Daylight overdraft fees and the Federal Reserve's payment system risk policy.

In April 1994 the Federal Reserve began charging fees for daylight overdrafts incurred in accounts at Federal Reserve Banks. The event was an important step in the Federal Reserve's ten-year program to control daylight overdrafts and their associated payment system risk. The fees produced a dramatic decline in overdrafts and led to significant changes in market practices, particularly in the government securities market. This article summarizes the results to date of the implementation of the Federal Reserve's daylight overdraft program, focusing on the effect of fees. It also highlights related policies for large-value payment systems in the private sector.

BACKGROUND

Daylight overdrafts are a form of intraday credit in which a holder of a deposit account at a bank or other depository institution runs a negative balance in its account during the day but ends the day with a balance equal to or greater than zero. An example of a daylight overdraft is the case in which a deposit account holder (1) makes withdrawals from the account in the early part of the day that exceed the account's opening cash balance and (2) does not make deposits sufficient to cover the withdrawals until later in the day. When the customer's withdrawals require its bank to send payments through Federal Reserve systems, this type of customer activity can, in turn, cause daylight overdrafts in the Federal Reserve account of the customer's bank. In addition, the bank's own activity, such as federal funds borrowing and lending and the associated payment activity, can also cause daylight overdrafts.

Historically, intraday credit extensions went largely unmeasured because attention within the banking industry was focused mainly on processing payments rather than on managing associated intraday risks. In addition, nearly unlimited intraday credit was available from the Federal Reserve at no cost. Banks normally accumulated payment instructions during the day and posted them to customers' accounts at the close of the banking day, a practice that can be viewed as reasonable and cost-effective given the operational and legal infrastructure at that time. As a result, however, settlement conventions related to the transactions most often characterized by same-day settlement--such as financing in the markets for federal funds and repurchase agreements--came to be based upon the availability of unlimited, free intraday overdraft credit from the Federal Reserve.

Daylight overdrafts in accounts at Federal Reserve Banks in fact arise from a variety of causes. The Federal Reserve's original overdraft policies were focused on intraday credit resulting from the transfer of funds through Fedwire, the Federal Reserve's electronic, real-time funds transfer system often used for large-value interbank payments. Overdrafts are, however, also caused by payments of banks and their customers that arise from transfers of government securities via Fedwire as well as from non-Fedwire transactions such as automated clearinghouse (ACH) payments, checks, and other payment activity posted directly to depository institution accounts at the Federal Reserve.

For many institutions, payments made on a given day may exceed that day's opening balance (the previous day's overnight balance) with the Federal Reserve. Indeed, in 1994, the total value of funds and securities transfers through Fedwire, ACH payments processed by the Federal Reserve, and checks cleared through the Federal Reserve averaged $1.5 trillion per day; in contrast, the average daily opening balance of all depository institution accounts at the Federal Reserve that year was $32 billion (depository institutions maintain balances to satisfy reserve requirements as well as to clear payments). Many of these payments were covered by funds in the accounts, but a significant portion were not. In 1994, for example, direct measures of daylight overdrafts show that the Federal Reserve extended about $50 billion in intraday overdraft credit on average during the day (table 1); and on any given day an average of about 2,400 institutions incurred daylight overdrafts in their accounts at the Federal Reserve out of a total of nearly 11,000 institutions with such accounts.

Private payment systems such as the Clearing House Interbank Payment System (CHIPS) also involve extensions of intraday credit.(1) Owned and operated by the private New York Clearing House, CHIPS currently has a membership of more than 100 banks, a large number of which are branches and agencies of foreign banks. Through CHIPS, members can send electronic payment messages to one another during the day. The value of the payment messages is settled on a multilateral net basis at the end of the business day. No accounts analogous to those at the Federal Reserve exist in CHIPS, but members are exposed to credit risk associated with payment messages they have received until settlement has been successfully completed each day. CHIPS currently handles a

greater dollar value of payments than does Fedwire--in 1994, about $1.2 trillion per day for CHIPS, and more than $800 billion per day for Fedwire funds transfers.

ORIGINS AND DEVELOPMENT OF THE PAYMENT SYSTEM RISK POLICY

The Federal Reserve's efforts to measure and control daylight overdrafts and other risks in the payment system date from the late 1970s and early 1980s.(2) During that period, public policy concerns were first raised about the size and growth of intraday credit related to large-value payment systems.

Because Fedwire payments are final and irrevocable, the Federal Reserve bears some degree of intraday credit risk when such payments are processed without sufficient funds in the sending bank's account.(3) Daylight overdrafts, if not repaid by, the end of the day, could readily become unsecured overnight overdrafts. The Federal Reserve strongly discourages overnight overdrafts by imposing high monetary penalties on them and by taking administrative action against institutions that incur them repeatedly.

Concerns about the magnitude of the credit risk borne by the Federal Reserve in regard to daylight overdrafts were heightened in the early 1980s. During that time, the Reserve Banks began to monitor the intraday account balances of institutions routinely and to collect data on institutions with the largest daylight overdrafts. These data indicated that aggregate daylight overdrafts on any given day often ran into the billions of dollars; furthermore, overdrafts for a small group of institutions often exceeded their capital by several times.

Also during the early 1980s, intraday credit risk exposures in private systems, such as CHIPS, drew the attention of the Federal Reserve and private participants. Under the rules then in effect, the default of a large participant in CHIPS before end-of-day settlement could have caused the unwinding of that day's net settlement on CHIPS, leaving some other participants with a large, sudden shortage of funds.(4) The potential systemic repercussions of such a scenario were substantial given the increasing volumes of large-value interbank transactions being processed over CHIPS.

Prompted by these concerns, the Federal Reserve Board and private-sector groups began studies of the causes and potential means of controlling payment system risk, including the risk arising from daylight overdrafts in Federal Reserve accounts. Between 1982 and 1988, they produced a series of reports that analyzed a variety of potential means of reducing overdrafts.(5) Such measures included quantitative limits on daylight overdrafts (caps), fees, collateral, and increases in minimum required account balances.

In April 1984 the Federal Reserve Board issued for public comment a proposed policy statement on reducing risk in large-dollar transfer systems. The stated objectives of the policy at that time were to contain the effects of a settlement failure, reduce the volume of intraday credit exposures, control the remaining credit risk, and promote the smooth operation of the payment system. The 1988 study by the Federal Reserve, Controlling Risk in the Payments System, also included objectives of rapid final payments, low operating expense for making payments, equitable treatment of payment system participants, and effective tools for implementing monetary policy. Although the Fedwire system is capable of automatically blocking funds transfers that would create an overdraft (and is currently used to do so in certain circumstances), the Federal

Reserve did not endorse an outright prohibition of daylight overdrafts as a necessary or desirable policy outcome. Indeed, concerns expressed then as well as now are that the prohibition of all overdrafts could seriously disrupt the U.S. money markets, especially the markets for federal funds and government securities.

As a result of these efforts, the Board in 1985 adopted a program of maximum limits, or net debit caps, for each institution on the combined intraday credit provided both by the Federal Reserve and by participants in CHIPS. These net debit caps were based on an institution's own assessment of its capacity to absorb and control daylight credit risks, including its financial condition and liquidity resources, controls on intraday credit extensions to customers, and its ability to monitor payments on an intraday basis.

Under the 1985 initiative, the level of the cap could be as high as three times an institution's regulatory capital. The Federal Reserve monitored daylight overdrafts daily as well as on average over each two-week reserve maintenance period, and the Reserve Banks took administrative measures aimed at reducing the overdrafts of institutions that repeatedly exceeded their caps. The Reserve Banks also retained the flexibility to deal with troubled institutions as circumstances required. The Federal Reserve has modified its policy on caps several times, most recently by incorporating a streamlined approach to assessing an institution's creditworthiness and by including an assessment of its operating controls and contingency procedures.

In 1989 the Federal Reserve issued a policy statement aimed at reducing systemic risk in private, large-dollar payment systems. In part, this policy addressed concerns that efforts to reduce the intraday credit risk borne by the Federal Reserve, such as by reducing net debit caps or charging fees for the use of Federal Reserve daylight credit, would cause risks to be shifted to private networks. By 1984 the New York Clearing House had implemented bilateral credit limits among participants in CHIPS, and by 1986 it had implemented net debit caps. In 1990, the New York Clearing House adopted explicit arrangements for loss sharing and took other steps to strengthen settlement procedures in CHIPS. As a result, the Federal Reserve eliminated its net debit caps on CHIPS positions as of early 1991; the CHIPS caps were thereafter monitored automatically within the CHIPS system.

Fee Policy on Daylight Overdrafts

Studies conducted by the private sector and by the Federal Reserve in the late 1980s concluded that charging fees to reduce daylight overdrafts and their associated risk would be preferable to other means, such as collateralization and more restrictive net debit caps. The Federal Reserve also expected fees on daylight overdrafts to provide market-oriented incentives that would distribute Federal Reserve intraday credit more efficiently, that is, to those institutions that value it most highly at a given time. In 1992, after considering public comments, the Federal Reserve Board announced its intention to charge fees for overdrafts, beginning in April 1994, as a supplement to the existing net debit cap policy, which the Board decided to retain in addition to fees. In effect, this decision left institutions with two moderating influences on their use of Federal Reserve credit: a credit limit (the net debit cap) and an explicit price on the use of that credit (the daylight overdraft fee).

The fee was set at a level intended to be comparable to the cost of measures institutions could take to avoid daylight overdrafts. The Board set the initial fee at an annual rate of 10 basis points (0.10 percent) of chargeable daily daylight overdrafts, effective April 14, 1994. The chargeable overdraft is the institution's average per-niinute daylight overdraft for a given day, less a deductible amount equal to 10 percent of its risk-based capital. The Board also scheduled increases in the fee, to 20 basis points in April 1995 and 25 basis points in April 1996, while reserving the right to change the the level of these fees or the timetable for their implementation in light of experience with the program.(6)

In early 1995, the Board determined that a smaller fee increase for April 1995, to 15 basis points, would be more appropriate than the doubling, to 20 basis points, originally planned. At that time, the Board stated that, rather than increase the fee again in April 1996, it intended to wait two years before evaluating the results of the April 1995 increase. The Board's decision to moderate, but not eliminate, the scheduled fee increase was based on three considerations. First, as discussed in greater detail below, the response by depository institutions and their customers to the 10 basis point fee had significantly reduced the use of Federal Reserve daylight credit. Second, the Board believed that some increase would provide a significant incentive for depository institutions and their customers to further evaluate and modify payment practices that create daylight overdrafts. Third, the Board was concerned that the more rapid increase in fees originally planned could cause a substantial volume of large-value payments to be shifted to less secure payment systems and thereby increase payment system risk.

EXPERIENCE WITH THE PAYMENT SYSTEM RISK POLICY: 1986-93

During the period between the imposition of caps in March 1986 and the implementation of fees in April 1994, daylight overdrafts in Federal Reserve accounts increased almost continuously. The increase is evident in two measures of aggregate daylight overdrafts, the peak and the per-minute average (chart 1). The peak daylight overdraft can be viewed as the Federal Reserve's maximum intraday credit exposure to all institutions combined, at any particular time during the day; and the average per-minute daylight overdraft is the Federal Reserve's average exposure to all institutions over the course of the day (and is the base upon which daylight overdraft fees are assessed).(7) During the 1986-93 period, peak and average overdrafts both grew at an average annual rate of about 12 percent; in fact, beginning in 1989, overdrafts increased dramatically despite the fact that cap levels had been reduced the year before.

Securities activity on Fedwire appears to have generated much of the rise in overdrafts. Overdrafts related to securities transfers were monitored separately because they were afforded special treatment under the net debt cap policy (as discussed below). Although the method of allocating total daylight overdrafts into a securities-related portion and a funds-related portion is based on an accounting procedure that can be viewed as somewhat arbitrary, the distinction between the two is important to an understanding of the major trends in overdrafts during this period.

Securities-Related Overdrafts

Concerns that caps on securities-related overdrafts might disrupt the government securities market led the Federal Reserve to apply net debit caps at the outset to only the portion of daylight overdrafts not related to transfers of securities through Fedwire. Between 1986 and 1993, the size of securities-related daylight overdrafts more than doubled (chart 2), and their proportion of total overdrafts swelled from about one-half to about two-thirds. In early 1988, the Federal Reserve adopted a $50 million limit on the size of securities transfers to discourage "position building" by dealers, a practice that had contributed to overdrafts.(8) Securities-related overdrafts continued to rise, however. In 1991, the Federal Reserve began including securities-related overdrafts within the measure of overdrafts subject to a cap. But the Federal Reserve allowed financially healthy institutions to exclude securities-related overdrafts from the cap by pledging collateral against those overdrafts, and it mandated pledging of collateral for institutions that breached their cap as a result of frequent and material securities-related overdrafts. Indeed, pledging collateral became standard practice among the small group of securities clearing banks whose customers' activity generated substantial daylight overdrafts. Thus, even after 1991, net debit caps did not have a significant effect on the majority of overdrafts related to securities transfers.

The most likely cause of the sharp rise in securities-related daylight overdrafts during the 1989-93 period was increased activity in the market for repurchase agreements (RPs). Securities dealers commonly use RPs for overnight or very short term financing of the securities they hold. In the case of overnight RPs, the borrower of funds (or the borrower's clearing bank in the case of nonbank dealers) typically delivers securities used as collateral to the lender in the afternoon and at the same time receives funds in return. In the morning the lender of funds returns the securities (collateral) to the borrower (or its bank) and receives its funds in exchange. When the borrower and the lender do not use the same clearing bank, this process involves tranfers of securities against payment via the Fedwire securities transfer system and typically generates overdrafts in the Federal Reserve accounts of the clearing banks. These overdrafts start in the morning and extend into the afternoon, when new RPs are arranged and settled.

No comprehensive measures of RP market activity are available, but data collected by the Federal Reserve on RP positions of primary dealers in U.S. government securities provides an approximate picture of trends in this market. Indeed, the correlation between the growth in dealer RP positions and securities-related daylight overdrafts over the 1989-93 period has been fairly close (chart 3).

Funds-Related Overdrafts

During the 1986-93 period, net debit caps appear to have been successful in restraining the growth of funds overdrafts (those related to Fedwire funds transfers as well as ACH payments, checks, and miscellaneous other payments posted to Federal Reserve accounts). Although not always binding for all institutions, caps provide an incentive to control intraday account balances. For example, to reduce the probability of breaching caps, some institutions have installed automated systems for managing and queuing payments they send.

Moreover, because net debit caps are proportional to each institution's capital base, the growth of funds-related overdrafts, if controlled, is likely to be limited by the overall increase in the capital of institutions that typically incur overdrafts. In fact, during the 1986-93 period, daylight overdrafts related to Fedwire funds transfers and other payments grew at 7 percent annually, on average, roughly in line with growth in aggregate equity capital at all U.S. commercial banks. In turn, that rate was less than the 8 percent growth rate of overall Fedwire funds transfer activity during this period.

The level of funds overdrafts (as well as of total overdrafts) was further affected by an October 1993 change in the method of measuring daylight overdrafts. The Federal Reserve's original method tended to create additional intraday credit ("float") because credits for certain types of non-Fedwire payments were posted before corresponding debits. This float tends to reduce measured overdrafts by providing an implicit source of intraday credit.

The new method of measuring daylight overdrafts was designed to satisfy four basic principles: the measurement method should (1) not provide intraday credit through float, (2) reflect the legal rights and obligations of parties to a payment, (3) allow institutions to control their use of intraday credit, and (4) not give a competitive advantage to the payment services offered by the Reserve Banks. The new method measured daylight overdrafts comprehensively by posting non-Fedwire payments, such as checks and ACH payments, according to a predetermined schedule based on the type of payment and the time it was processed. Fedwire payments continued to be posted to accounts at the time they were executed. At the time it was implemented, the 1993 change in the method of measurement reduced average net intraday float about $25 billion (as reflected in the level of average net intraday balances--aggregate positive balances less aggregate negative balances). The reduction in float reduced the implicit credit available to cover daylight overdrafts and therefore caused such overdrafts to increase, as measured, roughly $10 billion.(9)

EFFECT OF FEES ON OVERDRAFTS

The results of the initial implementation of daylight overdraft fees has constituted the most significant aspect of recent experience under the Federal Reserve's daylight overdraft program. The implementation of the 10 basis point fee on April 14, 1994, had an immediate and dramatic effect on daylight overdraft levels as well as on their typical intraday pattern (chart 4).

All measures of intraday overdrafts declined significantly beginning on April 14, 1994. Indeed, aggregate intraday peak overdrafts fell approximately 40 percent, from nearly $125 billion per day, on average, during the six months preceding April 14, to about $70 billion in the six months following April 14 (chart 5). Average per-minute overdrafts, the base measure upon which fees are assessed, also declined 40 percent, from $70 billion, on average, to $43 billion. Although these data are not adjusted for seasonality and other factors affecting longer-ter-in trends in overdrafts, the direction and general magnitude of the shift is unmistakable. The effect has been proportionally larger for securities-related overdrafts: Over the six months following April 14, securities-related daylight overdrafts decreased about 45 percent, and funds overdrafts decreased about 25 percent.

Preliminary evidence from the April 1995 increase in the fee, to 15 basis points, does not, to date, show a significant additional effect on the level of daylight overdrafts (chart 5 and table 2). Per-minute overdrafts have averaged $43 billion in the six months since the increase in the fee, the same level as during the analogous period in 1994. Institutions may have already implemented new systems and procedures in the period leading up to April 1994 based on an expectation that fees would eventually increase to 25 basis points. Alternatively, daylight overdraft levels may not be especially sensitive to relatively small changes in the level of the fee, or other factors may have offset any observable effect from the fee increase.

When viewed over the longer term, the overall reduction in overdrafts has been particularly striking. In constant-dollar terms, the implementation of fees has brought peak daylight overdrafts down below the levels of the mid-1980s (chart 6). Another useful scale by which to analyze long-term trends in overdrafts is the dollar value of payments made over the Fedwire funds and securities transfer system. Relative to the total dollar value of such payments, funds- and securities-related overdrafts have now fallen to about the levels of 1988 (chart 7). This pattern suggests that the explicit charges for the use of Federal Reserve daylight credit has resulted in a significantly more efficient use of such credit for a given volume of payments relative to most of the experience of the last decade.

DISTRIBUTION OF DAYLIGHT OVERDRAFTS AND DAYLIGHT RESERVES

In addition to reducing daylight overdrafts in the aggregate, fees have also led to some extent to a redistribution of daylight overdrafts across institutions. Historically, overdrafts have been highly concentrated among a few institutions. In the six months preceding the implementation of fees, for example, the ten institutions with the largest overdrafts accounted for 80 percent of total average overdrafts. The reduction in daylight overdrafts has been similarly concentrated: More than 90 percent of the Systemwide reduction in average overdrafts has come from the six institutions that typically incur average per-minute daylight overdrafts of more than $1 billion; overdrafts among these banks have fallen $25 billion overall, or 45 percent, over the six months before and after the implementation of fees. As a result, the ten institutions with the largest overdrafts now account for about 70 percent of overdrafts.

Under the daylight overdraft fee policy, net debit caps have remained in effect and are not related to the assessment of fees. For all but the largest institutions, net debit caps have continued to act as the more relevant factor constraining daylight overdrafts. For example, even after the 1995 fee increase, only about 90 institutions typically incurred fees of more than $100 in any two-week period. In contrast, caps seemed to be the effective boundary for management of daylight overdrafts for an average of more than 700 institutions (those with peak overdrafts of at least 75 percent of their net debit caps in any given two-week reserve maintenance period and who tended to incur zero or very low fees).

Another important aspect of reduced daylight overdrafts has been the effect on institutions that do not typically incur overdrafts but rather hold positive intraday balances. With the exception of activity that adds or removes intraday reserves from the entire banking system, such as open market operations, changes in intraday float, or changes in Treasury cash balances, Federal Reserve accounts constitute a "closed" system. Thus, in general, for each dollar of a negative balance, or overdraft, in one institution's account, one or more other institutions must hold a corresponding dollar of positive balances. Consequently, with the reduction in daylight overdrafts has come an overall reduction in positive intraday balances in accounts held at the Federal Reserve.

In the six months after the implementation of fees, positive intraday balances held by depository institutions in Federal Reserve accounts averaged about $70 billion, compared with $90 billion in the preceding six months. A small group of other institutions, particularly government-sponsored enterprises, also hold large positive intraday balances at the Federal Reserve and have seen a similar reduction in these intraday balances since the onset of fees. Depository institutions that typically carry substantial positive intraday balances at the Federal Reserve include custodian banks that hold securities and funds on behalf of institutional investors such as mutual funds. These custodians often provide overnight financing in the form of repurchase agreements secured by government securities. Changes in the timing of RP settlements as a result of daylight overdraft fees, discussed below, has caused the duration of these loans to be lengthened during a given twenty-four-hour period.

EFFECT ON MARKETS

The most noticeable market response to the implementation of fees on daylight overdrafts has been reflected in daily movements of government securities. In response to daylight overdraft fees, US. government securities dealers began arranging their financing transactions earlier in the morning and delivering securities used as collateral for RPs more quickly to their counterparties to cover overdrafts caused by early-morning repayment of maturing RPs. Traders reportedly facilitated faster back-office processing by pricing the securities to be used as collateral at the time of the trade rather than later in the morning, as had been the practice. Dealers also completed settlement of many secondary market trades in government securities earlier in the day. These activities not only reduced securities-related overdrafts but shifted the overall intraday peak overdraft to earlier in the day (chart 4).(10)

The anecdotal evidence on the earlier shift in trading and settlement practices in the government securities market is supported by data on the timing of securities transfers over Fedwire. Before April 1994, approximately 30 percent of the total daily value of securities transfers was processed by 10:00 a.m. eastern time, 40 percent by 12 noon, and 90 percent by 2:30 p.m. (chart 8), and Fedwire securities transfer operations were frequently extended to 4:00 p.m. or later because of surges in transfer volume in the early afternoon. These proportions began to rise significantly with the April 1994 implementation of overdraft fees and continued to rise with the April 1995 increase in fees. As a result, the Fedwire securities transfer system has been able to close, on average, within

approximately fifteen minutes of its designated final closing time of 3:00 p.m. In part because of this development, the Federal Reserve has now established a firm final closing time of 3:30 p.m., effective January 1996, for the Fedwire book-entry securities transfer system.

The timing of Fedwire funds transfers, however, has moved in the opposite direction, though in a much less dramatic fashion than securities transfers. After the implementation of daylight overdraft fees, the portion of the total daily value of Fedwire funds transfers originated by noon decreased approximately 5 percent, or roughly $40 billion, relative to the period six months before fees were implemented (chart 9). This pattern is not surprising given that the Federal Reserve account of an institution initiating a Fedwire funds transfer is debited (rather than credited as in the case of a securities transfer). As a result, institutions naturally have an incentive to delay less time-critical payments in order to reduce potential daylight overdrafts. The intraday pattern of funds-related overdrafts has likewise shifted to later in the day as a result of the later average transfer pattern. According to discussions with banks, this shift appears to have been caused more by increased use of their existing funds-transfer queuing capabilities rather than by more explicit measures, such as charging customers higher fees to send payments earlier in the day.

INCIDENCE OF FEES

In the first twelve months of the daylight overdraft fee program, the Federal Reserve collected a total of $18.5 million in fees from 280 depository institutions. Of these institutions, approximately 110 have incurred charges in at least one of every two reserve maintenance periods. Since the 50 percent increase in the fee in April 1995, overdraft charges have averaged $27 million at an annual rate, and about 120 institutions have incurred fees regularly.

Like daylight overdrafts, daylight overdraft charges have been highly concentrated among a few institutions. In fact, the concentration of fees is even greater than that of overdrafts because of the overdraft deductible and the waiving of fees of less than $25; these features effectively exempt many smaller-overdrafting institutions from daylight overdraft fees in most periods. The ten institutions with the largest charges accounted for 86 percent of the total. The largest U.S. banks (those with assets of more than $10 billion) paid, on average, 92 percent of total charges.

Discussions with larger banks indicate that those incurring the highest charges pass them on to those customers whose activity generates much of the banks' overdrafts. In particular, the securities clearing banks, who have experienced the largest reductions in overdrafts, have charged fees to their securities-dealer customers whose market activity, particularly RP transactions, creates overdrafts in the Federal Reserve accounts of the clearing banks. These customer fees are assessed against measures of the overdrafts in customer accounts. Of course, increases in costs to securities dealers could ultimately be passed on to dealers' customers or to the Treasury in the form of higher borrowing costs; given the amount of fees assessed for securities-related overdrafts relative to the volume of U.S. Treasury and federal agency securities transferred via Fedwire each year, however (roughly $14 million in charges at the current 15 basis point fee versus $145 trillion of securities transfers in 1994), the incremental effect of any fees ultimately passed on to the securities market would be small. Indirect costs associated with daylight overdraft fees include costs incurred by institutions to reduce overdrafts and avoid fees. Such costs could include the investments by banks in systems to assess fees on customers, automate clearing procedures, or improve control over their cash balances. In addition, some government securities dealers have also reportedly made investments to improve their back-office processing of government securities transactions. Institutions are generally reluctant to provide proprietary information about such costs; a reasonable assumption is that institutions are not likely to have spent more to avoid overdrafts than they would have paid in daylight overdraft fees. An upper bound on this amount is the additional amount of fees that institutions would hypothetically have been assessed in the absence of the dramatic reductions in overdrafts in April 1994. Based on the experience in the six months before the implementation of fees, this amount could have ranged roughly between $15 million and $40 million annually based on daylight overdraft fees of 10-25 basis points.

PRIVATE NETWORKS

Since it implemented daylight overdraft fees, the Federal Reserve has taken additional steps to help control risks in the payment system, most recently in the area of multilateral netting systems. The November 1990 Report of the Committee on Interbank Netting Schemes of the Central Banks of the Group of Ten Countries (commonly referred to as the "Lamfalussy Report," after the committee's chairman, Alexandre Lamfalussy) identified six minimum standards that should be met by multilateral netting systems. In December 1994 the Federal Reserve Board adopted a revised policy statement on risks in large-value multilateral netting systems, which incorporated the minimum standards in the Lamfalussy report. The Board also stated that it would work with each netting system separately to determine whether higher risk management standards would be appropriate for those systems presenting a potentially higher degree of systemic risk.

Shortly thereafter, in July 1995, the New York Clearing House announced changes to CHIPS rules with the goal of reducing risk while maintaining cost effectiveness and system efficiency.(11) The changes included a 20 percent reduction in CHIPS net debit caps to be implemented in three stages through early January 1997, an increase in the minimum amount of collateral pledged by a participant, a modification to loss-sharing procedures in the event of a multibank default, and certain additional procedures for liquidating collateral pledged by participants. These changes, when implemented, are expected to result in a further increase in the certainty of settlement for payments made via CHIPS.

POTENTIAL FUTURE DEVELOPMENTS AFFECTING DAYLIGHT OVERDRAFTS

Further changes in market practices, greater use of private payment systems, and further consolidation in the banking system all might affect the volume and distribution of daylight overdrafts in the future.

Changes in Market Practices

Additional changes in market practices, such as increased netting of securities transactions, may have the potential to further reduce daylight overdrafts. For example, the Government Securities Clearing Corporation (GSCC) has recently begun accepting RPs into its trade matching system, and it plans to begin netting RPs in the near future. Netting of RPs would reduce overdrafts to the extent that they are caused by movements of securities across Fedwire between GSCC members or their customers.

Some private-sector participants have suggested another potential change--the establishment of settlement "windows" in which the settlement of federal funds transfers would be more closely synchronized, thus reducing the duration of associated overdrafts. Other potential changes in market practices, such as increased use of "rollovers," or continuing contracts, in the federal funds market do not appear to have been implemented widely to date, and the prospects for adoption are unclear.

The development of an intraday funds market, in which institutions engage in explicit lending transactions for periods of several hours, may also be a potential means of reducing overdrafts in Federal Reserve accounts; however, institutions have suggested that such a development may not be cost-effective at the current fee level. Institutions may find other means of charging their counter-parties for the use of intraday credit, however, such as by negotiating differential transaction prices based on the timing of settlements for federal funds or RP transactions. Such practices require the ability to monitor closely and to enforce precise settlement times, capabilities that do not currently exist at many institutions.

Greater Use of Alternative Payment Systems

Greater use of private payment systems could further trim daylight overdrafts in Federal Reserve accounts by reducing the volume of funds transfers through the Federal Reserve. Whether daylight overdraft fees will drive such a process is unclear, as fees are only one of a large number of factors--including service pricing and other associated costs, risks, and patterns of usage--that may influence choices among payment and settlement systems. For example, although CHIPS has historically been used primarily to settle international transactions, for some purposes market participants may consider CHIPS and Fedwire to be fairly close substitutes. Although the growth in dollar volume on CHIPS has outpaced that on Fedwire over the past few years, the differential does not appear to have become more pronounced since the imposition of fees (chart 10).

Greater use of private systems for securities transfers could also reduce use of daylight credit in Federal Reserve accounts. Participants in such systems would have to weigh costs--such as those of posting additional collateral to support additional clearing activity, as well as differences in settlement timing and finality--against any benefits to be realized from reduced daylight overdrafts in Federal Reserve accounts. Several proposals have been made to clear Fedwire-eligible securities on existing private securities-clearing networks, but no such service has yet been implemented. Some anecdotal reports also cite greater use of "book" transfers of government securities, those in which securities are transferred on the books of a bank rather than through the Fedwire system. This practice may be feasible when, for example, two RP counterparties agree to use the same custodian bank (a "tri-party RP").

In addition, some reports have suggested greater use of what are primarily retail payment systems, such as the ACH system, to settle interbank money market transactions. These systems are not, however, designed with the high levels of security and control over the timing of payments that characterize large-value payment systems. Moreover, their use in large-value transfers would not necessarily reduce daylight overdrafts in Federal Reserve accounts relative to current methods because of differences in the intraday timing of payments posted to Federal Reserve accounts.

Increased Consolidation in the Banking Industry

The current trend toward greater consolidation in the banking industry may well continue. Moreover, in July 1997, nationwide interstate branch banking will become effective in states that have not explicitly "opted out." As a result, holding companies may convert some or all of their banks into branch banking organizations. Thus, many payments that are currently settled between two banks within a holding company may become transactions processed on the books of a single interstate bank. This may, in turn, result in fewer payments and associated daylight overdrafts being recorded on the books of the Federal Reserve Banks.

CONCLUSION

The introduction of daylight overdraft fees has reduced the level of daylight overdrafts in Federal Reserve accounts, the aggregate amount of daylight credit provided by the Federal Reserve, and the associated direct credit risk, with little evidence of disruption in the payment system or in the financial sector generally. That such a large reduction in overdrafts was the result of a relatively small fee suggests that the economic inefficiencies created by the provision of free daylight credit by the Federal Reserve were substantial. To date, the response to the fees has been largely reflected in changes in practices in the government securities market, and the incidence of fees has fallen primarily on the largest depository institutions and dealers in government securities.

When the Federal Reserve Board in April 1995 increased the daylight overdraft fee to 15 basis points, it recognized that significant progress had been made in reducing overdrafts. Additional time would be needed, however, to encourage the study and evaluation of further changes in practices and market conventions that could help reduce overdrafts. Thus, the Board stated that it will evaluate the desirability of any additional changes in the fee after two years in light of experience with the current fee and the overall objectives of the payment system risk program.

(1.) For more information about CHIPS, see Bank for International Settlements, Payment Systems in the Group of Ten Countries (December 1993), pp. 449-51; and Federal Reserve Bank of New York, "The Clearing House Interbank Payments System" (January 1991). (2.) A summary of daylight overdraft issues and the Federal Reserve's policies is in Terrence M. Belton, Matthew D. Gelfand, David B. Humphrey, and Jeffrey C. Marquardt, "Daylight Overdrafts and Payments System Risk," Federal Reserve Bulletin, vol. 73 (November 1987), pp. 839-52. (3.) Under the Federal Reserve's Regulation J, Fedwire funds tranfers are final and irrevocable when the Reserve Bank credits the account of the receiving bank or sends the advice of payment to the bank, whichever occurs first (12 CFR 210.31). (4.) A discussion of settlement failure scenarios in netting arrangements is in David B. Humphrey, "Payments Finality and Risk of Settlement Failure," in Anthony Saunders and Lawrence White, eds., Technology and the Regulation of Financial Markets (Lexington Books, 1986), pp. 97-120. (5.) Among the studies on controlling payment system risk were Association of Reserve City Bankers, Issues and Needs in the Nation's Payments System (Washington, 1982) and Risks in the Electronic Payments Systems (1983); Large-dollar Payments System Advisory Group, A Strategic Plan for Managing Risk in the Payments System: Report . . . to the Payments System Policy Committee of the Federal Reserve System (Board of Governors of the Federal Reserve System for the ... Advisory Group, 1988); and Task Force on Controlling Payments System Risk, Controlling Risk in the Payments System: Report ... to the Payments System Policy Committee of the Federal Reserve System (Board of Governors of the Federal Reserve System, 1988). (6.) These rates are quoted on a ten-hour basis, which is the current length of the operating day of the Fedwire funds transfer system. On a twenty-four-hour basis, the measure employed in the Federal Reserve's policy statement, the initial fee was 24 basis points, to be increased, in phases, to 60 basis points. (7.) The aggregate peak daylight overdraft for a given day is the greatest value reached by the sum of daylight overdrafts in Federal Reserve accounts for all depository institutions at the end of each minute during the day. (This measure is not to be confused with the composite peak daylight overdraft, a measure that aggregates all institutions' peak daily daylight overdrafts regardless of their timing.)

The aggregate average per-minute daylight overdraft for a given day is the sum of average per-minute daylight overdrafts for all institutions on that day. An institution's average per-minute overdraft for a given day is the sum of its overdrafts at the end of each minute in the standard operating day of the Fedwire funds transfer system divided by the number of such minutes. (8.) Securities dealers would hold securities until near the close of the Fedwire securities transfer system (2:30 p.m. or later) to make sure that they could complete large deliveries first and avoid costly failures to deliver. This practice delayed the receipt of funds into the dealers' accounts and exacerbated daylight overdrafts at major securities clearing banks. (9.) Although the dollar amount of increased overdrafts was not large relative to peak overdrafts of $130 billion at the time, the new measurement method made compliance with net debit caps difficult for those institutions, mostly smaller, whose payment activity consists primarily of checks or ACH payments. In 1994, the Federal Reserve modified its net debit cap classes and procedures somewhat to avoid undue burden on these institutions. (10.) To encourage efforts to reduce overdrafts in the months leading up to the implementation of fees in April 1994, the Public Securities Association designated the week of December 6-10, 1993, as "Daylight Overdraft Prevention Week."

The earlier trading in the RP market has not had an adverse effect on implementation of monetary policy. However, to reduce the impact of its own activity on daylight overdrafts, the New York Reserve Bank's Open Market Desk adopted a policy of returning securities held under repurchase agreements later in the morning, at 11 a.m., thereby delaying the receipt of funds from the banking sector. For a discussion of daylight overdraft fees in the context of RP market activity and open market operations, see "Monetary Policy and Open Market Operations during 1994," Federal Reserve Bulletin (June 1995), pp. 579-81. (11.) New York Clearing House Association, CHIPS: Settlement Finality Improvements, Rules and Documents (July 1995).

Heidi Willmann Richards, of the Board's Division of Reserve Bank Operations and Payment Systems, prepared this article.