we interpret Friedmans ultimate concern as a desire to more clearly understand the transmission mechanism of monetary policy, then it is obviously conceded that financial innovations impact significantly the Feds ability to manage the broader economy (e.g., Palley 2003). At the same time, concerns that the federal funds rate was a rather blunt instrument for managing the economy were pointed out well before the recent wave of financial innovations and the electronic money revolution by the likes of Minsky, Wray, Basil Moore, and by numerous others. To the extent that financial innovations in the payments system and elsewhere in the financial system might encourage economists to recognize the reality of modern money and to consider in more detail the actual operations and transmission mechanisms of monetary policy, research on the effects of such innovations may someday bear fruit even if there are substantial detours and wrong turns encountered along the way.

NOTES 1. Special thanks to Warren Mosler for helpful comments on a earlier version of this paper. All n remaining errors are mine. 2. The one caveat is that the concept of modern or sovereign money applies to economies that have flexible exchange rates. In the presence of fixed exchange rates, monetary policys overriding concern is maintenance of the fixed exchange rate; under such circumstances, the concept of modern or sovereign money does not apply (Wray 2003). 3. Banks can also contract with the Fed to hold required clearing balances during the maintenance period, which earn credits at the federal funds rate that can be used to pay for clearing and settlement costs incurred using the Feds services. Required clearing balances combine with required reserves to make up total required balances during a maintenance period. In recent years, required clearing balances have grown to outnumber required reserve balances as both required reserves and the federal funds rate have fallen. 4. The switch to lagged reserve accounting reduced uncertainty about the demand for reserve balances for the Fed in planning open market operations; it also reduced individual banks uncertainty about reserve needs throughout the maintenance period. In 2003, as discussed below, the Fed changed its discount window policy, which further reduced the possibility of substantial increases in the federal funds rate. 5. Even if reserve requirements were determined concurrently with the maintenance period, it is wellknown that neither banks nor depositors adjust portfolios within a period of time as short as the two week maintenance period. 6. While the demand for reserve balances is insensitive to interest rate changes in the short run, it is generally accepted that the sensitivity increases as the time horizon lengthens. However, such sensitivity should likewise not be interpreted within any sort of liquidity effect type of framework, either. Any correlation between the quantity of reserve balances and the federal funds rate is due to the fact that the Fed has accommodated an increased (reduced) demand for reserve balances after the federal funds rate had been reduced (increased) that had led deposits to rise (fall). The causation runs from changes in the federal funds rate target to medium- and longer-run adjustments by both banks (as the opportunity cost of holding excess balances changes) and the non-bank public (as the opportunity of holding deposits changes), which the Fed then accommodates. Thus, exogenous control over the quantity of reserve

17

balanceswhether via reserve targets or liquidity effectsis not possible in the short-run and occurs only indirectly via changes to the interest rate target over longer periods of time (Moore 1988; Wray 1998; Fullwiler 2003). 7. This is not to argue that ABRR would not be useful for other goals, such as providing countercyclical pressures on the creation of credit money. 8. Demiralp, et al (2004) find evidence of day-of-maintenance period and high payment flow day effects in overnight Eurodollar and repurchase agreement markets similar to those in the federal funds rate and demonstrate that arbitrage opportunities of only a few basis points are left unexhausted in these markets. 9. Fed Governor Ben Bernanke acknowledged, Loosely speaking, long-term interest rates embody the expectations of financial-market participants about the likely future path of short-term rates, which in turn are closely tied to expectations about the federal funds rate (2004, 4). 10. In countries without reserve requirements, such as England and Canada, the quantity of reserve balances is close to zero at the end of each day while intraday balances are integral for payment settlement. 11. Pending legislation would enable banks to make 24 sweeps per month between deposit accounts and money market deposit accounts, an increase from the six transfers allowed under current law; since there are no months with more than 23 business days, banks would eventually be able to sweep balances just before the end of each business day and thereby avoid any reserve requirements. This legislation, originally called the Small Business Interest Checking Act, had at the end of 2003 passed the House of Representatives but had not yet been passed by the Senate. The legislation also enables the Fed to pay interest on reserve balances. 12. Palley erred in suggesting that the quantity of reserve balances in circulation has anything to do with the Feds ability to set the federal funds rate. As discussed in the previous section, the Fed can set the rate at whatever level it wishes while also eliminating the rates variability regardless of the quantity of reserve balances in circulation. 13. While netted or gross settlement using deposits at clearing banks (rather than reserve balances) could (and does) involve the use of an asset whose nominal value is fixed, the fact that these deposits are FDIC insured (or at least that the liabilities of the banks involved are FDIC insured) and legally required to be convertible at par into currency are important reasons why their nominal val e does not change. If u assets/liabilities of varying creditworthiness were used in settlement, discounting would likely occur (Hawkins 2001, 100). 14. Note that this is consistent with the earlier discussion of reasons for variation in the federal funds rate. If the Fed were to be unwilling to supply intraday credit at the current low penalty rate, this is tantamount to raising the penalty rate and would put upward pressure on the federal funds rate as banks in need of reserve balances for settling payments entered the federal funds market in search of them. 15. The Fed does lend its securities to primary dealers on a regular basis at a slight penalty in order to aid smooth functioning of the repurchase agreement market; these actions, however, are m related to ore enabling smooth functioning of the payments system (in this case, securities settlement against payment) than to explicit market making in the repurchase agreement market. 16. Note that the interest earned on reserve balances would be paid in the form of additional reserve balances credited to banks reserve accounts. If reserve balances were no longer desired in payment

18

settlement or to meet reserve requirements, banks would similarly have no use for interest earned on reserve balances. 17. The discussion of the Treasurys tax collections and the TT&L system is based upon Financial Management Service (1998), United States Treasury (1999), and Federal Reserve Bank of New York (2001). 18. Daily changes in the Treasurys account at the Fed are shown to regularly be the largest and least predictable in the Federal Reserve Bank of New Yorks annual reports on open market operations. Absent manipulation of TT&L accounts by the Treasury, these changes would be far greater and far less predictable. 19. As Bell and Wray (2002-3) pointed out, Treasury auctions could operationally be settled via credits to TT&L accounts rather than via reserve balances immediately; this would enable sale of Treasuries without the usual accompanying drain of reserve balances and deposits. 20. An exception is FICC netted general collateral finance (GCF) repurchase agreements. GCF repos currently do not require any reserve balances for settlement because FICC will only handle those trades in which one of the two major clearing banks is involved (Bank of New York and JP Morgan Chase Bank). Prior to March 24, 2003, FICC would enable transactions in which both clearing banks were involved, which would require payment in reserve balances sent via Fedwire against the Treasuries changing ownership. Interbank GCF settlement was suspended by FICC due to concerns that either banks net liability might be more than the daylight overdraft allowed by the Fed, in which case the paying bank might delay some or all of its payment. See Christie (2003). 21. Note that a reduced demand for currency has nothing to do with whether the Fed can set interest rates, since the former would actually simplify the Feds open market operations as changes in currency in circulation are one of the key Fed balance sheet items that must be offset through open market operations. While banks purchase currency with reserve balances, given a tax liability payable in reserve balances, the elimination of the former motivation for holding reserve balances is immaterial to the ability to set interest rates. 22. Currency is purchased by banks from the Fed with reserve balances to replenish bank vault cash. The drain in reserve balances is offset by an open market purchase; consequently, as the public holds more currency, the Fed holds more Treasuries.

REFERENCES
Bank for International Settlements. 1989. Report on Netting Schemes. Committee on Payment and Settlement Systems Paper No. 2 (February). Bell, Stephanie. 2001. The Role of the State and the Hierarchy of Money. Cambridge Journal of Economics 25: 149-163. __________. 2000. Do Taxes and Bonds Finance Government Spending? Journal of Economic Issues, 34 (September): 603-620. Bell, Stephanie, and L. Randall Wray. 2002-3. Fiscal Effects on Reserves and the Independence of the Fed. Journal of Post Keynesian Economics 25 (Winter): 263-271.

19

Bennett, Paul, and Spence Hilton. 1997. Falling Reserve Balances and the Federal Funds Rate. Federal Reserve Bank of New York Current Issues in Economics and Finance, 3 (April). Ben Bernanke. 2004. Gradualim. Speech given at the University of Washington (May 20). Board of Governors. 2004. Fedwire Funds Service. http://www.federalreserve.gov/paymentsystems/fedwire. Christie, Rebecca. 2003. Risks Force Limitations on Popular Repo Market Product. The Wall Street Journal (March 12). Clouse, James A. and Douglas W. Elmendorf. 1997. Declining Required Reserve balances and the Volatility of the Federal Funds Rate. Federal Reserve Board, Finance and Economics Discussion Series (June). Costa Storti, Claudia, and Paul De Grauewe. 2001. Monetary Policy in a Cashless Society. Centre for Economic Polic y Research Discussion Paper No. 2696 (February). Cyree, Ken B., Mark D. Griffiths, and Drew B. Winters. 2003. On the Pervasive Effects of Federal Reserve Settlement Regulations. Federal Reserve Bank of St. Louis Review (March/April): 2746. Demiralp, Selva, Brian Preslopsky, and William Whitesell. 2004. Overnight Interbank Loan Markets. Federal Reserve Board, Finance and Economics Discussion Series (May). Federal Reserve Bank of New York. 2003. Domestic Open Market Operations during 2002. __________. 2000. Domestic Open Market Operations during 1999. __________. 2001. Fedpoint 21: Treasury Tax and Loan Program. http://www.newyorkfed.org/aboutthefed/fedpoint/fed21.html. Federal Reserve Bank of St. Louis. 2004a. http://research.stlouisfed.org/swdata Sweeps of Transaction Deposits into MMDAs.

__________. 2004b. Total Checkable Deposits. http://research.stlouisfed.org/fred2/data/TCDNS.txt Feinman, Joshua. 1993. Reserve Requirements: History, Current Practice, and Potential Reform. Federal Reserve Bulletin (June): 569-589. Financial Management Service. 1998. EFTPS http://www.fms.treas.gov/eftps/203fact.html Fact Sheet, (February). Website:

Freedman, Charles. 2000. Monetary Policy Implementation: Past, Present, and FutureWill Electronic Money Lead to the Eventual Demise of Central Banking? International Finance 3 (July): 211-227. Friedman, Benjamin M. 2000. Decoupling at the Margin: The Threat to Monetary Policy from the Electronic Revolution in Banking. International Finance 3 (July): 261-272.

20

__________. 1999. The Future of Monetary Policy: The Central Bank as an Army with Only a Signal Corps? International Finance 2 (December): 321-338. Fullwiler, Scott T. 2003. Timeliness and the Feds Daily Tactics. Journal of Economic Issues 37 (December): 851-880. Furfine, Craig. 2000. Interbank Payments and the Daily Federal Funds Rate. Journal of Monetary Economics 46 (August): 535-553. Goodfriend, Marvin. 2002. Interest on Reserves and Monetary Policy. Federal Reserve Bank of New York Economic Policy Review (May): 7784. __________. 2000. Can Central Banking Survive the IT Revolution? International Finance 3 (July): 189-210. Gormez, Yuksel, and Forrest Capie. 2000. Surveys on Electronic Money. Bank of Finland Discussion Paper No. 7/2000 (June). Government Accounting Office. 2002. Payment Systems: Central Bank Roles Vary, but Goals Are the Same. GAO-02-303 (February). Hakkio, Craig S. and Gordon H. Sellon, Jr. 2000. The Discount Window: Time for Reform? Federal Reserve Bank of Kansas City Economic Review, 85 (Second Quarter): 5-20. Hamilton, James. 1997. Measuring the Liquidity Effect. American Economic Review: (March): 8097. Hawkins, John. 2001. Electronic Finance and Monetary Policy. Bank for International Settlements Papers No. 7 (November). Hillery, Paula V. and Stephen E. Thompson. 2000. The Federal Reserve Banks as Fiscal Agents and Depositories of the United States. Federal Reserve Bulletin, (October): 251-259. Keynes, John Maynard. 1976 [1930]. A Treatise on Money. New York: Harcourt, Brace, and Company.

King, Mervyn. 1999. Challenges for Monetary Policy: New and Old. New Challenges for Monetary Policy: A Symposium Sponsored by the Federal Reserve Bank of Kansas City, 11-58.

Krieger, Sandra C. 2002. Recent Trends in Monetary Policy Implementation: A View From the Desk. Federal Reserve Bank of New York Economic Policy Review (May): 73-76. Lerner, Abba. 1943. Functional Finance and the Federal Debt. Social Research 10 (February): 38-51. Lovett, Joan E. 1978. Treasury Tax and Loan Accounts and Federal Reserve Open Market Operations. Federal Reserve Bank of New York Quarterly Review (Summer): 41-46.

21

McAndrews, James J. and Simon M. Potter. 2002. Liquidity Effects of the Events of September 11, 2001. Federal Reserve Bank of New York Economic Policy Review (November): 59-79.

McAndrews, James J. and Samira Rajan. 2000. The Timing and Funding of Fedwire Funds Transfers. Federal Reserve Bank of New York Economic Policy Review (July): 17-32.

Meyer, Laurence. 2000. Payment of Interest on Reserves and Fed Surplus. Testimony delivered before the Committee on Banking and Financial Servic es, U.S. House of Representatives, (May 3). http://www.federalreserve.gov/boarddocs/testimony/2000/20000503.htm Minsky, Hyman P. 1957. Central Banking and Money Market Changes. Quarterly Journal of Economics, 71 (May): 171-188. Moore, Basil. 1988. Horizontalists and Verticalists: Cambridge: Cambridge University Press. The Macroeconomics of Credit Money.

Mosler, Warren. 1997-8. Full Employment and Price Stability. Journal of Post Keynesian Economics, 20 (Winter): 167-182. Neely, Christopher J. 2004. The Federal Reserve Responds to Crises: September 11th Was Not the First. Federal Reserve Bank of St. Louis Review (March/April): 27-42. Nell, Edward J. and Mathew Forstater (eds.). 2003. Reinventing Functional Finance: Transformational Growth and Full Employment. Northampton: Edward Elgar. Palley, Thomas I. 2003. Monetary Policy in an Era of Financial Deregulation and Innovation. Presented at the 2003 Meetings of the Association for Evolutionary Economics. __________. 2001-2. The E-Money Revolution: Challenges and Implications for Monetary Policy. Journal of Post Keynesian Economics 24 (Winter): 217-233. __________. 2000. Stabilizing Finance: The Case for Asset-Based Reserve Requirements. Financial Markets Center Financial Markets and Society (August). Panigay Coleman, Stacy. 2002. The Evolution of the Federal Reserves Intraday Credit Policies. Federal Reserve Bulletin (February): 67-84. Sellon, Gordon H., Jr. 2002. The Changing US Financial System: Some Implications for the Monetary Transmission Mechanism. Federal Reserve Bank of Kansas City Economic Review (1st Quarter): 5-34.

22

Sellon, Gordon H., Jr., and Stuart E. Weiner. 1997. Monetary Policy without Reserve Requirements: Case Studies and Options for the United States. Federal Reserve Bank of Kansas City Economic Review, 82 (Second Quarter): 5-30. Stevens, Ed. 2002. Electronic Money and the Future of Central Banks. Federal Reserve Bank of Cleveland Economic Commentary (March 1). United States Treasury. 1999. Treasury Financial Manual, Volume IV. Woodford, Michael. 2001. Monetary Policy in the Information Economy. Economic Policy for the Information Economy: A Symposium Sponsored by the Federal Reserve Bank of Kansas City, 297-370. __________. 2000. Monetary Policy in a World without Money. International Finance 3 (July): 229260. Wray, L. Randall. 2004. International Aspects of Current Monetary Policy. Employment and Price Stability Working Paper No. 31 (March). Center for Full

__________. 2003. Seigniorage or Sovereignty? In Sergio Rossi and Louis Philippe-Rochon, eds., Modern Theories of Money: The Nature and Role of Money in Capitalist Economies, Aldershot: Edward Elgar, 84-102. __________. 1998. Understanding Modern Money: The Key to Full Employment and Price Stability. Aldershot: Edward Elgar.

23